Primary Hypothyroidism in Childhood Cancer Survivors: Prevalence, Risk Factors and Long-term Consequences

Abstract

Background.

Data on primary hypothyroidism and its long-term impact on health, cognition and quality of life (QOL) of childhood cancer survivors are limited. This study examined the prevalence of and risk factors for primary hypothyroidism and its associations with physical, neurocognitive and psychosocial outcomes.

Methods.

This was a retrospective study with cross-sectional health outcomes analysis of an established cohort comprising 2,965 survivors of childhood cancer (52.8 % male; median current age 30.9 years, 22.3 years since cancer diagnosis). Multivariable logistic regression estimated odds ratios (ORs) and 95% confidence intervals (CIs) for associations between primary hypothyroidism and cancer-related risk factors, cardiovascular disease risk factors, frailty, neurocognitive and QOL outcomes, social attainment and subsequent thyroid carcinoma. Associations between serum free thyroxine and thyrotropin levels at assessment and health outcomes were explored.

Results.

The prevalence of primary hypothyroidism was 14.7% (95% CI 13.5%-16.0%). It was more likely in females (OR 1.06; 95% CI 1.03-1.08), less likely in non-whites (OR 0.96; 95% CI 0.93-0.99), associated with thyroid radiotherapy (higher risk at higher doses), and more common if cancer was diagnosed at ≥15.0 vs <5 years of age (OR 1.05; 95% CI 1.01-1.09). Primary hypothyroidism was associated with frailty (OR=1.54; 95% CI 1.05-2.26), dyslipidemia (OR=1.52; 95% CI 1.14-2.04), impaired physical QOL (OR=1.66; 95% CI 1.12-2.48) and having healthcare insurance (OR=1.51; 95% CI 1.07-2.12).

Conclusions.

Primary hypothyroidism is common in survivors and is associated with unfavorable physical health and QOL outcomes. The impact of thyroid hormone replacement practices on these outcomes should be investigated further.

Precis:

Primary hypothyroidism affected 14.7% of 2,965 survivors of childhood cancer and was associated with adverse health and quality of life. Associations between thyroid function tests and health outcomes were explored.

Introduction

Primary hypothyroidism, thyroid hypofunction due to a disorder of the thyroid gland rather than the hypothalamic-pituitary axis,¹ is among the most common endocrine complications experienced by childhood cancer survivors with a reported prevalence of 15-20%.² Previous reports have described the prevalence of and risk factors for primary hypothyroidism in survivor populations.^2-7 However, data are scarce describing associations of primary hypothyroidism with adverse outcomes known to disproportionately affect survivors such as cardiovascular disease, frailty, neurocognitive deficits, psychosocial issues, and subsequent thyroid carcinoma.

The objectives of this study were to describe the impact of primary hypothyroidism in a large population of clinically assessed long-term childhood cancer survivors and evaluate potential associations with adverse physical, neurocognitive and psychosocial outcomes, as well as subsequent thyroid carcinoma. In addition, we sought to explore associations between these outcomes and serum levels of free thyroxine (FT4), and thyrotropin (TSH) above, within and below the normal ranges to investigate how these may inform thyroid hormone replacement therapy practices in survivors who require treatment for primary hypothyroidism.^{1, 8, 9}

Patients and Methods

The St. Jude Lifetime Cohort Study (SJLIFE)

SJLIFE is an IRB approved retrospective cohort study with prospective ongoing clinical follow-up of survivors of a pediatric malignancy treated at St. Jude Children’s Research Hospital, as previously described.⁹ The present analysis was limited to SJLIFE participants who were ≥18 years old, survived cancer for at least 10 years and had completed at least one campus visit.

