f». L
part?
3.1.... . .
3
gift»:

 

: ..
4.115 a

Laban. .

‘ “11.1451: all. 21:445'3114. 315.....zodnmw .M.fin.nvunow.;u....t.:7r§:: . . , , V ‘ r. ‘ ‘ . H. V , V . .na wanvn43

 

 

 

 

LIBRARY

Michigan State This is to certify that the
UnIve TSIIY thesis entitled

 

 

FISH CONSUMPTION AND MATERNAL HAIR MERCURY
LEVELS IN RELATION TO THE RISK OF PRETERM
DELIVERY

presented by

Fei Xue

has been accepted towards fulfillment
of the requirements for the

MS. degree in Ememiolgqy

 

 

Major Professor’s Signature
Wags 29:93
i /

Date

MSU is an Affirmative Action/Equal Opportunity Institution

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJClFlC/DntoDquBS-p. 15

 

FISH CONSUMPTION AND MATERNAL HAIR MERCURY LEVELS IN
RELATION TO THE RISK OF PRETERM DELIVERY

By

Fei Xue

A THESIS

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

MASTER OF SCIENCE
Department of Epidemiology

2003

ABSTRACT

FISH CONSUMPTION AND MATERNAL HAIR MERCURY LEVELS IN
RELATION TO THE RISK OF PRETERM DELIVERY

By

Fei Xue

Preterm delivery (PD) has continued to be the leading cause of fetal mortality and
morbidity despite extensive etiological studies and advancements in medical technologies.
Recently, fish consumption has been hypothesized to be related to decreased risk of PD
through the effect of omega—3 fatty acids (n-3 FA). However, other researchers have
indicated that mercury, which comes mainly from fish consumption for many human
populations has adverse effects on fetal development. This prospective cohort study was
based on the Pregnancy Outcome & Community Health (POUCH) Study and assessed
the association between maternal fish consumption during the first half of pregnancy and
the risk of PD among 1326 pregnant women, and the association between fish
consumption and hair mercury levels and the influence of hair mercury levels on the risk
of PD among 1031 women who agreed to give hair samples. Results showed that the top
10th percentile of hair mercury levels was significantly associated with increased risk of
very PD (<35 weeks) and the 4th quintile was significantly associated with increased risk
of moderately PD (36-37 weeks). Total fish consumption, consumption of canned fish
and sport caught fish were significantly associated with increased hair mercury levels.
Other maternal characteristics including age, ethnicity, Medicaid status and community
of enrollment were also associated with hair mercury levels after adjusting for fish

consumption. Total fish consumption was not found to be associated with the risk of PD.

ACKNOWLEDGEMENTS

This work would not have been possible without the instruction and support from my
advisor, Dr. Claudia Holzman, who provided me with continuing motivation and
encouragement. I would like to thank Dr. Larry Fischer for his precious ideas and the
funding which supported me during my two years of research and study. Dr. M. Hossein
Rahbar, my committee member, has also been abundantly helpful, and has assisted me in
numerous ways.

Many people have been a part of my graduate education, as friends,
teachers, and colleagues. Sincere thanks to Dr. Nigel Paneth, who inspired and
encouraged me in the study in Epidemiology. I would like to express gratitude to Dr.
Wilfried Karmaus, Dr. Jianping He, Dr. Wenjiang Fu, Dr. Joseph Gardiner and Dr.
Michael Collins for their insightful teaching and kindness. Mr. Sean Zhou, Dr. Yingzi
Deng and Mrs. Lora McAdams, I am grateful for all your friendship and supports.

I cannot end without thanking my parents in China and Mr. Lu Zi Wei, on whose

constant encouragement and love I have relied throughout my time in the US.

iii

TABLE OF CONTENTS

LIST OF TABLES ................................................................................. vi
LIST OF FIGURE ................................................................................. vii
LIST OF ABBREVIATIONS ................................................................... viii
INTRODUCTION
The Problem of Preterm Delivery ............................................................... 1
Fish Consumption and Adverse Pregnancy Outcomes ..................................... 4
Fish Consumption, Mercury Exposure and Fetal Development ......................... 18
Measurement of Exposure to n-3 FA and Mercury ........................................ 25
MATERNAL FISH CONSUMPTION IN RELATION TO PRETERM DELIVERY
Introduction ......................................................................................... 29
Methods
Population ....................................................................................... 30
Fish Consumption ............................................................................... 31
Gestational Age at Delivery .................................................................... 31
Analytic Strategy ............................................................................... 31
Results ............................................................................................... 32
Discussion ........................................................................................... 40
Studies Assessing Fish Consumption in Relation to Preterm Delivery ................. 41
Studies Assessing Blood n-3 FA levels in Relation to Preterm Delivery ............... 43
Experimental Studies and Clinical Trials ................................................... 45
Conclusion .......................................................................................... 47

MATERNAL MERCURY LEVELS IN HAIR IN RELATION TO FISH
CONSUMPTION AND RISK OF PRETERM DELIVERY

Introduction ........................................................................................ 49
Methods
Population ....................................................................................... 50
Fish Consumption .............................................................................. 51
Mercury Levels in Hair ........................................................................ 51
Gestational Age at Delivery ................................................................... 52
Analytic Strategy ............................................................................... 52
Results
Maternal Characteristics in Relation to Sampling .......................................... 53
Fish Consumption .............................................................................. 53
Total Fish Consumption in Relation to Maternal Characteristics ........................ 54
Mercury Levels in Hair in Relation to Maternal Characteristics .......................... 55
Mercury Levels in Relation to Fish Consumption .......................................... 56

Mercury Levels in Hair in Relation to Maternal Characteristics after Controlling for

iv

Total Fish Consumption ........................................................................ 57

Mercury Levels in Hair in Relation to the Risk of Preterm Delivery .................... 58
Discussion ............................................................................................ 68
Conclusion .......................................................................................... 74
SUMMARY ......................................................................................... 75
REFERENCE ...................................................................................... 79

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

LIST OF TABLES

Literature review of studies on consumption of fish/fish oil supplements in
relation to pregnancy outcomes including GA, BW, IUGR or PIH ................ 8

Literature review of previous studies on matemal/fetal mercury levels in relation
to GA or BW .............................................................................. 22
Maternal characteristics of study sample (N=1326) ................................. 35
Number of fish meals* (total fish and subtypes) in the first half of pregnancy..36
Maternal characteristics in relation to total fish consumption ...................... 37

Percentage and Odds Ratio of preterm deliveries in fish consumption groups
evaluated at multiple thresholds ........................................................ 38

Some demographic characteristics in the 1024 POUCH participants with
mercury assessments in sampled hair and in 197 participants with no hair
sampled .................................................................................... 59

Number of fish meals*** (total fish and subtypes) in the first half of
pregnancy .................................................................................. 60

Total fish consumption (meals/6 months) in relation to maternal
characteristics .............................................................................. 61

Table 10. Mercury levels (ppm range) in hair sample in relation to maternal

characteristics in both crude model and after adjusting for all the other
maternal characteristics ................................................................. 62

Table 11. Mercury levels in hair in relation to fish consumption ............................. 64

Table 12. Mercury levels (ppm range) in hair sample in relation to maternal

characteristics after adjusting for total fish consumption and other maternal
characteristics ............................................................................ 66

Table 13. The association between preterm delivery and mercury levels (ppm range) in

hair ....................................................................................... 67

vi

LIST OF FIGURES

Figure 1. Location of the five communities in this study ...................................... 69

vii

LIST OF ABBREVIATIONS

WHO .................................................................... World Health Organization
PD ................................................................................... Preterm Delivery
BW ....................................................................................... Birth Weight
LBW ............................................................................... Low Birth Weight
IUGR ............................................................. Intrauterine Growth Retardation
PIH ................................................................ Pregnancy Induced Hypertension
HPA ............................................................... Hypothalamic-Pituitary-Adrenal
CRH ............................................................. Corticotropin-Releasing hormone
PPROM ................................................ Preterm Premature Rupture of Membrane
MLCK .................................................................. Myosin Light Chain Kinase

IVF .............................................................................. In-Vitro Fertilization
GIFT ................................................................ Gamete Intrafallopian Transfer
n-3 FA .......................................................................... Omega-3 Fatty Acids
n-6 FA .......................................................................... Omega-6 Fatty Acids
EPA .................................................................................. Eicosapentaenoic
DHA ................................................................................ Docosahexaenoic

PGE2 ................................................................................. Prostaglandin E2
PGF2a .............................................................................. Prostaglandin F2a
PGI2 ....................................................................................... Prostacyclin
PG13 ...................................................................................... Prostaciclina
AA .................................................................................. Arachidonic Acid

viii

AAS .............................................................. Atomic Absorption Spectrometry

CVAAS .......................................... Cold Vapor Atomic Absorption Spectrometry
NAA ............................................................... Neuron Activation and Analysis
XRF .............................................................................. X-Ray Fluorescence
GC ............................................................................... Gas Chromatography
MSAFP ........................................................ Maternal Serum Alpha-Fetoprotein
POUCH ............................................. Pregnancy Outcome and Community Health
LMP ............................................................................ Last Menstrual Period
GLM .......................................................................... General Linear Models
USA EPA ...................................... United States Environmental Protection Agency
RfD .................................................................................... Reference Dose
PPM .................................................................................. Parts Per Million
DDT ........................................................... Dichloro Diphenyl Trichloroethane
PCB ....................................................................... Polychlorinated Biphenyl

INTRODUCTION

fie Problem oLPreterm Delivery

Preterm Delivery (PD), defined by the World Health Organization (WHO) as birth
before 37 completed weeks of gestation (1), remains the leading cause of perinatal
morbidity and mortality in obstetric practice in developed countries. PD is commonly
classified by clinical presentations as birth occurring after spontaneous preterm labor (in
approximately 50% of cases), spontaneous rupture of the membrane (in approximately
30%), and delivery of a premature infant as indicated for the benefit of either the infant or
the mother (in around 20%) (28). The prevalence of PD varies from 6% to 15% of all
deliveries across populations with different geographic and demographic features (2-5).
Three factors make PD a public health concern in the US: the substantial emotional and
economic costs to families and communities produced by prematurely delivered infants,
the higher rate of PD in low socio-economic status and ethnic minority patients, and the
continually increasing incidence despite of extensive efforts to predict and prevent PD.
Over the period from 1987 to 1998, the rate of PD increased from 10.2% to 11.6% of live
births in the US (6). Factors that have been suggested to contribute to the rising rate
include increased obstetric intervention (10), increased multiple birth driven by the use of
assisted reproduction techniques (1 1-13), raised prevalence of substance misuse in urban
areas (14), adverse socioeconomic factors (14-16) and changes in the better dating of
gestation with the use of ultrasound (16). Over the twenty years between 1975 and 1995,
PD increased by 3.6% among African-American women (from 15.5% to 16.0%), and by
22.3% among Whites (from 6.9% to 8.4%) (7). Though the racial gap is diminishing, it is

more likely to be explained by increased rate among whites.

Around 75% of perinatal deaths happen in prematurely delivered infants and more
than two thirds of these occur in infants delivered before 32 weeks’ gestation (3, 8-9).
The survival rate of preterm and Low Birth Weight (LBW) infants has been increased in
recent years, especially for those delivered extremely early in gestation. However,
because 80% of preterm infants are born at 32-36 weeks’ gestation, though their mortality
and morbidity are lower than extremely PD, the sheer number of this group makes the
largest contribution to the total perinatal mortality following preterm birth (46). Moreover,
the increased short-term morbidity and long-term physical and mental disability in
survivors of premature infants have compromised the overall reduction in perinatal
mortality. Hence, reducing the incidence of all PD through etiological research seems to
be a more powerful way of improving perinatal health despite the advances we have
made in neonatal medicine. Unfortunately, current approaches to predict and prevent PD
are unsuccessful and it is mainly due to an inadequate understanding of the underlying
pathogenesis. Four pathogenic processes have been supported by clinical and
experimental evidences: activation of the maternal or fetal hypothalamic-pituitary-adrenal
(HPA) axis, decidual-chorioamniotic or systemic inflammation, decidual hemorrhage,
and pathologic distension of the uterus (17). A variety of potential predictors of PD have
been researched in each of the pathogenic processes: maternal and fetal stress (18-19),
corticotropin-releasing hormone (CRH) (20) and increased estrogen concentration (21)
for the hypothesis of HPA activation; activation of the cytokine network leading to
increased uterotonin expression (22) and increased placental and membrane apoptosis (23)
for the inflammation hypothesis; thrombin and the genesis of Preterm Premature Rupture

of Membrane (PPROM) for the hypothesis of decidual hemorrhage (24-25); and

expression of myometrial gap junctions (connexin 43) (26) and increased myosin light
chain kinase (MLCK) activity (27) for the hypothesis of uterine distention. It is suggested
that each of these four pathologic processes may have unique biochemical and
biophysical pathway and distinct epidemiological profiles. However, they converge on a
final common mechanism, which lead to cervical dilation, membrane rupture and uterine

contraction.

Major risk factors for preterm birth include socioeconomic status, ethnicity, age,
multiple pregnancy, past reproductive history, substance misuse and infection. Social
disadvantage, usually evaluated by occupation, income and education level, is suggested
to be associated with an increased risk of PD (14, 8, 29-30). Possible explanations are
worse nutritional status, higher frequency of substance misuse and genital-tract infection,
limited access to qualified antenatal care, physically demanding work, and more frequent
cigarette smoking. Though part of the racial difference can be accounted for by
socioeconomic status, other factors such as maternal body size, customs, behaviors, age
distribution, exposure to racism and discrimination and some neighborhood level factors
may contribute to the PD gap attributed to race. Results from a study indicate that the
incidence of PD among whites is lowest in women aged 20-24 for the first birth, and 25-
29 for subsequent births (7). While among African-American women, the lowest rates of
PD for both first birth and subsequent births arise between 25 and 29 years of age (7).
Multiple pregnancy accounts for l2%-27% of all preterm births (2, 31). Increased use of
assisted conception technologies and increased intervention to deliver twins early in third
trimester resulted in increased PD rate due to multiple pregnancy over the past 20 years

(32-33). A previous history of preterm birth or delivery of a LBW infant is an important

risk factor for PD (34-35). Second (but not first) trimester pregnancy loss (36) and in-
vitro fertilization (IVF) and gamete intrafallopian transfer (GIFT) (37-39) have also been
linked with an increased risk of PD. Substance misuse, such as cocaine consumption
during pregnancy has been linked to increased risk of PD. However, this association
maybe confounded by lifestyle and other factors, e. g., high alcohol intake, cigarette
smoking and congenital infections ( 14, 40-41). Systemic maternal infections may
increase the risk of PD (42), and the association between spontaneous PD and genital-
tract infection, particularly intrauterine infections has been extensively researched.
However, the exact frequency of PD related to intrauterine infection, and the specific

organisms involved, are poorly understood.

Current therapeutic interventions to prevent PD include tocolytics, glucocorticoids,
antibodies, cervical cerclage and enhanced social support. To date, no method has been
shown to be associated with lowering rates of PD (43). Four types of nutritional
interventions have been widely studied: counseling, protein supplementation, caloric
supplementation and vitamin and or mineral supplementation. No evidence has been
provided that nutritional counseling changes diet habits of pregnant women, not to
mention the pregnancy outcomes (44). The provision of protein supplementation was
shown to be consistently associated with adverse outcomes (45). Also no compelling
evidence has been found that caloric supplementation is related to a reduction in preterm
births in developed countries (44). Studies on the supplementation of iron, zinc, folate
and combined vitamin have produced conflicting results. Up to now, no effective

intervention has been found in preventing PD.

Fish Consumption and Adverse Pregnancy Outcomes:

High consumption of fish and seafood, rich in long-chain omega-3 fatty acids (52-3 FA)
eicosapentaenoic (EPA, 20:5 0-3) and docosahexaenoic (DHA, 22:6 9-3) has been
found to explain the reduced rate of cardiovascular diseases in Greenland Inunits (47-48).
This fact has inspired researchers to explore the effect of fish oil on other problems of
human health. The risk of many adverse pregnancy outcomes, particularly PD, LBW,
Intrauterine Growth Retardation (IUGR), preeclampsia and Pregnancy Induced
Hypertension (PIH) have been assessed in relation to the consumption of fish and seafood

or the intake of fish oil supplements.

There is some biological basis for the hypothesis that fish consumption may reduce
the risk of PD because prostaglandins can play a role in the onset of labor in humans and
other primates (49). Prostaglandin E2 (PGE2) and Prostaglandin F2a (PGF2a), derived
from omega-6 fatty acids (52-6 FA), bring about contraction of myometrium, and cervical
ripening. These prostaglandins can trigger the initiation of labor throughout pregnancy,
and a small increase in these prostaglandins is detected in the amniotic fluid and plasma
right before the onset of parturition (50). Prostacyclin (PGI2) is found to have the
opposite effect, inhibition of myometrial contractility (51). Long chain 9-3 FA can delay
the initiation of parturition by down-regulating the formation of PGE2 and PGF2a,
promoting the synthesis of PG12 and PGI3 and thus leading to a more relaxed

myometrium.

There is an extensive biologic basis for the fish oil to decrease the risk of
preeclampsia, eclampsia and PIH. The exact cause of PIH is unclear, but a possible
explanation is that the impaired matemal-placental circulation is aggravated by an

interrupted balance in the production of vasoactive prostaglandin (thromboxane and PGI2)

during pregnancy (57). PGIZ, a potent vasodilator and platelet aggregation inhibitor
produced by endothelium, predominates over thromboxane A2, a potent vasoconstrictor
and platelet aggregator produced by platelets in normal pregnancy. In PIH, the ratio of
PGI2 to thromboxane A2 is reversed (58-59) and thus induces vasoconstriction of small
arteries and the activation of platelets (60). EPA (9-3 FA) derived thromboxane A3 is
physically less active than Arachidonic Acid (AA) (9—6 FA) derived thromboxane A2,
while Prostaciclina (PGI3) is equipotent with its AA derived counterpart PGIZ,

suggesting that fish oil might be effective in preventing preeclampsia and PIH (61).

The long chain 52-3 FA may increase birth weight (BW) either by prolonging
pregnancy (54) or by enhancing fetal growth rate (55-56). Besides delaying initiation of
spontaneous delivery resulting from altered balance between prostaglandin, 9-3 FA
lower thromboxane/prostacyclin ratio (55) and decrease blood viscosity (56) through the
similar pathway to prevent preeclampsia, resulting in improved placental blood flow and
increased fetal growth rate. The hypothesis that fish oil may reduce the risk of IUGR is

also based on the mechanism for 9-3 FA to increase fetal growth rate.

Most of epidemiological studies evaluating fish consumption or fish oil supplement
intake in relation to the risk of pregnancy outcomes were performed in Northern
European countries where the fish consumption levels are generally higher than the levels
in the United States and these studies produced inconsistent results. Table 1 is a brief
review of previous literature regarding the association between consumption of fish,
seafood or 0-3 FA supplements and the risk of adverse pregnancy outcomes, with an
emphasis on PD. These studies were collected by searching PUBMED using keywords

39 ‘6

such as “fish” or “omega-3 fatty acids” together with “preterm delivery , preterm birth”,

“birth weight”, “gestation” or “pregnancy”, and all the studies evaluating either

gestational age or BW as outcome variables were selected.

