Visual Abilities at 6 Months in Preterm Infants: Impact of Thyroid Hormone Deficiency and Neonatal Medical Morbidity

Abstract

Background:

Preterm infants are at risk for neonatal morbidity, transiently reduced thyroid hormone (TH) levels, and impaired visual abilities. To determine the interrelationship between these factors, we measured TH levels in the period ex utero and compared their visual abilities with those of term infants at 6 months (corrected) of age.

Methods:

The preterm group consisted of 62 infants stratified by gestational age: Group A (23–26 weeks, n = 10), Group B (27–29 weeks, n = 23), Group C (30–32 weeks, n = 19), and Group D (33–35 weeks, n = 10). Controls were 31 healthy full-term infants. In the preterm group, free thyroxine, triiodothyronine, and thyroid-stimulating hormone levels were measured at 2 and 4 weeks of life and 40 weeks postconceptional age. All infants were assessed for visual acuity, contrast sensitivity, and color vision using electrophysiological techniques.

Results:

Compared with controls, the preterm infants demonstrated reduced contrast sensitivity at low temporal frequencies and slower blue–yellow color processing. Groups did not differ from controls in visual acuity. In the preterm group, reduced contrast sensitivity and slow blue–yellow and red–green color vision processing were associated with low TH levels, low gestational age, and several medical morbidities.

Conclusions:

Our findings signify that some of the weak visual abilities in preterm infants can be accounted for, in part, by their reduced TH levels in the early postnatal period.

Introduction

P reterm birth affects a wide range of visual functions contributing to retinopathy of prematurity (ROP), strabismus, color vision deficits, reduced contrast sensitivity and visual acuity, and visual-field defects (1) as well as to weak visuospatial processing ability (2 –4). Although these deficits are usually attributed to the adverse circumstances surrounding the early birth in preterm children (5 –7), infants born preterm without any complications of prematurity may also show some form of visual impairment (8). Given the rise in and survival of an increasing number of extremely preterm infants (9,10), additional factors concerning the relationship between prematurity and reduced visual functioning need to be examined.

One potential factor contributing is the transient reductions in thyroid hormone (TH) level that typically follow preterm delivery (11,12). It is well established that TH is critical for normal brain development, including development of structures necessary for visual processing (13 –15). During pregnancy, maternal TH crosses the placenta into the fetal circulation (16). Because of the delayed maturation of the fetal thyroid system, maternal TH is the fetus' sole source of hormone in early gestation (17), and over the course of gestation, this reliance on maternal TH supply diminishes to assume full thyroid function at term. However, because preterm infants are severed prematurely from the mother and consequently her supply of TH, they often experience an insufficiency of TH in the period ex utero. This state of reduced TH, which is known as transient hypothyroxinemia of prematurity (THOP) (18), can also be exacerbated by neonatal illness (19) and illness severity (20). Relevantly, THOP has been linked to suboptimal neurodevelopment (21 –23), including weak visuospatial abilities (12,24). Although treatment with L-thyroxine has been shown to improve outcome in infants with THOP, a benefit was only seen in infants born before 29 weeks gestation (25,26) and the specific effects of TH treatment on subsequent visual functioning were not directly examined.

In this study, we used objective electrophysiological techniques involving visual evoked potentials (VEPs) to examine whether early postnatal TH levels are related to later visual abilities, namely contrast sensitivity, visual acuity, and color vision. In light of animal studies showing early TH insufficiency affects retinal development (15), we predicted a relationship between visual deficits and low early postnatal levels of TH in preterm infants and that this relationship would be further modified by factors relating to timing of birth and neonatal illness.

Methods

Participants

The preterm sample consisted of 67 infants recruited from the Sunnybrook Health Sciences Centre (SHSC) in Toronto, Canada, a University of Toronto-affiliated hospital supporting Levels I, II, and III neonatal nurseries. The infants were drawn from all eligible sequential cases at 23–35 weeks gestation between November 2004 and January 2007. One investigator (E.A.) informed all prospective parents about the study during child's first week of life, and if they were in agreement, she obtained via heelprick a sample of the child's blood (0.5 cc) at 2 weeks of life.

Controls (Group E, 37–42 weeks gestation) were 38 infants born during the same period in the SHSC Level I neonatal nursery. Based on careful review of hospital records by the investigator, only parents of healthy full-term infants were solicited via letter for participation.