Cancer Treatment Variables

Thyroid radiation exposure included the following fields: head, neck, cervical spine, craniospinal, and total body. Treatment fields were reconstructed on a computational phantom scaled to age at radiotherapy. Mean doses to the left and right thyroid lobes were calculated and the higher of the two was used in the analysis. ^{10, 11} The cumulative dose of alkylating agents was expressed using the cyclophosphamide equivalent dose (CED).¹² The cumulative dose of glucocorticoids was quantified as prednisone equivalent dose received in mg (1 mg prednisone = 0.15 mg dexamethasone). ¹³

Diagnosis of Primary Hypothyroidism

Serum FT4 and TSH levels were measured using an electro chemiluminescent immunometric assay (Roche Cobas 6000). Manufacturer provided normal ranges for adults were 1.0-2.1 ng/dL for FT4 and 0.4-4.0 mIU/L for TSH. In the absence of thyroid hormone replacement therapy, primary hypothyroidism was defined by a serum TSH > 4.0 mIU/L. Individuals on thyroid hormone replacement for primary hypothyroidism based on medical chart documentation but for whom laboratory data prior to the initiation of replacement were unavailable, were considered to have the condition as medications were not stopped before the SJLIFE assessment. As thyroid function cannot be reliably assessed in the context of hypothalamic-pituitary dysfunction, individuals not receiving thyroid hormone replacement with FT4 levels < 1.0 ng/dL and TSH levels ≤ 4.0 mIU/L and those with a documented diagnosis of central hypothyroidism were excluded from the study unless onset of primary hypothyroidism preceded that of superimposed central hypothyroidism.¹⁴ We excluded from analysis participants who: (1) were not on thyroid hormone replacement and had serum FT4 values > 2.1 ng/dL (regardless of TSH levels); and, (2) had a history of hyperthyroidism even if they subsequently became hypothyroid or euthyroid. Individuals whose thyroid function could not be determined based on history or laboratory studies were also excluded (Figure 1).

Figure 1:

Consort Diagram

Physical Health Outcomes

Height, weight, body mass index, waist circumference and blood pressure were obtained for all participants. Abdominal obesity was defined by waist circumference ≥102 cm in men and ≥88 cm in women or a waist-to-height ratio of more than 0.5.¹⁵ Hypertension, abnormal glucose metabolism and dyslipidemia were graded using the modified National Cancer Institute Terminology Criteria for Adverse Events (CTCAE) version v4.03.¹⁶ Grading took into account whether intervention was required and if so, its intensity. Associations were examined between primary hypothyroidism and grade 2-4 vs < grade 2 conditions. Physiological frailty, a phenotype associated with aging was defined by the presence of three or more of the following conditions: muscle weakness, low muscle mass, low energy expenditure, self-reported exhaustion and slow walking speed.¹⁷

Neurocognitive and Psychosocial Outcomes

Standardized objective testing used assess neurocognitive function included assessments of attention,^18-20 memory,²¹ executive function,^{19, 20} processing speed,^{19, 20} intelligence²² and academics.²³ Patient-reported neurobehavioral problems (task efficiency, memory, organization, emotional regulation) were assessed via questionnaire.²⁴ Health-related QOL (eight subscales: physical function, role limitations–physical, bodily pain, general health, vitality, social function, role limitations–emotional and mental health; two composite summaries: physical health and mental health) was evaluated using the Medical Outcomes Short-form 36 with scores ≤ 40 defined as impaired.²⁵ Social attainment was assessed by patient-reported educational attainment, employment, household income, marital status, independent living, and health care insurance access.

Subsequent Thyroid Carcinoma

Subsequent thyroid carcinoma was confirmed by histopathology reports. For the purpose of this study, we considered as hypothyroid only those whose diagnosis of hypothyroidism preceded their diagnosis of subsequent thyroid carcinoma by at least one year. Individuals with thyroid carcinoma as their primary diagnosis were excluded from this analysis.

Statistics

Four distinct analyses were conducted. First, the prevalence of primary hypothyroidism and associations with demographic and treatment-related risk factors were evaluated using thyroid function status from the most recent SJLIFE clinical assessment. Associations were investigated using logistic regression models including sex, race/ethnicity, age at cancer diagnosis, age at SJLIFE assessment, thyroid radiotherapy dose and CED.

