Role of thyrotropin‐releasing hormone test in re‐evaluation of congenital hypothyroidism

Abstract

The thyrotropin‐releasing hormone (TRH) test is useful for differentiating central and primary hypothyroidism, and is also valuable for diagnosing hypothyroidism. The threshold of the TRH test is usually set at 10–40 mIU/L. However, some experts are of the opinion that the TRH test has a limited role in evaluating hypothyroidism because of the clinical application of the new‐generation thyroid‐stimulating hormone (TSH) assay. We reviewed a case series to analyze the clinical use of the TRH test in the re‐evaluation of congenital hypothyroidism. In total, data on 228 children with eutopic thyroid glands and neonatal hyperthyrotropinemia under levothyroxine replacement were collected. Basal TSH levels were measured and the TRH test was performed at the age of 3 years for re‐evaluation of congenital hypothyroidism, and statistical analysis was performed. All of the patients were followed up to avoid over‐ or under‐treatment. At the age of 3 years, 31.6% of the patients still had hypothyroidism. There was no significant difference between basal TSH level and TRH test in the diagnosis of hypothyroidism (p = 0.23). The negative predictive value of the basal TSH level was 100%, however, the positive predictive value was only 43.6%. When the TSH level was near the upper limit of the normal range (4.5–8.5 mIU/L), the TRH test result had a better correlation with hypothyroidism than the basal TSH level (p = 0.03). The threshold of the TRH test set at 60 mIU/L had the greatest area under the curve, with a negative predictive value of 95.2% and a positive predictive value of 80.2%. Neonatal hyperthyrotropinemia was a risk factor for hypothyroidism. We suggest that the TRH test should be administered in children with a basal TSH value near the upper limit of the normal range, and the threshold of the TRH test should be set at 60 mIU/L.

Introduction

Newborn screening is now widespread, therefore, an increasing number of cases of congenital hypothyroidism (CH) are being reported, and it has become a leading cause of preventable mental retardation, with an incidence worldwide of around 1 in 3000–4000 [1], and 1 in 2000 in Taiwan [2]. Newborn screening is usually completed 24 hours after the newborns start to feed. Newborns are screened for CH with enzyme immunoassays of dried blood spots, and those with hyperthyrotropinemia are referred for further evaluation when they are 1–2 weeks old. After confirmatory tests, those with hypothyroidism are then started on levothyroxine replacements (10–15 μg/kg/day) immediately and receive thyroid scintigraphy or sonography for evaluation of thyroid development. Children with thyroid agenesis or ectopia usually need permanent levothyroxine treatment and are diagnosed as having permanent CH. Those with eutopic thyroid glands may only have transient neonatal hyperthyrotropinemia and are re‐evaluated at 3 years of age to decide whether hypothyroidism persists [3]. It has been suggested that hypothyroidism is an ongoing disease state starting with a positive thyrotropin‐releasing hormone (TRH) test, followed by TSH elevation, and subsequently to overt hypothyroidism and thyroid hormone failure [[4], [5], [6]]. Re‐evaluation mainly involves the levels of TSH, triiodothyronine (T3), thyroxine (T4), and free thyroxine (fT4), and the TRH test is also included in the re‐evaluation of CH [1].

The TRH test is also used to detect subclinical hypothyroidism in patients with goiter or depression or in women with infertility [[7], [8], [9]]. However, many experts regard that it now has a limited role after the clinical application of the new generation of TSH assays [[10], [11], [12]], and the TRH (protirelin) reagent is even unavailable in the United States currently [13]. We present a retrospective case analysis to determine whether the TSH assay alone is sufficient or whether the TRH test still has a role in the diagnosis of hypothyroidism.

