.u

3 1mg... 9. u
.1

. S. .. 3m): I: e x.

‘ hsdhjsq. a..." i ‘ .

alsvw. .1 <1ng . um" .Mﬁndkxma
.Lr‘é

I\..\|

aw 1%,“?
m . am. :(5 C

 

IT
. . If:
23.? .
.W. .ﬂnhﬁuak V
a u .

All...

. ..3nl..llh(m
(1.....unil. st»!
thd...” I)...
‘ fume...)le

‘. .uﬂnnumﬁm.
”gnu...” nil:

........,._<:. ﬁrm... . taming .mmauﬂunrmh ‘m.§ﬁ§

 

2. LIBRARY
10 0a: Michigan State
mversity

This is to certify that the
dissertation entitled

EPIDEMIOLOGY OF THYROID HORMONE
ABNORMALITIES IN EXTREMELY PREMATURE INFANTS

presented by

TING HONG

has been accepted towards fulﬁllment
of the requirements for the

Ph.D. degree in @idemioﬂy

 

 

 

Major Professor’s Signature

 

Putt}? "f , 2 3‘8
a 1

Date

MSU is an affinnative-action, equal-opportunity employer

— -.-~—c—---—----.-

.---—-—.-—.—-—-—-—n-----—.- —‘-_-l-l-l---I-l-l-v- — -.--.—-----—v—-—---— ..

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:lProj/Acc&Pres/CIRCIDateDue,indd

EPIDEMIOLOGY OF THYROID HORMONE
ABNORMALITIES IN EXTREMELY PREMATURE INFANTS
By

Ting Hong

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Epidemiology

2008

ABSTRACT

EPIDEMIOLOGY OF THYROID HORMONE ABNORMALITIES IN EXTREMELY
PREMATURE INFANTS

By
Ting Hong

Transient hypothyroxinemia of prematurity (THOP) is manifest by very low
levels of serum thyroxine (T4) and tri-iodothyronine (T3), but normal thyroid stimulating
hormone (TSH) levels from shortly after birth until up to 6 weeks after birth. THOP is
commonly seen among extremely premature infants born before 28 weeks of gestational
age (GA). Studies have shown that THOP could increase risk of cerebral palsy (CP), a
major neurodevelopmental disability outcome among children born prematurely, and may
also cause other neurodevelopmental disabilities at later ages. Recent evidence suggests
thyroid hormone (TH) supplementation may prevent some cases of CP and improve
cognitive outcomes among premature infants. Little has been written about the correlates
of baseline TH levels or postnatal TH levels in infants at risk for THOP.

From April 2005 to August 2007, we conducted a randomized clinical trial of TH
supplementation in preterm infants born between 24 and 28 weeks of GA. Neonatal free
T4 (FT4), total T4 (TT4), and TSH values were measured within 24 hours aﬁer birth but
prior to treatment (as baseline), and then on day 3, 7, 14, 21, 42, 56, and at discharge.
Maternal FT4, TT4, and TSH were measured shortly after delivery. FT4 was assessed by
direct equilibrium dialysis and TT4 and TSH by chemiimmunoluminescent assay.
Maternal and labor exposures were abstracted through medical records. Neonatal
information was kept in daily charts. Discharge information was recorded in study

discharge form.

This dissertation reviewed the characteristics and the correlates of neonatal
baseline and postnatal TH levels, and analyzed correlates of neonatal baseline and
postnatal TH levels (log-scale) among very premature infants using the THOP trial data.
Linear regression was used for analysis on correlates of baseline TH levels and mixed
model for correlates of postnatal TH levels.

Adjusted analyses showed that neonatal baseline FT4 levels related to GA
(percentage change; p-value: +l 7%/wk; <0.01), fetal growth ratio (~6%/0. 1; <0.01), race
(Black vs. Hispanic +30%; <0.01), multiple birth (-l4%; 0.06), and intrauterine grth
restriction (-18%; 0.06), while TT4 levels related to GA (+16%/wk; p<0.01), antenatal
steroid use (+22%; 0.03), magnesium use (+14%; 0.06), and prolonged rupture of
membrane (PROM) (+32%; <0.0l). Postnatal FT4 levels were not signiﬁcantly
associated with any factor examined. Postnatal TT4 levels on day 3-21 were all
signiﬁcantly lower than on day 56, ranging from 33-46% (p<0.01). Postnatal TT4 levels
also were associated with GA (+12%/wk; <0.01), duration of ventilator use (-5.6%/l Oday;
<0.01), and necrotizing enterocolitis (NBC) (-53%; 0.01).

With the exception of GA, different groups of factors were correlated with F T4
and TT4 levels in extremely premature infants. Baseline FT4 was more closely related to
ﬁxed attributes such as multiple birth, race, and fetal growth, while baseline TT4 was
more closely related to pregnancy conditions and treatments, such as antenatal steroid
and magnesium use, and PROM. Postnatal FT4 levels were not clearly affected by any
prenatal or postnatal exposures examined. Postnatal TT4 levels increased signiﬁcantly
over postnatal age and were signiﬁcantly associated with GA, duration of ventilator use,

and NEC.

To my beloved parents, Donghai Hong and Tianlan Ma,
and husband, Hong Chen

iv

ACKNOWLEDGEMENTS

On the path to my doctoral degree, I am greatly indebted to many people who
helped and supported me in different times and stages. First, I acknowledge my advisor,
Dr. Nigel Paneth, who guided me through both my Master’s and PhD. Dr. Paneth has
been exceptionally generous to me in his wise teaching, time, encouragement, and
patience. Over six and a half years, he taught me the great values of becoming a truthful
scientist, a passionate researcher, and a dedicated scholar. His persistence and passion in
cerebral palsy research set a role model for me and I will value forever my experience
and friendship shared with him.

I give special thanks to Madeleine and Richard Lenski. Since the start of my new
life at Michigan State University, Madeleine has been a tremendous help and comfort to
me in both research and personal life. She spent countless hours helping me polish my
English in speaking and writing. She encouraged me to explore many new things in life
and shared valuable life experiences and principles with me. She and Richard opened
their hearts and home to me where I learned to embrace another culture (and canine
creature). I am extremely grateful to have such dear friends and be part of their lives.

Another special thanks goes to Dr. Joseph Gardiner. With his wonderful
encouragement and afﬁrmation, I was able to build up my conﬁdence in research and
strengthen my accomplishments. With so many after-work hours and incredible patience,
he guided me through my dissertation analyses step by step. I am deeply indebted to his

mentorship and friendship.

Many have contributed on my path of growth, both individuals and
organizations. I could not list every name here, but express my sincere appreciation to
everyone that helped and loved me.

Lastly, but not least, I could never thank enough my loving parents for their
unconditional love. Without their support, understanding, and sacriﬁce, I could never
walk this far. I also want to thank my cousin, Yunlong Shi, for his love and care while
pursuing my American life, and my dear husband, Hong Chen, for his love, support, and

companionship.

vi

TABLE OF CONTENTS

LIST OF TABLES ' x
LIST OF FIGURES xiii
Chapter 1 Background 1
1.1 Physiology of thyroid hormone ............................................................................ l
1.2 Role of Iodine in thyroid physiology .................................................................... 3
1.3 Ontogenesis of hypothalamic-pituitary-thyroid system ........................................ 5
1.4 Maternal-Fetal thyroid hormone interaction ......................................................... 7
1.5 Postnatal thyroid hormone surge .......................................................................... 7
1.6 Transient hypothyroxinemia of prematurity ......................................................... 9
1.6.1 The different patterns of postnatal thyroid hormone levels in preterm
infants ...................................................................................................... 10
1.6.1.1 Cord blood thyroid hormone levels in preterm infants ..................... 10
1.6.1.2 Postnatal thyroid hormone changes in preterm infants ..................... 12
1.6.1.2.] Postnatal thyroid hormone changes in moderately preterm
infants .................................................................................... 12
1.6.1.2.2 Postnatal thyroid hormone changes in extremely premature
infants .................................................................................... 13
1.6.2 Prenatal factors that affect thyroid hormone levels in preterm infants at
birth ......................................................................................................... 15
1.6.2.1 Intrauterine growth restriction and THOP ........................................ 15
1.6.2.2 Histological chorioamnionitis and THOP ......................................... 16
1.6.2.3 Placenta insufﬁciency and THOP ..................................................... 18
1.6.2.4 Maternal thyroid hormone levels and THOP .................................... 18
1.6.2.5 Antenatal steroid treatment and THOP ............................................. 19
1.6.2.6 Other factors and THOP ................................................................... 20
1.6.3 Factors that affect postnatal thyroid hormone levels in preterm infants
................................................................................................................. 20
1.6.3.1 Postnatal illness and THOP .............................................................. 21
1.6.3.2 Medical procedure and THOP .......................................................... 22
1.7 How do thyroid hormones affect brain development? ...................................... 23
1.8 THOP and adverse neurodevelopmental outcomes .......................................... 24
1.9 Neurodevelopmental outcomes from randomized clinical trials of supplemental
thyroxine for THOP .......................................................................................... 25
SUMMARY OF CHAPTER ONE ........................................................................... 28
BIBLIOGRAPHY .................................................................................................... 48
Chapter 2 Methodology 57
2.1 Study Design ...................................................................................................... 57
2.2 Inclusion and exclusion criteria .......................................................................... 58
2.3 Adverse events .................................................................................................... 59

vii

2.4 Study Population ................................................................................................. 60

2.4.1 Early termination of group 2 ................................................................... 60
2.4.2 Unexpected dosing and coding errors in Madrid site ............................. 61
2.4.2.1 Impact of double dosing error and coding error in Madrid on study
group assignment ............................... - ............................................... 6 2
2.4.2.2 Regrouping procedure ....................................................................... 62
2.4.2.3 Termination of enrollment ................................................................ 62
2.4.2.4 Impact of the errors on study sample size ......................................... 63
2.5 Collection of biological specimens ..................................................................... 63
2.5.1 Blood specimens ..................................................................................... 63
2.5.2 Urine and breast milk specimens ............................................................ 64
2.5.3 Storage, transportation, and assessment of biological specimens ........... 64
2.6 Exposure of interest ............................................................................................ 65
2.7 Outcomes of interest ........................................................................................... 65
2.8 Statistical analysis ............................................................................................... 66
BIBLIOGRAPHY ..................................................................................................... 71
Chapter 3 Results: Population characteristics 72
3.1 Recruitment ........................................................................................................ 72
3.2 Comparison of consented and refused subjects .................................................. 72
3.3 Study cohort ........................................................................................................ 73
3.4 Maternal characteristic of the THOP cohort ....................................................... 74
3.5 Neonatal characteristic of the THOP cohort ....................................................... 75

Chapter 4 Correlates of thyroid hormone levels at birth among extremely

premature infants 81
4.1 Analytical sample ................................................................... ‘ ............................. 82
4.2 Exposures and outcomes .................................................................................... 82
4.2.1 Maternal exposures .................................................................................. 82
4.2.2 Neonatal exposures .................................................................................. 83
4.2.3 Log transformation of thyroid hormone levels ........................................ 84
4.3 Model selection approach ................................................................................... 84
4.4 Results ................................................................................................................. 85
4.4.1 Neonatal and maternal characteristics ..................................................... 85
4.4.2 Neonatal and maternal hormone levels .................................................... 86
4.4.3 Unadjusted association between neonatal exposures and neonatal baseline
thyroid hormone levels ........................................................................... 87
4.4.4 Unadjusted association between maternal exposures and neonatal baseline
thyroid hormone levels ........................................................................... 88
4.4.5 Correlates of neonatal F T4 level at birth ................................................. 88
4.4.6 Correlates of neonatal TT4 level at birth ................................................. 89
4.5 Discussion ........................................................................................................... 90
4.5.1 Correlates of neonatal baseline FT4 in premature infants between 24-28
weeks of gestational age ......................................................................... 91
4.5.2 Correlates of neonatal baseline TT4 in premature infants between 24-28
weeks of gestational age ......................................................................... 92

viii

4.5.3 Relationship between neonatal baseline thyroid hormone levels and

maternal thyroid hormone levels ............................................................ 93
4.6 Strength and limitation ........................................................................................ 94
4.7 Conclusion .......................................................................................................... 95
BIBLIOGRAPHY ...................................................... - ............................................. 1 16
Chapter 5 Correlates of postnatal thyroid hormone levels among extremely
premature infants born between 24-28 weeks of gestational age 118
5.1 Analytical sample .............................................................................................. 119
5.2 Exposures .......................................................................................................... 119
5.3 Outcome ............................................................................................................ 121
5.4 Statistical analyses ............................................................................................ 121
5.5 Results ............................................................................................................... 122
5.5.1 Prenatal and birth infant characteristics of 47 enrolled subjects ........... 122
5.5.2 Maternal characteristics of 47 enrolled subjects ................................... 122
5.5.3 Postnatal characteristics of 47 study subjects ....................................... 122
5.5.4 Neonatal baseline thyroid hormone levels of 47 study subjects ........... 123
5.5.5 Postnatal thyroid hormone levels of 47 study subjects ......................... 124
5.5.6 Correlates of postnatal FT4 levels in 47 study subjects ........................ 124
5.5.7 Correlates of postnatal TT4 levels in 47 study subjects ....................... 125
5.6 Discussion ......................................................................................................... 126
5.7 Strengths and limitations ................................................................................... 129
5.8 Conclusion ........................................................................................................ 130
BIBLIOGRAPHY ................................................................................................... 146

ix

LIST OF TABLES

Table 1.1 Percentages of transient hypothyroxinemia of prematurity .......................... 29
Table 1.2 Summary of studies reviewed in tables 1.3-1.11 .......................................... 30

Table 1.3 Cord FT4 (pmol/l) levels (mean i: SD) by gestational age groups reported by
different studies ............................................................................................................... 32

Table 1.4 Cord TT4 (nmol/l) levels (mean i SD) by gestational age groups reported by
different studies ............................................................................................................... 32

Table 1.5 Cord TSH (mU/l) levels (mean + SD) by gestational age groups reported by
different studies ............................................................................................................... 32

Table 1.6 Postnatal serum F T4 (nmol/l) levels (mean :1: SD) over postnatal age by
gestational weeks in moderately premature infants (29-36 completed weeks of GA) ...33

Table 1.7 Postnatal serum T4 (nmol/l) levels (mean 3: SD) over postnatal age by
gestational weeks in moderately premature infants (29-36 completed weeks of GA) ...34

Table 1.8 Postnatal serum TSH (mU/l) levels (mean i: SD) over postnatal age by
gestational weeks in moderately premature infants (29-36 completed weeks of GA) ...36

Table 1.9 Postnatal serum FT4 (pmol/l) levels (mean + SD) over postnatal age by
gestational weeks in extremely premature infants (§ 28 completed weeks of GA) ...... 37

Table 1.10 Postnatal serum T4 (nmol/l) levels (mean :t SD) over postnatal age by
gestational weeks in extremely premature infants (§ 28 completed weeks of GA) ...... 38

Table 1.11 Postnatal serum TSH (mU/l) levels (mean + SD) over postnatal age by
gestational weeks in extremely premature infants (E 28 completed weeks of GA) ...... 39

Table 1.12 Thyroid hormone levels (mean i: sem) by prenatal factors that are
signiﬁcantly correlated with newborn thyroid hormone levels at birth .......................... 39

Table 1.13 Comparison between two studies on the association between histological
chorioamnionitis and THOP ........................................................................................... 40

Table 1.14 Association between neonatal TSH (mU/l) levels (mean :t sem) and postnatal
exposures in preterm infants ........................................................................................... 40

Table 1.15 Association between neonatal T4 (nmol/l) levels (mean i sem) and postnatal
exposures in preterm infants ........................................................................................... 41

Table 1.16 Adverse neurodevelopmental outcomes of children with THOP ............... 42

Table 1.17 Study designs of six randomized clinical trials of thyroxine supplement
treatment ......................................................................................................................... 43

Table 1.18 Findings of six randomized clinical trials treating THOP infants with

thyroxine supplement ...................................................................................................... 43
Table 2.1 The six THOP arms ....................................................................................... 68
Table 2.2 Inclusion and exclusion criteria of the THOP trial ........................................ 68

Table 2.3 Deaths (total, < 7 days, and i 7 days) and mortality rates (total, < 7 days, and

.—>_— 7 days) by group assignment in the THOP trial as of July 21, 2006 .......................... 69
Table 2.4 The study group coding difference between Madrid site and the THOP triaé9
Table 2.5 Impact of Madrid double dosing and coding error on treatment groups ...... 69
Table 3.1 Reasons for declining or exclusion ............................................................... 77
Table 3.2 Characteristics of included and excluded subjects ....................................... 77
Table 3.3 Characteristics of the THOP cohort .............................................................. 78
Table 3.4 Treatment group assignment of the THOP cohort ........................................ 78
Table 3.5 Maternal characteristics of the THOP cohort ............................................... 79
Table 3.6 Neonatal characteristics of the THOP cohort ............................................... 80
Table 4.1 Neonatal and maternal characteristics of enrolled subjects ........................ 110
Table 4.2 Neonatal and maternal thyroid hormone levels at birth .............................. 110
Table 4.3 Neonatal thyroid hormone levels at birth by study site .............................. 1 11
Table 4.4 Maternal thyroid hormone levels by study site ........................................... 111

Table 4.5 P-values of pair-wise comparison between sites on thyroid hormone levels
....................................................................................................................................... 1 11

Table 4.6 Unadjusted associations between neonatal exposures and baseline serum Free
T4 and Total T4 levels (log-transfonned) ..................................................................... 112

xi

Table 4.7 Unadjusted associations between maternal exposures and baseline serum Free

T4 and Total T4 levels (log-transfonned) ..................................................................... 113
Table 4. 8 Adjusted association between matemal/neonatal exposures and baseline serum
Free T4 levels (log-transformed) .................................................................................. 114
Table 4.9 Adjusted association between matemal/neonatal exposures and baseline serum
total T4 levels (log-transformed) .................................................................................. 114
Table 4.10 Possible variations of FT4 and TT4 introduced by various time of collection
....................................................................................................................................... 115
Table 5.1 Neonatal and maternal characteristics of 47 study subjects ....................... 142
Table 5.2 Postnatal characteristics of 47 study subjects ............................................. 142
Table 5.3 Neonatal baseline thyroid hormone levels by study centers ....................... 143
Table 5.4 Neonatal baseline thyroid hormone levels by treatment groups ................. 143

Table 5.5 Correlates of postnatal FT4 levels in 47 preterm infants born between 24 and
28 completed weeks of gestational age ......................................................................... 144

Table 5.6 Correlates of postnatal TT4 levels in 47 preterm infants born between 24 and
28 completed weeks of gestational age ......................................................................... 145

xii

LIST OF FIGURES

Figure 1.1 The metabolism of human iodothyronines ................................................... 44

Figure 1.2 Cord serum FT4 levels by gestational age (plot based on data cited from

Williams et a1. 2005 (63)). (1 11le = 12.87 pmol/l) ..................................................... 45
Figure 1.3 Cord serum T4 levels by gestational age (plot based on data cited from
Williams et a1. 2005 (63)). (1 ug/dl = 12.87 nmol/l) ..................................................... 46
Figure 1.4 Cord serum TSH levels by gestational age (plot based on data cited from
Williams et a1. 2005 (63)) ............................................................................................... 47
Figure 2.1 Flow chart of the THOP enrollment .............................................................. 70
Figure 4.1 Flow chart of analytical sample selection ................................................... 96

Figure 4.2 Correlation between log-transformed FT4 and FGR in infants with IUGR

......................................................................................................................................... 97
Figure 4.3 Correlation between log-transformed FT4 and FGR in infants without IUGR
......................................................................................................................................... 98
Figure 4.4 Log-transformed FT4 distribution over FGR ............................................... 99
Figure 4.5 Mean FT4 (log-transformed) by FGR 10th percentile ........................... 100
Figure 4.6 Correlation between maternal and neonatal FT4 (log-transfonned) by

gestational week among 144 subjects ........................................................................... 101
Figure 4.7 Correlation between maternal and neonatal TT4 (log-transfonned) by

gestational week among 144 subjects ........................................................................... 104
Figure 4.8 Correlation between maternal and neonatal TSH (log-transformed) by

gestational week among 144 subjects ........................................................................... 107
Figure 5.1 Flow chart of analytical sample selection ................................................. 132

Figure 5.2 Mean structure of FT4 levels over postnatal age in 47 study subjects......133

Figure 5.3 Mean structure of F T4 levels over postnatal age in 47 study subjects by
treatment groups ............................................................................................................ 134

Figure 5.4 Mean structure ofTT4 levels over postnatal age in 47 study subjects ...... 135

xiii

Figure 5.5 Mean structure of TT4 levels over postnatal age in 47 study subjects by
treatment groups ............................................................................................................ 136

Figure 5.6 Mean structure of TSH levels over postnatal age in 47 study subjects ..... 137

Figure 5.7 Mean structure of TSH levels over postnatal age in 47 study subjects by
treatment groups ............................................................................................................ 138

Figure 5.8 Mean structure of FT4 levels over postnatal age in 12 study subjects enrolled
from Amsterdam site ..................................................................................................... 139

Figure 5.9 Mean structure of FT4 levels over postnatal age in 12 study subjects enrolled
from Amsterdam site by treatment groups .................................................................... 140

Figure 5.10 Mean structure of TT4/T30 and TBG levels over postnatal age in 47 study
subjects .......................................................................................................................... 141

xiv

Chapter 1 Background

In the past three decades, the development of laboratory techniques has improved
our understanding of perinatal thyroid physiology. However, as the survival of premature
infants increases, we are facing the new challenge of transient hypothyroxinemia of
prematurity, which is manifest by relatively low serum free thyroxine (T4), normal serum
thyroid stimulating hormone (TSH). Literature on premature thyroid is sparse, especially
on the correlates of transient hypothyroxinemia in prematurity newborns and how
premature thyroid reacts to postnatal life. Thus this dissertation aims to answer some of
these questions using data collected in a randomized clinical trial.

In this chapter, brief overview of thyroid hormones is provided, covering the
physiology of thyroid hormones, the role of iodine in thyroid physiology, and the
ontogenesis of the hypothalamic-pituitary-thyroid system. Then a brief description of
thyroid function changes in healthy term infants is given, followed by a review of the
changing patterns of thyroid hormones and prenatal/postnatal correlates of transient
hypothyroxinemia in premature infants. At the end, a summary discusses associations of
transient hypothyroxinemia with neurodevelopmental outcomes in premature infants, and
the rationale for attempting to discover whether there are beneﬁcial effects of thyroxine
supplements in this population.

1.1 Physiology of thyroid hormone (Figure 1.1)
The thyroid gland produces a variety of thyroid hormones, also called

iodothyronine, to provide the daily needs of the human body. Iodothyronine molecules

differ from each other by the number of iodine atoms and positions on the benzene-ring.
The major iodothyronines are:

1) 3, 5, 3’, 5’-L-tetraiodothyronine (T4),

2) 3, 5, 3’-triiodo-L-thyronine (T3),

3) 3, 3’, 5’-triiodothyronine (reverse T3 (rT3)),

4) 3, 3’-diiodothyronine (T2). Conversion among iodothyronines is catalyzed by
human iodothyronine selenodeiodinases.

Among all iodothyronines, T4 has the highest concentration in plasma and is the
only thyroid hormone exclusively secreted by the thyroid gland. T3 is also secreted by
the thyroid gland, but 80% of T3 (and almost all rT3) is produced by conversion from T4
in the liver, kidney, brain, skeletal muscle, and adipose tissue (1-4). In the circulation,
there are two forms of thyroid hormone: free and bound. Because free thyroid hormones
degrade rapidly, most T4 and T3 molecules are transported to target organs bound to
thyroid binding globulin (TBG). However, since only free thyroid hormone can be taken
up into the cell, bound T4 and T3 serve as a thyroid hormone reservoir to maintain
thyroid hormone homeostasis (1).

Transporting free thyroid hormone molecules into cells is an energy-dependent
process, mediated by the Na+/taurocholate cotransporting polypeptide and L-type amino
acid transporters. Free T3 (FT3) molecules enter cells and bind to speciﬁc nuclear DNA-
bound thyroid hormone receptors (TR) to regulate gene expression (5). With a 15-fold
higher binding afﬁnity to TR’s than T4, T3 is the major biologically active thyroid
hormone. Intracellular FT3 can either enter the nucleus and bind to the TRs, or be

degraded to T2 by type-3 selenodeiodinase (D3) (1). The intracellular free T4 (FT4)

molecule becomes active T3 by having a single 5’-iodine atom detached by type 2 (D2)
human iodothyronine selenodeiodinase. FT4 can also be converted to inactive rT3 by
removing a single S-iodine atom, a reaction catalyzed by D3. The activation and
degradation pathways of iodothyronines are illustrated in Figure 1.1.

l. .2 Role of Iodine in thyroid physiology

Iodine is an essential element in thyroid hormone synthesis and metabolism and,
in humans, is completely dependent on dietary intake. Iodine deﬁciency is often seen in
mountain area and lowland regions far from the oceans, such as central Africa, northern
China, Himalayas area, India, many countries in Europe, and South America, etc. Those
who consume only locally produced foods in these areas are at great risk for Iodine
deﬁciency.

When iodine supply is reduced, the human body takes compensatory reactions to
maximally conserve iodine and improve the efﬁciency of iodine use. These adjustments
occur at the hypothalamic, pituitary, thyroid, and peripheral tissue levels. For instance,
during iodine deﬁciency, a plasma T4 decrease may increase D2 expression in the central
nervous system (CNS), hypothalamus, and pituitary gland, which promotes the
conversion of T4 to T3. D3 expression in the CNS may decrease to allow T3 to stay
active for relatively a longer period in the face of iodine deﬁciency (6-8). This
preference for T3 during iodine deﬁciency is physiologically more efﬁcient as T3
consumes one iodine atom less than T4, but has a three-fold higher biological potency (1).

In human adults, when total iodine intake is less than 75 pg/day, the above
compensatory adjustments in thyroid function will be activated. Similar physiological

adjustments have been found in disorders in which there is difﬁculty utilizing iodide,

such as Hashimoto’s disease and Grave’s disease during antithyroid drug treatment
treatment (9, 10). Despite the compensatory mechanisms to help the human body to
preserve iodine and increase the efﬁcacy of iodine utilization, prolonged and severe
iodine deﬁciency will eventually lead to clinical hypothyroidism.

Since insufﬁcient iodine can cause hypothyroidism, one might assume that
excessive iodine would induce hyperthyroidism. However, the contrary is true.
Excessive iodine exposure inhibits thyroid hormone production and may cause transient
hypothyroxinemia. Prolonged excessive exposure to iodine places the thyroid gland at
risk of goiter or hypothyroidism. With iodide overload, the thyroid gland reacts ﬁrst by
increasing, and then by decreasing organiﬁcation of iodide, ultimately producing a
reduction in iodothyronine synthesis (11). This decrease intrathyroidal organiﬁcation of
iodide is known as the Wolﬂ-ChaikoﬂEﬂect. The decreased iodide organiﬁcation is
thought to result from the overly high concentration of inorganic iodide within the
thyroid gland, which possibly interferes with the effect of thyroid peroxidase (TPO).
Aﬁer experiencing the Wolﬂ-Chaikoﬂ Effect for a short time period, the normal thyroid
gland partially regains the ability to uptake iodide into the cell for iodothyronine
production. This process lasts until the iodide level falls below the threshold for
maintaining the Wolﬂ-ChaikoffEﬂect. This adaptation or escape phenomenon prevents
the development of hypothyroidism and iodide goiter. However, if excessive iodine
intake continues and iodine levels stay higher than the threshold, goiter or
hypothyroidism will eventually develop. Although this escape mechanism exists in fetus
or premature infants is not conﬁrmed, cases of infants born with goiters large enough to

cause respiratory distress have been reported as a result of maternal iodide ingestion (12-

14). Potassium iodide, once commonly used for cough treatment, was thought to be one
of the most frequent culprits (15). Thus prolonged overload iodine intake during
pregnancy or infancy (for preterm infants) may cause hypothyroidism or goiter and
should be avoided as much as possible (11, 16, 17).

1.3 Ontogenesis of hypothalamic-pituitary-thyroid system

The development of the fetal hypothalamic-pituitary-thyroid (HPT) system
includes three phases: 1) embryogenesis, lI) hypothalamic maturation, and III) maturation
of hypothalamic-pituitary-thyroid system function (18, 19). Although the time windows
overlap with each other, distinct differences exist in anatomy, histology, biochemistry,
and biological function across the three phases.

Phase I : embryogenesis of the pituitary and thyroid gland occurs during the ﬁrst
trimester. The thyroid embryo anlage becomes recognizable as early as one month after
conception and the embryogenesis of thyroid gland is the mostly completed by 10 to 12
weeks of gestational age (GA) (1, 20). Toward the end of this periOd, tubule-like
structures are seen in the fetal thyroid, soon after which the follicular structure appears
and then the follicles are ﬁlled with colloid. While the thyroid structure develops, the
fetal thyroid develops the ability to concentrate iodide and to synthesize iodothyronines
(1 , 18, 19, 21-22). Meanwhile, the fetal pituitary undergoes histological differentiation
and starts synthesizing TSH by 14 weeks’ gestation. Fetal serum T4, T3, and TSH
become measurable by the end of this phase (23-26), but concentrations remain low until
mid-gestation.

Phase II: hypothalamic maturation overlaps with Phase I. The time window, from

4-5 weeks of gestation to around 30-35 weeks of gestation, is when continuing

development of the hypothalamus and pituitary glands proceeds. Anatomically,
hypothalamic nuclei and ﬁbers of the supraoptic tract become visible by 12 to 14 weeks.
Biochemically, TRH, gonadotropin-releasing hormone, and somatostatin reach
measurable levels in hypothalamic tissue by 12 weeks, indicating that the hypothalamus
can synthesize hormones by then. From the 10-12th to the 35th week, hypothalamic
maturation continues as the pituitary-portal vascular system matures (18, 19).