 

05:22:: 02 .m

.2303

 

 

 

 

 

 

 

 

9.8m 855328 Saws :5 boon Eon .m
8 Ho: :5 £888 8 20¢ .vdm «833 8 N55
0:205? 38:25:23 83-30: 822352va .2»
SHOW? 3ch 5:8 .m> .325 .N 36828 382:2 .m Sm—SEm 80 8mm:
9:225: 38:8823 83-28: .6033: mo 08,—. .N 8082588 :0 axes—8&2
hHON .m> BHmm 8:8 .m> .325 A <2 .83:— Mo 28:0 ._ :8 mo 5885 N_nZ .moao -Eswam
$8 a: fie as
moon 632:8 N .m> 855225 .v
osoN :5 mN .m>
88o ӣ888 N .m> 835225 .m
osoN :5 2 .m> 25an Lo Ems? .v £83830: 2
sec 202:8 N .m> 880325 .N Egan: 8.22m .m BEEP—28
8:: cm: as «sea 53028 33 .m 58-8:
:m «in :5 2.25 5.5 5233 <2 woo—n E8 5:5 :2 :38? .mng .393 No:
88:83: 0830: H§o£:wmm ._ :5 3828 8 <29 ._ 80¢ mwmm 388m 83: :Eom
>330: >258
um am: 86:23:? 5c m mo ”38
mm.“ .36 HO osmm .wodnmc .m 1822280 .mowN 3265 5:835 .m 83:823.
N_ .N .vnd 5 $3 .cNgumO .N 8:88.288 832882 .338 .wowuz
mod .wmd 5 gen: .vmdnmo .2 8 cm: 830% 05:28:25 .N :6 EN “3.: 683 So:
”RE econ—Emoi 3:28:80 ._ @050: 8.205 A ovum—ESE 888m 88$ :85
once: 88a: 88 8m 28% 338853:
288% 8289880 $8880 \obnmomxm 823.88 88m

 

 

25.888 .255. a as; .836

 

.EE :0 ~53 3:203:89 .owa fi:o32mow
mag—o8 388:5 5:8on 2 :252 5 3:88:33 mo :mwhmc mo 83:82.88 :0 868m .8 2632 28823 A 2an

 

 

Cdnmv 5 82:00
E 2830— 08 520% 828328

5 8:2: 2 022 :82 .m
.5 82:00
:02 8283005 5 30300: AEN

._ 8 880 m S: .802: 388
8 2832 0:0 500% 828328

 

 

 

 

 

 

 

8 282w:— 2 28> :82 .N 8080858
.5 82:00 :05 0: ”N 82:0U
828328 8 3080— 98 65 HO =0
28 86 3. .8802: 688 8:930 .6 £22 58 .m 26 2 8.80 mmmuz
8 2830— 28 5:08 828328 :0228 $2.85 5303285 .N >20th ”<5 8082 A3:
8 28:2: 2 02? :82 ._ .x0m umfiN 0mm 8:02.280 5 m-: 8028325 .2800: $00: 80.0
8:838
8 822:9 .m
0.2:002mv mNNnZ
Emmoagm .0 .88
88 582022 8 8:
520% 82:00 08 828328 000505 88:88 28th ”QED 880—080
80802 2.88208 mg.— 88 45.5 .8 0—2: Ame: $00:
080220 025 8 8 00:20.28 02 <2 :0: 0:0 2.88205 5 w No.3 8882 8 :0803 083:0
0320 8.8: 02: 0.00 0.230 8885.8
2885 20:85:00 880220 880095 80208505 >005

 

 

"3582028 .555 a 28... .880

 

.55 :0 MOE 2:203:89 .0w0 8:0228m
88:88 880080 88:88 02 :0222 8 2:080:53. :0 202202 .8 82582800 :0 8880 8 3032 28825 ._ 085.

 

 

oocououbu 02 .m
082356 oz 4»

3:338

 

 

 

 

 

 

 

388.330 :55 a. .30". .m
05de voozfie asp—m EDP—adorn. .m Emmokétmm xv Aén US“
Gaodémo fiaovvaozeé gm 5% _ .o
§m.wm 3268 3on 8953; .N >28:on .m 88858 .02 ng
:Sduemo <2 8.833%: .N SE 23 .83; so:
g fl . _ m BBB. 95.» 52:35 ._ «633035 4 Em .ssfiS .338: 88: :35
EEV
“8 9a Eoc £39:
58 mo 536388
:83an 5.8 3. £8 55 .0
A8: EARS 0m 6-5m-.. .m 6,»:
385an E .8 3. >83 83 fig:
Age 922.: Cu “9E? .v .58 .8 833858 b8
:odvmv 050m .w 86m SE .m 320 am ”Hm
:qovnc Haven ”2 5 .m 8:328 coon .v =o €82
Agave Hmvo< Amooodve :32» 38%: .m $2 ”0....
Even Agave 058 .N Emmoansm .m =o 5d amass mnuz an
:odvnc pm 8 0m $599.. ._ <2 cm“ 8:330 ._ $2 ”0m .98 £33 83: :85
039.5 .25 fix 86 ”Beam cagofiiv
33mg“ who—vasomconv gag—go 3.2.50me EDGE—anon“ bun—am

 

 

"3855.5 335 a 23:. .325

 

 

.35 .5 ~52 .EEPEEB .owa Ecosflmow
955—05 388:6 >28:on 8 5:22 5 3:080:33. mo :mcEmm “o coca—cacao :0 8:56 Mo 3032 22885 ._ £an

10

 

0:12 6 $3
.w SN 3 832% ”$6 .N
Na .3 8 $3

 

3% n. B 3322: ”$5 ._ 220
“mo—:03: mo omega 2: 8 coca—8 “.0 x8 .Ewfix @003 30% can
5 «E 823 no 208065 oi Ecocflmow EEom REES:
.386an $2.3m £8336 58 E 2”:
v08 0538 go cocmfismcoa .xocmcwoa 36m bamm .N
REES: 55 @3385 «Em coo—n macaw wagon; Ewfi? :35an .m 5mm .NEHZ
Eco .moEooSo 28 conga—Sou .EwmoB HESS: Emmoaafim .N 53, 23.: ES: €83 A8: 28$
God 5053 53683 02 .Ewfi: 352a: own 38:33.0 ._ mo moconcofi ._ 3&8: 50.8520
“52928
a was 3508
05 55023
533260

mm; “6 Swan. .EwBB
.Sd 6 $3 .3 . Tame ac .L .v “5:33-23
3% .Ewfi: 5:3

 

 

 

 

 

 

 

 

.33 6 $3 .3335 53 L .N am“ 353:
cod dorm—Sacco xocmcwoa

a: 6 $3 .335 ”=3 a ._ 3:8:w ~62 .m 2:3 a Bang afiwnz
”8:058 ma coca—55:8 sac .wfixofim Emmon—tfi 33 .m mo 5:955:00 683 Go:
“6 63 2:53 an 32% 2: mam: £3. Eu”. beset E205 ._ us 223 Eamon £83 :85
cacao 56$ “.5. 3% 295m cantoeiv
333m 33:28:00 38850 $.58me Boss—smon— 33m

"883m :25

 

.EE .5 MOE .Ewfizfitfi .owa Ecocfimom
mafia—oi mos—8:5 >23:on 3 .5322 5 2580395 :0 5525 .0 8:33:00 co 362.9, .«0 2632 23835 A 053.

ll

 

 

Coodvmv 08:830me
:82 8 .5805: 80 hwd .m

 

 

 

 

 

 

 

 

 

3.9.0 ”583»?
5074 8 8080:: 80 mmA .v .Ew:0_ ovAuZ
Codvmv 00:88:80 000: .m 800:0)—
mfififiow? :33: :A 583 w ova .m £80— 585 .v 8:02
€80.03 023.023 23.638: .m a 8082:
8:03:80 823 :05 852 8003 528000 80:00
888%? 8 6:05 9?: 9m .N 353080 0:: m 80:
:0803 888%? w:08.m 083.0 308% .500— 00 08:0 ham .N mgr—Sow?
E Cooodvn: «ED 0:0 «Em 530A :20: 0:: 5:00 0003 0:00 0:0 8:9:
05208.90 .0w0 38808 0003 9:080? 8::on 88.8%? .m 80:38 803
6-5 86: 2208A 530:0 A awn :08330 -:0: 8 £05— <m A wfiom A :02 $00: aged
90.0"“:
wow 00 Ewfiaatfi :_ 088000 a 3
008008000: 83 ARE: «0 8:88 wEUAoEm .m
0003 8 80208 :0 Abomus 0005 8&8 0::
80808020 303 8 3:0 .N 38808 5 A05—
.Cooodvm .w8x08m 8:808 130% .N
.m A .oumv “008 .000— 8 8:38 daew :88 8:03 02 W2
53, 0030800 0:03 8:888:00 daew :0 20:3 .38 00 .0030 $5: 6003
8:808 0003 38802 A Eva—:00; uswA03nEm 8:083:00 A A3800: 8003805
809.5 88$ :02 08m 0383 Egan—>383
333m 80:88:00 380080 \08monxm 8030—008 35%
"mom—000m 80:00

 

 

.35 :0 MOE 4:803:80 6w: 3:0ufimom
@6208 880080 5:838 8 8:30: :_ 3:082:03 :0 55:90 00 8:83:00 :0 888m 00 3030: 28885 A 035.

12

 

$08830: mafia
003800 0:00 00:
0o 3.: 8a: 83
0:0 AMA: 0580.0
-:050_b0_3:%0:m
.00 2

.28 88.0“: .03 0203002080:

"0008 0: :A 0058000 0:0: E A000: 08R
08%.: 05 5A3 -5 6-5 << 0: <nA oNuZ .0000 8A 3

000030000 003 0w0 3:030:80 <2 0w0 :0000000 m: .00 0:0: 05. 308003 83: :85

 

000005 00000n~ 07AM 08m 0380m A00090050<V
3.30m 80:88:00 \080080 \0Smomxm 20:03:00 30%

 

 

 

 

 

 

 

"8:030 :25

 

.EE :0 MOE .Ewfiantfi .0w0 3:0:0000w
wig—0E 0080080 0800:0008 00 :0003: :_ 08080—0020 :0 5:220 00 5:083:00 :0 00680 hA0 3030: 0:80:00: A 030:.

13

 

 

 

 

 

 

 

 

 

 

14

 

.0586 868
85088 $389
.Ewfi? 285005 b0>=0n 800345 .m
.u8oficwa 0003 _0>0_ 00032 05 8:888 E0508 E0005 .v
53 £90. 00805 Bama— w£8m600 .Ewfi: 88088 MOB .m 55280:” wESc .NOSHZ
8 v0um:€8¢0w:=8=oo 05 .358; @5380 IE .N 30:0 0588 £082 AS 3 Ago:
mo 0:0: 8080050 0% 05 :8 Sm 85082 8698—0005 ._ 00 5:955:00 sown—such” E00803“
880% Ho: .8
080558 8: .m
Coodunc
8808 880953 0236 98000
05 5 :805ch 50:92:38
nmm 00508 53, to _0>0_ 8
“0:00 .Amoodnmv 38% 5:900:88
N. 3808 0:20.006 bucwocwamm .N
@006an
$on EEO—Eon 000m0u 2808 mac—9:0
05 E 280556 .soumfinmcoo 98 8:88 3E0? 880an .v
:0: Esta 5:5 to _0>0_ 8 8:88 .098 :28020 58505 .m $28:qu mom—HZ
0:00 Amodnmv anew 5395.200 .533 £563 Ema—0. :Em .N wctsc @8800 £00.89 2 fl 3
h 3808 E08006 baamoficwfi A .Esz 85082 Emfiafibm ._ 0o cosqfismcou conflaom Sam: :85
08005 082$ via 3% 03:8m $898533
3300M 3095850 \080830 \0Smomxm Eats—anon 32m
"00:55 3.555.880

 

.IE 00 M03 3:303:85 .0? 8,5388%
mics—08 00:58.8 58.80:” 8 :28—2 5 3.0080393, no smcEmc h«o 8:900:28 :0 00630 no 30300 0580005 ._ 058,—.

 

 

 

 

 

 

 

 

08880::
wES:
8083380
0:03 0:80 :8 40.50.08: <n~
0:: 5:3 :5 :00803 83 m<m 5:3 .388 8:88 :80 :0 m_0>0_ 0:8
8330008 05; .0008": w:088 :0: 5:80 .0? .0033: :0 080
:5 80.0"“: 5803 08850 .300: .5003 N 55:3 859%
w:088 0:30: 83 0:8: 0-: 88380:: 0300250 .mmuz .008: Am: C
\m: :33 :286008 :80E:w_m 88082 0mm 8:28:30 8 0:8: c-:\m-: 8:800: :00: :0:—O
08005 :0:08n_ 0—02 08m 0:08am A08Q:05:<V
3800”: 30086800 \080080 0:30:71": 828800.:— 355

 

 

"00:55 83.80-0080

 

.EE :0 ”:02 3:303:83 .0w8 8:28:00w
wag—0:: 0080080 30880:: 0: :28—0: :_ 3:082:95 :0 5:22.: :0 5:083:00 :0 00620 :0 3030: 0:08:35 ._ 038:.

15

 

Seven clinical trials, five cohort studies and three case—control studies have been
included in Table 1. All of the 13 studies performed on humans were conducted in
European countries with high consumption levels of fish or other seafood. Three
associations were evaluated: consumption of fish, seafood or 9-3 FA supplements in
relation to the risk of PD or other pregnancy outcomes; consumption of fish, other
seafood or 9-3 FA supplements in relation to FA levels in blood; and FA levels in blood
in relation to the risk of PD or other pregnancy outcomes. Thus, a hypothesis that fish
consumption decreases the risk of PD or other adverse pregnancy outcomes by increasing

blood FA levels could be tested by these studies.

Among the five clinical trials in humans (101-105), four (101- 102,104- 105) detected a
significantly decreased risk of PD and/or increased gestational age in the intervention
group given supplements containing fish oil or 9-3 FA during pregnancy. Both of the
animal experiments (52-53) found increased gestational age among animals treated with
fish oil supplements, while only in one study (52) the increase was statistically significant,
because a statistical test was not performed in the other one (53). In three out of the four
cohort studies (106,109-110) evaluating gestational age or the risk of PD as the outcome
variable, a significantly decreased risk of PD and elongated gestational age were related
to higher levels of fish consumption or enhanced long chain 52-3 FA levels in
maternal/cord blood. Two of the three case-control studies (111-112) did not observe the
association between gestational age and consumption of fish and seafood. Another case-
control study (113) reported increased gestational age in relation to increased 9-3/ 9-6
ratio among Denmark but not Faroe Island women, who consumed higher levels of

seafood. Both of the two studies evaluating fish consumption on maternal or cord blood

16

levels of FA (107, 109) were cohort studies, and they found either a significant increase
in blood 9-3 FA levels among women with higher levels of fish (107) or higher 9-6 FA

(109) among women with lower fish consumption levels.

The results from cohort studies and clinical trials were compatible with the hypothesis
that intake of fish oil during pregnancy, either in the form of supplements or fish meals,
elevates blood levels of 9-3 FA and elongates the pregnancy duration and reduce the risk
of PD. Two studies by Olsen et al (111, 113) observed a “ceiling effect” such that above
a certain high fish consumption level, enhanced intake of fish or seafood would not result
in a further increase in gestation age. This phenomenon may be explained by a limited
capacity for human gastro-intestinal system to digest and absorb fish oil or a balance of
P68 and PGI production, which is unchangeable by further 9-3 FA intake after reaching
a very high level. In addition, nutritional variable is likely to have misclassification error
when used as a surrogate of 9-3 FA intake, and this misclassification is likely to be
greater at higher range of scales (e. g. Women are more likely to accurately remember
whether they ate 1-2 fish meals than to remember whether they ate 10-15 fish meals
during the first half of pregnancy) Thus, the dose-dependent trend would tend to flatten
artificially at the highest range of fish consumption. Further studies are needed to
measure and define this “threshold level” in order to guide pregnant women to eat fish in

appropriate amount considering both the nutritional and potentially toxic contents in fish.

Major outcomes concerning fetal growth including IUGR, BW, birth length, placental
weight and head circumference were assessed in 13 studies (101-109, 52-53, 111-112).
Seven of the ten studies (102, 104, 52, 106, 107, 109, 111) evaluating BW observed

significantly increased BW or reduced risk of associated with enhanced maternal/cord

l7

blood 9-3 FA levels or intake of fish, marine food or fish oil supplements. Three studies
(101, 106, 112) assessed the effect of fish or fish oil supplement intake on the risk of
IUGR and none of them reported any significant findings. Four studies (101, 103, 105,

1 12) researched the risk of PIH or preeclampsia and only one (105) found a weak
protective effect from EPA and DHA intake. In summary, BW is the fetal growth-related
factor most consistently related to intake of fish or fish oil supplement. Nonetheless,
since BW increases almost linearly with gestational age during the later period of
pregnancy (61), whether the fish consumption-BW association is confounded by
gestational age needs further clarification, especially because only three ( 104, 107, 109)

of current 13 studies controlled for gestational age in the analysis.

Fish Consumption, Mercufl Exposure and Fetal Development

Mercury is a naturally existing element that is found in air, water and soil, and it is
generally released to the natural environment in an inorganic form by both natural and
anthropogenic sources (62-65). There is much concern about the potential for human
intoxication because of exposure to mercury in foodstuffs particularly fish. The most
serious toxicities to mercury arise from methylmercury residues in food (114). During
methylation inorganic mercury is converted into methylmercury by microbial action,
primarily in sediments of fresh and ocean water. Methylmercury readily enters the
aquatic food chain and bioaccumulates in predatory fish such as swordfish, pike and
ocean tuna (66). Larger and more long-lived fish tend to contain more methylmercury
and the contamination can be significant, for example, the total mercury in the edible

tissue of shark and swordfish can average as high as 1200 ug/kg (64). Consumption of

18

fish and marine food, though not the only source of methylmercury, is a very important
exposure to methylmercury for humans, especially for those populations near seawater
and rely on fish as main food source. Methylmercury is rapidly and extensively absorbed
through the gastrointestinal tract, and it is distributed throughout the body and easily

penetrates the blood-brain and placental barriers in humans and animals (66).

The critical target for methylmercury toxicity is the nervous system (66) and the
developing fetus may be at particular risk from the exposure to methylmercury. Studies
about epidemic poisonings in Japan (67) and Iraq (68) showed that maternal toxicity may
or may not be present during pregnancy for those offspring exhibiting adverse effects.
Offspring exposed to methylmercury in utero showed a variety of neurological
abnormalities including delayed walking, delayed onset of talking, cerebral palsy, altered
muscle tone and deep tendon reflexes, and reduced neurological test scores (67,68). The
pathological mechanism for the methylmercury-induced developmental neurotoxicity is
unknown, but several hypotheses were raised: changes in intracellular cytoskeletal
structure (69-71), oxidative stress (72-74), alternations to membrane function and signal
transduction (75), decreased protein production (76), and changes in neurotransmission
(77).

