The Quartile Levels of Thyroid-stimulating Hormone at Newborn Screening Stratified Risks of Neurodevelopmental Impairment in Extremely Preterm Infants: A Population Cohort Study

Background To evaluate whether thyroid-stimulating hormone (TSH) measured during newborn screening (NBS) at birth and at discharge can be surrogate markers for neurodevelopmental impairment (NDI) in extremely preterm infants. Methods The population cohort enrolled infants born <29 weeks’ gestation in 2008–2020 in southern Taiwan. Infants with a maternal history of thyroid disorders and infants who required thyroxine supplementation during hospitalization were excluded. TSH levels measured during NBS at birth and at term-equivalent age (TEA)/discharge were respectively categorized into the lowest quartile, the interquartile range, and the highest quartile, which were correlated to NDI outcomes. Results Among 392 patients with paired TSH data, 358 (91%) were prospectively followed until a corrected age of 24 months. At birth, infants with lowest-quartile TSH had higher NDI risks (odds ratio [OR] 2.3; 95% confidence interval [CI], 1.3–4.1, P = 0.004) compared to infants with interquartile-range TSH. Conversely, by TEA/discharge, infants with highest-quartile TSH had increased NDI (OR 1.9; 95% CI, 1.0–3.4, P = 0.03). By paired TSH categories, infants persistently in the lowest TSH quartile (48%; aOR 4.4; 95% CI, 1.4–14.5, P = 0.01) and those with a shift from interquartile range to the highest quartile (32%; aOR 2.7; 95% CI, 1.0–7.4, P = 0.046) had increased NDI risks compared with the reference with consistent interquartile-range TSH. Conclusion Extremely preterm infants persistently in the lowest-quartile TSH level at birth and at discharge had the highest NDI risk. TSH quartile levels measured during NBS may serve as a population surrogate biomarker for assessing NDI risks in infants born extremely preterm.


INTRODUCTION
Despite the improvement in survival rate, infants born extremely preterm continue to experience a high proportion of neurodevelopmental impairment (NDI) at follow-up, including cerebral palsy, hearing/visual impairment, and delayed motor and cognitive development, which could greatly impact children's long-term quality of life. 1,24][5][6] However, these tools are available only in research settings, and a population-level biomarker for NDI risks in extremely preterm infants is still lacking.
Newborn screening (NBS) programs have been implemented at population level to facilitate early diagnosis of inherited metabolic disorders including congenital hypothyroidism, for which the primary thyroid-stimulating hormone (TSH) screening strategy has been widely adopted. 7Congenital hypothyroidism, a disorder of thyroid hormone deficiency due to incomplete thyroid development or inadequate thyroid hormone production, is regarded as the leading cause of preventable intellectual disability in children. 7In hypothyroid status, the pituitary gland releases more TSH as regulated by thyrotropin-releasing hormone from the hypothalamus, to stimulate thyroxine secretion by the thyroid gland. 7Taking advantage of the feedback loop of hypothalamuspituitary-thyroid axis, a cut-off value defining elevated TSH has been applied in NBS worldwide, to identify neonates at risk of congenital hypothyroidism early for prompt thyroxine supplementation in confirmed cases, subsequently preventing sequelae of intellectual disability. 7,8n preterm infants, while most studies focus on high TSH levels suggesting hypothyroid status, the significance of low TSH levels remains unclear.In general, low TSH levels may suggest hyperthyroidism, but hyperthyroidism in neonates is rare and mostly associated with maternal thyroid disorders that can be differentiated from the history. 9The hypothalamus-pituitarythyroid axis is immature in extremely preterm infants, leading to a more intricate interpretation for TSH levels from NBS. 7 Our previous study demonstrated that extremely preterm infants who had lowest-quartile TSH levels by the first NBS conducted at 24-96 hours after birth were associated with lung and brain complications in the neonatal intensive care units (NICUs), 10 suggesting low TSH levels as a potential marker for brain dysfunction.However, the significance of low TSH levels for neurodevelopmental outcomes at long-term follow-up remains unclear.A clinical trial study of daily iodine supplementation in very preterm infants <31 weeks of gestation who had blood sampling on postnatal days 7, 14, and 28 and postmenstrual age 34 weeks showed that infants consistently in the top decile TSH levels had decreased cognitive and motor scores assessed using the Bayley Scales of Infant and Toddler Development (BSID), Third Edition (BSID-III) at age 2 years compared with those not in the top decile. 11In contrast, a single center study from Wisconsin enrolling very preterm infants <32 weeks of gestation did not identify an association between TSH percentiles by the NBS obtained at postmenstrual age 37 weeks and BSID-III neurodevelopmental outcome scores at corrected age 18-22 months. 12herefore, whether TSH can be a useful biomarker for NDI risks in infants born extremely preterm remains unclear.
In Taiwan, NBS for preterm infants includes the first screening sampled within the first few postnatal days and a second screening performed at term-equivalent age (TEA) or upon discharge.The NBS program in Taiwan uses primary-TSH-withbackup-thyroxine method, so the database contains only TSH levels.Thyroid function tests, including thyroxine levels, are examined in detail at referral hospitals for those with high TSH assessed during NBS.Using an extremely preterm population cohort enrolling infants born <29 weeks of gestation from a geographically defined area in southern Taiwan, the study hypothesized that the quartile levels of TSH measured during NBS at birth and at TEA/discharge are associated with different risks of NDI outcomes at follow-up in extremely preterm infants.