As per the standard procedure in Ontario until May 2006,* all infants were screened at 2–3 days of age for both phenylketonuria and congenital hypothyroidism, with the latter using the thyroid-stimulating hormone (TSH) test. Subsequently born infants were screened for congenital hypothyroidism (CH) and up to 21 other diseases. Any infant with a positive screen or known or suspected congenital or chromosomal abnormality (e.g., down syndrome) was excluded. Additional exclusion criteria were as follows: child or maternal history of thyroid disease; maternal exposure to organic solvents, alcohol, and other substances of abuse; maternal history of anticonvulsant or antirhythmic medication during pregnancy. Of the original 67 preterm participants, two in Group C were subsequently diagnosed with congenital hypothyroidism and one in Group D died while one infant from Group C, one from Group D, and seven from Group E (who participated in our study of infant attention at 3 months of age) dropped out prior to the 6-month vision assessment. Final sample sizes were Group A = 10, Group B = 23, Group C = 19, Group D = 10, and Group E = 31. There were 13 sets of twins in the study: 4 in Group B, 5 Group C, 3 Group D, and 1 Group E.

The overall study permitting recruitment, blood sampling, and medical chart review was approved by the SHSC Research Ethics Board. All subsequent follow-up phases were approved by The Hospital for Sick Children (HSC), where all of the assessments were carried out. Parents or guardians of all participants provided signed informed consent.

Measures

Demographic data were obtained on gestational age, birth weight, gender, and socioeconomic status (SES) using the Hollingshead index (27). For each preterm neonate, the following medical morbidities were recorded from the child's medical chart at SHSC: highest bilirubin point and age at peak point; presence of sepsis (positive blood or cerebral spinal fluid culture), respiratory distress syndrome, patent ductus arteriosus (PDA), bronchopulmonary dysplasia, or necrotizing entercolitis; evidence of severe white matter brain injury (periventricular leukomalacia, late ventriculomegaly, peri- or intraparenchymal hemorrhage); use of surfactants or phototherapy; and ROP. Except for ROP and bilirubin, for which the highest stage and zone and highest point and age at peak, respectively, were recorded, each morbidity factor was categorized as present or absent.

TH levels were determined for preterm infants at 2 and 4 weeks of age and 40 weeks postconceptional age (PCA). However, given the overlap, or near overlap, between the 4- and 40-week PCA sampling for Group D, they were only sampled at 2 and 4 weeks of life, and their 4-week results were substituted in the database for their 40-week PCA interval values. Later blood samples were drawn at SHSC or a suburban satellite hospital to which the infant was transferred. If the child was discharged before collection of the second or third samples, a nurse traveled to the home and drew the blood. All samples were assayed in the Department of Laboratory Medicine at HSC for levels of free thyroxine (fT4), total triiodothyronine (T3), and TSH using the Bayer Immuno 1 or Bayer IMS analyzer.

To assess vision, two noninvasive VEP methods were used: (i) sweep VEP for contrast sensitivity and visual acuity and (ii) transient VEP for color perception. The sweep method presented vertical black-and-white sinusoidal (striped) gratings using PowerDiva software (v.1.8.5; Smith Kettlewell Eye Research Institute, San Francisco, CA) to generate stimuli and record responses. To measure contrast thresholds, VEPs were recorded while infants viewed a grating of fixed spatial frequency (0.5 cycles per degree [cpd]), which was “swept” from low (0.5%) to high contrast (20%) at three temporal frequencies (6, 10, and 15 Hz). To measure visual acuity, infants viewed gratings of fixed contrast (80%) that were “swept” across spatial frequencies from 3 to 23 cpd. For further details on stimulus presentation and threshold estimation, see articles by Mirabella et al. and Till et al. (28,29). Average levels of contrast sensitivity and visual acuity across trials were provided by the PowerDiva system for each infant. However, if an average score was unavailable for a particular condition, we then derived individual trial scores using Power Diva tools and recorded the score with the largest signal to noise ratio of above 4, if possible (30).

The transient VEP system served to measure responses to chromatic stimuli. Stimuli consisted of three vertical sinusoidal (striped) gratings of low spatial frequency (0.5 cpd) presented in an onset–offset (100–400 ms) mode at 2 Hz. One grating was presented along a tritanopic confusion axis, which produced modulation of the short (S) wavelength sensitive cones; a second grating was presented along an axis orthogonal to this, which modulated long (L) and medium (M) wavelength sensitive cones selectively; and a third grating (achromatic condition) was presented with all cones modulated proportionally. The stimuli were generated using VisionWorks Stimulus Maker software (Vision Research Graphics, Durham, NH) and evoked responses were acquired on a Dell Dimension 4100 and analyzed using the NeuroScan 4.2 program (Compumedics Neuroscan and Compumedics DWL, Charlotte, NC). Stimuli were presented in a pseudo-random order and evoked responses were averaged over a minimum of 30 presentations per stimulus type. Recordings were deemed reliable if the P1 waveform components from each of the two averaged recordings were within 10 ms of each other. If not deemed reliable, no response was recorded in the statistics database for that condition. Waveform latencies were measured from pattern onset to peak of the component [see Ref. (29) for complete details].

