1h. .
. 1. .
u‘ , .. ,1 i r.

3.4 1&3 c5015.)“
..

9.
\r
2

1....
14..

. . ”WV“; A

‘Nbrhhxhlto
n.0,}: ‘

 

A. 37v.
: 25. 1.1.

(Ksa.4.!l

I'll

 

[Hilﬁib

 

\.
.4.

:i- ‘j } 9

This is to certify that the
dissertation entitled

PHYSICOCHEMICAL PROPERTIES AND RELATIVE
BIOAVAILABILITY OF FERRIC ORTHOPHOSPHATE IN
READY-TO-EAT CEREAL

presented by

Robin Sue Malone-Dickmann

has been accepted towards fulﬁllment
of the requirements for the

degree in Food Science and Human
Ph.D. Nutrition

 

 

ﬂu: '

Maj—or Profess r’s Signature

JMJLZS’; .2008

Date

 

MSU is an afﬁrmative-action, equal-opportunity employer

 

LIBRARY
Michigan State
Ul awersﬂy

 

--.-._.-—.-.—.-

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 Krlproj/AocaPrelelRCIDateDue.indd

 

PHYSICOCHEMICAL PROPERTIES
AND RELATIVE BIOAVAILABILITY OF FIVE SOURCES OF FERRIC
ORTHOPHOSPHATE IN READY-TO-EAT CEREAL
By

Robin Sue Malone-Dickmann

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Food Science and Human Nutrition

2008

ABSTRACT
PHYSICOCHEMICAL PROPERTIES
AND RELATIVE BIOAVAILABILITY OF FIVE SOURCES OF FERRIC
ORTHOPHOSPHATE IN READY-TO-EAT CEREAL
By

Robin Sue Malone-Dickmann

Ferric orthophosphate (FePO4) is a promising iron source for food fortiﬁcation
because of its light color and oxidative stability. However, FePO4 has had limited use in
cereal due its variable and often reported low bioavailability. An understanding of the
mechanisms underlying this variability may facilitate production of a F ePO4 source with
consistent bioavailability. The bioavailability of iron sources used for fortiﬁcation is
thought to be largely dependent on their solubility in the dilute HCl in the stomach.
Solubility is dependent on a combination of physicochemical properties, such as chemical
composition, particle size, surface area and crystal structure. This research determined
the relative bioavailability values (RBV) of two (control food) hydrogen-reduced iron
sources (Sources 2 and 3) and six commercially available FePO4 sources (Sources 4-9) in
ready-to-eat cereal (45% RDA, 0.27-0.30 mg iron/ g cereal) using the AOAC Rat
Hemoglobin Repletion Bioassay (ferrous sulfate standard, RBV 100%). The solubility of
the iron sources in dilute HCl was measured in vitro over the range of normal
physiological HCl concentrations (0.02, 0.05 and 0.1N) using the solubility method of
Shah and others (1977) that was modified to decrease assay variability (CV <1%).

Regression analysis found it to be a good predictor of RBV, R2 90%; P = 0.008). Particle

size distributions (PSD) and surface areas (SA) were measured by laser light diffraction
and nitrogen absorption, respectively.

The RBVs for the FePO4 sources ranged from 51 — 99 (P<0.05) and was higher
than that reported in the literature. Most of the PS distributions (PSD) were multimodal
and the sources could be grouped into three statistically different distributions (P<0.05).
The ﬁrst group had a uniform PSD (Median PS 16.5 um) and the lowest RBVs (51 &
60%); The second group was composed of a bimodal PSD of ﬁne particles (Median PS 2
pm) and intermediate RBVs (78 & 83%), and the remaining Source 4 (RBV 99%) had a
broad bimodal PSD (Median PS 9 pm) with one mode composed of larger 50 um
particles and the second mode composed of ﬁne 6 um particles. Regression analysis of
PS and SA were found to partially explain the variable RBV of FePO4.

In an effort to further explain the variability of RBV, the presence of amorphous
microstructure was investigated. Amorphous content is known to have important
mechanistic properties that affect the moisture absorption, reactivity and stability of many
materials. Very little data on the amorphous content of ferric orthophosphate exists in the
literature. Dynamic gravimetric vapor sorption was used to estimate the amount of
amorphous content (by moisture uptake). A regression model of median particle size and
moisture uptake versus solubility in 0.10 N HCl was found to be an excellent predictor of
solubility (R2 91%; P = 0.001).

In conclusion, two factors, mean particle size and moisture uptake (as determined
by amorphous content) were found to inﬂuence the solubility of FePO4, which may help

explain the variable RBV in the literature.

To my parents, William and Evelyn Malone

iv

ACKNOWLEDGMENTS

I would like to express my sincere appreciation to Dr. Dale Romsos, who
selﬂessly gave his time to mentor all of us in the Kellogg-MSU cohort, and my major
professor Dr. Gale Strasburg for their patience and never-ending encouragement and
support without which this achievement would not have been possible. They have both
been exemplary role models in their dedication to education, belief in their students and
high standards of professional integrity. I will continue to strive to meet their high
standards and share what I have learned with others.

I am grateful to my committee members, Dr. Maurice Bennink, Dr. James Steffe
and Dr. Alden Booren for their guidance and support. I would like to acknowledge all my
Kellogg mentors for their helpful suggestions and incredible support of this endeavor,
especially to Dr. Grace Lai for freely offering her good-natured insight, time and
technical expertise whenever I needed it. I would also like to thank Dr. Hsimin Huang, Dr.
Arti Arora and Dr. Donna Thede for their guidance, encouragement and certainty that I
would ﬁnish my dissertation. I am grateful to my Kellogg adjunct professors, Dr. Michael
McBurney and Dr. Neil Shay, who took time out of their busy schedules to provide
feedback on my research.

I give my sincere gratitude to the Kellogg Company and Michigan State
University for providing me with the opportunity to pursue a graduate degree through the
Kellogg-MSU cohort program. I am indebted to Dr. Donna Banks and Dr. Mark
Uebersax, who made this program a reality. Last but not least, thank you to all the

professors at MSU, who took the time to drive to Battle Creek to give thought-provoking

lectures in long evening sessions that accommodated the busy schedules of students
working full time at Kellogg’s.

Finally, I express my heartfelt gratitude to my family - my husband Glenn, and
children Thomas and Kathryn, who put up with late nights and a wife and mother who
was not always able to join in family activities and “have fun”. I am so thankful for their
love, support and understanding, which made it all worthwhile. I also want to thank my
brother Bill, who is an excellent teacher and statistician and who’s patience and humor
made statistics sister-ﬁ'iendly.

And special thanks to my dog, Penny, who never failed to offer her warm,

companionable presence during our many late night study and writing sessions.

vi

TABLE OF CONTENTS

LIST OF FIGURES ........................................................................................................ x
LIST OF TABLES ........................................................................................................... xii
LIST OF ABBREVIATIONS .......................................................................................... xiii
INTRODUCTION .......................................................................................................... 1
CHAPTER 1
LITERATURE REVIEW PART I: Biological Importance, Chemistry, Absorption and
Regulation, and Bioavailability ....................................................................................... 7
1.1 Importance of Iron ............................................................................................... 7
1.2 Iron Chemistry ..................................................................................................... 8
1.2.1 Solubility and Formation of ligands ........................................................... 10
1.3 Dietary Iron .......................................................................................................... 12
1.3.1 Heme and Non-Heme Iron .......................................................................... 13
1.3.2 Iron Absorption and Regulation .................................................................. 14
1.3.3 Iron Transport ............................................................................................ 16
1.3.4 Iron Regulation: Uptake and Loss .............................................................. 16
1.3.5 Factors Affecting Iron Absorption .............................................................. 17
1.3.5.1 Enhancers ........................................................................................ 17
1.3.5.2 Inhibitors ......................................................................................... 18
1.3.5.3 Micronutrient Interactions .............................................................. 19
1.4 Bioavailability ...................................................................................................... 19
1.4.1 Measures of in Vitro Bioavailability ........................................................... 20
1.4.2 Measures of In Vivo Bioavailability ........................................................... 24
LITERATURE REVIEW PART II: Iron Deﬁciency Anemia, Fortiﬁcation of Foods,
and Irons Sources ............................................................................................................. 27
1.5 Iron Deﬁciency Anemia ....................................................................................... 27
1.6 Iron Fortiﬁcation of Foods ................................................................................... 29
1.6.1 The Amount and Quality of Added Iron .................................................... 31
1.6.1.1 Recommended Dietary Intakes for Iron .......................................... 33
1.6.2 Effect of Iron on the Oxidatively Stability of Foods .................................. 35
1.7 Sources of Added Iron ......................................................................................... 38
1.7.1 Elemental Iron ............................................................................................. 41
1.7.2 Ferric Orthophosphate ................................................................................ 46
1.7.3 Bioavailability of Iron Forms used in RTE Cereal ..................................... 48
1.8 Physicochemical Properties and their Measurement .......................................... 50
1.8.1 Particle Size Analysis ................................................................................. 50
1.8.2 Surface Area Analysis ................................................................................ 53
1.8.3 Microstructure Structure and Moisture Sorption ....................................... 55

vii

CHAPTER 2
COMPARISON OF THE RELATIVE BIOAVAILABILITY OF SIX TYPES OF

F ERRIC ORTHOPHOSPHATE POWDERS IN READY-TO-EAT CEREAL ............ 58
2.1 Introduction ......................................................................................................... 58
2.2 Materials and Methods ......................................................................................... 61

2.2.1 Materials ..................................................................................................... 61
2.2.2 Methods ....................................................................................................... 62
2.2.2.1 Depletion Diet ................................................................................ 62

2.2.2.2 Repletion Diet ................................................................................ 63

2.2.2.3 Animals ......................................................................................... 65

2.3 Statistical Analysis ............................................................................................... 66
2.3.1 Hemoglobin Repletion Bioassay/slope ratio modeling ............................... 66
2.3.2 Hemoglobin Regeneration Efﬁciency ......................................................... 67
2.4 Results .................................................................................................................. 67
2.5 Discussion ............................................................................................................ 74
2.5.1 Hemoglobin Regeneration Efﬁciency versus Hemoglobin Repletion ....... 75
2.5.2 RBV of Iron treatments ............................................................................... 76

CHAPTER 3

PHYSICOCHEMICAL PROPERTIES OF F ERRIC ORTHOPHOSPHATE

INFLUENCING ITS SOLUBILITY IN DILUTE HCL ................................................. 79
3.1 Introduction .......................................................................................................... 79
3.2 Materials and Methods ......................................................................................... 85

3.2.1 Materials ..................................................................................................... 85
3.2.2 Methods ....................................................................................................... 86
3.2.2.1 Solubility in Dilute Hydrochloric Acid .......................................... 86
3.2.2.2 Polarized Light Micrographs ........................................................ 87
3.2.2.3 Particle Size by Laser Light Diffraction ........................................ 88
3.2.2.4 Surface Area Analysis by Nitrogen Adsorption ............................ 89
3.2.2.5 Moisture Uptake by Dynamic Vapor Sorption .............................. 90
3.2.2.6 Preparation of Amorphous Ferric Orthophosphate ........................ 90
3.3 Statistical Analyses .............................................................................................. 91
3.4 Results .................................................................................................................. 92
3.4.1 Solubility in Dilute Hydrochloric Acid ....................................................... 92
3.4.2 Particle Size, Surface Area and Polarized Light Microscopy ..................... 94
3.4.2.1 Particle Size Analysis .................................................................... 94
3.4.2.1.] Multinomial Curve Fitting of Particle Size Data .......... 95
3.4.2.2 Surface Area Analysis ................................................................... 95
3.4.2.3 Polarized Light Microscopy .......................................................... 96
3.4.3 Dynamic Vapor Sorption Isotherms ........................................................... 105
3.4.3.1 Semi-Quantitation of Amorphous Content .................................... 111
3.4.4 Regression Analysis and Prediction Of %RBV ........................................... 114
3.5 Discussion ........................................................................................................... 120

viii

CHAPTER 4
SENSORY PROPERTIES OF RTE CEREAL FORTIFIED WITH F ERRIC

ORTHOPHOSPHATE ................................................................................................... 128
4.1 Introduction .......................................................................................................... 128
4.2 Materials and Methods ......................................................................................... 129

4.2.1 Materials ..................................................................................................... 129
4.2.2 Methods ....................................................................................................... 129
4.3 Results .................................................................................................................. 130
4.4 Discussion ............................................................................................................ 131

SUMMARY AND CONCLUSIONS ............................................................................. 133

FUTURE RESEARCH .................................................................................................... 136

APPENDICES
A. Beckeline Test Results for Refractive Indices of Ferric Orthophosphate ............ 137
B. Surface Area Instrument Parameters ................................................................... 138

LIST OF REFERENCES ................................................................................................. 139

ix

LIST OF TABLES

Images in this dissertation are presented in color

Table 1.1 Cost and bioavailability of iron sources used for food fortiﬁcation ............. 39
Table 1.2 Cost, iron content, color and relative bioavailability of iron sources ........... 40
Table 1.3 Manufacturers and forms of food-grade elemental iron ............................... 45
Table 1.4 Literature Percent Relative Bioavailability Values (%RBV) for human,

rat and in vitro Studies ................................................................................... 49
Table 1.5 Inﬂuence of particle size on bioavailability of elemental iron powders ........ 50
Table 2.1 Identity of test iron sources for the hemoglobin repletion diets ................... 62
Table 2.2 Composition of iron depletion diet ............................................................... 63
Table 2.3 Composition of iron repletion diets .............................................................. 64
Table 2.4 Iron content of diets by analysis ................................................................... 65
Table 2.5 Hemoglobin repletion study data and RBV results .............................................. 70
Table 2.6 Mean Hemoglobin Regeneration Efﬁciency Values (HRES) Compared

to Relative Biological Values (RBVS) .......................................................... 72
Table 3.1 Iron Sources used for particle size and surface area measurements ............. 85
Table 3.2 Solubilities of iron sources in dilute hydrochloric acids solutions ................ 93
Table 3.3 Results of particle size and surface area analyses compared to RBV ............ 97
Table 3.4 Particle size sub-populations by multinomial curve ﬁtting .......................... 98
Table 3.5 Dynamic vapor sorption (DVS) results for Sources 4-8 ............................... 113
Table 3.6 ..Comparison of Actual (By Analysis) to Predicted RBV for Sources 4-8 ..... 120
Table 4.1 Vitamins and minerals added to test cereals .................................................. 129
Table 4.2 Sensory scores of RTE cereal made with iron Sources 1-9 ........................... 131
Table 1.0 Appendix A Reﬁ'active Indices for Ferric Orthophosphate Sources .............. 137

Figure 1.2

Figure 2.1

Figures 3.1

Figure 3.2

Figure 3.3.

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.11

Figure 3.12

LIST OF FIGURES

Images in this dissertation are presented in color

Molecular structure of ferric orthophosphate ........................................ 46

Dose-response of hemoglobin repletion in rats using nine different
sources of iron ........................................................................................ 73

Particle size distributions of Sources 4 — 8 by laser light diffraction
and polarized light photomicro graphs ................................................. 99

Plot of the percent change in mass due to moisture uptake at
incremental changes in percent RH over time for two isotherm
cycles for Source 4 ............................................................................... 107

Source 6: Isotherm mass plot showing areas of mass increase
(sorption) and mass decrease (desorption) ........................................... 108

Source 4: Isotherm mass plot for Showing areas of mass increase
(sorption) and mass decrease (desorption) ........................................... 109

Isotherm hysteresis plots for Sources 4-8 showing the change in

mass differences between sorption and desorption moisture uptake

and loss at each incremental change in %RH ..................................... 110
Mass plot showing three isotherm cycles for the spray-dried
(approximately 100% amorphous material) sample of ferric
orthophosphate .................................................................................... 1 12
Curvilinear (3.7a) and semi-log (3.7b) plots of Sources 4-8 ............... 115

Regression analysis of log 10 (solubility iron in HCl) vs. surface area for
Sources 4-8 .......................................................................................... 116

Regression analysis of log 10 (solubility iron in HCl) vs. median particle
Size for Sources 4-8 .............................................................................. 116

Regression analysis of log 10 (solubility iron in HCl) vs. % moisture

uptake for Sources 4-8 ........................................................................ 117
Regression analysis of % RBV vs. surface area for Sources 4-8 ......... 117
Regression analysis of % RBV vs. particle size for Sources 4-8 ......... 118

xi

Figure 3.13 Regression analysis of %RBV vs. moisture uptake for Sources 4-8 ...118

Figure 3.14 Regression model of % moisture uptake vs. median particle size to
predict % RBV for all lots of Sources 4-8 .......................................... 119

xii

AOAC
ATP
Aw
B.E.T.
DRI
DVS
EAR
FAO
FCP04
FF C
HgB
HRE
ICP
IDA
INACG
MPS
MOST
PS
PSD
RBV
RDA
RH
RTE
SA
SUSTAIN
USAID
WHO
W1C

LIST OF ABBREVIATIONS

Association of Ofﬁcial Analytical Chemists

Adenosine Triphosphate

Water Activity

Brunauer, Emmett and Teller (Theory)

Dietary Reference Intakes

Dynamic Vapor Sorption

Estimated Average Requirement

Food and Agricultural Organization of the United Nations
Ferric Orthophosphate

Food Chemical Codex

Hemoglobin

Hemoglobin Regeneration Efﬁciency

Inductively Couples Plasma

Iron Deﬁciency Anemia

International Nutritional Anemia Consultative Group
Mean Particle Size

The USAID Micronutrient Program

Particle Size

Particle Size Distribution

Relative Biological Value

Recommended Daily Allowance

Relative Humidity

Ready-tO—Eat (cereal)

Surface Area

Sharing United States Technology to aid in the Improvement of Nutrition
United States Agency for International Development
World Health Organization

Women Infant and Children Special Supplemental Nutrition Program

xiii

INTRODUCTION

Anemia remains the most prevalent micronutrient deﬁciency in the world. Iron
deﬁciency anemia (IDA) accounts for a large percentage of the total problem affecting
600 — 800 million people worldwide with women, infants and young children most at risk
(INACG 2001; Turner 2002). The World Health Organization (WHO) estimates that
approximately 50% of children, 42% of women and 26% of men are affected in
developing countries, while 2-3 8% of the population is at risk in developed countries
(Nalubola and Nestel 2000; Benoist 2001; Martinez-Navarrete and others 2002; Miret
and others 2003). In May of 2000, the WHO and the US. Agency for International
Development through the International Nutritional Anemia Consultative Group (INACG)
reexamined the nature and magnitude of the public health problem at the Belmont
Conference (Stoltzfus 2001b). The participants of the Belmont Conference undertook a
multidisciplinary critical review process to redeﬁne the nature of the public health
problem of IDA on the basis of its link to functional outcomes in human populations.
Conclusions drawn from the conference proceedings found that in the majority of cases,
the cause of IDA is diet related. Yet despite diet-intervention strategies implemented in
almost all countries, IDA remains endemic throughout the world and new strategies are
needed to address the problem (Stoltzfus 2001b).

In developing countries, government intervention programs focus on fortifying
cereal staples for poorer segments of the population. Industrialized countries fortify a
variety of foods, including cereal staples such as wheat ﬂour, corn meal and rice, ready-

tO-eat (RTE) cereal and products targeted at infants and young children. Foods are

enriched by law or are voluntarily fortiﬁed by food manufacturers. Elemental iron
powders are the most commonly used iron fortiﬁcants worldwide because they are
affordable, cause the least problems with food quality, and have good consumer
acceptance. However, conclusions drawn by experts at the Monterey Workshop, held in
Monterey, Mexico in 2000 to address conﬂicting data on absorption of elemental iron,
found that not enough is known about the extent to which elemental iron powders are
absorbed by the body (Hurrell 2002a; Turner 2002). As a result of the Monterey
Workshop, the ‘Sharing United States Technology to Aid in the Improvement of
Nutrition’ or SUSTAIN group commissioned a review of the literature published over the
past 45 years and found variable iron bioavailabilityl results. In addition, little
information was available on the precise types of iron used in these intervention
programs and nutritional studies.

Iron is a very reactive element and there are various types of iron, some Of which
are used as fortiﬁcants, including elemental iron and various ferric and ferrous iron salts.
The different iron forms have different degrees of reactivity, which generally impair the
organoleptic properties and shelf life of foods, thus limiting the iron forms used in foods
to those that are less reactive. Unfortunately, the physical properties that make an iron
source less reactive also are thought to make it less well absorbed. These challenges
make the prevention of iron deﬁciency through food fortiﬁcation difﬁcult {F airweather-
Tait, 2002 #66; Hurrell, 2002 #132; SUSTAIN Task Force, 2001 #29; Turner, 2002

#133}. Consequently, experts believe that food stability limitations and the conﬂict or

 

1 For the purposes of this research, bioavailability is deﬁned as the ability of the human body to
digest, absorb, and use an iron source for metabolic functions (W ienk KJH, Marx JJ, Beynen AC.
1999. The concept of iron bioavailability and its assessment. Eur J Nutr 38:51-75.

lack of literature information on bioavailability have hindered the successful worldwide
implementation of cereal enrichment programs. These barriers raise questions about the
efﬁcacy of iron fortiﬁcation (Lynch 2000; Hurrell 2002a; Turner 2002).

The Recommended Daily Intake (RDI) for iron is 18 mg/day (FDA 1999). Most RTE
cereals cannot be fortiﬁed with elemental iron above 25% RDI per serving. Empirical
experience at the Kellogg Company has shown that oxidatively sensitive cereal such as
whole-grain, non sugar-coated cereals often can not exceed 10% RDI even when less
reactive iron powders are chosen; however, certain forms of ferric orthophosphate have
been found to have minimal negative effects on product quality even at high fortiﬁcation
levels (>40% RDI/serving). Ferric orthophosphate is an affordable alternative iron
source that is used when color, density, and stability issues prevent the use of elemental
iron powders. Historically, nutritional experts have considered the bioavailability of ferric
orthophosphate to be poor (Shah and others 1977; Hurrell 1985; Forbes and others 1989;
Hallberg and others 1989; Kellogg 1997; Willis and Allen 1999; Nalubola and Nestel
2000; SUSTAIN Task Force 2001; Dary and others 2002). Although there is less
research data available on ferric orthophosphate, a review of the literature suggests that
the bioavailability information is as variable as that for elemental iron powders. And
unlike elemental iron powders, the physical and chemical properties that inﬂuence the
bioavailability of ferric orthophosphate are not well understood. In order to better
understand the factors inﬂuencing bioavailability, the ﬁrst aim of this research was to
determine the bioavailability of commonly used sources of ferric orthophosphate that on

visual inspection appeared to have different physical characteristics.

Six types of commercially available food- grade, ferric orthophosphate powders
were selected after screening for visual differences in color, appearance, ﬂow properties
and supplier—speciﬁed mean particle size. Examination of these sources by polarized light
microscopy revealed marked differences in microstructure and degree of crystallization
and presence of amorphous material. Ferric orthophosphate is a semi-crystalline salt. The
presence of amorphous structure in semi-crystalline materials has been given increasing
attention in the pharmaceutical sciences over the past ten years and it is now recognized
that even relatively low levels of amorphous material (<10%) affect stability and
dissolution characteristics (Buckton and Darcy 1999; Mackin and others 2002; Burnett
and others 2006). Therefore, the hypothesis of this research is that the amorphous content
of ferric orthophosphate is a critical determinant Of bioavailability due to its greater
capacity to absorb water and, therefore, increase solubility of this material in the gut.

In addition to the six sources of ferric orthophosphate, this study also included
ferrous sulfate as the standard of comparison that by deﬁnition has a relative
bioavailability (RBV) of 100%, hydrogen-reduced iron used commercially to fortify RTE
cereal, and encapsulated hydrogen-reduced iron that can be used at higher fortiﬁcation
levels due to its greater oxidative stability. Bioavailability was determined using the
AOAC Rat Hemoglobin Repletion Bioassay, which is a widely accepted method for
predicting bioavailability in human subjects and has shown excellent agreement with
human clinical studies (Forbes and others 1989). This method was suitable for
consistently screening a relatively large number of iron sources in one study.

The second aim of this research was to measure the physicochemical properties of

the iron sources used in the bioavailability study. Besides chemical composition,

properties such as particle size, surface area, porosity and microstructure all inﬂuence
important physicochemical properties, such as solubility. There is a great deal of
variation in the literature on the methodology used to measure the solubility of iron
fortiﬁcants and the outcome of these methods is strongly inﬂuenced by pH, temperature
and Shear. This variability makes inter-laboratory comparisons difﬁcult. For this research,
the solubility method of Shah (Shah and others 1977) was modiﬁed to decrease assay
variation by performing carefully controlled experiments in a dissolution apparatus used
to determine the solubility of dietary supplements in the gastrointestinal tract. A range of
dilute HCl concentrations that simulated differences in human gastric juice was studied as
a function of time to understand the effect of pH on solubility.

Particle size distributions were determined using laser light diffraction, and
surface areas were measured by nitrogen adsorption. Both these methods are widely used
to evaluate and specify elemental iron powders. Several techniques have been developed
to measure amorphous content, including dynamic (gravimetric) vapor sorption (DVS),
which measures the higher moisture uptake of amorphous versus crystalline forms of the
same material. DVS was chosen because of its sensitivity for very low levels of
amorphous content, and because it allows a direct measure of the moisture-induced
changes affecting the dissolution properties of ferric orthophosphate. Moisture uptake is
the main cause of food quality loss in RTE cereals. Solubilization of iron is a
prerequisite for absorption in the small intestine and is, therefore, critical to iron
bioavailability.

The ﬁnal aim of this research was to determine the relationship between the

physicochemical properties of ferric orthophosphate and their combined effect on

solubility as predictors of bioavailability and also on the organoleptic properties of a
standard RTE cereal. Regression analyses was used to determine how well solubility
predicts bioavailability and also what physical properties most impact solubility and
bioavailability.

This dissertation is organized into a series of chapters that build on information
from the previous chapter using the same lots of ferric orthophosphate throughout all the
experiments. Each chapter has a speciﬁc introduction, materials and methods, results and
discussion sections. In chapter 2, the bioavailabilities of six commercially available food-
grade ferric orthophosphate powders were determined in comparison with three other
iron sources. An in vitro solubility method was optimized in Chapter 3, and particle size
distributions and surface area were measured for the iron sources. Moisture uptake was
also measured by DVS in Chapter 3 to estimate the amorphous content of the ferric
orthophosphate samples. Finally, in Chapter 4, a standard RTE cereal was fortiﬁed at
45% RDA with each of the iron sources and the sensory attributes of the test cereals
evaluated. Sections common to the entire dissertation include the initial abstract,
introduction, literature review, ﬁnal conclusions, future research and the list of

references.

CHAPTER 1

LITERATURE REVIEW PART I
Biological Importance, Chemistry, Absorption and Regulation, and Bioavailability

1.1 BIOLOGICAL IMPORTANCE OF IRON

Iron is an essential nutrient to all living organisms whose cells rely on iron to play
a key role in a plethora of biochemical processes, which include electron transfer
reactions, gene regulation, binding and transport of oxygen and regulation of cell growth
and differentiation. Iron enables the ﬁnal energy-yielding steps of the electron-transport
chain, a complex metabolic system that functions in the terminal oxidation of nutrients to
form water and adenosine triphosphate (ATP). In addition, enzymes that catalyze the
production of amino acids, hormones and neurotransmitters require iron (Whitney and
others 1998b; IOM 2001; Beard 2006).

The body’s iron storage pool can contain as much as 1 to 4 g of iron. The majority
of iron is bound to three proteins in the human body: almost two-thirds is found in the
hemoglobin of circulating red blood cells, twenty-ﬁve percent is in the storage protein,
ferritin, and another ﬁfteen percent is in muscle myoglobin. Iron bound to tissue ferritin
serves as a reservoir to meet the body’s requirements of iron not provided by the diet.
Nearly all cells contain ferritin, although 60% of the body’s ferritin is found in the liver
and spleen. Approximately 95% of the iron stored in the liver is bound to fen'itin in the
heptocytes and approximately 5% is bound to hemosiderin in the Kupffer cells. The
remaining 40% is found in muscle tissues and the cells of the reticuloendothelial system.