Phase III: the elevation of fetal serum T4 concentrations marks the beginning of
the maturation of the hypothalamic-pituitary-thyroid system. Around 18-20 weeks’
gestation, pituitary and serum TSH begin to rise. Serum TSH concentrations peak early
in the third trimester and decline slowly from about 15 uU/ml at 30 weeks to 10 pU/ml at
term (27-29). F etal serum T4 and FT4 levels gradually climb up in the third trimesters
until term.

More importantly, the hypothalamic-pituitary-thyroid negative feedback system
starts regulating thyroid hormone production during this period. Infants born after 26 to
28 weeks of GA have been found to respond to exogenous TRH injections (40 pg) with
serum TSH elevations (30, 31). Also, increased cord and serum T4 levels after intra-
amniotic injection of T4 (700 pg) greatly suppressed both cord TSH levels and neonatal
TSH surges in term infants (32). These up and down-regulations of thyroid hormone
levels are mediated through hypothalamic stimulation of pituitary TSH secretion and
subsequent stimulation of the thyroid gland (19). The hypothalamic-pituitary-thyroid

feedback matures with gestational age and HPT development is complete at term (28).

1.4 Maternal-Fetal thyroid hormone interaction

The fetal thyroid hormone supply relies a great deal on maternal sources
throughout pregnancy, especially the ﬁrst two trimesters. It is well known that
transplacental passage of TSH from mother to fetus is negligible, but maternal T4 can
cross the placental barrier and freely enter the fetal circulation. T4 has thus been detected
in coelomic and amniotic ﬂuids prior to the onset of fetal thyroid function (22).
Moreover, measurable cord T4 levels are found in newborns with congenital
hypothyroidism (33). With increasing tissue needs resulting ﬁom fetal growth, maternal
thyroid hormone production rises as pregnancy proceeds. In the late second or early third
trimester, the source of fetal thyroid hormone supply gradually shifts, however, from
complete maternal dependence to partial maternal dependence, because of maturation of
fetal thyroid function and of hypothalamic-pituitary-thyroid regulation.

Can the fetus suffer hyperthyroxinemia when the mother transfers too much
thyroid hormone to the fetus? Most likely no, because large amount of type 3
monodeiodinase existed in placenta tissue, possibly in uterus as well, which prevents the
fetus from developing hyperthyroxinemia by increasing the degradation of T4 to rT3 or
T3 to T2. Nonetheless, there have been reports on neonatal thyrotoxicosis secondary to
maternal hyperthyroidism (34, 35). This could be due to placental transfer of
immunoglobulin, or possibly abnormality of neonatal thyroid.

1.5 Postnatal thyroid hormone surge

In term neonates, serum TSH levels surge immediately after birth and usually

peak at 30 minutes of postnatal life. A sharp decrease in serum TSH appears

immediately aﬁer the peak in the ﬁrst few hours, followed by a much slower decline until

TSH drops signiﬁcantly below cord level at about 48-72 hours after birth (36). Serum T4
levels rise to respond to the TSH elevation, reaching an apex at 24 hours after birth, and
then slowly decreasing to a plateau over the ﬁrst week of life (37-41). Serum T3
increases in a similar fashion to T4, except that its rise is not a direct resulted of the TSH
boost, but rather is produced by 1) reduced T3 degradation due to sudden loss of D3 from
placenta tissue, 2) decreased D3 expression in newborn liver and brain tissues, and 3)
increased peripheral conversion of T4 to T3 shortly after birth (19, 37, 41).

The acute TSH surge after birth is thought to be stimulated by the cold exposure
in the extrauterine environment, and the subsequent thyroid hormone elevations are
known to be responsible for non-shivering thermoregulation in newborns. The thyroid
hormones inﬂuence the brown adipose tissue (BAT) in heat production through many
physiologic pathways, such as by increasing uncoupling protein activity in BAT,
controlling lipid composition and the size of brown adipocyte, and increasing expression
of proteins involved in brown adipocyte metabolism (oxidation). All of these
mechanisms promote the production of high energy phosphate bonds that are ultimately
converted to heat in BAT (42). Results from many animal experiments support the role
of thyroid hormone in non-shivering thermoregulation in neonates. In addition to the
thyroid hormone surge triggered by cold exposure, it has been found that thyroid
hormone levels are positively related to rectal temperature, and that higher body
temperature are observed in euthyroid animals compared to animals with hypothyroidism
(43-46).

However, premature infants, especially extremely low gestational age infants (GA

< 28 completed weeks), are expected to have immature hypothalamic-pituitary-thyroid

system and less fat tissue. lmmaturity affects the thyroid hormone surge owing to the
lack of capacity to synthesize hormone and/or the lack of effectiveness of HPT hormone
regulation. The latter deﬁcit affects the thyroid hormone surge by reducing intracellular
conversion of T4 to T3 in BAT, which is rich in D2. A5 a result, the brown adipocyte is
smaller, has an altered lipid composition. There is a lower circulating free fatty acid
concentration, and eventually less heat production (18, 47-49).

1.6 Transient hypothyroxinemia of prematurity (Table 1.1)

Transient hypothyroxinemia of premature (THOP) infants is manifest by
relatively low free T4, normal serum TSH, and normal responses of TSH and T4 to TRH
(19). In general, THOP is a laboratory based diagnosis and is mostly asymptomatic.
However, signs of hypothyroidism, such as hypoactivity, lethargy, poor feeding,
constipation, jaundice, edema, and hoarse cry, have been observed in preterm infants with
THOP (50). Although these signs are not speciﬁc to hypothyroidism, the clinical
response to thyroxine replacement therapy, exclusion of other possible causes in addition
to the infants’ low serum T4 levels and normal TSH levels, have strongly suggested that
hypothyroxinemia was likely the cause. With increasing postnatal age, serum thyroid
hormone levels of preterm infants increase and catch up with term infants by 3-8 wks
postnatal age. The temporary presence of low thyroid hormone levels is the reason for
use of the term transient hypothyroxinemia.

Frequencies of THOP reported in different preterm cohorts have varied because
of differences in l) the gestational age groups studied, 2) the postnatal ages when tested,
3) the deﬁnition of transient hypothyroxinemia applied, and 4) the sample sizes. Table 1

summarizes published frequencies of THOP (50-59). Although variations exist across

studies, the trend is clear in showing that the THOP rate decreases dramatically as
gestational age increases. Within the same gestation group, the THOP rate declines over
postnatal age, although the effect of postnatal growth on thyroid hormone levels is not as
strong as that of gestational maturation.

1.6.1 The different patterns of postnatal thyroid hormone levels in preterm infants

In the past three decades, advances in laboratory techniques have improved our
knowledge of the physiology of the thyroid hormones. With the broad use of surfactant,
prenatal steroids, and the development in neonatal intensive care, more and more very
premature newborns survive. The physiology and pathophysiology of these premature
thyroid glands and HPT regulation systems have been studied, but are still not well
understood. This section provides a brief summary of the changing patterns of postnatal
thyroid hormones and TSH concentrations in preterm infants, especially extremely
premature infants, and the differences from term infants.
1.6.1.1 Cord blood thyroid hormone levels in preterm infants (Table 1.3-1.5; Figure
1.2-1.4)

At birth, preterm infants have low cord serum thyroid hormone levels. Many
studies have reported signiﬁcantly lower cord FT4, TT4, and TSH levels in preterm
infants than in term infants (60-63). The baseline hormone level differences between
preterm and term infants diminish as gestational age increases (64). Tables 1.3-1.5
summarize the cord FT4, TT4, and TSH levels reported in different populations by
different gestational age groups (2, 37, 54, 59, 65, 66). FT4 levels had slight increase
over gestational age with the highest values appearing at 36 weeks. TT4 levels increase

steadily along with gestational age. TSH levels remain nearly ﬂat over gestational age.

10

One Japanese study reported TSH levels being highest around 30-34 weeks, but the same
trend was not seen among other studies (66).

Williams et al. presented the cord thyroid hormone and TSH levels of 620 infants
in a cross-sectional fashion. Study infants were born between 23 and 42 weeks of GA
and enrolled from 11 Scottish neonatal intensive care units, as part of a cohort study of
transient hypothyroxinemia (63). Cord thyroid hormone levels of every single gestational
week were reported and results were difﬁcult to be included in table 1.3-1.5 because
other studies reported thyroid hormone levels by clustered gestational age groups. To
visualize the changing pattern, we have converted the thyroid hormone levels reported by
Williams into graphs. Figure 1.2-1.4 illustrate the approximate changing pattern of cord
FT4, TT4, and TSH with gestational age. FT4 reached the highest value around week 33
and then decreased thereafter. The turning point in this graph appeared much earlier than
what is observed in table 1.3, and the degree of decrement is much greater in William’s
population. These differences were probably caused by three factors. First, clustered
gestational age groups (table 1.3) can not show trends as clearly as individual gestational
weeks (ﬁgure 1.2). Second, the sample sizes in gestational weeks 35, 36, and 37 were 4,
9, and 10 respectively in Williams’ report. Small sample sizes might introduce variation
and might not accurately reﬂect the hormone levels of infants born within these
gestational weeks. Third, infants in Williams’ report were all admitted to neonatal
intensive care units, implying that they had different levels of illness. Their illness levels
might be related to their thyroid hormone levels and the relation is likely to be stronger in
term than in preterm infants. Patterns seen in both TT4 and TSH, in the Williams’ data,

are in agreement with those observed in table 1.4.

11

1.6.1.2 Postnatal thyroid hormone changes in preterm infants

After birth, as aforementioned, the TSH levels of term infants rise quickly and
reach an apex value within 30 minutes. Serum iodothyronine levels rise to respond to the
TSH stimulation and climax at around 24 hours after birth. Both TSH and thyroid
hormone levels drop after the peak levels but remain much higher than in adults.

The thyroid function of preterm infants behaves differently from term infants.
According to the maturity level, preterm infants can be categorized into two groups,
moderately preterm group (29-36 completed weeks of GA) and extremely premature
group (less than 28 completed weeks of GA). We will address these two groups
separately.
1.6.1.2.1 Postnatal thyroid hormone changes in moderately preterm infants (Table
1.6-1.8) (2, 54, 58, 64-71)

The moderately preterm group refers to infants born between 29-36 completed
weeks of gestational age. As these neonates are closer to term, their thyroid glands and
HPT regulation systems have gained a certain level of capacity to adapt to the extra-
uterine environment, but the potency of these system is not as great as those of term
infants. Baseline (at birth) thyroid hormones and TSH levels of this group are lower than
those of term infants and are positively associated with gestational age. The TSH surge is
seen in this population and rising T4 and T3 levels are observed in response to the TSH
surge. The surges of TSH and thyroid hormones in moderately preterm infants are

qualitatively similar to those of term infants, but lesser in degree ( 19, 72).

12

1.6.1.2.2 Postnatal thyroid hormone changes in extremely premature infants (Table
1.9-1.11) (2, 54, 57, 61, 64)

The extremely premature group consists of neonates born prior to 28 completed
weeks of gestational age. This population is delivered at the end of the second trimester
or the beginning of the third trimester. This time window coincides with the third phase
of the ontogenesis of the HPT system, during which the fetal thyroid gland and the HPT
regulation system begin to mature and function (see section 1.3) (18, 19). This process is
interrupted by parturition. The coordination between the fetal thyroid gland and the HPT
regulation system is not yet developed at birth. Clinically, TSH levels at baseline are
extremely low in this population and they increase with gestational age. The TSH surge,
is greatly suppressed or even absent among extremely premature infants (2, 61, 73). The
reasons for this absence could include a lack of TRH response to declining thyroid
hormone levels, or an immature HPT system that has limited TRH secretion.
Alternatively, the central nervous system may be too immature to respond to cold
stimulation after delivery (74).

Among extremely premature infants, F T4 concentrations start at a much lower
level compared to moderately preterm or term infants and the FT4 increase is transient
and minimal. In a randomized clinical trial, in which thyroxine supplement was provided
to preterm infants less than 30 weeks of gestational age, Van Wassenaer et al. reported a
small, insigniﬁcant increase (0.3 ng/dl) of the mean serum FT4 levels on day 1, followed
by a signiﬁcant decrease with a nadir appearing on day 7 among 100 preterm newborns
in the placebo group (75). In another large randomized clinical trial which examined the

effect of T3 and hydrocortisone on lung disease in very premature infants (23-29 weeks),

13

no thyroid hormone surge was seen in 128 subjects randomized to the placebo group (61).
In addition, the thyroid hormone concentrations barely changed over the ﬁrst two weeks
of life in this study cohort. A study of 22 infants born between 24 and 27 weeks of
gestational age, whose thyroid hormone levels were measured in cord blood and at 1, 7,
and 24 hours after birth, found small and transient increases consistent with other studies
(2)-

The total T4 increase is also greatly attenuated in these very low gestational age
newborns. Murphy et al. reported severely reduced TT4 elevation after birth in neonates
born between 24 and 27 gestational weeks (2). Hadeed et al. also found infants less than
1000 grams (GA < 30 wks) had small degrees of TT4 elevation. Table 1.7 summarizes
the previously reported serum TT4 values in the ﬁrst 3 weeks of life in different
gestational age groups. Cross-sectionally, TT4 levels increase with gestational age;
longitudinally, TT4 increases slightly by postnatal age, but with great variation across
different studies.

Interestingly enough, baseline rT3 levels (the inactive form of thyroid hormone)
remain much higher in extremely premature infants than in both moderately preterm and
term infants (2, 37, 68, 71). Cord rT3 levels decrease with increasing gestational age.
After birth, the serum rT3 levels gradually decrease to the same levels as term infants.
rT3 is also known as the major iodothyronine form in the fetus. This similarity between
extremely premature infants and fetuses strongly suggests that thyroid function and HPT
regulation are not functioning, only minimal functioning, in this population.

In summary, both baseline and postnatal thyroid hormone levels are low in

preterm infants. Hormone changes regularly seen in term infants, especially the hormone

l4

surges after birth, are suppressed in preterm infants. The degree of attenuation is the
greatest in extremely premature newborns, and sometimes the surges even disappear.
Transient hypothyroxinemia of prematurity reﬂects a delay in maturation of both thyroid
gland function and hypothalamic-pituitary-thyroid regulation system. The maturity of
newborn, namely gestational age, plays the most important role in this process.
1.6.2 Prenatal factors that affect thyroid hormone levels in preterm infants at birth
Besides prematurity, there are other prenatal factors that may affect the thyroid
function of preterm infants at birth, such as intrauterine growth restriction,
chorioamnionitis, placenta insufﬁciency, antenatal steroid use, etc. In this section we
brieﬂy summarize most of the prenatal factors previously discussed in the literature.
1.6.2.1 Intrauterine growth restriction and THOP (Table 1.12) (58, 60, 76, 77)
Intrauterine growth restriction (IUGR) is associated with transient
hypothyroxinemia in preterm infants. Many studies have compared the thyroid hormone
levels of small for gestational age (SGA) infants with those of apprOpriate for gestational
age (AGA) infants at birth or during neonatal period. Consistently low thyroid hormone
levels have been found in SGA infants independent of gestational age (41, 58, 71, 78).
However, in light of our limited knowledge of fetal thyroid development, and our
restriction to observational data from previous reports, it is hard to determine the causal
relationship between IUGR and THOP. Three mechanisms may explain the association
between IUGR and THOP. First, thyroid hormone is known to be critical in fetal growth.
In athyrosis cases or when fetal thyroid development was interrupted, intrauterine
hypothyroidism has been known to cause fetal growth restriction. Second, in fetus with

growth restriction from intrauterine insults, the thyroid gland could be one of the affected

15

organs and hypothyroxinemia might develop. Last, a common risk factor may affect both
fetal growth and thyroid gland development or function.
1.6.2.2 Histological chorioamnionitis and THOP (Table 1.12-1.13) (58, 60, 76, 77)

It is worth mentioning an interesting ﬁnding about the association of histological
chorioamnionitis (HCA) and low thyroid hormone levels in preterm infants. De Felice et
al. examined placentas of 155 preterm newborns (GA 29.36 3: 2.39 wks) and measured
the thyroid hormone levels of these infants on postnatal day 4 and day 40 (76). They
found that after adjustment for birthweight and respiratory distress syndrome, having
HCA increased the risk of having ﬁrst week TT4 (day 4) more than 1 standard deviation
below the mean by 32-fold (95% CI 8.9-115.6). Although this is the only study, to our
knowledge, designed for testing the relationship between asymptomatic HCA and
hypothyroxinemia, many other groups repeatedly have established the signiﬁcant impact
of intrauterine infection on elevated risks of preterm delivery and other neonatal
morbidities of the preterm infants (79-81). De Felice’s finding suggests that THOP may
be one of many consequences of a fetal inﬂammatory response to intrauterine infection.

However, in another study whose primary aim was to examine the association
between antenatal glucocorticoids and THOP, the opposite relationship was discovered
between chorioamnionitis (CA) and THOP. The unadjusted analyses showed that the
median TT4 level of 200 infants with CA was signiﬁcantly higher than that of 100 infants
without CA (TT4 6.6 vs. 6.2 ug/dl, p=0.02). Although whether this relationship
remained signiﬁcant in an adjusted model was not indicated in this report, it draws

attention to the unsettled effect of CA on THOP.

l6

Except for comparable prevalence rates of CA (63.2% (98/155) vs. 66.7%
(200/300)), we found several features in study designs and populations that differed and
which may explain their contradictory results (table 1.13). The most important difference
is that De Felice et al. included IUGR newborns, but Martin et al. did not. As noted above,
SGA infants have lower thyroid hormone levels than do AGA infants. CA is known to
be a risk factor for IUGR (81). Excluding IUGR subjects could artiﬁcially elevated TT4
levels in the chorioamnionitis population and attenuate the differences. In addition, De
Felice et al. studied moderately preterm infants, but Martin et al. studied extremely
premature infants. The latter are known to have very low thyroid hormone levels and a
minimal thyroidal response to the extra-uterine environment. Population difference could
thus be a second reason for the different results. Moreover, Martin’s group used the
thyroid hormone results assessed by the State newborn screening program and excluded
all subjects with no or delayed ﬁrst week thyroid hormone values, while De Felice et al.
obtained day 4 thyroid hormone values on every study subject. We know that overall
severity of illness tends to interfere with newborn screening and cause delays in blood
sample collection. According to Martin, the subjects with missing hormone values in
their study were less mature and had a higher mortality rate. We also know that sickness
severity is tightly bound to low thyroid hormone levels and CA may also contribute to the
sickness. In the Martin et al study, there might have been more subjects with CA
encountering missing hormone value problems than subjects without CA. This might
also raise the apparent hormone levels in chorioamnionitic infants. Finally, De Felice et

al. studied only infants with asymptomatic CA, while Martin’s group did not differentiate

17

asymptomatic from symptomatic. It might be that asymptomatic and symptomatic CA
might affect fetuses through different mechanisms.

Since the study conducted by De F elice’s group was designed to study the
association between HCA and THOP among premature infants, their ﬁndings provided us
evidence to believe that HCA may be associated with low thyroid hormone levels in
preterm infants. However, histological chorioamnionitis is a complicated condition and
may affect fetal growth in various ways. For further understanding of this association,
more studies with careful deﬁnition of histological chorioamnionitis need to be
conducted for this speciﬁc hypothesis.
1.6.2.3 Placenta insufﬁciency and THOP (Table 1.12) (58, 60, 76, 77)

Two small studies (n=62 and n=28) compared thyroid hormone levels of
moderately preterm infants born to preeclamptic mothers with “placenta deﬁciency” (PD)
assessed through ultrasonography and mothers without (60, 62). Both studies found low
cord thyroid hormone levels in infants born to preeclamptic mothers with placenta
insufﬁciency (PI). However, most infants born to preeclamptic mothers with P1 in these
two studies had IUGR. The association between PI and THOP thus might possibly have
been mediated by IUGR. Alternatively, THOP could also result from a condition other
than IUGR, primarily induced by P1. For instance, Buimer et al. found that low cord FT4
levels were not related to maternal FT4 levels, but to prenatal acidosis, which was an
outcome of PI (82).
1.6.2.4 Maternal thyroid hormone levels and THOP

It is well accepted that fetal thyroid hormone levels are strongly inﬂuenced by

maternal thyroid function. Will this also be true after delivery? Studies that examined

18

both cord and maternal serum thyroid hormone levels are needed to answer this question.
In a cross-sectional study, Hume et al. measured thyroid hormone levels (FT 4, T4, T3, &
TSH) of 428 mother-infant pairs at different gestational age (GA 15-42 wk) (3 7). The
correlation analyses revealed an overall lack of association between maternal thyroid
hormones and TSH levels at delivery with neonatal cord levels. However, in another
study engaged in newborn thyroid hormone changes in the ﬁrst 24 hours of life, the same
research team found a strong correlation between maternal TSH and neonatal TSH in 72
mother-infant pairs with gestational age between 24 and 34 weeks (2). The relationship
between maternal and neonatal thyroid hormone levels at birth thus remains unclear.
1.6.2.5 Antenatal steroid treatment and THOP

Controversies exist in literature with respect to the association between antenatal
steroid exposure and early neonatal thyroid hormone levels among premature infants.
Franklin et al. failed to ﬁnd any effect of antenatal steroids use on neonatal thyroid
function at birth or during the postnatal period (n=97) (83). But a recent study, in a much
larger preterm population (n=556; GA 23-28 wk), found a complete dose of antenatal
glucocorticoids increased ﬁrst-week mean T4 level by 0.81 ug/dl (p=0.03) (77). Except
for the difference in the number of subjects received antenatal steroids treatment (20 vs.
135), which gave more power in detecting association, the latter study also found much
less illness frequency among newborns receiving antenatal glucocorticoids treatment,
including lower rates of PDA, IVH, etc. These ﬁndings suggested that antenatal steroid
treatment could possibly promote the maturation of multiple organs, including the thyroid
gland, and reduce overall illness severity, both of which are known to be related to

improved serum thyroid hormone levels.

19

1.6.2.6 Other factors and THOP

A number of studies looked at other prenatal risk factors for THOP. One large
study (n=3 65) found that low thyroxine levels at newborn screening were related to older
postnatal age at test (OR=1.6/day, 95% CI 1.3-2.1), and maternal education higher than 12
years was protective (OR=O.4, 95% CI 0.2-1.0) (84). Although not tested in other studies,
we have already learned that illness severity is related to both delay in newborn screening
sample collection and low thyroid hormone levels. This might explain the relation
between older postnatal age at test and lower thyroxine levels. However, the role of
maternal education, a representation of social economic status, is unclear in relation to
neonatal thyroid hormone levels.

Other studies also examined the effects of the following factors on thyroid
hormone levels in preterm infants, such as gender, race, multiple birth, route of delivery,
birthweight < 1500 g, premature rupture of membrane, placenta infarction, gestational
diabetes, prenatal antibiotics/beta-2-sympathicomimetics therapy within 24 hours prior to
delivery, etc. No signiﬁcant association has been detected consistently (65, 68, 78).

1.6.3 Factors that affect postnatal thyroid hormone levels in preterm infants

After birth, postnatal exposures, in addition to residual effects from intrauterine
and labor exposures, add more complexity in revealing the association between postnatal
factors and THOP. As in adults, thyroid hormone levels in preterm infants decrease
dramatically as illness or nutritional deprivation becomes severe. Our limited knowledge
on the pathophysiology of how the neonatal thyroid gland responds to the extra-uterine
environment makes it difﬁcult to reveal the causal relations between neonatal illnesses

and THOP.

20

1.6.3.1 Postnatal illness and THOP (Table 1.14 -l.15) (54, 58)

Impaired lung function is the most commonly seen condition in preterm infants
with THOP. Compared to healthy preterm controls, signiﬁcantly lower cord and
postnatal serum T4 levels have been found in infants who subsequently developed
respiratory distress syndrome (RDS) and in those who suffered RDS, respectively (54, 58,
84). The T4 difference between preterm infants with and without RDS decreases as
gestational age increases, and no T4 difference is found between term infants with and
without RDS.

The association between RDS and THOP is intriguing. Some have suggested that
THOP could be secondary to RDS. However, the ﬁnding of suppressed cord thyroxine
levels in preterm infants who developed RDS later, as well as the lack of a T4 difference
between term infants with and without RDS did not support this hypothesis. The etiology
of this association is more likely to be intrauterine for premature infants. A few
mechanisms may explain the relationship.

First, intrauterine thyroid hormone deﬁciency may lead to a shortage of cortisol, a
key stimulator of pulmonary surfactant synthesis. Lack of cortisol would delay lung
maturation (85), produce surfactant deﬁciency and impose a greater risk of having RDS
in the premature population (58). Second, it could be that the overall immaturity in
premature infants leads to both RDS and THOP. Third, thyroid hormone dependent
pulmonary maturation could be hindered directly by postnatal hypothyroxinemia. The
severity of pulmonary illness could be worsened and its duration prolonged (84). Fourth,

once RDS occurs, it could possibly affect neonatal thyroid function and deteriorate

21

hypothyroxinemia, which in turn could slow down the thyroid hormone dependent
pulmonary maturation and RDS may become worse.

Interestingly enough, although T4 levels decrease in infants with RDS, TSH
showed very little difference between infants with and without RDS within the same
gestation groups(5 8). This ﬁnding suggested that the HPT system was immature for
hormone regulation in preterm infants. It could be that either the hypothalamus does not
respond to serum low thyroxine levels, or the pituitary reduces responding to TRH
stimulation.

Sepsis is another illness linked to THOP. It shares the same attributes with RDS
in relation to THOP, in that premature newborns with sepsis have lower serum thyroxine
levels. The association of sepsis with THOP appears to be independent of gestational age
and TSH is not affected by sepsis (54, 58). THOP has also been found to be related to
intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC), chronic lung disease
(CLD), and other neonatal disorders (4, 55, 84, 86-88).

Not many studies have investigated prenatal or postnatal risk factors of THOP,
and the consistency of published ﬁndings needs to be conﬁrmed by more observational
studies with larger populations and broader risk factors of interest.
1.6.3.2 Medical procedure and THOP

As aforementioned, the thyroid glands of premature infants do not have the ability
to escape from Wolﬂ-ChaikoﬂEjﬁct when exposed to excessive iodine exposure.
Thyroid hormone production will be suppressed and prolonged iodine exposure could
induce permanent hypothyroidism in this population. Topical iodine antiseptics for skin

(before or after surgery), umbilical cord disinfection, and iodine containing contrast

22

media are widely used in neonatal intensive care units. They are the major sources of
excessive iodine exposure for preterm infants. Khashu et al. reported severe
hypothyroxinemia in a premature infant after prolonged use of iodinated skin
disinfectants for a complex skin lesion (16). Weber et al. studied 23 infants exposed to
excessive iodine (89), among which eleven were through maternal iodine exposure
(iodine containing antiseptics used for vaginal application before or after delivery (n=4),
or caesarean section (n=5), or skin disinfectant applied in the second trimester of
pregnancy (n=2)) and twelve through neonatal exposure (topical iodine antiseptics used
for surgical intervention or umbilical cord disinfection (n=7), or contrast media used for
venous catheterization (n=4) or urography (n=1)). They have found that more than 50%
of premature infants who had THOP were exposed to iodinated skin disinfectants or
iodine-containing contrast media. The exposed neonates also had increased TSH levels
and urinary iodine excretion.

These ﬁndings suggest that iodine containing disinfectants, drugs, or contrast
agents should be used with great caution in premature infants, and that TSH monitoring is
advisable for detecting thyroid hormone suppression.

1.7 How do thyroid hormones affect brain development?

Thyroid hormones affect both fetal brain development and long term
neurodevelopment. The former outcome is more related to maternal thyroid hormone
levels during pregnancy since matemal-fetal thyroid hormone transfer is the major source
of fetal thyroid hormone supply. The latter outcome is more associated with neonatal

thyroid function after birth.

23

Three major molecular and cellular level ﬁndings support that thyroid hormone is
essential to fetal brain development. First, thyroid hormone receptors in the brain appear
early in fetal life (around 10 weeks after conception), along with detectable levels of T3
in the brain. Moreover, a large increase in the number of thyroid hormone receptors has
been found in parallel with neuroblast proliferation during the 2nd trimester. All three
discoveries strongly suggested that the fetal brain is a target organ of thyroid hormone
from early onset of pregnancy (90, 91).

Second, in ﬁrst-trimester moderate/transient hypothyroxinemia rat models,
neocortical genesis and neuron migration were found to be interrupted and altered (92).
The observed changes could be avoided by timely T4 infusion, and became irreversible
with delayed or absent T4 treatment. Absence of beneﬁt with delayed thyroxine
treatment implies that a critical window exists in brain development for the effects of
thyroid hormones. Similar ﬁndings from other authors reinforce the view that fetal
thyroid hormone deﬁciency can result in blurred layering in the neocortex, abnormal
distribution of callosal connections, decreased myelin deposition, and altered neuron
differentiation and axon arborization in fetal brain (93-97). Lastly, but not the least, free
T4 is the only iodothyronine that can cross the blood-brain barrier, undergo intracellular
conversion of T4 to T3, and modulate gene expression and protein production in brain
tissue (98). Thus T4 is critical in fetal brain development.

1.8 THOP and adverse neurodevelopmental outcomes (Table 1.16) (51, 53, 87, 99,
100)
THOP has repeatedly been shown to be related to adverse neurodevelopmental

outcomes. Three large observational studies linked ﬁrst week thyroid function of

24

premature infants and their later neurodevelopmental outcomes. All three studies
followed large cohorts of premature infants over a long period of time and examined
these children with widely accepted reliable neurological instruments. Each study found
different degrees of cognitive delay in early childhood (18-24 months) with THOP during
neonatal period, all were statistically signiﬁcant. The average reduction in the Bayley
mental development index (MDI) among children experiencing neonatal THOP varied
from 6.8 to 8.3 (87, 99). The cognitive delay persisted until later ages (age 6, 8, and 9), at
which time IQ reductions were found to vary from 6 to 8 points, and more school failures
were observed in these children than controls (51, 87, 99, 100). Neonatal THOP was also
found to increase the risk of cerebral palsy by 3.6 to 4.4 fold at age 2, and the risk of
motor development delay, especially walking disability, by 4.9 fold at age 6 (87). Their
ﬁndings were consistent with each other and strongly suggested that THOP during the
neonatal period is associated with increased risk of adverse neurodevelopmental
outcomes. However, the question of whether premature infants with THOP should be
treated with thyroxine can only be answered by a randomized clinical trial with valid long
term neurodevelopmental follow-up.

