Predictive factors of permanent versus transient congenital hypothyroidism: a pragmatic cohort study

Article information

Ann Pediatr Endocrinol Metab. 2025;30(3):149-156

Publication date (electronic) : 2025 March 20

doi : https://doi.org/10.6065/apem.2448126.063

Niki Dermitzaki^,1, Vasileios Giapros¹, Marianna Deligeorgopoulou², Vasiliki Rengina Tsinopoulou³, Eleni Kotanidou³, Maria Baltogianni¹, Foteini Balomenou¹, Anastasios Serbis²

¹Neonatal Intensive Care Unit, School of Medicine, University of Ioannina, Ioannina, Greece

²Pediatric Department, School of Medicine, University of Ioannina, Ioannina, Greece

³2nd Department of Pediatrics, AHEPA University General Hospital, School of Medicine, Aristotle University of Thessaloniki, Thessaloniki, Greece

Address for correspondence: Niki Dermitzaki Neonatal Intensive Care Unit, School of Medicine, University of Ioannina, Stavrou Niarchou Avenue, 45500, Ioannina, Greece Email: n.dermitzaki@uoi.gr

Received 2024 June 1; Revised 2024 September 24; Accepted 2024 November 12.

Abstract

Highlights

· TSH at diagnosis, TSH >5 μIU/mL during treatment, LT4 dose, LT4 needs >50 μg/day, and need for LT4 increases were identified as potential predictors for distinguishing transient from permanent congenital hypothyroidism.

Introduction

Congenital hypothyroidism (CH) is one of the most common endocrine diseases affecting the pediatric population and a potential cause of irreversible neurocognitive deficits if left untreated or treated after delay [1-5]. The implementation of neonatal screening programs has led to early diagnosis and prompt initiation of levothyroxine (LT4) replacement therapy, minimizing the risk of adverse neurodevelopmental outcomes [5,6]. It is estimated that 1:2,000 to 1:4,000 of neonates are born with CH [7], and the incidence of CH has increased progressively over recent decades. This increase might reflect the increased survival of preterm neonates and recent reductions in the cutoff values used in screening programs, such that screening now captures more subtle abnormalities of thyroid function, including transient hypothyroidism in patients with eutopic thyroid gland [2,4,8,9].

CH is classified as permanent or transient. Permanent CH (PCH) is usually the result of dysgenesis of the thyroid gland or, less commonly, thyroid dyshormogenesis and requires lifelong replacement therapy [2,4,6,10]. Although current European Society of Pediatric Endocrinology (ESPE) guidelines recommend thyroid imaging, preferably with a combination of modalities including ultrasound and scintigraphy, to diagnose thyroid dysgenesis in CH cases, treatment should not be delayed for imaging as it does not alter the initial approach [11].

Transient CH (TCH) is attributed to multiple causes, including transplacental transfer of maternal antithyroid antibodies or antithyroid drugs, immaturity of thyroid iodine organification, excessive or inadequate iodine, or mutations of the DUOX2 gene [6,10,12]. Recognition of transient thyroid hormone deficiency is not feasible in the neonatal period, and the therapeutic approach is essentially the same with that of PCH (i.e., prompt initiation of LT4 treatment) [13].

To facilitate differential diagnosis of PCH and TCH, and ensure timely discontinuation of therapy in TCH, a trial of LT4 withdrawal and reassessment of thyroid hormones after 4−6 weeks has been employed in children with unexplained CH [2,3]. In clinical practice, discontinuation of therapy is often scheduled at 2 or 3 years of age to avoid possible adverse effects of treatment cessation on neurodevelopment [14]. However, TCH typically improves within the first few months after birth, and replacement therapy should be discontinued or avoided [15]. Delayed recognition of TCH leads to frequent laboratory testing, increased family anxiety, and potentially excessive exposure to thyroid hormone, which has been associated with diverse neurocognitive outcomes [3,5,8]. To prevent unnecessary long-term treatment, the most recent ESPE consensus guidelines for CH recommend evaluation of thyroid hormones following phasing out or withdrawal of LT4 at as early as 6 months of age in infants with eutopic thyroid gland and thyroxine requirements <3 μg/kg/day [11].