Iron mobilized from tissue is transported by the iron transporter, transferrin. Transferrin

comprises a small pool of about 5 mg labile iron, and is normally 25-50% saturated with
iron. When iron stores are depleted, the saturation of transferrin will decline to less than
15%, resulting in reduced transport of iron to iron-binding proteins that take part in
essential reactions. The chemical reactions in which iron participates are classiﬁed as
oxygen transport and storage, electron transfer, and substrate oxidation-reduction. These
reactions are an integral part of metabolism and occur in all cells (Beard 2006). There
are four major classes of iron-containing proteins that play critical roles in metabolic
reactions: 1) The iron—containing, non-enzymatic hemoproteins (hemoglobin, myoglobin)
that function as ligands to bind and transport oxygen; 2) The iron-sulﬁir enzymes,
ﬂavoproteins and heme-ﬂavoproteins, that participate in single electron-transfer
reactions, primarily in energy metabolism; 3) The heme-containing enzymes associated
with various cofactors, which participate in electron-transfer reactions such as
cytochrome P450 complexes; and 4) the other iron-activated enzymes that lack a
porphyrin ring structure or iron-sulfur complex (Whitney and others 1998a; IOM 2001;

Lynch 2002; Beard 2006).

1.2 IRON CHEMISTRY

Iron is the fourth most abundant metal in the Earth’s crust and is a member of the
transition elements on the Periodic Table of the Elements. The most important property
of the transition metals is their ability to exist in different oxidation states and the fact
that they can easily move between reduced and oxidized states. This unique redox
chemistry is due to the gradual ﬁlling of the d electron orbitals and allows iron to take

part in electron transfer reactions. This property also allows iron to act as a chelator and

reversibly bind ligands such as oxygen, nitrogen and sulfur in biological processes (Hider
and Singh 1993). The electronic spin state and biological redox potential can vary
signiﬁcantly depending on the protein and ligand. This gives iron a great deal of
versatility to participate in a large number of reactions (Winterboum 1991; Beard 2006).

In nature, iron can exist in oxidation states ranging from —2 to +6 (Murphy and
Rousseau 1969; Keenan and Wood 1971; Arora 1997). The ferric form, F e+3 , and ferrous
form, F e+2 (and transient ferryl +4 states involved in biochemical oxidation reactions) are
the only forms that occur naturally in living tissues (Silver 1993). Elemental iron, FeO, is
not found naturally (except in rare instances), but is a by—product of several
manufacturing processes and is puriﬁed for use as a fortiﬁcant in foods. Many common
reducing agents such as ascorbic acid will reduce ferric to ferrous iron with the addition
of an electron, and exposure to oxygen will quickly oxidize ferrous iron back to the ferric
iron form. Iron’s ability to cycle between oxidation states and to reversibly bind ligands
form the basis for its two most important chemical properties that are critically important
to all life forms: 1) The redox potential between the two common oxidation states of iron,
Fe+3 and its reduced form Fe+2, allows oxidation processes centered on iron to be readily
coupled to metabolic processes; and 2) Iron’s high afﬁnity for oxygen. Iron-containing
proteins widely utilize these two properties for many life-sustaining ﬁmctions (Silver
1993; Beard and Dawson 1997; Miret and others 2003; Beard 2006). In contrast, the
properties of iron also allow it to catalyze and propagate deleterious reactions involving
oxygen and nitrogen making it a potential toxicant as well as an essential nutrient

(Dianzani 1991 ).

Very little free iron is present in living tissues since it readily participates in redox
reactions, which is the basis of iron’s inherent toxicity. Iron is most toxic when it is non-
speciﬁcally bound to the surface of proteins and membranes, as is the case in iron
overload (poisoning) and certain disorders of iron metabolism. In the presence of
molecular oxygen, weakly bound iron can cycle between its two oxidation states and this
can result in the production of the highly reactive OH' radical from the Haber-Weiss-
Fenton reaction shown below (Dianzani 1991; Hider and Singh 1993; Symons and

Gutteridge 1998; Beard 2006).

Fe+2 + 02 —> Fe+3 + 02'
202- + ZI‘I+ -—) H202 + 02

2 . .. - +
Fe+~ + H202 —> OH + 'OH + Fe 3
The potential toxicity of iron requires tight physiological homeostasis control to regulate
iron uptake, incorporation into cells and proteins, storage, release and transport (Dianzani

1991).

1.2.1 Solubility and Formation of Ligands

In acidic solutions (pH <7), ferrous ions are rapidly oxidized to ferric ions. Ferric
ions quickly form hydrates in acidic environments, e. g. F e(HzO)(,+3 and Fe(HzO)6+2. As
the pH increases (becomes more alkaline) and more hydrogen ions are removed, water
molecules give up protons to form the corresponding iron hydroxides, Fe(OH)2 (ferrous)
and Fe(OH)3 (ferric). Iron hydroxides decrease in solubility as the pH becomes more
alkaline and eventually will precipitate out of solution. Ferrous hydroxide will precipitate

out of an aqueous solution as a pale green gel having a solubility of about 10'1 M at pH 7

10

(Symons and Gutteridge 1998; Miret and others 2003). Ferric hydroxide is much less
soluble (10'16 M) at pH 7 and precipitates out of solution as a dark brown powder (rust).
The formation of sparingly soluble hydroxides (and especially ferric hydroxides) has
nutritional signiﬁcance, as the solubility of iron is the necessary ﬁrst determinant of
bioavailability.

In the acid environment of the stomach, soluble iron is hydrated. The hydrated
iron passes into the small intestine where bile and other digestive secretions increase the
pH to a slightly alkaline environment (pH 7). At this pH, iron hydroxides would
eventually precipitate out of solution. However, other components are present in foods,
called ligands, that act to keep iron complexed and soluble in the small intestine (Symons
and Gutteridge 1998; Miret and others 2003; Beard 2006).

In foods, iron forms soluble ligands with organic molecules such as amino acids,
peptides, carboxylic acids (e. g. citric and ascorbic acids), polyols (sugars), phosphates,
and with food-additive sequestrants like ethylenediaminetetraacetic acid, or EDTA.
Recent work by Huh and others (2004) found that the active fractions in cooked ﬁsh that
enhanced iron absorption were highly enriched with carbohydrate and contained
negligible amounts of protein or amino acids. The authors speculate that these enhancing
fractions may originate from glycosaminoglycans of muscle tissue. Depending on the
ligand, one iron atom can form from one to six bonds with the ligand. The stability of the
complex increases as the number of ligand bonds to the iron atom increases. A ligand that
forms more than two bonds, as in the case of EDTA, is called a polydentate ligand. A
metal complex formed with a polydentate ligand is very stable and is referred to as a

chelate. Soluble ligands naturally present in food are able to form weak chelates by

11

binding only a few of the six possible coordination sites of iron. These unstable ligands
keep gastric acid-solubilized ferric ions in solution at the pH in the duodenum and serve
as iron donors to mucin, which enhance reduce ferric iron to ferrous iron for absorption.
Certain other dietary constituents are able to form stable chelates or precipitates with
iron. Constituents that may form precipitates, such as oxalates, carbonates, phytates, and
tannates, may interfere with the binding of iron and mucin. The formation of chelation
complexes during digestion may be in part responsible for the conﬂicting research on the
biological availability of iron from foods (Lee and Clydesdale 1979; Benito and Miller
1998). Analogous to the binding of iron to food ligands during digestion, the endogenous
biological ligands for iron are oxygen, nitrogen and sulfur atoms in organic molecules
like hemoglobin, myoglobin, chlorophyll and ferritin (IOM 2001). The chemistry of
ligand formation controls the absorption and regulation of iron in living systems and

prevents non-speciﬁc binding of iron.

1.3 DIETARY IRON

There are four sources of iron in the diet: animal origin; plant origin; fortiﬁcation
iron; and contamination iron (e. g. dirt, minerals in water, iron cooking utensils). Most of
the world’s population relies on foods of plant origin and many populations do not have
access to fortiﬁed foods. The main plant-based food staples include wheat, rice, maize,
potatoes and beans. The endogenous iron in these foods has moderate to low
bioavailability (Morck and Cook 1981). The typical American diet contains about 50%

of its iron from grain products in which the iron concentration is typically between 0.1

12

and 0.4 mg of iron per serving. On the other hand, some fortiﬁed cereal products have

been found to contain more than 18 mg of iron in a single serving (Beard 2006).

1.3.1 Heme and Non-Heme Iron

There are two fundamental forms of iron in the diet: heme and non-heme iron.
Heme iron comes exclusively from animal-based food sources and non-heme iron is
supplied from inorganic iron compounds present in animal and plant-based foods. Heme
iron is well absorbed and only slightly inﬂuenced by dietary factors while the absorption
of non-heme iron is strongly inﬂuenced by its solubility and interaction with other meal
components during digestion. The two forms of iron follow different pathways for
absorption from the gut and incorporation into the intestinal mucosal cells. Both have a
common pathway out of the mucosal cells and into the plasma (Hallberg 1981; Benito
and Miller 1998; Whitney and others 1998b; Miret and others 2003).

Non-heme iron makes up the majority of iron in the diet but its availability for
absorption is very dependent on the properties of the iron compound, the meal
composition and other factors operating in the lumen of the stomach and proximal small
intestine. The iron status of the individual is the key determinant of the amount of iron
absorbed from a meal. The rate of absorption for heme iron is relatively constant at about
23% but can be as high as 45%, depending on iron status. The rate of absorption for non-
heme iron ranges from 2- 23% (Hallberg and Hulthen 2000). Dietary iron absorption is

increased approximately 4-fold once iron stores are totally depleted (Lynch 2002).

13

1.3.2 Iron Absorption and Regulation

Non-heme iron absorption depends on the solubilization of predominantly ferric
iron from the acid milieu of the stomach (IOM 2001). Most food iron is already in the
ferric form or it is quickly oxidized in the acid environment of the stomach. Ferric iron
must be kept solubilized by components in the diet capable of chelating the ferric iron
such as citrate and amino acids, and reduced to its ferrous form by compounds such as
ascorbic acid or the ferrireductase enzyme present on the mucosal cell surface of the
upper small intestine. This bioavailable iron is them absorbed in a 3-step process where
iron is ﬁrst taken up by the enterocytes across the cellular apical membrane by an energy-
dependent, carrier mediated process; absorbed iron is then transported intracellularly and
transferred across the basolateral cell membrane into the blood plasma. The mechanism
of transport through the enterocytes is not fully elucidated. However, it is a well accepted
that only soluble ferrous iron can be absorbed during digestion, and once solubilized and
absorbed, ferrous iron joins a common intracellular “labile iron pool” for use by the body
(IOM 2001; Miret and others 2003). The iron is available for binding by transferrin in
plasma and transported via transferrin throughout the body. Substantial evidence exists
to support the commonly held theory that all non-heme iron mixes together in a
intracellular “iron pool” in the cells of the upper gastrointestinal tract (Beard 2006). The
microenvironment in the gut provides a proton gradient directed toward the cell interior,
which together with the brush border membrane will provide the necessary driving force
for iron uptake (Benito and Miller 1998; Miret and others 2003).

Solubilized ferrous iron is a prerequisite to iron absorption. Iron uptake is

regulated by the total amount of iron ingested, the form of iron, the composition of the

14

 

meal and iron-status of the individual. Individuals with a high iron status will absorb
proportionally less of any amount of iron consumed than will an iron-deﬁcient individual.
Individuals with a lower iron status will absorb more of any dietary form of iron. This
process is called “selective absorption” and is the fundamental mechanism of iron
balance in humans (Beard 2006).

Ferric ions are rapidly converted to ferrous iron by a membrane-bound member of
the cytochrome P450 family, called Dcyth, which is present in abundance and is not a
rate-limiting factor (Beard 2006). Specialized transport mechanisms in the enterocytes at
the tips of the duodenal villi control iron absorption. Non-heme ferrous iron is transported
into the intestinal enterocytes by the divalent metal transporter, DMTl. Ferrous iron
taken up by DMTl and internalized via vesicle endocytosis is then either stored in
association with ferritin or exported into the plasma at the basolateral membrane via
transferrin. Homeostatic control of iron uptake and basolateral membrane transfer
appears to by controlled by the recently discovered plasma signal-protein, hepcidin.
Hepcidin is a putative plasma protein regulator of iron absorption. It is released from
liver into the plasma in concentrations proportional to the amount of liver iron stores (and
also cytokine regulation as part of the immune function). Hepcidin is associated with a
protein on the basolateral membrane called ferroportin (MTPl) which releases iron into
the plasma. Hepcidin appears to have two primary targets, the basolateral membrane of
the enterocytes and macrophages. As hepcidin concentrations vary, so does the release of
iron from ferroportin. This in turn regulates non-heme iron uptake from the lumen by
DMTI. When physiologic iron needs are high, there is a higher concentration of DMTl

on the enterocytes.

15

1.3.3 Iron Transport

Iron released into the plasma binds to transferrin. Transferrin’s primary function
is to move iron from one organ to another. Increased production of tranferrin occurs
when iron stores are low resulting in decreased plasma iron concentrations. Some iron is
delivered to the myoglobin of muscle cells and to other tissues while the bone marrow
takes up large quantities and incorporates iron into the hemoglobin of red blood cells.
The liver and spleen dismantle aging red blood cells that have a life expectancy of about
4 months. When iron in the diet is plentiful, the liver stores iron bound to ferritin and
breaks it down rapidly to supply iron when the need arises. When iron is limited,
hemosiderin is the main storage form of iron. Hemosiderin is a water-soluble breakdown
product of ferritin that releases iron more slowly to conserve its supply (Whitney and

others 1998a).

1.3.4 Iron Regulation: Uptake and Loss

The human body hoards iron very effectively and has no physiological means of
iron excretion. Iron absorption is the sole mechanism by which iron stores are managed.
The average adult stores 1 to 4 grams of iron, approximately 70—80 % of which is found
in erythroid cells such as hemoglobin and another 20-25% stored in association with
ferritin. About 1-2 mg are lost per day through the sloughing of intestinal and skin cells;
menstruation accounts for an average iron loss in females of about 2 mg per day.
Approximately 3 mg or 0.1% of the total iron in the body circulates in the plasma as an

exchangeable iron pool. Essentially all of this iron is chelated to transferrin to render the

16

iron soluble under physiologic conditions, prevent iron—mediated free radical toxicity and
facilitate iron transport into cells (Bridges 2007).

The body efﬁciently conserves iron and the primary regulator of iron homeostasis
is regulation of iron absorption at the enterocytes to approximate iron loss. Iron’s poor
solubility precludes excretion from the body as a means of maintaining iron homeostasis,
which is an important regulator of other minerals in the body. The predominant route of
iron loss is from the gastrointestinal tract (shed enterocytes, extravasated red blood cells
and biliary heme breakdown products) and amounts to 0.6 mg/d in adult males.
Urogenital and integumental losses have been estimated at 0.1-0.3 mg/d, respectively, for

adult males and menstrual blood losses range for 1.5 — 2.1 mg/d (Beard 2006).

1.3.5 Factors A jfecting Iron Absorption

Most naturally occurring non-heme food iron is in the form of ferric iron salts and
several dietary factors affect its absorption in the gut. Enhancers of absorption include
amino acids, animal proteins, ascorbic acid, increased hydrochloric acid secretion in the
stomach, and organic acids. Inhibitors include carbonates, calcium, egg yolk phosvitin,
ﬁber, oxalates, phytates, plant polyphenols, some ﬂavonoids and soy proteins (Hallberg

1981; Beard 2006).

1.3.5.1 Enhancers
Ascorbic acid is an important luminal enhancer of iron absorption. It acts as a
reducing agent at acid pH and as a low-afﬁnity ligand in the upper duodenum. Ascorbic

acid maintains iron in a soluble, low molecular weight form in the slightly alkaline

l7

environment of the upper small intestine (Hallberg and Hulthen 2000; Miret and others
2003). Ascorbic acid has also been shown to counteract the major inhibitors of iron
absorption such as phytic acid and polyphenols. Iron absorption from foods containing
phytates or polyphenols increases up to four—fold with the addition of ascorbic acid
(Hallberg and Hulthen 2000). Several studies have shown that the absorption of non
heme iron is increased signiﬁcantly from meals containing ascorbic acid. The effect is
greater for poorly water-soluble ferric iron sources (Forbes and others 1989; Moretti D
and others 2006). Animal proteins provide amino acids and peptides to form ligands with
iron that aid solubility and absorption during digestion. Organic acids maintain iron in its
ferrous form and also create ligands that enhance solubility and protect iron from
inhibiting factors. The low pH of the stomach acid is important for releasing iron from
macromolecules in food. The action of digestive enzymes, active in an acidic
environment, destroy high molecular weight ligands and create new, smaller and more

soluble ligands (Miret and others 2003).

1.3. 5.2 Inhibitors

Polyphenols and phytates are the major inhibitors of iron absorption (Hurrell
2002b; Lynch 2002). Polyphenols, phenolic compounds and phytates occur naturally and
are widespread in foods such as cereals, vegetables, spices, wine, chocolate, coffee and
tea. The lignin and phenolic constituents of insoluble ﬁber bind iron and the oxalates
present in certain grains (wheat bran) and vegetables (spinach and rhubarb) to form

highly insoluble complexes with ferric iron. Other inhibitors are present in eggs

l8

 

 

(phosvitin) and soy protein. Some ﬂavonoid compounds chelate iron similar to

polyphenols (Beard 2006).

I. 3. 5.3 Micronutrient Interactions

The potential risk of interactions between micronutrients competing for
absorption in the gastrointestinal tract should also be considered in a fortiﬁcation
program. At the naturally occurring levels of micronutrients present in foods, most
micronutrients appear to utilize speciﬁc absorptive mechanisms and are not vulnerable to
competitive interactions. At higher intake levels, competition between elements with
similar chemical characteristics and uptake-regulated processes can occur. The
interactions have clearly been demonstrated in absorption studies and to some extent been
conﬁrmed in supplementation studies (Sandstrom 2001). (Lynch 2000; Sandstrom 2001).
Several researchers have shown that the potential exists for calcium to decrease the
absorption of non-heme iron and this calcium fortiﬁcation levels should be considered in

foods fortiﬁed with both calcium and iron (Fairweather-Tait and Teucher 2002).

1.4 BIOAVAILABILITY

The terms bioavailability and absorption are often used interchangeably and the
deﬁnition of bioavailability can vary (Wienk and others 1999). In a comprehensive
review on the concept of bioavailability and its assessment by Weink and others (1999),
the authors discussed the topic of bioavailability from the perspective of three scientiﬁc
disciplines: nutritional science, animal nutrition and pharmacology. In human nutritional

sciences the concept of bioavailability is regarded as the efﬁcacy with which nutrients are

19

utilized. The view of animal scientists is similar except that it focuses on the nutritive
value of a diet to support growth and maintenance. In the ﬁeld of pharmacology,
bioavailability is deﬁned as the fraction of the active substance that reaches systemic
circulation after an oral dose is administered. This research will follow the deﬁnition of
Weink and others (1999) and use the concept of bioavailability from a nutritional
perspective: “Bioavailability is a function of digestibility, absorbability, and the ability
to use a nutrient for metabolic functions. ” The authors also make the important
distinction that bioavailability must be quantifiable if it is to be useful and that the
working deﬁnition of iron bioavailability is determined by the methods available to
measure it.

Bioavailability can be thought of as occurring in three stages: First is the
digestibility or the solubility of the iron source in the gut; second is how well the iron is
absorbed and delivered to the circulation; and third is how well the iron is processed by
the body once it reaches the circulation, target organs and cells for incorporation into
measurable functional entities. Iron bioavailability measures always assess one of these

three stages.

1.4.1 Measures of In Vitro Bioavailability

It is a generally accepted that only soluble ferrous iron can be absorbed during
digestion (Miret and others 2003). Tests to measure solubility can be simple dissolution
studies in various concentrations of hydrochloric acid or more complex methods that
mimic conditions during digestion. The methods that mimic digestion use a combination

of a stomacher (device that simulates the mixing of food contents in the stomach) and

20

dialysis to simulate the transport of soluble iron into the mucosal membranes of the
lumen (Forbes and others 1989). Subsequent soluble iron analysis is accomplished using

chromogens (or-or' dipyridyl, bathphenanthroline disulphonate, potassium dichromate and

ferrozine), atomic absorption spectroscopy, mass spectroscopy (stable isotopes) or y-
counting of radioisotopes (Wienk and others 1999). The physiological approaches use a
pepsin-HCl digestion step followed by neutralization using either bicarbonate or sodium
hydroxide and pancreatin-bile extracts to simulate the gastrointestinal tract (Larsson and
others 1997).

The greatest limitation of studies using these methods is their inability to model
the physiological processes of absorption and incorporation of iron. The dialysis
approach comes closer to simulating physiological conditions but cannot take into
account the in vivo complexity of the digestive process, such as variation in transit time,
roles of enzymes, pH, mucosal membranes and barriers to diffusion. Solubility studies
that evaluate the dissolution of different iron sources in dilute hydrochloric acid solutions
do not take into account the effects of food composition and physiological conditions. In
addition, the different HCl solubility methods have not been standardized with respect to
concentration (pH), incubation temperature, incubation time and method of iron analysis
(Harrison and others 1976; Rasmussen and others 1977; Shah and others 1977; Forbes
and others 1989; Hurrell and others 2002; Swain and others 2003). However, when any
one procedure is used for the analysis of a single type of iron, these individual methods
will give relative results that can be used to compare the solubilities of different forms of

the same material.

21

The solubility of several types of iron fortiﬁcants has been analyzed using HCl
solubility methods. The methods were developed for elemental iron sources, including
hydrogen and carbon monoxide-reduced, carbonyl and electrolytic iron and have been
applied to various iron salts such as ferric orthophosphate, ferric pyrophosphate, ferric
fumarate and others. Typically a small amount of elemental or iron salt containing 50 -
100 mg iron is added to 200 — 500 ml of dilute HCl at concentrations ranging from 0.02 -
0.25 N for various times and at varying temperatures. Different mixing techniques,
including the use of magnetic stir bars, water-bath shakers and wrist-action shakers have
been used to mix the samples during dissolution. For example, in a study by Forbes and
others (1989) using a modiﬁed method of Shah and others (1977), aliquots of electrolytic
iron and ferric orthophosphate were weighed to contain 50 mg of iron and added to 250
m1 of 0.02 N HCl (pH 1.7). The mixture was gently shaken for 30 minutes at a
temperature of 37° C. After 30 minutes dissolution time, between 60 -75% (30-375 mg)
of the electrolytic iron and between 1.4 —— 3.4% (0.7-1.7 mg) of the ferric orthophosphate
dissolved. Forbes and others (1989) also measured in vivo relative bioavailability of
electrolytic iron and ferric orthophosphate using the AOAC Rat Hemoglobin Repletion
Bioassay in two laboratories. They found RBVs from 66-78% and 24-34%, for reduced
iron and ferric orthophosphate, respectively. The authors concluded that in vitro
solubility in 0.02 N HCl underestimated the bioavailability of ferric orthophosphate in
human and animal studies. Both iron sources used in the experiments were produced in
the authors’ laboratory (bench-scale) and information on particle size and other physical

characteristics, such as surface area and density, were not given.

22

In contrast, Harrison and others (1976) measured the solubilities of ﬁve
commercial sources of ferric orthophosphate with known mean particle sizes (MPS) in
0.1 N HCl (pH 1.0) and found that sources with smaller MPS were more soluble.
Samples with a 15 um MPS had 11% of the iron analyzed dissolve while powders with a
MPS <1 um had solubilities of between 42 — 64% (Harrison and others 1976). Swain and
coworkers found two measures, solubility in dilute HCl and the surface area of reduced
iron powder, to be better predictors of in vivo bioavailability than particle size using the
rat hemoglobin repletion method (Swain and others 2003). Hallberg and coworkers
studied the dissolution rate of crystalline and amorphous ferric orthophosphate in HCl at
pH 1.0 (0.1 N), pH 1.5 (0.03 N), pH 2.0 (0.01 N) and pH 3.0 (0.001 N) at 30, 45 and 60
minutes and found that between 5 and 55% of the total iron in the samples dissolved in
the different concentrations of HCl. Results from this study also demonstrated that both
amorphous and crystalline ferric orthophosphate samples were completely soluble at pH
2 2.0. The authors’ method for establishing the presence of amorphous material was not
given (Hallberg and others 1989), and a quantitative measure of amorphous or crystalline
content was not provided. Researchers at the USDA, (Willis and Allen 1999)
hypothesized from their work with gypsy moth diets that the amorphous form of ferric
orthophosphate was necessary for human bioavailability. These authors developed a
solvent fractionation procedure to separate crystalline and amorphous ferric
orthophosphate.

The literature indicates that pH, particle size, surface area and the presence of
amorphous material inﬂuence solubility. However, information on the physical properties

of the iron sources used in the literature was very limited. Recently, The International

23

Anemia Nutritional Consultative Group (INACG), at the conclusion of their 2001
Symposium on global iron fortiﬁcation, concluded that not enough is known about the
extent to which elemental iron powders are absorbed by the body (Hurrell 2002a; Turner
2002). As a result of the Belmont Conference proceedings, The ‘Sharing United States
Technology to Aid in the Improvement of Nutrition’ or SUSTAIN commissioned a
review of the literature published over the past 45 years and found variable iron
bioavailability results. Little information was available on the precise types of iron used
in the various studies and programs. Consequently, experts concluded that conﬂicting or
lacking literature information on physicochemical properties and their relation to
bioavailability have hindered the successful implementation of cereal enrichment
programs worldwide and questioned the efﬁcacy of iron fortiﬁcation of foods (Lynch
2000; Hurrell 2002a; Turner 2002). Therefore, there is a need for standardized methods
and better understanding of the relationship of the physicochemical properties of iron

sources and solubility, which is an indicator of bioavailability.

1.4.2 Measures of In Vivo Bioavailability

Endpoint measures in animals and humans currently provide the best estimate of
bioavailability when solubility, absorption and the ability to use a nutrient for metabolic
functions are all in question. An endpoint method gives a quantitative measure of the
nutrient in a representative part of a metabolic function. In the case of iron, hemoglobin is
the target biomarker of iron of bioavailability.

Iron absorption was originally studied by chemical balance techniques. It was not

until the use of radio-labeled foods that absorption from individual foods was shown to

24

differ greatly. The concept of a common iron pool resulted from intrinsic double-radio
labeling experiments. When single foods were biosynthetically labeled with an intrinsic
iron tracer and mixed with an inorganic iron source labeled with a different radioisotope
(an extrinsic tracer), the observation was made that the absorption of the two tracers from
the doubly labeled foods was almost the same. An important implication of this method
was that all the components of the meal contributed to the bioavailability of non-heme
iron in the meal, including components that by themselves originally contain little or no
iron. This led to the theory of a common non-heme iron pool (Hallberg 1981). With the
recognition that different sources of non-heme iron form a common pool in the intestinal
lumen, it was no longer necessary to use the intrinsic tracer and a more convenient
method, using only the extrinsic tracer was developed. An extrinsic tracer alone may be
used because the small amount of radio-labeled iron added to the meal has been shown to
exchange with the intrinsic non-heme iron present, allowing determination of iron
absorption by the absorption of radio-labeled iron. Although use of an extrinsic tag has
allowed the study of the effect of individual components of meals on the bioavailability
of non-heme iron, conditions must be chosen to assure complete exchange (Forbes and
Erdman 1983). In the case of certain reduced iron powders and iron salts, like ferric
orthophosphate and ferric oxide, iron exchange with the extrinsic tracer in the common
iron pool may be in incomplete (Hallberg 1981). Other disadvantages of this method are
exposure of human subjects to radiation and the fact that it is based on a single meal or at
most two meals administered over a 30-day period (Forbes and others 1989; Benito and
Miller 1998; Wienk and others 1999; Miret and others 2003). Less expensive alternatives

to human studies are still needed, especially in vitro studies to screen and study factors

25

that affect bioavailability under controlled conditions that can’t be duplicated in living
systems.