Materials and methods

Patient selection

Term newborns in Southern Taiwan suspected of having CH after newborn screening were referred to Kaohsiung Medical University Hospital, Kaohsiung, Taiwan. TSH was measured with immunoradiometric assay (functional sensitivity: 0.03 mIU/L, intra‐assay variation < 5.7%, inter‐assay variation < 5.4%) and T3, T4, and fT4 were evaluated with radioimmunoassays (functional sensitivity: T3 10 ng/dL, T4 0.25 μg/dL, and fT4 0.05 ng/dL; intra‐assay variation: T3 < 12%, T4 < 5.6%, and fT4 < 8.15%; and inter‐assay variation: T3 < 10.1%, T4 < 14.3%, and fT4 < 14.9%). If the patient had hyperthyrotropinemia (TSH level > 20 mIU/L) and/or hypothyroxinemia (T4 level < 15 μg/dL), CH was diagnosed. According to thyroid scintigraphy (Tc 99m pertechnetate), we collected 228 cases with eutopic thyroid glands (including normal, goiter, and hypofunction) and 49 cases with thyroid agenesis or ectopia between 1989 and 2008. This study was approved by the Institutional Review Board of Kaohsiung Medical University Hospital (Number KMUHIRB‐20130070).

TRH test

The patients with eutopic thyroid glands received TRH tests once they reached the age of 3 years after stopping levothyroxine for 4 weeks. Thyroid hormones (fT4, T4, and T3, via the radioimmunoassay method) and TSH levels (immunoradiometric assay method) were measured immediately prior to when intravenous TRH (7 μg/kg) was administered, and the TSH levels were assessed again 30 minutes, 60 minutes, and 90 minutes later.

The patients with exaggerated peak TSH values (judged as > 40 mIU/L in this study) were regarded as being hypothyroiditic and kept receiving treatment and regular follow‐up visits. The other patients were classified as being euthyroiditic, and levothyroxine treatment was ceased. Serum thyroid function was assessed at least twice 1 month later with competitive enzyme‐linked immunoassay (detection limit: TSH: 0.01–100 mIU/L, T4: 0.5–30 μg/dL). If hyperthyrotropinemia (TSH level > 4.5 mIU/L) in combination with hypothyroxinemia (T4 < 11 μg/dL) developed, the patients were also regarded as having permanent CH and levothyroxine therapy was resumed. After 2 years follow‐up, those who were still receiving levothyroxine were classified as having permanent CH, and the other patients as having transient CH.

Statistical analysis

The results were analyzed by t tests and one‐way analysis of variance for comparisons between groups. SPSS statistical software version 17.0 (SPSS Inc., Chicago, IL, USA) was used for all statistical analysis. In one‐way analysis of variance, if the Levene test revealed homogenicity, then Scheffe's method was used for post‐hoc comparisons. However, if the Levene test revealed significant differences in homogenicity, then the Games–Howell method was used for post‐hoc comparisons. Continuous variables were summarized with means and standard deviations. A p value < 0.05 was considered to be statistically significant. Different cut‐off points of the TRH test were compared using analyses of receiver operating characteristic curves.

Results

Thyroid function in the newborn period

The TSH levels of the initial confirmatory test were significantly higher in the thyroid agenesis group (n = 19) than in the eutopic group (n = 228; mean ± standard deviation, 262.43 ± 200.42 mIU/L vs. 120.55 ± 76.66 mIU/L, p = 0.017), and the T3, T4, and fT4 levels were lower in the thyroid agenesis group (Table 1). However, all of them were poor predictors for permanent CH (Fig. 1).

Table 1.

Initial confirmatory test between those with eutopic thyroid glands, agenesis, and ectopia.

Scan	No.	TSH (mIU/L) a	T3 (ng/dL) b	T4 (μg/dL) c	fT4 (ng/dL) d
Eutopia	228	120.55 ± 76.66	130.19 ± 48.21	4.74 ± 3.00	0.61 ± 0.47
Agenesis	19	262.43 ± 200.42	60.71 ± 40.54	2.82 ± 2.65	0.27 ± 0.07
Ectopia	30	173.52 ± 134.72	124.46 ± 40.45	5.25 ± 3.03	0.59 ± 0.55

TSH = thyroid‐stimulating hormone; T3 = triiodothyronine; T4 = thyroxine; fT4 = free thyroxine.

^aeutopia < agenesis (p = 0.017).

^beutopia > agenesis (p < 0.001).

^ceutopia > agenesis (p = 0.028), ectopia > agenesis (p = 0.022).