Several studies have investigated potential predictors that could facilitate earlier reassessment in patients with a high probability of TCH before the age of 3 and avoid unnecessary prolonged treatment. These predictors include low birth weight, prematurity, male sex, negative family history of CH, morbidities necessitating intensive care unit admission, lower serum thyroid stimulating hormone (TSH) concentrations at the time of diagnosis, and lower LT4 supplementation requirements [1-5,11,16-20]. However, the ideal predictive factor or factors that can reliably distinguish between the 2 forms of CH remain to be determined.

In this cohort study, we assessed the performance of multiple potential discriminative predictors of PCH and TCH that we considered likely to be useful in clinical practice to individualize treatment plans and facilitate early cessation of replacement therapy.

Materials and methods

The registry of all infants with CH diagnosed and followed between 2018 and 2023 at the Pediatric Endocrinology Unit of the University Hospital of Ioannina, Greece, was retrospectively surveyed. This Endocrinology Unit covers all pediatric endocrinology cases in a well-defined area in Northwestern Greece. Of a total of 112 medical files, 88 were chosen according to the following inclusion criteria: preterm or term infants that were identified by abnormal newborn screening (NBS) or that were referred to our unit and underwent blood tests due to prematurity, maternal thyroid disease, prolonged jaundice, etc., within the first 2 months of life; confirmation of CH by blood tests that were performed in the hospital lab; LT4 treatment since diagnosis and for the next 3 years; and regular follow-up according to the relevant guidelines for at least for 3 years of treatment and 6 months after treatment cessation. Exclusion criteria were central hypothyroidism (n=1), thyroid aplasia (n=4), genetic or chromosomal abnormalities such as trisomy 21 (n=2), current use of medications known to affect thyroid function (n=1), and inadequate data or that were lost to followup (n=16). In addition, no twins or triplets were included in the study. The study was conducted in accordance with the Declaration of Helsinki and was reviewed and approved by the Institutional Scientific Review Board of Ioannina University Hospital (approval number: 409). Written informed consent was obtained from the parents of the participants.

Each medical file was reviewed, and the following data were identified and recorded: course of pregnancy and complications; thyroid status of the mother and need for relevant treatment during pregnancy; and perinatal history (Apgar score, perinatal complications, breathing difficulties, neonatal hypoglycemia, jaundice, feeding difficulties, weight gain, and medications affecting thyroid function).

The routine NBS procedure in Greece is to measure TSH concentrations in Guthrie card blood samples collected by heel prick between the 2nd and 3rd days of life. For preterm infants, additional blood samples are examined beginning on the 5th–7th days of life, as well as on the 14th day and before discharge. All TSH analyses are performed in a single national laboratory (National Institute of Child Health) by chemiluminescence using an Architect i1000-Abbott biochemical analyzer. When whole-blood TSH concentrations are between 7 and 14 μIU/mL, which correspond to approximately 14−28 μIU/mL in serum, the results are confirmed in a second sample. If TSH concentrations in the second sample are also elevated, a blood test for TSH and free thyroxine (fT4) is requested. If TSH in whole blood is >14 μIU/mL at any time, parents and referring doctors are informed to proceed immediately with blood tests and LT4 substitution therapy upon confirmation. Such confirmatory tests (TSH and fT4) were performed in our hospital’s lab by immunoassay (Immulite 2000, Diagnostic Products Corp., and ADVIA Centaur, Bayer Healthcare LLC, respectively). LT4 treatment (5−10 or 10−15 μg/kg/day in mild or severe cases, respectively) was immediately initiated upon confirmation of abnormal results [11]. Thyroid ultrasonography was performed during the first 6 months of life in 30 of 92 children, and no child underwent scintigraphy analysis.

Records of infants that had been referred to our unit within their first 2 months of life due to prematurity, maternal thyroid disease, prolonged jaundice, etc., were included in the study if they had abnormal TSH results according to our lab’s upper normal limit (i.e., patients with TSH >10 μIU/mL after the first week and TSH > 6 μIU/mL after the first 4 weeks of life) and who had been started on LT4 substitution therapy.

All infants that were started on LT4 substitution therapy were followed intensely as follows: every 2 weeks until TSH levels normalized and every 1−2 months thereafter for the first year, every 4 months during the second year, and every 6 months during the third year of life. Surveillance visits consisted of clinical (anthropometric and developmental parameters) and laboratory (TSH, fT4) evaluation, according to which the LT4 dose was adjusted. Additional TSH and fT4 assessments were performed 4–6 weeks after every dose modification.