In a comprehensive collaborative study by Forbes and others (1989) that
compared human radioisotope techniques with iron solubility methods and the AOAC
Rat Hemoglobin-Repletion Bioassay, they concluded that the standard AOAC rat model
was the economical method of choice for predicting iron bioavailability in humans.
Other reviews consider this method obsolete on the basis that the physiology of iron
absorption in rats is different from that than humans (Forbes and others 1989; Benito and
Miller 1998; Wienk and others 1999; Miret and others 2003). Dutra—de-Oliveira and
others (1995), compared the bioavailability of ferric EDTA, ferric bisglycinate and ferric
orthophosphate using the AOAC Rat Hemoglobin Repletion Assay. They measured mean
hemoglobin and hematocrit levels, transferrin saturation and liver iron stores. Results
showed that the iron from ferric orthophosphate was not as well incorporated into
hemoglobin or stored in the liver as the iron from soluble ferrous sulfate or other soluble

the ferric chelates (Dutra-de-Oliveira and others 1995).

26

LITERATURE REVIEW
PART II

Iron Deﬁciency Anemia, Fortification of Foods, and Iron Sources

1.5 IRON DEFICIENCY ANEMIA

Nutritional anemia remains the most prevalent micronutrient deﬁciency in the
world and iron deﬁciency anemia (IDA) accounts for a large percentage of the total
problem. Commonly cited estimates show that IDA affects 20-50% of the poorer
populations in Asia, Africa and Latin America and 2-28% in the poorer populations of
developed countries (Nalubola and Nestel 2000; Hurrell and others 2002; Martinez-
Navarrete and others 2002; Miret and others 2003). The most commonly affected
segments of the population are pregnant women (56%), school-age children (42%),
women (44%) and preschool-age children. Worldwide, the highest incidence of IDA
occurs in Saharan Africa and Southeast Asia (Nalubola and Nestel 2000). Many
estimates have been given on the total percentage of the world’s population suffering
from IDA with some estimates, according to the International Nutritional Anemia
Consultative Group (INACG), being questionably high. In 2001, INACG gave a
conservative estimate (extrapolated from 1985 data) of 600-800 million people suffering
from IDA worldwide.

In May of 2000, scientists at the Belmont Nutritional Conference sponsored by
the WHO and INACG, re—examined iron-deﬁciency anemia and the magnitude of the
public health problem. They evaluated the strength of the causal evidence linking iron-

deﬁciency to global health outcomes. Strong causal evidence was found to correlate

27

severe anemia with increased child and maternal mortality, and mild to moderate iron-
deﬁciency anemia with impaired childhood development (delayed psychomotor
development and impaired cognitive development) and decreased work productivity in
adults. The outcome of the proceedings concluded that “given the widespread prevalence
of mild to moderate iron-deﬁciency anemia and the public health importance of the
outcomes in question, there is an urgent need to elucidate these potentially causal
relationships” (Stoltzfus 2001b). Many intervention strategies were reported as not
working as well as predicted and experts recognized that multiple new strategies were
needed.

Following the Belmont Conference, a workshop was held in Monterey, Mexico in
September 2000 to address the conﬂicting data on absorption of elemental iron and to
evaluate its usefulness as a cereal fortiﬁcant. The SUSTAN group presented their
ﬁnding from a commissioned a review of the literature published over the past 45 years
that found variable iron bioavailability results. Experts at the Monterey Workshop
concluded that not enough is known about the extent to which elemental iron powders are
absorbed by the body (Hurrell and others 2002; Turner 2002). They found little
information available on the precise types of iron used in literature and in the intervention
programs and identiﬁed a need to better understand the factors inﬂuencing absorption
elemental iron (Hurrell 2002b).

Iron is a very reactive element and there are various types of iron, some of which
are used as fortiﬁcants, including elemental iron and various ferric and ferrous iron salts.
The different iron forms have different degrees of reactivity, which generally impair the

organoleptic properties and shelf life of foods, thus limiting the iron forms used in foods

28

to those that are less reactive. Unfortunately, the physical properties that make an iron
source less reactive also are thought to make it less well absorbed. These challenges
make the prevention of iron deﬁciency through food fortiﬁcation difﬁcult (SUSTAIN
Task Force 2001; Fairweather—Tait and Teucher 2002; Hurrell 2002b). The public health
importance of combating anemia has been widely recognized since the 1960’s and
measures to control anemia are implemented in almost all countries. The World Bank has
identiﬁed diet-based micronutrient interventions as among the most cost-effective of all
health interventions. Despite these efforts and the improvement seen in the nutritional
status of populations for other micronutrients, iron-deﬁciency remains endemic
throughout the world (Benoist 2001).

The effectiveness of a food fortiﬁcation program is dependent on both the
quantity and quality of the iron sources used, access to and frequency of use of the
fortiﬁed food by the at risk population and the composition of the whole diet of the target
population. The bulk of iron-deﬁciency cases are diet-related and despite the challenges,
diet-related intervention strategies remain important control measures. Although not as
effective as planned, they are responsible at least in part, for the reduction in IDA in
certain populations (ADA 2001; Turner 2002; FDA Department of Health and Human

Services 2003).

1.6 IRON FORTIFICATION OF FOODS
IDA has important negative health and economic consequences in countries
where it is a major health problem. For this reason, many countries are evaluating

strategies to combat iron deﬁciency and food fortiﬁcation of staple foods is considered

29

the most cost-effective, long-term approach (Stoltzfus 2001a; Hurrell 2002a). Cereal
staples include ﬂour made from wheat, corn, rice and staple foods made from these grains
such as porridge, bread, biscuits, pasta, and corn meal. Unlike the successful initiatives
to improve the intake of vitamin A, iodine and B vitamins using staple food vehicles,
enrichment with iron has not been as successful (Stoltzfus 2001a; Stoltzfus 2001b; Dary
2002; Hurrell 2002a). Several reasons combine to contribute to this lack of success. One
of the major reasons is that iron is a very reactive transition metal with multiple valence
states that allow it to participate in oxidation and reduction reactions and act as a catalyst
for undesirable reactions in food. The form of iron added to a food may change during
processing, storage and even during digestion (Lee and Clydesdale 1979). The reactivity
of iron, which enables iron to play a critical role in metabolic processes is associated with
the deterioration of organoleptic properties and shortened shelf life of foods. Thus, the
reactivity of iron limits the amount and type of iron that can be used for food fortiﬁcation,
usually limiting choices to iron less reactive forms. Unfortunately, the physical properties
that make an iron source less reactive also make it less bioavailable. These challenges
make the prevention of iron deﬁciency through food fortiﬁcation with non-heme iron
sources difﬁcult. (SUSTAIN Task Force 2001; F airweather-Tait and Teucher 2002).

In more developed countries, where the economy supports varied food choices,
food manufacturers voluntarily fortify processed products in addition to staple foods.
Government-subsidized programs help lower the purchase price of fortiﬁed foods for
poorer segments of the population and make these food items more widely accessible to
those who need it most. RTE cereals along with composite ﬂour, and products targeted

at infants and young children, play an important role in dietary intervention strategies.

30

Fortiﬁed RTE breakfast cereals are a major source of essential nutrients in many
countries and are of particular importance in the US, UK, Canada, Central and South
America and Australia where they play a signiﬁcant role in the diet and are consumed
regularly. Several govemment-sponsored initiatives are aimed at improving the
nutritional status of women and children and cereal plays an important role in these
programs. In the US, the School Breakfast Program and the Special Supplemental
Nutrition Program for Women, Infants and Children (WIC) are important initiatives to
safeguard the health of low—income women and children. WIC-eligible cereal products
must contain 45% of the RDA for iron per serving (270 ppm) (Oliveira and others 2002).

Achievement of this mandate is technologically challenging.

I. 6.1 The Amount and Quality of Added Iron

The FDA recommends that producers who voluntarily fortify products follow the
guiding principle of their position statement: “Nutrients may be added to a food to correct
a dietary insufﬁciency that is recognized by the scientiﬁc community to exist and known
to result in nutritional deﬁciency disease if sufﬁcient information is available to identify
the nutritional problem and the affected groups.” However, other than the requirement
that iron sources used to fortify foods be of food grade quality and compliant with the
quality standards set forth in the US. Pharmacopoeia (USP) or the Food Chemical Codex
(FCC), there are no regulations mandating the quality of iron to add from a
bioavailability perspective. The nutrition community highly recommends the use of iron
sources with known bioavailability and discourages the use of poorer quality forms

(Turner 2002).

31

Hydrogen-reduced iron is the most widely used iron source for food fortiﬁcation
due to its relatively low cost and availability in a range of particle sizes, surface areas,
porosity and purity. Its availability over a wide range of physical parameters allows food
manufacturers some control over the form used for a particular ingredient or product
application. As discussed earlier, bioavailability and reactivity are related to the
physicochemical properties of the iron form. Currently, particle size (and purity) is the
physical property most often used by food manufacturers to specify reduced iron powders.
Milling and sieving are inexpensive and fast procedures used to determine the particle
size fractions of reduced and electrolytic iron powders, which generally consist of dry
particles with a maximum particle size of 1 mm (Hurrell 2002a). The Food Chemical
Codex (FCC) guidelines require that reduced iron powders used to fortify food pass
through a 100-mesh screen or sieve (particle size < 149 um) and that electrolytic and
carbonyl powders pass through a 325-mesh Sieve (particle size < 44 um) (Government
2006). Most manufacturers set their speciﬁcations for reduced iron powder at the 325-
mesh screen size even though this higher quality standard is not stipulated by the FCC.
The SUSTAIN Task Force on the iron fortiﬁcation of cereal ﬂours recommends a
maximum 325-mesh speciﬁcation. According to the SUSTAIN Task Force’s 2002 report,
other physical parameters such as surface area, porosity and purity, have not been
sufﬁciently investigated to enable the formulation of clear quality guidelines for these
parameters (Hurrell 2002a). Although other forms of iron, such as the iron salts and iron

chelates are used to fortify foods, only reduced iron has an FCC guideline.

32

I. 6. 1.] Recommended Dietary Intakes for Iron

At the time of this writing, the Recommended Dietary Allowances (RDAS) of
essential nutrients, established by the Food and Nutrition Board (FNB) of the National
Academies, Institute of Medicine (IOM) are under revision. The RDAS have been re-
evaluated by the FNB periodically since their inception in 1943, with the latest (10m)
edition published in 1989. Over the last few decades, the original focus of the RDAS,
which was the prevention of classical nutritional deﬁciencies, has expanded to include
fortiﬁcation levels for optimal health and information that could be used for individual
dietary planning as well as population assessments. In the early 1990’s the FNB in
collaboration with Health Canada undertook the task of revising the RDAS and Canadian
Recommended Nutrient Intakes (RNIs). The outcome of this effort was a new framework
that consists of multiple sets of nutrient speciﬁc guidelines collectively referred to as
Dietary Reference Intakes or DRIS. The DRIS are composed of a family of four nutrient
reference values: 1) Recommended Daily Allowances (RDA), 2) Estimated Average
Requirements (EAR), 3) Adequate Intakes (AI), and 4) Tolerable Upper Intake Levels
(UL). The DRIS are slated to replace and expand the traditional RDAS in the US. and
Recommended Nutrient Intakes (RNIs) in Canada (Penland 2001).

The new DRIS for iron were published in 2001. The RDA for iron for boys and
girls between the ages of 9-13 is 8 mg/d (UL 40 mg/d). The RDA for boys 14-18 years of
age is 11 mg/d, while the RDA for girls of the same age is increased to 15 mg/d (UL for
adolescents is 45 mg/d).The RDA for healthy adults between the ages of 19 and 50 was
set at 8 mg/d for males and 18 mg/d for females (UL for adults is 45 mg/d). Men over the

age 0f 50 should continue to consume 8 mg/d and women over 50 years of age (or

33

postmenopausal) need to decrease their iron intake to 8 mg/d (IOM 2001). The new DRIS
for iron are lower for children, adolescent men and older adults as compared to the
previous RDAS. However, at the time of this writing, the original 1989 RDAS remain the
reference guidelines for food labeling.

The food labeling system uses Recommended Daily Intakes (RDIs) for vitamins
and minerals, based on the 1989 RDAS, for moderately active young adults consuming
2000 calories per day. Daily Reference Values (DRV) are estimated for nutrient
components that do not have an RDA, such as ﬁber, protein, total fat, saturated fat,
cholesterol, sodium, potassium, and carbohydrate. The RDI for iron is 18 mg/d for adults
and children (IOM 2001). Cereal products are typically fortiﬁed at between 10 — 25% of
the RDI for iron or 1.8 — 4.5 mg iron per typical 30-gm serving of RTE cereal (60 to 150
ppm iron). Iron present at levels as low as 0.1 ppm can decrease the lipid oxidation
induction period and accelerate oxidation rates (Nawar 1996). Sugar-coated cereals and
cereals high in natural antioxidants, like wheat bran, can withstand higher levels of iron
fortiﬁcation but most whole grain, low sugar cereals cannot withstand levels beyond 60
ppm or 10% of the RDI for iron due to the oxidatively sensitive polyunsaturated fats
inherent in the grain. A sugar-coating is also an excellent oxygen and moisture barrier for
cereal, which uncoated cereals lack. The W1C program requirement is 45% RDI per
serving or 270 ppm, which is an extremely challenging level of iron fortiﬁcation to meet

for any product.

34

I . 6.2 Effect of Iran on the Oxidative Stability of Foods

Lipid oxidation is one of main causes of food quality loss and spoilage and is of
great economic concern to the food industry. Oxidation of fat in foods leads to off-odors
and off-ﬂavors (rancidity) that shorten shelf life. Oxidative reactions can decrease the
nutritional quality of a food and certain products are potentially toxic (Nawar 1985).
Lipid oxidation is a complex process and proceeds by a classic free-radical chain
mechanism involving three discrete phases: 1) Initiation, the abstraction of a hydrogen
atom from the carbon or to the double bond in a methylene-interrupted, unsaturated fatty
acid in the presence of an initiator to form a carbon-centered lipid radical; 2) Propagation,
the lipid radical reacts with molecular oxygen to form peroxy radicals and lipid
hydroperoxides; and 3) termination, in which two radicals combine to form non-radical
end-products that do not propagate further reactions (N awar 1985; Arora 1997). The
initiation step requires the formation of free radicals and is thermodynamically
unfavorable, requiring an initiator. Transition metals that possess two or more valence
states with a suitable oxidation-reduction potential between them (iron, copper, cobalt,
manganese, nickel) act as major pro-oxidants. If soluble, reactive iron is present, even at
concentrations as low as 0.1 ppm, it can act as a catalyst and decrease the length of the
oxidation induction period and increase the rate of oxidation. When iron is used as a food
fortiﬁcant, added amounts often exceed 30 ppm and may exceed 200 ppm. Iron sources
used at such high levels must have properties that limit reactivity in the food but not
bioavailability in the gut. Few iron sources meet this criterion, often resulting in the use
of iron forms that are less bioavailable than desired in order to meet product quality

standards.

35

Other initiators of lipid oxidation are singlet oxygen ('02), ultraviolet light, and
high temperatures (Simic and others 1992). Lipid degradation in foods may be

propagated by multiple reactive species, which include hydrogen peroxide (H202), lipid
hydroperoxide radical (HOO'), superoxide anion (Oz' '), hydroxyl radical ('OH), peroxy

radical (ROO'), and the highly reactive alkoxyl radical (RO'). Several mechanisms for

metal catalysis are postulated:

I) Acceleration of h ydroperoxide decomposition to form alkoxyl radicals
M “t + ROOH —+ M ‘“*”*+ OH‘+ RO’
M “++ ROOH —’ M ‘”"’++ H++ R00

2) Direct reaction with unoxidized substrate
Mn+ +RH __, M(n-1)++H++R.

3) Activation of molecular oxygen to give singlet oxygen and peroxy radicals

_ -e'
M““+02 —> M‘“"”*+02‘ <
+

+H

102

HO;
(Nawar, 1985;(Fennema 1996))

One of the simplest free radical—generating reactions involving iron is:
O; + Fe (II) <—> '02— + Fe (III)

The oxygen radical formed in this reaction, superoxide anion, is a potent initiator of lipid

oxidation. The superoxide radicals recombine in the following reaction:

2 '02— + 2H+ -—) H202 + 02

36

 

Hydrogen peroxide generated via this reaction becomes a source of other radicals and
reacts with free metal ions like iron and copper, as well as some of their complexes,
thereby generating highly reactive free radicals. These are the Haber-Weiss and Fenton
reactions. (Simic and others 1992; IUPAC 2003). The Haber-Weiss cycle consists of the

following reactions:

H202 + OH —) H2O + 02- + H+ 311d

H202 + 02- —) 02 + OH + OH-

The second reaction has a negligible rate constant and it is believed that iron (11)

complexes catalyze the reaction via the Fenton reaction:
H202 + F e(ll) -—> ‘OH + OH“ + Fe(III)

Many complexes of iron with a low redox potential generate the superoxide
radical, 0;. Reducing agents, such as ascorbate, act to keep recycling ferric iron back to

reactive ferrous iron via the following reaction (Simic and others 1992; IUPAC 2003):
Fe (III) + AH— —> Fe (II) + 'A + H+

In this situation, the action of ascorbate to produce reactive ferrous ions is potentially

deleterious to biological molecules and foods. Lipid oxidation by ionic iron requires the

presence of ferrous iron and the subsequent production of oxygen radicals that go on to

produce hydroxyl radicals. Reactive oxygen species are required to abstract hydrogen

from the carbon or to site of unsaturation (double bond) on a lipid molecule to begin the

oxidation process (Decker and Hultin 1992; Simic and others 1992).

37

1.7 SOURCES OF ADDED IRON

Even though many compounds are available for use as iron fortiﬁcants, only ﬁve
compounds are commonly used: elemental iron powders, ferrous sulfate, ferric
orthophosphate, ferric pyrophosphate, and ferrous fumarate (Hurrell 1997). A few other
highly bioavailable sources are recommended, such as ferric EDTA, and ferric
bisglycinate, but these forms are cost-prohibitive in many food applications, and limited
due to their reactivity in others. Elemental reduced iron powder is the most commonly
used fortiﬁcant worldwide and particularly in cereals because of its relatively low cost
and low reactivity during processing and in the ﬁnished food during storage. Based on
1970 estimates that had not changed signiﬁcantly as of 2002 (SUSTAIN Task Force
2001; Hurrell and others 2002), it is estimated that reduced iron powder accounts for 40-
55% of the iron added to foods, followed by ferrous sulfate (25 -30%) and iron
phosphates (15-25%). Other compounds, like ferrous fumarate, contribute only 1 to 5%
of the iron added to foods (Lee and Clydesdale 1979; Nalubola and Nestel 2000). The
relative cost of the commonly used fortiﬁcants compared to ferrous sulfate (considered
the optimum fortiﬁcation source for bioavailability) is given in Table 1.1 (Hurrell 1985;
1997).

Table 1.1 provides a comprehensive list of the many iron sources available for
fortiﬁcation purposes, categorized by their relative water solubility. The forms listed are
classiﬁed as freely water-soluble, slowly water-soluble, poorly water-soluble and
insoluble. In general, water-soluble iron sources score higher in relative bioavailability
tests in rats and human studies because they are readily soluble during digestion.

However, the more water-soluble sources are also more prone to cause off-ﬂavor and off—

38

color problems due to their increased availability to catalyze and participate in

undesirable reactions.

Table 1.1 Relative cost (to ferrous sulfate) and bioavailability of iron sources used
for food fortiﬁcation '

 

 

 

 

 

 

Relative
Approximate Bioavailability Relative cost
Fe content (RBV) 2 compared to

Iron Source (%) Rat Human FeSO4
Freely water soluble

Ferrous sulfate- 7H 20 20 100 100 1. 0

Ferrous gluconate 12 97 89 5.1

Ferrous lactate l 9 106 4.1

Ferric ammonium citrate 18 107 - 4.1
Slowly water soluble

Anhydrous ferrous sulfate 33 100 100 0.65

Ferric saccharate 10 92 74 5.2

Ferric citrate 17 73 31 4.8
Poorly water soluble

Ferrous fumarate 33 95 100 1.3

Ferrous succinate 35 119 92 4.1

Ferrous tartrate 22 77 62 3 .9

Ferrous saccharate 10 92 74 5.2
Water insoluble

Ferric pyrophosphate 25 45-58 21-74 2.3

Ferric orthophosphate 28 6-46 25-32 4.1
Elemental iron powders

Electrolytic 98 44-48 5-100 0.5

Carbonyl 98 39-66 5-20 1 .0

Hydro gen-Reduced 97 24-54 1 3-148 0.2
Protected Compounds

NaFeEDTA 14 na 28-416 6.0

 

IAdapted from (Hurrell 1985) and (Hurrell 1997)

2 Percent Relative Bioavailability Value (RBV) is determined by a ratio comparison to of the iron
source to ferrous sulfate; ferrous sulfate is considered to have excellent bioavailability and a RBV
of 100% and is the reference standard used in most in vivo animal and human assays.

39

 

 

 

The differences in the cost of various iron fortiﬁcants can be quite large on a
manufacturing scale. Costs of the most commonly used sources are given in Table 1.2
along with information on iron content, color and bioavailability. Iron content is an
extremely important consideration as it can dramatically impact the cost of the raw
material and total amount of the iron source added. Equally important are the other
constituents present in the iron compound used. The sulfur, phosphate, amino acid,
organic acid, etc. moieties impart ﬂavor, affect pH, solubility, ionic strength and react

with other components of food.

Table 1.2 Cost, iron content, color and relative bioavailability of iron fortificants 1

 

 

 

 

 

 

 

 

 

 

 

 

 

Relative
Amount Cost Cost Bioavailabili

Iron Source (“/0 Fe) ($/Kg) (S/Kg Fe) Color in Man (%)
Ferrous sulfate 32 2.40 7.50 White 100
Ferrous 33 3.00 9.09 Red 100
fumarate
Ferric Yellow —
orthophosphate 28 5.00 17.86 pinkish white 24-32
Reduced iron 97 2.00 2.06 Black 13.853
Electrolytic
iron 98 6.70 6.84 Dark gray 5-100
Ferric EDTA 13 8.70 66.92 Dark yellow 100
Ferrous
bis a), cinate 20 20.26 101.30 Grey-green 100

 

I Adapted from (Nalubola and Nestel 2000)

2 Percent Relative Bioavailability Value (RBV) is determined by a ratio comparison of the
test iron source to ferrous sulfate; ferrous sulfate is considered to have excellent
bioavailability and a RBV of 100% and is the reference standard used in most in vivo
animal and human assays.

3 Dependent on the type and physical properties of reduced iron powder used

40

 

 

I. 7.1 Elemental iron

As mentioned previously, elemental (reduced) iron powder is the most widely
used iron fortiﬁcant. Cereal fortiﬁed with reduced iron serves as the reference against
which alternative iron sources like ferric orthophosphate are compared (SUSTAIN Task
Force 2001). The hydrogen-reduced form is the most common food-grade iron source. It
is produced by reduction of ground iron oxide with hydrogen or carbon monoxide at high
pressures and temperatures. Purity is dependent on the purity of the oxide used. Most
often, reduced iron is a byproduct of other industries since using mill scale produces a
higher purity iron powder than iron ore and is less expensive (Patrick 1985). The physical
properties of reduced iron vary widely and can range from smooth dense spheres of
varying size to spiked or smooth, irregularly shaped, dense or porous particles. The
surface area of the particles play a major role in the solubility rate and amount of iron
dissolved during digestion.

Impurities in food-grade iron also affect solubility. The most common impurities
consist of a mixture of ferrous and ferric oxides that form on the surface of iron powders
or they can be trapped during manufacture in the center of the particles. The surface of all
iron powders contain varying amounts of a mixture of ferrous and ferric oxides. Ferrous
oxide is nutritionally beneﬁcial since it is more soluble in gastric juice and the oxidized
areas provide attack sited for particle dissolution. However, iron oxide can further react
to form insoluble ferric hydroxides (rust) if exposed to humidity, which are not
bioavailable and are detrimental to product quality. Reactive ferric and ferrous oxides can
behave unpredictably during food processing causing negative organoleptic and oxidative

instability problems in the ﬁnished product.

41

In general, elemental iron is less well absorbed than other more water-soluble iron
sources (like ferrous sulfate) but this depends heavily on how the iron powder was
produced and its resultant physiochemical properties and impurities (Hurrell and others
2002; Turner 2002). All commercial elemental iron powders must contain >96% iron.
The powder characteristics that inﬂuence absorption are those that inﬂuence dissolution
at acid pH in the gastric juice of the stomach. Physical properties such as particle size
and microstructure (shape/density/porosity) ultimately affect the available surface area
and rate of dissolution. Different manufacturing processes produce the different iron
types and the subsequent grinding and sieving processes also contribute to the ﬁnal
characteristics of the powders. As mentioned earlier, all food grade iron powders must
conform to the Food Chemical Codex (FCC) speciﬁcations that require the material to
pass a 100-mesh (149 um) sieve. Food companies in the United States and Europe mainly
use smaller particle size powders that are < 50 pm in an effort to reduce organoleptic
issues (black specks in the product and graininess) and improve bioavailability (Hurrell
and others 2002; Turner 2002).

As discussed, reduced iron is not a single entity but varies widely depending on
its method of manufacture. Historically, reduced iron powder has been classiﬁed into
three categories of iron based on their method of manufacture and resultant physical
properties. A fourth category, atomized iron, has become a recognized type of iron in the
last ten years (Table 1.3). The four categories are brieﬂy described (Hurrell and others
2002), as follows:

1) Hydrogen-reduced iron and carbon monoxide-reduced (CO-reduced) iron

powders are made at elevated temperatures and pressures by using either hydrogen or

42

carbon monoxide to reduce iron oxide to its elemental state. The iron oxide can be either
naturally-occurring iron ore, Fe203 (hematite) or mill scale, Fe3O4 (magnetite), which is a
by-product of the steel industry. These forms have the lowest purity of the food-grade
iron powders at > 96% iron. Impurities can be mixtures of carbon, magnesium,
aluminum, Silicon, phosphorous, sulfur, chromium, manganese, nickel and copper,
usually present in the form of acid insoluble oxides and therefore would be expected to
reduce solubility. The particles typically have an irregular but smooth exterior surface
and porous internal structure. Milling and sieving is done under inert gas to reduce
surface oxidation (the greatest source of impurities of any iron powder) and produce
powders with a range of particle sizes, typically less than 150 pm.

2) Electrolytic iron is produced by electrolytic migration of iron from a pure iron anode
through a ferrous sulfate solution onto a stainless steel cathode. The thin brittle sheets
produced by this process are ground into powders having ﬁne particle sizes. Electrolytic
iron powders contain Signiﬁcantly fewer impurities than reduced iron powders at > 99%
iron and are composed of irregularly shaped particles with high surface areas that are
considered to have good to excellent bioavailability.

3) Carbonyl iron is made directly from reduced elemental iron or from scrap iron. The
iron source is reacted with carbon monoxide under heat and pressure to produce iron
pentacarbonyl, which is then decomposed under controlled conditions to yield iron
powder and carbon monoxide gas. A second reduction with hydrogen is done to reduce
the carbon impurities to produce powders containing > 98% iron. Carbonyl iron consists
of dense spheres with extremely small particle sizes (2-10 pm) whose structure is

characterized by concentric shells of iron arranged in an onion-like fashion. The particles

43

have a smooth outer surface less prone to oxidation than reduced and electrolytic iron
types.