All VEPs were administered at 6 months corrected age and measured using the International 10–20 system (31) with active electrodes placed over the infant's scalp at O1, Oz, and O2, and at Cz (reference) and Pz (ground). During testing, each infant sat on the parent's lap and viewed the monitor binocularly from a distance of 100 cm. Child's attention was attracted to the stimuli by dangling a small toy ∼2 cm from the screen. At the end of testing, each infant received an ophthalmologic examination for refraction and fundus examination purposes.

Data management and analyses

Any missing hormone and visual ability data were imputed by using the mean value of the child's gestational group or a further value adjusted for child's overall level of performance (31). For the contrast sensitivity data, replacement involved first adjusting the gestational-age group mean value by child's score relative to the subgroup mean for the other two temporal conditions if these data were available; otherwise, the gestational-age group mean was used. For visual acuity, we used the gestational-age subgroup mean. Regarding color vision, we imputed missing values accordingly: (a) for those infants who had a reliable achromatic response but no reliable chromatic VEP waveform, we substituted their missing red–green or blue–yellow latencies with the mean of the corresponding gestational-age group adjusted by child's achromatic latency relative to the one for the gestational-age group; (b) for infants without an achromatic latency, the gestational-age group mean was used. This set of procedures resulted in an average of four imputed values per gestational-age group. For each subgroup, all outliers were also replaced by winsorizing (32), an established procedure whereby the outlying value was replaced with the score of the nearest neighbor within the 95th confidence range.

One-way analyses of variance (ANOVA) was used to test for group differences in participant demographics, nonbinary indices of illness, and TH levels. A homogeneity of variance test was used to account for unequal sample sizes between groups, and for significant variables, we reported the Brown-Forsythe statistic. Post hoc analyses employed the Dunnett's T3 or Dunnett t-test, where applicable, comparing each preterm group with controls. Chi-square analyses evaluated group differences on binary (present/absent) indices of illness. Repeated measures ANOVA served to compare TH levels at the different time periods within the preterm group. Multivariate ANOVA tested for group and gender differences in sweep and transient VEP measures with the Dunnett t-test for post hoc analyses. If gender was significant for any measure, multivariate analysis of covariance (MANCOVA) with gender as covariate was used.

To assess the combined contribution of demographics (gestational age, birth weight, gender, SES), neonatal illness, and TH levels on visual functioning in the preterm group, each condition significantly differentiating preterm and control groups was independently correlated with each predictor variable. Factors showing significant associations were entered together into a multiple linear regression to determine their combined impact on visual ability. A two-tailed significance level of 0.05 was applied to all analyses.

Results

Table 1 shows participant demographics by gestational-age groupings. Groups differed significantly (p < 0.01) in gestational age, birth weight, SES, and gender composition but not mean APGAR* scores at 1 or 5 minutes. As expected, birth weight was strongly correlated with gestational age, but because birth weight is expected to covary with gestational age (infants born earlier in gestation are typically smaller), birth weight was not used as a covariate. Although SES was lowest in Groups A and B, it was not used as a covariate because it did not correlate with visual abilities. The results indicated that Group A had more females than males (90% vs. 10%) and Group C had more males than females (84% vs. 16%), whereas males and females were roughly equally distributed in the other three groups. Gender was therefore used as a covariate in assessing for group differences in visual ability.

Table 1.

Demographic and Morbidity Characteristics of the Sample

Characteristics	Group A: 24–26 weeks	Group B: 27–29 weeks	Group C: 30–32 weeks	Group D: 33–35 weeks	Group E: 37–42 weeks	p-Value
Sample size	10	23	19	10	31	—
Gestational age, weeks	25 ± 6/7	28 ± 1	31 ± 4/7	34 ± 9/7	39 ± 9/7	<0.001
Birth weight, grams	738.9 ± 128.7	1212.9 ± 267.2	1502.4 ± 267.2	2228.8 ± 366.5	3531.7 ± 472.9	<0.001
Gender, % male	10	56.5	84	50	51.6	0.005
SES rank	2.5	2.3	1.5	1.2	1.4	<0.001
APGAR score (1 minute)	5.0	7.0	7.0	8	NA	ns
APGAR score (5 minutes)	8.0	8.0	8.0	8	NA	ns
Sepsis	20	17.4	5.3	0	NA	ns
Respiratory distress syndrome	90	100	63.2	33.3	NA	<0.001
PDA	80	39.1	10.5	0	NA	<0.001
Surfactant use	40	69.6	26.3	0	NA	0.001
Bronchopulmonary dysplasia	11.1	4.5	0	0	NA	ns
Evidence of brain injury	50	56.5	36.8	0	NA	0.021
Necrotizing enterocolitis	0	0	5.3	0	NA	ns
Phototherapy use	50	100	90.1	28.6	NA	0.002
Retinopathy of prematurity	40	40	20	0	NA	0.075
Bilirubin
Highest point (μmol/L)	168.10 ± 20.21	213.45 ± 35.34	227.83 ± 34.78	195.75 ± 29.92	NA	<0.001
Age at peak (days)	4.10 ± 1.29	5.24 ± 1.79	5.33 ± 1.71	4.25 ± 2.38	NA	ns

Data are presented as mean ± standard deviation (SD) (gestational age, birth weight, and bilirubin), as medians (APGAR), and as percentage of the group demonstrating presence of each medical morbidity.