Discontinuation of LT4 therapy had occurred in all the study children after their third birthday. Their CH was considered transient if, after 6 months of LT4 discontinuation, serum TSH concentrations had not increased above 8 μIU/mL, measured at least twice during this 6-month follow-up period, because of which LT4 was not reinstituted. Alternately, CH was deemed permanent if reinitiation of LT4 had been necessary due to an increase of serum TSH concentrations to >8 μIU/mL at any time during follow-up or if the child developed symptoms of hypothyroidism with a serum TSH 6−8 μIU/mL.

Differences in biochemical parameters between the 2 groups of children were evaluated using either Student t-test or the Mann-Whitney U-test depending on the normality of the distribution of each parameter. When no parametric test was used, the results are presented as median and interquartile range; otherwise, values are expressed as mean±standard deviation. To eliminate the effect of multiple comparisons, the Benjamini-Hochberg procedure was utilized. Simple and multiple regression analyses (logistic regression) were performed to identify relationships among the parameters and to define independent relationships of thyroid status, either PCH or TCH, with parameters such as sex, gestational age, maternal thyroid disease, LT4 dosage, serum TSH concentration at diagnosis, total LT4 dose >50 μg/day, TSH >5 at any time during the treatment, and increase in LT4 dose more than 2 times over the treatment period. All parameters were tested with simple regressions to check for collinearity before entering multiple regression models. A level of significance of P<0.05 was set. It was estimated that a sample size of 88 would be sufficient to demonstrate a 20% difference in all the parameters examined, with a power >80% at a significance level of 0.05 [21]. All statistical analyses were performed using the Stat View software package of SAS Institute Inc.

Results

A total of 88 children was enrolled in the study. Based on the criteria described above, 35 of the 88 children (39.8%) included in the study were diagnosed with PCH and 53 (60.2%) with TCH. Patients diagnosed with TCH had significantly younger gestational age and birth weight than patients with PCH (35±3 weeks vs. 37±2 weeks, P=0.013 and 2,217±783 g vs. 2,635±670 g, P=0.004). There was no significant difference between the 2 groups in terms of neonates with birth weight deviations. Sex ratios did not differ significantly between the 2 groups; however, among females, TCH was diagnosed more commonly than PCH. History of maternal thyroid disease of any type was significantly more common in the TCH group (36% vs. 17%, P=0.003) (Table 1).

Table 1.

Clinical and laboratory parameters of children with permanent or transient congenital hypothyroidism

Doses of LT4 were invariably higher in the PCH group but declined from birth to 3 years in both groups (Fig. 1). The cutoff values of LT4 dose for PCH prediction for the various ages were those with maximum sensitivity and specificity. Fig. 2 depicts the receiver operating characteristic (ROC) curves for the LT4 dose at each study period, as used to discriminate between PCH and TCH.

Fig. 1.

Box and Plots for dose of levothyroxine (LT4) (μg/kg/day) on the vertical axis in the 2 study groups of permanent congenital hypothyroidism (PCH, red boxes) and transient congenital hypothyroidism (TCH, blue boxes).

Fig. 2.

Receiver operating characteristics (ROC) curves of each levothyroxine (LT4) dose (range, 1–7 μg/kg/day) at 6 months, 12 months, 1 year, and 2 years for discrimination between permanent congenital hypothyroidism (PCH) and transient congenital hypothyroidism (TCH). The optimal LT4 dose for discrimination at each time point was 3.8 μg/kg/day at 6 months (62% sensitivity, 96% specificity), 3.0 μg/kg/day at 12 months (64% sensitivity, 97% specificity), 2.6 μg/kg/day at 2 years (80% sensitivity, 98% specificity), and 2.5 μg/kg/day at 3 years (89% sensitivity, 98% specificity).

Optimal LT4 doses at each time point were 3.8 μg/kg/day at 6 months (62% sensitivity, 96% specificity), 3.0 μg/kg/day at 12 months (64% sensitivity, 97% specificity), 2.6 μg/kg/day at 2 years (80% sensitivity, 98% specificity), and 2.5 μg/kg/day at 3 years (89% sensitivity, 98% specificity).