4) Atomized iron is a recently introduced hydrogen-reduced iron powder that is
manufactured by an “atomizing” process that gives the iron particles higher surface area
due to a highly irregular and “spiked” surface structure. This iron is available in a range
of particle sizes, depending on the milling process, similar to H-and CO-reduced iron.
Quebec Metal Powders developed this relatively new process that can be found in the
Handbook of Powder Metal Technologies and Applications (1998).

Hydrogen—reduced iron and electrolytic iron are provided as well-characterized
powders that should differ only slightly from batch to batch. Carbon-monoxide-reduced
iron powders can have variable composition and particle sizes ranging from the least
expensive < 150 um product commonly used in developing countries to the higher
quality < 45 -50 um used in the US and Europe. The greatest advantage of using
elemental iron powder over other iron fortiﬁcants is the relative stability of the material.
However, experience fortifying with reduced-iron powders at the Kellogg Company has
found that H-reduced iron powders having a surface area > 10 m2/ g often produce
product defects associated with oxidative deterioration that increases with increasing

surface area in sensitive products.

44

Table 1.3 Manufacturers and forms of food-grade elemental iron1

 

 

 

 

 

 

 

 

 

Powder Type Manufacturer Particle Size
North American
H-reduced Hdganas (formerly 325-mesh2
Pyron)
Quebec Metal Powders
Reduced (QMP), Canada
(atomized)3 International Metal 325-mesh
Powders (IMP), India
325-mesh
CO-reduced 300-mesh
(sponge)4 HOganas AB, Sweden 100-mesh
325-mesh
OMG Americas, USA (39% <10 um, 25% 10-20 pm, 16% <20-
Electrolytic (Glidden, A131) 30 pm, 8% 30-44u\pm)
Carbonyl BASF, Germany 325-mesh
International Specialty
Carbonyl Products (ISP), USA 325-mesh (mean particle size 5 pm)

 

IAdapted from Hurrell and others 2002

2325-mesh means that ca. 95% of the particles are < 45 microns; 300-mesh means that ca.
95% of the particles at < 50 microns; 100-mesh means that ca. 95% of the particles are <
150 microns.

3Sponge iron is smooth on the surface and has a porous sponge-like interior with high
surface area and an RBV typically greater than 50 (Kellogg, 2006)

4Atomized iron has an irregular highly spiked surface that gives the particles a

comparable surface area to sponge iron even though its center is not porous (Kellogg,
2006)

45

1. 7.2 Ferric Orthophosphate

Elemental iron powder and ferric orthophosphate (Figure 1.2) are the most
widely-used iron sources for the fortiﬁcation of ready-to-eat (RTE) cereals because of
their low cost and limited negative effects on product quality. Hydrogen—reduced iron
powder (<50 um) costs approximately one to two dollars per pound and is approximately
one-half to one-third the cost of ferric orthophosphate. Food- grade elemental iron
contains a minimum of 96% iron compared to the 26% iron content of ferric
orthophosphate. This makes the cost of ferric orthophosphate six to seven times more
expensive than elemental iron to attain the same level of iron fortiﬁcation (Kellogg
2006). Due to cost and usage level, hydrogen-reduced iron in the US. and both
hydrogen-reduced and carbon monoxide-reduced iron outside the US. continue to be the
most commonly used forms of iron. However, for speciﬁc applications, ferric
orthophosphate is preferred over reduced iron due to its lighter color and lower density
(2.87 FePO4 g/cm3 versus 7.8 g/cm3, respectively) (Budavari and others 1996). Ferric
orthophosphate is also preferred for oxidatively sensitive foods and can be used at higher
levels than reduced iron powders, although its effect on product quality can be mixed

depending on the processing involved.

/° Fe“ 1’
O\ ‘O—P—O
[So LI,
of
‘3‘ /

Figure 1.1 Ferric orthophosphate (FePO4-x H20) is an odorless, yellow to buff to pinkish-white
powder composed of one Fe (III) atom complexed to from one to three phosphate groups and
hydrated with one to four molecules of water. The iron content typically ranges from 26- 32%.
The monobasic, dihydrate is most often used for food fortiﬁcation. It is prepared by reaction of
sodium phosphate with ferric chloride or ferric citrate (Madison Chemicals 1996; IUPAC 2003).

46

As mentioned previously elemental iron powder exhibits a range of reactivity and
bioavailability depending on its physical properties, such as particle Size and
microstructure (shape/density/porosity). Bioavailability can vary signiﬁcantly with iron
powders (RBV = 2 to greater than 80%). The elemental iron powders with a mean
particle size < 50 um are generally thought to have approximately 1/2 the bioavailability of
ferrous sulfate (RBV = 100%) (Waddell 1974; Harrison and others 1976; Hallberg 1981;
Patrick 1985; Forbes and others 1989; Beard and Dawson 1997; Wienk and others 1999;
Nalubola and Nestel 2000; Dary and others 2002).

The chemical properties that determine the reactivity and bioavailability of iron
phosphates are even less well understood. Although the volume of literature on the
bioavailability of ferric phosphates is much less extensive, RBV in animal and human
studies also vary widely with literature values from 6 to 100%. Because less is known
about the factors inﬂuencing the bioavailability of ferric orthophosphate, the nutritional
quality is thought to be unpredictable and considered by many experts to be poor (Hurrell
1985; Forbes and others 1989; Hallberg and others 1989; Beard and Dawson 1997; Willis
and Allen 1999; Nalubola and Nestel 2000; SUSTAIN Task Force 2001; Dary and others
2002; Moretti D and others 2006).

Despite its uncertain nutritional quality, ferric orthophosphate is often the only
choice for beverages, light colored and oxidatively sensitive applications. The use of
ferric orthophosphate in RTE cereal applications has had mixed success (Kellogg 2006).
Certain RTE cereal applications require exposure to high temperatures and pressures

during prolonged cooking, tempering, drying and toasting processes. Product quality

47

 

problems include off-colors, off-ﬂavors and oxidative instability in ﬁnished products.

Unpredictable quality and bioavailability have limited the use of this fortiﬁcant.

1. 7.3 Bioavailability of Iron F arms used in RTE Cereal

Ferrous sulfate is the highly soluble iron form used as a relative standard in
methods to assess bioavailability of iron sources in both animals and humans (Forbes and
others 1989; Wienk and others 1999). Ferrous fumarate also is considered to have good
bioavailability as are the chelated iron forms, ferric EDTA and bisglycinate. The
reported bioavailability of ferric EDTA is 1 to 1.5 times that of ferrous sulfate but the
cost of this ingredient (35 times the cost of hydrogen reduced iron) has limited the use of
this fortiﬁcant. Ferric bisglycinate is another form of chelated iron that has excellent
bioavailability, however it is expensive and inherently less stable than elemental iron due
to the reactivity of its amino acid component in certain applications where browning
reactions are likely to occur (Hurrell 1997; Nalubola and Nestel 2000; Kellogg 2006).

Due to its manufacturing process, electrolytic iron is > 98% pure and is less
variable batch to batch than other elemental iron forms. When its mean particle size is <
20 um, electrolytic iron has been shown to have a RBV > 70%. However, it is
approximately three times more expensive than reduced iron and so less often used.
Ferric orthophosphate is referenced to have poor bioavailability according to Nalubola
and Nestel (2000), Forbes and others (1989) and Harrison (1976), averaging 30% the
bioavailability of ferrous sulfate. These authors recommend that three times the amount
of ferric orthophosphate Should be added to compensate for its lower bioavailability but

state that its bioavailability is less variable than some elemental forms of iron Typical

48

 

RBVS in animal and human studies for electrolytic iron and ferric orthophosphate

fortiﬁcants are shown in Table 1.4. Table 1.5 gives the RBVS for different particle size

fractions of elemental iron as summarized by Hurrell and others (20020).

Table 1.4 Literature percent Relative Bioavailabilityl Values (%RBV) for human,

rat and in vitro studies 2’3

 

Relative Bioavailability Values (RBV)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ferric
Model and Method Used Electrolytic Iron Orthophos shate
55 Fe Fe ssFePO4 FePO4
Human
Double isotope, extrinsic tag 0.75 0.25
M
Hemoglobin depletion-repletion
assay
AOAC 0.66 0.25
AOAC 0.77 0.33
HRE (35 mg Fe/kg diet) 0.78 0.58
HRE (20mg/kg diet) 0.80 0.86 0.60 0.61
.48 .26
Rat .69 .26
Double isotope, extrinsic tag .47 .22
In Vitro .75 .04
Solubility .60 .03
.75 .28
In Vitro .58 .27
Dialysis .71 .46

 

 

 

 

 

IRelative Bioavailability = test Fe sample/FeSO4
2 Data are given from four laboratories participating in collaborative study
3 Adapted from Forbes, and others 1989

49

 

 

Table 1.5 Inﬂuence of particle size on the bioavailability of elemental iron powders 1

 

 

 

 

 

 

Production Method Particle Size (pm) RBV

Electrolytic 1 7-10 64
27-40 38

Electrolytic 2 0-10 76
10-20 75

20-40 48

>40 45

H-reduced 10-20 54
>40 34

CO-reduced 6-10 36
14-19 21

27-40 13

Carbonyl <4 69
4-8 64

 

I Adapted from Hurrell and others (20020)

1.8 PHYSICOCHEMICAL PROPERTIES AND THEIR MEASUREMENT

The physical attributes of iron sources used in food fortiﬁcation, such particle,
porosity size and surface area, inﬂuence their solubility during digestion and ultimately
determine how well an iron source is available for absorption. There are many techniques
available to measure the physical properties of iron sources. However, careful
consideration needs to be given to choose the correct method for the material in question.
In the case of iron fortiﬁcants, methods need to be cost effective, relatively fast and

reproducible as well as accurate.

1.8.1 Particle size analysis

Particle size is a very important physical property that inﬂuences the solubility,

reactivity and organoleptic properties of elemental iron fortiﬁcants and is considered

50

important for iron phosphate powders as well (Oetker 1992). It is often used as a quality
criterion for various iron fortiﬁcants (Budenheim 1995; Wienk and others 1999;
SUSTAIN 2001; Hurrell and others 2002). Controlling particle size will ensure that
particles do not exceed the limit for desired appearance nor impart undesirable
organoleptic attributes (dark specs, grainy mouth feel and metallic taste). Iron is a
reactive substance and particle size is a major determinant of the surface area available
for chemical reactions, and in particular, dissolution in a solvent such as water or dilute
hydrochloric acid. Particle size measurements have been found to be potential predictors
of relative bioavailability for reduced iron powders (Forbes and others 1989; Swain and
others 2003). However, there is limited information in the literature about the affect of
particle size on the solubility of ferric orthophosphate powders (Hallberg 1981; Swain
and others 2003). Supplier information suggests that a similar relationship between
particle size and solubility exists (Budenheim 1995).

There are several techniques used to measure particle size distribution, with the
most common being micrOSCOpy, image analysis, sedimentation, sieving, electrozone
sensing (Coulter Counter), and laser diffraction. Several factors should be taken into
consideration when choosing a technique: the Size range of the test material (minimum
and maximum), precision required, sample size and number of particles measured,
analysis time, and the media of choice for optimum sample dispersion (air or liquid,
aqueous or organic, electrolytic). It is also important to consider whether the technique
requires standards for calibration and whether knowledge of other properties of the
material are required, such as refractive index and density. Each type of analysis

produces a single value that describes mean particle size using the theory of equivalent

51

spheres (Rawle 2003). Low angle laser light diffraction has the capability to measure the
needed particle size distribution range (0.1-100) for iron sources and can analyze iron
powders in the organic dispersants needed to prevent agglomeration. Obtaining a
representative sample is always challenging and laser light scattering allows the analysis
of the entire sample within minutes so many samples can be practically tested. The
measurement does not require standards and is not temperature (viscosity) dependent.

Light scattering measures the volume of a “sphere” of particles passing by a
series of detectors set at different angles using the volume moment mean calculation
(Malvem 2000; 2003). The volume mean is equivalent to the mass mean, which is also
equivalent to a weight mean if the density of the sample remains constant. The volume
property tells you where the mass of the distribution lies as opposed to a number mean
that gives you a distribution based on the number of particles. These two aspects of a size
distribution each give valuable information and light scattering allows data collected as
the volume mean to be transformed to a number mean (using the Hatch-Choate
equations) with an acceptable increase in experimental error. This has the advantage of
providing reliable volume and number particle size distributions with one measurement
(Malvem 2003). In addition to particle size, the particle shape and porosity 2 of a
material contribute to the amount of available surface area (Lowell and Shields 1991).
For this reason, surface area measurements may provide more comprehensive

information if particle structure is complex.

 

2 Porosity is deﬁned as the surface ﬂaws, which are deeper than they are wide Lowell S, Shields JE. 1991.
POWDER: Surface Area and Porosity. Third ed. New York: Wiley and Sons Inc. 250 p..

52

I. 8.2 Surface area analysis

Surface area estimates can be calculated from particle Size measurements.
However, these values will at best establish the lower surface area limit because the
fundamental assumption of equivalent spheres does not take into account the highly
irregular nature (shape and porosity) of real surfaces (Lowell and Shields 1991). Most
surfaces are not smooth and the irregularities can exist in any number of topographies,
such as hills, valleys and surface ﬁssures. Holes, or pores, in the surface of a particle
may be shallow indentations or extend deep into the particle giving it a sponge like
appearance that creates a high surface area due to the additional exposed surface.

Because of the limitations of particle size methods for surface area measurement,
gas sorption techniques are used. Gas sorption probes surface irregularities by enveloping
a powder sample in thin layer of physically adsorbed gas molecules, called the adsorbate.
This technique is based on the tendency of solid surfaces to physically attract gas
molecules (physisorption). Surface area measurements rely on kinetic theories that
predict the number of molecules required to cover the surface of a solid with a monolayer
of a gas. However, a complete monolayer surface is never actually formed due to
different potential energies at different sites on a solid.

In order to overcome this, the most widely used procedure for determining surface
area, the Brunauer, Emmett and Teller (B.E.T) theory, enables an experimental prediction
of the number of molecules required to form a theoretical monolayer. The B.E.T. theory
assumes that a dynamic equilibrium exists between the ﬁrst adsorbed layer and upper
vapor layers and so that the actual location of the surface sites covered with one, two or

more layers might vary but the number of molecules in a layer remains constant and can

53

be estimated (Lowell and Shields 1991; Quantachrome 1997). The BET theory requires
the use of the BET equation.4

A linear plot of the BET equation terms 1/W((P0/P)-l)) versus the relative
pressure P/Po, is required. The linear range of the adsorption isotherm for nitrogen is
restricted to a narrow range of relative pressure (0.05 — 0.30) for most solids. The
equations for the Slope and the y-intercept are combined and the estimated weight of a
monolayer of adsorbed nitrogen is determined by solving for Wm.S In the BET
equation, C is the BET constant and represents the magnitude of the interaction between
the adsorbate and the solid surface. Nitrogen is the most widely used adsorbate because
of its intermediate C constant values of 50-250 for most surfaces. This acceptable range
of BET constants for nitrogen makes it possible to calculate its cross-sectional area,
which is needed for the ﬁnal surface area calculation (Lowell and Shields 1991;

Quantachrome, 1997).

 

‘ BET equation = (1M ((Po/P)-1) =(1/me) +(C-1/w... C) (P/Po)
Where, at 77.3° K using nitrogen:

W = weight of nitrogen gas molecules adsorbed at relative pressure P/Po
Wm = weight of nitrogen gas molecules adsorbed in estimated monolayer
P = equilibrium vapor pressure, in torr

P0 = saturated vapor pressure, in torr

C = BET constant = (s/i) = 1

s = slope of the BET plot, 1/W[(Po/P) — 1)] versus (P/Po)

i = intercept of BET plot

5 Wm = [(P/Po) — 1][(1/C) + ((C-1)/C) (P/Po)] = 1/(s+I)

54

Multiplying the weight of the gas molecules, Wm, times the adsorbate cross-
sectional area times Avogadro’s number and then dividing this quantity by the molecular
weight of the adsorbed gas gives the total surface area of the sample.6 The speciﬁc
surface area is obtained by dividing this number by the weight of the sample. The BET
theory continues to be the most universally applied surface area measurement because of
its simplicity and accuracy (Lowell and Shields 1991).

Surface area can be determined using either a multipoint or a single point BET
method. One data point alone is sufﬁcient to calculate surface area and is called the single
point BET method. This simpliﬁed approach is rapid and less expensive and often results
in little loss of accuracy (Lowell and Shields 1991). Multipoint methods require a
minimum of three data points for surface area measurements or as many as 40 data points
for in depth analysis of adsorption and desorption isotherms. Isotherms provide additional

information on the shape of pores and the pore size distribution.

1.8.3 Microstructure and moisture absorption

The physical state, or phase, of a food substance is an important characteristic that
inﬂuences its behavior over time and under different conditions, such as during
processing. There are three basic physical states: solid, liquid and gaseous. Most food

materials exist in either a solid or liquid state and in almost any combination of the two.

 

6 Surface area = WmNAwM

Where:

Wm = weight of the adsorbed nitrogen gas molecules

N = Avogadro’s Number = 6.023X1023 molecules/mol

Ac, = cross-sectional area for nitrogen gas at 77.3° K = 16.2 A2
M = molecular weight of nitrogen

55

The physical state of a solid is extremely sensitive to the presence of water, temperature
and pressure. Chemically pure compounds in foods, such as water and many organic and
inorganic compounds have exact temperatures where they transition between phases,
called phase transition temperatures. Solids may exist in molecularly ordered structures
described as crystalline or disorganized structures referred to as amorphous, or a
combination of both. Certain substances crystallize as hydrates and the presence of
hydration water affects phase transition temperatures, for instance, the temperature at
which the crystals melt. Materials that are at equilibrium do not change their physical
state over time and remain stable at a given temperature, pressure and water content. A
change in conditions, such as a temperature or humidity, may introduce a driving force
(non-equilibrium) and a change of phase results (Bhadeshia 2002).

Microscopic techniques are extremely useful in studying the physical states of
food materials, especially crystalline morphology. The crystalline state is the most
ordered state of molecular arrangement (Roos 1995). Transmission light microscopy
using polarized light if often used to explore crystal structure. Crystals display a
phenomenon called birefringence, also called anisotropy, when they are exposed to two
different refractive indices of plane-polarized light from two directions. A sequence of
interference colors will appear depending on the thickness and morphology of the crystal
structure. The colors yield information about the degree of crystallization and type of
crystal present. Ferric phosphate salts are in a class of molecules termed “rigid
molecules” and represent a large group of substances consisting of aggregates held
together by strong molecular forces, 6. g. covalent, ionic, and polar covalent bonds. This

class is made up of metals, oxides, salts, ceramics, silicate glasses and also rigid solids

56

that have a strongly associated, ordered network, such as diamonds. In order to melt this
class of molecules (phase change from solid to liquid) strong bonds must break and this
requires sufﬁcient heat and pressure and/or the presence of hydration water. The melted
version of crystals form an unorganized association of aggregates termed amorphous.
Birefringence is lost as crystals become amorphous (Wunderlich 1990; Roos 1995;
Genck and Bayard 1997).

Amorphous content refers to regions of molecular disorder within a semi-
crystalline solid that is formed during crystallization or typically during subsequent
processing, like milling and spray-drying; it is seldom a deliberate or controlled event.
Process-induced amorphous content is thought to be mostly at the powder surface. A
small amount (by weight) of amorphous surface structure can increase the surface area
signiﬁcantly (Buckton and Darcy 1999). Even relatively low levels of amorphous
material (<10%) are known to inﬂuence stability and dissolution characteristics (Mackin
and others 2002). Amorphous regions in crystalline material will absorb water to a
greater degree than crystalline regions and are often thermodynamically unstable. These
areas of molecular disorder and higher moisture content are subject to physical transitions
and chemical degradations. If exposed to the changes in humidity and temperature,
amorphous material may undergo spontaneous glass transition to its more stable
crystalline state. The presence of amorphous ferric orthophosphate may have important
mechanistic properties that determine its bioavailability and stability in food products

(Buckton and Darcy 1999; Mackin and others 2002; Burnett and others 2006).

57

CHAPTER 2

COMPARISON OF THE RELATIVE BIOAVAILABILITY OF SIX TYPES OF
FERRIC ORTHOPHOSPHATE POWDERS IN READY-TO-EAT CEREAL

2.1 INTRODUCTION

Anemia remains the most prevalent micronutrient deﬁciency in the world and iron
deﬁciency anemia (IDA) accounts for a large percentage of the total problem.
Approximately 600 — 800 million people are affected worldwide with women, infants,
and young children the most at risk (INACG 2001; Turner 2002). In the majority of cases,
the cause of IDA is diet related, yet despite diet-intervention strategies implemented in
almost all countries, IDA remains endemic throughout the world and new strategies are
needed to address the problem (Stoltzfus 2001b).

There are many forms of iron used for fortiﬁcation; however, elemental iron,
ferrous sulfate, ferric orthophosphate, ferric pyrophosphate and ferrous fumarate are the
ﬁve forms used almost exclusively on a large scale. Elemental iron powders are the most
commonly used iron fortiﬁcant because they are affordable, cause the least problems with
food quality, and have good consumer acceptance. The different iron forms have different
degrees of reactivity, which generally impair the organoleptic properties and shelf life of
foods, thus limiting the iron forms used in foods to those that are less reactive.
Unfortunately, the physical properties that make an iron source less reactive also are
thought to make it less well absorbed. These challenges make the prevention of iron
deﬁciency through food fortiﬁcation difﬁcult (SUSTAIN 2001; Fairweather-Tait and

Teucher 2002; Hurrell and others 2002). Experts believe that food stability limitations

58

and the conﬂict or lack of literature information on bioavailability have hindered the
successful implementation of cereal enrichment programs worldwide (Hurrell and others
2002; Lynch 2002; Turner 2002). In the US, the Special Supplemental Nutrition
Program for Women, Infants and Children (WIC) is an initiative to safeguard the health
of at-risk populations. WIC-eligible cereal products must contain 45% of the RDI for iron
per serving or 8.1 mg (> 27 ppm) and delivery of this level while maintaining food
quality is a technological challenge.

Ferric orthophosphate is an affordable iron source that has had limited use as an
iron supplement in ready-to-eat cereal due its variable bioavailability. It is often preferred
for beverages (due to its low density), light-colored foods (e. g. rice) and oxidatively
sensitive food applications, such as RTE cereals for the WIC program. RBVS for ferric
orthophosphate reportedly range from as low as 2% to greater that 80%. Due to this
largely unexplained variability, nutritional experts consider ferric orthophosphate to be
poor source for iron fortiﬁcation (Shah and others 1977; Hurrell 1985; Forbes and others
1989; Hallberg and others 1989; Willis and Allen 1999; Nalubola and Nestel 2000;
SUSTAIN Task Force 2001; Dary and others 2002). Interestingly, the bioavailability of
elemental iron, which is the most widely used iron fortiﬁcant in food in the world, also
varies signiﬁcantly. RBVS for reduced iron range from less than 13% to 100% (Waddell
1974; Harrison and others 1976; Hallberg 1981; Patrick 1985; Forbes and others 1989;
Wienk and others 1999; Nalubola and Nestel 2000; Dary and others 2002; Swain and
others 2003; Beard 2006).

The ofﬁcial method of the Association of Ofﬁcial Analytical Chemists (AOAC)

for determining bioavailability is the rat hemoglobin repletion bioassay/slope ratio

59

technique. This method provides a relative measure of bioavailability, called the Relative
Biological ValueZCRBV). The physiochemical properties of elemental iron vary according
to its method of manufacture and evidence indicates that physiochemical properties affect
bioavailability (Swain and others 2003). However, few bioavailability studies on reduced
iron and even fewer on ferric orthophosphate have addressed the physiochemical basis
for the inherent variability of these materials. In 2002, an expert panel concluded that
incomplete and inconsistent information exists on the bioavailability and physiochemistry
of elemental iron powders used by the food industry (Swain and others 2003). Limited
evidence in the literature indicates that this is also true of ferric orthophosphate powders.
Better understanding of the mechanism underlying the variable bioavailability of the iron
sources would provide more consistent and potentially improved nutritional value for
consumers.

The objective of this study was to compare the bioavailabilities of a variety of
sources of commercially available food grade ferric orthophosphate powders with other
sources of iron commonly used in food products. The ferric orthophosphate powders
were chosen to cover a range of powder types, differing in color, particle size distribution
and solubility in dilute hydrochloric acid. The study included, as references, ferrous
sulfate which has an assigned RBV of 100, hydrogen-reduced elemental iron, and
encapsulated hydrogen-reduced elemental iron. The relative bioavailability of the iron
powders was determined using the AOAC Rat Hemoglobin Repletion Bioassay/slope

ratio method. A modiﬁcation of the AOAC method, the Rat Hemoglobin Regeneration

 

2The Relative Bioavailability Value is determined using animal and human test methods. The
most commonly used techniques are the AOAC Rat Hemoglobin Repletion Bioassay/ Slope Ratio
and Human Double Isotope/Extrinsic Tag methods. The bioavailability of the test iron source is
compared to ferrous sulfate, a highly bioavailable standard iron fortiﬁcant with an assigned RBV
of 100%.

60

Efﬁciency (HRE) Bioassay, was compared to the AOAC Ofﬁcial method to account for

diet consumption differences between animals.

2.2 MATERIALS AND METHODS

2. 2.1 Materials

The bioavailabilities of nine iron sources were evaluated in this study (Table 2.1):
ferrous sulfate (Source 1), encapsulated hydrogen-reduced iron powder (Source 2),
hydrogen-reduced iron powder (Source 3), and six different types of ferric
orthophosphate powder (Sources 4—9). Ferric orthophosphate (F ePO4-x H20) is an
odorless, yellow to buff to pinkish white powder composed of one Fe (III) atom
complexed to from one to three phosphate groups and hydrated with one to four
molecules of water. The iron content typically ranges from 26- 32%. All iron sources
were commercially produced, food grade materials meeting Food Chemical Codex (FCC)
requirements for ferrous sulfate, hydrogen-reduced iron and ferric orthophosphate.
Source 1 (ferrous sulfate (FeSO4 o 7HZO) was purchased from Mallinckrodt Baker, Inc.
Phillipsburg, NJ, and is the AOAC Rat Hemoglobin Repletion Bioassay reference
standard against which the bioavailabilities of the other sources were compared. Source
2 (encapsulated, hydrogen-reduced iron powder) was supplied by International Flavors
and Fragrances, Inc. (IFF) (New York, NY) and is used to fortify oxidatively sensitive
products. Source 3 (hydrogen-reduced iron powder) was from the same source of
reduced iron powder used to produce the encapsulated iron powder (e. g. Source 2). The

six types of ferric orthophosphate powder were > 99% pure and had iron contents ranging

61

from 24.6 — 30.4 % by analysis, depending on waters of hydration and phosphate content.
Sources 4 -7 were supplied by Budenheim Chemische Fabrik (Mainz, Germany). Source
8 was purchased from Madison Chemicals, Inc. (Madison Township, NJ). Source 9 was
obtained from Wright Enrichment, Inc. (Crawley, LA). Iron sources were stored in

closed containers under ambient conditions.

 

 

 

TABLE 2.1

Identity of test iron sources for the hemoglobin repletion diets

Diet Iron Source

Source 1 Ferrous sulfate, heptahydrate (Standard)

Source 2 Encapsulated, hydrogen-reduced iron, IFF

Source 3 Hydrogen-reduced iron powder, IFF

Source 4 Ferric orthophosphate, Budenheim Chemische 53-80
Source 5 Ferric orthophosphate, Budenheim Chemische 53-81
Source 6 Ferric orthophosphate, Budenheim Chemische 53-82
Source 7 Ferric orthophosphate, Budenheim Chemische 53-85
Source 8 Fenic orthophosphate, Madison Chemicals

Source 9 Ferric orthophosphate, Wright Enrichment

2.2.2 Methods

Bioavailability was measured using the AOAC Rat Hemoglobin Repletion
Bioassay (AOAC 1990). The animal study was conducted at Covance Laboratories Inc.

in Madison, Wisconsin.