SES, socioeconomic status; ns, not significant; NA, not available; PDA, patent ductus arteriosus.

Table 1 also summarizes the different gestational age groups' medical morbidity findings. Significant differences were seen for presence of respiratory distress syndrome (p < 0.001), PDA (p < 0.001), use of surfactants (p < 0.01) or phototherapy (p < 0.01), any evidence of brain injury (p < 0.05), and highest point (μmol/L) of bilirubin (p < 0.001), all reflecting higher morbidity in the lower gestational age groups. A trend toward significance (p = 0.075) was seen for ROP stage, reflecting the higher incidence of ROP in infants born before 30 weeks gestation. No differences were seen in frequency of sepsis, bronchopulmonary dysplasia, necrotizing enterocolitis, and age at peak bilirubin.

Table 2 summarizes the 2-, 4-, and 40-week PCA TH levels for preterm groups. For TSH, no group differences and no significant changes in TSH over time were observed. However, a significant interaction was seen between group and time of blood sample (p = 0.039), reflecting the greater decline in TSH levels over time for infants born before than after 32 weeks gestation.

Table 2.

Thyroid Hormone Levels According to Gestational Age Group

Thyroid hormone	Group A: 24–26 weeks (n = 10)	Group B: 27–29 weeks (n = 19)	Group C: 30–32 weeks (n = 23)	Group D: 33–35 weeks (n = 10)	p-Value
TSH (mIU/L)
2 weeks	6.15 ± 5.54	5.02 ± 2.97	5.23 ± 2.78	3.69 ± 2.23	ns
4 weeks	4.05 ± 4.83	3.94 ± 2.07	3.44 ± 1.42	4.89 ± 3.17	ns
40 weeks PCA	5.83 ± 5.62	3.91 ± 1.81	3.34 ± 0.63	4.89 ± 3.17	ns
fT4 (pmol/L)
2 weeks	13.21 ± 5.13	18.58 ± 3.95	19.41 ± 3.17	22.55 ± 2.62	<0.001
4 weeks	16.58 ± 4.62	20.17 ± 3.53	22.11 ± 2.26	20.87 ± 2.22	<0.01
40 weeks PCA	18.59 ± 2.09	19.00 ± 2.58	18.74 ± 1.15	20.87 ± 2.22	0.051
T3 (nmol/L)
2 weeks	0.73 ± 0.25	1.08 ± 0.32	1.24 ± 0.33	2.22 ± 0.57	<0.001
4 weeks	1.34 ± 0.46	1.60 ± 0.33	1.63 ± 0.24	2.61 ± 0.41	<0.001
40 weeks PCA	2.77 ± 1.11	2.49 ± 1.41	2.01 ± 0.13	2.61 ± 0.41	ns

Data are presented as mean ± SD.

TSH, thyroid-stimulating hormone; fT4, free thyroxine; T3, triiodothyronine; PCA, postconceptional age.

For fT4, significant group differences were seen at all three time points (p < 0.05). Post hoc tests indicated that Group A had significantly lower 2-week fT4 levels than the other groups, whereas Groups B and C differed from D but not each other. Four-week fT4 levels were significantly lower in Group A than C. For 40 weeks PCA values, a trend (p = 0.051) was seen for lower fT4 values in Groups A, B, and C than Group D. Repeated measures ANOVA revealed significant effects of time of blood sampling (p = 0.022) and a significant group by time interaction (p < 0.001). The latter was accounted for by the larger group differences at 2 weeks of life than at 40 weeks PCA and by the larger increase in fT4 between 2 and 4 weeks in Group A than the others, thus signifying that fT4 increased more with time for infants born with earlier than later gestation.

For T3, significant group differences were seen at 2 and 4 weeks of life (p < 0.001), but not at 40 weeks PCA. Post hoc tests revealed that at 2 weeks, Group A's T3 levels were significantly lower than the others, and Groups B and C did differ from D. At 4 weeks of life, T3 levels were significantly lower in Groups A and B than Group D. Repeated measures ANOVA revealed significant effects of time since birth (p < 0.001) and a significant group by time interaction (p < 0.001), which reflected a greater increase in T3 with time for infants born with lower gestation.