Although most of the selected cutoff values had excellent specificities, the sensitivities were much lower. At 3 years of age, a cutoff of 2.5 μg/kg/day seemed to differentiate the 2 conditions in the study population due to its high sensitivity and specificity. Based on the ROC curves above and the cutoff values provided for each age group, clinicians can confidently predict PCH in cases that meet these criteria.

An absolute LT4 dose >50 μg/day was required for 28% of children in the PCH group, while none of the children in the TCH group required a similar dose. This finding implied that the need for an LT4 >50 μg/day at any age had a 100% specificity and strong positive predictive value (PPV) for PCH in our study cohort.

Laboratory TSH results at diagnosis were also a robust predictor of PCH. A serum TSH concentration of 13.3 μIU/mL was found to discriminate between PCH and TCH with a sensitivity of 72% and a specificity of 96%. Sensitivity increases, but specificity decreases at a cutoff of 10 μIU/mL.

Detection of TSH >5 μIU/mL at any time after treatment initiation was also a strong predictor of PCH, with a specificity of 96.2% and a relatively high sensitivity of 77%.

Children in the PCH group required increased LT4 doses significantly more often during the treatment period than the TCH group (5.0±1.5 times vs. 1.8±1.2 times, P<0.001) (Fig. 3). The need to increase the LT4 dose more than twice during follow-up was predictive of PCH with a sensitivity of 97.4% and a specificity of 71.6%. The results of this study suggest that combining this parameter with another that shows excellent specificity (i.e., serum TSH at diagnosis>13.3 μIU/mL, LT4 doses >50 μg/day or TSH>5 μIU/mL at any time, or LT4 doses above the higher cutoff at any age) could predict PCH with reasonable certainty in clinical practice.

Fig. 3.

The number of times that an increase in the dose of levothyroxine (LT4) was required after the initial dose for the 2 study groups with permanent congenital hypothyroidism (PCH) and transient congenital hypothyroidism (TCH).

Multiple regression analyses were performed, with thyroid status (PCH or TCH) as the dependent variable and the LT4 dose at each study period as the independent variable. Sex, gestational age, maternal thyroid disease, and serum TSH concentrations at diagnosis were also included in the same regression as additional independent variables or confounders. The serum TSH concentration at the time of diagnosis was a strong independent predictor of thyroid status (PCH or TCH; beta=-4.3; odds ratio [OR], 0.61; 95% confidence interval [CI], 0.49−0.77; P<0.001) after adjusting for sex, gestational age, and maternal thyroid disease. In the same model, female sex was independently correlated with TCH (beta=1.99; OR, 4; 95% CI, 1.02−15; P=0.04).

At the age of 6 months, the strongest predictor of thyroid status (PCH or TCH) remained the serum TSH concentration at diagnosis (beta=-2.9; OR, 0.004; 95% CI, 0.001−0.18, P=0.004), while the dose of LT4 at 6 months (beta=-2.9; OR, 0.004; 95% CI, 0.0001−0.18; P=0.004) and sex (beta=2.33; P=0.02) were also independent predictors.

At 1 and 2 years of age, the LT4 dose was the strongest predictor of disease, but TSH concentrations at the time of diagnosis remained an independent predictor. No significant correlations were found at 3 years of age.

A second set of multiple regression analyses was performed using most of the same variables but replacing LT4 dose with the parameters 'total LT4>50,' 'TSH>5,' and 'number of times the LT4 dose was increased.' These 3 parameters and the LT4 dose are interrelated, so they were entered into the model separately to avoid collinearity. Among the above-noted parameters, 'TSH>5' (beta=-3.6; OR, 0.025; 95% CI, 0.03-0.18, P<0.001) and ‘number of times the LT4 dose was increased’ (beta=-3.2; OR, 0.15; 95% CI, 0.06−0.48; P<0.001) were independent predictors of thyroid status (Table 2).

Table 2.

Results of multiple regression analysis

Regarding thyroid imaging studies, of the 30 children that were examined by thyroid ultrasonography, 4 were diagnosed with thyroid agenesis and one with thyroid hypoplasia. All 5 children were ultimately diagnosed with PCH. No ectopic thyroid glands were identified on ultrasound.