^deutopia > agenesis (p < 0.001).

Figure 1.

Receiver operating characteristic curve: initial confirmatory test versus permanent hypothyroidism (n = 280). The area‐under‐the‐curve (AUC) of thyroid‐stimulating hormone (TSH): 0.60 (95% CI, 0.51 to 0.69, p = 0.024). AUC of triiodothyronine (T3): 0.32 (95% CI, 0.24 to 0.41, p < 0.001). AUC of thyroxine (T4): 0.43 (95% CI, 0.34 to 0.52, p = 0.120). AUC of free‐T4 (fT4): 0.43 (95% CI, 0.34 to 0.52, p = 0.122).

Thyroid function in the eutopic thyroid group

The incidence of permanent CH was 31.6%, and there was no significant difference between the male and female infants (30.7% vs. 33.0%, n = 228, p = 0.807; Table 2). None of the three eutopic thyroid conditions (normal, hypofunction, and hyperfunction) were correlated to the final outcome (p = 0.468). The positive predictive rate (PPV) of basal TSH (threshold: 4.5 mIU/L) was 43.6%, and the negative predictive rate (NPV) was 100%. The PPV of peak TSH (threshold: 40 mIU/L) was 53.3%, and the NPV was 100%.

Table 2.

Incidence of transient HT and permanent CH according to sex, basal TSH, and peak TSH levels.

Variables		No.	Transient HT (%)	Permanent CH (%)	p
Male		137	95 (69.3)	42 (30.7)	0.807
Female		91	61 (67.0)	30 (33.0)
Thyroid scan	Normal	163	115 (70.6)	48 (29.4)	0.468
	Hyperfunction	49	30 (61.2)	19 (38.8)
	Hypofunction	16	11 (68.8)	5 (31.3)
Basal TSH	≤4.5 mIU/L	63	63 (100)	0 (0)	<0.001
	>4.5 mIU/L	165	93 (56.4)	72 (43.6)
Peak TSH	≤40 mIU/L	93	93 (100)	0 (0)	<0.001
	>40 mIU/L	135	63 (46.7)	72 (53.3)
	≤60 mIU/L	147	140 (95.2)	7 (4.8)	<0.001
	>60 mIU/L	81	16 (19.8)	65 (80.2)
Total		228	156 (68.4)	72 (31.6)

CH = congenital hypothyroidism; HT = hyperthyrotropinemia; TSH = thyroid‐stimulating hormone.

The distribution of transient hyperthyrotropinemia and permanent CH according to basal TSH and peak TSH values is shown in Fig. 2. The receiver operating characteristic curves for the basal and peak TSH levels of TRH test for the prediction of permanent CH are shown in Fig. 3A. The areas under the curve (AUC) of the basal and peak TSH levels was 0.950 versus 0.969, respectively (p = 0.23). For patients with a basal TSH level >4.5 mIU/L, the difference in AUC between the basal and peak TSH levels was significant (Fig. 3B; 0.917 vs. 0.948, respectively, n = 165, p = 0.23). Focusing on the patients with a basal TSH level between 4.5 mIU/L and 8.5 mIU/L, the difference in AUC between the basal and peak TSH levels was significant (Fig. 3C; 0.570 vs. 0.909, respectively, n = 76, p = 0.03).

Figure 2.

Distribution of transient hyperthyrotropinemia (HT) and permanent congenital hypothyroidism (CH) according to basal thyroid‐stimulating hormone (TSH) and peak TSH of thyrotropin‐releasing hormone (TRH) test in patients with thyroid eutopia (n = 228).

Figure 3.