1.9 Neurodevelopmental outcomes from randomized clinical trials of supplemental
thyroxine for THOP (Table 1.17-1.18) (75, 101-105)

Since the late seventies, six randomized clinical trials (RCT) have been done to
evaluate the effect of thyroid hormone supplement, among which three early studies
focused on neonatal mortality and morbidity, and three later studies also included
neurodevelopmental outcomes at later ages. One study detected a signiﬁcant protective

effect of thyroid hormone (RR=0.23, 95% CI 0.07 - 0.74) on neonatal mortality in

25

moderately preterm infants (104), but other studies, although they detected similar trends,
failed to ﬁnd statistical signiﬁcance (Chowdry et al. RR=0.46, 95% CI 0.05 - 4.38; Van
Wanssnaer et al. RR=0.67, 95% CI 0.36-1.24) (75, 103, 104). These negative studies
were underpowered to detect moderate effect of supplements on cognitive deveIOpment.

In a small RCT (n=44, 22 per group) supplementing preterm infants (GA < 32
wks) with two doses of 25 ug T3 for the ﬁrst two days of life, Amato et al. reported a
20% mean decrement of the fraction of inspired oxygen (F102) in T3 treated neonates
(peak F102 0.58 i 0.12 (T3 treated) vs. 0.78 :1: 0.09 (control) liters, p<0.05), however,
similar effects were not seen in other trials, including an earlier trial conducted by the
same team using T4 as the treatment (75, 101, 102, 105). The inconsistency could be due
ﬁrst to a different biological effect of T3 and T4 treatment. Second, Amato’s group
studied 44 premature infants with RDS, while others focused on general preterm
populations. By and large, thyroid hormone supplement had no beneﬁcial effect on
infant growth, duration of mechanical ventilation/CPAP, neonatal morbidities, such as
sepsis, patent ductus arteriosus, necrotizing enterocolitis, brain lesion, etc.

Among the three trials which studied neurodevelopmental outcomes at later ages,
Chowdry et al. found 11.4 points of MDI reduction (not statistically signiﬁcant) in
thyroxine treated group at 12 months of age and no MDI difference at 24 months.
However, their conclusion could be misleading because their sample size was extremely
small, 5 in treatment group and 3 in placebo group at 12-month follow up, and 2 in each
group at 24-month follow up (103). Vanhole’s group also examined Bayley MDI and
psychomotor development index (PDI), but at an age (7 months) too young for a reliable

CP diagnosis to be made (105).

26

The most important contribution was from Van Wassenaer’s group, because it
was by far the largest RCT of thyroid hormone supplementation in premature infants.
200 premature infants less than 30 weeks of gestational age were enrolled, with half
receiving 8 ug T4 per kilogram of body weight per day for 6 weeks and another half
receiving placebo. FT4, TT4, T3, rT3, TSH, and TBG were repeatedly measured at
postnatal day 0, 3, 7, 14, 21, 28, 35, 42, and 56. Standard Bayley examinations were
performed at 6, 12, and 24 months of corrected postnatal age. No difference was found
in neurodevelopment between two study groups. However, in subgroup analyses, treated
subjects who were born less than 27 weeks of gestation had an 18-point increment in
mean MDI (p=0.01). Among subjects born at or after 27 weeks, a 10-point decrease in
mean MDI was seen in the treatment group (p=0.03). These ﬁndings suggested that
thyroxine treatment affects newborns born before and after 27 weeks differently,
although non-pre-hypothesized interactions in a trial should be treated with caution,
infants below 27 weeks, i.e. extremely preterm infants, were indeed'short of thyroxine.
Thyroxine treatment appeared to beneﬁt this population and improve their
neurodevelopmental outcomes. Nonetheless, the sample size for the subgroup analysis in
infants < 27 weeks was relatively small, 13 in the thyroxine group and 18 in the placebo
group. To conﬁrm that thyroxine supplement is beneﬁcial to extremely premature infants,
a larger RCT is needed.

From April 2005 to March 2007, we conducted a RCT of thyroxine supplement
among infants born between 24 and 28 completed weeks of gestational age. This study
aimed to ﬁnd the optimal dosing schedule and route for thyroxine supplementation in

extremely premature infants. The methodology is described in detail in chapter 2. In

27

stead of the treatment effects, this dissertation will focus on the epidemiology of the

thyroid hormone abnormalities in very premature infants.

SUMMARY OF CHAPTER ONE

In summary, compared to term infants, the thyroid hormone levels of preterm
infants are much lower and are positively correlated with gestational age. The postnatal
hormone surges are in general attenuated, characterized by being qualitatively similar to
term infants but quantitatively less in moderately preterm infants (28-36 wks), and being

broadly suppressed in extremely preterm infants (g 28 wks). Independent of gestational

age, intrauterine growth restriction (-), histological chorioamnionitis (-), placenta
deﬁciency (-), and antenatal steroid use (+) were found possibly related to thyroid
hormone levels at birth, although the databases were modest for drawing conclusions.
Respiratory distress syndrome, sepsis, chronic lung disease, necrotizing enterocolitis, and
intraventricular hemorrhage were all found to be associated with low thyroid hormone
levels during the postnatal period, but the causal effect directions were uncertain.

Previous observational studies have suggested that neonatal transient
hypothyroxinemia of prematurity might impose greater risks of adverse
neurodevelopmental outcomes, such as delayed cognitive or motor development, school
failure, and cerebral palsy. Follow-up results from randomized clinical trials have
suggested that thyroxine supplement might improve neurodevelopmental outcomes in
preterm infants with neonatal transient hypothyroxinemia. As yet, however, the beneﬁts
of such treatment in terms of the survival, respiratory maturation, and decreased

morbidities/illnesses remain uncertain.

28

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 1.1 Percentages of transient hypothyroxinemia of prematurity.

Author, year Deﬁnition of THOP 1:21;!“ N Ggagvﬂgal Paglia/:1; age
Diamond, 1979 Serum T4 < 4.8 pg/dl Day 3-5 64 < 37 44
40 28-30 52
59 31-33 33
Hadeed, 1981 Cord T4 < 6.5 ug/dl Cord blood 92 34-36 12
215 28-36 25
Cord blood 18 1 1
Day I 32 3
Day 2 39 5
Uhrmann, 1981 Serum T4 < 3 ug/dl Day 3 31 28-36 16
M 7 50 18
Day 14 46 11
Day 21 44 2
R l' 36 g 33 69.4
1 53“" " Serum r4 < 6 ug/dl Day 3-60 64 34-36 40.6
110 2 37 17.2
Mercado, 1987 Serum T4 < 5 Eg/dl Week 1 292 - 6.5
Day 0 58
Day I 67
Day 3 83
Week 1 84
Mercado, 1988 Serum T4 < 5.1 ug/dl Week 3 108 23-31 63
Week 4 49
Week 5 48
Week 6 36
<26 , 83.3
26-27 70.8
Den Ouden. Serum T4 < 4.7 n /dl 28-29 44.1
1995 (38D below meals week 1 717 30-31 24.1
g 32 15.3
Total 32
Lucas, 1996 Serum T3 < 19mm Week 1 280 30.2 :t 0.2 21.8
23 40
24 31
25 31
Reuss, 1997 Serum T4 < 4 ug/dl Week 1 71 I 26 25
27 15
28 10.2
Serum T4 < lOth p of Week 1 91

Paul, 2000 babies screened on that day 247 < 1500 g
& TSH < 25 EIU/ml Week 2-4 59
§ 27 16
. Serum FT4 < 0.8 n /dl 28-30 4
Rabm, 2004 with normal TSH legvel Day 7‘14 1 14 31-33 0
Total 7.9

 

 

 

 

 

 

 

 

29

 

 

 

- :meip ”NAN am A 5528585: 28325 28.55:
- «E 8-8 a: 35528EBEEEEEE Sow SE:
850:. 2&2“: UsEE< mE «E six «2 24 a m; a $2 .22
. . . 3.32 .
1 1 :3 00m
05 :me me «e a. 2. SM .278 3.3 .913 «mi .mEo 32 a a
852: :32 _ 8255 :me «E «E 3-8 2 9. .N 3 88 .222
one .mme Ame: SAN .
mm 2:
22203 6m 82> 2.2:; E at .ﬂ «9 «E _ow-m_ ”a. woos 28 «8m :1
<2 322 38 :me «e «E 24.” «m NE. a 8% aces:
<2 OE. .mme «e «E 3.8 3 N ._ 3 ”N o 683 28 £2 .82:
<2 :2. .me «e «E 3.3 8 8-2 a 5223 :2 2e 52 58523:
<2 me «e «E 3-2 a. « A « .m .N ._ 3 32 ESE:
<2 :me «E «E 2-2 «m Nm «N « .N 3 a a ”803 28 SS 68%:
<2 Ewe .E «E 3-8 N 23 9. owe .9238 ES. .2283 ”3. a 82 .032
<2 :3 .Q. .E «E mam ”N a «_ .N. .m .m ._ o 585 28 M32 5:2
- one .mt .59 .me «e :18 3 m .2 a $2 .wcam
vac—:52 Oh .Om: n
. . . . . . . e . . m . . . . .EoELowm
QQEOEEESE :me me. me ME E «E 2v 2 S R «2 S. cm on S on 33. m
<2 2023222 :2. a «E 3-8 2 91mm .5: a 58 6.2558
- :me a. E «.N 4 am m2 9. « a 88 8:02 on
- GE .3 «F 2-8 3 8 .8 .8 a RS .385
<2 :3 .E «e «E 3-: cm W ._ a $2 .56
8503552 . . . . 2 dimgvecoz .
.— n n- n «moomatw
5528585. :me R L «e «E m. 9 we is a «N; E 683 28 «com o
- . . . - 8.: «so at: «-2: «Ma. .weioﬁm
RE :3 E «e a. E 2: 685 28 -eowaom
22138555662 :we at .mE «E «E 38 mm. .1 .2 .e .m .N ._ o 5:38: w v 88 £52
2;: . . . . . - . . . . .
amaocassZescm om: 2: E ME E «E E a S a e m 20:5 too; 28 88 sum
23% 525553 825 «E 3-2 «2 _ S: 283 5 8% 28 82 ans?
8M2 u. 2.5.5.3 68%: 3.5.53.— 9.5 mum Z 133:3 2.2—=92.» tee—a 5:3 25,—. .30» ice—i
e «E . . E .28580 . .

 

.:.2-m._nss seagusewseaﬂo «seam See;

30

.3833 c2958“ 3265 Bob 3:850 moan—am _

 

 

om: . .
3.2928. am 8s; .m«.« .:m.« «u: .3 «e «E N« ...N S .803 «.8 88 2.2:?
<2 OE. fake. «H «E. «E 2v 2: w e m « m N Mew/OENLOM .sscaaaemww
- :me «e «NAN «m _N .3 N .N .N ._ a 683 28 52 55.82:
- 05 at .mme .R «E. «3N m... .N «_ N .m .N ._ a «83 «.8 NE .5353:
05 . . . . .
3.2852 652.5 .:m E. wig. .me «e «E «m MN 3 «N «_ N E .803 «So EON 886%
<2 :3. «E «N 2N «_ a _ a «a: .5682
- :2. «e «-8 N3 8 .o« «N .2 .m a «83 28 N22 £82852
1 :mF .v: om V @_a chmcoohnvm E0332; M003 am: 1.— bam— .mmsoM
1 Emu. .v: NM V mom chmcoeow E03325 #003 «mu?— hOG~ .mmsoy—
mares 522.33 sea :3 «E mm v «: «E a a «8: 28 «SN «52
- :me «E «NAN SN «N.«_ .m a 88 .32
- :me .E «N «e NNNN s «_ e .N .m ._ o «85 28 82 «2.32
cm: . . , . .
32862 cm 8:5 Ewe .mE. at .me «e «E «m «N NN «N a N I: .802 25 «SN 29::
- om: .mme .me «e :me no. N« .91. .NN ._N N .N ._ e 2E3 2 3.3 .8852
.328 :1.— u. 25.532. .333 285.8: 926 own 2 22:8on coo—m .30» 5.5—5‘
. . 7255330 .

 

€253 .:._-N._azs 5 8532 863:0 $553 N. _ 22¢

31

23 o._ H Nam <0 5

a? 3 H 9% <0 : 23.2 <0 2 0.3 3-8 <0 « «83 28 6: .55 a 02:30:. 88:8: w
”0505 :03 <0 0 a? I H 3m <0 m a? S H 0.; <0 « a? 2.8 <0 m a? S H 0.3 <0 N 0.3 2-x <0 _

 

 

 

 

 

 

 

 

 

 

 

- - E H ”a 3 H 3 _ S H N« «SN .2922

- - - o. 5H2 awH0_ 032 608.02

mwé H 0.0 and H 0.x. ~_.m H o.» mum H 0.x. 0N H w.0 voom 68::

h w.m H 0.0 n QM H «.0 v m.” H «.3 N N; H ad. - £2 65::

- - m 06m H m0.0 . - - voom .383th
Sign 3-0m 3.-.». 3-3 3.3 .

3.5 on“ 3.83586 .30» 35.3

.3623 2.20006 .3 08.502 anew own Ecozﬁmow 3 am H 0025 «.35. 95:5 IE. You mg 030...

- - 2.09-0.8 owcﬁv : ads - - :3— .ESEE:

- - mm H 3 2 H on _ S H m0 33 ﬁes:

- - - 2 mm H 00 a 2 H av «32 600822

0.3. H 0.0: 0.5m H w.m: 0.0m H 0.3 v.0m H N. 5 0.0m H 0.00 voom 6:52

n o.0~ H vﬁﬂ m wﬁm H ”E: v Ema H 0.; N m.mm H www - 33 .0532

0 R H 3 _ - m 0.0m H 5.2 - - gem «88830
9.3“ 0m-mm 3-3. 033 SAN .

3.5 owa 3:05.336 .59» ..0_==<

3:030. Eobbﬁ .3 002002 anew 0mm .mcocﬁmow 3 Em H 2025 20>ﬂSOES EL. 200 v; 030%

- - ..N H 3: 3. H 0.: _ mm H «.2 «com .3822

o.m H 0.: N0 H No— mé H md— 00. H 0.3 «:0 H 3: voom 653$

N. 0.0 H VAN m <6 H ﬁwm v Wm H _._m N a.» H Ymm - 32 6:25

0 mé H M:— - M 5m H 0.2 - - voom .0880th
«9.5 0n-mm 3..-; 3-3 2-3 .

3.5 own 3:20.330 .30» ..05:<

 

.8653 2200.50 .3 00:0qu anew own .mcosﬁmow .3 Em H 0025 m_o>o_ 2:085 FE 050 m._ 030,—.

32

.oAHmdn <0 2 ”Nvém 3 ”NYC“ <0 2 ”:15” <0 .2 ”SH-mm <0 : ”ﬁx—Hﬁmm <0 @—
"mv-wm KTmN a mwmém <0 m momém <0 5 www-mm <0 0 MN. .Ho._m <0 m 3.3 <0 v ﬁmém <0 m mm._H©.oN <0 N :mA‘N <0 _

 

 

 

 

 

 

 

 

- - m m.m H rem: - voom .888th m 33>?
w.m H m.wm w.v H wdm 3: H New we H m.m_ mm? .0585

2 S m N a :83

3 En H Qmm a 9m H m.m~ - - 32 $938:

2 deNLm n _.wHw.C - - $2 .53..

2 m.mHo._m o_m.mHm.Nm mo.mHv.2 No.va.m_ 23— 6:85

2 as H Qmm a N6 H WE - - 33 $98.5: .233

m. 3: H 9.3 w m.o_ H mm v o6 H adm od H w.wm m3. £83<

- - mod H Q: - vocm 680%.th

: m.mHo_ wode.m_ - Fc.me.__ voi .550 wham—

~_ o8 H :m N. v6 H a.m_ - - $2 .53

- - .3 H EN 3 H gm 3cm ﬁes: Q

: o.mH£ cm.mHa.o_ - _ oéHmf 3o.— .5:U _ a

- - m ad H mém - voom 38$th

- - M: H EN as H ”.2 38 .2832 N. :3:

- - 3 H ”.8 3 H o. _ m 38 .3822 :3

- - m ad H ..w. - voom 680$th _ I

on A cmﬁn 3-3. cmén

 

0.30 can 3.53530

.59» ...cﬁ=<

”£38 «a «as .3558.—

 

.280 as”; 329:8 933
85%: BBaEEa 388258 E 333 Ecousmow 3 own 3358“. .83 am H 53:5 205. 2:085 Eh EEom Ema—“mom c; 2an

33

”36mm <0 3 momém <0 o 6m-mm <0 m ”N; H v.3“ <0 5

.o._ H mdm <0 3 mmvém <0 2 MhmM<0 E :vuwm <0 2 ”3.8%: Eh: own .mcouﬂmow Sm =mEm .2 JV.— H hgmm <0 I
mmmW <0 9 mvmuom <0 m £3 mmém v MN._Hw.am <0 m ”Scum <0 N ”Sum H :88 _

 

 

 

 

 

 

 

 

3 3. H is B E H 32 c E H 03 - _ $2 .zocwmeom 2 b5
- - H mm. H 3: - _ ME: a§§S
2 2: H22 : 33$: 5 ZNHW: m 2; 3.2 32 55::
2 mm H m0— 3 mm H Na: . .. _aa_ .gmxum: nv—ooa
- - v N? H VNE - voom 688830
- - - cm H mm 32 .8832
. H . . . . .
rewricﬁ ”$.32: - 123? #52,55 :5
2 2: H 3.: o H v.2: 52 H E - _ ma. 3:5
- - - 8 H S 32 5.2852
. . . . . . m E.
z _ HQ: 0 _ HHE eEH32 - _NM§ =8w§§
- - m 2; H 32 - _ $2 £582:
- - v 5.9“. H exho— - vcom 688$th
- - mm H m: E H 3w $8 “29:: _ 55
2 3m H o: w 9% H N: - N R H ”.2: 32 .85
- - - mm H 8 $2 .8852
2 I: H o. :1 2: H 3: 0.8 H E - _ $2 .wcem
- - 3 H HE S H mm 38 “has: n =8:
- - mm H a a H .R 38 .352
. . . _ :5:
- - v m 3 H w m3 - voom 38.8th
cm A 3.3 3:: 3-2

 

330 ads 1253380

.30» ...o.==<

”£38 an own .3258.—

 

.296 £83 323.8 8-20
3.5%: 0.53an 38835.: 5 £003 3:25QO 3 own 3258a 3>o Em H .825 20>“: A5085 Fr 82% 3958.; n; 2an

34

.o._ H mdm <0 2 ”inm <0 2 Mhmw<0 3 ”Swwm <0 2 ”35¢”: FEB own Ecosywamow he =me N. ”v..— H >.mm <0 I
Kmém <0 2 6mém <0 0 6m-mm <0 w mm; H 0.; <0 N. MmmW <0 0 mvmém <0 m 33 mmém v mN._H©.om <0 m ”5-5m <0 N 5.8 H :88 _

 

 

 

 

- - m ﬁn H S: .. _ who. dams—ED
- - - vm H E. wwﬂ .8882 m :33
- - v MEN H 3%. .. voom .383th
2 v.8 H QNM: : YMN H who N. S: H 3% m 9m H 0.0m 33 65::
2 mm H E. 2 om H mo - - 32 $23.8: N :33
- - m 0.: H _N_ - .mhi .55:th
cm A emumm vain Sim .
33m? 2285580 .32» .55:< ”:28. «a 93 .535ch—

 

6253 .920 v.83 83358 3%
3.5%: aha—SEE; 38825:. E 8303 35:8QO 3 own 3358a ..o>o AQm H 58:: 20>»: A5055 Fr 83% .8230; h; 033.

35

o._ H Ndm <0 2 ”Nvﬁm <0 0~ Km M<0 2 ”:lwm <0 3 ”$5.05 8.50 own Ecozﬁmow hc.2135 m. x..— H hdm <0 N_ m0mém <0 2 ”0m
-Vm <0 2 M0.m.~m <0 a mm; H 0.1m <0 w ”MAW—m <0 N. mmmw <0 0 ”3.6m <0 0 ”mmém <0 v ”N; H 0d~ <0 m xmém <0 N ”Sow H 538—

 

 

 

 

 

 

 

 

 

: N. H w: 2 2 H ”.2 0 mm H ..2 - _ $2 ﬁgmaeom 2 a“:
- - n no H 0.». - _ $2 .5562:
2 aé H 0.0 2 Wm H Wm m Wm H Em m ad H v.0. mwi .053:
2 ma Hem : 33.: hMEH mi SHE: m3. .832 3.83
- - - 0 H .2 wmﬂ 60852
- . - H 3N H 03 38 58826

3 H E 2 H am 3 H ..o - $2 ,wscm
m— — m han—
E 90 H 5.2 a #6 H 08 u N 0.x H 0.x woo. 60:0
- - - 0 H 2 $2 .8852 Q
2 2 H3. 2 3 HS: 0 GHQ: - Ewe accwmeom m a
- - m ..m H 2: - _ m8. .EaEE:
- - v 2.0 H 3.2 - voom .388th
- - 3 H 3 3 H 3 $8 .2932 _ b5
3 2; H 9mm 0 Q: H E: - N 3 H M; 32 £26
- - - m H 2 $2 .8852
m. 0H H 3K 3 H w... ”m H NS - _ $2 $55
- - 3 H 3. mm H E 38 is: b :5:
- - Nd H EN ..2 H we 38 2&5: :5
- - H 3.2 H $3 - 38 £83.50 _ =

3 A 3.3 3.-.». :33 .
0.30 owa 3:20.380 .59» .85:< w==m3 an own .335ch—

 

.~<o% 3.83 329:8 033
£50.05 230808 b88009: E 9.83 Ecousmow 3 own 3358a B>o Em H 53:: £05. 25:: :mH E38 3358A M: 22?

36

.23 bN-0N <0 m £3 aN..MN <0 N ”E3. H :38 2

 

 

N vd H 0.: _ NOON ask/Em N :33

2 H 3 88 8322
N _ _, 28>»
M 2 H 2 . 32 £53

. . .339
NHoHoS 328. .m :3
E H <2 88 29:2
3H EN 38 .2832 b :5:
3 H EN 88 229:2 _ :5:
E - E

.3?» ..85=< “:38. «a on: 33585

 

930 um.“ 3:30.330

 

.208 283 320.28 ”N WV
acme: 23$an 3qube E 8.83 Ecowsmow >0 own Ema—“mom 8>o 5m H 525 232 $6.23 SE 623 Ema—“mom a; wink W

o._ H Ndm <0 5 ”NY?” <0 0_ Km M<0 2 ”IN-mm <0 3 $2305 628 own Ecocﬁmow 5.0 =mEm 2 x».— H b.mm <0 N_ m0231... <0 : m0m
-Vm <0 3 M0m.Nm <0 9 mN._ H 0.; <0 w ”QT—m <0 5 mmmw <0 0 mvmém <0 0 Mmmém <0 v ”N; H 0.9N <0 m MEAN <0 N ”83 H 53::

 

 

 

 

- - m N.o H m.m - _ whﬂ “enacts:
- - v m6.— H om.m - EON .383th m 23>?
- - - v H 2 ”M3— 60882
- - N.o H Ym - 8.2 65:55

W _ N 283
t o.oHo.0 N. m._ H m.N mo.mH0.a ma.mHv.o_ 32 6:82:

0m A 0m-mm vmém awrwN .
95 can 3.83380 .80» 35.2 9:38 3 «mg .535ch—

 

9 28H .280 283 329:8 3.83
$505 0.58622 388002: E 8.83 _m=o_§mow 3 own 30582 B>o am H 5380 m_o>o_ 25:: :mH 838 3358; w; 203;

3.3 omém <0 N 2:8 H :38 _

 

 

 

 

 

33 H v.2: 33. H ﬁmo mﬁm H wdh Wmv H awn odm H own N.Nm H 0.3 3a. 530% 2.33
£2 :2 £2 £2 £2 3 w
mm H 8 32 62202 m :33
N Nav. an wéuh _ NOON .mwzzmmm N 3003
e— H ON MW©— 593.52 — gum;
N 3 H 23 _ .88 .335
a H mm 32 .8852 m b“:
R H am 38 ﬁes:
: H ow $2 “8852 _ 55
N 3 H 98 _ 82 .355
X H g $8 .2932 h :8:
R H S $8 .3832 _ :5:
Ma 2 gﬁwﬁwhmzsmoo mu m 3% 55.2 958. a on.“ 32.8..

 

$0.8 383 3238 ﬂ WV
3.8%: 053805 32:2on 5 $33 Ecouﬁmom 3 own 33:38 $>o 5m H 53:: £32 £20.55 Fr 62% Emﬁmom o _. _ 038:

38

 

 

 

 

 

 

 

 

_No.o u a 2383 N8 2235 3 :2»: EL \ H 88 SON .582
8.2 :3 ,mm omd H and mm; H SN 5w: F: \ v 89 Sam dozen on
.0 £3 £0 - 2:888:385 838.32: + 9.8888822? 838.3%:

88 v a 2 H 2 S. H 3 .553 :2‘ \ 38
.9: no H 5.8 We H 5.x 8&1 Fr \ .280 moon 22$
28 v a N8 H N2 N8 H 8 :28 Eu \ 35

2:38 - 88.2.88 8.88..— +No=£ocov «:8an
8.: ﬁond— vqu: 23 pm M 32
H... 2 H 8.8 I H 3 23 8-8 :81 H: m 2.8 ._ 88.8

. a . . . . H ._ M
moov nonw moHov 233V u

2:38 .. ”—0:- + :03 95 own 2888: \ $58. .80» .8225.
938...?» 2:830 .885...— 885380 8 ans 8858a

 

2:5 8 282

2888: 28%.: 5826: 53> 8822.80 9.83in 0.8 85 8083 88:08 3 A88 H 825 288— 288.8; RENEE N _._ 0389

mEﬂE 228808 3088on E 8.83 .2823me N3 own 88588 88 Em H 825 208. 25:: IE; 828 88585

VH :

. ax.
NVOHC'
5H:
Nodeé
mmBHmw—
mH Z
m._Hm.m
8H2
mdeQN
m._Hc.m
N.N.Hm.w

38?: 5-: own 883.380

.23 2-8 <0 N 23 oNAN <0 N “Ea H 82: _

82 68982
_ NSN .858

89 .8882
_ NSN 832m

82 32%
89 .8882
EON .3932
82 .8882
_ NSN 932m
HEN 229:2
EON 229:2

.89» ..85:<

n :33
N 8.33

_ 18>?

N 88

_ :8

b :8:
_ :8:

9:83 8 own .8858.—

 

.208 283 3298 NN WV

_ _._ 038%

9
3

 

 

 

 

 

 

 

2 3.38.2 2.12: 2 sa:
.2 SHH 2.2 EH 3.8 m8:
.. £28m + £28m
.m: :.mHoM.m~ ENHwVE o. N3Q
.m: 2d H 3.3 and H 3: m 80 $2 .=o:wm:8m
- 2:968? 29.52: foawhﬂ—mom + 0:885? 2952: ESEEmoy—
.m: mHE vH: M803
.m.:. me. 5H2 .803
.2 2H: HH: :8
2 mHQ 8H2 in:
.m: w H S C H 3 5.5 :< mom: 8:882
u 2885 288:8:— o:=a.m= + 288:. 08.582: 25:»:
2:38 032.8.» chamomxm wEamu. an own .8258; .80» ..8_=:<

 

.aﬁdE 882m E 88898 385% cc: A88 H :muE%m_o>o_ 938%:mg. 3280: 5232. 8:282}: : Z 033.

 

.888: 8 628:0 £28533
2: E m=uo 820:8:808bom
2.5.8 .88 com

mo>
mo>
mo>
:2
oz
2;

3:
oz
:2

mm;

m“;

2;

2% 22% .22.: a? 2555 5 ...22 533-93
88:28-8: 2 E Eu: :8 3:88:82 8203:2888boa o_ M

38$ .38 m2

0:: .8 383 :2: 2: :8 29:2 02

028% 288: 3:882

MOD—

mE:o_:EaE8:o 28:88:3m

82:83:88? 380800

“a: .8 82> 2c 2: S 8:: £25
38:8 :Emonm

5:2 82%:
E. .+ <08 8: 222m

 

8o. 78m :2 H E: @ “832:8
2-2 93 H 2.3. 35 2 22830
22 .522 22 .88..— m8 23.3:

 

~53... :8 mE:o_:ancoso 1882082 5232. 8:382: 2: :o 86:? oz: 5950: 828980 m _._ 28H

0
4

 

 

 

 

 

 

 

 

 

26 v a one a a: who a :2 3.:

.2 :3 a we.“ «.3 H 3.” RM

:3 v m 3: a m; :2 a 3m 8.:

85 v m m: a 2: 32 a 2: 2w :31 E \ 2 >5

86 v m a: a 3.: So a v3 3.;

.2 NE a 3.: 8a a 8.: am

:3 v m 5m a 3.2 gm a a; 3-3

86 v a ”Na « 8.x 3 a a? mmw .2? EL m b5

. £30m + £39m

mod v a one a 8.» NE a 3% .55

.2 2.. a 3: mg a $5 SM

85 v m o; um 2..” $5 a as 3.:

85 v A. who a :5 Rd a and 2W 6&1 E \ 2 >8

83 v a .2 a :2: 85 a m3 3.:

.2. 3; a 2...: a: a 9.2 2M

53. v m who a 32 who a m; 03m

:55 v m 2: a w? as « Sm mm w 5% E \ m E $2 .__o=w§§

. «ESE—mm 323:. ESE—38¢ + 0595:? $9.35 ESP—Emu:

86 v a 8 a 2 mm a a m v.83

53. v a mm a mm mm a a _ :83

83 v a 3 a 8 mm a S m g

53va m3? 2 is :8

83 v a 8 a E M: a mm 5.3 E was .8852

u 093%? 0:52—50:— oE—En: + omaoﬂﬁ 0:92—50:— ops—Fm:
2:57: 93.32; 2:3an 93v ems wits .80» £2.22

.3833on an own .355ch—

 

Qcﬂﬁ Essa E 85893 .9933 was 98m H $25 m_o>o_ 9:085 El .828: 5233 :28682 m 2 Egg

4]

383:. 35.5.0353 3039:9338 - 3563.233. 353:. .0 330m 30.33. ”EA. m 3:35 3338?. 33.5 ..o 30... ”mag... 5
W5333 5v .3o:.m-3:o..3:8m Hm..m o m5.33.3.0 8.. mm3.3m 8:03.33. 3.3.83 30mg» w 2
383:. 32:33.53 33:2: - 3583.232. 353:. .0 830m 38.395 ”5.2 v 38333 3.23 :. 3333:3538 33:82. .N .5. 333.3 m 4
3833. 33:32. 3:3 own 3:233me 30.. 3033.33 N 30333 30:... .o 8532: 3:33 mic—3.0:. 330350.350 33:82. 03.3.2: 8.. 3033.3 .