Participants
Infants born <29 weeks of gestation in 2008-2020 and admitted to all the NICUs in Tainan City were enrolled.The city, with a population of nearly 2 million people in southern Taiwan, is served by three regional hospitals, one tertiary center, and one academic tertiary center of the university hospital for the intensive care of preterm neonates.Infants whose mothers had thyroid disorders, such as hyperthyroidism or hypothyroidism, and infants who required thyroxine supplementation during hospitalization were excluded.
In addition, the TSH levels measured during NBS of all term neonates born at gestational age ≥37 weeks in the city in one randomly-selected month during the study period were obtained for comparison.
The study was approved by the review board of National Cheng Kung University Hospital (ER-98-135).Parents signed the consents for data collection when infants were hospitalized in the NICUs and at neurodevelopmental follow-up visits.

TSH measurement by NBS program for infants born preterm
The national NBS program for congenital hypothyroidism in Taiwan utilizes the primary-TSH-with-backup-thyroxine method.Since 2008, in addition to the screening performed in the early postnatal days, preterm infants have received additional TSH screening at TEA/discharge.Therefore, preterm infants included in this study underwent TSH screening tests at two time points: at birth (24-96 hours of age) and at TEA/discharge (postmenstrual age 35-44 weeks).
Whole-blood TSH levels in the dried blood spot specimens were measured using a time-resolved fluoroimmunoassay (AutoDELFIA kit of neonatal human TSH, Perkin Elmer, Turku, Finland) at the NBS Center of the Chinese Foundation of Health, a semi-government non-profit organization for NBS program in Taiwan.Blood TSH levels ≥40 µU/mL were considered positive for congenital hypothyroidism.Infants with TSH levels ≥10 µU/mL but <40 µU/mL were classified as suspected congenital hypothyroidism necessitating re-examinations.Blood TSH levels <10 µU/mL were considered normal.Infants whose TSH levels ≥10 µU/mL were referred to pediatric endocrinologists for further evaluations. 7,13In Taiwan, the same cutoff value applies to infants born at term and preterm, and for early neonatal and TEA/discharge screenings.