Second, associations were examined between primary hypothyroidism and cardiovascular disease risk factors, frailty, neurocognitive parameters, QOL and social attainment measures as outcomes in a cross-sectional fashion using results from the most recent clinical assessment. Covariates included sex, race/ethnicity, age at cancer diagnosis, age at clinical assessment, abdominal obesity, hypertension, abnormal glucose metabolism, dyslipidemia, growth hormone deficiency,¹⁴ CED, cumulative glucocorticoid dose, intrathecal or high-dose methotrexate, anthracycline exposure, cranial radiotherapy dose, chest radiotherapy dose, meeting CDC physical activity guidelines (≥ 150 minutes of moderate or vigorous physical activity per week ; this covariate was used only in analyses pertaining to cardiovascular disease risk factors), tobacco use and alcohol consumption. Since a proportion of survivors were treated for non-surgical primary hypothyroidism at outside institutions, we repeated the analysis using multiple imputation (based on sex, thyroid radiotherapy dose and age at SJLIFE) of thyroid function status in these individuals. We also repeated the analysis after the exclusion of these survivors.

Third, FT4 and TSH levels from the most recent assessment were examined to identify associations with cardiovascular risk factors, frailty, neurocognitive parameters, QOL and social attainment outcomes. This analysis utilized TSH and FT4 levels obtained on thyroid hormone replacement therapy when applicable, independent of underlying thyroid status, and within 6 months of the assessed outcome. The reference segments used for comparisons were the upper middle normal range of 1.6-1.9 ng/dL for FT4 and lower middle normal range of 0.4-1.9 mIU/L for TSH.

Fourth, associations between risk of subsequent thyroid carcinoma and demographic characteristics, primary cancer treatment, lifestyle factors and preexisting primary hypothyroidism were examined.

In all analyses, clinical relevance and statistical considerations (having adequate numbers per category) were used to determine variable categorization. Multivariable analyses incorporating variables with p ≤0.10 were subsequently conducted with results expressed as adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs).

Results

Of 4225 potentially eligible survivors, 3228 (76%) completed clinical assessments. Variables with significant differences between participants and non-participants included sex, thyroid radiotherapy dose, high dose intravenous methotrexate dose and CED (Tables 1 and Supplemental A1). After applying exclusions (Figure 1), 2965 were evaluable for investigation of hypothyroidism and physical health outcomes and 2025 for the neurocognitive and psychosocial outcomes. Median age at assessment and time since initial cancer diagnosis were 30.9 (range 18.1-64.6) and 22.3 (10.4-49.8) years, respectively. With 436 affected individuals, the prevalence of primary hypothyroidism was 14.7% (95% CI: 13.5%-16.0%); n=116 (26.6%) were diagnosed with non-surgical primary hypothyroidism outside our institution. Thyroidectomy, performed for the treatment of thyroid nodules, thyroid cancer or goiter was noted in 31.9% of participants with primary hypothyroidism (n=139, including eight who had partial thyroidectomy). Thirty-two individuals with primary hypothyroidism were untreated including 13 with subclinical or compensated hypothyroidism (serum TSH > 4.0 mIU/L and normal FT4). All but eight treated participants were on levothyroxine.

Table 1:

Participant Characteristics

	Participants N=3228
Variable	No.	%
Sex
Male	1693	52.45
Female	1535	47.55
Race / Ethnicity
Non-Hispanic White	2653	82.19
Non-Hispanic Black	466	14.44
Hispanic	66	2.04
Other	43	1.33
Age at Evaluation, years
16-25.9	975	30.20
26-35.9	1312	40.64
36-45.9	733	22.71
46-55.9	188	5.82
≥ 56	20	0.62
Median (range)	30.68 (18.05-64.63)
Primary Cancer Diagnosis
Leukemia	1185	36.71
Lymphoma	617	19.11
Central nervous system tumor	346	10.72
Embryonal tumor	527	16.33
Bone and soft tissue sarcoma	426	13.20
Carcinoma	48	1.49
Other	79	1.49
Age at Cancer Diagnosis, years
0 – 4.9	1200	37.17
5 – 9.9	760	23.54
10 – 14.9	744	23.05
≥15	524	16.23
Median (Range)	7.36 (0-24.79)
Thyroid Radiotherapy Dose, Gy
None	1739	53.91
> 0 and <20	960	29.76
≥20 and <30	273	8.46
≥30 and <40	105	3.25
≥40	149	4.62
Dose unknown	2	-
Cyclophosphamide Equivalent Dose (mg/m2)
0	1342	41.78
> 0 to < 4000	287	8.94
≥ 4000 to < 8000	561	17.47
≥ 8000 to < 12000	500	15.57
≥ 12000	522	16.25
Dose unknown	16	-
Intrathecal Methotrexate
No	1950	60.41
Yes	1278	39.59
High-dose IV methotrexate (g/m2)
No	2355	72.96
Yes	873	27.04
Glucocorticoid Cumulative Dose (Prednisone Equivalent, mg)
None	1727	54.65
> 0 to < 4000	781	24.72
≥ 4000 to < 8000	85	2.69
≥ 8000 to < 12000	415	13.13
≥ 12000	152	4.81
Dose unknown	68	-
Cranial Radiotherapy Dose, Gy
None	2134	66.11
> 0 to < 20	395	12.24
≥ 20 to < 35	393	12.17
≥ 35	306	9.48
Dose unknown	-	-