 

 

. . mm a 3 mag
3 own 3 3.33 8 033:: ..o 33 3 o h .E . .

a . 3 3.2.- com. 3:3.

33.8.3: .m own 3 30.0 3 v3.3 3.0. 3.13 m 3.0..-. ... 3:3 3 c m.-m 33.2.63 em 3 5.). w w m
.m own 3 G. T. .. .o awn 3 3. T... .38 @3828 N 33. .5350
.0 $3 .m. .HmO. 3530:3933 030.2357. .0 $3 .m. .umOv 23.3.. .oosom 3:23: @333 ..o 3.3: 3.0..-3d :oQ 3:3 mom. .532).
.E M: a 3 352..-: 3 Sn. .33.: em“ a 3 358-3 m mm; .63: a 3 3. 52 6.83... 82 .32 933
3.3—«5.83.3.3 33323.8: LOIP .350 Ann.“ .323 O. ..O....—. Aowa 3.33 3388.5 95...?! 3.0;! ...33 ...23:<

 

333:8. .55.... 33.3 3.2.3.20 ..o 8:58.30 335830353835: uﬁo>3< o .. . 2.3...

 

 

58:. 322. am 3 v E 8.3 v 3m 8.. 82 .23.
=85 322. mm m v E 83.3: 3. va<0 a own. 333 ..m 033 $3 AN 33 Mom 33. .5350 5: Ba 32 .5362
see: no v E Omavam s N... a No... <0 93 82 .32 803
525.3: .53: a. 3: 36.333: a: 295m :3» .353.

 

.39.... 3.3 53.3.3.0 (.0 3808.5 .85.:3035323: 83%... o ... 033 (.1

.xooE EoEQEgoc 3:2: ”52
:86 n 3 358 2 a 52 anew x3 Rm
:3 n 3 358 E P a: a8.» VE EN v

 

.9552 £55 .5 <Qm .NOE £80m co Soto oZ ”$5.88 v3 .9: 3m— ...omcummmzz §>
.an £303 98 .NOE .89 tau: so Soto oZ Ame—“:08 C .9: 32 63:53
@993 SEES??? 3255853 353 - a: .swspcoﬁm
538w co Soto 02 $5.88 vm um. m: .md 33— .waonu

Godve :3, _ a 8: x35 - $2 .222

NOE V33 :0 Soto oZ - wag 63E<

2.353.: a: 33:9 .350 05.53.: .3 «99:3 Esau Nae—ham :3» £2.22

 

EoEoEQE 25859: .23 3:85 AOIF wanna: 2th 28:50 noﬁEovcﬁ 5m .8 mecci w _._ Bash

45.585 .«0 x83 “m5 2: Ban om: Ho: Ev 3. 5:3 wﬁw: m. 8 8322: 33 omoc 05x0??? m
£83 $5 of £53, 36 3:0ch m cm; 95% comm N
6:8 mmRv E 3:38 was 3083 5:5 3:35 ontgnaE mm? “:25me 9:88 3 nucwﬁmm 803 on? 38.33 at“. _

 

 

£83 0 E 9:? w 3. v 8:2: 32 52333 §>

383 N E mam: cm 2. - a N 38 82 .2o:§>

xﬁm .928: 2:5 .8 omen >35 mH w: m + E. w: mm w ooﬁm v 8 mm v _ QOv 52 593235
3.83 E m E 9%: 2 mm - 2 :E 52 5386

9% N x 5 a 83E E w: 2 2 v $8 $2 .222

E vm can _ 6 880 N Fr w: on E” - om wim— wwm: .onE<

559:... 2.0.5.3.; «mac: 9%?“ 3.53580 Aeneas—d b55395 Z 39m ...c5=<

 

€253: “SEQ—33m 05x85: .3 ﬂat“ E356 noﬁEoccﬁ xi .6 mcwioc 33m E Z 2an

43

Figure 1.] The metabolism of human iodothyronines.

3, 5, 3’, 5’-
Tetraiodothyronine (T4)
I I Nm
|
HO 0 CHZ—CH—COOH
35’3" 131,132 45’) D3 (D1) -1(5)

Triiodothvronine (T3)

I I NH2
I
HO 0 CHZ—CH—COOH
l
I
HO 0©W CH2 —C|H— COOH
3, 3’, 5’-Triiodothyronine

Reverse T3 (rT3)

 

D3 (D1) - I (5) D1, D2 -I(5’)

3, 3’- Diiodothyronine (T2)

 

44

Figure 1.2 Cord serum FT4 levels by gestational age (plot based on data cited from
Williams et al. 2005 (63)). (l ng/dl = 12.87 pmon)

21<

20‘ A

19*

FT4 (pM/L)

17-

16'

 

A

 

A
a
15

T T V I T T I T I T I I I r I f T T I

24 25 26 27 28 29 30 31 32 33 34 35 36 37 .38 39 40 41 42
Gestational Age (Week)

45

Figure 1.3 Cord serum T4 levels by gestational age (plot based on data cited from
Williams et al. 2005 (63)). (l pg/dl = 12.87 nmol/l)

TT4 (nM/L)

140‘

130“

120*

110‘

100‘

90*

80-

70‘

 

 

60‘,

1 I I I I I I I I I I

24 25 26 27 28 29 30 31 32 33 34 35 36 37 -38 39 40 41 42

Gestational Age (Week)

46

Figure 1.4 Cord serum TSH levels by gestational age (plot based on data cited from
Williams et al. 2005 (63)).

TSH (mIU/L)

 

A

 

I I T T I I I

4 ,
I I I I r I I I I I I

24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

Gestational Age (Week)

47

10.

ll.

BIBLIOGRAPHY

Larsen PR, Davies TF, Schlumberger M, Hay ID. Williams Textbook of
Endocrinology. Chapter 10 THYROID physiology and Diagnostic Evaluation of
Patients with Thyroid Disorders, 2002.

Murphy N, Hume R, van Toor H, et al. The hypothalamic-pituitary-thyroid axis in
preterm infants; changes in the ﬁrst 24 hours of postnatal life. J Clin Endocrinol
Metab 2004;89:2824-31.

Pavelka S, Kopecky P, Bendlova B, et al. Tissue metabolism and plasma levels of
thyroid hormones in critically ill very premature infants. Pediatr Res
1997;42:812-8.

Simpson J, Williams FL, Delahunty C, et al. Serum thyroid hormones in preterm
infants and relationships to indices of severity of intercurrent illness. J Clin
Endocrinol Metab 2005;90: 1271 -9.

Lazar MA. Williams Textbook of Endocrinology. Chapter 4 Mechanism of
Action of Hormones That Act on Nuclear Receptors, 2002.

Campos-Barros A, Meinhold H, Walzog B, Behne D. Effects of selenium and
iodine deﬁciency on thyroid hormone concentrations in the central nervous
system of the rat. Eur J Endocrinol 1997;136:316-23.

Meinhold H, Campos-Barros A, Walzog B, Kohler R, Muller F, Behne D. Effects
of selenium and iodine deﬁciency on type I, type II and type III iodothyronine
deiodinases and circulating thyroid hormones in the rat. Exp Clin Endocrinol
1993;101:87-93.

Peeters R, F ekete C, Goncalves C, et al. Regional physiological adaptation of the
central nervous system deiodinases to iodine deﬁciency. Am J Physiol Endocrinol
Metab 2001;281:E54-61.

lzumi M, Larsen PR. Triiodothyronine, thyroxine, and iodine in puriﬁed
thyroglobulin from patients with Graves' disease. J Clin Invest 1977;59:1105-12.

Larsen PR. Thyroidal triiodothyronine and thyroxine in Graves' disease:
correlation with presurgical treatment, thyroid status, and iodine content. J Clin
Endocrinol Metab 1975;41 : 1098-104.

Theodoropoulou A, Vagenakis AG, Makri M, Markou KB. Thyroid hormone

synthesis and secretion in humans after 80 milligrams of iodine for 15 days and
subsequent withdrawal. J Clin Endocrinol Metab 2007;92:212-4.

48

I7.

18.

19.

20.

21.

22.

23.

24.

25.

Ayromlooi J. Congenital goiter due to maternal ingestion ofiodides. Obstet
Gynecol 1972;39:818-22.

C arswell F, Kerr MM. Hutchison JH. Congenital goitre and hypothyroidism
produced by maternal ingestion of iodides. Lancet l970;l:l24l-3.

lancu T, Boyanower Y, Laurian N. Congenital goiter due to maternal ingestion of
iodide. Am J Dis Child 1974;128:528-30.

Bostanci I, Sarioglu A, Ergin H, Aksit A, Cinbis M, Akalin N. Neonatal goiter
caused by expectorant usage. J Pediatr Endocrinol Metab 2001 ;14:1 161-2.

Khashu M, Chessex P, Chanoine JP. Iodine overload and severe hypothyroidism
in a premature neonate. J Pediatr Surg 2005;402El-4.

Markou K, Georgopoulos N, Kyriazopoulou V, Vagenakis AG. Iodine-Induced
hypothyroidism. Thyroid 2001 ;1 1:501-10.

Fisher DA, Dussault JH, Sack J, Chopra IJ. Ontogenesis of hypothalamic--
pituitary--thyroid function and metabolism in man, sheep, and rat. Recent Prog
Horm Res 1976;33:59-116.

Fisher DA, Klein AH. Thyroid development and disorders of thyroid function in
the newborn. N Engl J Med 1981;304:702-12.

Brown R, Larsen PR. Thyroid Disease Manager©. Chapter 15. Thyroid Gland
Development and Disease in Infancy and Childhood: Endocrine Education, Inc.,
2005.

Ballabio M, Nicolini U, Jowett T, Ruiz de Elvira MC, Ekins RP, Rodeck CH.
Maturation of thyroid function in normal human foetuses. Clin Endocrinol (Oxf)
1989;31:565-71.

Contempre B, Jauniaux E, Calvo R, Jurkovic D, Campbell S, de Escobar GM.
Detection of thyroid hormones in human embryonic cavities during the ﬁrst
trimester of pregnancy. J Clin Endocrinol Metab 1993;77:1719-22.

Costa A, Arisio R, Benedetto C, et al. Thyroid hormones in tissues from human
embryos and fetuses. J Endocrinol Invest 1991;14:559-68.

Thorpe-Beeston JG, Nicolaides KH. Fetal thyroid function. Fetal Diagn Ther
1993;8z60-72.

Thorpe-Beeston JG, Nicolaides KH, F elton CV, Butler J, McGregor AM.

Maturation of the secretion of thyroid hormone and thyroid-stimulating hormone
in the fetus. N Engl J Med 1991;324:532-6.

49

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

Thorpe-Beeston JG, Nicolaides KH, McGregor AM. Fetal thyroid function.
Thyroid 1992;2z207-17.

Fisher DA, Hobel CJ, Garza R, Pierce CA. Thyroid function in the preterm fetus.
Pediatrics 1970;46:208-16.

Hashimoto H, Sato T, Horita S, Kubo M, Ohki T. Maturation of the pituitary-
thyroid axis during the perinatal period. Endocrinol Jpn 1991;38:151-7.

Oddie TH, Fisher DA, Bernard B, Lam RW. Thyroid ﬁmction at birth in infants
of 30 to 45 weeks' gestation. J Pediatr 1977;90:803-6.

Jacobsen BB, Andersen H, Dige-Petersen H, Hummer L. Thyrotropin response to
thyrotropin-releasing hormone in ﬁIllterm, euthyroid and hypothyroid newborns.
Acta Paediatr Scand 1976;65:433-8.

Jacobsen BB, Andersen H, Dige-Petersen H, Hummer L. Pituitary-thyroid
responsiveness to thyrotropin-releasing hormone in preterm and small-for-
gestational age newborns. Acta Paediatr Scand 1977;66:541-8.

Klein AH, Hobel CJ, Sack J, Fisher DA. Effect of intraamniotic ﬂuid thyroxine
injection on fetal serum and amniotic ﬂuid iodothyronine concentrations. J Clin
Endocrinol Metab 1978;47:1034-7.

Vulsma T, Gons MH, de Vijlder JJ. Matemal-fetal transfer of thyroxine in
congenital hypothyroidism due to a total organiﬁcation defect or thyroid agenesis.
N Engl J Med 1989;321:13-6. '

Markham LA, Stevens DL. A case report of neonatal thyrotoxicosis due to
maternal autoimmune hyperthyroidism. Adv Neonatal Care 2003;3z272-82; quiz
283-5.

Walker P, Dussault J H, Hart IR, Langelier P, Szots F. Thyrotoxicosis detected in
a mass-screening program for neonatal hypothyroidism: demonstration of
placental transfer of an immunoglobulin with marked lipolytic activity. J Pediatr

1977;91:400-3.

Fisher DA. Thyroid function in premature infants. The hypothyroxinemia of
prematurity. Clin Perinatol 1998;25:999-1014, viii.

Hume R, Simpson J, Delahunty C, et al. Human fetal and cord serum thyroid
hormones: developmental trends and interrelationships. J Clin Endocrinol Metab
2004;89:4097-103.

Abuid J, Klein AH, Foley TP, Jr., Larsen PR. Total and free triiodothyronine and
thyroxine in early infancy. J Clin Endocrinol Metab 1974;39:263-8.

50

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

Abuid J, Stinson DA, Larsen PR. Serum triiodothyronine and thyroxine in the
neonate and the acute increases in these hormones following delivery. J Clin
Invest 1973;52:1195-9.

Erenberg A, Phelps DL, Lam R, Fisher DA. Total and free thyroid hormone
concentrations in the neonatal period. Pediatrics 1974;53:21 1-6.

Jacobsen BB, Hummer L. Changes in serum concentrations of thyroid hormones
and thyroid hormone-binding proteins during early infancy. Studies in healthy
fulltenn, small-for-gestational age and preterm infants aged 7 to 240 days. Acta
Paediatr Scand 1979;68:411-8.

Polk DH. Thyroid hormone effects on neonatal thermogenesis. Semin Perinatol
1988;12:151-6.

Bird JA, Clarke L, Symonds ME. Inﬂuence of thyrotrophin-releasing hormone on
thermoregulation in newborn lambs. Biol Neonate 1998;73:52-9.

Cabello G. Endocrine reactivity (T3, T4, cortisol) during cold exposure in preterm
and full-term lambs. Biol Neonate 1983;44:224-33.

Heasman L, Clarke L, Symonds ME. Inﬂuence of thyrotropin-releasing hormone
administration at birth on thermoregulation in lambs delivered by cesarean. Am J
Obstet Gynecol 2000;183:1257-62.

Polk DH, Callegari CC, Newnham J, et al. Effect of fetal thyroidectomy on
newborn thermogenesis in lambs. Pediatr Res 1987;21:453-7.

Gluckman PD, Sizonenko SV, Bassett NS. The transition from fetus to neonate--
an endocrine perspective. Acta Paediatr Suppl 1999;88:7-11.

Liggins GC. The role of cortisol in preparing the fetus for birth. Reprod Fertil Dev
1994;6:141-50.

Nathanielsz PW. Thyroid function in the fetus and newborn mammal. Br Med
Bull 1975;31:51-6.

Mercado M, Szymonowicz W, Yu VY, Gold H. Symptomatic hypothyroxinemia
with normal TSH levels in preterm infants. Clin Pediatr (Phila) 1987;26:343-6.

Den Ouden AL, Kok JH, Verkerk PI-I, Brand R, Verloove-Vanhorick SP. The
relation between neonatal thyroxine levels and neurodevelopmental outcome at
age 5 and 9 years in a national cohort of very preterm and/or very low birth
weight infants. Pediatr Res 1996;39:142-5.

51

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

Hadeed AJ, Asay LD, Klein AH, Fisher DA. Signiﬁcance of transient postnatal
hypothyroxinemia in premature infants with and without respiratory distress
syndrome. Pediatrics 1981;68:494-8.

Lucas A, Morley R, F ewtrell MS. Low triiodothyronine concentration in preterm
infants and subsequent intelligence quotient (IQ) at 8 year follow up. ij
1996;312:1132-3; discussion 1133-4.

Mercado M, Yu VY, Francis 1, Szymonowicz W, Gold H. Thyroid function in
very preterm infants. Early Hum Dev 1988;16:131-41.

Paul DA, Leef KH, Stefano JL, Bartoshesky L. Thyroid function in very-low-
birth-weight infants with intraventricular hemorrhage. Clin Pediatr (Phila)
2000;39:651-6.

Rabin CW, Hopper AO, Job L, et al. Incidence of low free T4 values in premature
infants as determined by direct equilibrium dialysis. J Perinatol 2004;24:640-4.

Reuss ML, Leviton A, Paneth N, Susser M. Thyroxine values from newborn
screening of 919 infants born before 29 weeks' gestation. Am J Public Health
1997;87:1693-7.

Romagnoli C, Curro V, Luciano R, et al. Serial blood T4 and TSH determinations
in low birth weight infants. Inﬂuence of gestational age, birth weight and neonatal
pathology on thyroid function. Helv Paediatr Acta 1982;37:331-44.

Uhrmann S, Marks KH, Maisels MJ, Kulin HE, Kaplan M, Utiger R. Frequency
of transient hypothyroxinaemia in low birthweight infants. Potential pitfall for
neonatal screening programmes. Arch Dis Child 1981;56:214-7.

Belet N, Imdat H, Yanik F, Kucukoduk S. Thyroid function tests in preterm
infants born to preeclamptic mothers with placental insufﬁciency. J Pediatr
Endocrinol Metab 2003;16:1 131-5.

Biswas S, Buffery J, Enoch H, Bland JM, Walters D, Markiewicz M. A
longitudinal assessment of thyroid hormone concentrations in preterm infants
younger than 30 weeks' gestation during the ﬁrst 2 weeks of life and their
relationship to outcome. Pediatrics 2002;109:222-7.

Fetter WP, Waals-Van de Wal CM, Van Eyck J, Samson G, Bongers-Schokking
JJ. Thyroid hormone concentrations in preterm infants born to pre-eclamptic
women with placental insufﬁciency. Acta Paediatr 1998;87:186-90.

Williams FL, Mires GJ, Barnett C, et al. Transient hypothyroxinemia in preterm

infants: the role of cord sera thyroid hormone levels adjusted for prenatal and
intrapartum factors. J Clin Endocrinol Metab 2005;90:4599-606.

52

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

76.

Adams LM, Emery JR, Clark SJ, Carlton EI, Nelson JC. Reference ranges for
newer thyroid function tests in premature infants. J Pediatr 1995;126:122-7.

Carrascosa A, Ruiz-Cuevas P, Potau N, et al. Thyroid ﬁmction in seventy-ﬁve
healthy preterm infants thirty to thirty-ﬁve weeks of gestational age: a prospective
and longitudinal study during the ﬁrst year of life. Thyroid 2004;14:435-42.

Hirano T, Singh J, Srinivasan G, Pildes R. Post-natal thyroid function in low birth
weight infants; a longitudinal assessment of free thyroxine and thyroid hormone
binding globulin. Acta Endocrinol (Copenh) 1985;110:56-60.

Chen JY. Thyroid function in healthy and sick neonates. Zhonghua Yi Xue Za Zhi
(Taipei) 1994;54:51-6.

Etling N, Padovani E, Gehin-Fouque F, Tato L. Serum and urine thyroid hormone
levels in healthy preterm and small for date infants on the ﬁrst and ﬁfth day of life.
Helv Paediatr Acta 1984;39:223-30.

Harkavy KL, Enecio CE. Free thyroxine levels in hospitalized newborns:
depressed levels in critical, nonthyroidal illness. J Perinatol 1991;] 1:117-21.

John R, Bamforth FJ. Serum free thyroxine and free triiodothyronine
concentrations in healthy fullterm, preterm and sick preterm neonates. Ann Clin
Biochem 1987;24 ( Pt 5):461-5.

Uhrmann S, Marks KH, Maisels MJ, et al. Thyroid function in the preterm infant:
a longitudinal assessment. J Pediatr 1978;92:968-73.

Diamond FB, Jr., Root AW. Thyroid function in the preterm infant. Adv Exp Med
Biol 1991;299:227-41.

Fuse Y, Shimizu M, Uga N, Fujii T, Irie M. Maturation of feedback control of
thyrotropin in premature infants. J Dev Physiol 1990;14:17-22.

Rooman RP, Du Caju MV, De Beeck LO, Docx M, Van Reempts P, Van Acker
KJ. Low thyroxinaemia occurs in the majority of very preterm newborns. Eur J
Pediatr 1996;155:211-5.

van Wassenaer AG, Kok JH, de Vijlder JJ, et al. Effects of thyroxine
supplementation on neurologic development in infants born at less than 30 weeks'
gestation. N Engl] Med 1997;336:21-6.

De Felice C, Bagnoli F, Toti P, et al. Transient hypothyroxinemia of prematurity
and histological chorioamnionitis. J Perinat Med 2005;33:514-8.

53

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

Martin CR, Van Marter LJ, Allred EN, Leviton A. Antenatal glucocorticoids
increase early total thyroxine levels in premature infants. Biol Neonate
2005;87:273-80.

Martin CR, Van Marter LJ, Allred EN, Leviton A. Growth-restricted premature
infants are at increased risk for low thyroxine. Early Hum Dev 2001;64:119-28.

Dammann O, Leviton A. Maternal intrauterine infection, cytokines, and brain
damage in the preterm newborn. Pediatr Res 1997;42:1-8.

Dammann O, Leviton A. Inﬂammation, brain damage and visual dysfunction in
preterm infants. Semin Fetal Neonatal Med 2006;11:363-8.

Williams MC, O'Brien WF, Nelson RN, Spellacy WN. Histologic
chorioamnionitis is associated with fetal growth restriction in term and preterm
infants. Am J Obstet Gynecol 2000;183:1094-9.

Buimer M, van Wassenaer AG, Ganzevoort W, Wolf H, Bleker OP, Kok JH.
Transient hypothyroxinemia in severe hypertensive disorders of pregnancy.
Obstet Gynecol 2005;106:973-9.

Franklin RC, Purdie GL, O'Grady CM. Neonatal thyroid function: prematurity,
prenatal steroids, and respiratory distress syndrome. Arch Dis Child 1986;61:589-
92.

Reuss ML, Paneth N, Lorenz J M, Susser M. Correlates of low thyroxine values at
newborn screening among infants born before 32 weeks gestation. Early Hum
Dev 1997;47:223-33.

Cuestas RA, Engel RR. Thyroid function in preterm infants with respiratory
distress syndrome. J Pediatr 1979;94:643-6.

Huang CB, Chen F S, Chung MY. Transient hypothyroxinemia of prematurity is
associated with abnormal cranial ultrasound and illness severity. Am J Perinatol
2002;19:139-47.

Reuss ML, Paneth N, Pinto-Martin JA, Lorenz JM, Susser M. The relation of
transient hypothyroxinemia in preterm infants to neurologic development at two
years of age. N Engl J Med 1996;334:821-7.

Williams FL, Ogston SA, van Toor H, Visser TJ, Hume R. Serum thyroid

hormones in preterm infants: associations with postnatal illnesses and drug usage.
J Clin Endocrinol Metab 2005;90:5954-63.

54

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

99.

100.

Weber G, Vigone MC, Rapa A, Bona G, Chiumello G. Neonatal transient
hypothyroidism: aetiological study. Italian Collaborative Study on Transient
Hypothyroidism. Arch Dis Child Fetal Neonatal Ed 1998;79zF70-2.

Bemal J, Pekonen F. Ontogenesis of the nuclear 3,5,3'-triiodothyronine receptor
in the human fetal brain. Endocrinology 1984;] 14:677-9.

Dobbing J, Sands J. Timing of neuroblast multiplication in developing human
brain. Nature 1970;226:639-40.

Auso E, Lavado-Autric R, Cuevas E, Del Rey FE, Morreale De Escobar G, Berbel
P. A moderate and transient deﬁciency of maternal thyroid function at the
beginning of fetal neocorticogenesis alters neuronal migration. Endocrinology
2004;145:4037-47.

Berbel P, Guadano-Ferraz A, Martinez M, Quiles JA, Balboa R, Innocenti GM.
Organization of auditory callosal connections in hypothyroid adult rats. Eur J
Neurosci 1993;5: 1465-78.

Gravel C, Hawkes R. Maturation of the corpus callosum of the rat: 1. Inﬂuence of
thyroid hormones on the topography of callosal projections. J Comp Neurol
1990;291:128-46.

Gravel C, Sasseville R, Hawkes R. Maturation of the corpus callosum of the rat:
11. Inﬂuence of thyroid hormones on the number and maturation of axons. J Comp
Neurol 1990;291:147-61.

Lavado-Autric R, Auso E, Garcia-Velasco JV, et al. Early maternal
hypothyroxinemia alters histogenesis and cerebral cortex cytoarchitecture of the
progeny. J Clin Invest 2003;] l 1:1073-82.

Lucio RA, Garcia JV, Ramon Cerezo J, Pacheco P, Innocenti GM, Berbel P. The
development of auditory callosal connections in normal and hypothyroid rats.
Cereb Cortex 1997;7:303-l6.

Morreale de Escobar G. The role of thyroid hormone in fetal neurodevelopment. J
Pediatr Endocrinol Metab 2001;14 Suppl 6: 1453-62.

Lucas A, Rennie J, Baker BA, Morley R. Low plasma triiodothyronine
concentrations and outcome in preterm infants. Arch Dis Child 1988;63:1201-6.

Meijer WJ, Verloove-Vanhorick SP, Brand R, van den Brande J L. Transient

hypothyroxinaemia associated with developmental delay in very preterm infants.
Arch Dis Child 1992;67:944-7.

55

10].

102.

103.

104.

105.

Amato M, Guggisberg C, Schneider H. Postnatal triiodothyronine replacement
and respiratory distress syndrome of the preterm infant. Horm Res 1989;32:213-7.

Amato M, Pasquier S, Carasso A, Von Muralt G. Postnatal thyroxine
administration for idiopathic respiratory distress syndrome in preterm infants.
Horm Res 1988;29:27-30.

Chowdhry P, Scanlon J W, Auerbach R, Abbassi V. Results of controlled double-
blind study of thyroid replacement in very low-birth-weight premature infants
with hypothyroxinemia. Pediatrics 1984;73:301-5.

Schonberger W, Grimm W, Emmrich P, Gempp W. Reduction of mortality rate in
premature infants by substitution of thyroid hormones. Eur J Pediatr
1981;135:245-53.

Vanhole C, Aerssens P, Naulaers G, et al. L-thyroxine treatment of preterm
newborns: clinical and endocrine effects. Pediatr Res 1997;42:87-92.

56

Chapter 2 Methodology

This chapter includes four major aspects of the Phase 1 Clinical Trial of Thyroid
Hormone in Extremely Premature Infants (THOP): 1) study design; 2) study population;
3) exposures of interest and outcomes of interest; 4) statistical analysis. In the last
section, two research questions of this dissertation are framed, followed by their
analytical samples statistical methods used.

2.1 Study Design (Table 2.1)

The Thyroid Hormone in Extremely Premature Infants study, a multi-center
randomized clinical trial, is designed to establish the optimal dosing schedule for very
preterm infants to increase plasma T4 and T3 levels to the normal range without
increasing risks of morbidity or mortality, and without suppressing TSH. This trial was
conducted at the neonatology intensive care units of three hospitals: Maria F areri
Children’s Hospital at Westchester Medical Center, Valhalla, New York; Emma
Children’s Hospital, Amsterdam, Netherlands; and University Hospital La Paz, Madrid,
Spain. The data management team at Michigan State University, East Lansing, Michigan,
was responsible for data management and monitoring.

The THOP trial was designed to have six study arms. Four treatment groups
received 4 or 8 pg/kg/day of T4 by bolus or by continuous infusion in the 6 weeks of life,
and also received 1 pg/kg/day of T3 by continuous infusion in the ﬁrst 2 weeks of life. A
ﬁfth group received 30 ug/kg/day of potassium iodide (K1), in the form of SSKI drops,
via enteral route, for 6 weeks. The last group was a control group, subjects of which

received Dextrose 5% in Water (DSW) by continuous infusion for 6 weeks. Treatment

57

was required to begin within 24 hours after birth. The six study treatment groups were
coded as 1, 2, 3, 4, 5, and 6. The contents of group 1-6 were only revealed to study
pharmacists and the data management team. Patients, clinicians, and study coordinators
were blinded to the treatment throughout the study period, except the group 6, which
revealed itself by the enteral route.

From April 2005 to March 2007, 168 preterm infants born between 24 and 28
completed weeks of gestational age were enrolled. The randomization procedure was
ﬁrst stratiﬁed by study centers to ensure the enrollment ratio is 2: l :1 in New York,
Amsterdam, and Madrid respectively. Gestational age (below or above 26 weeks) and
gender (male or female) were also used in stratiﬁcation to minimize imbalance in
gestational age and gender across six study groups at enrollment.