(A) Receiver operating characteristic (ROC) curve: basal thyroid‐stimulating hormone (TSH) and peak TSH versus permanent hypothyroidism for patients with thyroid eutopia (n = 228). The area‐under‐the‐curve (AUC) of basal TSH: 0.950 (95% CI: 0.925 to 0.976, p < 0.001). AUC of peak TSH: 0.969 (95% CI: 0.951 to 0.987, p < 0.001). AUC of basal TSH & peak TSH: t = 1.44, p = 0.23. (B) ROC curve: basal TSH and peak TSH versus permanent hypothyroidism for patients with thyroid eutopia and basal TSH level >4.5 mIU/L(n = 165). The AUC of basal TSH: 0.917 (95% CI: 0.875 to 0.959, p < 0.001). AUC of peak TSH: 0.948 (95% CI: 0.918 to 0.978, p < 0.001). AUC of basal TSH & peak TSH: t = 1.44, p = 0.23. (C) ROC curve: basal TSH and peak TSH versus permanent hypothyroidism for patients with thyroid eutopia and basal TSH level 4.5∼8.5 mIU/L (n = 51). The AUC of basal TSH: 0.570 (95% CI: 0.28 to 0.86, p = 0.590). AUC of peak TSH: 0.909 (95% CI: 0.80 to 1.000, p = 0.003). AUC of basal & peak TSH: t = 4.522, p = 0.033.

Threshold of the TRH test

The threshold of the TRH test was tested with the method of approach. A threshold between 50 mIU/L and 70 mIU/L had a higher AUC (Table 3), and a threshold of 60 mIU/L had the highest AUC (Table 4). The NPV of peak TSH (threshold: 60 mIU/L) was 95.2% and the PPV 80.2% (Table 2).

Table 3.

AUC of receiver operating characteristic curves: TRH test peak of 40–80 mIU/L versus permanent hypothyroidism for patients with eutopic thyroid glands (n = 228).

Variables	AUC	Standard deviation	p	95% confidence interval
				Lower limit	Upper limit
Peak 40	0.798	0.028	0.000	0.743	0.853
Peak 50	0.858	0.025	0.000	0.809	0.907
Peak 60	0.900	0.025	0.000	0.852	0.948
Peak 70	0.874	0.030	0.000	0.816	0.932
Peak 80	0.855	0.032	0.000	0.792	0.919

AUC = area under the curve; TRH = thyrotropin‐releasing hormone.

Table 4.

AUC of receiver operating characteristic curves: TRH test peak of 55–65 mIU/L versus permanent hypothyroidism for patients with eutopic thyroid glands (n = 228).

Variables	AUC	Standard deviation	p	95% confidence interval
				Lower limit	Upper limit
Peak 55	0.886	0.024	0.000	0.839	0.933
Peak 56	0.889	0.024	0.000	0.842	0.935
Peak 57	0.895	0.024	0.000	0.848	0.942
Peak 58	0.894	0.024	0.000	0.846	0.942
Peak 59	0.894	0.024	0.000	0.846	0.942
Peak 60	0.900	0.025	0.000	0.852	0.948
Peak 61	0.893	0.026	0.000	0.843	0.943
Peak 62	0.893	0.026	0.000	0.843	0.943
Peak 63	0.889	0.026	0.000	0.838	0.941
Peak 64	0.893	0.026	0.000	0.841	0.944
Peak 65	0.886	0.027	0.000	0.833	0.939

AUC = area under the curve; TRH = thyrotropin‐releasing hormone.

Fig. 4 describes the distribution of transient HT and permanent CH according to basal TSH and peak TSH values in the patients with eutopic thyroid glands with a basal TSH level <20 mIU/L and peak TSH <100 mIU/L (n = 183).

Figure 4.

Distribution of transient hyperthyrotropinemia (HT) and permanent congenital hypothyroidism (CH) according to basal thyroid‐stimulating hormone (TSH) and peak TSH in patients with thyroid eutopia with basal TSH < 20 mIU/L and peak TSH < 100mIU/L (n = 183). The higher and lower transverse dashed lines mean peak TSH 60 and 40 mIU/L, respectively; the left and right longitudinal dashed lines mean basal TSH 4.5 and 8.5 mIU/L, respectively.

Discussion

Children with hyperthyrotropinemia in newborn screening are at an increased risk for the development of subclinical hypothyroidism in early childhood, with a reported prevalence rate up to 70% [14]. The current study excluded the patients with thyroid agenesis and ectopia, and around one‐third of those with eutopic thyroid glands needed lifelong levothyroxine treatment, which is consistent with a previous report [15].