2. 2.2.1 Depletion Diet
The composition of the low-iron animal diet supplied by Harlan Teklad (Madison,

Wisconsin) is given in Table 2.2. The animals also received ad-libitum a deionized water

source that was analyzed for iron content at the start and periodically throughout the

 

 

study.
TABLE 2.2
Composition of iron depletion diet 1

_I_ngredient Amount (g/kg)
Casein, high protein 200.0
DL-methionine 3.0
Corn starch 150.0
Sucrose 554.3
Corn oil 50.0
Mineral mix (TD 81062) 35.0
Vitamin mix, AIN-76A (TD 40077) 10.0
Choline bitartrate 2.0
Ethoxyquin (antioxidant) 0.01

 

1 Diet TD 0396 purchased from Harlan Teklad, Madison, WI

2. 2. 2.2 Repletion Diets

All iron sources were added to a ﬂaked, com-based RTE cereal product before the
pressurized cooking step at the point during processing when reduced iron powder is
typically added. The cereal was manufactured in approximately 90-kg batches (Kellogg
Company, Battle Creek, Michigan) and was used as the vehicle to deliver the iron source
to the test animals during the hemoglobin repletion study. The cereal was made from
whole corn grits, malted barley ﬂavor, high fructose corn syrup, sugar, salt and the
following added vitamins per 30 g cereal: A (500 IU), C (6 mg), D (40 IU), B-12 (1.5 pg),
B-l (0.4 mg), B-2 (0.4 mg), B-6 (0.5 mg), niacin (5 mg), and folate (0.1 mg). Nine cereal
batches were formulated to each contain one of the nine iron sources at 8 mg iron/30 g

cereal. A tenth batch of cereal was made with no added iron and was considered a low-

63

iron cereal due to the presence of approximately 4 mg of naturally occurring iron coming
from the malt ﬂavoring.

The iron repletion diets were formulated to contain 60% iron-free rat basal diet
and 40% cereal (Table 2.3). Iron-fortiﬁed cereal and the low-iron (no added. iron) cereal
were blended to contain 0, 6, 12, 24 or 48 mg iron per kg repletion diet. The iron content
of each resultant test diet was measured at Covance Laboratory using an inductively

coupled plasma (ICP) method (AOAC International 2000) (Table 2.4).

 

 

 

 

TABLE 2.3
Composition of iron repletion diet 1
Egredients Amount (glkg)
60% Iron Free Basal Rat Diet
Casein, high protein 333.3
DL-methionine 5.0
Sucrose 499.9
Corn oil 83.3
Mineral mix, iron deﬁcient as per depletion diet 58.3
Vitamin mix, as per depletion diet 16.7
Choline bitartrate 3.3
Ethoxyquin (antioxidant) 0.017
40% Test CM
0, 6, 12, 24 or 48 mg
supplemental iron/kg
Fortiﬁed cereal repletion diet
Unfortiﬁed cereal To equal 40% of the diet

 

miet was prepared in 1 kg batches to contain 60% (600 g) Iron-Free Diet and 40% (400 g)
mixture of fortiﬁed and unfortiﬁed cereal to equal target iron doses

64

TABLE 2.4 Iron Content of Diets, by Analysis ‘ (mg/kg Diet)
Targt Iron Dose Levels (mg)

 

 

 

 

 

 

 

 

 

Diet (Iron Source) 0 6 12 24
Diet 1 (Source 1, Standard)2 4 i 0.04 10 i1 16 i 2 25 i 1
Diet 2 4 1 1 17 30
Diet 3 4 1 1 15 27
Diet 4 4 1 1 17 27
Diet 5 4 1 l 16 28
Diet 6 4 11 16 29
Diet 7 4 10 16 28
Diet 8 4 10 14 25
Diet 9 4 10 16 27

 

1Cereal contains approximately 4 mg of naturally occurring iron from malted barley syrup.
2 Diet 1 had the following number of replicate analyses per dose level: 2 (dose 0), 4 (dose 6 and
12) or 5 (dose 24); Diets 2-9 were analyzed singly.

2. 2. 2.3 Animals

Male Hsd: Sprague Dawley® SD® rats were procured at 21-days of age from
Harlan Sprague Dawley, Inc., Madison, Wisconsin. They were single-housed in wire-
bottomed, stainless steel cages under a 12-hour light/12-hour dark cycle. Upon arrival,
the rats were immediately fed ad libitum the depletion diet for 15 days. Blood was
collected ﬁ'om the jugular vein at the end of the depletion period and analyzed for
hemoglobin concentration. The animals were then randomly assigned to test diet groups
at 36 days of age with 10 animals per group. The repletion diets were fed ad libitum
during a 2-week repletion period. At the end of the repletion period, the animals were
bled and the ﬁnal hemoglobin levels determined. The animals were fed their respective
diets from clear glass jars to allow easy inspection on the amount and condition of the
food. Fresh food was offered on day 1 and day 8. Individual animal body weight data
were recorded on days 1, 8 and 15 and individual food consumption data were recorded

weekly during the test period.

65

2.3 STATISTICAL ANALYSIS
2. 3.1 Hemoglobin Repletion Bioassay/Slope Ratio Modeling

Linearity of the regression curves was determined for each iron source separately
and a multiple regression model was used to determine the slopes of the nine iron sources
compared to a no-iron-added control diet. A model-based Residual Variance Method with
an AIC Fit Statistic was used to determine if the y-intercepts were simultaneously equal
(response at dose 0) to the reference Diet 9 and a contrast test was performed to see
which diets were signiﬁcantly different in terms of the change in hemoglobin repletion
values. The following equation (Forbes and others 1989) was used to calculate the

relative per cent biological values (RBV)3 of the diets:

[Slope of Test Diets (2-9)]
% RBV = X 100
Slope of Standard Diet

 

The equation used for the standard error was:

 

—1-\/Squ +RZSEj,
q

where q is the slope of the standard diet, and p is the slope of the test diet. The standard
errors of p and q are shown as SE, and SE.,. The relative bioavailability is R and R = p/q,

calculated using the SAS output.

 

3 Where,

Slope of Standard Diet 1 (Ferrous Sulfate) = 0.62

Slope Change Due to Test Diet Iron Treatments 2-9 = -0.35, -0.35, -0.01, -0.14, -0.30, -
0.10,-0.25,-0.19, respectively

66

2.3.2 Hemoglobin Regeneration E fficiency
The hemoglobin regeneration efﬁciency (HRE) ratio is determined by taking body
weight and food consumption into consideration in the following formula (Forbes and

others 1989):

[HGB-Fe (mg) — HGB-Fe (mg) . . . ]
HRE 2 ﬁnal Inrtlal X 100
Total Iron Consumed (mg)

 

Where,

HGB-Fe(n'g)=BodyWeightTestAnirral(kg)x[0'075B100da‘) x —i’1"él3'35Fe x MB ]
BodyWeight(l<g) HGB(g) Blood(L)

 

HGB = hemoglobin
HGB-Fe = hemoglobin iron
2.4 RESULTS

The Objective of this research was to compare bioavailabilities of different
sources of ferric orthophosphate that had different physiochemical properties to other
sources of iron fortiﬁcants (reduced iron and ferrous sulfate). A RTE ﬂaked-com cereal
was fortiﬁed with different iron sources and added to a standard rat diet in varying
amounts to achieve the desired fortiﬁcation levels. The differing amounts of cereal in the
rat diet and the food quality differences due to the physiochemical properties of the
different ferric orthophosphate types had the potential to inﬂuence animal acceptance of
the diet. The HRE model and AOAC standard repletion method were compared to

account for any differences in animal diet consumption.

67

Hemoglobin repletion study data (target iron dose, analyzed iron content, animal
weight, food intake, hemoglobin gain) and RBV results are summarized in Table 2.5.
Initial hemoglobin values averaged 5.0 i 0.51 g/dl. Hemoglobin Regeneration Efﬁciency
and RBV results are compared in Table 2.6. RBVS are graphed in Figure 2.1. The
unfortiﬁed cereal had approximately 4 mg endogenous iron from the malt ﬂavoring
(AOAC International 2000). Therefore, 4 mg of iron was subtracted from the analyzed
iron content of all diets. A linear lack—of—ﬁt test was performed on the Diets 1-9 (SAS
model-based, Fixed Effects Standard Error Method) comparing the iron content of the
diet versus the change in hemoglobin at the end of the repletion study. Signiﬁcant lack-
of-ﬁt occurred at p S 0.05 when the highest test dose (e. g. 48 mg iron/kg diet) was
included in the regression analysis. When the highest iron dose was excluded, linearity
was achieved. As a result, statistical analysis was performed on the data minus the
highest dose and the no-iron- added dose level was included to increase the statistical
power of the test.

After adjusting for endogenous iron, the SAS Residual Variance analysis of the
iron content versus the change in hemoglobin found the y-intercepts of the diets to be
simultaneously equal. RBVS were found to be signiﬁcantly different for all diets except
diets 2 and 3 (encapsulated and non-encapsulated reduced iron powders, respectively);
Diets 1 and 4 (standard ferrous sulfate and Budenheim ferric orthophosphate powder 53-
80, respectively); Diets 5 and 7 (Budenheim ferric orthophosphate powders 53-81 and
53-85, respectively). RBVs for the ferric orthophosphate powders ranged from 51 to 100.
Reduced iron and encapsulated reduced iron powders both had an RBV of 43, thus

indicating that encapsulation had little effect on bioavailability. Ferrous sulfate was

68

established as the standard with an RBV of 100; it was not statistically different from
ferric orthophosphate Source 4 (RBV 99).

Comparison of the RBV and HRE values (Table 2.6) indicated that the two
techniques gave Similar results; however, the RBV method was able to detect more
differences between the diets and there was more overlap in the HRE data. Diets 1, 4, 5
and 7 had the highest HRE values and were not statistically different from each other;
Diets 2, 3, 6 and 8 had the lowest HRE values and also were not statistically different

from each other.

69

Table 2.5. Hemoglobin repletion study data and Relative Bioavailability Value

(RBV) results

Targeted Dose3 (mglkg diet)

 

 

 

 

 

Dietl (Iron Source) RBV:l:SD4
and Animal Data2 0 6 12 24

Diet 1 (Source 1, standard) 100 a
Analyzed iron content 2 (mg/kg 4 i 0.04 10 i 1 16 i 2 25 i 1

Diet)

Animal Data

Body Weight Gain (g) 65 i 9 85 i 9 97 i 9 102 i 7
FoodIntake (mg) 1651L 11 191i15 2101-21 224i13
Hemoglobin gain (g/l) -21 i 5 -2 i 9 22 i ll 44 1r 11

Diet 2 (Source 2, encap red iron) 43:5 d
Analyzed iron content (mg/kg 4 l l 17 30

Diet)

Animal Data

Body Weight (g) 65 .+_ 9 86 i 8 95 i 5 103 i 7

Food Intake (mg) 172:10 174:15 193i 17 222i12
Hemoglobin gain (g/l) -21 i 5 -11 i 4 -2 i 6 15 i 6

Diet 3 (Source 3, red iron) 43:5 d
Analyzed iron content (mg/kg 4 l l 15 27

Diet)

Animal Data

Body Weight (g) 65 i 9 74 i 6 79 i 8 92 i 6

Food Intake (mg) 172:10 174i 15 193:17 222:12
Hemoglobin gain (g/l) —21: 5 -13 i 4 -6 i 5 11 i 5

Diet 4 (Source 4) 99:9 3
Analyzed iron content (mg/kg 4 l l 17 27

Diet)

Animal Data

Body Weight (g) 65 i 9 87 i 7 94 i 7 100 i 7

Food Intake (mg) 172 i 10 194 i 14 220 i 30 225 i 13
Hemoglobin gain (g/l) -21 i 5 -3 i 6 16 i- 6 53 :t 6

Diet 5 (Source 5) 7gi7 3"
Analyzed iron content (mg/kg 4 11 16 28

Diet)

Animal Data

Body Weight (g) 65 i 9 79 i 5 91: 6 97 i 6

Food Intake (mg) 172:10 197i 12 203i7 212:13
Hemoglobin gain (g/l) -21 i 5 -5 i 3 6 i 5 40 i 6

 

I Mean iSD of 10 animals per study group per Source tested

2 Diet 1 was analyzed in 2 (dose 0), 4 (dose 6 and 12) or 5 (dose 24) replicates; Diets 2-9 were

analyzed singly.

3 Cereal contains approximately 4 mg of naturally occurring iron from malted barley syrup.
4 Means with different letters are statistically different (95% CL).

70

Table 2.5 (continued)
Hemrﬂbin repletion study data and Relative Bioavailability Value (RBV) results

 

Targeted Dose3 (ngkg diet)

Diet1 (Iron Source)

 

 

 

 

and Animal Bang 0 6 12 24 %RBViSD‘
Diet 6 (Source 6) 51:5 Cd
Analyzed iron content2 (mg/kg 4 1 l 16 29

Diet)

Body Weight (g) 65 : 9 77 : 4 84 : 6 95 : 7

Food Intake (mg) 172: 10 184:8 193:6 211 :13

Hemoglobin gain (g/l) -21: 5 -10 : 4 -1 : 6 22 : 9

Diet 7 (Source 7) 83g ab
Analyzed iron content (mg/kg 4 10 16 28

Diet)

Body Weight (g) 65 : 9 82 : 6 93 : 5 101 : 7

Food Intake (mg) 172 : 10 182 : 7 207 : 12 223 : 18

Hemoglobin gain (g/l) -21: 5 -5 : 4 9 : 7 40 : 5

Diet 8 (Source 8) 60 : 6 her
Analyzed iron content (mg/kg 4 10 14 25

Diet)

Body Weight (g) 65 : 9 74: 6 81 : 9 93 : 9

Food Intake (mg) 172: 10 180: 15 186:9 202:8

Hemoglobin gain (g/l) -21: 5 -10 : 3 -3 : 5 20 : 6

Diet 9 (Source 9) 72 : 7 abc
Analyzed iron content (mg/kg 4 10 16 27

Diet)

Body Weight (g) 65 : 9 82 : 2 84 : 3 98 : 6

Food Intake (mg) 172:10 194: 10 199: 12 226: 13

Hemoglobin gain (g/l) -21: 5 -8 : 5 2.5 : 3 34 : 6

 

I Mean :SD of 10 animals per study group per Source tested

2 Diet 1 was analyzed in 2 (dose 0), 4 (dose 6 and 12) or 5 (dose 24) replicates; Diets 2-9 were
analyzed singly.

3 Cereal contains approximately 4 mg of naturally occurring iron from malted barley syrup.

4 Means with different letters are Statistically different (95% CL).

71

Table 2.6 Mean1 Hemoglobin Regeneration Efﬁciency (HRE) Ratios
compared to Mean Relative Bioavailability Values (RBV)2

 

 

Diet (Iron Source) % RBV : SD HRE : SD
Diet 1 (Source 1, standard) 100 a 100 a
Diet 2 (Source 2, 43 : 5 f 45 : 3 d:
encapsulated reduced iron)

Diet 3 (Source 3, reduced 43 : 5 f 35 : 3 e
iron)

Diet4 (Source 4, ferric 99 : 9 a 93 : 5 a
orthophosphate)

Diet 5 (Source 5, ferric 78 : 7 b 76 : 4 ab
orthophosphate)

Diet 6 (Source 6, ferric 51 e 5 e 48 : 8 be“
orthophosphate)

Diet 7 (Source 7, ferric 83 r 7 b 82 : 5 3"
orthophosphate)

Diet 8 (Source 8, ferric 60 : 6 d 55 : 4 °d
orthophosphate)

Diet 9 (Source 9, ferric 72 : 7 ° 65 : 4 bc
orthophosphate)

 

I Mean :SD of 10 animals per study group per Source tested
2 Means within a column with different letters are signiﬁcantly different (p<0.05)

72

Wm do: Mo 8958 E803“. 2:: maﬁa £8 E nose—no.2 EnﬁmoEo: mo owzommoromon ﬂu 933m

 

 

can 9:95 82: .c 28.30 :2.
on mw on my 0.. m o

 

Dc... 9 320
0:3 0 35
0:3 x. #90
mE3 0 ~20
0:3 m «20
o:3 v «£0
0:3 m «20
0:3 N «20
0:3 F «20
m ~05
w “0.0 I
n ~20
w «20
m “20
1 ~20
m «20
N no.0
w «20

 

 

 

 

 

 

(1M)89HUIUIBS

tom

XXO+

O

 

 

on

 

 

73

DISCUSSION
2. 5.1 Hemoglobin Regeneration E ﬂiciency versus Hemoglobin Repletion

The AOAC Rat Hemoglobin Repletion method was designed to compare the
bioavailabilities of different iron fortiﬁcants added to a standardized diet (Wienk and
others 1999). The shape of the hemoglobin repletion curve that shows that relationship
between hemoglobin levels as a function of dietary iron intake is sigrnoidal, becoming
ﬂat over time, making the length of the repletion period critical. Comparisons are made
on the linear portion of the curve and total dietary iron intake over the course of the study
drives the outcome of this measure. Despite equal iron contents in the diet, dietary intake
may differ among the groups if different amounts of the diet are consumed, contributing
to erroneous results. The HRE approach was developed to study intrinsic iron
bioavailability from different food sources that may be consumed preferentially (Forbes
and others 1989). The HRE method utilizes a correction factor for different iron intakes
by dividing whole-body hemoglobin gain by the amount of iron that is consumed.

The AOAC hemoglobin repletion method to determine RBV and the HRE
method results are in close agreement with similar standard errors. However, the
hemoglobin repletion/slope ratio method was able to detect more differences between the
diets. Overall, the hemoglobin repletion/slope ratio model is a less complex technique
with fewer measurements and appears to give accurate results less inﬂuenced by animal
and food intake measurement error. Variable food analysis data indicate that analytical
error may have been introduced by non homogeneity of the rat diets and that improved
sample blending and more replicate analyses were needed. Target levels of iron in the

diets were 0, 6, 12, 24 and 48 mg/iron per kg diet (excluding 4 mg/kg endogenous iron).

74

After grouping the analytical results for all iron sources by target level, standard
deviations were : 0.78 (n=12), 1.2 (n=12), 2.0 (n=13) and 7.4 (n=13), respectively. The
HRE takes into account animal growth and food consumption. However, the potential
error in these measurements can outweigh the beneﬁts of this added information.

In a similar study by Forbes and others (1989) for the International Nutritional
Anemia Consultative Group ((INACG), the standard AOAC hemoglobin repletion/slope
ratio method was compared with the HRE model for both ferric orthophosphate and
electrolytic elemental iron. Both techniques were compared to results obtained by double
isotope studies in humans. The AOAC Hemoglobin Repletion assay was run in two
laboratories. RBVS of electrolytic iron were 66% (: 2%) and 78% (: 3%) and ferric
orthophosphate were 25% (: 2%) and 34% (: 3%), respectively. The HRE assay
performed in one laboratory showed values of 78% (: 4%) and 58% (: 3%) for
electrolytic iron and ferric orthophosphate, respectively. When compared to ferrous
sulfate in a farina-based meal fed to humans, the double isotope, intrinsic tag method
found RBVS 75% for electrolytic iron and 25% for ferric orthophosphate. Forbes and
others (1989) concluded that the standard AOAC hemoglobin repletion method served as
a reliable predictor of iron bioavailability in man.

Forbes and others (1989) also looked at the effect of vitamin C on iron absorption
from the farina meals in both humans and rats using the double isotope method. Ascorbic
acid caused more than a threefold increase in the absorption of ferric orthophosphate.
Slightly less of an increase was seen in electrolytic iron and ferrous sulfate. The authors
hypothesized that fenic orthophosphate beneﬁted more from the presence of vitamin C

than the soluble ferrous sulfate and elemental iron because ascorbic acid, a reducing

75

agent, caused an increase in the Fe (11) concentration by reducing the Fe (III) content of
ferric orthophosphate. However, the relative absorption of ferrous sulfate to ferric
orthophosphate and electrolytic iron was the same. The authors concluded that the nature
of the meal had little inﬂuence on bioavailability measurements in humans and so the
theory of absorption of iron from a common non heme iron pool in the double-isotope
study was viable.

In contrast, in a recent double-isotope study by Moretti and others (2006) that
studied the effect of subject iron status, food matrices, processing and ascorbic acid on
RBV. An experimental micronized form of ferric pyrophosphate was compared to ferrous
sulfate in a wheat-milk infant cereal with and without ascorbic acid and in a processed
and unprocessed rice meal. Ascorbic acid (molar ratio of 4:1 to iron) increased iron
absorption more than twofold. As expected, iron status was a highly signiﬁcant predictor
of the RBV of fenic pyrophosphate (P < 0.001). RBV varied ﬁ'om 15 to 62% and the
authors concluded that assigning a single RBV to a poorly soluble iron source, like ferric
pyrophosphate, may be of limited value in evaluating their suitability for food
fortiﬁcation (Harrison and others 1976; Swain and others 2003; Moretti D and others

2006).

2. 5.2 RB Vof Iron Treatments

The RBV of the iron sources used in this study were determined on reduced iron
and ferric orthophosphate powders that were added to RTE cereal grains prior to
processing. During manufacture, the iron sources were exposed to high pressure cooking

at temperatures greater than 240 °F for longer than 60 minutes and then exposed to

76

forming, drying and toasting processes. The RTE cereal was fortiﬁed with vitamins and
minerals including 10% DV of vitamin C, which meant the repletion diets used in this
dissertation, including the standard ferrous sulfate diet, each contained 80 mg vitamin C
per kg animal feed (Materials, Repletion Diets 2.2.2.2).

The powders were chosen based on visual differences, such as varying color and
ﬂow properties, which indicate physical property differences that may inﬂuence how they
behave during food processing. Properties such as surface oxidation, particle size and
surface area are known to inﬂuence the solubility and reactivity of elemental iron,
causing oxidative instability in foods that affect taste, texture, appearance and shelf life.
Factors that inﬂuence the behavior of fenic orthophosphate during processing include
particle size, surface area and crystal structure. The same physiochemical properties that
inﬂuence food quality during manufacture and over shelf life also inﬂuence the
physiochemistry of the iron sources during digestion, which ultimately inﬂuences
bioavailability. Results of this study Show that all the ferric orthophosphate powders
tested had bioavailabilities equal to or better than a typical reduced iron powder
commonly used to fortify foods worldwide.

A potential mechanism explaining the physiochemical variability of ferric
orthophosphate powders may be the presence of amorphous material, which occurs to
varying degrees as a result of differing manufacturing practices. Amorphous materials
are known to have higher solubility and dissolution rates and decreased chemical and
physical stability as compared to their more crystalline counterparts (Williams, 2003).
The presence of amorphous material may result in increased reactivity during cereal

processing. Soluble iron is a prooxidant that catalyzes lipid oxidation and other

77

deleterious oxidative reactions such as the fading of natural colors and degradation of
oxidatively sensitive vitamins. Conversely, it may also favorably inﬂuence
bioavailability. Differences in the physiochemical properties of the ferric orthophosphate
powders used in this study, such as particle size, crystal structure and amorphous content,
are thought to have inﬂuenced the bioavailability outcome by altering solubility.
Therefore, the next steps in this research project were to characterize the physiochemical
properties of the powders and study their effect on the solubility of ferric orthophosphate
in dilute hydrochloric acid solutions to better deﬁne the factors inﬂuencing

bioavailability.

78

CHAPTER 3

PHYSICOCHEMICAL PROPERTIES OF FERRIC ORTHOPHOSPHATE
INFLUENCING ITS SOLUBILITY IN DILUTE HCl

Particle Size, Surface Area and Amorphous Content

3.1 INTRODUCTION

Bioavailability of iron can be thought of as consisting of three components: ﬁrst is
the solubility of the iron source in the gut; second is how well the iron is absorbed and
delivered to the circulation; and third is how well the iron is processed by the body once
it reaches the circulation, target organs and cells for incorporation into measurable
functional entities. Bioavailability must be quantifiable and the working deﬁnition of iron
bioavailability is often relative as determined by the methods available to measure one of
these three stages. Endpoint measures currently provide the best estimate of
bioavailability when all three aspects of bioavailability, e. g. solubility, absorption and
utilization, need to be considered. However, in vitro methods that give a good estimate of
the bioavailability obtained from endpoint measures and that are economical and avoid
the use of animals and people are desirable. Non-heme iron absorption depends on the
solubilization of ferrous, fenic iron, or elemental from the acid milieu of the stomach.
depending on the form of iron in the food (IOM 2001). Ferric iron must be reduced to its
ferrous form by reducing agents such as ascorbic acid or ferrireductase enzyme present
on the mucosal cell surface of the upper small intestine. This bioavailable iron is then
absorbed in a 3-step process by the enterocytes, transported intracellularly and transferred
into the plasma. It is well accepted that only soluble ferrous iron can be absorbed during

digestion, and once solubilized and absorbed, ferrous iron joins a common intracellular

79

“iron pool” for use by the body (Miret and others 2003). Therefore, in vitro methods that
measure the solubility of fortiﬁcants are promising for use as estimates of bioavailability
and to gain an understanding of the properties that inﬂuence solubility.

Several researchers have demonstrated that the RBV of ferric orthophosphate is
highly variable (Harrison and others 1976; Swain and others 2003; Moretti D and others
2006). The same is true of reduced iron, which is the most commonly used iron
fortiﬁcant worldwide (Waddell 1974; Harrison and others 1976; Hallberg 1981; Patrick
1985; Forbes and others 1989; Beard and Dawson 1997; Wienk and others 1999;
Nalubola and Nestel 2000; Dary and others 2002). The SUSTAIN group commissioned a
review of the literature published over the past 45 years and concluded that little
information was available on the precise types of iron used in the literature and that not
enough is known about the extent to which different elemental iron powders (and other
iron fortiﬁcants with low solubility) are absorbed by the body (Hurrell 2002a; Turner
2002)

The physical attributes of mineral sources used in food fortiﬁcation, such as
particle size and surface area, inﬂuence their solubility during digestion. Tests to measure
solubility can be simple dissolution studies performed at various concentrations of
hydrochloric acid, or more complex methods that simulate the digestive environment
(Harrison and others 1976; Shah and others 1977; Forbes and others 1989; Swain and
others 2003; Wortley and others 2005). Using the same procedure for a single type of
iron, a given method will yield relative results that can be used to compare the solubility
of different forms of the same material. However, even within the same laboratory,

results have been variable (Forbes and others 1989; Swain and others 2003; Kellogg

80

2006; Moretti D and others 2006). Standardized methods to measure solubility and other
physicochemical properties such as particle size and surface area for intra-laboratory
collaborative studies are needed. For the affordable iron sources, most of the research on
the physical properties inﬂuencing bioavailability has been done on elemental iron with
much less work done to investigate the factors affecting the bioavailability of ferric salts,
like ferric orthophosphate. It is important that the factors inﬂuencing the bioavailability
of the widely used, affordable iron fortiﬁcants, like reduced iron and ferric
orthophosphate, are well elucidated. In a recent study by Moretti and others (2006), the
relative bioavailability of a micronized, dispersible fenic pyrophosphate varied markedly
(15 — 62 % RBV) with the food matrix and iron status of the human test subjects. The
authors concluded that assigning a single RBV to poorly soluble iron compounds may be
of limited value in evaluating their suitability for food fortiﬁcation (Moretti D and others
2006)

Two physical parameters, particle size and surface area, are thought to be
important factors in the solubility of iron fortiﬁcants (Harrison and others 1976; Hurrell
2002b; Swain and others 2003) Therefore, one objective of this study was to measure
these attributes and their relationship to the solubility. A third physical parameter,
amorphous content, is known to play an important role in the bioavailability of ferric
orthophosphate in plants and insects, and is hypothesized to be an important factor for
efﬁcacy in humans (Willis and Allen 1999). Amorphous content refers to regions of
molecular disorder within a semi-crystalline solid. Amorphous regions in crystalline
material will absorb water to a greater degree than crystalline regions and are often

thermodynamically unstable. These areas of molecular disorder and higher moisture

81

content are subject to physical transitions and chemical degradations. The presence of
amorphous ferric orthophosphate may have important mechanistic properties that
determine its bioavailability and stability in food products (Buckton and Darcy 1999;
Mackin and others 2002; Burnett and others 2006).