Besides the adverse effect on fetal neurodevelopment, animal experiments also showed
reduced neonatal weight gain among animals treated with high does of methylmercury (78-
80). To explore a possible intrauterine growth inhibiting effect from mercury for human,
Sikosky et al (81) studied the inorganic mercury levels in newborn hair samples and reported a
statistically significant inverse correlation between mercury concentration in fetal hair and

BW. However, no the analyses lacked other covariates and the result may be explained by

19

confounding. A study in Greenland reported high maternal and offspring methylmercury

blood concentrations associated with low mean BW (82). Other studies detected no
association between BW and methylmercury levels in hair or blood (83-85). Because BW and
gestational age are correlated and most previous studies did not control for gestational age (81,
82, 84, 85), the detected BW-mercury association can at least partially be due to the
confounding from gestational age.

The exact pathological mechanism linking gestation age with mercury intake is unknown,
but if such relationship exists, the “threshold effect” detected in the fish consumption-
gestational age association could be explained by the accumulated toxicity from mercury
when the fish consumption level is very high. Studies investigating the influence of mercury
level on gestational age are few and none of them observed significant results (82, 83, 87).
The only study assessing the risk of PD (87) evaluated the difference in the incidence of PD
between the group occupationally exposed to 0.06-0.l mg/m3 of metallic mercury and a non-
exposed group, and did not report a significant result.

Table 2 is a literature review of studies investigating BW and gestational age in relation to
exposure to mercury. Except the study measuring metallic mercury exposure in relation to the
risk of PD (87), all the other studies evaluated methylmercury in blood or hair samples. No
convincing evidence was provided for gestational age or BW to be related to maternal or fetal
mercury level in blood or hair. Noticeably, five of the six cohort studies have a small sample
size (81-84, 86) and they may not have had enough power to detect subtle but important
differences. In addition, various kinds of samples including maternal and cord blood, maternal
hair and infant hair, were used to quantify mercury levels and this also may have contributed

to the inconsistency of the results concerning levels of mercury exposure in association with

20

gestational age at delivery or BW. The measurement of mercury exposure will be discussed in
more details in the next section. Two studies evaluated the association between intake of
seafood and mercury levels (82, 85) and both found marine food consumption to be
significantly related to elevated maternal and cord blood levels of methylmercury. All three
studies assessing the correlation between maternal and fetal mercury levels (81, 82, 86)
concluded that fetal mercury levels in blood or hair are significantly associated with maternal
levels. This result supports the previous reports that methylmercury can easily cross placental

barriers in humans and animals (66).

21

 

80880—000

 

 

 

 

 

 

 

 

 

 

:88: 00 0:0808 8 0:8 .:008:m0m
8008000080 388083502 :0 3:808 a
88088 503:0: 8:80:88: 0:: 80>00 u_0>0_ :m:uZ
8800883 38003083 0008080080 080m .N 3:00:08 :8: .008: 3%:
8 80:: m: 80:83:: 0: E .N-: 08 88082 ::w:03::::m .: 808 80802 8:33: 300:: :882
0.0.08: 08 .8 000383
8 080:000 3: 0088080008
83 300: .:0 3:00:08
000:: 8 08808 :8 Agmné
80088080 303 8 380
.Sooodvm .mmdumv
008.0800 3830—0 0:03 000:: $0888
0:00 0:8 000:: 88088 8 m_0>0_ 080—083 0:088 00
3:00:02 dooodvm .mfionmv 0:8 80:83:00 .m
808 800: :0 008:0: :03 .0008 80:: .0003 00: :n2 3me
008—0800 0:03 80880800 .0008 88088080 0: .008: 600:
3:00:08 000:: 88082 00020000 8803885 3:00: 3:00:02 .: 80303 0088805
.:00:0w 8:00 88:88:00
80:08:88: .880: 88088 8003 8808?: .m 000803 $7.2
080300 0:: 00 38 80.08 :0: 000883 280358: .N 00 80880800 .008: 3me £008
00 m_0>0_ 3:00:08 880m .m-: 88088 .3880 08 8:008:80 ._ 80:03 0:00 8:030: 80.30880
0835 8:08": 07.00 08m 038% 0:898:83
8:30”: 800000800 \080080 8:30:30 8008:0003 308m
88005 80:00

 

 

08:03:80 :0 08 8:008:38 0: :0088: E £08: 38008 8808088 :0 8003 30030:: 00 3030: 0:08:35 .N 008,—.

22

 

80:038.:

:008 0:: 008008 :0:

0:0 000: 08:08 :0 08:8 0::.
dodunc :0::08 0:0 Sedan:
w8:::::0 0:: :0 3:00:08

000:: 0:8: :83 008888
83 880388: :008 33 .N
0008800000 3:00:08 000::
0:: 3: :0 000: 800: :0 :808
:0 :0:80: :0::0 3: 00:80:00

:0: :: 0m0 8:000:80 .: .w8:::::0 :w8:::::0

886": .36": :.:0:80::: 0:: :0 x0: 0:0 80:88

008088 0:03 ::0>0: 0:0 88:08: :0 88:88:00

3:00:08 000:: 888088083: 0::0:0m:0 3:00:083::08
.Aooodumv 000:: 8:0:0088088 008000: 000:: 0:0 .000: cbmuz :me
8 ::0>0: 3:00:08 8:8: 0: :0 :30 H 000:: ::w:03::::m :0. 08:08 :0 08:8 .008: 800::
0088: :: 000: 08:08 :0 00:08: :8: :.:0::0:>: 0w0 8:000:80 .: 3:0:003 0w0:0>< 8:880: w:0::0:0n:

 

.::0:::00 :0 08:: 8
:0:: :88: 30000088: :0: 83
:00:w 00890 0:::0 :w8:::::0 8
880388: 30: :0 00:00:00: .N
.:0>0: 800088: 0 :000: :0:

0:0 @008 :0:800 0:0 3:00:08 :0

 

 

 

 

 

 

 

888 00-00.: 0: 00890 :08» ::0>0: 33u7:
0:: 8038: 3:03:00 80:08: 880388: 30.: .N 3:00:08 0:888 .008: Gm:
:0 00:00:08 8 00:80:00 .: <7: 3:03:00 808:: .: 0: 0:08:08: :000:0000 :33: 0":
080:5 8:00”: ::0: 08m 00083 08900880.:
3:08”: ::00:00::00 880080 08898: 8000080: 308m
u8:005 80:00

 

 

880358: :0 0w0 8:000:8w 0: :000:0: 8 ::0>0: 3:0008 88:80:08 :0 8:08: 8030:: :0 3030: 0:08:08: .N 030:.

23

 

 

.mcozmbcoocoo
332 53, 95%: Va

wcoEm Ewfiafihn 532 90m
< ”EwfianEn m: “Em 5830:
a .3 an: 05 E cocfiaooaoQ
Daub... 52,33 scam—oboe

in: 3280:

 

 

 

 

 

 

 

 

8.02: Enumfiwfi flea—oboe can 388%: SLHZ
£33355 203 has .8580: E 32.5 205— .393 28 Gme
tea has .3059: 5 ASE. <2 fiwBBnEm >589: :38. 3383 3&8:
9835 885m fig 35 295m GEQSEHZV
$38M $029850 $8850 \ohsmoaxm anflsmom 33m
”835m $8.30

 

 

.Ewmoafihn 8 own 38:2me 8 sous—2 5 £05. @585: .auoméofie :0 863% $6395 mo 3032 22825 .N £an

24

Measurement at ExQFosure to 9—3 FA and Mercufl

Questionnaires and interviews concerning frequency of meals of fish or marine food during
pregnancy are most often used to measure fish consumption. However, the estimate of 9-3 FA
intake by these methods is generally imprecise because of the portion size, distribution of fish
species in meals, and food nutrient content are only approximations to the true value (106).
Intake of supplements of fish oil or 9-3 FA is sometimes taken into account as part of the
sources of 9-3 FA, but this part of 9-3 FA intake is difficult to scale to levels of fish
consumption in analysis. Classification according to types of fish and marine food is
conducted in some but not all studies. Fish and sea mammals differ with respect to the
chemical composition of their 9-3 FA content (96) and considerable variation has been shown
among fish living at different latitudes (97). These differences of 9-3 FA in chemical
composition may result in varied biological activities. Levels of fish or seafood consumption
are usually categorized according to the range of the consumption level in each study
population. Thus, cut points for the high or low levels of fish consumption in different studies
vary and the results are difficult compare across populations. Furthermore, questionnaires and
interviews concerning fish consumption may increase women’s awareness of their intake of
fish, and they alter their intake of fish in the duration of the pregnancy.

Direct measurement of biomarkers of 9-3 FA in serum or erythrocytes is another approach
to evaluating the exposure to 9-3 FA. The most frequently used indicators include EPA, DHA
and the ratio of 9-3 FA to AA (9-6) (9-3/ 9-6 ratio). Fatty acids concentrations in serum and
erythrocytes are a result of dynamic interaction between absorption, degradation and for some
of them, catabolism, and changes in transplacental passage (98-99). Levels of FA in cord

serum seem to be stable during the last few weeks of normal gestation (99-100), though

25

physiological changes during pregnancy and the time of sampling and measurement should be
taken into account in analysis.

Sources of mercury exposure can be categorized into environmental, industrial and
agricultural (115). Mercury has a relatively high vapor pressure and the air over mercury and
its ore deposits generally contains enhanced levels of mercury (115). The combustion of fossil
fuels also releases traces of mercury into the atmosphere and these add to the much larger
quantities present in the atmosphere than the natural vaporization processes from the earth’s
surface (116). Considerable evidence has shown that manmade pollution of rivers, lakes and
estuaries has increased the mercury levels in fish, but such pollution has not contributed
significantly to the mercury levels in ocean fish (115). Individual lakes or watercourses can
also be contaminated by underlying mineral deposits containing mercury, which leach into the
water under natural circumstances (117). The direct pollution of water by industrial sources is
likely to affect fish more than other foods (115). The industrial sources that cause the transfer
of mercury-containing wastes to water or mud in fishery areas constitute sources of most
direct concern of mercury contamination. The principle industrial sources include the chlor-
alkali industry using the mercury process, the pulp and paper industry and other industries
using mercury (118). Other industrial sources such as mining, smelting and refining of ores
are also significant but the effects are usually localized. The amount of mercury in the food
supply from normal agricultural use has been usually small compared with that from other
sources, but alkyl, alkoxyalkyl, aryl, and inorganic mercurial fungicides have been used for
seed dressing, as turf fungicides and in orchards (115). Although in recent years, the use of
alkyl and aryl mercurial fungicide has been restricted or forbidden in many countries,

accidents have occurred from the misuse of seeds treated with alkyl mercury compounds

26

(115). Other sources of mercury include pharmaceuticals containing either inorganic or
organic mercury compounds, and mercurials used as preservatives in cosmetics and toiletries
(115).

The most common biological samples analyzed for mercury are blood, urine, and scalp
hair. Blood, urine and exhaled air are common to assess occupational mercury exposure,
which is primarily elemental mercury (87). Blood and scalp hair are the primary indicators
used to assess methylmercury exposure (87). Blood is a good indicator medium for estimating
methylmercury because methylmercury freely distributes throughout the body. However,
because an individual’s intake may vary, blood levels may not necessary reflect mercury
intake over time (88-89). Methylmercury is incorporated into scalp hair at the hair follicle in
proportion to its content in blood (90). Once incorporated into hair, the methylmercury is
stable, and therefore, gives a longitudinal history of blood methylmercury levels (90).
However, the analysis of hair mercury levels may be confounded by absorption of mercury
vapor onto the hair strands (91). The hair-to-blood ratio in humans has been estimated as
approximately 250:1 expressed as ug Hg/g hair to mg Hg/L blood, but some difficulties in
measurements, interindividual variation in body burden and metabolism, differences in hair
growth rates, and variations in fresh and saltwater fish intake have led to varying estimates
(92-93).

The most common methods used to determine mercury levels in blood, urine and hair of
humans and animals include atomic absorption spectrometry (AAS), neuron activation and
analysis (NAA), X-ray fluorescence (XRF) and gas chromatography (GC). Among these
methods, only GC is able to distinguish methylmercury from the total mercury content (94-

95). The Magos and Clarkson method of AAS can estimate methylmercury by subtracting the

27

inorganic mercury content from the total mercury. However, hair contains primarily
methylmercury from consumption of fish or other aquatic foods and thus the content of
inorganic mercury is very low or absent, so the measurement of hair methylmercury levels
should not be largely influenced if the method cannot distinguish methylmercury from

inorganic mercury.

28

MATERNAL FISH CONSUMPTION IN RELATION TO PRETERM

DELIVERY

Fei Xue, Claudia Holzman, Larry Fischer, Hossein Rahbar

Introduction:

PD, which occurs in l 1% of U.S. birth, continues to be a leading cause of infant
mortality and morbidity and excess healthcare costs. Lowering the rate of PD has been an
elusive goal, in part because there are many clues regarding the underlying causes of PD,
but few answers. Wide variations in PD rates across countries and between subgroups
within the US. have prompted investigators to consider maternal diet as one potential
factor affecting the risk of delivering prematurely. Though total caloric intake in
developed countries (131), and diet supplementation (132) have not been associated with
lower PD rates, it is possible that components of the diet, such as vitamins, minerals, or
essential fatty acids may be protective for certain pathways leading to PD.

Researchers have hypothesized that intake of high levels of 9-3 FA may be protective
for preterm birth because 9-3 FA inhibit the synthesis of 9-6 eicosanoids from AA (1 12-
1 13), a 20-carbon unsaturated fatty acid produced from membrane phospholipids. These
9-6 ecosanoid derivatives are thought to play a role in the initiation and completion of
term and preterm labor (1 14). Fish and marine food are the main source of 9-3 FA in
many human populations. The reduced rates of LBW in fish-eating Northern European
populations, such as those in Faroe Islands, offer ecological support for the protective
effects of 9-3 FA on the risk of delivering LBW babies, many of whom are preterm (54,

124).

29

Studies of fish consumption during pregnancy and risk of PD are conducted mainly in
regions near seawater with high levels of fish intake. These studies have produced mixed
results, and do not provide insights into populations with lower levels of fish intake. In
some regions, it may also be important to weigh the potentially harmful effects posed by
fish contaminants (e. g. mercury) against the potential benefits of fish consumption as a
source of 9-3 FA.

In this prospective study from five Michigan communities, we evaluate fish intake in
the first half of pregnancy and the risk of very preterm (<35 weeks) and moderately
preterm (35-36 weeks) delivery. Our study offers an opportunity to assess the variability
of fish intake in pregnancy in relation to maternal characteristics (e. g. age, ethnicity,
education, Medicaid insurance status, and community of residence) and link these
findings to the rates of PD. In this way we examine the impact of fish consumption on
length of gestation and the potential for varying levels of fish consumption to contribute

to ethnic and social class disparity in PD rate.
Methods:

Population:

The sample includes participants in the (Pregnancy Outcome and Community Health)
POUCH Study who were enrolled from September 8, 1998 through July 31, 2001during
the 17’h to 26th weeks of pregnancy (130). Participants were recruited from 52 clinics in
five Michigan communities 1, 2, 3, 4 and 5 (include five major cities and their
surrounding areas). Eligibility criteria are maternal age greater than 14 years,
participation in screening for maternal serum alpha-fetoprotein (MSAFP) levels between

15-20 weeks’ gestation, proficiency in English and singleton pregnancy with no known

30

congenital or chromosomal abnormalities at the time of recruitment. Women with a
history of diabetes prior to pregnancy were excluded. Approximately one third of all
eligible women participated in the POUCH Study. Among the 1,336 women enrolled in
this time period, ten were excluded (five lost to follow-up, five infants with congenital
anomalies identified at birth), leaving a total of 1,326 women.

Fish consumption:

Participants provided dietary information about their fish consumption during the
previous six months through an in-person interview at the time of enrollment. Since
women were enrolled at 16-26 weeks’ gestation, this period of recall approximately
coincided with the first half of pregnancy. Women were asked about frequency of
consumption of shellfish, canned fish, sport caught fish, purchased fish and other fish.
Consumption of types of fish was reported as number of meals per day, week, month or
six months and scaled to meals per six months.

Gestational Age at Delivery:

Gestational age at delivery was determined by the date of delivery and the gestational
age estimated at the time of MSAFP screening. This estimate was based on the date of
the first day of the last menstrual period (LMP), or on an early ultrasound (< 20 weeks)
estimated gestational age, the latter given preference when the two estimates disagreed by
more than 2 weeks. Very preterm was defined as delivery before 35 weeks’ gestation and
moderately preterm was delivery at 35-36 weeks’ gestation.

Analytic Strategy:
Total fish consumption was calculated by summing up the frequency of consumption

of all types of fish. The frequency of fish consumption was assessed both as continuous

31

variables and as quartiles. Descriptive statistics including mean, standard deviation, range
and quartiles were used to describe the distribution of total fish consumption and
consumption of subtypes of fish. The General Linear Models (GLM) and mixed
procedure was used to assess associations between maternal characteristics and maternal
fish consumption expressed on a continuous scale. Univariate and multivariate logistic
regression models were used to test the association between maternal total fish
consumption and the risk of all PD (<37 weeks), moderately PD (35-36 weeks) and very
PD (<35 weeks). Maternal characteristic, which were found to be significantly associated
with total fish consumption, were controlled for in the multivariate logistic regression
model. To explore the influence of high levels of fish consumption on the risk of PD, cut
points at the top 10‘h and 5‘h percentiles, and at thresholds where effects were noted in
two previous studies (226 meals/6 months (106) and 2104 meals/6 months (112)) were

used. All analyses were conducted using the SAS software (133).
Results:

In this study sample, 44% of women were less than 25 years of age, 24% were
African-American, 48% had education of less than or equal to 12 years, 46% were
Medicaid insured, 28% were primiparous, 41% were from community four, 18% were
from community five, 17% were from community two, 15% were from community one
and 9% were from community three (Table 3).

Data concerning consumption of shellfish, canned fish, bought fish, sport caught fish
or other fish were missing for some women respectively, therefore fewer women
(N=1302) were included in calculation of frequency of total fish consumption. The mean

total fish consumption was 19.7 meals/6 months. Three women reported unusually high

32

fish consumption, ranging from 364-548 fishmeals per six months. After removing these
three outliers, the mean level of total fish consumption was 19.6 meals/6 months, or
about three meals of fish per month. The mean levels of consumption of shell fish,

canned fish, bought fish, sport caught fish and other fish were 3.7, 8.5, 6.3, 0.7 and 0.4
meals/6 months respectively, suggesting that canned fish and bought fish contributed
most to total fish consumption (Table 4). During the first six months of pregnancy, 10.9%
of women did not consume any fish and 50% ate at least 9.0 meals of fish. Only 9.2%
reported consuming sport caught fish.