Subjects who expired or who withdrew from the THOP study prior to 7 days after
birth were considered as non-completers. As the study was designed to monitor hormone
changes over time (from birth to discharge), non-completers would not provide sufficient
information on hormone levels, we replaced them with the next eligible subject.
Information of non-completers was kept in the database for analysis. A note was
provided to identify the reason for not being able to complete the study treatment.

2.2 Inclusion and exclusion criteria (Table 2.2)

We screened all preterm infants who were born in or transferred to the three study
centers during the enrollment period and whose gestational age was between 24 and 28
completed weeks. Gestational age estimation was based on the last menstrual period and

was also conﬁrmed by obstetric ultrasound scans.

58

Infants born to women less than 18 years of age were excluded because of the
complexity of getting parental consent for an underaged mother. Infants with congenital
thyroid disease or major/lethal anomalies, such as trisomy 13 and 18, CNS structural
anomaly, and so on, were excluded. Infants born to mothers who had thyroid disease,
who planned to breastfeed meanwhile taking anti-thyroid medications/supplemental iodine in
therapeutic doses, or who had substance abuse history during the current pregnancy were all
excluded. We also avoided extremely sick infants, for instance, liver failure, sepsis, severe
cardiopulmonary disease, etc. Neonates who were enrolled in another clinical trial, or have no
parental consent or assent from treating neonatologist were also excluded.

2.3 Adverse events

As in all clinical trials, adverse events were closely monitored throughout the
study period and were generally reported within 24 to 48 hours of detection. In this trial,
we classiﬁed adverse events into two categories: 1) severe adverse events, which include
death and brain parenchymal lesion on ultrasound; 2) mild/moderate adverse events,
which include changes in thyroid gland function, pulmonary cardiovascular status,
metabolic status, gastrointestinal system status, central nervous system status, and skin
system status. lnfonnation of each adverse event was recorded in detail, including the
starting and ending time points, symptoms of the adverse event, concomitant medications
used for treating the adverse event, and the outcome of the adverse event. Whether and
when the study treatment was suspended, stopped, or restarted were also documented.

The attending neonatologist was responsible for ﬁlling the study adverse event
form online. An instant email notiﬁcation was sent to three parties: the data management
team at MSU, which would update the mortality report if it was a death; the Data Safety

Monitoring Board (DSMB), formed by clinicians, statisticians, and project ofﬁcers

59

appointed by the funding agency of this study---the National Institute of Health to
oversee the THOP trial; and the Independent Medical Monitor (IMM), who, a
neonatologist in another institute blinded from study treatment, was responsible for
making an independentjudegement of whether the adverse event was related to the study
treatment based on the information provided in the study adverse event form.

2.4 Study Population (Figure 2.1)

During the two-year enrollment period, 239 preterm infants born between 24 and
28 completed weeks of gestational age were screened. 188 (188/239, 78.7%) newborns
were eligible for enrollment. 168 (89.4%) neonates, born to 151 women, were consented
and randomized. A total of 32 study subjects expired before hospital discharge, among
which 7 expired before day 7 of life. Six subjects withdrew from the study, four of
whom withdrew before day-7 of life and 2 withdrew at day 12 and day 41 of life
respectively. The total number of non-completers is l 1.

2.4.1 Early termination of group 2 (Table 2.3 and Figure 1)

As of July 21, 2006, 113 subjects were enrolled in the THOP trial. 19 subjects
expired, 4 of which died prior to day 7 of life. Table 2.3 lists the deaths and mortality
rates of each group at that time. At the time 20 subjects were enrolled in group 2. No
death occurred before 7 days of life and the mortality rate of this group was 30%, which
was higher than any other group, although not statistically signiﬁcant. This high
mortality rate raised concern among the DSMB members, which believed that the group
2 dosing level put patients in this group at an unacceptable risk. On July 21, 2006, the
DSMB notiﬁed the THOP study team to stop recruiting study subjects into Group 2. The

decision was immediately carried out in three study sites. The randomization program

60

was subsequently modiﬁed to adapt to the change and to maintain the balance across the
rest ﬁve study groups.
2.4.2 Unexpected dosing and coding errors in Madrid site (Table 2.4)

On October 24, 2006, a double dosing error was reported by the study P] from the
Madrid site. In the THOP trial, study drug was shipped directly from the pharmaceutical
company to each study site. The local pharmacist of each study site is responsible for
diluting the study drug into two different doses used in the THOP trial. In each package,
a standard manufacture dilution instruction was included, however it was different from
the THOP dilution protocol written in the manual of procedure. The hospital pharmacist
in Madrid site followed the manufacture dilution instruction instead of the study manual
of procedure, which omitted one dilution step and led to double dosing error in all
treatment groups. The same dilution procedure had been applied in Madrid since the
beginning of the THOP trial in Madrid, until the error was detected.

While investigating the dosing error, a second error was found in Madrid.
Unbeknownst to the data center, the Madrid site, at the randomization stage, used a group
coding system different from the one assigned to all study sites. In itself, this violation
would not have violated the principles of randomization, since all groups were assigned
randomly from a central minimization/randomization program. However, part of the
study subjects enrolled in Madrid site were classiﬁed into wrong groups. Table 2.4
compared the difference of the group coding between Madrid site and the THOP study.

As a result of the double-dosing error, we lost all subjects in two 4 ug/kg/day T4
groups in Madrid and accidentally created two new groups, 16 ug/kg/day T4 by

continuous or bolus infusion.

61

2.4.2.1 Impact of double dosing error and coding error in Madrid on study group
assignment (Table 2.5)

While one might have expected all six Madrid groups to differ from the protocol
because of the Madrid coding error, as seen in table 2.5, only three actually did. This is
because 1) group 6 was the K1 group in all three sites, and 2) Madrid coded group 2 as B4
group and group 3 as C4 group, whereas they were supposed to be B8 and C8 groups
respectively. The double dosing error restored these two groups to their original
assignment.
2.4.2.2 Regrouping procedure

After the errors were identiﬁed, we regrouped the study participants in Madrid
and put the mis-grouped subjects into the correct dosing groups based on the true
treatment they received. The regrouping procedure followed the rules below:

Subjects assigned to group 1 in Madrid were regrouped to group 4 of THOP;
Subjects assigned to group 2 in Madrid remained in group 2 of THOP;
Subjects assigned to group 3 in Madrid remained in group 3 of THOP;
Subjects assigned to group 4 in Madrid were regrouped to group 7 of THOP;
Subjects assigned to group 5 in Madrid were regrouped to group 8 of THOP;
Subjects assigned to group 6 in Madrid remained in group 6 of THOP.

Two subjects were enrolled in Madrid after the dosing error was corrected. One
subject was randomized to group 6, and therefore remained in group 6 of THOP. The
other subject was randomized to Madrid group 3, which was group 1 of THOP after
regrouping.
2.4.2.3 Termination of enrollment and treatment in Madrid

On October 25, 2006, The DSMB decided to close Madrid site for both

enrollment and treatment. This decision turned the two last-enrolled subjects, who

62

received correct treatment dose from the beginning, into non-completers by ending the
treatment before 7 days of life and also withdrew the study treatment from another one
subject, who had been enrolled for 12 days at that time. _

2.4.2.4 Impact of the errors on study sample size

After regrouping based on the true treatment each subject received, Madrid had
subjects in groups 1, 2, 3, 4, and 6 of THOP. No subject in Madrid received 4 ug/kg/day
T4 treatment except the last enrolled subject, who was enrolled after the dosing error was
corrected.

The total loss of study sample size was 17, of which 15 received incorrect
treatment dose and 2 became non-completers by withdrawing from treatment prior to day
7 of life. The effective sample size of the THOP trial went down from 139 to 124. The
New York and Amsterdam sites continued enrollment until the end of the trial.

2.5 Collection of biological specimens
2.5.1 Blood specimens

A series of 2-milliliter (m1) neonatal blood sample were collected for every study
subject at enrollment (day 0), on day 3, 7, l4, 2], 42, 56, and at discharge. Serum Free
T4 (F T4), Total T4 (TT4), Total T3 (1T3), TSH, thyroid binding globulin (TBG), and
cortisol levels were measured in neonatal blood specimens. Neonatal serum hormone
results were available for 164 study subjects. Four subjects did not have hormone results
because of either early withdrawal from the study or insufﬁcient specimen volume for
laboratory testing.

A 3-ml maternal venous blood was collected at enrollment. FT4, TT4, and TSH

were measured in maternal blood samples. Blood samples were not obtained for 25

63

women because some mothers were discharged home early or project manager missed
maternal visits. Hormone testing results of one woman was excluded in the analysis
because her blood sample was taken 18 days after delivery, which was too far away from
delivery and may not reﬂect the thyroid function at delivery. Maternal serum hormone
results were available for 125 mothers.
2.5.2 Urine and breast milk specimens

A series of 2 to 5-ml urine samples were collected for each study subject at
enrollment (day 0), on day 7, 21, 42, 56, and at discharge. A single 2 to 5-ml urine
sample was taken from participating mothers at enrollment. A series of 2-ml breast milk
samples were collected from mothers whose breast milk was available on day 3, 7, 14, 21,
42, and 56. Iodine levels were measured in all urine and breast-milk samples.
2.5.3 Storage, transportation, and assessment of biological specimens

All specimens were stored at -20 degrees Celsius (°C). Blood specimens were
transported in dry ice by over-night FedEx to the Quest Diagnostics, at Chantilly,
Virginia, where all hormone assessments were completed. Direct equilibrium dialysis
(DED) was used to assess serum FT4 levels and chemiimmunoluminescent assay was
used for other hormone assessments. In a situation when sample volume was insufﬁcient
to complete the tests of all six biomarkers, we prioritized FT4 testing and then TSH, TT4,
TT3, TBG, and cortisol.

Urine and breast-milk specimens were transported in dry ice to the Molecular
Endocrinology Lab at the Institute of Biomedical Research “Alberto Sols”, Autonomous
University of Madrid, in Madrid, Spain. Ammonium Persulfate method and Chloric Acid

Digestion method were used for urine and breast milk iodine assessment respectively.

64

2.6 Exposure of interest

We are interested in three groups of exposures in this clinical trial. First, intra-
uterine exposures. At enrollment, a concise questionnaire was used to collect maternal
information on health history/behavior (e.g. gravidity, parity, addiction, etc.) and current
pregnancy (e. g. preeclampsia, fever, HELLP syndrome, etc.). Medication prescribed to
mothers prior to delivery was also recorded (e.g. antibiotics, magnesium, steroids, etc.).
Second, effects of breast-feeding. Iodine, immunoglobulins, and very low levels of
thyroid hormones would be transported to the neonates through breast-milk (1, 2). We
measured the iodine levels in breast-milk, maternal urine, and neonatal urine and adjusted
the iodine levels in thyroid hormone analysis. Third, neonatal exposures. Neonatal birth
characteristics, such as birthweight, gestational age, multiple birth status, and Apgar
score at 1 and 5 minutes, were collected at enrollment for each study subject. Daily
records (such as vital signs, blood gases, calorie intake, growth, respiratory support, etc.)
and illness episodes during neonatal period were also recorded in detail.
2.7 Outcomes of interest

As the ultimate goal of this study was to look for the optimal dosing schedule of
thyroid hormone treatment for very preterm infants to increase plasma T4 and T3 levels
to normal range, without increasing risks of morbidity or mortality, and without
suppressing TSH, the primary outcomes were the neonatal thyroid hormone levels (FT4,
TT4, TT3, and TSH) at different time points, death, severe morbidity such as brain lesion.

Most studies measured serum FT4 concentrations by analog immunoassay or FT4
index methods, both of which could be biased by the serum protein levels and/or protein-

bound T4 concentrations (3). Hence low FT4 values of preterm infants could be partially

65

attributed to reduced T4 binding to proteins. DED method is by far the least biased
method by changes in serum proteins or protein binding of T4 and we believe it can most
accurately reﬂect the serum FT4 levels in extremely premature infants (4).

Each death was reported to the study P], the data management team, the DSMB,
and the IMM within 24 hours of occurrence. The DSMB received appropriate mortality
analyses update from the data management team within 1-2 days.

Brain lesions (such as parenchymal lesion, hemorrhagic lesions, periventricular
leukomalacia (PVL), hydrocephalus, etc.) were diagnosed through cranial ultrasound
scans. All subjects received three cranial ultrasound scans at postnatal day 1-3, day 7-10,
and > 4 weeks. All scans were sent to one central ultrasound reader, who was blinded
from treatment assignment throughout the study period.

Secondary outcomes include neonatal morbidities, such as chronic lung disease,
necrotizing enterocolitis, retinopathy of prematurity, etc., which were recorded in the
study discharge/death form.

2.8 Statistical analysis
In this dissertation, two research questions will be studied. First, what are the

correlates of thyroid hormone levels at birth among extremely premature infants (§ 28

wks)? Linear regression (with log-transformation) will be used for both unadjusted and
adjusted analyses.
Second, what are the correlates of postnatal thyroid hormone levels among

extremely premature infants (._<_ 28 wks)? Repeatedly measured neonatal thyroid

hormone levels will be treated as an outcome in a whole. Proc Mixed procedure will be

used for analysis to account the between and within subjects variations. Backward model

66

selection approach will be applied. All analyses will be carried out using SAS soﬁware

9.1.3.

67

Table 2.1 The six THOP arms.

 

 

Study arm Treatment Group abbreviation

1 Continuous Infusion T4(4ug/kg/d) X 42days + C4
Continuous Infusion T3 (lug/kg/d) X l4days

2 Bolus InﬁJsion T4 (8ug/kg/d) X 42 days + B8
Continuous Infusion T3 (lug/kg/d) X 14 days '

3 Continuous Infusion T4 (8ug/kg/d) X 42 days + C8
Continuous Infusion T3(lug/kg/d) X l4days

4 Placebo (DSW infusion) Placebo

5 Bolus Infusion T4 (4ug/kg/d) X 42 days + B4
Continuous Infusion T3 (lug/kg/d) X 14 days

6 Iodine Supplementation alone (SSKI) (30pg/kg/d) X 42 days Iodine

 

Table 2.2 Inclusion and exclusion criteria of the THOP trial.
Inclusion criterion:

Infants born between 24 and 28 complete weeks of gestational age (GA, conﬁrmed
by early obstetric ultrasound scan).

Exclusion criteria:

1. Mother of newborn is less than 18 years of age.
2. Maternal or congenital thyroid disease (e.g. propylthiouracil, PTU).
3. Women who plan to breast feed and who take anti-thyroid medications or
supplemental iodine in therapeutic doses.
4. Substance abuser (heroin, methadone, alcohol > 102 per day, or other).
5. Known disorders of fatty acid metabolism or inherited metabolic diseases.
6. Liver Failure (e.g. enzymes in the thousands).
7. Newborns with known major or lethal anomalies such as:
1) Known chromosomal anomalies T risomy l3 & 18.
2) CNS structural anomaly (e. g. hydrocephalus, migration disorders, etc. ).
3) Major congenital malformations requiring surgery.
4) Structural congenital heart disease (e. g. cyanotic heart defects).
5) Anomalies of the gastrointestinal tract (e. g. gastroschesis).
8. Newborn with arrhythmia.
9. Proven sepsis at birth (clinical signs, positive body ﬂuid culture, and antibiotic
treatment > 5 contiguous days).
10. Death expected within 48 hours of birth due to severity of initial cardiopulmonary
disease.
11. A concurrent clinical trial with another randomized drug.
12. Other concern by treating physician that either mandates or prohibits study
treatment.
13. Absence of treating neonatal physician's assent.
14. Absence of parental written consent.

 

68

Table 2.3 Deaths (total, < 7 days, and g 7 days) and mortality rates (total, < 7 days, and
2 7 days) by group assignment in the THOP trial as of July 21, 2006.

 

 

Study group 1 2 3 4 5 6 p-value Total
Number of subjects 19 20 20 19 19 16 113
All deaths 2 6 2 4 . 2 3 0 3 19
(Mortality %) (10.5) (30) (10) (21.1) (10.5) (18.8) 50 (16.8)
Deaths < 7 days 1 0 2 1 0 o 4
(Mortality %) l (5.3) (0) (10) (5-3) (0) (0) (3-5)
Deaths 2 7 days 1 6 0 3 2 3 3 15
(Mommy %) 2 (5.6) (30) (0) (16.7) (10.5) (18.8) 0“ (13.8)

 

l The total number of subjects enrolled in each group was used as the denominator;
2 The total number of subjects survived g 7 days in each group was used as the denominator.

3 p-value by exact analysis (Pearson test).

Table 2.4 The study group coding difference between Madrid site and the THOP trial.

 

 

 

smdy Gm" THOPGmup cadnlléadrid
1 C4 Placebo
2 B8 B4
3 C8 C4
4 Placebo C8
5 B4 BB
6 Iodine Iodine

 

Table 2.5 Impact of Madrid double dosin and coding error on treatment groups.

 

 

 

 

 

 

 

Madrid group Impact of Madrid Impact of Madrid double dosing error and
coding double dosing error coding error in comparison with the THOP
on the groups protocol grouping standard

1 Placebo 1 Placebo 4 Placebo

2 B4 2 B8 2 B8

3 C4 3 C8 3 C8

4 C3 4 C16 No corresponding group — named as group 71
. 2

5 I38 5 I316 No corresponding group — named as goup 8

6 Iodine 6 Iodine 6 Iodine

 

 

 

 

 

 

 

1 Group 7: Continuous infusion T4 (16 ug/kg/d) X 42 days + continuous infusion T3 (1 pg/kg/d) X 14

days;

2 Group 8: Bolus infusion T4 (16 ug/kg/d) X 42 days + continuous infusion T3 (1 ug/kg/d) X 14 days;

69

 

Figure 2.1 Flow chart of the THOP enrollment.

 

 

239 newborns between 24-28 wks screened

 

 

 

 

 

205 newborns eligible for enrollment

 

 

 

\37 subjects refused
r

 

 

I68 newborns consented and enrolled

 

 

h

 

\34 subjects were ineligible
r

—>Born to 151 mothers

 

 

i I
Placebo Iodine B4 C4 88 C8
(n=27) (n=24) (n=24) (n=24) (n=20) (n=25)

 

 

 

 

 

 

 

 

 

 

 

 

 

L__ 144 effective study subjects __I

born to 129 mothers

70

7 subjects expired prior to 7 days of life (non-completers)

4 subjects withdrew consent prior to 7 days of life (non-completers)

: ............ .: - =
+3 816 §;+‘ C16 g=ls7§tudy
(0:6) (n=7) subjects

‘0 ooooooooooooo

o IIIIIIIIIIII $

Born to 141
mothers

BIBLIOGRAPHY
Smith VC, Svoren BM, Wolfsdorf J I. Hypothyroidism in a breast-fed preterm
infant resulting from maternal topical iodine exposure. J Pediatr 2006;149:566-7.
van Wassenaer AG, Stulp MR, Valianpour F, et al. The quantity of thyroid
hormone in human milk is too low to inﬂuence plasma thyroid hormone levels in

the very preterm infant. Clin Endocrinol (Oxf) 2002;56:621-7.

Nelson JC, Weiss RM, Wilcox RB. Underestimates of serum free thyroxine (T4)
concentrations by free T4 immunoassays. J Clin Endocrinol Metab 1994;79:76-9.

Rabin CW, Hopper A0, Job L, et al. Incidence of low free T4 values in premature
infants as determined by direct equilibrium dialysis. J Perinatol 2004;24:640-4.

7]

Chapter 3 Results: Population characteristics

This chapter describes the general characteristicsof the THOP cohort.
Information covers: 1) attributes of 239 potential subjects screened for eligibility in the
two-year enrollment period; 2) maternal characteristics, such as demographic information
(e. g. maternal age, ethnicity) and health history during current pregnancy; and 3) neonatal
characteristics, such as demographic and birth attributes.

3.1 Recruitment (Table 3.1)

Among 239 potential subjects we screened, 168 (70%) were successfully enrolled
in this clinical trial. The major reasons for declining participation is that parents of 36
(15%) potential subjects were either not interested in research, afraid of research, too
stressed to decide, or wanted incentives to participate. 2 (0.9%) refused because of
religious beliefs.

33 potential subjects were ineligible for enrollment, among which 3 (1.3%) were
contacted after the study eligibility time frame (24 hours after birth), 10 (4.2%) were very
sick at birth and death was expected within 48 hours, 8 (3.3%) had mothers younger than
18 years old, and another 8 (3.3%) had mothers with diseases listed in the exclusion
criteria. Ineligible for enrollment due to other reasons, such as maternal addiction (I,
0.4%), mother too sick to approach (1, 0.4%), and iodine exposure prior to delivery (2,
0.9%) were all under 1%.

3.2 Comparison of consented and refused subjects (Table 3.2)
We compared study site, race, gender, birthweight, and gestational age of subjects

who were enrolled in the THOP study versus those who were not. Subjects of Hispanic

72

origin were less like to be enrolled in the THOP study, with a relative risk of 0.4 (95% CI
0.17-0.83) compared to whites, 0.3 (95% CI 0.1-0.7) compared to blacks, and 0.2 (95%
Cl 002-15) compared to other race category. Except for race, other factors did not seem
to inﬂuence parental decision making in trial participation.

3.3 Study cohort (Table 3.3-3.4)

During the two-year enrollment period, we recruited a total of 168 subjects. 77
(45.8%) were enrolled in New York, 52 (31%) in Amsterdam, and 39 (23.2%) in Madrid.
Parents of four subjects withdrew their consent within 7 days after birth and another two
withdrew after 7 days of life. Seven subjects (4.3%) expired prior to day 7 of life and 25
(15.9%) more died after 7 days. About 63% of the study population is white, 24% is
black, 9% is Hispanic, and other ethnicities account for about 4%.

Birthweight of all study subjects (n=168) ranged between 440 to 1550 grams with
a mean of 851 grams and a standard deviation of 190.7 (se=l4.7). Gestational age ranged
between 24 to 27.9 completed weeks with a mean of 26.2 weeks and a standard deviation
of 1.1 (se=0.1).

Among the 168 enrolled subjects, we consider 144 of them as effective study
subjects, since 9 subjects either expired or withdrew from the study prior to day 7 of life
and 15 subjects received 16 ug/kg/day thyroxine due to an unexpected dilution error as
aforementioned in chapter 2. With the early termination of group 2 (Bolus 8 pg group) at
group size of 20, we enrolled 4 more subjects to complete the projected sample size of

144. These four subjects were randomized to the placebo group (n=3) and the continuous

8 ug group (n=1).

73

3.4 Maternal characteristic of the T HOP cohort

The 168 subjects were born to 151 women. Maternal information is complete for
150 women. 73 (48.3%) were enrolled from New York site, 42 (27.8%) were enrolled
from Amsterdam site, and 36 (23.8%) were enrolled from Madrid site. 62% of mothers
were white, 25% were black, and 8% were Hispanic.

During the current pregnancies of the 150 mothers, 4 (2.7%) had HELLP, 20
(13.4%) had preeclampsia, 16 (10.7%) had fever (2 100.4 F) during pregnancy, 40
(28.8%) had preterm premature rupture of membrane (ROM) for more than 24 hours
prior to onset of labor, and 69 (46%), 124 (82.7%), and 42 (28%) were treated with
antibiotics, antenatal steroids, and magnesium sulfate respectively prior to delivery.
These mothers had an average age of 29.5 years, ranging between 18 and 47 years old.

Except for HELLP syndrome, we found significant differences across three study
sites on all maternal characteristics we studied. This suggests that population or clinical
management difference may exist among three sites. NY tended to have younger
mothers than other sites (p=0.0034). Both black and Hispanic enrollees were mostly
recruited from New York site and the major race group in both Amsterdam and Madrid is
white.

NY also had the highest rates of preeclampsia, fever during pregnancy, and ROM
greater than 24 hours. The large difference in preeclampsia (PE) rates between NY and
Amsterdam in this study, although not signiﬁcant, could have several explanations. First,
the urinary excretion of 300 mg of protein or higher in a 24-hour urine collection is
required for PE diagnosis in Amsterdam, but not in the US. Given the potential difficulty

of obtaining a 24-hour urine sample, this criterion may have restricted women from

74

getting a preeclampsia diagnosis. On the other hand, less restrictive criteria may have led
to an overestimation of preeclampsia in NY. Second, the racial composition is quite
different. A large proportion of NY subjects were blacks and Hispanics, while
Amsterdam had very few blacks and no Hispanics, both of whom are known to have
higher risk of preeclampsia. Third, clinical practice in Amsterdam allows fetal demise,
which possibly could have caused fewer preeclamptic women to carry the fetus to
delivery. Potentially this difference in practice has introduced selection bias in our
sampling and an under-representation of preeclamptic pregnancy.

More than half the women in NY received antibiotics (68%) and magnesium
sulfate (56%) treatment prior to delivery, while only 20-30% of women received
antibiotics in the other two sites and magnesium sulfate treatment was rare in Amsterdam
(0%) and Madrid (5.6%). Both treatments are related to management of premature ROM,
preeclampsia, and preterm labor, therefore differences were partially related to the low
rates of preeclampsia in Amsterdam, and low rates of prolonged premature ROM in
Amsterdam and Madrid. In addition, clinical practice may have contributed to the
difference, for instance, Magnesium sulfate may not be a standard treatment for
preeclampsia and preterm labor in Amsterdam (need to clarify with Amsterdam).

3.5 Neonatal characteristic of the THOP cohort (Table 3.6)

Signiﬁcant birthweight and gestational age differences existed across study sites.
The average birthweight of NY subjects was 150 grams less than that of Amsterdam
subjects. The mean gestational age was 3.5 days younger in NY and Madrid than in
Amsterdam. With bigger babies, Amsterdam also had the highest FGR level among three

sites.

75

The characteristics of the 168 study subjects across three study sites are given in
table 3.6. No gender difference was found in the THOP study across three study sites.
The racial composition differed signiﬁcantly by study site. Madrid had the highest
number of subjects in the whites group and NY had the highest number of subjects in
both black and Hispanic groups. The difference is not surprising as it reﬂects the racial
composition of the three cities where study sites are located.

In both Amsterdam and Madrid, more than 80% of the subjects had Apgar score
at 1 minute (APGARI) higher than 3, but in NY only 65.7% did. The difference in
APGARI across study sites was signiﬁcant, possibly because Amsterdam had bigger or
more mature babies who might perform well at birth. After resuscitation, Apgar score at
5 minute (APGARS) of all subjects in Amsterdam and Madrid went up to 4 and above,
while APGARS of ﬁve subjects in NY remained under 3. The Apgar score difference by
sites was no longer signiﬁcant at ﬁve minutes.

Findings through comparisons among study sites suggest that there are differences
in both population and clinical practice across sites. We will adjust for the effect of study

sites in later analyses.

76

Table 3.1 Reasons for declining or exclusion.

 

Reasons for declining

Number of potential subjects Percentag_e__

 

 

 

Not interested, or afraid of research, or wants 36 15.1%
payment, or too stressed to decide
Religious beliefs 2 0.9%
Reasons for exclusion
Eligible time frame expired (< 24 hr after birth) 3 1.3%
Death expected within 48 hours 10 4.2%
Mother < 18 years old 8 3.3%
Maternal disease during pregnancy 8 3.3%
Maternal substance abuse 1 0.4%
Mother is too ill after delivery 1 0.4%
Iodine exposure prior to delivery 2 0.9%
Total number of enrolled subjects 168 70%
Total number of screened subjects 239 100%
Table 3.2 Characteristics of included and excluded subjects.
Included (n=168) Excluded (n=7l) p-value
N (%) N (%)

Study site

New York 77 (65.6) 42 (35.3)

Amsterdam 52 (77.6) 15 (22.4) 0,1516 ‘

Madrid 39 (73.6) 14 (26.4)
Race/ethnicity

White 107 (71.3) 43 (28.7)

Black 40 (78.4) 11 (21.6) 1

Hispanic 15 (48.4) 16 (51.6) 0°02“

Other 6 (85.7) I (14.3)
Gender

Male 93 (72.1) 36 (27.9) 1

Female 75 (68.2) 35 (31.8) 050%

Mean (Se) Mean (Se)

Birthweight (g) 852.2 (14.7) 853.9 (25.3) 0.9526 2
Gestational age (wk) 26.2 (0.1) 26 (0.1) 0.2708 2

 

1 P-value by Chi-square test;
2 P-value based on t-test.

77

Table 3.3 Characteristics of the THOP cohort.

 

 

N (%)

Study site

New York 77 (45.8)

Amsterdam 52 (31)

Madrid 39 (23.2)
Withdrawal

Prior to 7 days 1 4 (2-4)

After 7 days 2 2 (1-2)

Total 6 (3.6)
Race/ethnicity

White 107 (63.7)

Black 40 (23.8)

Hispanic 15 (8.9)

Other 6 (3.6)
Gender

Male 93 (55.4)

Female 75 (44.6)
Birthweight (g)

Mean (se) 851 (14.7)

Median (range) 827 (440, 1550)
Gestational age (week)

Mean (se) 26.2 (0.1)

Median (range) 26.2 (24, 27.9)
Total 168

 

l The denominator is the number of subjects who did not withdraw within the ﬁrst 7 days of life;

2 The denominator is the number of subjects who survived for the ﬁrst 7 days of life;

Table 3.4 Treatment group assignment of the THOP cohort.

 

Number of

 

 

Treatment Number enrolled Noncompleter ] effective subject
Study groups N (%) N (%)
Placebo 28 (18.3) 1 27 (18.7)
Iodine 26 (17.0) 2 24 (16.7)
Bolus 4 pg 25 (16.3) 1 24 (16.7)
Continuous 4 pg 27 (17.7) 3 24 (16.7)
Bolus 8 11g 20 (13.0) 0 20 (13.9)
Continuous 8 pg 27 (17.7) 2 25 (17.3)
Total 153 9 144
Accidental groups
Bolus 16 pg 7 (46.7) 1 6 (46.2)
Continuous 16 pg 8 (53.3) I 7 (53.8)
Total 15 2 13
Total 168 11 157

 

l Noncompleter refers to subjects who expired or withdrew consent prior to day 7 of life.