Neonatal morbidities in NICUs
Demographic and perinatal data, including the use of dopamine due to its potential interaction with TSH secretion, 14 were recorded.Neonatal morbidities were reviewed, including severe brain injury (grade III intraventricular hemorrhage with ventriculomegaly and/or intraparenchymal hemorrhage, and cystic periventricular leukomalacia), necrotizing enterocolitis (NEC) stage ≥2a, 15 hemodynamically-significant patent ductus arteriosus (hs-PDA) requiring surgical interventions, bronchopulmonary dysplasia (BPD; defined as oxygen treatment >21% for at least 28 days followed by an assessment at 36 weeks of postmenstrual age or discharge), 16 and severe retinopathy of prematurity (ROP; defined as stage ≥2 plus disease or requiring treatment in either eye). 17

Neurodevelopmental outcomes
Neurodevelopmental outcomes were evaluated at a corrected age of 24 months by a team of pediatric neurologists and child psychologists at the university hospital, who were blinded to the TSH data when performing assessments.Infants born before 2011 were assessed using the BSID, Second Edition (BSID-II), and those born after 2011 were evaluated using BSID-III.Cerebral palsy was defined as abnormal muscle tonicity and gross motor function limitations with a gross motor function classification system score ≥2.Hearing impairment referred to permanent hearing loss that affected the ability to understand or communicate, or with a threshold higher than 55 dB in the better ear.Visual impairment represented bilateral blindness.Cognitive delay was defined as a mental developmental index <70 using BSID-II or a cognitive composite score <85 using BSID-III, and motor delay was defined as a psychomotor developmental index <70 using BSID-II or a motor composite score <85 using BSID-III. 18NDI encompassed any of the following deficits: cerebral palsy, hearing/ visual impairment, or delayed cognitive or motor development. 19,20

Statistical analyses
Density plots were used to present TSH distribution levels.The Wilcoxon signed-rank test was employed to compare the paired TSH data collected at birth and at TEA/discharge, and the Mann-Whitney test was used to compare the TSH data of preterm infants at TEA/discharge with those of term neonates.Clinical characteristics of infants with and without neurodevelopmental follow-up Newborn TSH Screening for Neurodevelopmental Outcomes in Prematurity were compared using Fisher's exact test for categorical variables and the Mann-Whitney test for continuous variables.TSH levels obtained through NBS at birth and at TEA/discharge were respectively divided into three groups based on quartiles: the lowest quartile, the interquartile range, and the highest quartile, according to previous literatures of thyroid function tests on neuropsychiatric outcomes. 21The associations of NDI with different TSH quartile groups during screening at birth, TEA/ discharge screening, and paired TSH quartile patterns were first examined using univariable logistic regression models, followed by multivariable logistic regression model adjusted for clinical factors that might associate with NDI, including gestational age, sex, maternal educational status, use of dopamine, total counts of neonatal morbidities, and the year of birth.For the quartile groups categorized by paired TSH data that showed increased odds for NDI, univariable logistic regression models were used to examine their associations with individual neonatal morbidity.The significance level was set at P < 0.05 for 2-sided hypothesis tests.The statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, Inc, San Diego, CA, USA) and R version 4.0.2(R Foundation for Statistical Computing, Vienna, Austria).

Enrollment of participants
A total of 484 infants with extremely preterm birth underwent NBS for TSH at 24-96 hours of age.After excluding 32 infants who died during hospitalization and 1 infant whose TEA/ discharge screening was not conducted, 451 infants had TSH screening at TEA/discharge.Infants excluded from analysis encompassed 13 infants whose mothers had a thyroid disorder, 27 infants who received thyroxine supplementation between the first and second TSH screening, and 19 infants whose TEA/discharge screening tests were performed earlier than 35 or later than 44 weeks of postmenstrual age.Among the remaining 392 infants, three patients died after discharge, one patient did not have complete clinical information for analysis, and 30 patients were lost to follow-up (Figure 1).