Discussion

In this study cohort, over half of the infants with CH were diagnosed with TCH. The high proportion of transient cases underscores the need to identify reliable predictors to enable early discontinuation of replacement therapy. The prevalence of TCH was higher in this cohort compared with previously published studies that included thyroid dysgenesis cases (12%−36.5%; references [3,4,6]). However, these studies included only neonates diagnosed with CH by neonatal screening test and those born mostly at term. Studies including exclusively patients with CH and eutopic thyroid gland reported a higher proportion of TCH (39.4%−65%; references [8,18,22]). The current cohort includes infants diagnosed based on abnormal NBS and those who underwent testing due to risk factors such as prematurity, maternal thyroid disease, or medication. These factors are associated with TCH and may have contributed to its high prevalence in the study population [9,15,23] given that half of the neonates included in the study were preterm, neonates in the TCH group had a lower gestational age and birth weight, and maternal thyroid disease was more common. These findings highlight the need to discriminate between the 2 conditions as early as possible.

The neonatal screening CH test cutoff in Greece is currently set at TSH >7 μIU/mL in whole blood, which is equivalent to >14 μIU/mL in serum. However, the cohort also included neonates referred to the pediatric endocrine clinic as a high-risk population and diagnosed with CH by abnormal-for-age serum TSH concentrations measured in a laboratory. Therefore, neonates with TSH concentrations below the neonatal screening test cutoff were also included.

This study identified the serum TSH concentration at initial diagnosis as an independent predictor of the differentiation between TCH and PCH. The mean TSH concentration was significantly higher in the PCH group, consistent with previous reports [2,18,19,24]; however, other authors have reported that discrimination based on TSH concentrations is unreliable, as overlap between the 2 groups is often observed [3,4,25]. It should be noted these studies excluded patients with thyroid dysgenesis, which is associated with higher serum TSH concentrations [3,4]. Messina et al. [4] reported that TSH at diagnosis was not significantly different between children with a eutopic thyroid gland and TCH or PCH; however, when comparing the subgroup of patients with an ectopic gland to those with a eutopic gland, the difference was significant irrespective of TCH or PCH diagnosis.

According to the results of our study, a cutoff serum TSH concentration at diagnosis<than 13.3 μIU/mL could predict TCH; this is much lower than the previously proposed cutoff (38.4−75 μIU/mL) [3,19,22]. We hypothesized that this discrepancy arises from differences in the study population: our study included infants with negative neonatal screening tests for CH diagnosed through laboratory testing based on risk factors, which led to lower median serum TSH concentrations compared to those in other studies.

The LT4 doses (μg/kg/day) in our study cohort were consistently larger in the PCH group, but the LT4 dose in both the PCH and TCH groups declined from birth to 3 years, similar to the findings of previous studies [2,3,5,6,18]. The LT4 dose was recognized as an independent predictor for discrimination between TCH and PCH at 6 months and at 1 and 2 years of age. The optimal LT4 dosage at different ages that could assist in the differential diagnosis was assessed. The results of this study indicate the clinicians can accurately predict PCH in patients whose LT4 requirements exceed the following cutoffs: 3.8 μg/kg/day at 6 months, 3.0 μg/kg/day at 12 months, 2.6 μg/kg/day at 2 years, and 2.5 μg/kg/day at 3 years, which are consistent with the cutoffs reported in previous studies [3,22]. Cho et al. [22], in their retrospective study, proposed that a cutoff of 3.25 μg/kg/day at 1 and 2 years of age could predict PCH in patients with eutopic glands. Messina et al. [4] reported that LT4 doses above 4.9, 4.27, and 4.70 μg/kg/day could predict PCH with the highest specificity (100%), at 12, 24, and 36 months, respectively, while the LT4 doses with the highest sensitivity (100%) were 1.70, 1.45 and 0.98 μg/kg/day. Another retrospective study suggested a cutoff of 2.2 μg/kg/day after the sixth month of age to predict PCH in patients with eutopic thyroid. In the same study, the authors reported that LT4 doses > 4.6 μg/kg/day can differentiate cases with agenesis or ectopic gland from those with eutopic gland after the sixth month of age [6]. The variability in the reported cutoff values for differentiating PCH from TCH in the relevant studies may be due to genetic, demographic, or other differences in the study populations, as well as different causes of PCH leading to thyroid insufficiency of various severity requiring differing LT4 doses.