Fundus ophthalmoscopy showed no abnormality in any infant. However, when the groups were examined for refractive error, they differed (p < 0.05) in spherical error, but not cylindrical error size, in both left and right eyes. Post hoc analyses revealed that Group A was different from the other groups. The mean of this group was myopic compared with the normal hyperopic error of 6-month olds (33) (Table 3).

Table 3.

Refractive Error Results for Preterm and Full-Term Infants

Variable	Group A: 24–26 weeks (n = 10)	Group B: 27–29 weeks (n = 21)	Group C: 30–32 weeks (n = 15)	Group D: 33–35 weeks (n = 10)	Group E: 37–42 weeks (n = 23)	p-Value
Spherical error (D)
Right eye	−0.33 ± 1.45	0.38 ± 0.94	0.60 ± 1.07	0.80 ± 1.14	0.97 ± 0.72	0.019
Left eye	−0.23 ± 1.40	0.45 ± 0.94	0.58 ± 1.10	0.73 ± 1.08	1.01 ± 0.75	0.032
Cylindrical error (D)
Right eye	+0.69 ± 0.24	+0.82 ± 0.45	+0.92 ± 0.56	+0.89 ± 0.84	+0.70 ± 0.71	ns
Left eye	+0.88 ± 0.53	+0.67 ± 0.46	+0.82 ± 0.55	+0.89 ± 0.84	+0.42 ± 0.59	ns

D, dioptres.

On the sweep VEP, groups differed significantly (p < 0.01) in contrast sensitivity only at the slow (6 Hz) temporal frequency (Table 4). Post hoc tests revealed lower sensitivity (p < 0.05) for Groups A, B, and C than controls. Although the other temporal frequencies were not different among groups, contrast sensitivity at the medium (10 Hz) temporal frequency was lower for all preterm groups than controls and a linear relation between detection and gestational-age grouping was observed. Groups did not differ in the fast (15 Hz) temporal frequency condition, reflecting the poor contrast detection of all groups. There was no effect of gender on sweep VEP results. There were no significant group differences in visual acuity.

Table 4.

6-Month Visual Function Results According to Gestational-Age Group

Visual ability	Group A: 24–26 weeks (n = 10)	Group B: 27–29 weeks (n = 23) ^a	Group C: 30–32 weeks (n = 15) ^a	Group D: 33–35 weeks (n = 10) ^a	Group E: 37–42 weeks (n = 26) ^a	p-Value
Log contrast thresholds
6 Hz temporal frequency	1.74 ± 0.22^b	1.72 ± 0.21^b	1.72 ± 0.37^b	1.87 ± 0.14	1.98 ± 0.27	0.002
10 Hz temporal frequency	1.68 ± 0.24	1.67 ± 0.20	1.71 ± 0.29	1.76 ± 0.17	1.83 ± 0.29	ns
15 Hz temporal frequency	1.60 ± 0.14	1.52 ± 0.17	1.53 ± 0.31	1.54 ± 0.21	1.66 ± 0.28	ns
Visual acuity (cycles per degree) thresholds	14.88 ± 4.86	12.86 ± 2.56	13.12 ± 3.64	11.98 ± 3.03	12.51 ± 3.05	ns
Achromatic latency (ms)	120.60 ± 11.96^b	110.09 ± 5.69	114.42 ± 8.08	105.71 ± 12.66	108.60 ± 7.70	0.004
Blue–yellow latency (ms)	177.11 ± 22.01^c	156.21 ± 13.41	154.64 ± 13.48	157.68 ± 5.29	153.74 ± 11.50	0.001
Red–green latency (ms)	165.29 ± 20.65	136.74 ± 14.66^b	147.70 ± 18.73	141.58 ± 19.66	155.74 ± 24.40	0.011

Data are presented as mean ± SD.

Sample size decreased in these groups on chromatic measures such that for Group B, n = 14; Group C, n = 12; Group D, n = 7; and Group E, n = 20.

p < 0.05, post hoc analysis.

p < 0.001, post hoc analysis.

Groups differed significantly (p < 0.05) on achromatic, blue–yellow, and red–green transient VEP measures (Table 4). Post hoc analyses indicated that on the achromatic and blue–yellow conditions, these group differences were mainly driven by longer latencies in Group A infants, whereas they were driven by the shorter latencies of Group B in the red–green condition. Controlling for gender did not affect the group differences on any condition. However, a main effect of gender on red–green latencies (p = 0.037) revealed that males had shorter latencies than females, contrary to expectation. As these results are based on reliable red–green responses, it is unlikely that our results can be explained by congenital red–green color deficiencies.