In univariate analyses women 25 years and older had a significantly higher mean total
fish consumption compared to that of women less than 25 years of age (Difference=5.2
meals/6 months, 95% CI 1.5, 8.9). Using women enrolled from community one as
reference, women from community five had a significantly higher mean total fish
consumption (Difference=8.4 meals/6 months, 95% CI 0.5, 16.3). Mean levels of total
fish consumption were also higher in African-Americans compared with whites and
others (Difference=3.6 meals/6 months, 95% CI -0.7, 8.0) and the women with more than
12 years’ education compared to women with 12 or few years (Difference=3.6 meals/6
months, 95% CI -0.1, 7.3), though the results were not statistically significant at P=0.05.
After adjusting for all other maternal characteristics, the adjusted mean total fish
consumption level continued to be significantly higher in women 25 years and older
(Difference=5.9 meals/6 months, 95% CI 1.6, 10.2). In the adjusted model, mean levels
of total fish consumption were suggested to be higher among African-American women

(Difference=4.3, 95% CI -0.5, 9.2) and among women enrolled from community 5

33

(Difference=7.7, 95% CI -0.2, 15.6), though the result is no longer significant after
controlling for other maternal characteristics (Table 5).

Overall, 10.6% of women delivered before 37 weeks’ gestation and 4.1% delivered
before 35 weeks’ gestation. Modeling fish consumption thresholds at >2.0 (25th
percentile), >9.0 (50th percentile), >25.0 (75th percentile), 2 26.0 (Olsen study), 252 (top
10th percentile), 276 (top 5th percentile), and 2104 (Kesmodel study) fishmeals per six
months failed to reveal a statistically significant association between maternal fish
consumption and PD (Table 6). In separate analyses assessing the association between
PD and maternal consumption of bought fish, sport caught fish, canned fish, shellfish and
other fish, high intake of any one subtype of fish did not appear to be significantly

associated with the risk of PD.

34

Table 3. Maternal characteristics of study sample (N=1326)

 

 

 

 

 

 

 

 

 

 

Maternal Characteristics % (N)
Age:
<25 44 (580)
>=25 56 (746)
lEthnicity:
white 68 (902)
Afr-Americanl 24 (3 1 8)
Other 8 (106)
Education (yrs)**:
<=12 48 (631)
>12 52 (691)
Medicaid“:
Ye 46 (613)
Nil 54 (712)
ICommunity:
l 15 (198)
17 (231)
3 9 (117)
i 41 (536)
S 18 (244)
Total Fish Consumption***:
(meals/6 months) < 38 (495)
6-4 35 (452)
>=24 27 (355)
lPrimiparous“:
Ye;l 28 (364)
N 72 (961)

 

 

* Data missing for 1 woman.
** Data missing for 4 women.
*** Data missing for 24 women.

35

.8an w-m 80 mEmmmE Sam 3*

.3808 0 80 £88 00m SE. 088 8:08:88 :6 E80 5; 8:83 m **
£088 0 80 £88 8 80:8: 8 028m 05 .2288 0 8 £88 .883 00% 80 £88 8 088090 *

 

 

 

 

 

 

 

 

 

0.0 0.0 0.0 m.mw_-0 0.0 v.0 .12.. am: .856

0.0 0.0 0.0 0.000 06 5.0 1.22.0 23:8 team

0.0 0._ 0.0 0.m0m-0 2: m0 $.45: Easem—

0.N_ 0.N 0.0 WNM: -0 WE Wm enigma 02:80

Qm 0.0 0.0 me _ -o 9.00 Nam uni—8E =25

0.0m 0.0 0.~ m4: m0 NWN 0.9 Aim—8:08 098:8: 5E 88,—.

0.3 0.0 0.N 028-0 0.mm 5.2 88.38.80 5E 88%
8:88.80 5mm 8.3.8980 50m 8:88.80 5mm mum—8m .8588: Em E82 .85 .8 850E380

 

 

 

 

 

 

 

 

228038 00 :2 “Eu 2: E 309903 28 0mm 88: .8088 5mm 00 82:32 .v 2an

36

use 2: s 30:5 .28 05 2 é was? .2...

 

 

 

o_.ovm “.3."
8.9:...
16.2.3-0: 203.20...» 0.2.3083. 6.8.3808 m
53.3: 3.23:: 82.3092 0.8.302: a
6.2.8-3: séwefia 6.8.0202: 38.3002
8.2.4.284 8.22.38 886.282 2.8.208:
-- -- 3.8.0282 6.88032 :8: "8:25.28:
8.12-02- 823-04.? 8.826002 8.85.282 :z
-- -- 0.8.8.258 6.8.3002 8» ”28:52
8.5.233 $323-8: 38828.8 8.8.2258 7:?
-- -- 0.3.303: €82.25: Efiuv 8:32.:

 

3.3.0 .m.0-v mé

3.8.0 $.00 0am

SEN 6.x: w.m~

3.0m .50: WNN

:3tvfi<-:3_.c<

 

 

 

 

-- -. 8.8.30n2 8.8.3022 35808.33 2225::
1318.03 232.20% 0.8.3098 28:28.2 mNuA

.. .. 8.3 .440 0: G2 .30 :2 muv 2280 8;:
.....=..08m:_.0< 8030 ***03m:_.0< 8.5.5 83288820 8:882

 

8 ems €28: 2: :5: 853:8

 

8 em.“ «:82

 

 

8:08.880 02% 88,—. .8 8.8285

 

 

 

80:08:88 :2.“ 880 2 8022 5 80288820 3:882 .m 835.

37

.60: A803 :0 0o cum—O m .m E 08: E020 SD 1*...
.02 0 0880 .06 06 0088.030 .0 =0 88 E020 05 .1...
65:9: 0 hum 280: 002.0 8m E2: 82.: 00:03:00 on? 20:30 m wins—80m .1.
50:08:08 05 0:60:50 .0? REES: 08 0838.6. .0.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

008 .08 008 .08 60 .08 60.08 60 .08 60.0.8 60: 60 E 603 62:; 86m 6000:8020
0c 0o 0o 0o 0o 0c
-- -- -- -- -- -- 886.8 6008 0800868000.; mmv £2 mop
A: 0.8 60.08 6.8 .08 60.08 60 .08 60.080008 2800.8 8808 68008. 8N 5.13.6006
ad 0o 00 00 00 00
-- -- -- -- -- -- 606.8 6800.8 600008 6800: 8v £300
60 .0000 A: .98 8.0 .500 $0 .500 G0 .500 $0 H.008: NNN 88 New 35 oém Amwmv mém o.mNA 0000:8020
0o 00 00 00 o0 o0
-- -- -- -- -- -- 08008 8808 0300.8 08808 3% a?
60 .08 60.0828 .08 60.08 60 .08 60.0868 30. 686.8 6800.8 66802. 0? 20008020
0o 00 00 00 00 00
-- -- -- -- -- -- 686.66 680.60. 680.8 088006 odw £8
60 08 60.0.8 0020.0 6.8.08 60 .08 6008080006 8008 E000: E808 QNA 6000:8020
3 0o 00 60 N0 00
-- -- -- -- -- -- 600: 60060 00808 0.8038 cam 58
so 0068 mo 020 85:28 020085620025 80203:;
Edna Fonm 0.000”; m0oun0 o0oun0 8.30 6800. 6806 020600280008 $803828 000E038.
82206.... 2005 @0303. 2:5 ,._.0§m=_.00< 205 903 8v a? 8.8 a? 8v a? RN 385 80:05:28 am
$03 mmv 303 3.3 83> bmv ::Boi E0205 E0895
85 033.632 02 :05

 

 

 

 

 

 

 

 

 

 

 

62903:: 2022: :0 088390 anew 0000005528 002.0 E 8:03—20 8585 00 E08 b3 28% v.25 0cm owflcoobm .c 290,—.

38

.GOS 388 E 6 com—O m .m 5 wow: “Eon EU in:
AN— 3 38$ .3 Ho EcoEmoM .D E cum: “Eon SD 13*
.3258 0 Sq £38 nmc 8m 55 29: uofismcoo 0:3 20:50 m @62me **
.bEaEEoQ was 55:50 .owa 3539: 28 Sunni *

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

AoaSoVQWSS <2 <2 2.9082559 5% 83d 53 253 EN :35
3 Wm <2 <2 ad ad
: -- -- -- -- -- 293$ $8982 Amflvwsafiflswg ::v 23532
Sade$2.8$.~.m.e§.m.oeSmwdinmwé 53 $3 63 €33 3m 03:83
2 I 3 0.. A: 3
-- -- -- -- -- -- 233$ 8830 5:23:23? Ev 53$
cu 9&3 mo 22V $528 95550285 mgofiefi
Edna Edna 3on2 mnoum one”; wvdum an; $83 smcodimogvow iceaaacou 552$
%23€< 2:5 $23.? 256 ,*B§.€< 256 8.3 mmv 3.3 Ram 9? RV 3.3 RN £85 829528 in
8:5 mmv 33 3.2 $15 hmv ::Bntn— ::0qu 5535
F5 233352 =< ash

 

 

 

 

wEocmEE 29:2: 8 wean—«>0 mmsohw coca—bacon sac E 353% E2805 ho EH3 mo 23% 225 can owficoobm .o 035.

39

law

This study found that older women, African-American women and women enrolled
from Community five had higher levels of total fish consumption in the first half of
pregnancy. Evidences showed that maternal fish consumption was higher in women with
more years of education in the unadjusted model but the result did not sustain after
adjusting for other maternal characteristics. Maternal fish consumption did not differ
between women with Medicaid insurance and those without. Levels of total fish
consumption and consumption of types of fish were not associated with the risk of all PD,
very preterm or moderately PD. High levels of total fish consumption, including the top
10‘h percentile, the top 5th percentile and the same cut points as the other two previous
studies (_>_26 meals/6 months and 2104 meals/6 months) was not associated with the risk
of PD.

Researchers have hypothesized that intake of high levels of Q-3 FA may be protective
for preterm birth, possibly because Q-3 FA inhibit the synthesis of eicosanoids from AA
(1 12-113), whose derivatives play a role in the initiation and completion of term and
preterm labor (1 14). Since fish and marine food are the main source of Q-3 FA in many
human populations, this study have considered frequency of fish consumption during the
first 6 months of pregnancy as a potential predictor of PD. However, some other factors
such as the portion size, fish preparation, fish types that were not assessed in this study
and use of fish oil supplement could affect levels of n3-FA intake, so using fish
consumption as a proxy for Q—3 FA intake may have limitations. Direct measurement the
blood level of Q-3 FA in maternal or fetal blood would be preferred to test this

hypothesis.

40

Generally low levels of fish consumption in this study population also reduced the
power to detect an association between fish consumption and the risk of PD. Despite the
small number of women who delivered before 35 weeks’ gestation and consumed high
levels of fish, the analysis showed hints of a protective effect for very PD from increased
levels of total fish consumption from the 25% percentile to the top 10th percentile (OR
ranged from 0.7 to 0.9). At levels beyond the cut point of the study by Kesmodel et al
(2104 meals/6 months), fish consumption seemed to be positively related to the risk of
very PD (OR=2.3), however the sample size at high consumption levels was small and
confidence interval was wide (0.7, 8.0). Data also showed consistently protective effects
for PD using the cut point at the top 10‘h percentile (252 fish meals/6 months). However,
the results were not statistically significant. This may be explained by the fact that only
ten women with total fish consumption levels as high as over 52 fishmeals per six months
delivered before 37 weeks of gestation. For example, with a current sample size of 1302
and using the prevalence of PD in this study, we have only a power of 41% to detect the
difference in the risk of PD at the cut point of 52 fishmeals per 6 months at the
significance of 0.05. To reach a power of 0.80, a sample size of 3287 would be required.

Several previous studies have been undertaken to explore whether maternal fish
consumption, intake of Q-3 FA supplement or levels of Q-3 FA in blood during
pregnancy could change the risk of premature labor. Most of them were conducted in
European countries, such as Denmark, Finland, Faroe Islands and England where levels
of fish consumption are generally high and these studies produced mixed results.

Studies Assessing Fish Consumption in Relation to Preterm Delivery:

4]

In 1993, S. F. Olsen et al (111) conducted a case-control study and found that BW and
birth length of the newborn increased with the frequency of seafood dinner meals during
pregnancy, but such an effect tended to disappear with more than 3 meals/week. The
same analyses for pregnancy duration were not significant though exhibited the same
trend. In a more recent cohort study (106), Olsen et al estimated daily intakes of fish and
Q—3 FA from frequency of fish consumption and found that the decreasing occurrence of
LBW, preterm birth and IUGR was associated with increasing fish consumption in early
pregnancy. They also found that these associations were mainly apparent at the lower end
of the exposure, particularly for mean duration of gestation. Both studies detected a less
significant association at higher levels of fish consumption. It might be related to the
“ceiling effect”, which means after reaching a certain high level of exposure, incremental
increases in exposure will no longer bring substantial changes to outcomes. The more
recent cohort study (106) has a similar study deign as our study, however the population
in Olsen’s study has a higher general consumption of fishmeals (76.8% versus 60.7%
women consume at least 1 fish meals/month). In addition, the fish types in Olsen’s study
were mainly seafood, which may contain higher levels of Q-3 FA.

Another cohort study by Olsen et a1 investigated the influence of dietary nutrient
density of marine 9-3 FA at 30'h week of pregnancy and found no difference in quintiles
of 9-3 FA nutrient density and gestational age at delivery, BW and birth length (125).
Other case-controls studies on the effect of consumption of fish and seafood on the risk
of PD have likewise not detected a significant association with these birth outcomes. U.
Kesmodel et al (112) detected no association between fish intake or intake of fish oil and

the risk of PD or posterm delivery. Bjerregarrd et al found in Denmark that consumption

42

of marine mammals was not significantly associated with LBW (108). Philippe
Grandjean et al reported that in Faroe Island, frequencies of main meals with fish, whale
meat and whale blubber during pregnancy were not related to gestational age, BW or
placental weight (107).

The studies assessing the dietary intake of fish in relation to gestational age at delivery
produced mixed but largely negative results. Five possibilities might contribute to the
explanation of the inconsistency of the results. First, other Q-3 FA sources than fish may
exist in some populations. Second, retrospective questionnaire is not an accurate method
to measure the frequency of fish consumption. Third, the amount of fish intake during
each fishmeal varies among individuals and areas. Fourth, fish type and fish preparation
may influence the content of Q-3 FA in each fishmeal. Fifth, other maternal
characteristics which differ over different populations, such as ethnicity, socioeconomic
status, physical activity and diet other than fish may influence the risk of PD.

Studies Assessing Blood 9-3 FA Levels in Relation to Preterm Delivery:

Evidences have shown that fish and seafood are main n3-FA source for many human
populations. S. Reddy et al proved that the proportion of EPA and DHA were lower in
plasma phospholipids, plasma free FA and total plasma lipids of the vegetarians
compared with those from omnivores (109). In the Olsen’s study (113), the average ratio
of Q-3 FA and Q-6 FA among Faroese women who consumed higher levels of marine fat
was significantly greater than that of Danish women. Grandjean et al also observed that
the maternal and cord serum concentration of EPA increased with maternal marine food

intake, though the tendency was less clear for DHA (107).

43

In contrast to the largely negative results from studies measuring dietary fish intake,
more studies correlating maternal or cord blood 9-3 FA levels and PD detected
significant associations, and most of them were cohort studies. Because of the
contradictory biological effects of 9-3 FA and Q-6 FA on the initiation of spontaneous
labor, the ratio of levels of 9-3 FA to Q-6 FA (the (3/6) ratio) is commonly used as a
reliable single biochemical measure of Q-3 FA level in blood. In an early cohort study by
Olsen et al (110), results showed that gestational age was significantly related to the (3/6)
ratio in phosphatidylethanolamine among 18 women with spontaneous onset of delivery
after controlling for maternal age, weight, marital status or smoking. And in the study by
Olsen et al in 1991 (113), the association between gestation age and the (3/6) ratio was
significant in Danish women but not in Faroese women though marine food consumption
was higher in Faroe Island. The result of this study reinforced the possibility of a “ceiling
effect” in the fish consumption-PD association. A recent cohort study by Grandjean et al
(107) reported that an increase in the relative concentration of DHA in cord serum
phospholipids by 1% resulted in an increased duration of gestation by 1.5 days (95% CI
0.7-2.2), though BW adjusted for gestational age may decrease at high intake level.
Another cohort studies by Olsen et al (125) assessed both dietary fish intake and Q-3 FA
relative to AA (FA-ratio) and indicated no difference in mean gestation length, BW or
birth length across quintiles of the FA-ratio. Possible explanation is that in-person
interview to collect dietary fish consumption data could enhance the awareness of the
participants to increase their intake of marine Q-3 FA. Also the “ceiling effect” cannot be
excluded because the mean fish consumption in this population was as high as 2.5

meals/week.

44

Although studies on blood levels of Q-3 FA produced more exciting findings, it is
difficult to interpret the results from the public health perspective. Whether or not to
encourage pregnant women to eat fish in order to decrease the risk of PD cannot be
decided solely by these results. Lower levels of fish consumption could be accompanied
by a higher consumption of vegetable oil, which contains high levels of Q-6 FA, so
biomarkers such as blood Q-3 FA levels and 3/6 ratio may just reflect levels of vegetable
oil intake rather than fish consumption levels. Intake of fish oil supplement and
individual absorption and metabolism potential can influence the correlation between fish
intake and blood levels of Q-3 FA. Additionally, blood levels of FA might have daily
fluctuations and do not accurately reflected tissue levels.

Experimental Studies and Clinical Trials:

Experimental studies were performed to test the effect of fatty acids in preventing
premature birth and most showed consistently significant results. In the experiment by
Olsen et a1 (52), 15% MaxEPA supplement (F0) and 15% arachis oil supplement (AO)
were given to Lewis rats. Results showed significantly higher levels of Q-3 FA and lower
levels of Q-6 FA, longer gestational age (P<0.01) and interestingly, lower BW (P<0.05)
in FO-rats than AC-rats. The decreased BW in the FO-rats may be explained by the lower
maternal weight gain during pregnancy and the lower food consumption observed in FO-
rats (P<0.01). Baguma-Nibasheka et al found that infusion of fish oil concentrate
emulsion 3ml/kg per day from 124 days of gestation in sheep delayed both the onset of
labor and the time of delivery (53).

Randomized clinical trial is generally deemed as the “golden standard” for

epidemiological studies because of its advantage of controlling for potential confounders

45

and biases in study design. Most clinical trials on fish oil supplements detected
consistently protective effect for premature birth. The earliest published controlled trial of
fish oil was performed during 1938-1939 in London ( 126-129) and reinterpreted in 1990
by Olsen et al (105). 2510 of the 5022 participants were given an estimate of 0.1 g/day
EPA plus DHA supplement for an average length of 20 weeks. The odds of delivering
before 40 weeks of gestation was decreased by 20.4% (P=0.00083) in the treatment group.
In 1992, Olsen et al (104) found that the average gestational age, BW and birth length
were consistently the greatest in the group treated with 2.7 g/d 9-3 FA (P=0.0006, 0.07
and O. 1, respectively) and lowest in the group treated with olive oil though the risk of
preterm and posterm delivery did not differ significantly among the control group and the
two treatment groups. Borod et al found in a clinical trial that DHA in each lipid class
was increased among women who consumed 1 dozen eggs/week, each containing
approximately 135 mg DHA during 24-28 weeks’ gestation and decreased in those
consuming 1 dozen regular eggs/week fewer (102). The results also showed fewer
newborns with LBW or preterm birth in intervention group though the result is not
statistically significant (102). Olsen et a1 did not detect significant effects of Q—3 FA
supplement starting from 33 weeks of gestation on the risk of PD, but they did find
significantly decreased recurrence risk of PD among women who were given the
supplement during the second half of pregnancy (P<0.05), and significantly delayed
spontaneous delivery (P=0.002) in two intervention groups (101). In a randomized
double-blinded placebo controlled trial, Onwude et al (103) observed no difference in the
occurrence of BW less than the 3rd centile and duration of pregnancy between

intervention group and the control group. However, compared with other intervention

46

studies, the study by Onwude et al has a smaller sample size and might not be able to
detect subtle but important differences.