Clinical characteristics and neurodevelopmental outcomes
A total of 358 (91%) infants who had neurodevelopmental outcomes were included in the analysis.The 30 infants lost to followup had larger birth weights and lower rates of multiple gestation or receiving prenatal steroids compared to the infants who underwent follow-up assessments (Table 1).The 358 infants were born at a median gestation of 27 weeks, with a median birth weight of 956 grams, and were composed of 203 (57%) male and 155 (43%) female infants.Among them, 66 (18%) infants were diagnosed with severe brain injury (Table 2).NDI was observed in 82 (23%) infants, including 20 (6%) of cerebral palsy, 8 (2%) of hearing impairment, 1 (0.3%) of visual impairment, 57 (16%) of cognitive delay, and 55 (15%) of motor delay.After the NBS at TEA/ discharge, 3 (0.8%) infants received thyroxine supplementation.

NDI risks by TSH quartiles of NBS at birth and at TEA/discharge
The TSH data obtained through NBS at birth and at TEA/ discharge periods were categorized into the lowest quartile, the interquartile range, and the highest quartile.Using the group with TSH in the interquartile range as reference, we examined the NDI risks of infants in the lowest and highest quartiles of TSH.Based on NBS at birth, infants with TSH in the lowest quartile had higher odds of NDI compared with the reference group (37% vs 20%; odds ratio [OR] 2.3; 95% confidence interval [CI], 1.3-4.1,P = 0.004) (Table 3).Conversely, based on NBS at TEA/ discharge, infants with TSH in the highest quartile had higher odds of NDI compared with reference (30% vs 19%; OR 1.9; 95% CI, 1.0-3.4,P = 0.03).

NDI risks by patterns of paired TSH quartiles
We then examined the NDI risks based on the patterns of paired TSH quartiles at the two time points.Using infants who had interquartile-range TSH during both NBS at birth and at TEA/ discharge as the reference, we found that infants with TSH in the lowest quartile at birth followed by any quartile at TEA/ discharge were more susceptible to NDI (lowest to lowest quartile: 48%; OR 5.2; 95% CI, 1.9-14.6,P = 0.001; lowest quartile to interquartile range: 32%; OR 2.7; 95% CI, 1.1-6.4,P = 0.03; lowest to highest quartile: 35%; OR 3.1; 95% CI, 0.9-9.8,P = 0.05) (Table 4).Infants who had TSH in the interquartile range during NBS at birth but increased to the highest quartile at TEA/discharge also had higher odds of NDI (32%; OR 2.7; 95% CI, 1.1-6.4,P = 0.03).After adjusting for gestational age, dopamine usage, sex, maternal education below college level, total counts of neonatal morbidities, and the year of birth, infants whose TSH persistently in the lowest quartile (adjusted OR 4.4; 95% CI, 1.4-14.5,P = 0.01) and those with a shift from interquartile range to the highest quartile (adjusted OR 2.7; 95% CI, 1.0-7.4,P = 0.046) continued to have increased odds of NDI.

TSH quartile patterns and neonatal morbidities
We further analyzed whether the paired TSH quartile patterns with increased NDI risks were correlated with any specific neonatal morbidities.Compared with the reference group, infants with TSH persistently in the lowest quartile were associated with severe brain injury, whereas infants whose TSH increased from interquartile range to the highest quartile and other groups did not show an association with any specific neonatal morbidity (Figure 4).When adjusted for severe brain injury, NDI remained significantly increased in infants whose TSH persistently stayed in the lowest quartile (aOR 4.1; 95% CI, 1.3-13.3,P = 0.02).

Main findings
In the population cohort of infants born extremely preterm, the risk of NDI at 24 months of corrected age was associated with the changing patterns of TSH quartiles assessed during NBS from birth to TEA/discharge.The cohort used TSH data obtained through routine NBS program as a population-level marker for NDI.Infants enrolled in the study were followed up prospec-tively, with a follow-up rate of more than 90% at 24 months of corrected age.Our findings indicate that, compared to infants with consistent interquartile-range TSH, infants with TSH persistently in the lowest quartile had a 4-fold risk, and those whose TSH elevated from interquartile range to the highest quartile had a nearly 3-fold risk of NDI.These results suggest that TSH obtained through routine NBS is useful to stratify NDI risks, allowing for early identification and potential early intervention in high-risk infants.