To date, little information is available on the amorphous content of ferric
orthophosphate powders and its relationship to bioavailability in man. Quantiﬁcation of
amorphous content is analytically difﬁcult, especially for contents less than 10%. Over
the past ten years, several techniques have been developed that have improved sensitivity
to measure low amounts of amorphous material. Bulk analytical techniques, such as
differential seaming calorimetry, and powder X-ray diffraction, which measure a
properties of the entire sample, are less sensitive than techniques that preferentially
measure properties of the amorphous content only. In bulk measurements, the amorphous
content becomes a small part of the total signal, and consequently it is difﬁcult to
quantify with conﬁdence (Buckton and Darcy 1999). The detection limits for amorphous
content with bulk techniques will generally have a lower cut off of 5—10%. In addition,
techniques which preferentially investigate the properties of the powder surface, where
amorphous material may predominate, are more sensitive. A powerful way of
investigating surface phenomena is by dynamic (gravimetric) vapor sorption (DVS),
which preferentially probes amorphous (over crystalline) regions. DVS is the preferred
means in the pharmaceutical industry for measuring low levels of amorphous material
(Buckton and Darcy 1999). This technique measures the mass change due to preferential
solute (water) uptake by an amorphous phase under controlled conditions (Burnett and

others 2006). A small amount of amorphous material on the surface of a particle (by

82

weight of the sample) can have a very substantial surface effect and, in the case of iron,
cause solution behavior that is very different from material in the crystalline state
(Buckton and Darcy 1999). These differences in amorphous content behavior can
signiﬁcantly alter the solubility of the iron source.

In the previous chapter, substantial differences in the bioavailability of different
sources of ferric orthophosphate were observed. Solubility is known to play an important
role in bioavailability. Iron sources with low water-solubility must ﬁrst be solubilized in
digestive juices before they can enter the common iron pool for absorption. This chapter
investigates the in vitro solubility of the ﬁve of the sources of fenic orthophosphate
studied in Chapter 2. The method of Shah and others (1977) was adapted to evaluate the
solubility of ferric orthophosphate over the range of hydrochloric acid concentrations
found during normal human digestion (0.02 — 0.10 N HCl). Although this method was
originally developed for elemental iron powders (Shah and others 1977; Hallberg and
others 1989; Swain and others 2003; Hoppe and others 2006), it has been applied to
various non-heme iron sources, like ferric phosphates (Forbes and others 1989; Hallberg
and others 1989).

There is wide variability in the data in the literature on the physicochemical
properties and RBV for ferric orthophosphate, and standardized methods are needed to
allow comparable results between laboratories. Thus, the inﬂuence of particle size,
surface area and amorphous content on the solubility of the ferric orthophosphate samples
was investigated in this study. Due to the potentially important effects of the presence of
amorphous material on moisture uptake and, therefore, solubility and reactivity during

cereal processing and over shelf life, a sensitive technique to measure amorphous content

83

was needed. RTE cereals area dry products and moisture uptake is their main mode of
shelf-life failure along with oxidative rancidity. DVS was chosen as a sensitive measure
of amorphous content and also because it directly measured the effects of exposure to

moisture uptake at different relative humidity.

84

3.2 MATERIALS AND METHODS

3.2.1 Materials

Solubility, particle size, surface area and isotherm measurements were made on
the same lots of ferric orthophosphate powder that were used in Chapter 2, identiﬁed as
Sources 4 — 8 (Table 3.1). New lots of these materials were also analyzed for Sources 4,
5 and 6 to determine lot-to-lot process variation. Different production lots were not
available for Source 7 and 8. Iron sources were commercially produced, food grade
materials meeting Food Chemical Codex (FCC) requirements for fenic orthophosphate.
The ferric orthophosphate powders (fenic (III) —orthophosphate (FePO4 X H20, mono —
tribasic, molecular weight 150.82, anhydrous) were > 99% pure and had iron contents
ranging from 24.6 — 30.4 % by ICP analysis, depending on waters of hydration and
phosphate content. Sources 4 -7 were supplied by Budenheim Chemische Fabrik (Mainz,
Germany). Source 8 was purchased from Madison Chemicals, Inc. (Madison Township,

NJ). Iron sources were stored in closed containers under ambient conditions.

 

TABLE 3.1

Iron sources used for particle size and surface area measurements
Sample

Identification # Lots Iron Source

 

Source 4 2 Ferric orthophosphate, Budenheim Chemische 53-80
Source 5 2 Ferric orthophosphate, Budenheim Chemische 53-81
Source 6 3 Ferric orthophosphate, Budenheim Chemische 53-82
Source 7 l Ferric orthophosphate, Budenheim Chemische 53-85
Source 8 1 Ferric orthophosphate, Madison Chemicals

 

85

3.2.2 Methods
3. 2.2.1 Solubility in Dilute Hydrochloric Acid

The solubility method used for the iron sources was based on the method of Shah
and others (1977). The method was modiﬁed to decrease assay variation by performing
experiments in the dissolution apparatus described in the USPC 2006 Ofﬁcial Method for
Dietary Supplements #2040 Disintegration and Dissolution of Dietary Vitamin and
Mineral Supplements (USPC 2006). The apparatus consists of a 1000-m1 low-forrn
beaker on a basket-rack assembly, which contains the dissolution solution and a
thermostat to control heating of the solutions between 35° and 39°C. Sample mixing is
precisely controlled by a device that vertically rotates the beaker a set distance (not less
than 5.3 cm and not more than 5.7 cm) with controlled frequency rates (rpm).

Sources 4-8 were weighed to contain approximately 2.5 mg iron/ml acid solution,
e. g. approximately 1.123 g ferric orthophosphate powder was dissolved in 450 ml of
standardized HCl (Fisher Scientiﬁc Catalog #SA54-4). Three acid concentrations, 0.02 N,
0.05 N and 0.10 N were used in the experiments. The acid solutions were maintained at a
temperature of 37°C : 1° and mixed in the dissolution apparatus at 200 rpm for 45
minutes. After dissolution, a 20-ml aliquot was removed and sequentially ﬁltered through
a Whatman #40 ﬁlter and a Whatman #42 ﬁlter. One ml of the ﬁltrate was transferred to
a 25-ml volumetric ﬂask and diluted to volume with more dissolution solution. The
amount of dissolved iron was determined by inductively coupled plasma (ICP) at
Covance Laboratory in Madison, WI. (AOAC International 2000). Ferric chloride (1 mg
iron/ml) was dissolved in the same concentration acid solution as the sample and used as

the standard stock solution. Standard concentrations were prepared to cover the range of

86

0.2 to 10 pg iron/ml and a standard curve was used to quantitate the amount of iron in the

samples.

3.2.2.2 Polarized Light Micrographs

Ferric orthophosphate Sources 4-8 were examined by polarized light microscopy at
a magniﬁcation of 360X. Photomicrographs were taken under optical conditions of
crossed polarizing ﬁlters and ﬁrst order red compensation that resulted in a light red
background. The polarization phenomena of the crystals were displayed for the two
principal vibration directions, using an optical shift of 0.576 um. The crystals showed a
sequence of interference colors depending on their thickness, birefringence, and
orientation relative to the direction of the polarizing ﬁlters and the compensation plate

(Genck and Bayard 1997)

87

3. 2. 2.3 Particle Size by Laser Light Diffraction

The particle size distributions of the ferric orthophosphate samples were
determined by Particle Technology Labs, Ltd. Method MM324.01 using a Malvem
Mastersizer 2000 Laser Diffractor. The refractive indices (R1) of the iron sources were
determined by Bayard Development Company using the Beckeline Test4. Results are
given in the Appendix A. An average RI was used for samples having a range of
refractive index values. Samples were prepared for analysis by placing an amount from
the bulk sample into a scintillation vial to reach an obscuration of 10-20%, or
approximately 0.003 — 0.020% volume depending on the sample. Approximately 0.5 cm
of Aerosol OT (Fisher Scientiﬁc) was added to the sample. The scintillation vial was
ﬁlled with 20 m1 of hexane (Fisher Optima Grade), and was placed in an ultrasonic bath
(Branson 3) for 2 minutes. Samples were checked for clarity and visually inspected using
a microscope to verify adequate dispersion. The suspension was transferred by pipette
into the Malvem Hydro 200S recirculator bath (ﬁlled with Fisher Optima grads hexane)
using an amount to reach an obscuration level in the range of 10 to 20 percent. Four 12-
second analyses were made at one-minute intervals and results are reported as the
average of the four analyses. The analysis was repeated Six times for each lot of ferric
orthophosphate. Results are reported on a volume (mass) basis. Instrument and software

parameters are given in Appendix I, Table 2.

 

4 The Beckeline test measured the refractive indices for the sodium d line (589 nm) of the ferric
orthophosphate samples based on standard immersion methods.

88

The following six values, termed the mean volume statistics, were measured:
mean particle size (MPS) diameter in micrometers; three cumulative percent statisticss,
10%, 50% and 90%, which are the percentages of the total sample smaller than the
indicated sizes in micrometers; uniformity statistic, which is an indicator of the normalcy

of the distribution and the span, a value representative of the width of the distribution.

3.2.2.4 Surface Area by Nitrogen Adsorption

Surface area measurements were made by Particle Technology, Ltd. Method
AU225.01 using a 3-point Brunauer, Emmett and Teller (BET) surface area analysis
performed on a Quantachrome Autosorb-l instrument. Approximately 2-3 grams of
sample was prepared for analysis by outgassing under a helium purge at 25°C for 16
hours until the samples were thoroughly dry and free of surface contamination. Analysis
was carried out using nitrogen as the adsorbate gas maintained at a temperature of 77.4
°K over a range of ﬁve B.E.T. points (0.10, 0.15, 0.20, 0.25, and 0.30). Instrument
parameters are given in Appendix B, Table 3. The analysis was repeated six times for

each lot of ferric orthophosphate.

 

5 The volume statistics (D[v,0.10], D[v,0.50], D[v,0.90]) are based on the volume moment mean:
D[4,3] = Sum of the equivalent diameters4/sum of the equivalent spherical diameters3. D =
spherical diameter and [4,3] refer to the power of the sum of the diameters for the numerator and

denominator, respectively.

89

3.2.2.5 Moisture Uptake by Dynamic Vapor Sorption

Dynamic Vapor Sorption (DVS) analyses on ferric orthophosphate powders were
performed in the laboratory of Dr. Daryl Willams in the Department of Chemical
Engineering, Imperial College, London, UK as part of a Kellogg Company-funded
research project (Heng and Williams 2006). The method developed by Heng and
Williams (2006) used the Dynamic Vapor Sorption DVS-l (Surface Measurement
Systems, UK) equipped with a Cahn D-100 microbalance (sensitivity 0.1 pg).
Sorption/desorption analyses were conducted on the same ferric orthophosphate powders
(Source 4-8) used throughout this dissertation. Approximately 100 g of fortiﬁcant was
weighed in the microbalance module and pre-conditioned at 0% relative humidity (RH)
for 300 minutes. Following preconditioning, the relative humidity (RH) was
incrementally increased on average every 60 minutes as follows: 0, 10, 20, 30, 40, 50, 60,
70, 80 and 90% RH and then reversed to complete one sorption/ desorption isotherm
cycle over approximately 2000 minutes. Two cycles were performed on each of the
sources to determine if the samples underwent crystallization changes during analysis and

to determine optimum cycle-time.

3.2.2.6 Preparation of Amorphous Ferric Orthophosphate

A 100% amorphous ferric orthophosphate powder was not readily available from
a commercial source. Therefore, an attempt was made to prepare a 100% amorphous
fenic orthophosphate standard from Source 4, which appeared to be the source with the
highest amorphous content based on solubility and microscopic analysis. In order to

make as close to a 100% amorphous sample as possible, Source 4 was dissolved in dilute

90

aqueous 0.10 N HCl. This solution was spray dried using a Buchi B-290 apparatus
(Buchi, Oldham, UK) and collected by a cyclone system (Heng and Williams 2006) and
analyzed immediately after collection using the same experimental conditions and

parameters.

3.3 STATISTICAL ANALYSIS

Solubility analyses were conducted in triplicate on Sources 4-8 while six
replicates were performed for particle size and surface area measurements. Additional
lots of material (Table 3.1) were not available for all of the ferric orthophosphate sources
resulting in an unbalanced data set and reduced ability to detect differences between lots.
A least squares analysis of variance was performed using the Proc Mixed Procedure in
SAS. Physical properties, such as mean particle size, were considered a function of the
source of the iron and therefore ﬁxed effects while lot was considered a random effect.
Multinomial decomposition of the particle size data was done using the nonlinear curve
ﬁt algorithm in Peak Fit software from Systat Software, Inc. The determination of
moisture uptake (amorphous content) by DVS measurements was semi-quantitative due
to the instability and hydroscopicity of the amorphous standard during analysis.
Regression analysis was used to correlate median particle size, surface area and moisture
uptake versus percent solubility and percent RBV. Percent RBV was correlated to
solubility and log 10 solubility and a multiple regression model was used to evaluate

median particle size and moisture uptake as predictors of solubility and % RBV.

91

3.4 RESULTS
3. 4.1 Solubility in Dilute H Cl

There was a marked difference in the solubilities of the iron sources (Table 3.2).
Iron solubility was calculated as the amount (pg) iron that dissolved per gram of iron
analyzed. The ferric orthophosphate samples had an average iron content of 27.8% (26.4-
29.3% : 1.09%). Approximately 0.29 mg total iron (1.12 g of ferric orthophosphate) was
analyzed per sample. Each source was dissolved at three dilute hydrochloric acid
concentrations, 0.02 N, 0.05 N and 0.10 N. The range of dissolved iron in the three HCl
solutions was as follows: 101 pg to 4.1 mg in 0.02 N HCl; 103 pg to 24.7 mg in 0.05 N
HC1; and 231 pg to 195 mg in 0.1N HCl. Expressed as a percentage of iron dissolved
(weight basis), the rank solubility of the iron from the most to the least soluble was
Source 4 (19.5%) > Source 7 (1.0%) > (Source 5 (0.29%) > Sources 6 and 8 (<0.05%).

This trend was seen at all dilute acid concentrations tested.

92

29.26%

2953. 0:" a:

 

 

 

 

 

 

 

 

ﬁnca
$3 $3 $3.: Dame... 8.8:
Ba: 4.22: ESE. :82
Been
No no: 2N4 am Econ
35 one: a. _ we: 85:9 Econ
.88 $8... o u .. .2 $ 8... 2 u... a: $8... 3 a :N are _ .3
a 3.:8m
.38 $2... 3. a RN; $N§ E a 83. $8: w: u 83: sue a :3
h Quascm
.33 $8... N a NN_ $8... 8 u. NON $8... me n. omm not. =<
$8... N .i. No. $ 8... m 44 a: $8... 2. .l. eNm ANue m .3
$8... m n 3 $8... _ a mNN $3... we a 8.... Te N s3
$8... a n § $8... ON a 3N $8... 2 u NoN are L :3
e ooh—cw
and: $3... SN a N8 $2... 3. a BS $3... SN a 8% ﬁne _ an
m ooh—om
anon $:..e 3a: a $3. $5: ONE. 4 SEN $3.2 3: a 25.5 a3 =<
$8.: 3 A New $23. 8N a 24.9. $3.: 82 a Sea: are N :3
$3.... o .1. .53? $2... o 44 .84.; $2.: 0 a SS _N ANteV _ .3
v eon—Sm
cm H >5. 828.5 cm a g .5233: am a awe 828.5 :m .4 awe .3 E...
5:333235 :9: .x. .8..— cozomma :9: e\e E:— quommE :9: e\e eon—1028mm: «3.8m
cacao: 6: 2 Na... 6: z 8... .0: z 3

 

gaunt—.850 29¢.

 

m>m¢ S ESQ—:8 3 233.3. man 3.5—:2...ch 33:. E 8958 :9: .3 85.525 ﬁn 03:.

93

3. 4.2 Particle Size, Surface Area and Polarized Light Microscopy
3. 4. 2. 1 Particle Size Analysis

The results of the laser diffraction measurements for particle size are
shown in Table 3.3 along with the % RBVS for each source (Chapter 2, Table 2.4)
and surface area results. A conﬁdence level of 95% was used. The differences
between lots were found to be signiﬁcant and random. Based on the PS
distribution data, the sources were grouped into three statistically different
distributions (P < 0.05).

The ﬁrst group, Sources 6 and 8, had larger particle size distributions
(Median PS 17.7 and 15.5 pm, respectively) and the lowest bioavailability (RBV
51 and 60%, respectively). Source 8 had the most uniform distribution and was
the only source that lacked a population of ﬁne particles. Polarized light
microscopy showed Source 8 to have a uniform population of bireﬁingent
particles, indicting a much higher degree of molecular order (crystal structure).

The second group, Sources 5 and 7, had bimodal distributions of ﬁne
particles. Both Source 5 and 7 had median PS of 2 pm and intermediate RBVS
(78 & 83%, respectively). Microscopic examination revealed a mixture of small,
amorphous and crystalline particles with varying degrees of crystallinity.

Source 4 was distinctly different from the other two PS distribution groups
and comprised the third group. This source appeared to be the most amorphous
of all the sources and had a bimodal distribution with an overall median PS of 9.3
pm and an RBV of 99%. Source 4 behaved differently in dilute acid solutions,

with 100 to lOOO-times the solubility of the other sources (depending on pH).

94

Two other particle size statistics, span and uniformity are given in Table
3.3. The Span parameter showed that Sources 4, 5 and 7 had the broadest particle
size distribution while Sources 6 and 8 had the narrowest distribution. The
uniformity parameter showed a similar trend with Sources 6 and 8 displaying a

single, more Gaussian-shaped distribution than the other sources.

3.4.2.1.1 Multinomial Curve Fitting of Particle Size Data

Peak ﬁtting software was used to estimate the presence of normally-
distributed particle size populations within the distributions for each iron source.
The MPS and the percentage of particles making up each normally-distributed
subpopulation were determined for each lot of Source 4, 5, 6, 7 and 8. Results are
Shown in Table 3.4. Sources 4, 5 and 7 had bimodal distributions with the
following MPS and percent composition for each subpopulation (percent range
over lots): Source 4, MPS 6pm (33-34%) and 51pm (66-70%). %). Source 5,
MPS 2 (IS-43%) and 18pm (57-82%); and Source 7, MPS 2 (29%) and 17pm

(72%). Source 6 had a trimodal distribution (MPS 0.8 pm (0.1%), 3pm (0.3-1 .8%)

and 21 pm (98-100%). Source 8 had a single, uniform distribution with a mean

particle Size of 17pm (100%).

3. 4. 2.2 Surface Area Analysis
Surface area measurements separated the ﬁve sources into two instead of

three signiﬁcantly different groups, which were not different within a group. The

95

surface area for Sources 4, 5 and 7 were approximately 10-times higher than the

surface areas for Sources 6 and 8 (Table 3.3).

3. 4.2.3 Polarized Light Microscopy

Microscopic examination visualized the presence of different particle size
populations evident from particle size distribution analysis (Shown in Figure 3.1a-
c with the particle size distributions). Polarized light microscopy revealed the
presence of birefringent crystalline material showing different degrees of
crystallinity among the sources. Source 4 (Figure 3.1a) has two populations of
particles, ﬁne particles with very small internal features and some larger entities
(15 —75 pm) with a trace of birefringence. Sources 5 and 7 (Figure 3.1b and d) are
composed of predominantly very small (1 -5 pm) isotropic particles indicating an
amorphous molecular structure. A few larger particles (10-50 pm) are intermixed
and Source 5 shows a tendency to form clumps. Two populations of particles are
seen in the micrograph for Source 6 (Figure 3.1e). The smaller (3-10 pm) are
isotropic with very low bireﬁingence and amorphous structure. The larger
particles (15-50 pm) display more birefringence and have more deﬁned crystal
structure. Source 8 (Figure 3.1e) has a more deﬁned crystal structure than the rest
of the samples with mostly bireﬁingent crystallites consisting of several small
crystals growing radially from a common grain and a uniform particle size (10-15

pm) with a few larger (25 pm) grains.

96

859: 2.9 onoaxnnomem 8:98: 58.8815: 95.... .3 $9.. o.._c> 286... 9.66: ”3:39.985 9:23. c

o5m=m>m $53.22 cozosuoa .cocotﬁ new Encamocaocto 2.8. ..o 83. 30.. new 850m

2-9893. oEoEomEmsg 8:98 mom coach... .5 856.9% ”we? oomtzm m

Boom .oN_m.o.mms_ Eozmsc $62888. ..oﬁEEoEo new 5.995.: on. .2 $99.50
on. .0 E3... on. .0 826a on. o. .92 Si ocm 5.95.6 ﬁctocam n o .mEoEEmE 50.52% E2953 9.. do E:w\em.o.oEm_o

22353 on. .0 .55 .l. Sin. “com... 2.9.8:. 9:29 on. :0 comma 9m 28.0.30 .Smdio 43.0.39 8.3sz 9:29 of.

A83 588$: 33 29:3 :moE ”mn=2

NM ‘3’

EBoEc bﬁozmzﬂw 9m :oEEoo E Bozo. 505.2. 2822 P

 

 

 

 

97

 

 

 

canoe Na... + cmNd 2. a R503 o... ..n no.9 a... a no.9 a... a come. are. P .3
gala
n88 I. a no: a... A Scam. e... a coN e... a. some N... n... cam are. . .3
Maﬁa

can... . . a oN N... a 80mm m. u out a . a 8%.... A... .4 com. 93 __<
N... a on N... a 0% ..o a of N... 44 an N... a 0.9 are. m .3
a... a E e.. a com ..o .4 v.9 ..c n ma 2. a RON are. N .3
o... n m. e... a m. B ..o a at N... a N... N... H.H we. 8...... . .3
claw

n3: ..N a some. .3 a one. N... n oN e... a come 3 a com. 33 __<
m... a Q: o... a o. E o... .4. we a... a. a... e... a o... are. N .o.
N... .4 Sm. o... 4 ON o... A NN a... a ad N... a NN are. . .3
maﬁa...

canoe .4... c... no.3 N..N a con e e a on... ..c 4. no. N. .4 cFN 93 __<
..o a mo. N.m A new. e... n N... ..o a oN N.. a tom are. N .3
a... a 4.3 ..N a cam e... .4 ma 2. a me I n m. .N are. . .3
.33
$9.. .8... 5%.. So .5... B... .E... .2... .5... .2» .5... N.o._ .2...

o>.__.nn=a>no_m 3.2. 35 8.3.9.. 86 8.8%... gm 8.3.2.. n was. 350m

9320: m v a. 2:59.... v a. 2:890: v a. oEEum ounco><
costs» .23 .o $8 .23 co $8 .23 .0 $.. P
Ecomvgcm vice—.1 v; 3.3.35 «:33: cogs—p.830

 

>m¢ 2: 8 team—:8 £925... «9:. 3&3... can maotsehﬁc on: 222.:— ..o 3:53— md 03:.

TABLE 3.4 Particle size sub-populations by multinomial curve fitting

 

Peak 1 Area Peak 2 Area Peak 3 Area
Source Lot MPS (1.1m) (%) MPS (pm) (%) MPS (pm) (%)
Source 4 Lot 1 6 34 50 66 - -
Lot 2 6 33 51 70 - -
Source 5 Lot 1 2 18 - - 18 82
Lot 2 2 43 - - 17 57
Source 6 Lot 1 0.9 0.1 2 0.3 19 99.9
Lot 2 0.6 0.1 4 3 23 96.6
Lot 3 0.7 0.1 4 1.8 20 98.1
Source 7 Lot 1 2 29 - - 17 72
Source 8 Lot 1 - - - - 17 100

 

98

Figure 3.1a-e. Particle size distributions by laser light diffraction and polarized light
micrographs of iron Sources 4-8, magniﬁcation 360X.

99

ooom ooow

 

9“... m. <m
.28. E: on 35.. was.
.23. E: o :22 was.
E: rN._<._.O.. wan—S—

m H 8 >mm

E1 .35 22th

oo

_.

 

 

or

3.0

 

m6

o._.

ON

mN

o.m

md

Volume (%)

100

E: .35 22th

 

 

   

ooom o2... cop or F v.0 E... o
m...
oé
m3
ON
ON
o.m
Wm
0.».

ENE o. <m .
312:5 t 35.. on... m ..
.032. E: N 33.. was. e...
E3 0 ..<..O._. mn=>_
N H ..N >mm m...

 

 

m 3.50m .5...” 9.: E

(%) 9LUFIIOA

101

E1 .oN_m o_o_tmn_

 

 

 

coon ooor o9 or 09 rd 5... o

oé

QN

od

o6

oh

QNE N <m o...
03...... E: 8 m .2... mm...

.28 E: .V 33.. was. 3
.02.... E: o... :3. mn__>_

E3 Ow 31:30.. $15— .
m H 5 >9. .. m

 

 

m 3.50m .0...» 9.39".

(%) ewnIOA

102

E1 65m m_o_tmn_

ooom 000—. on: or

 

QNE : <m
@omt E: 2 $3.. was.
§mmv E: N 3.5.. was.
E: m #90... man—)—

m u. 8 >mm

oé To rod

 

     

 

h mohzom .38 2:9".

md

o._.

mé

o.N

mN

o.m

md

9v

mé

Qm

m6

0.0

(%) ewnIOA

103

0000 002

E1 .35 m_o_tmn_

00.. or F v.0 5.0

 

 

 

ENE no <m

§80
E: 5 $23 was.

E: 0_. ._<._.O.r mn=>_
0 H 00 >mm

  

t

 

0 350w .m_..w.u5 . _u_

 

0N

0.v

0.0

0.0

0.0V

0.N_.

0.:

(%) eumlOA

104

3.4.3 Dynamic Sorption Isotherms

Dynamic vapor sorption (DVS) quantiﬁes small amounts of amorphous material
in an otherwise crystalline solid. The amount of moisture gain and loss in a sample is
measured while relative humidity (RH) is cycled in a step-wise fashion at constant
temperature, in this case from 0-100% RH, generating an isotherm for the material.

A minimum of two consecutive isotherm cycles were run on each source of ferric
orthophosphate. Samples were held at each change in humidity until the sample weight
(moisture gain) stabilized. Within amorphous regions, there was a substantial sorption of
water vapor and increase in mass due to moisture uptake (and moisture release during
desorption). Mass plots showing the percent change in mass due to moisture gain plotted
against time for two isotherm cycles are shown for Source 4 in Figure 3.2. The two
isotherm cycles were identical, indicating that Source 4 did not undergo any moisture
induced collapse of amorphous structure and/or recrystllization with increasing humidity.
These events would have caused marked differences between the two consecutive
isotherms. This was true for all the sources tested, indicating that the material adsorbed
moisture but would be stable when stored at different relative humidities.

Mass plots for Source 6 and Source 4 are shown in Figure 3.3, 3.4, respectively,
Source 6 and Source 4 had moisture uptakes of 1.8% i 0.8 and 11.8% i 3.5, respectively.
The biggest mass sorption increase for most of the sources occurred between 70-90% RH.
Isotherm hysteresis plots (Figure 3.5) display the moisture sorption and desorption
behavior and showed that the three lots of Source 4 had markedly different isotherms
than the rest of the sources with a greater maximum moisture uptake and larger hysteresis.