The proportional contributions of demographic factors, TH levels, and medical morbidities to visual functioning were examined only for those endpoints on which preterm infants performed significantly more poorly than full-term controls (see Table 5). In the slow (6 Hz) temporal frequency contrast sensitivity condition, the combination of sepsis, PDA, and fT4 at 40 weeks PCA best explained this ability, accounting for 20% of its variance; the strongest predictor was PDA. On the transient VEP, achromatic latencies were best explained by the combination of gestational age, phototherapy use, T3 at 2-week, and fT4 at 40-week PCA, which accounted for 45% of its variance. Although not significant at an individual level, the 40-week PCA fT4 value was the strongest predictor of achromatic responses. For blue–yellow latencies, the combination of gestational age, PDA, bronchopulmonary dysplasia, and fT4 at 2 weeks accounted for 28% of its variance, with bronchopulmonary dysplasia being the strongest predictor. Red–green latencies were best explained by gender and 2-week TSH, fT4, and T3 levels, which accounted for 34% of its variance; the effects of fT4 at 2 weeks were highly significant.

Table 5.

Summary of Regression Analyses for Predictors of 6-Month Visual Function

Variable	β	SE	t Statistic	p-Value
Contrast Sensitivity (6 Hz)
Sepsis	−0.154	0.094	−1.632	ns
PDA	−0.151	0.065	−2.336	0.023
fT4 (40 week PCA)	0.026	0.014	1.913	0.061
Achromatic latency
Gestational age	0.181	0.142	1.279	ns
Phototherapy use	8.321	5.324	1.563	ns
T3 (2 weeks)	−6.425	5.332	−1.205	ns
fT4 (40 week PCA)	−1.604	0.967	−1.658	ns
Blue–yellow latency
Gestational age	−0.202	0.169	−1.201	ns
PDA	4.631	7.015	0.660	ns
BPD	18.046	12.334	1.463	ns
fT4 (2 weeks)	−0.330	0.650	−0.508	ns
Red–green latency
Gender	3.544	5.637	0.629	ns
TSH (2 weeks)	1.487	0.794	1.874	0.069
fT4 (2 weeks)	−1.936	0.698	−2.773	0.009
T3 (2 weeks)	0.589	5.637	0.095	ns

SE, standard error; BPD, bronchopulmonary dysplasia.

Discussion

TH is essential for normal visual development (13). Because infants born preterm are at risk of both transiently reduced TH levels (18) and impaired visual processing (1), we investigated whether differences in their neonatal TH levels could account for visual processing deficits at 6 months (corrected) age. Regression analyses revealed that it was not gestational age alone, but rather gestational age plus low TH levels as well as several medical morbidities that predicted poor contrast sensitivity and slow color vision processing.

A primary finding of this study was that contrast sensitivity effects were mainly driven by infants born at 23–32 weeks gestation, whereas color vision effects were confined to infants born before 30 weeks gestation. These findings suggest that different visual abilities may have discrete time points when affected by medical morbidity and when TH is needed. In our study, lower contrast sensitivity was associated with the presence of sepsis and PDA in the early neonatal period, as well as with decreased fT4 levels at 40 weeks PCA. Although previous research has shown that sepsis in extremely preterm infants contributes to vision impairment at 18–22 months (34) and TH deficiency during pregnancy to large contrast sensitivity deficits in offspring of hypothyroid women or with congenital hypothyroidism (35), this study is the first to demonstrate that low TH, particularly low fT4, is associated with contrast sensitivity deficits in 6-month-old preterm infants.

Lower 2-week T3 and 40-week PCA fT4 levels were associated with longer achromatic latencies in the transient VEP assessments, as were gestational age and phototherapy use. Although animal literature provides strong support for a role of TH in retinal development, the human literature describing such a relationship is virtually nonexistent. Regarding color vision, research with rodents has shown that in the developing retina, TH regulates cone opsin expression (15) and TH receptors are required for the development of green cone photoreceptors (36). Although preterm infants are known to be at increased risk of developing color vision deficits (1), this is the first work to examine the association between TH deficiency and color vision development in children known to be at increased risk of blue–yellow deficits (37). We presently showed that longer preterm blue–yellow latencies were associated with lower 2-week fT4 levels, as well as with shorter gestation and presence of PDA and bronchopulmonary dysplasia. Although preterm infants' red–green vision is usually comparable to that seen in infants born at term (38), with the exception of those infants born with the shortest gestation, we found that the preterm group as a whole had faster red–green responses than controls. In preterm infants, longer red–green latencies were associated with lower 2-week fT4 and T3 levels. Thus we demonstrated an association between hypothyroxinemia due to preterm birth and color vision development.

We did not find any effect for visual acuity, consistent with the extant literature showing either accelerated development (39) or no effect of birth before 32 weeks gestation (18).