Clinical trials are superior to cohort and case-control studies in controlling for
confounders and testing casual relationship between exposure and outcome. However,
clinical trials also have problems, such as the compliance of participants and the
estimation of etiological relevant time window. Furthermore, most previous clinical trials
used fish oil or Q-3 FA supplement as the intervention assuming that Q-3 FA would be
the critical component in fish consumption that affects risk of PD. However fish oil pills
may not be comparable to dietary fish intake in terms of type and amount of fatty acids

and the combined effect of other nutrient contents in fish.
Conclusion:

The etiology and risk factors for PD has long been a field of active research. The
present study investigated whether eating fish protects women from PD in Michigan area
where general fish consumption is much lower than areas near seawater. We found no
significant association between maternal fish consumption during the first half of
pregnancy and the risk of PD. Most of previous studies were conducted in north
European countries where general consumption of fish and seafood are higher, but few
studies have been done in the USA. These studies provided evidence that fish
consumption, intake of fish oil supplement and increased blood Q-3 FA levels are
associated with longer gestation duration, higher BW and decreased risk of PD, but the
protective effect may not increase after reaching a certain high level of exposure. Future

studies need to clarify etiological relevant time for fish consumption to influence

47

gestational age and take into concern effects of the fish type, fish preparation and intake

of fish oil supplement.

48

MATERNAL MERCURY LEVELS IN HAIR IN RELATION TO FISE

CONSUMPTION AND RISK OF PRETERM DELIVERY

Fei Xue, Claudia Holzman, Larry Fischer, Hossein Rhabar
Introduction:

PD has continued to be the leading cause of fetal mortality and morbidity in the US
despite recent advancements in medical technology. Extensive epidemiological studies
have been done to explore the etiological risk factors for PD, covering from infection,
gene, stress, environmental exposure to diet and nutrition. However, effective prevention
of PD is unfortunately still an unattainable objective. Data from epidemic poisonings in
Japan (67) and Iraq (68) showed that infants with intrauterine exposure to methylmercury
developed marked fetal developmental delays; especially neurodevelopmental disorders
while their mothers experienced little or no overt signs of toxicity. Three forms of
mercury including elemental mercury, inorganic mercury and methylmercury can present
a human health hazard, but only methylmercury has been confirmed by the United States
Environmental protection Agency (US EPA) as developmental toxicant in human.
Clarkson et al indicated that the developing fetus might be 5 to 10 times more sensitive
than the adults to the toxicity of mercury (135). Though previous researches have
consistently suggested the influence of mercury on fetal neurodevelopment,
epidemiological studies evaluating mercury levels in relation to BW and gestational age
are scant and mixed results have been reached. Foldspang et al (82) reported that higher
maternal and offspring methylmercury concentrations were associated with low mean

BW. However, the same association was not detected in other studies (83-86, 136). Small

49

sample size, selection of various biological samples and methods of mercury
measurement can contribute to the inconsistency of these research results.

Fish and seafood, especially the species high in the food chain, are an important
source of methylmercury exposure. Previous studies have shown that approximately 95 %
of the methylmercury in fish ingested by volunteers is absorbed from the gastrointestinal
tract (137—138). In recognition of the hazard of methylmercury exposure, the US EPA has
established a reference dose (Rfl)) for methylmercury at 0.3 ug/kg/day, which is
equivalent to consumption of 19 pg per day of Methylmercury for a 62 kg woman (139).
Most previous studies on this topic were conducted in populations either exposed to
catastrophic pollution or with a diet composed mainly of fish or seafood and thus the
results can not be generalized to populations with moderate levels of fish consumption if
a threshold effect is taken into account. To better understand the effect of fish
consumption and mercury on pregnancy outcomes, we conducted a prospective cohort
study among 1024 pregnant women from five Michigan communities to investigate fish
consumption during the first half of pregnancy in relation to mercury levels in hair and

the effect of mercury levels on the risk of PD.
Methods:

Population:

The POUCH Study, a prospective study enrolls pregnant women at 16 to 26 weeks’
gestation from 52 participating clinics in five Michigan communities, including the major
city and surrounding areas. Eligibility criteria for the POUCH Study include maternal age
greater than 14 years, screening for MSAFP levels between 15-20 weeks’ gestation,

proficiency in English and singleton pregnancy with no known congenital or

50

chromosomal abnormalities at time of recruitment, and no history of diabetes mellitus
prior to pregnancy. This study included POUCH participants who were enrolled from
September 8, 1998 through July 31, 2001. Of the 1336 women enrolled in this study
(approximately one third of all eligible women), five women were lost to follow-up and
five infants had congenital anomalies identified and thus were excluded. At the time of
enrollment women meet with a study nurse who conducts interviews and collects biologic
samples including scalp hair. Women had to have loose hair (e.g. no woven hair) longer
than 3 inches in order to have their hair sampled. Hair samples were not collected in 197
POUCH participants because they declined (N=69, 35%) or had hair that was too short or
not loose (N=128, 65%). Resources for measuring hair mercury levels were available for
the first 1,024 hair samples. The remaining 105 hair samples have not been assessed to
date.

Fish Consumption:

Dietary information about fish consumption during the first half of pregnancy was
collected from all 1336 participating women through in-person interview during the mid-
pregnancy visit at 16-26 weeks’ gestation. Women were asked about frequency of
consumption of shellfish, canned fish, sport caught fish, bought fish and other fish.
Consumption of types of fish was reported as number of meals per day, week, month or
six months and scaled to meals per six months. Total fish consumption was calculated by
summing up consumption of all types of fish.

Mercury Levels in Hair:
Approximately 100 strands or more of hair were sampled from each woman during the

mid-pregnancy interview. The hair samples were cut close to the scalp from the posterior

51

vertex region and the length of hair that was analyzed was measured and represented the
period of time that the subject was pregnant prior to sampling (calculated using an
lcm/month of hair growth). Cold Vapor Atomic Absorption Spectrometry (CVAAS) was
used to quantify total mercury levels in hair.

Gestational Age at Delivery:

Gestational age at delivery was determined by the date of delivery and the gestational
age estimated at the time of MSAFP screening. This estimate was based on the date of
the first day of the LMP, or on an early ultrasound (< 20 weeks) estimated gestational age,
the latter given preference when the two estimates disagreed by more than 2 weeks. Very
preterm was defined as delivery before 35 weeks’ gestation and moderately preterm was
delivery at 35-36 weeks’ gestation.

Analytical Strategy:

Chi-square test was used to assess the differences in maternal characteristics in
women with hair sampled and those without. The distribution of total fish consumption
and consumption of types of fish were described using descriptive statistics including
mean, standard deviation, range and quartiles. Levels of total fish consumption in relation
to maternal characteristics were measured using GLM. Univariate and multivariate
logistic regression models were applied to assess the association between maternal
characteristics and mercury levels in hair. Mercury levels were categorized into quintiles
and the first quintile was used as reference in the analysis. GLM was used to evaluate fish
consumption in relation to mercury levels in hair. In the analysis, mercury levels were
transformed to natural log (log ppm) to adjust for the right skewness, and the results were

transformed back to mercury levels parts per million (ppm) for easier interpretation. Total

52

fish consumption was analyzed in relation to mercury levels after controlling for maternal
characteristics, and consumption of each type of fish was analyzed after controlling for
maternal characteristics and consumption of other types of fish. Total fish consumption
was included in a logistic regression model when assessing the association between
maternal characteristics and quintiles of mercury levels to determine if mercury
originated from sources other than fish consumption. The risk of PD, moderately PD and
very PD in relation to quintiles and the top 10th percentile of mercury levels in hair was
evaluated using logistic regression model. Total fish consumption and maternal
characteristics, which were found to be significantly related to mercury levels in hair,
were controlled for as potential confounders in the adjusted logistic regression model. All

analyses were conducted using the SAS software (133).
Results:

Maternal Characteristics in Relation to Sampling:

The distribution of maternal characteristics including maternal age, ethnicity,
education, Medicaid status, smoking during pregnancy and community were found to be
significantly different between the group of women with hair sampled (N=1024) and
those without (N=197). Compared with the women who did not give hair samples,
women with samples were more likely to be older than 25 years of age (59% vs. 45%),
whites (73% vs. 33%) and other ethnicity (non-African-American) (9% vs. 6%), have
more than 12 years education (58% vs. 35%), not smoke during pregnancy (84% vs.
75%), and enrolled from community 2 (18% vs. 8%), 3 (10% vs. 9%) and 5 (19% vs.
15%) (Table 7).

Fish Consumption:

53

The mean level of total fish consumption in this study population was 20.4 meals/6
months. After removing two outliers whose total fish consumption levels were as high as
364 and 547.5 meals/6 months, the mean level of total fish consumption was reduced to
19.6 meals/6 months. The mean consumptions of types of fish from the highest to the
lowest were 9.1 meals/6months for canned fish, 6.3 meals/6 months for bought fish, 3.8
meals/6 months for shellfish, 0.8 meals/6 months for sport caught fish and 0.3 meals/6
months for other fish. The distribution of total fish consumption and the consumption of
types of fish were slightly right skewed and mean levels were all higher than median
levels. Half of the women consumed at least 3 meals/6 months of canned fish, at least 1
meal/6 months of shellfish and bought fish. 10.5% of the women did not consume any
fish and more than 75% of them did not consume any sport caught fish or other fish
during the first 6 months of pregnancy (Table 8).

Total Fish Consumption in Relation to Maternal Characteristics:

Compared with women less than 25 years of age, mean levels of total fish
consumption were 5.6 meals/6 months (95% CI 1.3, 9.9) higher among women _>_ 25
years of age in the unadjusted model and 5.8 meals/6 months (95% CI 0.8, 10.7) higher
after adjusting for all the other maternal characteristics. Mean total fish consumption
level was 5.7 meals/6 months (95% CI 0.1, 11.2) higher among African-American
women and 7.5 meals/6 months (95% CI 1.3, 13.6) higher after adjusting for all the other
maternal characteristics. Using gestational age of less than 20 weeks at enrollment as
reference, women who were enrolled at 20—24 weeks’ and _>_24 weeks’ gestation had
lower levels of total fish consumption both in the unadjusted model (Mean Difference=-

l 1.0 meals/6 months, 95% CI -18.1, -4.0; and Mean Difference=-l4.9, 95% CI-22.6, -7. l ,

54

respectively) and after controlling for all the other maternal characteristics (Adjusted
Mean Difference=-9.8 meals/6 months, 95% CI -16.9, -2.7; and Adjusted Mean
Difference=—l3.7, 95% CI -21.5, -5.8, respectively). Mean total fish consumption among
women enrolled from community 5 was 10.3 meals/6 months (95% CI 1.1, 19.4) higher
than that of women from community 1 in the unadjusted model and 9.2 meals/6 months
(95% CI 0.1, 18.4) higher after adjusting for all the other maternal characteristics (Table
9).

Mercury Levels in Hair in Relation to Maternal Characteristics:

The range of mercury levels in hair was from 0.013 ppm to 2.50 ppm. After adjusting
for all the other maternal characteristics and using the first quintile of mercury level as
reference, women 25 years and older were more likely to be in the fourth quintile
(Adjusted OR=1.96, 95% CI 1.21 , 3.17) and the fifth quintile (Adjusted OR=2.13, 95%
CI 1.22, 3.71) as compared to women under 25 years of age. Compared with African-
American women, whites and women of other ethnicities were more likely to be in the
third quintile (Adjusted OR=1.89, 95% CI 1.10, 3.26), fourth quintile (Adjusted OR=1.93,
95% CI 1.08, 3.44) and fifth quintile (Adjusted OR=1.94, 95% CI 1.00, 3.73). Women
not insured by Medicaid were at significantly increased risk of being in the second
quintile (Adjusted OR=1.93, 95% CI 1.15, 3.25), third quintile (Adjusted OR=1.82, 95%
CI 1.09, 3.05), fourth quintile (Adjusted OR=1.90, 95% CI 1.09, 3.32) and fifth quintile
(Adjusted OR=3.05, 95% CI 1.69, 5.49). The distributions of hair mercury levels and fish
consumption levels in women from community one and three were very similar to each
other. They are also geographically close, so community one and three were combined

into one group in the analysis. Using community five, the community with the lowest hair

55

mercury levels as reference, and after adjusting for all the other maternal characteristics,
women enrolled from community two were consistently more likely to be in the second
quintile (Adjusted OR=3.93, 95% CI 1.76, 8.77), third quintile (Adjusted OR=2.38, 95%
CI 1.05, 5.36), fourth quintile (Adjusted OR=5.73, 95% CI 2.39, 13.72) and fifth quintile
(Adjusted OR=7.34, 95% CI 2.52, 21.37) of hair mercury level. In the adjusted model,
women enrolled from community one and three were at increased risk of being in the
second quintile (Adjusted OR=2.21, 95% CI 1.10, 4.42), the fourth quintile (Adjusted
OR=3.50, 95% CI 1.64, 7.44) and the fifth quintile (Adjusted OR=6.00, 95% C1233,
15.45) of hair mercury levels than community five. Interestingly, after adjusting for all
the other maternal characteristics, women from community four were at significantly
reduced risk of being in the third quintile of hair mercury levels (Adjusted OR=0.47, 95%
CI 0.27, 0.82) than community 5 though the mean mercury level in community 5 was
lower. Mercury levels in hair were associated with education, smoking before and during
pregnancy and gestation age at enrollment in the univariate analysis, but the statistical
significance of the association disappeared after controlling for other maternal
characteristics (Table 10).

Mercury Levels in Relation to Fish Consumption:

Using total fish consumption of less than 6 meals/6 months as reference, mean
mercury levels were significantly increased among women with 6-24 meals/6 months
(P<0.0001) or 2 24 meals/6 months (P<0.0001) of total fish consumption after
controlling for all maternal characteristics. Mean mercury levels were higher among
women who consumed 6-24 meals/6 months (P<0.0001) or 2 24 meals/6 months

(P<0.0001) of canned fish, women who consumed 6-24 meals/6 months of bought fish

56

(P=0.02) and women who consumed 224 meals/6 months of sport caught fish (P=0.007)
after adjusting for all maternal characteristics and the consumption of other types of fish.
Increase mercury levels in hair were found to be significantly related to consumption of
6-24 meals/6 months of shellfish (P<0.0001), 224 meals/6 months of bought fish (P=0.03)
and 6-24 meals/6 months of sport caught fish (P=0.02), but the relationships disappeared
after controlling for all the other maternal characteristics and consumption of other fish
types (Table 11).

Mercury Levels in Hair in Relation to Maternal Characteristics after Controlling

for Total Fish Consumption:

After adjusting for other maternal characteristics and levels of total fish consumption,
the association between mercury levels in hair and maternal age, ethnicity, Medicaid
status and community remained to be significant. Women 2 25 years old were at elevated
risk of being in the fourth quintile (OR=1.83, 95% CI 1.12, 3.02) and fifth quintile
(OR=2.05, 95% CI 1.15, 3.66) of hair mercury levels compared with African-American.
Whites and women of other ethnicity were at higher risk of being in the third quintile
(OR=2.14, 95% CI 1.22, 3.74), the fourth quintile (OR=2.58, 95% 1.38, 4.85) and fifth
quintile (OR=2.72, 95% CI 1.33, 5.54) of hair mercury levels. Women not insured with
Medicaid were found to be consistently at increased risk of being in the second to the
fifth quintile of mercury levels in hair (OR=1.96, 95% CI 1.16, 3.30; OR=1.87, 95% CI
1.11, 3.14; OR=1.90, 95% CI 1.08, 3.37; OR=3.21, 95% CI 1.73, 5.98, respectively for
the second to the fifth quintile). The risk of being in the second to the fifth quintile of
mercury levels in hair was significantly higher among women enrolled from community

two than community five (OR=3.97, 95% CI 1.78, 8.86; OR=2.53, 95% CI 1.10, 5.79;

57

OR=6.56, 95% CI 2.65, 16.27; OR=8.69, 95% CI 2.83, 26.67, respectively for the second
to the fifth quintile). Women enrolled from community one and three continued to be at
increased risk of being in the second quintile (OR=2.36, 95% CI 1.17, 4.79), fourth
quintile (OR=4.59, 95% CI 2.05, 10.29) and the fifth quintile (OR=10.65, 95% CI 3.58,
31.70) compared with women enrolled from community five. Women enrolled from
community four were at significantly lower risk of being in the third quintile of mercury
level (OR=0.51, 95% C1029, 0.91) than women from community five (table 12).
Mercury Levels in Hair in Relation to the Risk of Preterm Delivery:

The association between mercury levels in hair and the risk of PD was assessed both
in univariate and multicovariate models that adjusted for total fish consumption and
maternal characteristics that were found to be significantly associated with hair mercury
levels (i.e. maternal age, ethnicity, Medicaid status and community). Compared with
women in the first quintile of hair mercury level, women in the fourth quintile of mercury
level were at significantly greater risk of delivering moderately preterm in the adjusted
model (Adjusted OR=2.80, 95% CI 1.03, 7.59). Women in the top 10th percentile of hair
mercury levels were at significantly higher risk of delivering very preterm when
compared to women in the lowest 90% in both the univariate model (OR=2.35, 95% CI
1.10, 5.05) and the adjusted model (Adjusted OR=2.95, 95% CI 1.30, 6.69). Interestingly,
the third quintile was associated with significantly decreased risk of very PD in both
univariate model (OR=0.20, 95% CI 0.06, 0.73) and adjusted model (OR=0.22, 95% CI

0.06, 0.81), however, the confidence intervals were wide (table 13).

