The present study found that TSH levels were significantly higher in SGA newborns than AGA ones. The association between higher TSH levels and SGA birth was further confirmed by the following multivariate regression analysis after adjusting for potential confounders, including the history of conditions, procedures, and medications before the initial TFT.
Taking advantage of the TFT results, our discovery have confirmed and expanded the findings of previous studies. Bosch-Giménez et al. reported higher TSH concentrations in SGA neonates, but their data were from NBS which measured TSH exclusively, and the TSH level range was set to <7.5mU/L . In our study, thyroid hormone values were from those infants with normal thyroid function diagnosed by one or more TFTs, hence our TSH values were higher than theirs. Franco et al. reported that TSH concentrations are significantly higher only in term SGA infants compared to AGA ones, whereas serum concentrations of T4 are lower in both preterm and term SGA infants . FT4 concentrations were significantly higher in the SGA group, although the significance did not hold in multivariate linear regression later. Nonetheless, our conclusions are in line with the two studies that SGA babies have a higher incidence of transient hypothyroidism and need close follow-up.
In the present study, the higher percentages of low 1-min Apgar score and Caesarian section delivery in the SGA group were seen. SGA infants do not have relatively mature functions of organ systems that AGA infants possess, so they are more susceptible to birth asphyxia and difficult deliveries, which tend to elevate TSH levels. Lower Apgar score, known as an indicator of asphyxia at delivery, has been associated with higher TSH levels [14, 15]. However, Rashmi et al. found the lowest TSH levels in infants born by elective Caesarian section delivery compared to other modes of delivery, suggesting it may be a factor decreasing TSH levels .
Besides altered thyroid hormone concentrations, we presented different kinds of thyroid dysfunction diagnosed by one or more TFTs in Table 2. The SGA infants showed a significantly higher incidence of transient hypothyroidism with dTSHe. Recently, a Japanese study showed that SGA birth is the only independent risk factor for the development of TH with dTSHe in the preterm infants weighing less than 2000g . Another study revealed that the prevalence of CH with dTSHe is associated with SGA birth in extremely and early premature infants (GA <30wk) . Our finding further underlines preterm SGA infants are prone to this disorder and should be closely followed up.
Additionally, we presented the rates of starting on L-T4 replacement therapy and follow-up in the SGA and AGA groups. The rate of SGA infants receiving the therapy was significantly higher than that of AGA ones, indicating the SGA infants suffered more from a deranged thyroid hormone secretion. The follow-up rate within the first 6 months of life was significantly higher in the SGA group, but the significance did not persist beyond. Compared to the AGA group, the SGA group was more likely to have thyroid dysfunction and longer treatment, but in most cases, the alterations were transient and reverted to normal during the following months. The incidence affected by hypothyroidism in the two groups did not show a significant difference in the long term.
Several lines of research addressed TSH elevation in fetuses and children born SGA. Thorpe-beeston et al. found that significantly higher levels of TSH and lower levels of T4 and FT4 in SGA fetuses, and suggested the associations may be explained by degrees of fetal hypoxemia and academia respectively . The authors also revealed some SGA fetuses may have abnormal thyroid function.
Radetti et al. found that TSH concentrations are significantly higher in children born SGA, and 20% SGA children have TSH levels above the upper limit of the normal range, whereas no difference was found for FT4 . De Kort et al. found higher TSH levels within the normal range in preterm short SGA children, but mean FT4 is not significantly different . Although these findings cannot lend direct support to our conclusion, they distinctly suggest that TSH elevation may play a similar role at different stages of the developmental process.
As thyroid hormones have been credited with a wide range of important physiologic functions, the etiology of higher TSH levels in SGA infants is worth exploring. It is postulated that many factors may involve, including immaturity of the hypothalamic-pituitary-thyroid axis, uterine stress with growth restriction, less efficient thermogenic response, or non-thyroidal illness. However, it is difficult to figure out the temporal pattern of thyroid hormone alteration caused by SGA stunting and to distinguish it from that of AGA infants, as either NBS or TFT is generally done at separate time points. Furthermore, exposure to various medications in hospitalization and the inability to regulate iodine balance may be other common reasons .
Given the importance of proper thyroid function, especially for the overall development in early and later childhood, some cohort studies investigated the impact of high neonatal TSH levels on long-term developmental and cognitive sequelae.
Differential effects of preterm and SGA birth on cognitive and motor development have been noted: SGA birth is associated with cognitive ability, as measured by IQ and reading comprehension, while motor ability was additionally associated with preterm birth . Furthermore, Trumpff et al. reported lower verbal IQ scores in preschool children with high TSH values between 10-15 mU/L in univariate analysis, but the result did not hold after adjusting for confounding factors . Chung et al. confirmed preterm infants with persistently high TSH levels have worse neurological outcomes compared to those with transiently high TSH levels . Nonetheless, inconclusive results demonstrate the clinical significance of neonatal TSH elevation is still under debate [23, 24]. Considering the potential cognitive risks in infancy and childhood, preterm SGA infants with TSH elevation should be closely followed up.
The main strength of the present study is that we both evaluated the normal thyroid hormone concentrations and thyroid dysfunction in the study population.
First, the Z-score of BW for GA and sex was applied to quantify the association of altered thyroid hormone levels and SGA birth. Most previous studies interpret altered thyroid hormone levels and thyroid function from the perspective of prematurity, and analyze the data based on either GA or BW. The present study elucidated the thyroid alterations and disorders from the perspective of BW adequacy for GA.
Second, the primary and following TFTs enabled us to fully interpret the status of thyroid hormone secretion and thyroid disorders in the preterm newborns. Serum TFT measures TSH and FT4 simultaneously, which is considered the ideal screening approach to investigate the thyroid function. Evidence indicates that the screening based on measuring only TSH levels cannot diagnose transient hypothyroidism, so its clinical significance is fairly limited . Therefore, measuring both FT4 and TSH in screening tests is advisable.
Third, a detailed collection of clinical risk factors facilitated the adjustment for these potential confounders in the multivariate regression analysis. As our study population was preterm SGA and AGA neonates admitted into NICU, the clinical confounding effects might show false associations if we failed to control them.
The present study has several limitations. First, the number of cases is relatively small compared to large-scale NBS, and further studies including more cases are warranted to confirm the findings. Second, it is a retrospective, single-center study in nature, and the data were extracted from the NICU database, thus our results may not be generalizable to other populations. Third, as the study subjects were hospitalized in the NICU, some confounding factors such as illness severity, procedures, and medications were inevitably introduced . Although we have applied multivariate regression analysis to adjust for these potential confounders, bias and misclassification may not be ruled out, so the results should be interpreted cautiously.