Main Analysis

In multivariable analysis, primary hypothyroidism was associated with female sex, age ≥ 15 years at primary cancer diagnosis compared to diagnosis at <5 years of age and was more likely in participants of white race compared to others (Table 2). Of note, participants whose hypothyroidism occurred after thyroid surgery (n=102) were excluded from this analysis. Primary hypothyroidism was independently associated with any radiotherapy dose to the thyroid in a dose-related fashion (Table 2). Primary hypothyroidism was independently associated with frailty, dyslipidemia, self-reported impaired physical function (Table 3) and having healthcare insurance (OR 1.51; 95%CI 1.07-2.12; p=0.02).

Table 2:

Risk Factors of Primary Hypothyroidism- Multivariable Analysis

	Primary Hypothyroidism (Original Cohort)						Primary Hypothyroidism (after exclusion of survivors diagnosed with non-surgical hypothyroidism outside of St. Jude)
Characteristic	Total N	N	%	OR	95%CI	P	Total N	N	%	OR	95%CI	P
Sex
Male	1526	144	9.44	1.00			1478	96	6.50	1.00
Female	1334	190	14.24	1.06	1.03-1.08	<0.0001	1266	122	9.64	1.04	1.02-1.06	<0.0001
Race/Ethnicity_
Non-Hispanic White	2350	291	12.38	1.00			2242	183	8.16	1.00
Other	510	43	8.43	0.96	0.93-0.99	0.003	502	35	6.97	-	-	-
Age at Cancer Diagnosis, years
0 – 4.9	1072	68	6.34	1.00			1043	39	3.74	1.00
5 – 9.9	666	59	8.86	0.98	0.96-1.01	0.28	651	44	6.76	1.00	0.92-1.02	0.68
10 – 14.9	650	101	15.54	1.02	0.99-1.05	0.13	618	69	11.17	1.02	0.99-1.04	0.19
≥15	472	106	22.46	1.05	1.01-1.09	0.01	432	66	15.28	1.05	1.03-1.07	<0.0001
Age at Evaluation, years
16 – 25.9	856	71	8.29	1.00			848	63	7.43	1.00
26 – 35.9	1180	136	11.53	0.98	0.96-1.01	0.15	1143	99	8.66	-	-	-
36 – 45.9	641	97	15.13	0.99	0.96-1.02	0.41	589	45	7.64	-	-	-
=or >46	183	30	16.39	0.96	0.91-1.01	0.09	164	11	6.71	-	-	-
Thyroid Radiotherapy Dose, Gy
None	1631	56	3.43	1.00			1596	21	1.32	1.00
> 0 to <10	678	53	7.82	1.05	1.03-1.08	<0.0001	662	37	5.59	1.05	1.03-1.07	<0.0001
≥10 to < 20	184	49	26.63	1.27	1.21-1.33	<0.0001	173	38	21.97	1.24	1.19-1.29	<0.0001
≥ 20 to < 30	219	97	44.29	1.49	1.43-1.55	<0.0001	198	76	38.48	1.45	1.40-1.50	<0.0001
≥30 to < 40	80	40	50.00	1.59	1.49-1.70	<0.0001	61	21	34.43	1.40	1.32-1.49	<0.0001
≥40	66	39	59.09	1.75	1.63-1.87	<0.0001	52	25	48.08	1.59	1.49-1.71	<0.0001
Cyclophosphamide Equivalent Dose (mg/m2)
None	1187	123	10.36	1.00			1143	79	6.91	1.00
> 0 to < 4000	253	39	15.42	1.00	0.96-1.04	0.86	240	26	10.83	0.98	0.95-1.02	0.37
≥ 4000 to < 8000	507	63	12.43	0.99	0.97-1.02	0.73	487	43	8.83	0.99	0.97-1.02	0.54
≥ 8000 to < 12000	441	42	9.52	1.00	0.97-1.03	0.94	425	26	6.12	1.00	0.97-1.02	0.72
≥ 12000	459	66	14.38	0.99	0.96-1.02	0.35	436	43	9.86	0.98	0.95-1.00	0.10