78

Table 3.5 Maternal characteristics of the THOP cohort.

 

 

New York Amsterdam Madrid value Total
(n=72) (n=42) (n=36) 9' (n=150)
Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Maternal age (yr) 28 (5.9) 31.6 (4.9) 30.2 (5.9) 0.00341 29.5 (5.8)
N (%) N (%) N (%) N (%)
Maternal race
White 29 (39.7) 32 (76.2) 33 (91.7) <0.00012 94 (62.2)
Black 28 (38.4) 9 (21.4) 1 (2.8) 38 (25.2)
Hispanic 11 (15.1) 0 (0) 2 (5.6) 13(8.6)
Other 5 (6.8) 1 (2.4) 0 (0) 6 (4)
HELLP 4 (5.6) 0 (0) 0 (0) 0.15062 4 (2.7)
Preeclampsia 14 (19.4) 1 (2.4) 5 (14.3) 0.03553 20 (13.4)
Fever ( 2 100.4 F) 2 1 7
during pregnancy 12 (16.7) 4 (9.5) O (0) 0,0303 16( 0. )
Antibiotics 2
administration 49 (68.1) 13 (31) 7 (19.4) <0.0001 69 (46)
Antenatal steroids . 3
administration 61 (84.7) 38 (90.5) 25 (69.4) 0.0409 124 (82.7)
Magnesium 3
administration 40 (55.6) 0 (0) 2 (5.6) <0.000] 42 (28)
“Pm" 0‘ 25 (40.3) 9 (21.9) 6 (16.7) 0.02303 40 (28.8)

membrane 2 24 hr

 

1 P-value based on F-test;
2 P-value by exact analysis (Pearson test);
3 P-value by Chi-square test.

79

 

 

Table 3.6 Neonatal characteristics of the THOP cohort.
New York Amsterdam Madrid Total
(n=77) (n=52) (n=39) P'va'“ (n=168)
N (%) N (%) N (%) N (%)
Gender (male) 35 (45.5) 34 (65.4) 24 (61 .5) 0,05571 93 (55.4)
Race
White 30 (39) 41 (78.9) 36 (92.3) < 0,00012 107 (63.7)
Black 29 (37.7) 10 (19.3) 1 (2.6) 40 (23.8)
Hispanic 13 (16.9) 0 (0%) 2 (5.1) 15 (8.9)
Other 5 (6.5) 1 (1.9%) 0 (0) 6 (3.6)
C-section 48 (63.2) 7 (13.5) 22 (56.4) < 0.00011 77 (46.1)
Multiple birth
Singleton 66 (86.8) 32 (61.5) 32 (82.1) < 0.00012 130 (77.8)
Twins 10 (13.2) 14 (26.9) 7 (17.9) 31 (18.6)
Triplets 0 (0) 6 (11.5) 0 (0) 6 (3.6)
Intrauterine rowth
restriction g 13 (16.9) 3 (5.8) 4 (10.3) 0.1505 20 (11.9)
Apgar score at 1 min
5 3 25 (34.3) 9 (18.8) 6 (16.7) < 0.00011 40 (25.5)
4 - 6 29 (39.7) 12 (25) 24 (66.6) 65 (41.4)
Apgar score at 5 min
5 3 5 (6.7) o (0) o (0) 0.13622 s (3)
4-6 13 (17.6) 9(17.3) 5 (3.2) 27(16.5)
Apgar score change from 1 min to 5 min
S 3 to (4 - 6) 7 (9.6) 7 (4.6) 2 (5.6) 0.00162 16 (10.2)
(4 - 6) to Z 7 23 (31.5) 10 (20.8) 21 (58.3) 54 (34.4)
S3t027 l3(l7.8) 2(4.2) 4(11.1) 19(12.1)
Mean (SD) Mean (SD) Mean (SD) Mean (SD)
Birthweight 796.9 (177.1) 948.4 (202.3) 830.9 (151.4) < 0.00013 852 (190.8)
Gestational age 26 (1.1) 26.5 (0.8) 26 (1.2) 0,03523 26.2 (1 .1)
Fetal growth ratio 0.82 (0.2) 0.93 (0.2) 0.85 (0.1) 0.00013 0.86 (0.2)

 

1 P-value by exact analysis (Pearson test);
2 P-value by Chi-square test;
3 P-value based on F-test.

80

Chapter 4 Correlates of thyroid hormone levels at birth among extremely

premature infants

Transient hypothyroxinemia of prematurity (THOP) is manifest by very low
levels of serum thyroxine (T4) and tri-iodothyronine (T3), but normal thyroid stimulating
hormone (TSH) levels from shortly after birth until up to 6 weeks after birth. The
disorder is seen with increasing frequency in infants born before 32 weeks of gestation,
and especially below 28 weeks (1-5). The major cause of THOP is thought to be the
immaturity of the thyroid gland and the hypothalamic-pituitary-thyroid axis in preterm
infants. Despite variations in study populations and THOP deﬁnitions, studies have
consistently shown clear decreases in THOP as gestational age increases.

Many studies have suggested that hypothyroxinemia during the neonatal period in
premature infants may be associated with neurodevelopmental disabilities (1, 6-9).

Preventing and treating THOP in extremely premature newborns (g 28 wks) has the

potential to decrease the risk of adverse neurodevelopmental outcomes. It seems
reasonable to suppose that premature infants vary in their thyroid hormone levels at birth,
but little is known about the correlates of such variation. In this chapter, we use data
from the Phase 1 Study of Thyroid Hormone in Extremely Premature Infants to
investigate associations between prenatal exposures and newborn thyroid hormone levels
at birth (baseline). The examination of correlates of thyroid hormones can lead to a better

understanding of THOP.

81

4.1 Analytical sample (Figure 4.1)

To reduce the effect of the correlation between infants born to the same mother,
we only included the ﬁrst born of any multiple birth set if more than one infant in this set
was enrolled. Among the 168 enrolled subjects, 131 were singleton and 37 were multiple
births. The multiple births included 13 pairs of twins (the second-born excluded), ﬁve
single twin enrolled from ﬁve different twin pairs (zero excluded), and two births from
two triplet sets (four excluded). After the exclusion of 17 multiple births, seven more
subjects were removed from this analysis because no baseline thyroid hormone levels
were available. The ﬁnal analytical sample thus excluded 24 of the 168 enrolled subjects,
leaving 144 infants born between 24 and 28 completed weeks of gestational age.
Neonatal baseline FT4 levels were available for 144 subjects, TT4 levels were available
for 128 subjects, and TSH levels were available for 135 subjects. Missing data were due
to insufﬁcient sample volume, because of which multiple hormone testing could not be
completed.

4.2 Exposures and outcomes
4.2.1 Maternal exposures

Four types of maternal information were obtained from maternal interview,
medical records abstraction, and laboratory tests: 1) maternal demographic information
(race, age); 2) previous pregnancy history (gravida, parity); 3) current pregnancy and
delivery history (route of delivery, preeclampsia, prolonged premature rupture of
membrane, and antenatal antibiotic/steroids/magnesium use prior to delivery); and 4)

maternal thyroid function tests (F T4, TSH, and TT4).

82

Maternal venous blood (3 ml) was collected at enrollment for thyroid function
tests. Among 144 mothers, blood samples were not obtained for 22 women due to early
discharge. F T4, TT4, and TSH were measured in maternal blood samples. Three women
were missing TT4 and two were missing TSH results. The time of collection varied
from 0-18 days aﬁer delivery. Half of the specimens (52%) were drawn within 24 hours
after delivery, 31% were drawn within 3 days of delivery, and 16% within a week. Test
results for one woman whose blood sample was taken 18 days after delivery were
excluded ﬁom the analysis, because such a sample may not reﬂect thyroid function at
delivery.

4.2.2 Neonatal exposures

Neonatal birth characteristics consisted of exposures with two different attributes:
1) ﬁxed exposures, e.g. birthweight (BW), gestational age (GA), gender, race, fetal
growth ratio (FGR), intrauterine growth restriction (IUGR), multiple birth status, and
Apgar score at 1 minute (APGARI) and 5 minute (APGARS); and 2) changing exposures,
e.g. the change of Apgar score from 1 minute to 5 minute.

F GR was deﬁned as the ratio of birthweight to the median birthweight of infants
of the same gestational week. IUGR was deﬁned as birthweight below 10th percentile of
the same gestational week. Both FGR and IUGR were generated using an external
gestational-age and plurality-speciﬁc birthweight standard, derived by Alexander et al.
from the 1991-1995 US Natality Data (10).

At 1 minute and 5 minutes after birth, neonatal Apgar scores were evaluated for
each subject on a 0 to 10 scale, with 0 being the worst and 10 being the best. We divided

Apgar scores into three groups: 5 3, 4 - 6, and Z 7. To understand how neonates changed

83

from 1 minute to 5 minute, we created a new variable, Apgar score change from 1’ to 5’,
based on Apgar scores at these two time points. Subjects were categorized into four
groups: from S 3 to 4 - 6, from 5 3 to Z 7, from 4 - 6 to 2 7, and no change.

4.2.3 Log transformation of thyroid hormone levels

Thyroid hormone levels of subjects at birth serve as the only outcome variable in
this analysis. We tested the normality of the distribution of both neonatal and maternal
FT4, TT4, and TSH levels. Results suggested that all ﬁve hormones followed the
lognormal distribution. Therefore, a logarithmic transformation (base e) was applied to
all hormone results and the report below is based on log-transformed hormone values.
4.3 Model selection approach

Unadjusted analyses were carried out to study the associations between each
maternal/neonatal exposure and neonatal FT4/TT 4 levels, as continuous variables.
Linear regression was used to test relationships involving continuous exposures. One-
way ANOVA was used to test associations involving categorical exposures. Unadjusted
associations between maternal/neonatal exposures and baseline FT4/IT 4 levels are
reported ﬁrst, before adjusted associations. .

Multivariable model selection involved four major steps. First, we included all
variables that were related to the neonatal baseline thyroid hormone levels in unadjusted
analyses at a signiﬁcance level of 0.1. Second, we eliminated variables with adjusted p-
values less than 0.05 from the full model, one at a time, beginning with the least
signiﬁcant variable. Third, we added each one of the remaining variables, which were
excluded at the ﬁrst step and dropped at the second step, back to the model one at a time

starting with the most signiﬁcant variable in univariable analyses. If any variable showed

84

a signiﬁcant association with the outcomes after adjusting for other variables, it would be
added into the model and the third step would be repeated one more time with the new
model. Finally, when every variable that was signiﬁcantly (01=0.05) related to the
outcomes was included, we increased the signiﬁcance level to 0.1 and included any
variable that was biologically meaningful but not statistically signiﬁcant (01=0.05).
4.4 Results
4.4.1 Neonatal and maternal characteristics (Table 4.1)

Birthweight of the 144 subjects ranged from 440 to 1550 grams with a mean of
855 grams and a standard deviation of 198.3 grams. Gestational ages ranged from 24 to
27.9 completed weeks with a mean of 26.2 weeks and a standard deviation of 1.1 weeks.
Fetal growth ratio varied from 0.43 to 1.33 with a mean of 0.86 and a standard deviation
of 0.2. F GR was deﬁned as the ratio of birthweight to the median birthweight of infants
of the same gestational week and was generated using an external gestational-age and
plurality-speciﬁc birthweight standard, derived by Alexander et al. from the 1991-1995
US Natality Data (10).

55% of the enrolled subjects were male. The majority of this p0pulation was
White (62.5%). 45.8% of subjects were born by C-section, 14% were one of a multiple-
birth set, and 13% had intrauterine growth restriction. About 27% of subjects had Apgar
scores at 1 minute below 4 points. At ﬁve minutes, only 3% remained under 4. Apgar
scores from 1 minute to 5 minutes were improved in 57% of the subjects. The rest
remained in the same Apgar score range (e.g., 0-3, 4-6, 7-10). About 12% of subjects
had the most improvement, changing from less than 4 points at 1 minute to greater than 7

points at 5 minutes.

85

The 144 subjects were born to 144 mothers, whose age ranged from 18 to 47
years with a mean of 29.5 years and a standard deviation of 5.8. 14% of mothers had
preeclampsia and 10% reported fever during the pregnancy. Prior to delivery, 46%, 83%,
and 30% of mothers were treated with antibiotics, antenatal steroids and magnesium
sulfate, respectively. About 30% of women had rupture of membranes for longer than 24
hours (ROM > 24 hours) prior to onset of labor.

4.4.2 Neonatal and maternal hormone levels (Table 4.2 - 4.5)

Neonatal and maternal serum FT4 levels were similar and both averaged around
1.5 ung. The mean maternal TT4 level was close to two times higher than that of
neonates, while the mean maternal TSH level was only half the neonatal level.

We studied hormone levels across the three study sites. Results in NY and
Amsterdam were comparable. Madrid had the lowest neonatal but the highest maternal
thyroid hormone levels, with signiﬁcant differences in neonatal FT4 and maternal FT 4
and TT4. Lower iodine levels in Spain may explain the low neonatal thyroid hormone
levels, but do not explain the higher maternal thyroid hormone levels. Study site will be
included in the multivariable analyses to account for site differences.

Both neonatal and maternal thyroid hormone levels were right-skewed and
followed Iognonnal distribution. We reported thyroid hormone using the original scales
in Table 4.2-4.5, and log transformed hormone levels were used in linear regression

analyses (Table 4.6-4.9).

86

4.4.3 Unadjusted association between neonatal exposures and neonatal baseline

thyroid hormone levels (Table 4.6)

Baseline FT4: Gestational age was signiﬁcantly correlated with neonatal baseline
FT4 levels (p< 0.0001), but birthweight was not. F GR was inversely associated with F T4
(p=0.01 17). Subjects enrolled in the NY site had the highest FT4 values and subjects
from Madrid the lowest. Female subjects tended to have higher baseline FT4 levels than
males. Black infants had the highest baseline hormone levels, while Hispanics had the
lowest (p=0.032). Multiple birth status also was related to neonatal baseline FT4 levels,
but at the border of signiﬁcance (p=0.0649). In subgroup analysis, singleton births had
signiﬁcantly higher baseline thyroid hormone levels than did twin births (p=0.0432).
APGARI was directly, but not signiﬁcantly associated with FT4 (p=0.0804), but
APGARS was not. With respect to Apgar score change from 1 to 5 minutes, stable
subjects whose Apgar scores remained in the same category had the highest FT4 levels
(p=0.0810). The lowest FT4 levels were seen in subjects who had the most dramatic

Apgar score improvement, i.e., from —_<_ 3 to 2

Baseline TT4: Compared to neonatal baseline FT4, fewer variables were
correlated with baseline TT4 levels. Birthweight and gestational age were both
signiﬁcantly related to TT4 levels at birth. APGARI (p=0.0098) and Apgar score change
from 1 minute to 5 minute (p=0.0445) were signiﬁcantly related to TT4 levels in the
same fashion as to FT4 levels. Interestingly, IUGR, a widely accepted risk factor for low

thyroid hormone levels in neonates (5, 1 1-13), was not related to either FT4 or TT4 levels.

87

4.4.4 Unadjusted association between maternal exposures and neonatal baseline

thyroid hormone levels (Table 4.7)

Log-transformed maternal TT4 levels were inversely correlated with neonatal
baseline F T4 levels (p=0.0218). No other maternal exposures were related to neonatal
baseline FT4 levels at signiﬁcance level of 0.1.

However, several maternal exposures were signiﬁcantly correlated with neonatal
baseline TT4 levels in unadjusted analyses. Infants born to women who received
antibiotics (p=0.0648), antenatal steroids (p=0.0112), and magnesium sulfate (p=0.0809)
treatment prior to delivery had higher baseline TT4 levels. Women who had ROM > 24
hours (p=0.0005) prior to onset of labor also had higher TT4 levels at birth.

4.4.5 Correlates of neonatal FT4 level at birth (Table 4.8, Figure 4.2-4.3)

Previously, we found that subjects enrolled in the three centers had different
attributes reﬂecting variations in population and clinical management across centers. We
therefore studied the association between study sites and neonatal baseline thyroid
hormone levels. No signiﬁcant results were found. We also adjusted for study centers in
the ﬁnal linear regression model and center was insigniﬁcant (p>0.1). As study center
has no effect on the outcomes of interest, we did not include center in the ﬁnal model.

Gestational age, fetal growth ratio, race, multiple birth status, and intrauterine
growth restriction were signiﬁcantly correlated with neonatal baseline FT4 levels.
Newborn baseline F T4 levels rose as gestational age increased. Every one week increase
in gestational age increased FT4 by 17% (p < 0.0001). F GR was found inversely
correlated with baseline F T4 levels. Every 10% increase in fetal growth ratio resulted in

a 6% reduction in baseline FT4 levels. IUGR was not signiﬁcantly related to neonatal

88

F T4 level in the unadjusted model, but became marginally signiﬁcant after adjusting for
gestational age and other factors. Subjects with growth restriction had an 18% reduction
in FT4 level compared to subjects without.

On average, FT4 levels of Blacks were 14% higher than Whites (p=0.0570) and
30% higher than Hispanics (p=0.0016, data not shown). Subjects born as one of a
multiple-birth set had on average a 14% reduction in FT4 levels compared to singleton
births (p=0.0642).

It is interesting to see that although both F GR and IUGR represent fetal growth,
they were related with FT4 levels in different directions. To further understand why FGR
was negatively, while IUGR was positively, correlated with FT4 levels, we plotted log-
transforrned FT4 levels over F GR values by IUGR status. Figure 4.2-4.3 suggests that
for subjects who had IUGR, there was no linear relationship between FGR and neonatal
baseline FT4 levels (regression p=0.7684, r2=0.0056); for subjects who did not have
IUGR, the higher the F GR, or birthweight, the lower the FT4 levels (regression p=0.0034,
r2=0.0069).

4.4.6 Correlates of neonatal TT4 level at birth (Table 4.9)

As with baseline FT4 levels, gestational age remained directly signiﬁcantly
correlated with neonatal TT4 levels at birth adjusted for other factors. For every one
week increase in gestational age, TT4 increased about 16%.

Neonatal baseline TT4 levels were associated with antenatal steroid use,
magnesium use prior to delivery, and ROM > 24 hours prior to onset of labor with
adjusting for gestational age. Infants born to women with prolonged ROM had a 32%

increase in TT4 levels (p < 0.0001). Infants who received antenatal steroids treatment

89

had a 22% increase in TT4 levels (p=0.0272). Infants born to women who received
magnesium sulfate prior to delivery had an average increase of 14% in TT4 levels
(p=0.0266). Although both antenatal steroid use (p=0.006) and rupture of membrane for
greater than 24 hours (p=0.034) are signiﬁcantly correlated with magnesium sulfate use
prior to delivery, these three factors remain signiﬁcant in the model after adjusting for
each other, which suggests that these three factors were all independently related to
neonatal baseline TT4 levels.
4.5 Discussion

In the literature, correlates of neonatal F T4 at birth have seldom been studied,
mainly because most studies used thyroid hormone levels obtained from various local
newborn screening programs, which normally do not test F T4 values. In our study, we
had the opportunity to study both neonatal FT4 and TT4 levels at birth. Our results
suggest that neonatal baseline FT4 and TT4 levels were associated with distinct groups of
factors: 1) F T4 levels were associated with ﬁxed attributes, such as fetal growth ratio,
multiple birth status, race, and IUGR; 2) TT4 levels were associated with pregnancy
conditions, such as antenatal steroids/magnesium use and ROM > 24 hours prior to onset
of labor.

Gestational age was positively associated with both baseline FT4 and TT4 levels
adjusting for other factors. This ﬁnding is consistent with previous reports that have

shown this association not only in serum samples at birth, but also in cord blood (13-18).

90

4.5.1 Correlates of neonatal baseline FT4 in premature infants between 24-28
weeks of gestational age (Figure 4.4-4.5)

An interesting discovery is the inverse relationship between fetal growth ratio and
neonatal baseline FT4 levels. FGR is an index representing intrauterine growth by
comparing birthweight with the gestational week speciﬁc median birthweight. A slight
decreasing trend in FT4 levels over FGR is shown in ﬁgure 4.4. Infants with lower
F GR’s have higher FT4 levels, and vice versa.

More interestingly, both F GR and IUGR represent fetal growth, one would
assume these two variables should relate to baseline FT4 levels similarly. However, in
this analysis, presence of IUGR, deﬁned as the lowest 10th percentile of birthweight for
gestational age, had an opposite effect. In them, FT4 levels were lower rather than higher.

To further explore the opposite association between F GR and IUGR in relation to
baseline FT4 levels, we plotted log FT4 levels over FGR within IUGR stratum (ﬁgure
4.2-4.3) and found a signiﬁcant inverse relationship between log FT4 levels and F GR
only among infants who did not have IUGR (p=0.0034). Figure 4.5 (mean of log-
transformed F T4 by every 10th percentile of FGR) revealed a quadratic shape between
FT4 levels and FGR percentiles. FT4 levels were low in the lowest 10th percentile,
gradually increased in the second 10th percentile, then decreased until the highest
percentile. This illustrates the opposite direction of effect of FGR on FT4 within IUGR
and within non-IUGR babies. In effect, infants with the lowest and highest FGR had low
FT4 levels, while infants in the middle had higher FT4 levels.

Few studies have examined race and multiple birth status in relation to thyroid

hormone levels at birth. In our study, race (p=0.0167) and multiple birth status

91

(p=0.0642, marginally signiﬁcant) were both correlated with baseline FT4 levels after
adjusting for gestational age and other factors. Black infants had the highest F T4 levels
at birth compared to Whites (p=0.0570) and Hispanics (p=0.0016, not shown in Table).
Singleton births had higher FT4 levels than multiple births. The mean FT4 level
difference between singletons and twin births was especially signiﬁcant (p=0.0489, not
shown in Table). To our best knowledge, no one has ever reported any signiﬁcant
associations between race/ethnicity and multiple birth status in relation to baseline
thyroid hormone levels. We do not know whether our ﬁndings are a true reﬂection of
biological relationships or were due to chance because we do not have a large sample in
testing F T4 level differences across race/ethnicity group (power = 71%) or multiple birth
status (power = 56%). Future research with a larger sample size stratiﬁed by social
demographic factors and race proﬁle /multiple birth status is needed.
4.5.2 Correlates of neonatal baseline TT4 in premature infants between 24-28

weeks of gestational age

Martin et al. reported that newborns with a complete dose of antenatal steroid
treatment had a mean increase of 0.81 11ng in serum TT4 levels (19). Our results
support this ﬁnding. Infants who received antenatal steroid treatment prior to delivery
had a 22% increase in neonatal TT4 levels. Both results, although not comparable,
agreed on the directionality of the association between antenatal steroids and newborn
baseline TT4 levels.

We also found that ROM > 24 hours prior to onset of labor was related to higher
(32%) TT4 levels. Magnesium use prior to delivery, although marginally signiﬁcant, was

also associated with higher TT4 levels. Although a signiﬁcant correlation existed

92

between ROM > 24 hours and magnesium use, both factors remained signiﬁcant after
adjusting for each other and additional variables, which suggests that their effects on
neonatal baseline TT4 levels were independent.

Other factors, such as gender, route of delivery, Apgar score at 1 minute, and
antenatal antibiotic use, have been reported in the literature to be associated with neonatal
FT4 and TT4 levels. In our unadjusted analyses, we found F T4 levels, at the signiﬁcance
level of 0. 1, were associated with gender and Apgar score at 1 minute, and TT4 levels
were associated with Apgar score at 1 minute and antibiotic use prior to delivery. All
these associations disappeared after adjusting for gestational age and other factors.

4.5.3 Relationship between neonatal baseline thyroid hormone levels and maternal

thyroid hormone levels (Figure 4.6 - 4.8)

Hume et al. reported an overall lack of association between neonatal cord and
maternal thyroid hormone levels (16). We looked at the relationship between neonatal
baseline and maternal thyroid hormone levels. Maternal TT4 and neonatal FT 4 levels
were correlated, but the signiﬁcance disappeared after adjusting for other factors. No
associations in FT4 or TT4 levels were observed between maternal and neonatal serum
samples within different gestational week groups (ﬁgure 4.6-4.8).

Murphy et al. found a strong relationship between maternal and neonatal TSH
levels (18). Signiﬁcant associations between maternal and neonatal TSH within each

gestational week group were also seen in our study (ﬁgure 4.8).

93

4.6 Strength and limitation (Table 4.10)

Our study collected information on 168 neonates born between 24 and 28
completed weeks of gestational age and is the largest study for this unique premature
population. We also are the ﬁrst group to study a broad range of prenatal exposures and
baseline thyroid hormone levels at birth.

Unlike most previous studies, we measured serum FT4 levels by the direct
equilibrium dialysis (DED) method. Studies have shown that analog immunoassay or
FT4 index methods, commonly used by earlier studies, could be biased by the
concentrations of serum proteins and protein-bound T4 (3, 20). The DED method can
maximally remove bias from serum protein levels, thus provides us the most accurate
FT4 levels.

Baseline F T4 levels were measured between birth and 24 hours of life. One may
question whether the time of collection matters in this population. The literature has
shown that in term infants, thyroid hormone surges soon after birth. TSH surges within
half an hour after birth and decreases soon after its maximum level. Serum T4 and T3
levels increase quickly following TSH surge and peak at 24 hours of life. Various time
points of sample collection may result in capturing FT4 and TT4 levels in different stages
of the increasing phase, which may not reﬂect the baseline thyroid hormone levels of the
study subjects.

Several studies have shown that among extremely premature infants, the
hypothalamic-pituitary-thyroid system may not respond to the extra-uterine environment
in the same way or at the same level of magnitude as in the term or moderately preterm

infant. Hormone surges were found to be attenuated, and some even disappeared, due to

94

immaturity of the thyroid gland and the hypothalamic-pituitary-thyroid system. To test
whether time of sample collection affected our results, we calculated the mean and
standard deviation of neonatal FT4 and TT4 levels on day 0 for all study participants and
on other days among the subjects who did not receive thyroid hormone treatment (Table
4.10). We found that FT4 levels hardly changed between day 0 and day 3, which
suggested that time of collection on day 0 would not affect the associations we observed.
TT4 levels began to decline for about 10-20% by day 3, which implied that most of the
relationships we found in the current analyses would be at most mildly affected by time
of collection.
4.7 Conclusion

In conclusion, we found that except for gestational age, different groups of factors
were associated with baseline F T4 and TT4 levels among premature infants born between
24-28 weeks of gestational age. Baseline FT4 levels were associated with ﬁxed attributes,
such as race, fetal growth ratio, intrauterine growth restriction, and multiple birth status;
while baseline TT4 levels were associated with conditions such as antenatal steroids and
magnesium treatment, and ROM > 24 hours prior to onset of labor.

Neonatal baseline F T4/1T 4 levels were not related to maternal F T4/TT 4 levels,

however, TSH levels of neonates and mothers were positively correlated with each other.

95

Figure 4.1 Flow chart of analytical sample selection.

168 subjects enrolled at birth

 

 

F \
131 13 twin pairs+ 2 triplet
singletons 5 single-twins=31 sets=6
1-0 1-13 1-4
131 18 single 2 single
singletons twin births triplet births

1-7 1-0 1-0

124 singletons 18 single twins 2 single triplets

with hormone with hormone with hormone
results results results
\ J

 

 

Analytical sample size = 144

96

Figure 4.2 Correlation between log-transfonned FT4 and FGR in infants with IUGR.

1.504
1.251
1.004
075*

0.504 ’ '

LogFT4

0.25 ' '

0.001 0

-0.25*

-0.50‘

 

-0.75‘,f..e....,...... .
0.50 0.75 1.00 1.25 1.50

 

97

Figure 4.3 Correlation between log-transformed FT4 and FGR in infants without IUGR.

 

 

1.50‘
1.251 ' .
C .
1.001 . c o .
. C
o
o o
o o
0.75‘ o o
o o o
o o
. .. . .
'3- 0'50‘ o o u o o .0 o . .
LI. 0 a o one o o c
U) c c o a o co 0
3 0.25. o o u o -0 a o
c o o o o a
can a o o o o a o o
0.00‘ o o. to. e o o o o
o a on o
-0.‘25 o o o I o
. . . .
-0.50* o o
0751. ......... . ......... . ......... .s - . .
050 075 1.00 125 1.50
FGR

98

Figure 4.4 Log-transformed FT4 distribution over FGR.

 

 

 

 

1.50j
1.25‘ ° 0
. .
1.001 o o .
on c
c
o a a
o o
0.751 o o
c o e c
o o o
_ o no a o
3 ““““
E 050‘ :-~"-L-_- u c a .0 c . .
LL. 0 I o e“‘--o.p_o. o o c
U ‘ c I o o o o --‘t-~-.__ co 0
3 0.25. o e o e ca 0 u o u-"I .......
o I o I o ; _________
cone 0 o o o o o o -.o .........
(1004 o o 00 a- o o o e 0 ~
0 e c o o
-0.254 . . . I .
. . C . . .
-0.50‘ o o
. o
075*, ......... . ......... . ......... , .........
0.50 0.75 1.00 1.25 1.50
FGR
R—square=0.0439

Adjusted R-square=0.0371

99

Figure 4.5 Mean FT4 (log-transformed) by F GR 10th percentile.

1.3‘

 

..l \

Log FT4
C)
- an

 

 

 

 

-0'5 ‘1 1 T 1 T j r 1 1 T
10m 2001 30th 40th 50th 6001 70th _ 80th 90m 100m
Fetal Growth Ratio Percentile

100

Figure 4.6 Correlation between maternal and neonatal FT4 (log-transfonned) by
gestational week among 144 subjects.