58

Table 7. Some demographic characteristics in the 1024 POUCH participants with
mercury assessments in sampled hair and in 197 participants with no hair sampled.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Maternal Characteristics Hair Sampled No Hair Sampled
% (N) (N=1024) % (N) (N=197)
Maternal Age+: <25 yrs 41 (422) 55 (109)
>=25 yrs 59 (602) 45 (88)
Maternal Ethnicity+: White, Non-Latino 73 (752) 33 (66)
African-Ameri 18 (183) 61 (120)
Othe 9 (89) 6 (11)
Education-H <=12 yrs 42 (430) 65 (129)
>12 yrs 58 (590) 35 (68)
Medicaid“ +: Yes 43 (436) 65 (129)
No 57 (587) 35 (72)
Smoking before Pregnancy“: Yes 26 (268) 33 (64)
No 74 (751) 67 (133)
Smoking during Pregnancy* +: Yesj 16 (161) 25 (48)
No 84 (859) 75 (149)
Preterm Delivery: <35 wks 4 (44) 5 (9)
35-36 wks 6 (57) 9 (18)
>=37 wks (Ref) 90 (923) 86 (170)
lGestation Age at Enrollment: <20 wks 14 (141) 11 (21)
20-24w 59 (606) 61 (121)
>=24 ks 27 (277) 28 (55)
Total Fish Consumption“: <6 meals/6 months 37 (376) 38 (74)
6-24 meals/6 months 34 (347) 34 (67)
>=24 meals/6 months 29 (288) 28 (55)
lCommunity+: 1 15 (148) 23 (46)
2 18 ( 185) 8 (16)
3 10 (100) 9 (17)
4 38 (393) 45 (88)
5 19 (198) 15 (30)

 

 

 

+ The difference is statistically significant between women with hair sampled and those

without at P<0.05.
* Data missing for 1-13 women.

59

.2258 o 2 uofiom 28 3:58 0 ho 5:9: £33 5% Hum mm cotomom 5....
690th 203 2:58 0 Ba £38 com 55 EOE cocmfismug 5mm :38 5:5 5an N .I.
69:85 am: some 28 5:53 5; .8 BE 50:92:28 :mc :38 28 5an 2 28 @528 Sam ...

 

 

 

 

 

 

 

 

 

od od od ané 5m md Ans: * 53 .550

0.0 od od odoé mw wd ANS: *smE Ew=a0 team

CE 0; 0.0 meTo 5.2 m6 825 turn Ewscm

o.m_ Qm od meTO 92 _d €55 *ami moo—Emu

ed 0; od mdwTO v.3 w.m 33: iii :25

odm od o.m WENé Nam 0.2 $8: A**co>oEP_ 22:58 5:— :83.

0.8 od Qm mHVmé :m VON :3: 2.3.553 M£6225 55 .85.
25:8th 25:3th oEEuhum mew—Em Eva—£59 E32 AZV :mE we Egan—£3.50

5m“. 5cm 5mm Em

 

 

 

 

 

 

 

 

.223:on Co has :5 2: E @2523 98 :mc :30: 2.3232: cmc Co 23:52 .w 033.

60

.033 2: E 83:93 .050 2: z: :8 335.25. .22..

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.98:3— 2: EEK 35.559

 

—U 9&3 .532

 

 

55955—80 Amw— =33.

 

Aodvm .1.
.modvm *
LE: .3: «5 LE: .3: 2: A3... fin: 0.: Son .3: o8 m
3.: .3: mm 8.: .3: M3 8.3 .3: 0.2 8.8 ._ .2: 0.2 a
2.: .3: ca 2.2 .3: a: 8.9. .3: 2: $3 .3 : 3: m
2 .: .3: Z 3.: .3: nm 6.8 .3: 3m 88 .3: as N
-- -- SM: .3: 3m 3.: .N.O:w.m;co:: _ "8:55ch
13:29:: :2. 1.3:. .33 3:. cl: .2: v.2 3.8 .3: E: a? 3R
:3: 5.8: as- $8.? .32: 3: GR .2: «.8 3% .3: a: a? 3.8
-- -- 3mm .3: :2 30m .3: in c969? 8v 35:58:: a :3 5:58:
a: .3: mm A; .3: 3 SE .3: SN 33 .3: 3m :2
-- -- :8 .3: v.8 38 .5: QB 8» 5:23.: was: 322::
Q: .3: E- 5. .3: to- 3.8 «a: 2m :1: .3: :8 A2
-- -- 21m .3: EN 63 .2: 2: 8: 5:23: 2&2. 33:5
2: .3: «o- 6:. .3: no 93m .wd: EN 38 .5: new cz
-- -- $3 .2: UR 3mm .3: 3N 8: ”38:52
a: .3: Z a: .3: a: 6.8 .2: :2 8.3 .3: N: a: 2A
-- -- $3 .3: SN 68 do: :2 a: fluv assuag—
...G.2 .2: ms *8: .3: mm 2.2 .3: can can do: caustusfififiz
-. .. 2.2 .5: EN Aim .5: v.2 22.8 a 8.23 535::
LE: .3: 3 L? .m: 3 3% .3: SN 82 .3: :2 3R
.- .. 2.8 6.23:218 £2: :2 muv gas: a»...
**...:8m=_.:< $5.5 £56522 35.5 35:89:52.0 353::

 

 

.mozmcofifiwso BEBE: 8 cog—8 E 65:9: 023:: 5:983:00 5: 130,—. .m 2an

61

s=moEmfimoow 82a 08 .85 28

08—0 29.5 oofiEa—mvu can 3&3— E—EO Eat 5an 95:5 205— corms—3:8 :mc :39 98 :2 5 £32 >582: 582 33......

.5603 m 88 wage «an 3.1....
.5an _ .59. $588 Sam .22....
.5803 v he 95%: 3mm .13..
.RD=BQEE mo 58833 3 £o>o_ .6388 2

x23 8.5885: 203 3.38 05 was $2.386 Em: “8 as? 2 £8?wa 2: E wo_ 383m: 8 388838“ 803 £90— .95qu .1.

.wodumnqfi “a “gummy—mu £838:me ._.

 

 

 

 

 

 

 

 

 

:3 .88 6982.8 €n1€8 83 .38
$5 £5 93 $6 $3 .88 Rd We, $339552;
-- -- -- -- $3 .88 mad as: oz 2&2. 8.22.5
a} .83 an». .38 $3 .88 33 ~38
$3 .3.“ $3 a3 8% .28 03 Avz
-- -- -- -- 83 .2 .8 m; as: new "iiafifiz
$3.38 23.38 33.68 3.168
NE W: 8: 3o 93 £8 £6 a: 2A
-- -- -- -- 83 .28 8 .o 8.82.» Snv ae.3=§
35.8.: 3.m.8.: 82.2.: 63 .28
is.— ..nfi .33 84 G3 .88 v3 5:5 a. 33.5
-- -- -- -- $3 .28 :d as: =85E<$< ӣ355
32.3.: 5.338 @0838 63 .88
*2.” .83 81 3o 83 .28 and mNuA
-- -- -- -- a; .2 .8 w; as: muv A288 93
86% 3%.? 8.943 82%? 938 :8 83 2. 83.39820 .2532
A3838 @0858 £888 2:828 _33 r852 :82
2:55 .3 2:55 5.. 2:55 Em 2:55 EN

 

 

 

 

 

no Quma “=3 33% 36C

 

 

agar—883:0

REES: .850 05 =m 5m Magma?“ Baa 83380820 BEBE: 8 cons—2 E 0383 SE 5 Gwen: Eng 6%: 8.50.82 .2 2an

62

 

=wo.n..m.woow 820 o... .35 8...

30.0 203 005.823. 8... 2.5.8 8.8.0 :88 .5an macs... 80>... 8.385880 5... .38 8.... 8.... 5 £90— 8.35... :82 *...............

.5an m .8 $.68... 8.5 ............._.
.5.an . .8 wag... «HQ 1......
.5503 v .8 mafia. San— .........

ACCESS...— 80 53.598 9. m_o>o_ .0580... o.
80.... 808.085... 825 3:52 on. E... .3238? Em... .8 75.8.. o. 89.3.. of 5 mo. .33.... 8 38.8.0.5... 9.2... $05. b.3802 ......
.modnanmfi .m .goEcmE >=moumzfim ..

 

 

 

 

 

 

 

 

 

 

_.m...m. an... .3; .3... .2... .38 .8... .3...
1...... Em... on. L... 83.8.83... 3.
5...... $8 an... .8”... 83 .2... E... .2...
$3 5.5 .83 E... a... .88 2.... a
83.38 am. .38 .9558 .3... 8.8
8.. 8.... *3... a... .8688 a... v
-- -- -- -- .2... 8.88... m 5.5.5.5
8m. .38 .3... .88 83.8.8 8819.8
8... 8.... a... 9.... .8688 .3 .9...» 3R
.2. .38 .8. .88 22.5.8 8.1.88
3.... 8... mm. mm... .36....88... a...» 3.... .5588...
-- .3... .38 mm... €89.38v on... 5.3.89
.8888 .8838 ..o.m..m.8 .8888
8.. 8... a... m... .8888 a... .8.» 32.82.2892.
-- -- -- -- .3... mm. 8 mm. o a»... A.7. use... micam
38...... 8.3m... 3.3.3.. 88%.... BE... 2.... 9...? .5... 8.25.9820 359...:
.3888 .mm...4.~..8 .3888 a. .... a. .8 .3... has: :82
2:55 .3 2.2.5 5.. 2.2.5 Em 85.5 .5.
.0 8m. 2... 3.3. .25 _

 

 

 

 

338.8855

.388... .550 on. E. .8 mafia? .08.. 838.883.... BEBE: o. .5820. 5 29:3 .6... E flows... FEB .05. .9682 .o. 28....

63

29......» 5-. .8 923.... San. ......

.modnmfii .2 .gogawa 88.2.28 ..

 

 

 

 

 

 

 

 

 

 

 

 

$9.80.. .3 8.683.... 0.35.... «E... mE. wfim: 40$? 6:... 8.3— $8.0

.3. .88 .3. 8.8 an. ......8 .8. .28
8... a... a... no... 8.. 3.223....
-- -- .: -- Com. c7. 9.82. mixofim
.23 .2... .28 .2... .3... ...... an... 8....
L... 2.... .2... .2... oz
-- -- -- -- a»... 8.. 2.2228...
2.2 .88 .3. .28 .3. .88 .8. .88
3... .2 9.... 8.. 2. SA
-- -- -- -- c»... a. Suv .......=2.8=....
8mm .8... .3... .2... .2... .8... 2... .28
a... .23 1.... a... .25 2 2......
-- -- -- -- co... 2852......4 5.25....
.8... .2... .8... .2... an... .28 .2... .88
.2... .2... a... 3... RR
-- -- -- -- ca... mav ".28.. «a...
83.38 .2838 .3838 3.828 8.2.2.826 252...).
2.2.5 22 2.2.5 2... 2.2.5 in 2.2.5 .2.

 

28.9.5250 cm... 8 5.2.... c. .2... a. m.o>o. >588). .2 0.8....

 

E 5. .33.... 38 ._.._.

.mo.ous....m a Eocema ragga ..

 

 

 

 

 

 

 

 

 

8;. .38 an... .38 a. .v .38 851:...
*3... am... 3.. .....m.~ .3.
5.3 .88 2.3. .38 RE .248 G...» £58
.39.” *emé *mm.N *bafi N
3.3 $8 sod .38 .3... .88 6.... .38
an. 8.. hm... S... a
-- -- -- -- m “5:55:50
3.... .88 .3. .38 Ed .38 .3... .38
we... mm... 8.. cm... a...» 3R
an. .88 SK .38 :3 .38 an. .38
we... 8.. on. S... a...» 3.8 1.558.... 2
-- -- -- -- cum. 9.3:? a»... 5.380.
8.3 .38 a? .mp8 82 .38 .88 .88
8.. NS .3. 3... 8.» 15:23....
-- .. -- -- c3: c2 mum—.6 wfixcfim
83.3.8 3885.8 @2158 83.2.8 8.3232226 353...).
2.2.5 .3 2.2.5 5.. 3.2.5 Em 2.3.5 .5
2.29.2 8 5.8288 0.35.... .2... 2.. 3...... .583 .5; 3.3. $3

 

 

 

dosmasmcoo am... 8 .8328 a. an: 5 £03. .9552 .3 033.

65

Table 12. Mercury level (ppm range) in hair sample in relation to maternal characteristics

after adjusting for total fish consumption and other maternal characteristics.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Odds Ratio amd 95%CI, using the first
quintile (0.013-0.13) as reference
2nd Quintile 3rd Quintile 4th Quintile 5th Quintile
Maternal Characteristics (0.12-0.19) (0.20-0.26) (0.27-0.38) (0.39-2.50)
Age (years): <25 (Ref) -- -- -- --
>=25 0.97 1 .43 1.83* 2.05 *
(0.60, 1.56) (0.88, 2.32) (1.12, 3.02) (1.15, 3.66)
Ethnicity: Afri-American (Ref) -- -- -- _-
White & Other 1.32 2.14* 2.58* 2.72*
(0.79, 2.21) (1.22, 3.74) (1.38, 4.85) (1.33, 5.54)
Education": <=12 yrs (Ref) -- -- -- --
>12 yrs 1.00 0.99 1.21 1.44
(0.62, 1.62) (0.60, 1.64) (0.71, 2.05) (0.81, 2.54)
Medicaid“: Yes (Ref) -- -- -- --
No: 1.96* 1.87* L90“ 3.21*
(1.16, 3.30) (1.11, 3.14) (1.08, 3.37) (1.73, 5.98)
Smoking before No (Ref) -- -- -- --
regnancy** Yes 0.65 0.81 0.83 0.63
(0.35, 1.21) (0.44, 1.51) (0.42, 1.65) (0.29, 1.35)
Smoking during No (Ref) -- -- -- --
pregnancy“ Yes 1.44 1.78 1.72 1.07
(0.69, 3.03) (0.85, 3.70) (0.78, 3.75) (0.41, 2.80)
IGestation Age <20Wks (Ref) -- -- -- --
At Enrollment" 20-24 Wks 0.81 1.56 1.02 0.68
(0.42, 1.59) (0.74, 3.27) (0.50, 2.10) (0.33, 1.38)
>=24 Wks 0.90 1.20 0.89 0.64
(0.44, 1.84) (0.52, 2.77) (0.41, 1.91) (0.29, 1.42)
Community: 5 -- -- -- -_
4 0.82 0.51* 1.05 1.32
(0.46, 1.46) (0.29, 0.91) (0.55, 2.00) (0.62, 2.82)
2 3.97* 2.53* 6.56* 8.69*
(1.78, 8.86) (1.10, 5.79) (2.65, 16.27)I(2.83, 26.67)
1&3 2.36* 1.95 4.59* 10.65*
(1.17, 4.79) (0.92, 4.12) (2.05, 10.29)](3.58, 31.70)

 

* Statistically significant at alpha=0.05.

** Data missing for 1-7 women.

66

 

6:8 “$32 2 3.39:8 $338 20585 .3 and. 3....
65:8: 3632 2 uoSquU 5..
damage—5:3 saw 38 ::m bazfifionv £83,: 23602 53:50 .6? 3:88: @622: 83883820 3:53.: :8 333.43.. *

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

83. Gem 8?“ 62 An: 62 a»... a? an: a?
.88 .38 .28 .88 .88 .88 .88 .88 .38 .38
m3 mg 8: $6 85 So «as 8... 8: «no .522:
3.8 2v 80>
-- -- -- -- -- -- -- -- -- -- $8
38 RR :8
am: a? 8% :8: am.» 8; a; 83 83 63
.88 .88 .38 .88 .88 .88 8.8 8.8 .88 .28
33 moo S: 3o 3." on: 5.: 5o 8: 8.: .822:
88 8-2 .82
.. -- -- -- -- -- -- -- -- -- A88
38 RR .58.
:3 E .N 2 3 83 $3 a: :2 so: am: 83
.88 .38 .88 .28 .88 .88 .38 .58 .88 .38
a: 8: mm: 8.0 9...: mod mod «2 Ed Rd .522:
so: RV =<
3982.1 SE.
8236.... 8.5 833.8... 820 823%... one: 8338.... 8.5 .8583. 880
89828 83.38 3:838 6:888 83.88 98 3?
._.._....o__a85: 3.0855 .5 $0858 54 $0855 Em $085.5 EN :23:
:5— moH 8.565
6 88 2: 2:: ”25

 

 

:8: E 898: 8:3 832 >582: ::m bog—ow E585 .5qu: 53:68:: 2:. .2 033.

67

Discussion:

The mean level of total fish consumption was 19.6 meals/6 months after removing two
outliers. 11% of women did not eat any fish and 90% of them did not eat any sport caught
fish. Older women, African-American women, women enrolled earlier in gestation and
women enrolled from community five were found to eat more fish during the first half of
pregnancy after controlling for other maternal characteristics. Increased hair mercury
levels were found to be significantly associated with older maternal age, ethnicities other
than African-American, not being insured by Medicaid and enrollment from community
1-3 after adjusting for maternal characteristics. Mercury levels in hair were significantly
related to higher total fish consumption and consumption of canned fish and sport caught
fish. After adjusting for total fish consumption and other maternal characteristics, the
association between hair mercury levels and maternal characteristics including maternal
age, ethnicity, Medicaid status, and community persisted. The top 10‘h percentile of hair
mercury level was found to be significantly associated with increased risk of very PD.
The fourth quintile was found to be significantly associated with enhanced risk of
moderately PD.

Later gestational age at enrollment was observed to be significantly associated with
lower total fish consumption levels, most likely due to maternal characteristics of women
who delay prenatal screening or would be difficult to enroll into the study in a timely
manner. Women enrolled from community five were found to have the highest levels of
fish consumption. Below is a map regarding the location of the five communities in this
study (Figure 1). Community five is the closest to a Great Lake among the five

communities and perhaps this influences availabilities and preferences for fish. Culture

68

and food preference may account for the higher fish consumption levels among older
women and African-American women.

Because tuna comprises a large part of canned fish, the detected association between
the consumption of canned fish and increased hair mercury levels is compatible with
previous reports that methylmercury readily enters the aquatic food chain and
bioaccumulates in predatory fish such as swordfish, pike and ocean tuna (66). Though the
study population generally ate very low levels of sport caught fish, consumption of sport
caught fish continued to be significantly related to increased hair mercury levels after
adjusting for consumption of other types of fish. The result may be explained by local
pollution in nearby waters, but confounding from other factors such as life style and

outdoor activities cannot be excluded.

Figure 1. Location of the five communities in this study

 

Similar to the association between maternal age and fish consumption, older women
were also found to have higher hair mercury levels, and the association can not be
completely explained by elevated fish consumption. The higher hair mercury levels in

older women can not be explained by bioaccumulation as well because the half-life of

69

 

methylmercury is approximately 70-80 days in the human body (137) and so hair
mercury level can not reflect methylmercury exposure over years. However, aging may
have influence on the absorption, metabolism and excretion of mercury, though no
previous reports on these mechanisms have been found. The association between
Medicaid insurance and lower mercury levels in hair cannot be explained by ethnicity,
age and fish consumption. Some factors related to higher socioeconomic status might
contribute to an elevated exposure to mercury.