Previous studies have reported difficulty in managing hypothyroidism in patients with PCH, as evidenced by the elevation of TSH levels above the reference interval during treatment and the subsequent need to increase the LT4 dose [3,4,12,26,27]. During the follow-up treatment period, up to 80% of patients with PCH had serum TSH concentrations greater than 5 μIU/mL, while only 3.8% of patients with TCH experienced such levels. Marr et al. [3] reported TSH levels above the reference levels in 91.6% and 18.3% of treated patients with PCH and TCH aged 6 months or older, respectively. Moreover, the need for escalating LT4 doses at any time during the treatment course has been associated with PCH [3,4,9,26], and is supported by the findings of the present study, as patients with PCH more frequently required increased LT4 doses compared to TCH patients.

The absolute daily dose of LT4 was recognized as a parameter with a substantial PPV for diagnosing PCH. During the followup period, 28% of PCH patients required an LT4 dose greater than 50 μg/day at some point. In contrast, none of the TCH patients required a dose that high. This practical threshold has also been identified in previous studies [3,27,28]. For example, in their retrospective study, Marr et al. [3] found that LT4 doses greater than 50 μg/day were associated with a PCH diagnosis in almost all cases.

In summary, predictors of PCH with high specificity that were identified in the current study included age-dependent LT4 dose cutoffs, an absolute dose of LT4 greater than 50 μg/day, serum TSH higher than 13.3 μIU/mL at diagnosis, and TSH higher than 5 μIU/mL at any time during treatment. In addition, the need to increase the LT4 dose more than twice during follow-up was recognized as a predictor with high sensitivity for diagnosing PCH. Thus, in clinical practice, according to the results of the current study, PCH can be confidently diagnosed in patients who require more than 2 increases in LT4 dose combined either with an LT4 dose above the age-specific cutoff value, an absolute LT4 dose greater than 50 μg/day, a serum TSH greater than 13.3 μIU/mL at diagnosis, or a TSH higher than 5 μIU/mL at any point during follow-up. Conversely, the lack of these characteristics supports the diagnosis of TCH.

The current study identified pre-treatment serum TSH concentration as the strongest independent predictor for discrimination of PCH and TCH at the time of diagnosis and at 6 months of age. Moreover, LT4 dose was an independent prognostic factor at 6 months of age and the most robust predictor at the ages of 1 and 2 years. In addition, the serum TSH concentration at diagnosis remained a significant predictor for discrimination between PCH and TCH at that age. Additionally, at any time during follow-up, serum TSH concentrations above 5 μIU/mL and the number of times the LT4 dose needed to be increased were identified as independent predictors of a PCH diagnosis.

The main limitations of the study were its retrospective nature and the relatively small sample size. The study was conducted in a single center, treating patients from a well-defined geographic region. Therefore, it is difficult to generalize the results to other populations. In addition, due to follow-up only until 3 years of age, the incidence of TCH may have been underestimated. However, it was a clinical study that includes patients referred to our pediatric endocrine clinic for further testing due to a variety of predisposing factors for CH. Thus, the children enrolled in this study comprised a representative population of infants with hypothyroidism within a defined geographical area, who were referred to our clinic, and who were treated by a single pediatric endocrinologist. Moreover, all patients were treated in the same clinic by a single experienced consultant pediatric endocrinologist, avoiding variations in clinical practice and ensuring consistent management and follow-up.

In conclusion, in the present study, various predictors for discrimination between PCH and TCH were recognized, including serum TSH concentration at diagnosis, LT4 dose, frequency of increased LT4 dose, and TSH concentrations during treatment. Applying these criteria in clinical practice could facilitate timely treatment withdrawal in cases with a strong suspicion of TCH. This treatment issue has become of particular importance in recent years as reductions in screening cutoff values and the birth of more numerous premature infants with immature thyroid function have led to increased numbers of TCH cases. The present study could be a useful addition to the few other studies that have investigated potential factors to differentiate PCH from TCH under different or similar clinical settings. A future systematic review may provide conclusive and universal evidence and contribute to the development of new guidelines for the management and follow-up of patients with CH.

Article information Continued