Neonatal IogFT4

Neonatal IogFT4

1.51

1.0‘

0.51

0.0‘

 

Gestational age [24. 25)

 

1.51

1.0‘

0.51

0.01

-0.51

 

-1.0

1

-15 -1.0 -05 0.0 0.5 1.0 1.5 2.0 2.5
Maternal IogFT4

Gestational age [25. 26)

j T I

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
Maternal logFT 4

101

Figure 4.6 Correlation between maternal and neonatal FT4 (log-transfomted) by
gestational week among 144 subjects. (cont’d)

Neonatal IogFT4

Neonatal IogFT4

1.01

0.5-

0.0“

 

Gestational age [26. 27)

 

1.5“

1.01

0.51

0.01

 

T T T

-15 -1.0 .05 0.0 0.5 1.0 1.5 2.0 2.5
Maternal IogFT4

Gestational age [27, 28)

T T l I W

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
Maternal logFT 4

102

Figure 4.6 Correlation between maternal and neonatal FT4 (log-transformed) by
gestational week among 144 subjects. (cont’d)

 

 

GA (week) I} coefﬁcient P-value
24-25 -0.1365 0.5321
25-26 0.0528 0.7212
26-27 -0.01 18 0.9025
27-28 0.1698 0.3032

 

103

Figure 4.7 Correlation between maternal and neonatal TT4 (log-transformed) by
gestational week among 144 subjects.

3.03

2.51

2.01

1.5‘

Neonatal logTT4

1.01

 

0.5-
0.5

3.0'

2.51

.N
0

Neonatal logTT4
2?

1.01

 

0.5‘.
0.5

Gestational age [24, 25)

1.0 1.5 2.0 2.5 3.0
Maternal logTT4

Gestational age [25. 26)

1.0 1.5 2.0 2.5 3.0
Maternal logTT4

104

Figure 4.7 Correlation between maternal and neonatal TT4 (log-transformed) by
gestational week among 144 subjects. (cont’d)

Neonatal logTT4

Neonatal logTT4

Gestational age [26. 27)

 

 

 

 

3.0
2.5 o
O 0 ~.
2.04 .0 . .
1.5 I o .0 .
1.04
0.51. . . . . .
0.5 1.0 1.5 2.0 2.5 3.0
Maternal logTT4
Gestational age [27. 28)
3.04
2.51 ° 0
2.0 . I '
15 °'
1.04
0.51. . . r r .
0.5 1.0 1.5 2.0 2.5 3.0

Maternal logTT4

105

Figure 4.7 Correlation between maternal and neonatal TT4 (log-transfonned) by
gestational week among 144 subjects. (cont’d)

 

 

GA (week) 6 coefficient P-value
24-25 -0.3051 0.0697
25—26 -0.1601 0.4757
26—27 0.2113 0.0538
27-28 -0.0029 0.9817

 

106

Figure 4.8 Correlation between maternal and neonatal TSH (log-transfonned) by
gestational week among 144 subjects.

Neonatal logTSH

Neonatal logTSH

Gestational age [24. 25)
3.5‘
3.01
2.51
2.01
1.51 ' a.
1.0« ' - ' - -
0.51
0.01
-O.51 .
-1.01
-151
-2.0*
-2.51

-3.0, . . . . . T T T . . r . .
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Maternal logTSH

 

Gestational age [25, 26)
3.5‘
3.0‘
2.53 .
2.01 '-
1.5‘ '
1.0« ‘ ° °
0.51 '
0.01 °
-0.5<
_1_0.
-1.51
-2.0'
-2.51

‘3.0" r 1 r r r 1 r r r 1 ﬂ 1 1
-3.0 -2.5 -2.0 -1.5 —1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Maternal logTSH

 

107

Figure 4.8 Correlation between maternal and neonatal TSH (log-transformed) by
gestational week among 144 subjects. (cont’d)

Neonatal logTSH

Neonatal logTSH

3.5
3.01
2.5-
2.0-
1.5
1.03
0.5-
0.0-

-05

-103

-15

-20

-2.5«

 

-3.0

Gestational age [26. 27)

 

ﬁ

-30 -25 -20 -15 -10 -05 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

3.54
3.0‘
2.5‘
2.0‘
1.5‘
1.0‘
0.5‘
0.0‘
-0.51
~1.0’
-1.51
-2.0'
-2.5*
-3.0‘r

 

Maternal logTSH

Gestational age [27, 28)

 

-3.0

-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

Maternal logTSH

108

Figure 4.8 Correlation between maternal and neonatal TSH (log-transformed) by
gestational week among 144 subjects. (cont’d)

 

 

GA (week) 13 coefﬁcient P—value‘
24-25 0.2591 0.3767
25-26 -0.1410 0.8089
26—27 0.0551 0.7522
27-28 -0.0670 0.8232

 

109

Table 4.1 Neonatal and maternal characteristics of enrolled subjects.

 

 

 

 

 

 

 

 

 

Neonatal characteristics N=144
Birthweight (g) Mean (SD) 855 (198.3)
Median (Range) 830 (440, 1550)
Gestational age (wk) Mean (SD). 26.2 (1.1)
Median (Range) 26.3 (24, 27.9)
Fetal growth ratio Mean (SD) 0.86 (0.2)
Median (Range) 0.87 (0.43, 1.33)
N (%)
Gender Male 79 (54.9)
Race White 90 (62.5)
Black 35 (24.3)
Hispanic 13 (9)
Multiple birth (twins and triplets) 20 (13.9)
C-section 66 (45.8)
Intrauterine growth restriction (IUGR) 18 (12.5)
Apgar score at l min 5 3 36 (26.5)
4 - 6 56 (41.2)
7-10 44 (32.3)
Apgar score at 5 min S 3 4 (2.8)
4 - 6 26 (18.3)
7-10 112 (78.9)
Apgar score change from 1 min to 5 min S 3 to (4 - 6) 16 (11.8)
(4 - 6) to 2 7 46 (33.8)
S3toZ7 16(11.8)
Maternal characteristics =144
Maternal age (yr) Mean (SD) 29.5 (5.8)
Median (range) 29 (18, 47)
N (%)
HELLP syndrome 4 (2.8)
Preeclampsia 20 (14)
Fever during pregnancy 15 (10.4)
Antibiotics administration 66 (45.8)
Antenatal steroids administration 119 (82.6)
Magnesium administration 42 (29.2)
ROM > 24 hr 40 (30.1)
Table 4.2 Neonatal and maternal thyroid hormone levels at birth.
Hormone N Mean (SD) Median Range u, Pgrcentilteh u,
10 25 75 90
Neonatal
FT4 (pg/d1) 144 1.5 (0.6) 1.3 (0.5, 3.5) 0.8 1.1 1.7 2.4
TT4 (ng/dl) 128 6.8 (2.7) 6.5 (2.5, 17.4) 3.9 4.7 8.2 10.6
TSH (mlU/ml) 135 5.2 (3.8) 4.3 (0, 22.8) 1.5 2.7 6.7 10.6
Maternal
FT4 (pg/d1) 122 1.4 (0.8) 1.3 (0.2, 8.5) 0.8 1.0 1.7 2.1
TT4 (ng/dl) 119 11.5 (2.5) 11.6 (2.3, 17.6) 8.1 10.1 13.4 14.4
TSH (mIU/ml) 120 2.3 (1.6) 1.9 (0.1, 7.4) 0.7 1.1 3.1 4.9

 

110

Table 4.3 Neonatal thyroid hormone levels at birth by study site.

 

 

 

H /S't N Mean M d' R Percentile l
ormone le (SD) e tan ange 10th 25th 75th 90th p va ue
FT4 (ug/dl)
New York 70 1.6 (0.7) 1.4 (0.5, 3.5) ~ 0.8 1.0 2.0 2.6
Amsterdam 40 1.5 (0.5) 1.3 (0.8, 2.8) 1.0 1.1 1.8 2.2 0.0206
Madrid 34 1.2 (0.3) 1.3 (0.6, 1.9) 0.8 1.0 ' 1.4 1.6
TT4 (ng/dl)
New York 69 7.1 (3.1) 6.6 (2.5, 17.4) 3.9 4.6 8.3 12.3
Amsterdam 34 6.7 (2.2) 6.5 (3.4, 12.0) 4.2 4.8 8.4 10.1 0.2804
Madrid 25 6.0 (2.1) 6.0 (2.6, 10.6) 3.3 4.4 7.4 8.7
TSH (mIU/ml)
New York 67 5.6 (3.9) 4.8 (0.2, 17.5) 1.4 2.8 7.2 11.8
Amsterdam 38 4.9 (4.1) 3.7 (0.8, 22.8) 1.8 2.3 5.8 10.6 0.3711
Madrid 30 4.5 (2.9) 4.3 (0, 14.5) 1.3 2.5 5.8 7.6

 

Table 4.4 Maternal thyroid hormone levels bystudy site.

 

Percentile

 

 

Hormone/Site N Mean (Se) Median Range 10th 25th 75". 9011. p value
F T4 (pg/d1)
New York 63 1.2 (0.5) 1.2 (0.2, 2.3) 0.8 0.9 1.5 2.0
Amsterdam 32 1.5 (1.3) 1.3 (0.2, 8.5) 0.8 1.1 1.6 2.0 0.0043
Madrid 27 1.9 (0.5) 1.9 (0.9, 2.9) 1.0 1.5 2.1 2.7
TT4 (ng/dl)
New York 61 11.1 (2.5) 11.0 (2.3, 15.5) 8.1 9.8 12.9 14.0
Amsterdam 33 11.2 (2.5) 11.5 (5.7, 15.2) 7.1 10.2 12.6 14.1 0.0026
Madrid 25 13.0 (2.0) 12.7 (9.1, 17.6) 10.6 11.5 14.1 15.5
TSH (mIU/ml)
New York 61 2.2 (1.6) 1.8 (0.1, 6.6) 0.6 1.0 2.5 4.4
Amsterdam 33 2.3 (1.4) 2.2 (0.2, 5.9) 0.8 1.1 3.2 4.0 0.4977
Madrid 26 2.6 (1.9) 1.9 (0.2, 7.4) 1.1 1.3 3.4 5.6

 

Table 4.5 P-values of pair-wise comparison between sites on thyroid hormone levels.

 

 

Comparison Neonatal Maternal
FT4 TT4 TSH FT4 TT4 TSH
NY vs. Amsterdam 0.4342 0.5794 0.3348 0.1245 0.8079 0.6039
NY vs. Madrid 0.0055 0.1 120 0.1978 0.0011 0.0009 0.2421

Amsterdam vs. Madrid 0.0651 0.3309 0.7199 0.1006 0.0049 0.5355

111

Table 4.6 Unadjusted associations between neonatal exposures and baseline serum Free
T4 and Total T4 levels (log-transformed).

 

 

Log FT4 (n=144) P value Log TT4 (n=128) P value
Fixed attributes
13 (se) P value _ [3 (se) P value
Birthweight (100 g) 0.01 (0.02) 0.4156 0.04 (0.02) 0.0237
Gestational age (wk) 0.14 (0.03) < 0.0001 0.12 (0.03) 0.0001
Fetal growth ratio -0.51 (0.2) 0.0117 -0.08 (0.2) 0.7206
mean (se) mean (se)
Study site
New York 0.36 (0.05) 0.0555 1.87 (0.05) 0.3240
Amsterdam 0.35 (0.05) 1.86 (0.06)
Madrid 0.17 (0.05) 1.74 (0.08)
Gender
Male 0.26 (0.04) 0.0783 1.79 (0.05) 0.1527
Female 0.37 (0.05) 1.89 (0.05)
Race
White 0.29 (0.03) 0.0320 1.77 (0.04) 0.1042
Black 0.44 (0.07) 1.92 (0.07)
Hispanic 0.08 (0.14) 1.93 (0.14)
Other 0.26 (0.27) 2.04 (0.17)
Multiple birth
Singleton 0.33 (0.04) 0.0649 1.86 (0.04) 0.2372
Twins or triplets 0.16 (0.07) 1.74 (0.08)
Route of delivery
Cesarean section 0.36 (0.04) 0.1677 1.83 (0.05) 0.8084
Vaginal delivery 0.27 (0.05) 1.85 (0.05)
Intrauterine growth restriction '
Yes 0.35 (0.11) 0.6461 1.84 (0.09) 0.9477
No 0.30 (0.03) 1.84 (0.04)
Changing attributes
Apgar score at 1’
S 3 0.21 (0.06) 0.0804 1.71 (0.05) 0.0098
4 - 6 0.30 (0.05) 1.82 (0.06)
2 7 0.41 (0.06) 1.98 (0.06)
Apgar score at 5’
S 3 0.36 (0.17) 0.9418 1.82 (0.10) 0.2824
4 - 6 0.29 (0.09) 1.74 (0.06)
2 7 0.32 (0.03) 1.87 (0.04)
Apgar score change from 1 min to 5 min
5 3 to 4-6 0.22 (0.1 1) 0.0810 1.68 (0.08) 0.0445
5 3 to Z 7 0.17 (0.09) 1.70 (0.08)
4-6 to Z 7 0.28 (0.05) 1.81 (0.07)
No change 0.41 (0.06) 1.94 (0.05)

 

112

Table 4.7 Unadjusted associations between maternal exposures and baseline serum Free
T4 and Total T4 levels (log-transformed).

 

 

LogiT4 (n=144) P value Log TT4 (n=128) P value
mean (se) mean (se)

Primigravida

Yes 0.36 (0.06) 1.89 (0.06)

No 0.29 (0.04) 0.3216 1.81 (0.04) 0.2893
Preeclampsia

Yes 0.42 (0.08) 1.91 (0.60)

No 0.29 (0.04) 0.1797 1.83 (0.04) 0.4170
Fever

Yes 0.36 (0.11) 1.99 (0.11)

No 0.30 (0.03) 0.6037 1.82 (0.04) 0.1160
Antibiotics use during pregnancy

Yes 0.29 (0.05) 1.90 (0.05)

No 0.33 (0.04) 0.5038 1.78 (0.04) 0.0648
Antenatal steroids use

Yes 0.31 (0.03) 0.8763 1.88 (0.04) 0.0112

No 0.30 (0.09) 1.64 (0.08)
Magnesium use

Yes 0.33 (0.07) 1.93 (0.06)

No 0.30 (0.04) 0.7579 1.80 (0.04) 0.0809
Rupture of membrane 2 24 hr

Yes 0.32 (0.07) 2.03 (0.07)

No 0.29 (0.04) 0.6812 1.76 (0.04) 0.0005

13 (86) B (86)

Maternal age (yr) 2e-3 (6e-3) 0.7259 5e-3(6e-3) 0.4197
Maternal gravida 0.01 (0.01) 0.4784 -5e-3(0.02) 0.7472
Maternal log Free T4 -2.4e-4 (0.07) 0.9974 0.01 (0.07) 0.9036
Maternal log Total T4 -0.31 (0.13) 0.0207 0.11 (0.13) 0.4065
Maternal log T SH -0.02 (0.05) 0.5832 0.06 (0.05) 0.1827

 

113

Table 4.8 Adjusted association between maternal/neonatal exposures and baseline
serum Free T4 levels (log-transformed).
Estimate % of 2 Overall

 

 

(n=144)l change SE t Value p-value p-value3

Intercept -3.57 - 0.73 -4.-9l <.0001 <.0001
Gestational age (wk) 0.16 17% 1‘ 0.03 5.89 <.0001 <.0001
Fetal growth ratio -0.063 6% 1 0.023 -2.76 0.0067 0.0067
Race/ethnicity

Black 4 0.13 14% 1 0.07 1.92 0.0570

Hispanic -0.23 20% l 0.10 -2.28 0.0240 0.0167

Other 0.04 4% T 0.14 0.27 0.7851

White ref - - - -
Multiple birth

Multlple birth -0.15 14% 1 0.08 -1.87 0.0642 0.0642

Singleton ref - - - -
Intrauterine growth restriction

Yes -0.20 18% l 0.11 -1.88 0.0627 0.0627

No ref - - - -

 

1 Adjusted for everything else in the table.

2 P-value by Wald-test.

3 P-value by F ~test.

4 The % of change in comparison of Black vs. Hispanic is 30% (p=0.0016).

Table 4.9 Adjusted association between maternal/neonatal exposures and baseline
serum total T4 levels (log-transformed).
Estimate % of 2 Overall

 

SE t V l -
(n=144)l change a "e p "'“e p-value3
Intercept -2. l 7 - 0.77 -2.81 0.0058 0.0058
Gestational age (wk) 0.15 16% T 0.03 5.16 <.0001 <.0001
Antenatal steroids
Yes 0.20 22% T 0.09 2.24 0.0272 0.0272
No ref - - - -
ROM > 24 hours
0
Yes 0.28 32 A. T 0.07 4.22 <.0001 <.0001
No ref - - - -
Magnesium sulfate
0
Yes 0.13 14 /o 1‘ 0.07 1.88 0.0625 0.0625
No ref - - - -

 

1 Adjusted for everything else in the table.
2 P-value by Wald-test.
3 P-value by F-test.

114

Table 4.10 Possible variations of FT4 and TT4 introduced by various time of collection.

 

 

 

FT4 TT4

Postnatalday N Mean (SD) N Mean (SD)

Complete cohort 0 160 1.45 (0.59) 144 6.63 (2.72)
0 50 1.51 (0.64) 47 6.42 (2.80)

3 51 1.64 (0.96) 41 5.32 (2.43)

7 51 1.84 (1.34) 45 4.44 (1.90)

Placebo + Iodine group 14 49 1.52 (0.81) 44 4.45 (2.16)
21 48 1.50 (0.77) 42 5.29 (2.42)

42 48 1.78 (0.96) 46 6.98 (2.35)

56 43 1.64 (0.96) 39 7.26 (2.13)

0 25 1.45 (0.5) 25 6.52 (2.591

3 27 1.35 (0.44) 22 5.12 (2.21)

7 27 1.70 (1.20) 23 4.77 (2.00)

Placebo group only 14 27 1.46 (0.99) 26 4.32 (1.97)
21 24 1.48 (0.73) 21 5.44 (2.34)

42 26 1.43 (0.88) 25 6.49 (1.56)

56 25 1.47 (0.59) 22 6.89 (2.27)

 

115

10.

11.

BIBLIOGRAPHY

Den Ouden AL, Kok JH, Verkerk PH, Brand R, Verloove-Vanhorick SP. The
relation between neonatal thyroxine levels andneurodevelopmental outcome at
age 5 and 9 years in a national cohort of very preterm and/or very low birth
weight infants. Pediatr Res 1996;39:142-5.

Hadeed AJ, Asay LD, Klein AH, Fisher DA. Signiﬁcance of transient postnatal
hypothyroxinemia in premature infants with and without respiratory distress
syndrome. Pediatrics 1981;68:494-8.

Rabin CW, Hopper AO, Job L, et al. Incidence of low free T4 values in premature
infants as determined by direct equilibrium dialysis. J Perinatol 2004;24:640-4.

Reuss ML, Paneth N, Lorenz JM, Susser M. Correlates of low thyroxine values at
newborn screening among infants born before 32 weeks gestation. Early Hum
Dev 1997;47:223-33.

Romagnoli C, Curro V, Luciano R, et al. Serial blood T4 and TSH determinations
in low birth weight infants. Inﬂuence of gestational age, birth weight and neonatal
pathology on thyroid function. Helv Paediatr Acta 1982;37:331-44.

Lucas A, Morley R, Fewtrell MS. Low triiodothyronine concentration in preterm
infants and subsequent intelligence quotient (IQ) at 8 year follow up. ij
1996;312:1132-3; discussion 1133-4.

Lucas A, Rennie J, Baker BA, Morley R. Low plasma triiodothyronine
concentrations and outcome in preterm infants. Arch Dis Child 1988;63:1201-6.

Meijer WJ, Verloove-Vanhorick SP, Brand R, van den Brande JL. Transient
hypothyroxinaemia associated with developmental delay in very preterm infants.
Arch Dis Child 1992;67:944-7.

Reuss ML, Paneth N, Pinto-Martin JA, Lorenz JM, Susser M. The relation of
transient hypothyroxinemia in preterm infants to neurologic development at two
years of age. N Engl J Med 1996;334:821-7.

Alexander GR, Kogan M, Martin J, Papiemik B. What are the fetal growth
patterns of singletons, twins, and triplets in the United States? Clin Obstet
Gynecol 1998;41 :1 14-25.

Jacobsen BB, Hummer L. Changes in serum concentrations of thyroid hormones
and thyroid hormone-binding proteins during early infancy. Studies in healthy
fullterm, small-for-gestational age and preterm infants aged 7 to 240 days. Acta
Paediatr Scand 1979;68:411-8.

116

13.

14.

’15.

16.

17.

I8.

19.

20.

Martin CR, Van Marter LJ, Allred EN, Leviton A. Growth-restricted premature
infants are at increased risk for low thyroxine. Early Hum Dev 2001;64:119-28.

Uhrmann S, Marks KH, Maisels MJ, et al. Thyroid function in the preterm infant:
a longitudinal assessment. J Pediatr 1978;92:968-73.

Carrascosa A, Ruiz-Cuevas P, Potau N, et al. Thyroid function in seventy—ﬁve
healthy preterm infants thirty to thirty-ﬁve weeks of gestational age: a prospective
and longitudinal study during the ﬁrst year of life. Thyroid 2004;14:435-42.

Hirano T, Singh J, Srinivasan G, Pildes R. Post-natal thyroid function in low birth
weight infants; a longitudinal assessment of free thyroxine and thyroid hormone
binding globulin. Acta Endocrinol (Copenh) 1985;110:56-60.

Hume R, Simpson J, Delahunty C, et al. Human fetal and cord serum thyroid
hormones: developmental trends and interrelationships. J Clin Endocrinol Metab
2004;89:4097-103.

Mercado M, Yu VY, Francis 1, Szymonowicz W, Gold H. Thyroid function in
very preterm infants. Early Hum Dev 1988;16:131-41.

Murphy N, Hume R, van Toor H, et al. The hypothalamic-pituitary-thyroid axis in
preterm infants; changes in the ﬁrst 24 hours of postnatal life. J Clin Endocrinol
Metab 2004;89:2824-31.

Martin CR, Van Marter LJ, Allred EN, Leviton A. Antenatal glucocorticoids
increase early total thyroxine levels in premature infants. Biol Neonate
2005;87:273-80.

Nelson JC, Weiss RM, Wilcox RB. Underestimates of serum free thyroxine (T4)
concentrations by free T4 immunoassays. J Clin Endocrinol Metab 1994;79:76-9.

lI7

Chapter 5 Correlates of postnatal thyroid hormone levels among extremely

premature infants born between 24-28 weeks of gestational age

Studies have shown that differences in postnatal thyroid hormone levels exist
between extremely preterm infants (< 28 completed weeks of gestational age) and both
term and moderately preterm infants (29-36 completed weeks of gestational age). At
birth, the TSH levels of term infants peak within 30 minutes, followed by a sharp
decrease (1-3). Serum T4 and T3 levels increase quickly following the TSH surge and
peak at 24 hours after delivery. The thyroid hormone surges of moderately preterm
infants are similar to term infants, but on a much smaller scale (1, 4). For extremely
preterm infants, however, the thyroid hormone surges have been found greatly attenuated,
or even absent (5, 6).

It is not well known, however, among extremely preterm infants, whether prenatal
and postnatal exposures are associated with the postnatal thyroid hormone levels.

Though previous ﬁndings have suggested that postnatal thyroid hormone levels are
associated with gestational age, postnatal age, respiratory distress syndrome, and sepsis,
they have not taken into account the correlation among hormone measurements within
each subject in their analyses (7-10). In addition, very little is known about the
association between necrotizing enterocolitis (NBC) and decreased postnatal thyroid
hormone levels.

In this chapter, we use data collected from the Phase 1 Study of Thyroid Hormone
in Extremely Premature Infants 1) to observe the postnatal thyroid hormone level changes

over time in premature infants born between 24 and 28 weeks of gestational weeks; and 2)

118

to study the correlates of thyroid hormone levels of premature infants during the neonatal
period, adjusting for correlation among hormone measurements within each subject. The
results of this analysis may help identify infants at risk for THOP and possibly reduce the
risk of adverse neurodevelopmental outcomes through timely supplementary treatment.
5.1 Analytical sample (Figure 5.1)

Among the 168 study subjects, we excluded 121 participants in this analysis for
the following reasons: 1) received thyroxine supplementary treatment (i.e. subjects in the
four thyroxine supplementary treatment groups), 2) expired or withdrew from the study
prior to day 7 of life, 3) were not the ﬁrstbom of any multiple birth set if more than one
infant in this set was enrolled, and 4) data were missing on postnatal thyroid hormone
results, maternal history, obstetric information, or postnatal exposures.

Among the 54 study subjects who did not receive thyroxine treatment, 28 were
randomized to the placebo group and 26 to the potassium iodine (KI) group. One subject
in the placebo group was excluded from the analysis because of death before 7 days of
life. Two subjects in the K1 group were excluded because parents withdrew consent
before 7 days of life. Among the remaining 51 subjects, 40 were singletons and 11 were
multiple births. The multiple births included one pair of twins (the second-bom
excluded), 4 single twin enrolled from four different twin pairs (zero excluded), and ﬁve
births from two triplet sets (three excluded). We excluded a total of seven infants from
multiple births. The ﬁnal analytical sample included 47 subjects.

5.2 Exposures
The ﬁndings from chapter 4 suggest that baseline F T4 levels were associated with

gestational age, race, fetal growth ratio, intrauterine growth restriction, and multiple birth

119

status. Baseline TT4 levels also were associated with antenatal steroid and magnesium
treatment, and rupture of membrane for greater than 24 hours (ROM > 24 hours) prior to
the onset of labor.

In this analysis, we would like to explore whether these factors will continue to
have similar associations with postnatal thyroid hormone levels over time. In addition,
previous studies have shown that postnatal illnesses, such as respiratory distress
syndrome, chronic lung disease, and sepsis, are associated with low serum thyroxine
levels in preterm infants. Hence, we will study these factors as well in relation to
postnatal thyroid hormone levels.

Three types of exposure variables were studied: 1) prenatal and birth exposures
(e.g., antenatal medication, gestational age, fetal growth ratio); 2) postnatal
characteristics during hospitalization (e.g., death, postnatal age, length of hospital stay,
duration of ventilator/oxygen use, chronic lung disease, NEC, receiving antibiotic
treatment for greater than 5 consecutive days (antibiotic > 5 days)); and 3) neonatal
thyroid hormone levels at birth. Most exposures were ﬁxed attributes (e.g., gestational
age, fetal growth ratio). NBC and antibiotic > 5 days were time dependent variables and
were both binary exposures, which were absent at birth and became present when the
disease occurred.

Prenatal and birth information were collected at birth through maternal interview.
Neonatal baseline thyroid hormone levels were tested within 24 hours after birth but
before the study treatment was administered. Postnatal information was collected on a

daily basis and summarized and recorded in discharge/death form at discharge or death.

120

5.3 Outcome

Postnatal thyroid hormone levels collected on day 3, 7, 14, 21, 42, and 56 were
the outcome variables in this analysis. We tested the normality of neonatal FT4, TT4,
and TSH distributions and found that all hormones followed the lognormal distribution.
Therefore, a log transformation was applied to all hormone results and the report below is
based on log transformed hormone values.

Missing postnatal thyroid hormone results. Because blood specimens were
collected for each study subject on postnatal day 3, 7, 14, 21, 42, and 56, the 47 subjects
in our study would potentially have a total of 282 measurements in hormone levels.
However, results of 12 subjects were partially incomplete because of death (11 samples
of 5 subjects) and missing records (7 samples of 7 subjects). Most samples (85%, n=224)
had all three hormone results (FT4, TT4, and TSH) available, 11% (n=30) of the samples
had two test results available, and 4% (n=10) had only one test result available for
analysis. Missing data were due to insufﬁcient sample volume, because of which
multiple hormone testing could not be completed. Missingness was considered as
random.

5.4 Statistical analyses

As the outcome variables of this analysis are a series of repeated hormone levels
on different postnatal days, we used analysis for repeated measurements to study the
associations between exposures and outcomes. A repeated measurement analysis allows
us to account for the correlation among hormone levels collected at different time points
within each study subject. The Mixed procedure in SAS software (ver. 9.1.3) was used in

all analyses.

121

5.5 Results
5.5.1 Prenatal and birth infant characteristics of 47 enrolled subjects (Table 5.1)

Birthweight of the 47 subjects ranged from 440 to 1325 grams (mean :1: standard
deviation (SD): 864 g i194). Gestational age ranged from 24 to 27.9 completed weeks
(mean i SD: 26.2 weeks :1: 1.2). Fetal growth ratio varied from 0.51 to 1.24 (mean 3: SD:
0.87 :1: 0.15). More than half (60%) of the analytical sample was male. The majority of
the population was White (62%). 15% of the subjects were one of a multiple-birth set,
36% were born through C-section, and 9% had intrauterine growth restriction. About
25% of subjects had an Apgar score at 1 minute below 4 points. By ﬁve minutes, the
Apgar scores of most infants (56%) had improved and only 2% remained under 4. About
16% of the subjects had a large improvement in Apgar scores, changing from less than 4
points at 1 minute to greater than 7 points at 5 minutes.
5.5.2 Maternal characteristics of 47 enrolled subjects (Table 5.1)

The 47 subjects were born to 47 mothers, whose ages ranged ﬁom 19 to 40 years
(mean :t SD: 29.5 years i: 5.7). During this current pregnancy, 89% of the mothers
received antenatal steroid and 26% received magnesium treatment prior to delivery.
About 24% of women had ROM > 24 hours prior to onset of labor. Both neonatal and
maternal characteristics of the 47 study subjects were similar to those attributes in the 144
subjects studied in previous chapter.
5.5.3 Postnatal characteristics of 47 study subjects (Table 5.2)

The 47 subjects were followed until discharge or death. The weight at the end of
follow-up ranged from 750 to 6335 grams (mean i SD: 2797g i1045). On average,

subjects stayed in the neonatal intensive care unit for 92 days before discharge or death.

122

Except for one outlier case hospitalized for 343 days, most subjects (72%) stayed in the
hospital for under 100 days and 26% stayed for more than 100 and less than 200 days.
Duration of ventilator use varied from 0 to 300 days (mean :1: SD: 28 d :t 53). About 30%
of the subjects were on a ventilator for less than 28 days and only three (7%) were on a
ventilator for above 60 days. The outlier case mentioned above was on a ventilator for
300 days. Compared to ventilator use, subjects were on CPAP for a shorter time. The
mean duration on CPAP was 21 days, varying from 0 to 100 days.