Sources 8 and 6, and Sources 5 and 7 display similar hysteresis behavior.

105

For Sources 4-8, the maximum moisture uptake varied from as low as 1% for one
lot of Source 6 to as high as 14.3% for one lot of Source 4 (Table 3.5). The following
rank order of increasing moisture uptake (mean of multiple lots) occurred for Sources 4-
8: Source 8 (1.2%) < Source 6 (1.8% i 0.8) < Source 7 (3.4%) < Source 5 (3.5% i 0.5) <

Source 4 (11.8% i- 3.5).

106

@298 8.850% 025

80 2:: #96 EM E850 E mowcmso 035E805 8 380: 8:22: 8 26 mag: E owcmao E850 of 00 BE ”v bouzom .~.m 95E...—

 

(°/.) HUIGBJel

 

0:.

2.
ow.
of
3.

E .
8
cm .

 

 

2: .

82 s... c

 

 

, s.

,- .2
22 m
-8. m
.3. e
, 3. w...
. 8.

. 3..

 

3:0 36:23; 332$. Ill. .2:
3.0 m8: 4

 

 

107

.Acozﬁomog 3:280 mmaE 28 080980 098.85 3,2: me 285 wEBoﬁ 83 8:6 8.853— 6 850m .m.m unsure

 

 

(%) Ha 166121

:55 25h

 

 

 

 

 

 

 

83. 82 82 com. a
O . . E a -. El. iéii l if .-l p - w.QQ
2- . 2:
an. .22
8- w 4.2:
-92: mo
3. w
-92:
cm . w
- z:
8. . w
-u Form
Oh i . \II
.4 SJ.
8. I. -32
8. -33
car. - «a?

 

.x. 36:5; 932$ <
0 «venom :\: moms. 4

 

 

108

.3059583 @3280 $9: 93 €209.80 8:82: 85:. 00 988 wEBoﬁ 80 83 HS: Ebﬁoﬂ ”v 830m in 9.sz

 

 

 

 

 

 

 

 

 

 

.828
3.52::
o8~ 82. 82. 8» o
o r — p n P
S .«2.
on.
+8
8.
m. 8..
M O?-
m 8. 8..
m 8. a:
2-
.NS
8..
8. 6:
8?. w:

 

3:283: ozﬁomd
v oogzom ﬁx: mums. 4

(3;) sum U| oBunqa

 

 

109

63.x. E 098:0 EEEQSE some 3 A32 05 8.80: 0.53650
5:90ch 05 8:98 2:568 5253 woos—68.00% 93E 5 omSEo 2: @526 $20 £35093 goﬁoﬁ ”0+ mooEow .m.m 0.. =3..—

 

 

 

 

 

 

 

 

 

 

30 :1 893
om om ON 00 cm O? on ON or O
_ i O
- md
- —.
i m. —.
o ooSomlOl Q i N W
n mouaowlll .. m.N
ow monsow i
am oohaomll I m W
W mm mohaomIOI S.
A on mouaomlwxl i mm.m “/N
am mousomi I V (\
0v moSow Lil
8. 850m no: r mé
av moiowlli H
-i . m

 

 

110

3.4.3.1 Semi- Quantitation of Amorphous Content

Repetitive isotherms are shown for the amorphous standard that was prepared
from Source 4. This highly amorphous material appeared to be unstable and sublime,
collecting on instrument components during measurements that accounted for a slight
mass loss at the end of isotherm cycles 1 and 2 (Figure 3.6). Mass equilibration was not
reached above 10% RH during the lOOO-minute cycle time of each isotherm due to the
hydroscopicity of the material. A semi-quantitative estimate of the amorphous content of
Source 4-8 was done by ratio of the moisture content of Source 4-8 to the standard
moisture content at 10% RH. Results are given in Table 3.5. The maximum moisture
uptake of the spray-dried material was markedly different than Sources 4-8 and was

estimated to be approximately 60% of the sample by weight.

111

 

.wotomno mm? :oumﬁzﬁmboou Ho 8038 3908 oZ .N can _ 8.93 50:33 :03

£3 $2 39: szm a was 36.3 8.5:: 003 on“ macaw 3820 a :82 8: 20 was 0382: 8 concuaoo $8.: 2: ”03:08:00,020
3.5.“ we 20an 2.2338 msozﬁoﬁa $02 buﬁﬁdxoaamv Btu->2“? 05 80 833 @0503 09:: @0396 020 $22 6.0 9.5“:—

 

00°?

1'

(%)HutaﬁJeL
O O O O C
no N to In

O
0’

00—. -

manganese—05.5 2.9.2 macs—95:2 e\..00_.

 

E25 2.5.

00mm 0000 oomﬂ OOON 00m r 000 —. 00m 0

 

ID
‘—

2 a
(%)wawmaﬁueuo

G—‘FV'O *7?
\-

Q
‘—

3... £2.22. 62.23. <
Ao\0v mwNS— Q

 

 

112

 

 

$5 on T 0v $5 00 T on 0.~ NA _ w 89:80
zigzag-Nomi----------...»...mm.mmH-9- .................... N um .............................. ww- .................... ._ ................... N mamma-
gmx 0w T 00
- ............................................................................. 8 “m-------------------------w.....-mwn_ .............................. $808.88?-
de on T 0v 32M 00 T 00 N8 Wm m o 880m
fog 0w T 00
- - n; 0.0 m 0 880m
$3M 0m T 0v 33m 00 T on 0.m M: _ o 8800
$3M 0w T 00
- ............................................................................. m. mmcnmm .............................. $8.888?-
XEM 0m T 0v ﬁrm 00 T on We QM m m 880m
$5 0w T 00
$5 on T 00 :85 co T 00 m.m ~.m 0 m 880m
- ............................................................................ ma ......................... m. mam-ma: ............................. %Ewﬁi-
REM 0m T 0v X: 00 T 00 Wm _ md m w 880m
$3M 0m T 0m Kim 00 T on w.mm m4: _ v 880m
8.8.82. .80 :\.. I: 8.88.: .80 .x. I: 0x0 0x. 8.36 84 83:35:03
88.80 m=8=088< 8.3.5 «.8882 «9:80
3883mm

 

82.8w 8.: 8.. 8.30.. Am>av .8338 .89; 8:88»: .m.m «3:.

113

3. 4.4 Regression Analyses and Prediction of % RB V

Solubility results given in Table 3.2, showed that relative bioavailability
increased rapidly with increasing solubility for all the iron sources at all acid
concentrations. Figure 3.7a shows a regression plot of solubility versus % RBV.
Due to the rapid (nonlinear) increase in % RBV, the data are presented in Figure
3.7b as a semi-log plot of % RBV versus log 10 solubility. Solubility was found to
be a good predictor of relative bioavailability (R2 90%; P = 0.008).

The physical properties of surface area, median particle size and moisture
uptake were each plotted against log 10 solubility in 0.1N HCl and % RBV. These
relationships are shown in Figures 3.8 — 3.13. Regression analyses of moisture
uptake versus log 30 solubility or % RBV using a power law equation had the best
fit of the data (R2 0.9527) followed by surface area (R2 0.8504) and particle size
(R2 0.3663). The median particle size had the highest correlation of all the
particle size volume statistics, however the correlation was poor.

A regression model of median particle size versus % moisture uptake was
used to predict % RBV and a 2-D plot of the results for both predicted and
actual %RBV is shown in Figure 3.14 and given in Table 3.6. The adjusted R2 for
the model was 93.0%; P = 0.035 (moisture uptake P = 0.031; median particle size
P = 0.055). Estimated % RBVS for the five sources using the regression model
were found to closely estimate the actual % RBV. Nine lots of ferric
orthophosphate were tested from the 5 Sources, including lots that did not have

known % RBVS.

114

RBV vs. Solubility in 0.1 N HCI

 

 

 

 

 

 

Figure 3.73 RBV = 20.67 + 15.15 |0910(0.1 NHCL) Regression
_ _ 95% CI
120- ____95°/o PI
100_ 5 5.90690
R-Sq 92.8%
> 80- R-Sq (adj) 90.4%
m
(I
60-
40
0 50000 100000 150000 200000
Solubility in 0.1 N HCI, p/g Figure 3.8a

 

 

 

RBV vs. Log Solubility in 0.1 N HCI

 

 

 

 

 

 

 

ﬁgure 3.7b RBV = 20.67 + 15.15 log10(o.1 NHCL1) Regression
_ _95°/o CI
120-
____95°/o PI
S 5.90690
10°“ R-Sq 92.8%
R-Sq (adj) 90.4°/o
m 804
60-
40- X
100 1000 10000 100000
Solubility in 0.1 N HCI - log10 Scale Figure 3-8b

 

 

Figure 3.7 a and b. Curvilinear (3.7a) and semi-log (3.7b) plots of RBV versus
Solubility in 0.1 N HCl;

115

 

Log1o(SolubiIity Iron in HCI) vs. Surface Area

1000000 0
100000 2
9 10000 4 :0
3E? 6A;
e - 2 1000 0 a w
:2: 28
a ; :3 100 10 g >
° °r‘-’ 12 a
-‘ .2 10 ..
o 14
1 16
y=-4.01x+20.7 0.1 18

 

R2 = 0.8504 s
ouree

 

- Solubility Iron 0.1N HCl - Surface area -—Unear (Surface area) I

 

 

 

Figure 3.8 Regression analysis of log 10 (solubility iron in HCI) vs. surface area
for Sources 4-8

 

Log1o(Solubility Iron in HCI) vs. Median Particle Size
1000000 ; .1
100000 '“
10000
1000
100
10

1

0.1

Log1o (mcg
Dissovled Iron lg
Total Iron)
(60m) 0213
amped WIN"

 

y = 2.8131 + 0.87
R2 = 0.3663

Source

 

l- Solubility Iron 0.1N HCI - Median Particle Size —Linear (Median Particle Size)

 

 

 

Figure 3.9 Regression analysis of log 10 (solubility iron in HCl) vs. median
particle size for Sources 4-8

116

 

Log1o(Solubility Iron in HCI) vs. Moisture Uptake
1000000
100000
10000
1000

_n
o
C

Total Iron)

..5
0

Logic (meg
Dissovled Iron lg
exeidn ammo“

_L

 

0.1
y = 1122:1313

R2 = 0.9521 Source

 

l- Solubility Iron 0.1N HCl - Moisture uptake (%) —Power (Moisture uptake (%))i

 

 

 

 

Figure 3.10 Regression analysis of log 10 (solubility) vs. % moisture uptake for
Sources 4-8

 

 

% RBV Iron vs. Surface Area
0
i a
S 6 i
a m
> 8 I:
2 10 3
,\° 12 ‘5
14 g
16
18
y = 401x + 20.7 4 5 7 6 3
R2 = 0.8504 Source

 

- RBV - Surface area —Linear (Surface area)

 

 

 

 

Figure 3.11 Regression analysis of % RBV vs. surface area for Sources 4-8

117

 

 

 

 

% RBV Iron vs. Median Particle Size
100 .,
3
c
a.
c 5'
o 3
h

; 1o 2
In a
I! <2
°‘° ;
ﬁ'
0

1

y = 2.8111 + 0.87

R2 = 0.3663 Source
[- % RBV - Median Particle Size —Linear (Median Particle Size)

 

 

 

Figure 3.12 Regression analysis of % RBV vs. particle size for Sources 4-8

 

% RBV Iron vs. Moisture Uptake

% RBV Iron
(%) axeidn amnion

 

y = 11.m-1.3313 4 5 7 6 8
R2 = 0-9527 Source

 

 

 

 

- RBV - Moisture uptake (%) —Power (Moisture uptake (%))

 

 

Figure 3.13 Regression analysis of % RBV vs. % moisture uptake for Sources 4-8

118

wé 88:8 .8 $2

:a 8m >mM ..\.. 8:58 3 own 2098 .362: use 8.3% 2:588 ..\.. .8 E88 coﬁmocwom :4” unsur—

 

 

 

>m¢ .33 I 1..
>9. BEBE i0

 

 

 

 

 

 

n8... 0 a 338 «a 8.3.3 "3.8:. 3n.

325 Sr. 0.35.... EB:

 

mood u a cam 0.25... :98:
Rod 0. a 9.9% 93%.).

 

 

SEED 08382 .x. m=m8>
oﬁm 23th 2352 98¢ >mm ..\.. 868m 8 6on coamewom

119

Table 3.6 Comparison of Actual (By Analysis) to Predicted % RBVS for Sources 4—8

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Actual Predicted
Source & % RBV :1; SD Regression Model Median Particle Moisture Uptake
Lot (By analysis)‘ % RBV Size (mcg)2 (%)3

Source 4

Lot 1 99:9 86 9.3 14.3

Lot 2 86 9.3 9.3
Source 5

Lot 1 78i7 80 2.0 3.2

Lot 2 83 1.8 3.9
Source 6

Lot 1 51:5 55 17.1 1.8

Lot 2 52 9.4 1.0

Lot 3 53 17.7 2.5
Source 7

Lot 1 83:6 81 2.0 3.4
Source 8

Lot 1 6016 55 15.5 1.2

 

 

 

 

 

 

1 RBV Determined by AOAC Rat Hemoglobin Repletion Bioassay (Chapter 2, Table 2.5)
2 Median Particle Size, mcg (Chapter 3, Table 3.3)
3 Moisture Uptake, % (Chapter 3, Table 3.5)

3.5 DISCUSSION

The results of this research demonstrate how widely the physicochemical
properties of ferric orthophosphate vary and offer insight into the factors contributing to
the variable bioavailability documented in the literature (Hurrell 1985; Forbes and others
1989; Willis and Allen 1999; Nalubola and Nestel 2000; SUSTAIN 2001; Dary and
others 2002; Beard 2006; Moretti D and others 2006). The results of Chapter 2
demonstrate this variability for six commercially available sources of ferric

orthophosphate that were found to have markedly different RBVS as determined by the

120

AOAC Rat Hemoglobin Repletion Bioassay. It is impractical and cost prohibitive to use
in vivo RBV testing to study the physiochemical properties of iron sources. Therefore, in
Chapter 3, experiments were performed that characterized the physical attributes thought
to inﬂuence bioavailability, namely particle size, surface area, and amorphous content
using an in vitro solubility method that was shown to provide a good estimation of in
vivo RBV.

Solubility in dilute HCI is an important determinant of bioavailability and an in
vitro method that provides an accurate and reproducible measure of solubility would be
useful in predicting the RBV for a variety of iron sources. Several researchers have
developed and/or modiﬁed in vitro solubility methods and used them to predict the RBVS
of iron fortiﬁcants. To date, it has been difﬁcult to compare results between laboratories
and the conclusions drawn from these studies recognized the need for ﬁirther method
development to reduce the assay variability for routine use (Harrison and others 1976;
Forbes and Erdman 1983; Forbes and others 1989; Swain and others 2003; Moretti D and
others 2006).

In the study by Forbes and others (1989), an in vitro method was evaluated to
measure the solubility of iron (% dissolved iron by weight) of electrolytic iron and fenic
orthophosphate after a 30 minute dissolution period in 0.02 N HCI at 37° C. Electrolytic
iron was 60 to 75% soluble, while only 3 to 4% of the ferric orthophosphate dissolved.
Swain and coworkers (2003) determined the RBVS of six elemental iron powders by the
AOAC hemoglobin repletion assay and measured the surface area by nitrogen absorption.
Solubility in dilute HCI at pH 1.0 and 1.7 was determined over various dissolution

periods. RBVS were found to be signiﬁcantly less than ferrous sulfate (P < 0.05). Better

121

RBV predictability was achieved when the method was modiﬁed to use a lower pH and
less dissolution time (R2 = 0.82, pH 1.0 for 30 min). Surface area measurements ranged
from 90 to 370 mZ/kg and were highly predictive of RBV (R2 = 0.80). Harrison and
others (1976) separated (by nitrogen elutriation) iron powders into different particle size
fractions and found that reduced iron fractions of 7-10 pm and 27-40 um had RBVS of
68-75% and 27-29%, respectively, using the AOAC hemoglobin repletion bioassay. In
the same study, the RBVS of ﬁve ferric orthophosphate samples with different average
particle size ranging from 1 to 14 um (estimated by microscopy) had RBVS between 6
and 46%. RBV was positively correlated with solubility in 0.10 N HCl (R = 0.99) and
negatively correlated with mean particle size (R = 0.95). The authors concluded that
reduced iron should have a mean particle size < 10 microns to achieve a RBV that
approaches 50%.

In this study, the method developed by Shah and others (1977) and modiﬁed by
Forbes and others (1989) was developed further to reduce the variability of the assay. The
method was modiﬁed to incorporate a dissolution apparatus commonly used to measure
the solubility of solid vitamin and mineral preparations (U SPC 2006). This equipment
carefully controls temperature, mixing shear and time, which allowed precise control of
the assay parameters to allow optimization. The 250-ml volume required in the
dissolution apparatus allows for larger samples sizes, which is useful for obtaining a
representative sample of a nonhomogeneous material, 6. g. iron fortiﬁcants with a range
of particle sizes and amorphous contents. Three different HCl concentrations (0.02, 0.05
and 0.1 N), covering the normal physiological pH range found in the stomach, were

tested at different shear rates and dissolution times. The 0.10 N HCl concentration was

122

found to give the most reproducible measure of solubility (CV < 1%), although the
coefﬁcient of variation for all the concentrations was S 5% (Table 3.2). This was in
agreement with the study by Swain and others (2003). The method development in this
study resulted in a validated, robust technique that can potentially be used for intra-
laboratory comparisons. Regression analysis found the method to be a good predictor of
RBV (R2 = 90%, P = 0.008).

It is generally accepted that particle size plays an important role in the
bioavailability of iron fortiﬁcants, particularly reduced iron powders. Less is known
about the relationship of particle size and bioavailability for the other forms of iron.
However, suppliers routinely specify the average particle size of their ingredients,
typically using sieving techniques, for insoluble iron sources like fenic orthophosphate.
A smaller particle size is thought to result in greater solubility in the gut and therefore
higher bioavailability. However, results of this study did not ﬁnd the particle size of ferric
orthophosphate to be well correlated to either solubility or bioavailability with the median
particle size giving the best ﬁt of the particle size distribution statistics (R2 = 0.3663).

Sources 6 and 8 were not statistically different by PSD results and had the largest
median PS (median PS 17.7 and 15.5 urn, respectively). These sources, as expected, had
the lowest bioavailability of the sources tested (RBV 51 and 60%, respectively). Sources
5 and 7 were also not statistically different and had the smallest median PS (2 pm) and
higher RBVS (78 & 83%, respectively) than Source 6 and 8. Source 4 was distinctly
different from all the other sources. It had a median PS of 9.3 pm and a markedly higher
RBV of 99%. Multinomial curve ﬁtting identiﬁed two normal distributions of particles

within the distribution for Source 4. One population comprised 33-34% of the sample and

123

 

was made up of particles with a mean PS of 6 pm. The second population of was

composed of large particles with a mean PS of 50 um, comprising 66-70% of the sample.
The high RBV was also reflected in the high solubility of Source 4, which had 100 to
1000-times the solubility of the other sources. Its drastically higher solubility could not
be accounted for by particle size alone and indicated that other physical properties were
playing an important role in its behavior.

Average particle size measurement is often used as a simple and inexpensive
quality speciﬁcation to indicate higher quality (higher solubility in the gut) iron sources,
particularly for reduced iron powders, but it may be misleading when it is the only quality
measure used to predict bioavailability. Surface area provides a more comprehensive
measure, taking into account both surface and internal structure. It is often used in the
pharmaceutical industry as a predictor of dissolution behavior of drug and supplement
preparations in the body. This research found surface area (R2 = 0.8504 ) to be a better
predictor than particle size (R2 = 0.3366) of solubility and bioavailability. Sources 4, 5
and 7 had 10-times higher surface area than Sources 6 and 8, yet the SA of Source 4 was
only slightly larger than the SA for 5 and 7 and could not completely explain the marked
difference in its solubility.

Surface area is inﬂuenced by several physical properties, such as particle size and
microstructure. Semi-crystalline powders, like ferric orthophosphate, can have varying
amounts of disorder in their microstructures. Amorphous content refers to regions of
molecular disorder within a semi-crystalline solid that is formed during crystallization,
typically during subsequent processing, like milling and spray-drying; it is seldom a

deliberate or controlled event (Buckton and Darcy 1999). Process-induced amorphous

124

‘ﬁ—

content is thought to be localized mostly at the powder surface, increasing surface area.
Amorphous regions in crystalline material will absorb water to a greater degree than
crystalline regions and are often thermodynamically unstable. The presence of
amorphous particle structure is known to be an extremely important determinant of the
bioavailability of active pharmaceutical ingredients and has been given increasing
attention in the pharmaceutical sciences over the past ten years (Mackin and others 2002;
Burnett and others 2006). It is now recognized that even relatively low levels of
amorphous material (<5 %) affect stability and dissolution characteristics (Buckton and
Darcy 1999; Mackin and others 2002; Burnett and others 2006). Therefore, it was
hypothesized that the amorphous content of ferric orthophosphate may be an important
determinant of bioavailability in humans and may also be responsible for negative effects
on food.

Gravimetric vapor sorption (DVS) was chosen as the technique to measure the
amorphous content of ferric orthophosphate powders. This technique is routinely used to
measure amorphous contents below 5 -10% and the amount of amorphous material in the
ferric orthophosphate powders was unknown. In addition, moisture uptake is the main
mode of product quality failure in RTE cereal making DVS a potentially promising tool
to study shelf life and stability in RTE cereal.

In order to quantitate the amorphous content of a material using DVS, an
amorphous standard with known amorphous content is needed and a 100% amorphous
standard is typically used. A fenic orthophosphate standard was not available
commercially at the time of this work. Therefore, the standard was made using Source

4, since the physicochemical behavior of this material indicated a higher amorphous

125

content. The standard material made from Source 4 was found to be extremely
hydroscopic and volatile during analysis. However, semi-quantitative results were
possible, providing an estimate of the amorphous content of the fenic orthophosphate
powders. The amorphous content was calculated directly from the moisture uptake data
by a ratio comparison of the moisture uptake of the standard to the moisture uptake of the
sample under identical experimental conditions. The instability of the standard was
evidenced by sublimation that caused a slight decrease in mass during repeated isotherm
cycles during an analysis. In addition, mass equilibration was closely approximated but
not fully reached above 10% RH during the 1000-minute isotherm cycle-time used in the
method. This was due to the high moisture absorption capacity of the standard. Semi-
quantitative estimations of amorphous content were made using the data at 10% RH.

The amorphous contents of the samples ranged ﬁ'om < 2% to more than 15% (by
weight). Source 4 appeared to contain at least 5 times more amorphous material than the
other sources, accounting for its high solubility and resultant high bioavailability.
Solubility increased rapidly as the amorphous content increased requiring a semi-log
regression analysis. A comparison of the data ﬁt for surface area, moisture uptake (e. g.
amorphous content), or median particle size versus % RBV found moisture uptake to
have the best correlation (R2 = 0.9527) to %RBV, followed by surface area (R2 =
0.8504) and particle size (R2 = 0.3663).

Bioavailability is dependent on the solubility of the iron source, and solubility
depends on several interdependent physicochemical factors. A 2D multiple regression
model of log 10 solubility versus surface area and moisture uptake showed these two

properties to be excellent predictors of solubility (R2 = 0.9333, P = 0.001). However, this

126

regression model did not meet the 95% conﬁdence level (R2 .8480, P=0.078) when used
to predict %RBV. A possible explanation is that although surface area is highly
correlated to solubility, its value is dependent on the sum of the nitrogen absorption of all
the particles present. SA does not distinguish amorphous particles from the less soluble
particles that are much larger in number forming the bulk of the material. A multiple
regression model of median particle size and moisture uptake was found to be a better
predictor of % RBV (R2 = 0.9333, P = 0.035) indicating the interaction between these
two properties may be a more sensitive measure of the soluble particles than surface area.
This model was used to predict the % RBV for four additional lots of the following
sources that did not have actual % RBVS from AOAC Rat Hemoglobin Repletion data :
Source 4 (1 lot), Source 5 (1 lot), Source 6 (two lots) with %RBVs of 86, 83, 52 and 53,
respectively. These values were very close to %RBVs for the tested lots of the same
sources.

A multiple regression model of median particle size and moisture uptake versus
solubility was also a good predictor of solubility (R2 = 0.9100, P = 0.001). Median
particle size, where 50% of the sample is less than the indicated size (um), gave the best
correlation to % RBV and solubility as compared to the mean particle size, the 90% and
the 10% statistics as well as the uniformity statistics. This indicates that the amorphous
material may not be uniformly distributed throughout the material and may reside in the

smaller particle size fraction.

127

CHAPTER 4
SENSORY PROPERTIES OF RTE CEREAL FORTIFIED WITH DIFFERENT
SOURCES OF FERRIC ORTHOPHOSPHATE
4.0 INTRODUCTION

Hydrogen reduced iron is the most widely used iron source for food fortiﬁcation
due to its relative low cost and availability in a range of particle sizes, surface area, porosity
and purity. Its availability over a wide-range of physical parameters allows food
manufacturers some control over the form used for a particular ingredient or product
application. As discussed earlier, bioavailability and reactivity are related to the
physicochemical properties of the iron form and its effect on product quality and shelf life.
The different iron forms have different degrees of reactivity, which generally impair the
organoleptic properties and shelf life of foods, thus limiting the iron forms used in foods to
those that are less reactive. Unfortunately, this also limits the use of more bioavailable iron
sources due to their negative effect on quality.

Due to the need to optimize food quality and bioavailability, the effect of the iron
sources used in this research. on food quality was determined. The sensory attributes of a
standard RTE ﬂaked-com cereal made with the different iron sources were compared to the
sensory attributes of cereal made with a commercially available, encapsulated reduced-iron

source known to have minimal effects on food quality.

128

4.] MATERIALS AND METHODS
4.2.1. Materials

A standard com-based, RTE-cereal was fortiﬁed with iron Sources 1-9, which
included ferrous sulfate (Source 1), encapsulated, hydrogen-reduced iron (Source 2,
control), hydrogen-reduced iron (source 3) and ferric orthophosphate (Source 4-9). Source
2 was encapsulated with zein (corn protein), designed to survive pressurized cooking and
milling processes and to dissolve in stomach acid.

Products were produced in a research pilot plant in 200-lb batches. Cereal was
fortiﬁed at the 45 % RDI level for iron for adults (29 mg iron/ 100g food) with each of the
Sources 1-9. The cereals were also fortiﬁed with the following vitamins and minerals,

typical for the standard RTE cereal:

Table 4.1 Vitamins and minerals added to the test cereals

 

 

 

 

 

 

Vitamin or Vitanrin or Vitamin or
Mineral RDI Mineral RDI Mineral RDI
A 15 B1 25 B; 25
B6 25 Biz 25 C 25
D2 10 E na Fe 45
Folic acid 25 Niacin 25 Zn 25

 

 

 

 

 

 

 

4. 2. 2. Methods

Eleven test batches of RTE cereal were produced in a research pilot plant using a
standard commercial formula and processing conditions for a ﬂaked, corn based cereal.
One lot of each iron source was used to make one 200 lb batch of test cereal. The cereal
was made from whole corn grits cooked in a pressure vessel with added water, ﬂavor,

sweeteners and fortiﬁcants. All iron sources were added to the cereal cooker along with the

129

corn grits and other ingredients except for heat sensitive or chemically incompatible
fortiﬁcants (vitamins A, D2, E, B1, C), which were sprayed on after cooking. Cooking
temperatures and time exceeded 250°F and 50 minutes, respectively. The cereal made with
encapsulated reduced-iron (Source 2) served as the control food to which all other test
cereals were compared. The appearance, taste and texture of the cereals were compared to
the control cereal by the six-member taste panel, who ranked the cereals by consensus on a

scale of l —10, with 10 matching or exceeding the quality of the control product.