Although this study is the first of its kind to examine how early TH levels influence several core visual abilities in preterm infants, it is not without limitations. First, and most notably, the sample sizes were small in Groups A and D, for whom recruitment was more difficult than anticipated. Indeed, the lack of significance in the 10 Hz contrast sensitivity condition, despite a clear difference between subgroups and a stepwise increase with gestational age, may reflect a lack of power rather than a lack of effect. Second, we experienced loss of some data due to equipment problems. Although a few mothers did bring their babies back when the system was fixed, this was not possible for all cases. Third, a few babies did not have a complete set of TH measurements, despite our concerted efforts to obtain these.

In conclusion, hypothyroxinemia in preterm infants, particularly their low postnatal fT4 levels, is associated with suboptimal visual processing at 6 months corrected age. We hope our findings will encourage further study of the role of TH on brain development following preterm birth as well as the effects of TH supplementation on visual development of extremely preterm infants.

Footnotes

Acknowledgments

The authors are grateful to Kelly Nash and Meagan Williamson for carrying out this study and its many facets, Dr. Christina Gerth for conducting the ophthalmologic examinations, and Jennifer Vaughan for collecting blood samples at SHSC and from discharged infants in their homes and for sending them to HSC. Funding for this research was provided by the March of Dimes, RESTRACOMP, and the Vision Science Research Program.

Disclosure Statement

The authors declare that no competing financial interests exist.

*

This was when the Provincial screening laboratory moved from Toronto to the Children's Hospital of Eastern Ontario in Ottawa and the program was expanded to 22 diseases.

*

APGAR scores (appearance, pulse, grimace, activity, respiration) are used to assess the health of a newborn immediately after birth.

References

O'Connor

, Wilson

, Fielder

. 2007. Ophthalmological problems associated with preterm birth. Eye, 21:1254–1260.

Vohr

, Garcia-Coll

, Mayfield

, Brann

, Shaul

, Oh

. 1989. Neurologic and developmental status related to the evolution of visual-motor abnormalities from birth to 2 years of age in preterm infants with intraventricular hemorrhage. J Pediatr, 115:296–302.

Waber

, McCormick

. 1995. Late neuropsychological outcomes in preterm infants of normal IQ: selective vulnerability of the visual system. J Pediatr Psychol, 20:721–735.

Deforge

, Andre

, Hascoet

, Fresson

, Toniolo

. 2009. Impact of very preterm birth on visuospatial processes at 5 years of age. Arch Pediatr, 16:227–234.

Edmond

, Foroozan

. 2006. Cortical visual impairment in children. Curr Opin Ophthalmol, 17:509–512.

Dammann

, Leviton

. 2006. Inflammation, brain damage and visual dysfunction in preterm infants. Semin Fetal Neonatal Med, 11:363–368.

Jakobson

, Frisk

, Knight

, Downie

ALS

, Whyte

. 2001. The relationship between periventricular brain injury and deficits in visual processing among extremely-low-birthweight (<1000g) children. J Pediatr Psychol, 26:503–512.

Downie

ALS

, Jakobson

, Frisk

, Ushycky

. 2003. Periventricular brain injury, visual motion processing, and reading and spelling abilities in children who were extremely low birthweight. Intern Neuropsychol Soc, 9:440–449.

Martin

, Kung

, Mathews

, Hoyert

, Strobino

, Guyer

, Sutton

. 2008. Annual summary of vital statistics: 2006. Pediatr, 121:788–801

10.

Joseph

, Huang

, Liu

, Ananth

, Allen

, Sauve

, Kramer

. 2007. Reconciling the high rates of preterm and postterm birth in the United States. Obstet Gynecol, 109:813–822.

11.

Uhrmann

, Marks

, Maisels

, Friedman

, Murray

, Kulin

, Kaplan

, Utiger

. 1998. Thyroid function in the preterm infant: a longitudinal assessment. J Pediatr, 92:968–973.

12.

Rovet

, Simic

. 2008. The role of transient hypothyroxinemia of prematurity (THOP) in development of visual abilities. Semin Perinatol, 32:431–437.

13.

Harpavat

, Cepko

. 2003. Thyroid hormone and retinal development: an emerging field. Thyroid, 13:1013–1019.

14.

Roberts

, Srinivas

, Forrest

, Morreale de Escobar

, Reh

. 2006. Making the gradient: thyroid hormone regulates cone opsin expression in the developing mouse retina. Proc Natl Acad Sci USA, 103:6218–6223.

15.

Forrest

, Reh

, Rüsch

. 2002. Neurodevelopmental control by thyroid hormone receptors. Curr Opin Neurobiol, 12:49–56

16.

Chan

, Rovet

. 2003. Thyroid hormones in fetal central nervous system. Fetal Matern Med Rev, 14:177–208.

17.