During the study treatment (up to 42 days postnatal) and follow-up period (up to
discharge or death), eight (17%) subjects expired, 35 (75%) were still on oxygen
supplement past postnatal age 28 days, 17 (36%) were still oxygen dependent at postnatal
age 36 weeks, 22 (47%) received one or more antibiotic treatments for ﬁve or more
consecutive days, and three (6%) developed NBC.

5.5.4 Neonatal baseline thyroid hormone levels of 47 study subjects (Table 5.3-5.4)

Average baseline FT4, TT4, and TSH levels of the 47 subjects were similar to
those of the 144 subjects discussed in chapter 4, except that no hormone level differences
existed across three study centers. Hormone levels also were compared between placebo
and K1 groups. The F T4 levels of the K1 group tended to be slightly higher and TSH
levels slightly lower than that of placebo group, but no signiﬁcant difference was found
in any of the three hormones between two study groups. In the adjusted analysis, we
included baseline thyroid hormone levels to test any association between baseline and

postnatal thyroid hormone levels.

123

5.5.5 Postnatal thyroid hormone levels of 47 study subjects (Figure 5.2-5.7)

In general, mean FT4 levels of the 47 study subjects appeared reasonably stable
over the ﬁrst 56 days of life. The mean FT4 levels increased approximately 0.5 ng/dl
over the ﬁrst week, decreased back to the baseline level, and then remained ﬂat.
Compared to the placebo group, subjects of the K1 group had slightly higher FT 4 levels at
all time points except on day 21. The FT4 level differences between two groups were not
statistically signiﬁcant.

Mean TT4 levels of the 47 subjects decreased about 2 jig/d1 in the ﬁrst week,
stayed at the lowest level (4 ug/dl) until the end of the second week (day 7-14), and
gradually rose after the end of the second week (day 14). The mean TT4 level at
postnatal day 56 was slightly above the baseline level. Similar to FT4, TT4 levels of the
K1 group were slightly higher than those of the placebo group except on day 7. The TT4
level differences between two groups were not statistically signiﬁcant either.

Mean TSH levels of the 47 subjects initially decreased abOut 3.5 mU/L in the ﬁrst
three days, then increased until around day 21, but still below the baseline mean value,
and plateaued thereafter. TSH levels of the K1 group were lower than that of the placebo
group at most time points except on day 14 and 21. No signiﬁcant differences between
TSH levels of two groups were detected. Regardless of the lack of a statistically
signiﬁcant difference between the placebo and K1 group for each hormone studied, we
included treatment assignment in the adjusted analysis.

5.5.6 Correlates of postnatal FT4 levels in 47 study subjects (Table 5.5)
We studied the correlates of postnatal FT4 levels with adjustment for correlation

among measurements within each subject and other factors. Although the mean FT4

124

level on day 56 was higher than those on other days (n.s.), the mean F T4 levels had little
variation over postnatal growth. We also observed that the FT4 levels of the K1 group
were slightly higher than that of the placebo group at most of the time points, however,
the difference was not statistically signiﬁcant. Among the rest of the variables we tested,
no variable was correlated with FT4 levels over 56-day postnatal period.

5.5.7 Correlates of postnatal TT4 levels in 47 study subjects (Table 5.6)

TT4 levels were positively and signiﬁcantly associated with postnatal age. Figure
5.4 showed that the TT4 levels decreased in the ﬁrst week and increased steadily since
the end of the second week. The TT4 difference between the two study groups also was
tested but no signiﬁcant association was found.

Among the prenatal exposures we studied, gestational age remained positively
and signiﬁcantly correlated with TT4 levels over time after adjusting for correlation
among hormone levels within each study subject and other factors. This association was
not seen in FT4 levels. Baseline TT4 levels were not found to be associated with
postnatal TT4 levels.

Among the postnatal exposures, duration of ventilator use was negatively and
signiﬁcantly correlated with TT4 levels. On average, every 10 days of ventilator use was
associated with a 5.6% decrease in TT4 levels (p < 0.0001). No association was found
for either CPAP use or oxygen use. NEC also was signiﬁcantly associated with TT4
levels. Subjects who developed NEC had a 53% reduction in TT4 levels (p=0.0167). No

other postnatal exposure variables were signiﬁcantly associated with postnatal TT4 levels.

125

5.6 Discussion

It has consistently been reported that thyroid hormone surges in extremely
premature infants (< 28 weeks of gestation) are greatly attenuated, or even absent (2, 5).
In our study, no postnatal thyroid hormone surge was observed in either FT4 or TT4.
However, we only have data at birth and day 3 of life. Due to the limited time points of
sample collection, it is possible that we missed the known 24-hour thyroid hormone surge
seen in term infants.

In a thyroxine supplement trial conducted in Amsterdam, Van Wassenaer et al.
found that the mean plasma F T4 concentrations of 100 preterm infants (25-30 wk) in
their placebo arm increased slightly (about 0.4 ng/dl) from birth on day 1 followed by a
nadir on day 7, and then increased again (6). Our results differed from their ﬁndings in
F T4 changing pattern. We found the mean FT4 level on day 7 was not the lowest, but in
fact the highest among FT4 measurements.

We further explored the hormone changes over time by three study sites (Figure
5.8-5.9). We found that the FT4 levels of the 12 Amsterdam subjects followed a pattern
similar to that described by Van Wassenaer’s group. On the other hand, the mean F T4
levels of subjects from both New York (n=25) and Madrid (n=10) were highest on day 7;
New York was lowest on day 21, and Madrid lowest on day 14. The variation by site
could be due to population differences, or due to a slight difference in gestational age
between the 12 Amsterdam subjects and those of New York and Madrid (25-28 vs. 24-28
weeks of GA). This also could explain why our ﬁnding among Amsterdam subjects is
consistent with that of Van Wassenaer and colleagues. We next examined the FT4 levels

over time by treatment group in the Amsterdam subjects and found no effect of treatment.

126

This was expected given that our Amsterdam FT4 trend was similar to the placebo group
studied by Van Wassenaer et al. Due to the differences by site, our data suggest that for
infants under 28 weeks of gestation, there is hardly any pattern in FT4 changes in the
early neonatal period, but modest variation exists during the ﬁrst three weeks of life.

In chapter 4, baseline F T4 levels of preterm infants were found to be associated
with a group of attributes (e.g., gestational age, race, fetal growth ratio), but postnatal
FT4 levels were not associated with any of these factors. In addition, postnatal exposures
(e.g., ventilator or oxygen use, postnatal illnesses) did not seem to affect serum FT4
levels either. Postnatal F T4 has been rarely studied and we do not have a clear
understanding of the role of F T4 in postnatal life. As thyroid hormone levels could be
affected by many factors, this lack of association may be due to the limited variables we
examined in this analysis. It also is possible that since the times of sample collection
were ﬁxed, while the measurement of exposures was not, we may have missed the
etiologically relevant time window for certain exposures studied‘(l 1).

After a decrease in the ﬁrst week of neonatal life, TT4 tended to rise gradually
with increasing postnatal age. We hypothesize that this increase was related to the
increasing production of both thyroxine and serum thyroid binding globulin (TBG)
concentrations in preterm infants. Each TBG molecule has one iodothyronine binding
site, hence, the T4 binding capacity of TBG in normal human serum is equivalent to its
blood concentration (12). If our hypothesis is correct, we would expect increasing
production of TBG as neonates grow. If the increasing rate of TT4 and TBG were equal,
we would observe ﬂat TT4/TBG ratio over time; however if 1T4 production was greater

than that of TBG, we would see increasing TT4/TBG over time. In Figure 5.10, the mean

127

TBG level of the 47 subjects increased steadily from 15 to 21 ug/ml. The TT4/1" BG ratio,
which paralleled the change of TT4, increased as well after an initial decrease. These
ﬁndings suggest a decrease in thyroxine synthesis in the ﬁrst week of life, but not in TBG
synthesis. In addition, both TT4 and TBG production increased in postnatal life and the
increase of TT4 was greater than that of TBG.

In our study, TT4 levels were negatively associated with duration of ventilator use.
Ventilator use reﬂects pulmonary function. In the literature, respiratory distress
syndrome (RDS) has repeatedly and consistently been reported to be associated with low
TT4 levels in preterm infants. Impaired lung function has been seen to be the most
common condition in preterm infants with hypothyroxinemia. Many groups have
reported that signiﬁcant reductions in serum T4 levels were found in preterm infants who
had RDS or who subsequently developed RDS (8-10). In our analysis, we used duration
of ventilator use, duration of CPAP use, and oxygen dependence at 36 weeks postnatal to
reﬂect neonatal lung function. After adjustment, a 5.6% reductiOn in 1T4 levels was
associated with each 10-day increase of ventilator use.

We have three possible mechanisms to explain the relationship between impaired
lung function and neonatal hypothyroxinemia. During early fetal life, thyroxine has been
known to be a key factor to fetal growth. Intrauterine thyroid hormone deﬁciency may
lead to a shortage of cortisol, a key stimulator of pulmonary surfactant synthesis. Lack of
cortisol would delay lung maturation (7), produce surfactant deﬁciency, and impose a
greater risk of having lung problems in premature infants (10). Although intrauterine
thyroxine deﬁciency could impede fetal lung development, one thing to keep in mind is

that the condition itself is very likely a result of poor pregnancy conditions but not a

128

cause. Second, prematurity could lead to both lung underdevelopment and
hypothyroxinemia. Third, thyroid hormone-dependent pulmonary maturation could be
hindered directly by postnatal hypothyroxinemia. The severity of pulmonary illness
could be worsened and its duration prolonged (9). Unfortunately, given the restriction of
the data, the time order of the relationship between impaired lung function and
hypothyroxinemia cannot be resolved.

Simpson et al. classiﬁed preterm infants by severity level of postnatal illnesses
(prematurity, chronic lung disease, severe respiratory disease, cerebral pathology, patent
ductus arteriosus, and NEC) and found a statistically signiﬁcant reduction in postnatal
TT4 levels among newborns with maximal intensive care (13). In our study, subjects
with NEC had a 52% reduction in TT4, compared to those without. Although our
ﬁndings are similar to Simpson’s, results are not directly comparable because Simpson et
al. analyzed the effect of postnatal illnesses as a group, while we speciﬁcally looked at
NEC. Our results were limited because the power to detect the aSsociation between NEC
and TT4 levels was restricted by our study objectives, which aimed to establish the
optimal dosing schedule for very preterm infants. Only three out of 47 subjects
developed NEC during the 56-day follow-up period. Such a small sample size can only
suggest a possible association between NEC and reduced postnatal TT4 levels. Future
study with larger sample size is needed to conﬁrm this ﬁnding.

5.7 Strengths and limitations

The present analyses have several strengths that must be considered. Studies have

measured postnatal thyroid hormone levels longitudinally, however, unlike our study,

very few have accounted for the correlation among measurements within each subject in

129

their analyses of hormone trends. Even fewer studies have examined the correlates of
postnatal thyroid hormone levels using analysis for repeated measurements. In an
improvement over previous studies, we assessed the serum FT4 concentrations using the
Direct Equilibrium Dialysis method, which maximally avoided possible biases from
changes in serum proteins or protein binding of T4, and accurately reﬂected the serum
FT4 levels in extremely premature infants. An additional advantage of our analyses is
that we had very few subjects with missing outcome data and all missingness was random,
and hence should not bias our results. Finally, the trends of postnatal thyroid hormone
observed in this study can be generalized to extremely premature infants.

We analyzed data from this clinical trial as a prospective cohort study. Restricted
by the original study design, we can only examine hormone results on 47 subjects from
the two non-treatment groups. This limited the power to detect associations between
exposures of interest and postnatal thyroid hormones levels in our study. Further, even if
power had been sufﬁcient for our analyses, it is possible that we may have missed the
etiologically relevant time window for certain exposures studied. In addition, as expected,
we had some lost to follow-up due to deaths (5, 10.6%), however, the number was
minimal and is unlikely to affect our results.

5.8 Conclusion

In summary, among preterm infants born between 24 and 28 weeks of gestational
age: 1) a postnatal FT4 surge did not exist in this study population and hardly any pattern
in FT4 levels variations over time was present; 2) a postnatal TT4 surge was not observed
and "1T4 levels remained lowest in the ﬁrst week before increasing; 3) postnatal thyroid

hormone levels increased over postnatal age and only the elevation in serum TT4 levels

130

was statistically signiﬁcant; 4) potassium iodine treatment did not affect postnatal thyroid
hormone levels compared to placebo; 5) postnatal FT4 levels were not clearly affected by
any prenatal or postnatal exposures examined; and 6).postnatal TT4 levels were
signiﬁcantly associated with gestational age, postnatal age, duration of ventilator use, and
NEC diagnosis.

Due to the limited knowledge in this ﬁeld, though this study contributes to the
understanding of postnatal thyroid hormone level changes and correlates, prospective
cohort studies with larger sample sizes are needed to more rigorously evaluate these

associations.

I31

Figure 5.1 Flow chart of analytical sample selection.

168 enrolled subjects

 

/

114 received
thyroxine
supplement

1

l

 

Not eligible for
analysis

 

 

A
N
28 received 26 received
placebo potassium
iodine (KI)
81 Chad < 7 $2 withdrew
days < 7 days
82 multiple 82 multiple
birth sets birth sets
v 17

 

25 eligible for 22 eligible for
analysis analysis
\ J

 

Y
Analytical sample size = 47

132

Figure 5.2 Mean structure of F T4 levels over postnatal age in 47 study subjects.

FT4 (ng/dl)

14H
l——l
.14

 

 

YrﬁﬁYTrrvfflv vv'

5 10 15 20 25 30 35 40 45 50 55 60
Postnatal age (day)

133

Figure 5.3 Mean structure of F T4 levels over postnatal age in 47 study subjects by
treatment groups.

 

 

 

 

 

 

5~ .
3 Treatment Groups —' Placebo (05W)
; -- lodine(Kl)
4£
1
1
1
1
a: 3?
U .
\
O)
C
V
V
’—
L1.
1 .1.
11
0%. -

 

 

0 5 10 15 20 25 30 35 40‘ 45 50 55 60
Postnatal age (day)

134

Figure 5.4 Mean structure of TT4 levels over postnatal age in 47 study subjects.

TT4 (meg/d1)

18:
17%
161
15%
143
133
125
11%

 

IIYYYYII‘VYIVYYY'T'I‘I’IYYYII" YIYYYIIVIIITIifY‘Ij

0 5 10 15 20 25 30 35 40 45 50 55 60
Postnatal age (day)

.4
.4

135

Figure 5.5 Mean structure of TT4 levels over postnatal age in 47 study subjects by

treatment groups.

1T4 (mcg/dl)

18‘
17:

 

 

Treatment Groups — Placebo (05W)
-— Iodine (Kl)

 

 

 

 

 

.4

..1

YYY’Y—IIvvt‘YTUTTTV—T‘Tth'TIIV!VVIYVYYIYYYV'VYVTI

25 30 35 40 45 50 55 60
Postnatal age (day)

136

Figure 5.6 Mean structure of TSH levels over postnatal age in 47 study subjects.

TSH (mU/L)
3.271. 9°. 5.0..

#01

J
I— 1

 

O—KNOJ

 

YITYYTIVIVVIVYVVIYYYYYYIVIlvavlvvvafTTl[rvayvﬁvv'rvTTj

30 35 40 45 50 55 60
Postnatal age (day)

0.1
01
_r.
O
..s
01
N
O
N
01

I37

Figure 5.7 Mean structure of TSH levels over postnatal age in 47 study subjects by
treatment groups.

TSH (mU/L)

 

Treatment Groups — Placebo (05W)
'- Iodine (Kl)

 

 

 

4...;
O_L

‘lAAJl‘A‘IA

 

LAJLAAAIAA

41A]

 

O-wanhU'lm'NmtO

 

'r'] VVVVVVVVV I vvvvvvvvv [r1 TrrvYY'leI'VIII'VYYVIYIVVIfTYI"

25 V30 35 4O 45 50 55 60
Postnatal age (day)

OH
(1'1
..L
O
.1.
01
N
o

138

Figure 5.8 Mean structure of FT4 levels over postnatal age in 12 study subjects enrolled
from Amsterdam site.

 

 

51
d
4
r
a
4
4
a
.
.
4.
.4
1
4
<
4
4
1
4
3..
a 4
'C .
\ 4
8’ ‘
v ‘
e
v 1
|— .
m 4
2‘1
4
1
‘1'
a
4
A...
1 L l
1
1-1
4
4
01
YYYTVIY’YDY[’YVYY'IYVV'Y'VV‘IV'VYT'Y—V’YIYVVV'VV 'VIVVYVI V'YV'T’V’T

 

0 5 10 15 20 25 30 35 4O 45 50 55 60
Postnatal age (day)

139

Figure 5.9 Mean structure of FT4 levels over postnatal age in 12 study subjects enrolled
from Amsterdam site by treatment groups.

 

 

 

 

 

 

 

5* .
1 Treatment Groups — Placebo (05W)
1 -- lodine(Kl)
41
1
1
c: 3?
'c 1
\
C» 1
C .
V 4
q- .
“1': :
2". /\
Ill \
/ \
1' \
;. x --.im-t— -------- a ———————
: T T
1. l i i—éE
1
03......-.... . ......... . .............. . . .

 

0 5 1O 15 20 25 30 35 40 45 50 55 60
Postnatal age (day)

140

Figure 5.10 Mean structure of TT4/TBG and TBG levels over postnatal age in 47 study
subjects.

 

 

 

 

1-01 . 140
0.9{
0.8} I
l ‘ :30
0.7: I
o 0'63 E
E 3 3’
V:- C20 E,
l: : 0
_ m
1 l—
310
0.1 [
00 .............. .10
0 5 10 15 20 25 30 35 40 45 50 55 60

Postnatal age (day)

141

Table 5.1 Neonatal and maternal characteristics of 47 study subjects.

 

 

 

 

 

 

 

 

 

Neonatal characteristics =47
Birthweight Mean (SD) 864 (194)
Median (Range) 860 (440, 1325)
Gestational age Mean (SD) 26.2 (1 .2)
Median (Range) 26.4 (24, 27.9)
Fetal growth ratio Mean (SD) 0.87 (0.15)
Median (Range) 0.87 (0.51, 1.24)

N (%)
Gender Male 28 (60)
Race White 29 (62)
Black 13 (28)
Hispanic 4 (9)
Multiple birth (twins and triplets) 7 (15)
C-section 17 (36)
Intrauterine growth restriction (IUGR) 4 (9)
Apgar score at l min S 3 ll (25)
4 - 6 17 (39)
Apgar score at 5 min 5 3 l (2)
4 - 6 5 (1 l)
Apgar score change from 1 min to 5 min 5 3 to (4 - 6) 3 (7)
(4-6)toZ7 15(34)
5 3 to Z 7 7 (l6)

Maternal characteristics N=47
Maternal age (yr) Mean (SD) 29.5 (5.7)
Median (range) 29 (19, 40)

N (%)
Antenatal steroids administration 42 (89)
Magnesium administration 12 (26)
ROM > 24 hr 11 (24)

Table 5.2 Postnatal characteristics of 47 study subjects.

Postnatal characteristics N=47
Discharge weight (gram) Mean (SD) 2797 (1045)
Median (range) 2614 (750, 6335)
Duration of hospital stay (day) Mean (SD) 92 (53)
Median (range) 85 (15, 343)
Duration of ventilator use (day) Mean (SD) 28 (53)
Median (range) 9 (0, 300)
Duration of CPAP use (day) Mean (SD) 21 (19)
Median (range) 19 (O, 100)

N (%)
Death 8 (17)
Oxygen dependent > 28 days 35 (75)
Oxygen dependent at postnatal 36 weeks 17 (36)
Necrotizing enterocolitis 3 (6)
Germinal matrix/intraventricular hemorrhage 15 (33)
White matter damage 5 (l l)
Treatimeith antibiotics > 5 consecutive days 22 (47)

 

l42

Table 5.3 Neonatal baseline tlmoid hormone levels bystudy centers.
Percentile

 

 

 

 

 

 

Hormone/Site N Mean (SD) Median Range 10th 25th 75th 90th p value
Baseline FT4 (pg/d1)
New York 24 1.6 (0.8) 1.4 (0.7, 3.5) 0.7 1.0 1.9 2.4
Amsterdam 12 1.5 (0.6) 1.4 (1.0, 2.8) 1.1 1.1 1.9 2.3 0.7390
Madrid 9 1.4 (0.4) 1.3 (0.8, 1.9) 0.8 1.1 1.6 1.9
Total 45 1.5 (0.6) 1.4 (0.7, 3.5) 0.8 1.1 1.8 2.4
Baseline TT4 (ng/dl)
New York 24 6.5 (3.2) 5.9 (3.1, 17.4) 3.5 4.3 7.2 9.]
Amsterdam 10 6.4 (2.5) 5.8 (3.4, 10.2) 3.8 4.3 8.4 10.2 0.9474
Madrid 8 6.8 (2.5) 7.3 (3.5, 10.9) 3.5 4.5 8.4 10.9
Total 42 6.5 (2.9) 5.9 (3.1, 17.4) 3.5 4.3 7.5 10.1
Baseline TSH (mlU/ml)
New York 23 6.1 (4.5) 4.5 (1.3, 17.1) 1.4 2.5 10.3 11.8
Amsterdam 12 4.1 (2.8) 3.8 (0.8, 12.0) 1.8 2.5 4.7 5.4 0.3544
Madrid 8 5.5 (2.7) 5.6 (1.3, 10.4) 1.3 4.0 6.6 10.4
Total 43 5.4 (3.8) 4.4 (0.8, 17.1) 1.4 2.8 7.1 11.5
Table 5.4 Neonatal baseline thyroid hormone levels by treatment groups.
Percentile
81:31:31 N Mean (SD) Median Range 10th 25;]. 75th 90th p-value
Baseline FT4 (pg/d1)
Placebo 23 1.4 (0.5) 1.3 (0.7, 2.8) 1.0 1.0 1.6 1.9 0 166
KI 22 1.7 (0.8) 1.6 (0.7, 3.5) 0.8 1.1 2.0 2.4 '
Total 45 1.5 (0.6) 1.4 (0.7, 3.5) 0.8 1.1 1. .8 2.4
Baseline TT4 (ng/dl) '
Placebo 23 6.5 (2.6) 5.9 (3.1, 13.0) 3.5 4.3 7.5 10.2 0 976
K] 19 6.5 (3.2) 5.9 (3.4, 17.4) 3.5 4.3 8.0 9.1 '
Total 42 6.5 (2.9) 5.9 (3.1, 17.4) 3.5 4.3 7.5 10.1
Baseline TSH (mlU/ml)
Placebo 23 6.0 (3.3) 5.4 (1.3, 13.6) 2.2 3.9 7.1 11.8 0 303
Kl 20 4.8 (4.3) 3.1 (0.8, 17.1) 1.3 2.2 5.8 11.2 '
Total 43 5.4 (3.8) 4.4 (0.8, 17.1) 1.4 2.8 7.1 11.5

 

143

Table 5.5 Correlates of postnatal FT4 levels in 47 preterm infants born between 24 and
28 completed weeks of gestational age.

 

 

 

 

Estimate (n=47) SE t Value p-value Overall p-value
Intercept 2.22 1.57 1.41 0.1665 0.1665
Postnatal age (day) .
Day 3 -0.14 0.12 -1.14 0.2540
Day 7 -0.06 0.12 -0.48 0.6285
Day 14 -0.08 0.12 -0.65 0.5196 0 8500
Day 21 -0.15 0.12 -1.17 0.2430 '
Day 42 -0.07 0.12 -0.55 0.5828
Day 56 ref - - -
Study group
Potassmm iodine 0.03 0.11 0.31 0.7564 0.7564
Placebo ref - - -
Prenatal exposures
Gestational age (wk) -0.04 0.06 -0.63 0.5295 0.5295
Fetal growth ratio -0.60 0.41 ~1.48 0.141 1 0.1411
Race
Black -0.24 0.13 -1.82 0.0710 0.2190
Hispanic 0.20 0.19 1.02 0.3089
Other -0.19 0.37 -0.52 0.6021
White ref - - -
Baseline thyroid hormone levels
Log (Baseline F T..) 0.22 0.17 1.35 0.1775 0.1775
Postnatal exposures
Duration of hospital stay (day) 2.2 e—3 2.4 e-3 0.90 0.3670 0.3670
Duration of ventilator use (day) -2.2 e-3 2.0 e-3 -1.11 0.2679 0.2679
Duration of CPAP use (day) -3.3 e-3 4.2 e-3 -0.78 0.4339 0.4339
Oxygen dependent at postnatal 36 weeks
Yes 0.01 0.14 0.10 0.9237 0.9237
No ref - - -
Receiving antibiotic treatment > 5 consecutive day
Yes 0.12 0.12 1.02 0.3075 0.3075
No ref - - -
N ecrotizing enterocolitis
Yes 0.32 0.31 1.04 0.2995 02995
No ref - - -

 

* adjusted for the rest of the variables.

144

Table 5.6 Correlates of postnatal TT4 levels in 47 preterm infants born between 24 and
28 completed weeks of Estational age.

Estimate (n=47) SE t value p-value Overall p-vaIue

 

 

 

 

Intercept -2.29 1.10 -2.08 0.0467 0.0467
Postnatal age (day) .

Day 3 -0.45 0.10 -4.62 < 0.0001

Day 7 -0.62 0.09 -6.75 < 0.0001

Day 14 -0.55 0.10 -5.74 < 0.0001

Day 21 cm 0.10 -4.17 < 0.0001 < 0'00"]

Day 42 -0.07 0.09 -0.82 0.4126

Day 56 Ref - - -
Study group

Potassium iodine 0.04 0.08 0.49 0.6243

Placebo Ref - - - 0'6243
Prenatal exposures
Gestational age (wk) 0.14 0.04 3.22 0.0016 0.0016
Antenatal steroids use prior to delivery

Yes -0.01 0.13 -0.08 0.9336

No Ref _ _ _ 0.9336
Magnesium sulfate use prior to delivery

Yes 0.06 0.09 0.66 0.5111 0 5] 11

No Ref - - - '
Rupture of membranes > 24 hours prior to delivery

Yes 0.1 1 0.10 1.09 0.2778

No Ref _ _ _ 0.2778
Baseline thyroid hormone levels
Log (Baseline TT4) 0.07 0.13 0.55 0.5854 0.5854
Postnatal exposures
Duration of hospital stay (day) 0.8 e-3 1.7 e-3 0.49 0.6253 0.6253
Duration of ventilator use (day) -5.8 e-3 1.4 e-3 -4. 18 < 0.0001 < 0.0001
Duration of CPAP use (day) 1.3 e-3 3.3 e-3 0.41 0.6860 0.6860
Oxygen dependent at postnatal 36 weeks

Yes 0.1 1 0.10 1.05 0.2956

No Ref _ _ _ 0.2956
Receiving antibiotic treatment > 5 consecutive day

Yes 0.06 0.10 0.57 0.5680

No Ref _ _ _ 0.5680
Necrotizing enterocolitis

Yes -0.76 0.30 -2.55 0.0117

No Ref _ - - 0.01 17

 

* adjusted for the rest of the variables.

145

10.

ll.

12.

BIBLIOGRAPHY

Fisher DA, Klein AH. Thyroid development and disorders of thyroid. function in
the newborn. N Engl J Med 1981;304:702-1-2.

Hume R, Simpson J, Delahunty C, et al. Human fetal and cord serum thyroid
hormones: developmental trends and interrelationships. J Clin Endocrinol Metab
2004;89:4097-103.

Jacobsen BB, Hummer L. Changes in serum concentrations of thyroid hormones
and thyroid hormone-binding proteins during early infancy. Studies in healthy
fullterm, small-for-gestational age and preterm infants aged 7 to 240 days. Acta
Paediatr Scand 1979;68:411-8.

Diamond FB, Jr., Root AW. Thyroid function in the preterm infant. Adv Exp Med
Biol 1991;299:227-41.

Murphy N, Hume R, van Toor H, et al. The hypothalamic-pituitary-thyroid axis in
preterm infants; changes in the first 24 hours of postnatal life. J Clin Endocrinol
Metab 2004;89:2824-31.

van Wassenaer AG, Kok JH, de Vijlder JJ, et al. Effects of thyroxine
supplementation on neurologic development in infants born at less than 30 weeks'
gestation. N Engl J Med 1997;336:21-6.

Cuestas RA, Engel RR. Thyroid function in preterm infants with respiratory
distress syndrome. J Pediatr 1979;94:643-6.

Mercado M, Yu VY, Francis 1, Szymonowicz W, Gold H. Thyroid function in
very preterm infants. Early Hum Dev 1988;16:131-41.

Reuss ML, Paneth N, Lorenz JM, Susser M. Correlates of low thyroxine values at

newborn screening among infants born before 32 weeks gestation. Early Hum
Dev 1997;47:223-33.

Romagnoli C, Curro V, Luciano R, et al. Serial blood T4 and TSH determinations
in low birth weight infants. Inﬂuence of gestational age, birth weight and neonatal
pathology on thyroid function. Helv Paediatr Acta 1982;37:331-44.

White E, Hunt JR, Casso D. Exposure measurement in cohort studies: the
challenges of prospective data collection. Epidemiol Rev 1998;20:43-56.

Larsen PR, Davies TF, Schlumberger M, Hay ID. Williams Textbook of

Endocrinology. Chapter 10 THYROID physiology and Diagnostic Evaluation of
Patients with Thyroid Disorders, 2002.

146

13. Report of working group of the British Association of Perinatal Medicine and
Neonatal Nurses Association on categories of babies requiring neonatal care.
Arch Dis Child 1992;67:868-9.

147

S

LIBRARlE

UI VERSITY

STATE

MICHIGAN

| l

7348

1 lllll‘

3 1293

* 111

6
5
9
2
0

 

 

 

 

 

 

 

 

 

.
. . v
i . . {I}
. .