4.3 RESULTS

There were obvious food quality differences among the test cereals with regards to
appearance, odor and ﬂavor. Results are shown in Table 4.1. Sources 1 and 4 had the
greatest negative effect on food quality and were ranked 1 with extreme sensory defects
(off-color, gray-green spots, metallic aftertaste). Cereal made with Source 8 had equivalent
product quality as the control cereal (Source 2), which had a bright yellow color, toasted
corn aroma and ﬂavor. The following rank scores for the test cereals fortiﬁed with
decreasing order of quality: Source 8 (10) > Source 2 (10) > Source 6 (8) > Source 9 (7) >

Source 7 (6) > Sources 3 and 5 (5) > Sources 4 and 1 (1).

130

Table 4.2. Sensory scores of RTE cereal made with iron Sources 1-9

 

Iron Source

ID

Sensory Description

Sensory
Score*
(1-10)

 

Ferrous sulfate

Source 1

Extreme presence of grayish green
discoloration, metallic off-odor and metallic
taste with bitter after taste

 

Encapsulated H-reduced iron

Source 2

Control iron source used as a quality standari
for comparison with other iron sources

10

 

H-reduced iron

Source 3

Poor quality but less grayish green
discoloration, off-odor and metallic off-
ﬂavor than Source 1 and 4

 

Ferric orthophosphate

Source 4

Extreme presence of grayish green
discoloration, metallic off-odor and metallic
taste with bitter aftertaste

 

Ferric orthophosphate

Source 5

Poor quality but less grayish green
discoloration, off-odor and metallic off-
ﬂavor than Source 1 and 4 Noticeable gray
spots on ﬂakes.

 

Ferric orthophosphate

Source 6

Slight dullness to ﬂakes and a few grayish
areas, good initial taste and no off-odor or
aftertaste

 

Ferric orthophosphate

Source 7

Slight grayish off-color, dull color and
aftertaste evident

 

Ferric orthophosphate

Source 8

No gray spots, slight dullness, no off odor
and good toasted corn ﬂavor

10

 

 

Ferric orthophosphate

 

Source 9

 

Slight dullness to ﬂakes and gray off-color

 

7

 

 

* 6 Panel Members by Consensus, Food Scores: 1 = worst quality; 10 = best quality

4.4 DISCUSSION

The food was packaged in high density polyethylene liners containing 0.05 lbs

BHT antioxidant per ream. Sensory evaluation took place three-weeks after production,

which is typical practice to allow products to equilibrate before sensory testing.

Fortiﬁcation with Source 8 yielded the best quality food and was a potential match to

standard cereal made with encapsulated reduced iron. It was interesting that the cereal

made with highly soluble ferrous sulfate was similar to that made with ferric

orthophosphate, Source 4. These products had obvious product defects with grayish-

131

green discoloration, a strong metallic off-ﬂavor and a bitter aftertaste. Texture, color and
toasted cereal ﬂavor of the products were diminished. In a sensory study by Lim and
lawless (2006) on the detection thresholds and taste qualities of iron salts, detection
thresholds for several soluble iron salts were between 20.5 — 99.2 pm. The authors
concluded that the taste of iron salts can be adequately described by a combination of
four basic tastes as well as metallic and astringent if only the last two terms are used by
panelists.(Lim and Lawless 2006).

The RBV of the Source 1 and Source 4 were not statistically different and were
the highest of the sources tested (RBV 100% and 99%, respectively). All the iron sources
used in the research had RBV greater than 50, which is generally accepted practice in the
cereal industry to be the lower RBV cut-off for iron fortiﬁcants. As expected, good
quality scores increased as the RBV decreased. Food sensory scores below 8 were not
considered acceptable for commercial production, which limited the commercially viable

iron sources to Sources 8 and 6.

132

SUMMARY AND CONCLUSIONS

The hypothesis of this research dissertation is that the amorphous content of ferric
orthophosphate is a critical determinant of solubility and therefore its bioavailability.

In order to test this hypothesis, the ﬁrst objective of this research was to determine the
relative bioavailability of a variety of commercially available food-grade, ferric
orthophosphate powders using in vivo and in vitro methods . The in vivo approach used in
this research was unique in that the RBVS of the of the iron sources were processed in a
RTE-cereal. Typically, in the AOAC Rat Hemoglobin Repletion Bioassay, the test iron
sources are added to the animal diets as supplements, and are not tested in a food so the
effects of the food matrix and food processing parameters were not considered.

The in vivo RBV was determined for six commercial sources of fenic
orthophosphate and a rapid, reproducible in vitro method was established to estimate
RBV using the solubility of ferric orthophosphate in dilute HCI. The RBVS of the test
sources were found to be variable (RBV 51-99%). However, the overall RBVS were
higher that those reported in the literature. As expected, the solubilities of the sources in
dilute HCI also varied (0.02 — 21%) and were found to be highly correlated with RBV (R2
90%, P = 0.008). The reproducibility in the in vitro solubility assay was excellent (CV <
1%).

The second objective of this work was to understand the physicochemical
properties that inﬂuence solubility and ultimately bioavailability. Particle size and
surface area are more commonly used to characterize the physicochemical behavior of

materials, such as solubility. However, in the last lO-years, the importance of amorphous

133

content, particularly in the pharmaceutical industry, has become increasingly recognized
as techniques to measure it accurately became available.

Gravimetric vapor sorption (DVS) was chosen as the technique to measure the
amorphous content of fenic orthophosphate powders due to its sensitivity and underlying
principle. Vapor sorption or moisture uptake is the main mode of product quality failure
in RTE cereal making DVS a potentially promising tool to study shelf life and stability in
RTE cereal.

Semi-quantitative measurements found the six sources of fenic orthophosphate to
have low to intermediate amorphous contents (2-20%). Amorphous content (moisture
uptake) had the strongest correlation to solubility followed by surface area and then
particle size. However, a regression model of median particle size and moisture uptake
was found to give the best estimates of RBV. Thus, the combination of median particle
size and moisture uptake appeared to be a more sensitive predictor of RBV than surface
area and moisture uptake. The observation might be due to the sensitivity of the median
particle size statistic for distinguishing amorphous material versus surface area, which
measures a bulk property. More work is needed to reproduce and conﬁrm these results
using a validated, quantitative method for determining amorphous content.

Relating the physicochemical properties that determine solubility and ultimately
bioavailability to food quality, is the ﬁnal aim of this research. A standard RTE cereal
was fortiﬁed with the test sources of ferric orthophosphate and evaluated for food quality
against a control cereal. The control was fortiﬁed with a commercially available,
encapsulated H-reduced iron powder known to make excellent quality products. Food

quality was determined by sensory panel testing and only Source 6 and 8 produced food

134

comparable to the control cereal. Product defects included grayish- green discoloration,
off-odor, a strong metallic off-ﬂavor and a bitter aftertaste. Interestingly, cereal made
with highly soluble and reactive ferrous sulfate was similar to that made with ferric
orthophosphate, Source 4.

In conclusion, solubility appears to be a critical determinant of the bioavailability
and reactivity of ferric orthophosphate and is dependent on the interaction of particle size,
surface area and perhaps most importantly, the presence of amorphous microstructure.
The effect of varying amounts of amorphous material on solubility is not linear with
solubility, increasing rapidly in a power law relationship as the amorphous content
increases between 3 and approximately 20%. The %RBV of ferric orthophosphate was as
high or higher than the H-reduced iron typically used to fortify RTE-cereal.

Ferric orthophosphate (F ePO4) is a promising iron source for food fortiﬁcation
because of its light color and oxidative stability. However, FePO4 has had limited use in
cereal due its variable and often reported low bioavailability. This research contributes
toward a better understanding of the physicochemical properties that are determinants of
its bioavailability. An understanding of the mechanisms underlying this variability may
facilitate production of a FePO4 source with consistent bioavailability and also offer an

affordable iron source for applications where improved oxidative stability is needed.

135

FUTURE RESEARCH

Suggestions for future research:

Further work is needed to validate the DVS technique using a well-characterized
amorphous standard and test the RBV in vitro prediction models against new lots of ferric
orthophosphate. These tools are needed to answer the following question: How closely
can the amorphous content and particle size of ferric orthophosphate powders be
controlled within cost and variability limitations?

One of the test sources of ferric orthophosphate (Source 4) was markedly different
than the other sources and shed valuable insight on the properties inﬂuencing
bioavailability. However, there were no sources available that had an amorphous content
between 3 and approximately 20%. In addition, the bioavailability and food quality
varied markedly at amorphous content between 1 -— 3%. Can particle size and amorphous
content be optimized to produce material within wider process allowances that maintains
improved bioavailability?

Finally, Can the degree of amorphous material present be manipulated to optimize
food quality and bioavailability? We do not know if it is a matter of the degree of
disorder in all the particles present or if there is a speciﬁc soluble population that is

determining bioavailability.

136

APPENDIX A

Beckeline Test Results for Refractive Indices of Ferric Orthophosphate

Analysis was performed by Bayard development Company, 1425 W. Summerdale Ave.
Chicago, IL 60640, 773-728-0531

The refractive indices for the sodium (1 line (5 89 nm) were determined for the ferric
orthophosphate sources using the standard microscopic immersion method.

Table 1.0 Refractive Indices for Ferric Orthophosphate Sources

 

Iron Source

Refractive Index

Observations

 

Source 4, Lot 1

5%: 1.680 — 1.705
50 — 55%: 1.680
40 - 45%: 1.632 - 1.675

Isotropic mixture

 

Source 4, Lot 2

3-5%: 1.680 — 1.706
60 — 65%: 1.680
30 — 35%: 1.630 — 1.675

Isotropic mixture

 

Source 5, Lot 1

1.681

Isotropic — very small

 

 

 

 

 

 

 

 

 

fused crystallites
Source 5, Lot 2 1.680 ISOthlC — VCD’ small
fused crystallites
Source 6, Lot 1 1.687 _ 1.704 Brrefnngent sphereulrtes
Source 6, Lot 2 1.674 — 1.709 Very pogrly formed
sphereulrtes
Source 6, Lot 3 1.675 — 1.707 Very P09F1)’ formed
sphereulrtes
Isotropic - very small
Source 7’ La 1 1.682 fused crystallites
Source 3, Lot 1 1.687 _ 1.704 Brrefnngent sphereulrtes

 

137

 

APPENDIX B

3-Point B.E.T Surface Area Analysis - Instrument Parameters

Analyses were performed at Particle technology labs, Ltd., Downers Grove, IL 6-515,
630-969-2703

Instrumentation: Quantachrome® Autosorb AS-l Static-Pressure Analyzer
Sample weight: approximately 2.2 -2.6 grams (four decimal places)
Experimental parameters:

Adsorbate: Nitrogen

Cross-section area: 16.2 Angstroms 2 /molecule

Non Ideality: 6.580E-05

Molecular Weight: 28.135 g/mol

Outgas Temp: 25° C

Outgas Time: 16 hours

Analysis Time: approximately 38-83 minutes (sample dependent)

Bath Temp: 77.4 °K

138

LIST OF REFERENCES

1998. Handbook of Powder Metal Technologies and Applications. Materials Park, Ohio:
American Society of Metals.

ADA. 2001. Position of the American Dietetic Association: Food fortiﬁcation and dietary
supplements. Journal of the American Dietetic Association 101(1):115-25.

AOAC. 1990. Bioavailability of Iron: Rat Hemoglobin Repletion Bioassay. AOAC,
editor. 15th ed. Arlington, Virginia: AOAC. 1098-9 p.

AOAC International. 2000. Iron Test by ICP. In: Covance Laboratory Method #661,
editor. Ofﬁcial Methods of Analysis of AOAC International. 17 ed. Gaithersburg,
MD.

Arora A. 1997. Evaluation of Oxidative Stability of Lipids By Fluorescence
Spectroscopy and Characterization of Flavonoid Antioxidant Chemistry
[Dissertation for Doctor of Philosophy in Food Science and Human Nutrition].
East Lansing: Michigan State University. 207 p.

Beard J. 2006. Iron. In: Bowman BA, Russell R, M., editors. Present Knowledge in
Nutrition. Ninth ed: ILSI Press. p 430-44.

Beard J, Dawson H. 1997. Iron. In: CDC“ BL, Sunde RA, editors. Handbook of
Nutritionally Essential Mineral Elements. New York: Marcel Dekker, Inc. p 275-
334.

Benito P, Miller D. 1998. Iron Absorption and Bioavailability: An Update Review.
Nutrition Research 18(3):583-603.

Benoist B. Introduction: Iron Deﬁciency Anemia: Reexarnining the Nature and

Magnitude of the Public Health Problem. In: Beard J, Stoltzfus R, editors; 2001;
Belmont, MD. American Society for the Nutritional Sciences. p 564S.

Bhadeshia HKDH. 2002. Differential Scanning Calorimetry. University of Cambridge
Materials Science and Metallurgy.

Bridges KR. 2007. Iron Absorption, Iron Transport and Cellular Uptake. Joint Center for
Sickle Cell and Thalassemic Disorders, Harvard Medical School.

Buckton G, Darcy P. 1999. Assessment of disorder in crystalline powders - a review of
the analytical techniques and their application. International Journal of
Pharmaceutics 179: 141-5 8.

Budavari S, O'Neil MJ, Smith A, Heckelman P, Kinneary J, editors. 1996. The Merk
Index. 12 ed. Whitehouse Station, NJ: Merck & Co., Inc.

139

Budenheim C. 1995. Technical Bulletin on ferric Orthophosphate.

Burnett D, Thiehnann F, Williams D. 2006. Characterizing the Amorphous State in a
Pharmaceutical Powder using Vapour Sorption Analysis. In: Dickrnann R, editor.

research presentation ed: Surface Measurement Systems, Allentown, PA, USA. p
1-36.

Dary O. 2002. Staple Food Fortiﬁcation with Iron: a Multifactorial Decision. Nutrition
Reviews 60(7):(II)S34-S41.

Dary O, Freire W, Kim S. 2002. Iron Compounds for Food Fortiﬁcation: Guidelines for
Latin America and the Caribbean 2002. Nutrition Reviews 60(7):(II)S50-S61.

Decker EA, Hultin H0. 1992. Lipid Oxidation in Muscle Foods via Redox Iron. In:
Angelo AJ S, editor. Lipid Oxidation in Food. ﬁrst ed. Washington, DC: American
Chemical Society. p 33-4.

Dianzani MU. 1991. Dietary Prooxidants. In: Dreosti IE, editor. Trace Elements,
Micronutrients and Free Radicals. Totowa, New Jersey: Humana press. p 91.

Dutra-de-Oliveira JE, Freitas MLS, Ferreira J F, Goncalves AL, Marchini J S. 1995 . Iron
From Complex Salts and Its Bioavailability to Rats. Intemat J Vit Nutr Res
65:272-5.

F airweather—Tait S, Teucher B. 2002. Iron and Calcium Bioavailability of Fortiﬁed Foods
and Dietary Supplements. Nutrition Reviews 60(12):360-7.

FDA. 1999. The Food Label. US. Food and Drug Administration.

FDA Department of Health and Human Services. 2003. Nutritional Quality Guidelines
for Foods, Subpart B--Fortiﬁcation Policy Statement of Purpose. In: Code of
Federal Regulations Title 21 V, Chapter 1, 104.20 editor: US Government Printing
Ofﬁce via GPO Access.

F ennema OR, editor. 1996. Food Chemistry. 3rd ed: Marcel Dekker, Inc. 1069 p.

Forbes AL, E C, Adams M, Amaud J, Chichester C, Cook J, Hanison B, Hurrell R, Kahn
S, Morris E, Tanner J, Whittaker P. 1989. Comparison of in Vitro, Animal, and
Clinical Determinations of Iron Bioavailability: International Nutritional Anemia
Consultative Group Task Force Report on Iron Bioavailability. Am J Clin
Nutr(49):225-38.

Forbes R, Erdman J. 1983. Bioavailability of Dietary Iron in Man. Annual Reviews in
Nutrition 3:213-31.

Genck WJ, Bayard M. 1997. Microanalysis of Ferric Orthophosphate Samples. Richton
Park, IL: Genck International.

140

Government US. 2006. Title 21, Iron, elemental 184.1375, Food and Drugs,. In: Code of
Federal Regulations, editor: National Archives and Records, Administration,
Ofﬁce of the Federal Register.

Hallberg L. 1981. Bioavailability of Dietary Iron in Man. Ann Rev Nutr 1:123-47.

Hallberg L, Hulthen L. 2000. Prediction of dietary iron absorption: an algorithm for
calculating absorption and bioavailability of dietary iron. Am J Clin Nutr
71 :1 147-60.

Hallberg L, Rossander-Hulthen L, Grarnatkovski E. 1989. Iron fortiﬁcation of ﬂour with
a complex ferric orthophosphate. Am J Clin Nutr 50:129—35.

Harrison BN, Pla GW, Clark GA, Fritz J C. 1976. Selection of iron sources for cereal
enrichment. Cereal Chemistry 53:78-84.

Heng J, Williams DR. 2006. Ferric Orthophosphate DVS Analysis Report London, UK:
Imperial College London.

Hider RC, Singh S. 1993. Iron chelators of clinical signiﬁcance. In: J .Silver, editor. Iron
chelators of clinical signiﬁcance. ﬁrst ed. London: Blackie Academic &
Professional. p 275.

Hoppe M, Hulthen L, Hallberg L. 2006. The relative bioavailability in humans of
elemental iron powders for use in food fortiﬁcation. Eur J Nutr 45(1):37-44.

Hurrell RF. 1985. Nonelemental Sources. In: Clydesdale FM, Wiemer KL, editors. Iron
Fortiﬁcation of Foods. New York: Academic Press. Inc. p 39-51.

Hurrell RF. 1997. Preventing Iron Deﬁciency Through Food fortiﬁcation. Nutrition
Reviews 55(6):210-21.

Hurrell RF. 2002a. Fortiﬁcation: Overcoming Technical and Practical Barriers. T13
American Society for Nutritional Sciences Journal Nutrition, 132:806S-12S.

Hurrell RF. 2002b. How to Ensure Adequate Iron Absorption from Iron-Fortiﬁed Food.
Nutrition Reviews 60(7):S7-S15.

Hurrell RF, Bothwell T, Cook JD, Dary O, Davidsson L, Fairweather-Tait S, Hallberg L,
Lynch S, Rosado J, Walter T, Whittaker P. 2002. The Usefulness of Elemental
Iron for Cereal Flour Fortiﬁcation: A SUSTAIN Task Force. Nutrition Reviews
60(12):391-406.

INACG. Why Iron is Important and What to Do About It: a New Perspective; 2001;
Hanoi, Vietnam. Micronutrient Global Leadership, a project of the Ofﬁce of
Health, Infectious Disease and nutrition, Bureau for Global Health, US Agency
for International.

141

IOM. 2001. DRI: Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron,
Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon,
Vanadium and Zinc. FNB Panel on Micronutrients, editor. First ed. Washington
DC: National Academy Press

p290-3 78.

IUPAC. 2003. IUPAC Compendium of Chemical Terminology, Electronic Version:
Haber-Weiss and Fenton Reactions. In: Nic M, Jirat J, Kosata B, editors: ICT
Prague, Czech Republic.

Keenan CW, Wood J H. 1971. Metals 1: Chemical Properties and Production. General
College Chemistry. Fourth ed. New York: Harper and Row. p 429-48.

Larsson M, Minnekus M, Havenaar R. 1997. Estimation of Bioavailability of iron and
Phosphorous in Cereals using a Dynamic In Vitro Gastrointestinal Model. .1 Sci
Food Agric 74299-106.

Lee K, Clydesdale F. 1979. Iron Sources Used in Food Fortiﬁcation and Their Changes
Due to Food Processing. CRC Critical Reviews in Food Science and
Nutrition(February 1979):] 17-53.

Lim J, Lawless HT. 2006. Detection thresholds and taste qualities of iron salts. Food
Quality and Preference 17:513-21.

Lowell S, Shields J E. 1991. POWDER: Surface Area and Porosity. Third ed. New York:
Wiley and Sons Inc. 250 p.

Lynch S. 2002. Food Iron Absorption and Its Importance for the Design of Food
Fortiﬁcation Strategies. Nutrition Reviews 60(7):S3-S6.

Lynch SR. 2000. The effect of calcium in iron absorption. Nutrition Research Reviews
13:141-58.

Mackin L, Zanon R, Park JM, Foster K, Opalenik H, Demonte M. 2002. Quantiﬁcation
of low levels (<10%) of amorphous content in micronized active batches using

dynamic vapor sorption and thermal microcalorimetry. International Journal of
Pharmaceutics 231 2227-3 6.

Madison Chemicals I. 1996. Ferric Orthophosphate. Madison Chemicals, Inc. p 1-2.

Malvem. 2000. Laser diffraction for particle size analysis - why use the Mie theory? :
Malvem Instruments.

Malvem. 2003. Mastersizer 2000 - the science behind the technology. 2003 MRK502-01
ed. Worcestershire, United Kingdom: Malvem Instruments.

142

Martinez-Navarrete N, Camacho MM, Martinez-Lahuerta J, Martinez-Monzo J, Fito P.
2002. Iron deﬁciency and iron fortiﬁed foods - a review. Food Research
International 35 :225-3 1.

Miret S, Simpson RJ, McKie aT. 2003. Physiology and Molecular Biology of Dietary
Iron Absorption. Annu Rev Nutr 23:283-301.

Morck TA, Cook JD. 1981. Factors Affecting the Bioavailability of Dietary Iron. Cereal
Foods World 26(12):667-72.

Moretti D, Zimmermann M, Wegrnuller R, Walczyk T, Zeder C, Hurrell R. 2006. Iron
status and food matrix strongly affect the relative bioavailability of ferric
pyrophosphate in humans. Am J Clin Nutr 83(3):632-8.

Murphy DB, Rousseau V. 1969. Transition Metals and Complexes. Foundations of
College Chemistry. New York: The Ronald Press Company. p 569-83.

Nalubola R, Nestel P. 2000. Manual for Wheat Flour Fortiﬁcation with Iron. Arlington,
VA: US. Agency for International Development (USAID). Report nr OMNI
HRN-5122-A-OO-3046—00 and MOST HRN-A-00-98-00047-00.

Nawar W. 1985. Lipids. In: Fennema OR, editor. Food Chemistry. 2 ed. new York:
Marcel Dekker, Inc. p 205-10.

Nawar W. 1996. Lipids. In: Fennema OR, editor. Food Chemistry. 3 ed. New York:
Marcel Dekker, Inc. p 275-6.

Oetker RA. 1992. Ferric Orthophosphates for Fortiﬁcation. Mainz, Germany: Chemische
Fabrik Budenheim.

Oliveira V, Racine E, Olimsted J, Ghelﬁ LM. 2002. The WIC Program: Background,
Trends, and Issues. Washington, DC: USDA
Report nr FANRR-27. 1-40 p.

Patrick J. 1985. Elemental Sources. In: Clydesdale FM, Wiemer KL, editors. Iron
Fortiﬁcation of Foods. New York: Academic Press. p 31-8.

Penland JG. 2001. N ew Dietary Reference Intakes for Micronutrients Released: GF
Human Nutrition Research Center makes Major Contribution. Grand Forks
Human Nutrition Research Center of the USDA, Agricultural Service.

Quantachrome. 1997. Autosorb-1 Operating Manual. Chapter IV Theory and Discussion:
Quantachrome, Corporation. p 64-7.

Rasmussen E, Hallberg L, Rossander L. 1977. Absorption of Fortiﬁcation Iron. Br J Nutr
372375-88.

Rawle A. 2003. Basic Principles of Particle Size Analysis. Malvem Instruments Limited.

143

Roos YH. 1995. Phase Transitions in Foods. Taylor SL, editor. First ed. San Diego:
Academic Press, Inc.

Sandstrom B. 2001. Micronutrient interactions: effects on absorption and bioavailability.
Brittish Journal of Nutrition 85(Suppl. 2):Sl8l-S5.

Shah BG, Giroux A, Befonjie B. 1977. Speciﬁcations for reduced iron as an additive. J
Agric Food Chem 25:592-4.

Silver J. 1993. Introduction to Iron Chemistry. In: J .Silver, editor. The Chemistry of Iron.
ﬁrst ed. London: Blackie Academic & professional. p 1-24.

Simic MG, J ovanovic SV, Niki E. 1992. Mechanisms of lipid Oxidative Processes and
Their Inhibition. In: Angelo AJ S, editor. Lipid Oxidation in Food. Washington,
DC: American Chemical Society. p 14-31.

Stoltzfus R. 2001a. Deﬁning Iron-Deﬁciency Anemia in Public Health Terms: A time for
Reﬂection. The Journal of Nutrition 131(2S-II):565S-7S.

Stoltzfus R. Iron-Deﬁciency Anemia: Reexamining the Nature and the Magnitude of the
Public Health Problem. In: Beard J, Stoltzﬁrs R, editors; 2001b; Belmont, MD.
American Society for Nutritional Sciences. p 564S-7S.

SUSTAIN. 2001. Guidelines for Iron Fortiﬁcation of Cereal Food Staples. Washington,
DC: SUSTAIN: Sharing United States Technology to Aid in the Improvement of
Nutrition. 1-7 p.

SUSTAIN Task Force. 2001. Guidelines for Iron Fortiﬁcation of Cereal Food Staples.
Washington, DC: SUSTAIN: Sharing United States Technology to Aid in the
Improvement of Nutrition. 1-7 p.

Swain J H, Newman SM, Hunt JR. 2003. Bioavailability of elemental iron powders to rats
is less than bakery—grade ferrous sulfate and predicted by iron solubility and
particle surface area. J Nutr 133(11):3546-52.

Symons M, Gutteridge J. 1998. Free Radicals and Iron: Chemistry, Biology and Medicine.
ﬁrst ed. Oxford: Oxford University Press. 242 p.

Turner L. 2002. Monterrey Workshop Summary: Evaluating the Usefulness of Elemental
Iron Powders. Nutrition Reviews 60(7):S16-S7.

USPC. 2006. Disintegration and Dissolution of Dietary Supplements. USPC, Inc Ofﬁcial
Methods. p 1-9.

Waddell J. 1974 Bioavailability of Iron Sources. Nutrition for Food Executives: 80-6.

144

 

Whitney EN, Cataldo CB, Rolfes SR. 1998a. An Overview of Nutrition. Understanding
Normal and Clinical Nutrition. 5 ed. Belmont: Wadsworth Publishing Company.
p 15-24.

Whitney EN, Cataldo CB, Rolfes SR. 1998b. The Trace Minerals. Understanding Normal
and Clinical Nutrition. 5th ed. Belmont, California: Wadsworth Publishing
Company. p 449-63.

Wienk KJH, Marx JJ, Beynen AC. 1999. The concept of iron bioavailability and its
assessment. Eur J Nutr 38:51-75.

Willis RB, Allen PR. 1999. Measurement of amorphous ferric phosphate to assess iron
bioavailability in diets and diet ingredients. Analyst 1242425-0.

Winterboum CC. 1991. Free Radical Biology of Iron. In: Dreosti IE, editor. Trace
Elements, Micronutrients, and Free Radicals. Totowa, New Jersey: Humana Press.
p 53-76.

Wortley G, Leusner S, Good C, Gugger E, Glahn R. 2005. Iron availability of a fortiﬁed
processed wheat cereal: comparison of fourteen iron forms using an in vitro
digestion/human colonic adenocarcinoma (CaCo-2)cell model. Br J Nutr
93(1):65-71.

Wunderlich B. 1990. Thermal Analysis. First ed. San Diego: Academic Press, Inc.

145

   

M'11111111111111iiiililillililijiliii'Es