Calvo

, Jauniaux

, Gulbris

, Asuncion

, Gervy

, Contempre

, Morrale

. 2002. Fetal tissues are exposed to biologically relevant free thyroxine concentrations during early phases of development. J Clin Endocrinol Metab, 87:1768–1777.

18.

Fisher

. 1998. Thyroid function in premature infants. Clin Perinatol, 25:999–1014.

19.

Williams

, Ogston

, van Toor

, Visser

, Hume

. 2005. Serum thyroid hormones in preterm infants; associations with postnatal illnesses and drug usage. J Clin Endocrinol Metab, 90:5954–5963.

20.

Simpson

, Williams

, Delahunty

, van Toor

, Wu

, Ogston

, Visser

, Hume

. 2005. Serum thyroid hormones in preterm infants and relationships to indices of severity of intercurrent illness. J Clin Endocrinol Metab, 90:1271–1279.

21.

Reuss

, Paneth

, Pinto-Martin

, Lorena

, Susser

. 1996. The relation of transient hypothyroxinemia in preterm infants to neurologic development at two years of age. N Engl J Med, 334:821–827.

22.

Ishaik

, Asztalos

, Perlman

, Newton

, Frisk

, Rovet

. 2000. Hypothyroxinemia of prematurity and infant neurodevelopment: a pilot study. J Dev Behav Pediatr, 21:172–179.

23.

Simic

, Asztalos

. Rovet J 3-month neurocognitive development in preterm infants: impact of thyroid hormone deficiency, neonatal medical morbidity. Thyroid, 19:395–401

24.

Hinton

. 2006 Cognitive development in children with transient hypothyroxinemia of prematurity. Presented at the International Neuropsychological Society Annual Meeting, Boston, 22.

25.

Van Wassenaer

, Kok

, de Vijlder

, Briët

, Smit

, Tamminga

, van Baar

, Dekker

, Vulsma

. 1997. Effects of thyroxine supplementation on neurologic development in infants born at less than 30 weeks' gestation. N Engl J Med, 336:21–26.

26.

Briet

, Van Wassenaer

, Dekker

, De Vijlder

, Van Baar

, Kok

. 2001. Neonatal thyroxine supplementation in very preterm children: developmental outcome evaluated at early school age. Pediatrics, 107:712–718.

27.

Hollingshead

. 1975. Four Factor Index of Social Status. Yale Unviersity Department of Psychology: New Haven, CT.

28.

Mirabella

, Kjaer

, Norcia

, Good

, Madan

. 2006. Visual development in very low birth weight infants. Pediatr Res, 60:435–439.

29.

Till

, Rovet

, Koren

, Westall

. 2003. Assessment of visual functions following prenatal exposure to organic solvents. Neurotoxicology, 24:725–731.

30.

Lauritzen

, Jorgensen

, Michaelsen

. 2004. Test-retest reliability of swept visual evoked potential measurements of infant visual acuity and contrast sensitivity. Pediatr Res, 55:701–708.

31.

Engels

, Dierh

. 2003. Imputation of missing longitudinal data: a comparison of methods. J Clin Epidemiol, 56:968–976.

32.

Bollinger

, Chandra

. 2005. Iatrogenic specification error: a cautionary tale of cleaning data. J Labor Econ, 23:235–257.

33.

Banks

. 1980. Infant refraction and accommodation. Int Ophthalmol Clin, 20:205–232.

34.

Stoll

, Hansen

, Adams-Chapman

, Fanaroff

, Hintz

, Vohr

, Higgins

. 2004. Neurodevelopmental and growth impairment among extremely low-birth-weight infants with neonatal infection. JAMA, 292:2357–2365.

35.

Mirabella

, Westall

, Asztalos

, Perlman

, Koren

, Rovet

. 2005. Development of contrast sensitivity in infants with prenatal and neonatal thyroid hormone insufficiencies. Pediatr Res, 57:902–907.

36.

, Hurley

, Dierks

, Srinivas

, Salto

, Vennstrom

, Reh

, Forrest

. 2001. A thyroid hormone receptor that is required for the development of green cone photoreceptors. Nat Genet, 27:94–98.

37.

O'Connor

, Fielder

. 2007. Visual outcomes and perinatal adversity. Semin Fetal Neonatal Med, 12:408–414.

38.

Dobson

, Quinn

, Abramov

, Hardy

, Tung

, Siatkowski

, Phelps

. 1996. Color vision measured with pseudoisochromatic plates at five-and-a-half years in eyes of children from the CRYO-ROP study. Invest Ophthalmol Vis Sci, 37:2467–2474.

39.

Norcia

, Tyler

, Piecuch

, Clyman

, Grobstein

. 1987. Visual acuity development in normal and abnormal preterm human infants. J Pediatr Ophthalmol Strabismus, 24:70–74.