In most situations, basal TSH level is a simple and accurate index for the diagnosis of primary hypothyroidism. All children with normal basal TSH values in this study had normal growth and cognitive function without levothyroxine treatment after the age of 3 years. Clinicians face a dilemma, however, when the basal TSH level is near the upper limit of the normal range. Subclinical hypothyroidism may be difficult to assess because it is usually asymptomatic and often fluctuates at the basal TSH level [[16], [17]]. In the current study, both the basal and peak TSH levels had an excellent NPV of 100% for hypothyroidism, however the PPV of basal TSH was only 43.6%, whereas that of the TRH test was slightly higher at 53.3%. These children were followed closely to ensure a normal thyroid function and prevent over‐diagnosis. The linear correlation of basal TSH and peak TSH was in direct proportion (Fig. 2). There were two slopes of the correlation between basal and peak TSH, that is, one below 60 mIU/L of basal TSH level and another above. It may have been due to dilution of the samples. They both had a good NPV, therefore, we assumed that if there was a difference in the prediction rate between basal and peak TSH level, it would be in the group with the lower basal TSH values, and we found that the greatest difference was in the group with a basal TSH level higher but near the upper limit of the normal range. The patients with a basal TSH level of 4.5–8.5 mIU/L had the greatest difference in AUC from peak TSH in the prediction of permanent hypothyroidism. This indicated that if a child had a basal TSH level higher but near the upper limit of normal range, further evaluation with the TRH test would help to raise the PPV.

Previous studies have suggested that peak TSH values above 10–40 mIU/L in the TRH test are exaggerated [[18], [19]]. However, we suggest that a threshold of 60 mIU/L has the greatest efficacy, because it raised the PPV from 53.3% to 80.2%, while only slightly decreasing the NPV from 100% to 95.2% compared with a threshold of 40 mIU/L (Fig. 4).

Considering their near‐normal thyroid function during a previous follow‐up, even with poor therapeutic compliance, six children initially presenting with thyroid agenesis on scintigraphy also participated in re‐evaluation when 3 years of age. All six were found to have functional thyroid glands when they were 3 years old and successfully discontinued therapy. Two of them had no recorded history of an abnormal birth and the other four were hospitalized after birth for different conditions including maternal syphilis, congenital heart disease, ileal atresia, with intestinal perforation combined with maternal drug abuse, and congenital diaphragmatic hernia with pneumothorax. They were all term neonates, thus, the initial thyroid isotope uptake may have been impaired via the mechanism of escape from the Wolff–Chaikoff effect after exposure to iodine antiseptics with medical intervention [20]. In addition, a eutopic thyroid gland is more likely to be detected by ¹²³I scintigraphy over the Tc99 m used in the current study [[21], [22]]. This means that for patients with thyroid agenesis, and especially those with concurrent medical problems, further confirmation with sonography is advised.

The incidence of thyroid hemiagenesis is estimated to be approximately 0.05% in normal children, with the majority being left lobe defects [23]. Three cases of thyroid hemiagenesis were included in our study (2 of the left lobe and 1 of the right lobe). There were no significant differences in the initial confirmatory tests and the TRH tests at 3 years of age between these six children and those with eutopic thyroid glands (p = 0.999, data not shown).

There were some limitations to this retrospective study. Lacking a double‐blind design, those with an exaggerated TSH response would preferably be regarded as having subclinical hypothyroidism and continue to receive levothyroxine treatment when they presented with equivocal results during follow‐up. Further investigations are still needed.

In conclusion, neonatal hyperthyrotropinemia is a risk factor for subclinical hypothyroidism. Of the patients with eutopic thyroid glands, approximately 30% needed permanent treatment. We suggest that the TRH test should be administered in children with a basal TSH value near the upper limit of the normal range to rule out subclinical hypothyroidism. A threshold of the TSH peak of 60 mIU/L has good efficacy for diagnosis.

Acknowledgments

The authors thank Dr. Liang‐Yeu Chen (E‐Da hospital) and Dr. Chiou‐Lian Lai (KMUH) for their abundantly helpful and invaluable assistance.