Table 3:

Associations between Primary Hypothyroidism and Physical Outcomes: Multivariable Analysis

	Frailty**
Characteristic	Total N	N*	%	OR	95%CI	P
Primary Hypothyroidism
Absent	2529	158	6.34	1.00
Present	436	47	11.01	1.54	1.05-2.26	0.03
	Dyslipidemia**
Characteristic	Total N	N*	%	OR	95%CI	P
Primary Hypothyroidism
Absent	2529	402	15.90	1.00
Present	436	121	27.75	1.52	1.14-2.04	0.005
	Physical Function ≤ 40**
Characteristic	Total N$	N*	%	OR	95%CI	P
Primary Hypothyroidism
Absent	1783	266	15.90	1.00
Present	242	64	27.83	1.66	1.12-2.48	0.01

^*Missing values were ignored

^**All analyses were adjusted to age at evaluation, sex, race/ethnicity, abdominal obesity, hypertension, diabetes mellitus, dyslipidemia (except the analysis of dyslipidemia as an outcome), growth hormone deficiency, cyclophosphamide equivalent dose, high dose intravenous methotrexate, intrathecal methotrexate, cumulative glucocorticoid dose, cranial radiotherapy dose, tobacco use and risky alcohol consumption. The analyses of dyslipidemia and physical function were adjusted for chest radiotherapy exposure. The analysis for dyslipidemia was also adjusted for anthracycline exposure and meeting CDC physical activity guidelines.

^$$The number of participants was limited by the need to meet additional eligibility criteria for neuropsychological testing.

Compared to participants with high normal FT4 (1.6-1.9 ng/dL), those with FT4 of 0.9-1.29 ng/dL or 1.3-1.59 ng/dL were more likely to have abdominal obesity (OR 1.83; 95% CI 1.45-2.32; p<0.0001 and OR 1.38; 95% CI 1.09-1.74; p=0.007, respectively) and were less likely to have frailty (OR 0.64; 95% CI 0.42-0.96; p=0.03 and OR 0.56; 95% CI 0.37-0.86; p=0.008, respectively). FT4 ≥ 2.0 ng/dL was associated with impaired memory on neuropsychological testing (OR 2.44; 95% CI 1.09-5.47; p=0.03) compared to participants with high normal FT4 levels (Supplemental Table A2).

Overall, FT4 serum levels < 1.6 ng/dL were independently associated with QOL impairment. Compared with FT4 1.6-1.9 ng/dL, FT4 levels < 0.9 ng/dL were significantly associated with risk for reduced general health (OR 3.20; 95% CI 1.09-9.43; p=0.03) ; FT4 0.9-1.29 was associated with role limitations due to physical health problems and bodily pain (OR 1.56; 95%CI 1.04-2.35; p=0.03 and OR 1.63; 95% CI 1.10-2.41; P=0.01 respectively) while FT4 1.3-1.59 was associated with role limitations due to physical health problems (OR 1.67; 95% CI 1.10-2.51; p=0.01) (Supplemental Table A3).