Higher fish consumption was found among African-American women, however non-
African—American women had significantly higher hair mercury levels after controlling
for all the other maternal characteristics. Similarly, women enrolled from community 5
were found to consume the highest levels of fish, but their hair mercury levels were the
lowest among the 5 communities. In addition, all the relationships between maternal
characteristics and hair mercury levels remained significant after controlling for total fish
consumption levels. It is generally accepted that fish is the major source of
methylmercury for human. These results suggest that there may be other sources of
methylmercury in addition to fish, or more likely the measurement of fish intake or
subtypes of fish by maternal diet recall for the past 6 months leads to some
misclassification of fish consumption levels. The personal interview used in our study did
not take into concern portion size, cooking methods, other fish types and the intake of
fish supplements, thus the analyses may not be able to completely control for fish
consumption. Besides this, other methylmercury sources may influence the mercury
exposure in this study sample. Localized industrial, especially those chlor-alkali industry

using the mercury process, the pulp and paper industry and other industries using

70

mercury (118) and agricultural pollution including alkyl, alkoxyalkyl, aryl, and inorganic
mercurial fungicides have been used for seed dressing, as turf fungicides and in orchards
(1 15) could contribute to mercury exposure, but how the exposure to the pollution varied
with age, ethnicity and Medicaid status is unknown. The exposure to other sources of
mercury including pharmaceuticals and preservatives in cosmetics and toiletries (115) are
more likely to differ with maternal characteristics and further studies need to clarify these
factors in relation to personal characteristics and their effect on human health.

One previous review of human data on the developmental effects of methylmercury
exposure concluded that maternal hair levels of 10 to 20 ppm are potentially harmful to
fetal development (66). The range of hair mercury levels among this study population
were 0.013 ppm to 2.5 ppm, therefore were considerably lower. However, the top 10"I
percentile of mercury levels in hair (055-250 ppm) was found to be significantly
associated with increased risk of very PD after controlling for potential confounders. The
biological mechanism for higher mercury exposure to be related to early delivery is
unknown. Few studies have assessed mercury levels in relation to risk of PD, and all of
them produced negative results (82, 83, 86). Foldspang et al (82) evaluated average
weekly intake of marine food, and blood methylmercury concentrations of mothers and
offsprings immediately after delivery, and concluded that gestational age at delivery was
not related to either number of meals of local food or to the blood mercury concentration.
Fu et al (86) measured exposure to metallic mercury levels in an occupation based study
and reported that the difference in incidence of PD between the group exposed to 0.06-

01 mg/m3 of mercury in air and control group did not reach a significant level.

71

Grandjean et al (83) found that cord blood total mercury levels did not vary by
gestational age at delivery.

Three underlying factors may explain the inconsistency of results between our study
and previous studies: First, most of previous studies were based on small sample size
(range 182-704) (81-84, 86) and they might not have had enough power to detect subtle
but important differences. Second, none of these studies measured hair mercury levels.
Metallic mercury level in air is not equivalent to methylmercury exposure. Blood levels
may not be a good measurement of mercury intake over time (88-89). In our study, we
measured mercury levels in the length of hair that reflects the exposure during the first
half of pregnancy. Once incorporated into hair follicles, the methylmercury is chemically
stable, and therefore, gives a longitudinal history of blood methylmercury levels (90).
Previous studies did not control for fish consumption. Fish intake is associated with
mercury levels in hair, and previous studies have reported that Q-3 FA in fish can delay
the initiation of parturition by down-regulating the formation of PGE2 and PGF2a,
promoting the synthesis of PG12 and P013 and thus leading to a more relaxed
myometrium (49-51). Thus, in previous studies fish consumption may have had a
protective effect for PD and thereby the mercury-gestational age association toward null.

The study suggested a “U” shaped curve in the association between mercury levels in
hair and the risk of PD. The possible explanation is that at lower levels of fish
consumption, the effect of nutrients in fish, such as 9-3 FA predominates, but when
certain higher levels of fish consumption are attained the effect of methylmercury can

overcome the effect of Q-3 FA. Although current study adjusted for total fish

72

 

 

consumption in the analysis, the effect of fish consumption may not have been entirely
controlled for because of the crude measurement of levels of fish consumption.

There were some maternal characteristics that differed between all POUCH enrollees
and these with hair sampled. Women with hair sampled were older women, less likely to
smoke during pregnancy, had higher education, were not insured with Medicaid, and
more likely to be whites and other ethnicities than African-American. Among the 197
women without hair sampled, only 35% declined and the other 65% were due to either
short hair or elaborated hairstyles such as woven hair, which were more prevalent among
African-American women. African-American women in POUCH tended to be younger,
lower educate, and insured by Medicaid. So this might explain some of the differences in
characteristics noted between those with and without hair samples. However, the
disproportional distribution of these maternal characteristics should not significantly bias
the study results because of the sheer larger number of women with hair sampled
compared to those without (1024 vs. 197). Furthermore, levels of total fish consumption,
an important factor suggested to be associated with both mercury levels in hair and the
risk of PD, did not differ significantly between women with hair sampled and those
without.

We used CVAAS to measure mercury levels in hair, however this measurement
cannot distinguish methylmercury from the total mercury content. Though hair is a good
indicator medium for methylmercury exposure (90), analysis of hair mercury levels may
be confounded by absorption of mercury vapor onto the hair strands (91 ). Nonetheless,
this situation mostly occurs if women had occupational exposure as in a dentist’ office or

a factory containing metallic mercury. Since this is a community-based study sample of

73

women, the contamination of vapor mercury in hair is probably minimized compared
with that found in occupational studies. In addition, all hair samples were washed before
analyses and thus chances to be contaminated by mercury vapor were minimized. Also in
our analysis, fish consumption, an important source of methylmercury exposure for
humans, was found to be significantly associated with hair mercury levels, suggesting
that the measurement of hair methylmercury levels in our study was not significantly

confounded by the contamination of vapor mercury.
Conclusion:

In brief, this study showed that total fish consumption, and consumption of canned
fish and sport caught fish were significantly associated with increased hair mercury levels.
Maternal characteristics including age, ethnicity, Medicaid status and community of
enrollment were significantly associated with mercury levels in hair after taking account
of fish consumption. Women in the top 10th percentile of methylmercury levels in hair
(O.55-2.50 ppm) were at to increased risk of Delivery very preterm. The consumption of
fish and other seafood may provide with beneficial nutrients such as 9-3 FA and
Selenium, and adverse substances such as methylmerucury. Studies need to consider
effects from both exposures when interpreting evidence regarding the role of fish
consumption during pregnancy. Though fish consumption is an important source of
methylmercury for human, results of this study indicate other potential methylmercury
sources, which distribute differentially with certain maternal characteristics, especially
those related to socioeconomic status and place of residence. Further studies need to
clarify these sources and their importance in increasing methylmercury in the human

body and influencing human health.

74

SUMMARY

In order to evaluate the influence of fish consumption on the risk of PD, the two papers
included in this thesis separately investigated two elements, which have been suggested
by previous studies to have opposite effects on gestational age at delivery. The first paper
measured frequency of fish consumption during the first half of pregnancy to provide
indirect evidence that high levels of Q-3 FA, which comes mainly from fish and seafood,
may prevent PD. Our study did not show the risk of PD, moderately PD or very PD to be
significantly associated with levels of fish consumption at. However, there were hints that
the risk of very PD may be reduced with increased total fish consumption, and the risk of
PD, moderately PD and very PD seemed to be lower among women in the top 10"1
percentile of total fish consumption, but the results were not statistically significant.

The second study examined mercury levels in hair reflecting the growth during the
first half of pregnancy, and the risk of PD, moderately PD and very PD. Results showed
that women in the top 10‘h percentile of hair mercury levels were significantly more
likely to delivery very preterm.

Adjusted analyses were performed in both studies. Higher fish consumption levels
were reported in older women, African-American women, women enrolled from
community 5 and women enrolled after 20 weeks’ gestation. Total fish consumption and
consumption of canned fish and sport caught fish were found to be significantly
associated with increased mercury levels in hair. Higher mercury levels in hair were
found in older women, non-African-American women, women not insured with Medicaid
and women enrolled from community 1-3 even after adjusting for levels of fish

consumption. Adjustment for total fish consumption and other maternal characteristics

75

did not influence the significance of the detected association between hair mercury levels
and the risk of PD.

The range of hair mercury levels in our study was much lower than the range of 10 to
20 ppm that is considered potentially harmful to fetal development (66). However, our
study detected a significantly increased risk of very PD in women in the top 10th
percentile of hair mercury levels. No previous study has reported significant association
between mercury exposure and gestational age at delivery. Our study may have been
well-suited detect this association because it had a larger sample size than other studies
and controlled for of fish consumption to test the persistence of the effect and attempt to
control for the potential protective effects of fatty acids from fish.

The studies confirmed fish consumption, especially marine predators such as tuna as
one of important sources of methylmercury intake for human being. The significance of
sport caught fish can be accounted for by local mercury pollution released into nearby
waters and accumulated in fish. However, some other maternal characteristics including
maternal age, ethnicity, Medicaid status and community of enrollment were also found to
be associated with mercury levels even after controlling for fish consumption. Some of
the remained effect of maternal characteristics can be explained by inaccuracy of the
measurement of fish consumption ( 106), others can be other resources of mercury
exposure such as local pollution, pharmaceuticals and preservatives in cosmetics and
toiletries, and occupational exposure. Based on these results, reducing fish consumption
during gestation alone may not be enough for pregnant women to decrease
methylmercury exposure. Further efforts need to clarify other methylmercury sources and

establish effective methods to manage these sources.

76

 

Though there were hints of reduced risk of PD among women with elevated total fish
consumption levels, fish consumption was not shown to be significantly associated with
the risk of PD. This result is incompatible with the reported protective effect of Q-3 FA
in fish for PD (112-114) and findings from previous epidemiological studies (54, 124).
Toxic effects of other contents in fish, inaccuracy in measurement of fish consumption,
and generally low fish consumption levels in this study population may explain such
inconsistency. Although fish is the main dietary source of Q-3 FA, it also contains
hazardous elements such as methylmercury, PCB and DDT, the toxic effects of which
can compromise the benefits from Q-3 FA. The contents of these toxics in fish tissues
and the association between fish consumption and the risk of PD can vary according to
geographic location and extent of local pollution of waters and fish tissues. Compared
with previous studies conducted in European countries that found association between
maternal fish intake during pregnancy and gestational age, fish consumption levels in this
study population are considerably lower. Thus the number of women delivered
prematurely and ate high levels of fish during pregnancy is small and enough power may
not have been reached. Measurement of dietary fish intake has always been a challenge
for researchers because most of these measurements rely on the recall of participants and
inaccuracy in the recall is unavoidable, especially when the consumption levels are high
or a long time has elapsed between fish intake and the interview. Although our personal
interview inquired the consumption of types of fish, the processing and cooking methods,
which can also influence the nutrient content, were too complex to be standardized and
documented. Furthermore, our study only assessed the fish consumption levels of the first

half of pregnancy, however, this period of pregnancy may not be the important etiologic

77

 

 

time window for fish consumption to influence the gestation age at delivery. Fish
consumption levels before or after this period of time may also be important in deciding
the risk of preterm delivery.

Overall, the two studies in this thesis provided evidences about moderately high levels
of fish consumption and methylmercury intake on the risk of PD. Through the
investigations, we hope to demonstrate the relative benefits and hazards of fish
consumption during the first half of gestation on one of the most important pregnancy
outcomes, PD. The results are not comparable to studies on populations taking seafood as
main food resource or data from epidemic methylmercury poisonings. Our studies also
showed that selected maternal characteristics are not only related to fish consumption
levels, but also associated with methylmercury levels even after controlling for fish
consumption. These findings are important because effective interventions cannot be
taken until these potential methylmercury resources are better identified. In order to
provide further insight into the role of Q-3 FA and methylmercury in fish in on gestation
age, future studies should include information on Q-3 FA levels in maternal and fetal

blood, intake of fish oil supplement, occupational mercury exposure and local pollution.

78

10.

ll.

12.

13.

REFERENCE

Lumley, J. (2003). "Defining the problem: the epidemiology of preterm birth." Bjog
110 Suppl 20: 3-7.

Iannucci, T. A., P. G. Tomich, et al. (1996). "Etiology and outcome of extremely
low-birth-weight infants." Am J Obstet Gynecol 174(6): 1896-900; discussion
1900-2.

Creasy, R. K. (1993). "Preterm birth prevention: where are we?" Am J Obstet
Gynecol 168(4): 1223-30.

Leung, T. N ., V. J. Roach, et al. (1998). "Incidence of preterm delivery in Hong
Kong Chinese." Aust N Z J Obstet Gynaecol 38(2): 138-41.

Monaghan, S. C., R. E. Little, et al. (2000). "Preterm birth in two urban areas of
Ukraine." Obstet Gynecol 95(5): 752-5.

Mattison, D. R., K. Damus, et al. (2001). "Preterm delivery: a public health
perspective." Paediatr Perinat Epidemiol 15 Suppl 2: 7-16.

Ananth, C. V., D. P. Misra, et al. (2001). "Rates of preterm delivery among Black
women and White women in the United States over two decades: an age-period-
cohort analysis." Am J Epidemiol 154(7): 657-65.

Morrison, J. J. and J. M. Rennie (1997). "Clinical, scientific and ethical aspects of
fetal and neonatal care at extremely preterm periods of gestation." Br J Obstet
Gynaecol 104(12): 1341-50.

Demissie, K., G. G. Rhoads, et al. (2001). "Trends in preterm birth and neonatal
mortality among blacks and whites in the United States from 1989 to 1997." Am J
Epidemiol 154(4): 307-15.

Meis, P. J., J. M. Ernest, et al. (1987). "Causes of low birth weight births in public
and private patients." Am J Obstet Gynecol 156(5): 1165-8.

Joseph, K. S., M. S. Kramer, et al. (1998). "Determinants of preterm birth rates in
Canada from 1981 through 1983 and from 1992 through 1994." N Engl J Med
339(20): 1434-9.

Romero R, AthadeN, Maymon E, Pacora P, Bahado-Singh R. “Premature Rupture
of the membranes.” In: Reece A, Hobbins J, eds. Medicine of the fetus and mother.

Philandelphia: Lippincott-Raven. ( 1999): 158 1-625.

Sawdy, R. J. and P. R. Bennett (1999). "Recent advances in the therapeutic
management of preterm labour." Curr Opin Obstet Gynecol 11(2): 131-9.

79

14.

15.

l6.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

Olsen, P., E. Laara, et al. (1995). "Epidemiology of preterm delivery in two birth
cohorts with an interval of 20 years." Am J Epidemiol 142(11): 1184-93.

Mercer, B. M., R. L. Goldenberg, et al. (1996). "The preterm prediction study: a
clinical risk assessment system." Am J Obstet Gynecol 174(6): 1885-93; discussion
1893-5.

Kramer, M. S., L. Goulet, et al. (2001). "Socio-economic disparities in preterm
birth: causal pathways and mechanisms." Paediatr Perinat Epidemiol 15 Suppl 2:
104-23.

Lockwood, C. J. and E. Kuczynski (1999). "Markers of risk for preterm delivery." J
Perinat Med 27(1): 5-20.

Greig, P. C., A. P. Murtha, et al. (1997). "Maternal serum interleukin-6 during
pregnancy and during term and preterm labor." Obstet Gynecol 90(3): 465-9.

Harger, J. H., A. W. Hsing, et al. (1990). "Risk factors for preterm premature
rupture of fetal membranes: a multicenter case-control study." Am J Obstet Gynecol
163(1 Pt 1): 130—7.

McLean, M., A. Bisits, et al. (1995). "A placental clock controlling the length of
human pregnancy." Nat Med 1(5): 460-3.

McGregor, J. A., G. M. Jackson, et a1. ( 1995). "Salivary estriol as risk assessment
for preterm labor: a prospective trial." Am J Obstet Gynecol 173(4): 1337-42.

Coleman, M. A., J. A. Keelan, et al. (2001). "Predicting preterm delivery:
comparison of cervicovaginal interleukin (IL)-1beta, IL-6 and IL-8 with fetal
fibronectin and cervical dilatation." Eur J Obstet Gynecol Reprod Biol 95(2): 154-8.

Kakinuma, C., C. Kuwayama, et al. (1997). "Trophoblastic apoptosis in mice with
preterm delivery and its suppression by urinary trypsin inhibitor." Obstet Gynecol
90(1): 117-24.

Rosen, T., E. Kuczynski, et a1. (2001). "Plasma levels of thrombin-antithrombin

complexes predict preterm premature rupture of the fetal membranes." J Matern
Fetal Med 10(5): 297-300.

Elovitz, M. A., J. Baron, et al. (2001). "The role of thrombin in preterm parturition."
Am J Obstet Gynecol 185(5): 1059-63.

Miyoshi, H., M. B. Boyle, et al. (1996). "Voltage-clamp studies of gap junctions

between uterine muscle cells during term and preterm labor." Biophys J 71(3):
1324-34.

80

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

Word, R. A., J. T. Stull, et al. (1993). "Contractile elements and myosin light chain
phosphorylation in myometrial tissue from nonpregnant and pregnant women." J
Clin Invest 92(1): 29-37.

Tucker, J. M., R. L. Goldenberg, et al. (1991). "Etiologies of preterm birth in an
indigent population: is prevention a logical expectation?" Obstet Gynecol 77(3):
343-7.

Macfarlane, A., S. Cole, et al. (1988). "Epidemiology of birth before 28 weeks of
gestation." Br Med Bull 44(4): 861-93.

Koupilova, I., D. Vagero, et al. (1998). "Social variation in size at birth and preterm
delivery in the Czech Republic and Sweden, 1989-91." Paediatr Perinat Epidemiol
12(1): 7-24.

Gardner, M. O., R. L. Goldenberg, et al. (1995). "The origin and outcome of
preterm twin pregnancies." Obstet Gynecol 85(4): 553-7.

Ventura, S. J., J. A. Martin, et al. (1999). "Births: final data for 1997." Natl Vital
Stat Rep 47(18): 1-96.

Joseph, K. S., A. C. Allen, et al. (2001). "Causes and consequences of recent
increases in preterm birth among twins." Obstet Gynecol 98(1): 57-64.

Carr-Hill, R. A. and M. H. Hall (1985). "The repetition of spontaneous preterm
labour." Br J Obstet Gynaecol 92(9): 921-8.

Kristensen, J ., J. Langhoff-Roos, et al. (1995). "Implications of idiopathic preterm
delivery for previous and subsequent pregnancies." Obstet Gynecol 86(5): 800-4.

Goldenberg, R. L., S. K. Mayberry, et al. (1993). "Pregnancy outcome following a
second-trimester loss." Obstet Gynecol 81(3): 444-6.

Lumley, J. (1993). "The epidemiology of preterm birth." Baillieres Clin Obstet
Gynaecol 7(3): 477-98.

Fisch, B., L. Harel, et al. (1997). "Neonatal assessment of babies conceived by in
vitro fertilization." J Perinatol 17(6): 473-6.

Perri, T., R. Chen, et al. (2001). "Are singleton assisted reproductive technology
pregnancies at risk of prematurity?" J Assist Reprod Genet 18(5): 245-9.

Kramer MS, Attributable causes of low birth weight. In: Battaglia F, Falkner F,

Garza C, et al., eds. Maternal and extrauterine nutritional factors: their influence on
fetal and infant growth. Madrid: Ediciones Ergon, 1996: 349-57.

81

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

Golding, J. (1995). "Reproduction and caffeine consumption--a literature review."
Early Hum Dev 43(1): 1-14.

Madinger, N. E., J. S. Greenspoon, et al. (1989). "Pneumonia during pregnancy: has
modern technology improved maternal and fetal outcome?" Am J Obstet Gynecol
161(3): 657-62.