NDI markers in preterm infants
NDI is one of the most concerned outcomes in extremely preterm infants. 22Predicting NDI risks at discharge may lead to earlier initiation of interventions, which can be more effective in alleviating impairments compared to initiating interventions at school age. 23Several models have been proposed to predict NDI in clinical settings for preterm populations, mostly relying on a combination of physiological profiles at admission and neonatal morbidities before discharge. 225][26][27] As biomarkers used at population level to Newborn TSH Screening for Neurodevelopmental Outcomes in Prematurity forecast NDI in extremely preterm infants are rarely reported, we took advantage of the NBS database and proposed TSH levels as a potential early marker for NDI.

NBS for mortality and morbidities at acute stage in NICUs
NBS, initially used to identify neonates at risk of inherited metabolic disorders, has started to contribute as metabolomic profiles in evaluating mortality and major morbidities associated with preterm birth in NICUs, 10,28,29 but it's association to neurodevelopmental outcomes at follow-up stage has not been established before. 12Our study extended the application of NBS beyond the acute stage in NICUs and into follow-up stage at 24 months of corrected age, validating the value of NBS in forecasting neurodevelopmental outcomes in the extremely preterm population.
Vulnerability of the immature hypothalamus-pituitary-thyroid axis in preterm infants Extremely preterm infants are born during the critical window when TSH secretion is not fully coordinated by upstream or downstream regulatory/feedback hormones. 30Our results showed that, even when these extremely preterm infants reached TEA, the density plot of TSH distribution at TEA/discharge still differed significantly from the density plot of term babies, suggesting that the extrauterine maturation process experienced by extremely preterm infants is not equal to the intrauterine maturation process in full-term pregnancy.
Critical illnesses and neonatal morbidities may further complicate the maturation process of the axis. 31We found that infants with TSH persistently in the lowest quartile, in whom nearly half of them had NDI at a corrected age of 24 months, were associated with severe brain injuries.3][34][35] However, even after adjusting for severe brain injury, we still observed a 4-fold risk of NDI, suggesting that the association between persistently low TSH levels and NDI may not be fully mediated through the presence of severe brain injury.

Low TSH and brain function alteration
7][38][39] For example, in a study enrolling nearly 30,000 adult subjects, low TSH levels were correlated to clinically relevant depression irrespective of subclinical or overt hyperthyroidism/hypothyroidism, suggesting low TSH levels as a marker for altered mental function independent of thyroxine levels. 38According to a population cohort in California enrolling mostly term babies, infants with a pattern of lower neonatal TSH levels were at higher risk for developmental delay. 21n addition, autistic children with delayed social-emotional development were shown to exhibit decreased TSH secretion upon thyrotropin-releasing hormone stimulation compared with controls. 39Therefore, our findings of increased NDI risks in extremely preterm infants with persistent low TSH levels may illustrate a blunted response of pituitary gland to the upstream hypothalamic control in the situation of deviated neurodevelopment. 39creasing TSH and NDI risks Previous studies have focused on the relationship between mildly elevated TSH levels suggesting subclinical hypothyroidism and neurodevelopmental outcomes in preterm infants.An iodine supplementation trial examining serial TSH levels collected on postnatal days 7, 14, and 28 and at postmenstrual age 34 weeks during hospitalization showed that infants with persistently high TSH levels in the top decile had inferior scores in cognition and motor function. 11However, the design of a clinical therapeutic trial and differences in laboratory techniques for TSH measurements across different hospitals may complicate the generalization of their findings.Interpreting NBS TSH data for outcome correlation could be more accessible, as it covers all preterm graduates from NICUs since NBS programs conduct nationwide screening using uniform techniques with limited trials of sampling.
Our study excluded infants with thyroid dysfunction necessitating thyroxine supplementation, and we still noticed that infants with an increasing pattern of TSH from interquartile range to the highest quartile had a higher risk for NDI, without an association to any specific neonatal morbidity.Previous studies have found that increased TSH levels but without thyroxine deficiency are more common in preterm and low-birth-weight infants. 40,41This phenomenon might be related to fetal growth restriction that changes the amount and isoform of hormone receptors in the immature brain and alters the function of the hypothalamus-pituitary-thyroid axis. 40The clinical significance of this rising TSH pattern remains unclear, 42 but an association with neurodevelopment was suggested by our findings.Further investigations are required to understand the mechanism linking isolated increased TSH concentration and neurodevelopmental outcomes in infants born extremely preterm.