Levels of TSH within the lower half of the normal range were associated with better QOL compared to values below the normal range and those in the upper half of the normal range. TSH <0.4 mIU/L was independently associated with role limitations due to physical health problems (OR 2.50; 95% CI 1.37-4.58; p=0.003), while TSH 2.0-2.99 mIU/L was associated with reduced social functioning (OR 1.58; 95% CI 1.08-2.32; p=0.02) and reduced mental health [(subscale: OR 1.74; 95% CI 1.21-2.51; p=0.003);composite summary: OR 1.60; 95% CI 1.10-2.33; p=0.01)] when compared to TSH 0.4-1.9 mIU/L (Supplemental Table A3).

Subsequent thyroid carcinoma occurred in 79 participants. Among them, ten did not receive radiotherapy to the thyroid prior to the diagnosis of thyroid carcinoma; none of those ten participants had primary hypothyroidism prior to the diagnosis of thyroid cancer. The remaining 69 were treated with radiotherapy, of whom 20 had previously been diagnosed with primary hypothyroidism (median 11.67 years; range 2.20-27.36). In adjusted models, subsequent thyroid carcinoma was associated with female sex (OR 1.88; 95% CI 1.15-3.07; p=0.01) and radiotherapy doses >0-<20 Gy ( OR 7.78; 95% CI 3.23-18.74; p <0.0001) and ≥20-< 35 Gy (OR: 10.99; 95% CI 4.37-27.64; p<0.0001). The association with a prior history of primary hypothyroidism was not significant.

Imputation Analysis

65.3% of participants whose thyroid function status was imputed based on sex, thyroid radiation dose and age at SJLIFE were no longer considered to have primary hypothyroidism. (Supplemental Table A4).

Utilizing the imputed thyroid status resulted in the following associations becoming statistically significant: impaired academic function (OR 1.75; 95% CI 1.08-2.84; p=0.02), slowed processing speed (OR 1.69; 95% CI 1.04-2.73; p=0.03), abdominal obesity (OR 1.66; 95% CI 1.18-2.32; p=0.003) and subsequent thyroid carcinoma (OR 16.06; 95% CI 5.77-44.64; p<0.0001). Associations that lost statistical significance after imputation included those between primary hypothyroidism and frailty (OR 1.60; 95% CI 0.99-2.58; p=0.06), self-reported impaired physical function (OR 1.56; 95% CI 0.94-2.57; p=0.08) and having healthcare insurance (OR 1.47; 95% CI 0.97-2.23; p=0.07).

Outcomes after Exclusion of Survivors with Non-surgical Hypothyroidism Diagnosed outside of St. Jude

Risk factors independently associated with primary hypothyroidism remained unchanged except for Race/Ethnicity which was no longer significant (Table 2). Primary hypothyroidism was independently associated with frailty (OR 1.78; 95% CI 1.14-2.77; p=0.001) and abdominal obesity (OR 1.38; 95%CI 1.03-1.85; p=0.03) but no longer with dyslipidemia (OR 1.37: 95% CI 0.98-1.92; p=0.06). The association with self-reported impaired physical function (OR 1.86; 95% CI 1.17-2.95; p=0.009) remained significant while primary hypothyroidism became significantly associated with role limitations due to physical function (OR 1.58; 95% CI 1.01-2.47; p= 0.047; Supplemental Table A5). The association with having healthcare insurance was no longer significant (OR 1.29: 95% CI 0.89-1.87; p=0.18). A history of preexisting primary hypothyroidism was independently associated with subsequent thyroid carcinoma (OR 1.18; 95% CI 1.15-1.20; p< 0.0001).

Discussion:

This report provides a comprehensive evaluation of primary hypothyroidism in a large and well-characterized cohort of childhood cancer survivors with extended follow-up. The current study has addressed knowledge gaps pertaining to the impact of primary hypothyroidism on survivors by showing independent associations with physiological frailty and QOL. The present study is also unique in its ability to explore associations between serum levels of FT4 and TSH and adverse health outcomes.