Slattery, M. M. and J. J. Morrison (2002). "Preterm delivery." Lancet 360(9344):
1489-97.

Kramer, M. S. ( 1993). "Effects of energy and protein intakes on pregnancy
outcome: an overview of the research evidence from controlled clinical trials." Am
J Clin Nutr 58(5): 627-35.

Rush D. Effects of changes in protein and caloric intake during pregnancy on the
growth of the human fetus. In: Chalmers I, Enkin MW, Keirse MJNC, eds.
Effective care in pregnancy and childbirth. Vol. 1. Oxford, England: Oxford
University Press, 1989:255-80.

Kramer, M. S., K. Demissie, et al. (2000). "The contribution of mild and moderate
preterm birth to infant mortality. Fetal and Infant Health Study Group of the
Canadian Perinatal Surveillance System. " J ama 284(7): 843-9.

Bang, H. O., J. Dyerberg, et al. ( 1976). "The composition of food consumed by
Greenland Eskimos." Acta Med Scand 200(1-2): 69-73.

Dyerberg, J ., H. 0. Bang, et al. (1978). "Eicosapentaenoic acid and prevention of
thrombosis and atherosclerosis?" Lancet 2(8081): 117-9.

Lange, A. P. ( 1984). "Induction of labour." Dan Med Bull 31(2): 89-108.

Geirsson, R. T. and I. A. Greer (1990). "Prostaglandins: a key factor in human
labor." Acta Obstet Gynecol Scand 69(5): 371-3.

Omini, C., G. C. Folco, et al. (1979). "Prostacyclin (PGIZ) in pregnant human
uterus." Prostaglandins 17(1): 113-20.

Olsen, S. F., H. S. Hansen, et a]. (1990). "Fish oil versus arachis oil food
supplementation in relation to pregnancy duration in rats." Prostaglandins Leukot
Essent Fatty Acids 40(4): 255-60.

Baguma-Nibasheka, M., J. T. Brenna, et al. (1999). "Delay of preterm delivery in
sheep by omega-3 long-chain polyunsaturates." Biol Reprod 60(3): 698-701.

82

 

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

Olsen, S. F., H. S. Hansen, et al. (1986). "Intake of marine fat, rich in (n-3)-
polyunsaturated fatty acids, may increase birthweight by prolonging gestation."
Lancet 2(8503): 367-9.

Andersen, H. J ., L. F. Andersen, et al. (1989). "Diet, pre-eclampsia, and intrauterine
growth retardation." Lancet 1(8647): 1146.

Olsen, S. F., J. Olsen, et al. (1990). "Does fish consumption during pregnancy
increase fetal growth? A study of the size of the newborn, placental weight and
gestational age in relation to fish consumption during pregnancy." Int J Epidemiol
19(4): 971-7.

McParland, P., J. M. Pearce, et al. (1990). "Doppler ultrasound and aspirin in
recognition and prevention of pregnancy-induced hypertension." Lancet 335(8705):
1552-5.

Ylikorkala, O. and U. M. Makila (1985). "Prostacyclin and thromboxane in
gynecology and obstetrics." Am J Obstet Gynecol 152(3): 318-29.

Fitzgerald, D. J ., W. Rocki, et al. (1990). "Thromboxane A2 synthesis in
pregnancy-induced hypertension." Lancet 335(8692): 751-4.

Olsen, S. F. and N. J. Secher (1990). "A possible preventive effect of low-dose fish
oil on early delivery and pre-eclampsia: indications from a 50-year-old controlled
trial." Br J Nutr 64(3): 599-609.

Cunningham, F. G. and J. W. Williams (2001). Williams obstetrics. New York,
McGraw-Hill.

WHO. Inorganic Mercury Environmental Health Criteria 118. World Health
Organization, 1991.

ATSDR. Toxicological Profile for Mercury. Agency for Toxic Substances and
Disease Registry, 1994.

WHO. Methylmercury Environmental Health Criteria 101. World Health
Organization, 1990.

WHO. Mercury-Environmental Aspects, Environmental Health Cruteria 86. World
Health Organization, 1989.

Gilbert, S. G. and K. S. Grant-Webster (1995). "Neurobehavioral effects of

developmental methylmercury exposure." Environ Health Perspect 103 Suppl 6:
135-42.

83

67.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

Harada, M. (1978). "Congenital Minamata disease: intrauterine methylmercury
poisoning." Teratology 18(2): 285-8.

Marsh, D. O., T. W. Clarkson, et al. (1987). "Fetal methylmercury poisoning.

Relationship between concentration in single strands of maternal hair and child
effects." Arch Neurol 44( 10): 1017-22.

Sager, P. R. (1988). "Selectivity of methyl mercury effects on cytoskeleton and
mitotic progression in cultured cells." Toxicol Appl Pharmacol 94(3): 473-86.

Sager, P. R. and D. W. Matheson (1988). "Mechanisms of neurotoxicity related to
selective disruption of rnicrotubules and intermediate filaments." Toxicology 49(2-
3): 479-92.

Vogel, D. G., P. S. Rabinovitch, et al. (1986). "Methylmercury effects on cell cycle
kinetics." Cell Tissue Kinet 19(2): 227-42.

Bondy, S. C. and M. McKee (1990). "Prevention of chemically induced
synaptosomal changes." J Neurosci Res 25(2): 229-35.

LeBel, C. P., S. F. Ali, et al. (1990). "Organometal-induced increases in oxygen
reactive species: the potential of 2',7'-dichlorofluorescin diacetate as an index of
neurotoxic damage." Toxicol Appl Pharmacol 104(1): 17-24.

Sarafian, T. and M. A. Verity (1991). "Oxidative mechanisms underlying methyl
mercury neurotoxicity." Int J Dev Neurosci 9(2): 147-53.

Bartolome, J. V., R. J. Kavlock, et al. (1987). "Development of adrenergic receptor
binding sites in brain regions of the neonatal rat: effects of prenatal or postnatal
exposure to methylmercury." Neurotoxicology 8(1): 1-13.

Syversen, T. L., G. Totland, et al. (1981). "Early morphological changes in rat
cerebellum caused by a single dose of methylmercury." Arch Toxicol 47(2): 101-1 1.
Slotkin, T. A. and J. Bartolome (1987). "Biochemical mechanisms of.

developmental neurotoxicity of methylmercury." Neurotoxicology 8(1): 65-84.
Fuyuta, M., T. Fujimoto, et al. (1978). "Embryotoxic effects of methylmercuric
chloride administered to mice and rats during orangogenesis." Teratology 18(3):
353-66.

Geyer, M. A., R. E. Butcher, et al. ( 1985). "A study of startle and locomotor activity

in rats exposed prenatally to methylmercury." Neurobehav Toxicol Teratol 7(6):
759-65.

84

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

92.

93.

Inouye, M., K. Murao, et al. (1985). "Behavioral and neuropathological effects of
prenatal methylmercury exposure in mice." Neurobehav Toxicol Teratol 7(3): 227-
32.

Sikorski, R., T. Paszkowski, et al. (1986). "Mercury in neonatal scalp hair." Sci
Total Environ 57: 105-10.

Foldspang, A. and J. C. Hansen (1990). "Dietary intake of methylmercury as a
correlate of gestational length and birth weight among newborns in Greenland." Am
J Epidemiol 132(2): 310-7.

Grandjean, P., K. S. Bjerve, et al. (2001). "Birthweight in a fishing community:
significance of essential fatty acids and marine food contaminants." Int J Epidemiol
30(6): 1272-8.

Marsh, D. O., M. D. Turner, et al. (1995). "Fetal methylmercury study in a Peruvian
fish-eating population." Neurotoxicology 16(4): 717-26.

Bjerregaard, P. and J. C. Hansen (1996). "Effects of smoking and marine diet on
birthweight in Greenland." Arctic Med Res 55(4): 156-64.

Fu, W. Z. (1993). "[Effects of mercury exposure on reproduction in female
workers]." Zhonghua Yu Fang Yi Xue Za Zhi 27(6): 347-9.

United States. Environmental Protection Agency. Office of Air Quality Planning

and Standards. (1998). Mercury Study Report To Congress... Volume V: Health
Effects Of Mercury And Mercury Compounds... US. Environmental Protection

Agency... December 1997. [8.1., s.n.

Sherlock, J. C. and M. J. Quinn ( 1988). "Underestirnation of dose--response
relationship with particular reference to the relationship between the dietary intake
of mercury and its concentration in blood." Hum Toxicol 7(2): 129-32.

Sherlock, J. C., D. G. Lindsay, et al. (1982). "Duplication diet study on mercury
intake by fish consumers in the United Kingdom." Arch Environ Health 37(5): 271-
8.

Phelps, R. W., T. W. Clarkson, et al. (1980). "Interrelationships of blood and hair
mercury concentrations in a North American population exposed to
methylmercury." Arch Environ Health 35(3): 161-8.

Francis, P. C., W. J. Birge, et al. (1982). "Mercury content of human hair: a survey
of dental personnel." J Toxicol Environ Health 10(4-5): 667-72.

Birke, G., A. G. Johnels, et al. (1972). "Studies on humans exposed to methyl
mercury through fish consumption." Arch Environ Health 25(2): 77-91.

85

94.

95.

96.

97.

98.

99.

100.

101.

102.

103.

104.

105.

Skerfving, S. (1974). "Methylmercury exposure, mercury levels in blood and hair,
and health status in Swedes consuming contaminated fish." Toxicology 2(1): 3-23.

Von Burg, R. and H. Rustam (1974). "Electrophysiological investigations of
methylmercury intoxication in humans. Evaluation of peripheral nerve by
conduction velocity and electromyography." Electroencephalogr Clin Neurophysiol
37(4): 381-92.

Cappon, C. J. and J. C. Smith (1982). "Chemical form and distribution of mercury
and selenium in edible seafood." J Anal Toxicol 6(1): 10-21.

Hansen, J. C., H. S. Pedersen, et al. (1994). "Fatty acids and antioxidants in the
Inuit diet. Their role in ischemic heart disease (IHD) and possible interactions with
other dietary factors. A review." Arctic Med Res 53(1): 4-17.

Ackman RG. (1990). “Variability of fatty acids and lipids in seafoods. In: Omega 3

News.” Boston: Department of Preventive Medicine, Massachusetts General
Hospital, 1990; 5:2-4.

Bjerve, K. S., A. M. Brubakk, et al. (1993). "Omega-3 fatty acids: essential fatty
acids with important biological effects, and serum phospholipid fatty acids as
markers of dietary omega 3-fatty acid intake." Am J Clin Nutr 57(5 Suppl): 8018-
8058; discussion 805S-806.

Hoving, E. B., C. M. van Beusekom, et al. (1994). "Gestational age dependency of
essential fatty acids in cord plasma cholesterol esters and triglycerides." Pediatr Res
35(4 Pt 1): 461-9.

Van Houwelingen, A. C., M. M. Foreman-van Drongelen, et al. (1996). "Essential
fatty acid status of fetal plasma phospholipids: similar to postnatal values obtained
at comparable gestational ages." Early Hum Dev 46(1-2): 141-52.

Olsen, S. F., N. J. Secher, et a1. (2000). "Randomised clinical trials of fish oil
supplementation in high risk pregnancies. Fish Oil Trials In Pregnancy (FOTIP)
Team." Bjog 107(3): 382-95.

Borod, E., R. Atkinson, et a1. (1999). "Effects of third trimester consumption of
eggs high in docosahexaenoic acid on docosahexaenoic acid status and pregnancy."
Lipids 34(Suppl): S231.

Onwude, J. L., R. J. Lilford, et al. (1995). "A randomised double blind placebo
controlled trial of fish oil in high risk pregnancy." Br J Obstet Gynaecol 102(2): 95-
100.

Olsen, S. F., J. D. Sorensen, et al. (1992). "Randomised controlled trial of effect of
fish-oil supplementation on pregnancy duration." Lancet 339(8800): 1003-7.

86

106.

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

117.

Olsen, S. F. and N. J. Secher (1990). "A possible preventive effect of low-dose fish
oil on early delivery and pre-eclampsia: indications from a 50-year-old controlled
trial." Br J Nutr 64(3): 599-609.

Olsen, S. F., Secher, N. J .: Low consumption of seafood in early pregnancy as a
risk factor for preterm delivery: prospective cohort study. ij: 3241447, 2002.

Grandjean, P.; Bjerve, K. S.; Weihe, P.; Steuerwald, U.: Birthweight in a fishing
community: significance of essential fatty acids and marine food contaminants. Int J
Epidemiol, 2001, 30:1272-8.

Peter Bjerregaard and Jens C. Hansen: Effect of smoking and marine diet on
birthweight in Greenland. Arc Med Res 1996; 55:156-164.

S. Reddy, T.A.B Sanders and O. Obeid: The influence of maternal vegetarian diet
on essential fatty acid status of the newborn. European Journal of Clinical Nutrition
(1994) 48, 358-368.

Olsen, S. F.; Hansen, H. S.; Jensen, B.; Sorensen, T. 1.: Pregnancy duration and the
ratio of long-chain n-3 fatty acids to arachidonic acid in erythrocytes from Faroese
women. J Intern Med Suppl, 1989, 225:185-9.

Olsen S.F., Philippe Grandjean, Pal Weihe, Torbjom Videro: Frequency of seafood
intake in pregnancy as a determinant of birth weight: evidence for a dose dependent
relationship. J Epidemiol Community Health 47: 436-40, 1993.

Kesmodel, U.; Olsen, S. F.; Salvig, J. D.: Marine n-3 fatty acid and calcium intake
in relation to pregnancy induced hypertension, intrauterine growth retardation, and
preterm delivery. A case-control study. Acta Obstet Gynecol Scand: 76: 38-44,
1997.

Olsen, S. F.; Hansen, H. S.; Sommer, S.; Jensen, B.; Sorensen, T. I.; Secher, N. J .;
Zachariassen, P.: Gestational age in relation to marine n-3 fatty acids in maternal
erythrocytes: a study of women in the Faroe Islands and Denmark. Am J Obstet
Gynecol 1991; 164:1203-9.

Lu, F. C., Berteau, P. E. & Clegg, D. J. (1972) “International Atomic Energy
Agency, Vienna, Techn. Rep.” Ser. 137.

“Sixteenth Report of the Joint FAO/WHO Expert Committee on Food Additives,
Wld Hlth Org. techn. Rep. Ser.”, 1972, No. 505; FAQ Nutrition Meetings Report
Series, 1972, No. 51.

Weiss, H. V., M. Koide, et al. (1971). "Mercury in a Greenland ice sheet: evidence
of recent input by man." Science 174(10): 692-4.

87

118.

119.

120.

121.

122.

123.

124.

125.

126.

127.

128.

129.

130.

Wershaw, R. I. (1970) “US. Geological Survey”, Profes. Paper 213, p. 29.
Environmental Research (1971) “Hazards of mercury”, 4, 26.

McCormack MC. The contribution of low birth weight to infant mortality and
childhood morbidity. N Engl J Med 312:82-90, 1985.

Committee to Study the Prevention of Low Birth Weight, Division of Health
Promotion and Disease, Institute of Medicine. Preventing Low Birth Weight.
National Academy Press, Washington, DC, 1985.

Hansen H S and Olsen S. F. Dietary n-3 fatty acids, prostaglandins, and prolonged
gestation in humans. PP 305-317 in Biological Membranes: Aberrations in
Membrane Structure and Function. (A Karnowski, A Leaf, L Bolis edsO AR Liss
Inc., New York, 1988.

Ferretti A, Flanagan V P, Reeves V B. Occurrence of prostaglandin E3 in human
urine as a result of marine oil ingestion: gas chromatographic-mass spectrometric
evidence. Boichimica et Biophysica Acta 959: 262-268, 1988.

Mitchell M D. Prostaglandins during pregnancy and the perinatal period. Journal of
Reproduction and Fertility 62: 301-315, 1981.0lsen SF, Joensen HD. High livebom
birthweight in the Faroes: comparison between birthweights in the Faroes and
Denmark. J Epidemiol Community Health 1985; 39:27-32.

Olsen SF, Joensen HD. High livebom birthweight in the Faroes: comparison
between birthweights in the Faroes and Denmark. J Epidemiol Community Health
1985; 39:27-32.

Olsen, S. F.; Hansen, H. S.; Secher, N. J.; Jensen, B.; Sandstrom, B.: Gestation
length and birth weight in relation to intake of marine n-3 fatty acids. Br J
Nutr(1995), 73, 397-404.

People’s League of Health (1942 a). Nutrition of expectant and nursing mothers.
British Medical Journal ii, 77-78.

Pe0ple’s League of Health (1942 b). Nutrition of expectant and nursing mothers.
Lancet ii, 10-12.

People’s League of Health (1943). A safe milk supply memorandum: the
compulsory pasteurization of milk. London: Hereford Times Ltd.

People’s League of Health (1946). The nutrition of expectant and nursing mothers

in relation to maternal and infant mortality and morbidity. Journal of Obstetrics and
Gynecology of the British Empire 53, 498-509.

88

131.

132.

133.

134.

135.

136.

137.

138.

139.

140.

141.

142.

Holzman, C., B. Bullen, et al. (2001). "Pregnancy outcomes and community health:
the POUCH study of preterm delivery." Paediatr Perinat Epidemiol 15 Suppl 2:
136-58.

Cohen, G. R., L. B. Curet, et al. (2001). "Ethnicity, nutrition, and birth outcomes in
nulliparous women." Am J Obstet Gynecol 185(3): 660-7.

Villar, J ., A. M. Gulmezoglu, et al. (1998). "Nutritional and antimicrobial
interventions to prevent preterm birth: an overview of randomized controlled trials."
Obstet Gynecol Surv 53(9): 575-85.

Copyright (c) 1999 SAS Institute Inc., SAS Campus Drive, Cary, North Carolina
27513, USA. All rights reserved. l

Allen, K. G. and M. A. Harris (2001). "The role of n-3 fatty acids in gestation and
parturition." Exp Biol Med (Maywood) 226(6): 498-506.

Clarkson, T. W. (1992). "Principles of risk assessment." Adv Dent Res 6: 22-7.

Davidson, P. W., G. J. Myer, et al. (1999). "Association between prenatal exposure
to methylmercury and developmental outcomes in Seychellois children: effect
modification by social and environmental factors." Neurotoxicology 20(5): 833-41.

Aberg, B., L. Ekman, R. Falk, U. Greitz, G. Persson and J. Snihs. (1969).

“Metabolism of methylmercury (203Hg) compounds in man: Excretion and
distribution. Arch. Environ.” Healh 19:478-484.

Miettinen, J. K. (1973). “Absorption and elimination of dietary (Hg-H) and
methylmercury in man.” In: Mercury, Mercurial, and Mercaptans, M.W. Miller and
T.W. Clarkson, Ed. Springfield, IL. P. 233-243.

US. EPA. (1992). “Methylmercury - IRIS (Integrated Risk Information System).
U. S. Environmental Protection Agency”.

Baumgartner, Werner S., et al., “Detection of drug use by analysis of hair.”

Proceedings of the 35lh Annual Meeting-Society Nuclear Medicine. Journal of
Nuclear Medicine Supplement. Vol 29. No. 5. 1998, R980.

89

ll