Limitations
Limitations of this study include the lack of thyroxine data for infants with normal-range TSH levels according to the primary-TSH-with-backup-thyroxine NBS strategy.Therefore, we cannot further explore the role of thyroxine deficiency or excessiveness in NDI among infants with normal-range TSH at NBS. 43 Additionally, the study included a relatively small number of cases despite being derived from a population study, which was limited by the low rate of extremely preterm birth affecting 2-5 per 1,000 pregnancies. 44The low incidence of extremely preterm birth restricted this study from examining TSH levels in more details.Therefore, the generalizability of our findings may be limited because a proper cutoff value to define low or high TSH is not available.Therefore, recruiting a larger number of infants born extremely preterm with neurodevelopmental follow-up data is necessary to validate the association between the changing patterns of TSH levels from early postnatal period to TEA and NDI risks.
In conclusion, extremely preterm infants persistently in the lowest-quartile TSH level at birth and at discharge had the highest risk of NDI at follow-up.The TSH quartile levels determined through NBS may serve as a population surrogate biomarker for assessing the risk of NDI in infants born extremely preterm, aiding neurodevelopmental monitoring in targeted subgroups of infants for potential early intervention.

Figure 2 .
Figure 2. Density plots of TSH levels by NBS in infants born extremely preterm and at term.Preterm infants: early neonatal screening at 24-96 hours after birth (green) and term-equivalent age (TEA)/discharge screening at postmenstrual age 35-44 weeks (blue).Term neonates: newborn screening at 48-72 hours of age (red).
interval; NA, not available; TEA, term-equivalent age; TSH, thyroid-stimulating hormone.a Adjusted for gestational age, sex, use of dopamine, maternal education below college level, total counts of neonatal morbidities, and the year of birth.b No infant had neurodevelopmental impairment in this group.

Figure 3 .
Figure 3. Heatmap of predicted probabilities of NDI by paired TSH quartile groups and sex.Median predicted probabilities of NDI by TSH quartile groups were adjusted for gestational age, sex, use of dopamine, maternal educational status, total counts of neonatal morbidities, and the year of birth.Male infants (A) had higher predicted probabilities of NDI than female infants (B).No infant with TSH transiting from the highest quartile at birth to the lowest quartile at TEA/ discharge had NDI.* Statistically significant compared with the reference group with consistent interquartile-range TSH levels.

Figure 4 .
Figure 4. Neonatal morbidities and paired TSH quartile patterns.Reference: TSH in the interquartile range by both NBS at birth and at TEA/discharge.Low -Low: TSH persistently in the lowest quartile.Medium -High: TSH shifting from the interquartile range to the highest quartile.Other groups: the remaining 6 TSH patterns.

Table 1 .
Differences in demographic and perinatal data between infants with follow-up assessment and those lost to follow-up

Table 2 .
Neonatal morbidities and neurodevelopmental outcomes of 358 extremely preterm infants who received neurodevelopmental assessments at follow-up

Table 3 .
Neurodevelopmental impairment rates by TSH quartiles of newborn screening at birth and at term-equivalent age/discharge

Table 4 .
Risks for neurodevelopmental impairment by the paired quartile patterns of TSH