At 14.7%, the prevalence of primary hypothyroidism was within the range reported by others.² This was substantially higher than the 4.6% reported in the non-cancer survivor population ²⁶. The high proportion of individuals with surgically induced hypothyroidism is likely due to the extended follow-up of individuals at high risk for secondary thyroid neoplasia such as survivors of Hodgkin lymphoma.³ The great majority (95.5%) of individuals with primary hypothyroidism were treated, which may indicate a higher level of comfort with the management of this condition by community providers in comparison to other endocrine late effects. ¹⁴ The small number of individuals with untreated subclinical hypothyroidism (n=13) is surprising given a previously reported prevalence of nearly 25% among survivors;²⁷ this discrepancy may reflect the resolution of cases over time, ²⁷ or the practice of more readily initiating treatment in long-term survivors with subclinical hypothyroidism (and who could thus not be identified as subclinical in our cohort) due to concerns that chronically elevated TSH may increase the risk for thyroid neoplasia. ^{27, 28}

Overall, associations found between primary hypothyroidism and chronic health conditions such as dyslipidemia and impaired QOL mirrored those reported in the general non cancer survivor population.⁸ The only significant association with social attainment pertained to having health care insurance. Hypothyroidism may have provided impetus for obtaining or maintaining insurance in a subset of otherwise healthy survivors; its treatment may have been facilitated by access to insurance. The independent association with frailty has not been previously reported in childhood cancer survivors. The loss of statistical significance after the imputation of thyroid function status suggests that this finding was primarily driven by survivors who were diagnosed after transitioning community-based care. Conversely, independent associations observed after imputation with impaired academic performance and slowed processing speed suggest a possible impact of primary hypothyroidism on cognitive outcomes in a subset of survivors. These associations thus deserve further study.

The exploratory analysis of correlations between Free T4, TSH levels and health outcomes was conducted independently of the underlying thyroid status; its results could thus reflect the impact of thyroid hormone replacement rather than of hypothyroidism per se. The analysis showed contrasting trends at high-normal FT4 levels: better outcomes for obesity and QOL and worse outcomes for frailty. The association with frailty has been reported in elderly community dwelling individuals. ²⁹ FT4 above the normal range was associated with memory impairment. Survivors of thyroid carcinoma may require several years of thyroid hormone replacement at TSH -suppressive doses, frequently resulting in high normal or mildly elevated FT4 levels. ³⁰ The overall impact of this strategy on other aspects of survivor health, such as frailty and cognition, deserves further study. While management guidelines suggest maintaining TSH within the normal range in patients treated for hypothyroidism,³⁰ our data suggest that values in the lower half of the normal range are associated with better QOL.

Despite the known property of TSH as a stimulus for thyroid neoplasia, ²⁸ we were not able to identify an independent association between preexisting hypothyroidism and subsequent thyroid carcinoma in the main analysis. This investigation is however complicated by the relatively small number of events, and the fact that thyroid cancer may have been present for many years before being detected. However, the significance of the association after imputation of thyroid status and after the exclusion of survivors diagnosed with non-surgical hypothyroidism outside of St. Jude does not allow the definitive exclusion of a possible contribution of preexisting primary hypothyroidism to the risk of subsequent thyroid carcinoma. This association thus deserves further evaluation, preferably in prospectively followed and systematically assessed cohorts. This study confirmed previously observed associations between survivor and treatment factors, primary hypothyroidism^3-7 and subsequent thyroid carcinoma. ^{3, 31, 32}.

Several study limitations have to be noted. These include the number of nonparticipants, the fact that treatment was not stopped to verify the diagnosis of hypothyroidism in participants treated outside of our institution, the absence of thyroid auto-antibody measures, difficulties in determining thyroid status resulting in participant exclusions, intra-individual variability of thyroid function tests, the fact that these were not necessarily obtained on the day of the neuropsychological assessment but within a six months window, potential interference of factors such as concurrent medications and co-morbid conditions with hormonal assays, and the cross-sectional analysis of associations with health outcomes. While associations between primary hypothyroidism and adverse health outcomes are concerning, some of these results may be the consequence of shared risk factors, such as central nervous system neoplasia and /or cranial radiotherapy rather than demonstrating causal relationships and should thus be validated in other cohorts.

In summary, primary hypothyroidism is commonly observed in childhood cancer survivors and is associated with unfavorable physical health and quality of life outcomes despite replacement therapy. Observed interactions between laboratory measures of thyroid function and obesity, frailty, neurocognitive impairment and reduced QOL suggest a need to further examine the impact of thyroid hormone dosing practices on outcomes in long-term survivors.