Thyroid disorders have emerged as one of the most common immune-related adverse events (irAE), yet optimum management and biomarkers to predict vulnerable individuals remain to be explored. High-dose glucocorticoid (HDG) therapy is routinely recommended for irAEs. However, systematic analysis of the impact of glucocorticoid therapy on the outcome of immune-checkpoint inhibitor (ICI)–induced thyroid disorders is lacking. We analyzed 151 patients with or without ICI-related thyroid disorders. We divided the patients with ICI-related thyroid disorders into two subgroups: those with and without HDG treatment. Our results showed no significant differences between HDG and no HDG groups in terms of the median duration of thyrotoxicosis: 28 (range, 7–85) and 42 (range, 14–273) days, the median time to conversion from thyrotoxicosis to hypothyroidism: 39 days (range, 14–169) and 42 days (range, 14–315) days, the median time to onset of hypothyroidism: 63 (range, 21–190) and 63 (range, 14–489) days, and the median maintenance dose of levothyroxine: 1.5 (range, 0.4–2.3) μg/kg/day, and 1.3 (range, 0.3–2.5) μg/kg/day. The median pretreatment TSH was 2.3 (range, 0.3–5.2) mIU/L and 1.7 (range, 0.5–4.5) mIU/L in patients with and without ICI-related thyroid disorders, respectively. Baseline TSH was significantly higher in patients who developed ICI-related thyroid disorders (P = 0.05). Subgroup analysis revealed significantly higher baseline TSH in male but not in female patients with ICI-induced thyroid dysfunction. Our results show that HDG treatment did not improve the outcome of ICI-related thyroid disorders.

Unleashing the immune response through the use of immune-checkpoint inhibitors has demonstrated significant improvement in overall survival in a broad spectrum of advanced solid as well as liquid malignancies (1–6), but this treatment increases autoimmunity and causes immune-related adverse events (irAE; refs. 7, 8). Thyroid disorders have emerged to be very common irAEs (9). Early identification and proper management of immune-checkpoint inhibitor (ICI)–induced thyroid disorders will help to prevent severe consequences such as thyroid storm or myxedema coma, and improve quality of life. High-dose glucocorticoid therapy (HDG) is routinely recommended for moderate–severe irAEs (10, 11). However, systematic analysis of the impact of HDG on the outcome of ICI-induced thyroid disorders is lacking. In the context of management of hypothyroidism, initiation with 1.6 μg/kg/day was recommended by the American Society of Clinical Oncology (ASCO) clinical practice guideline (11), but no study has evaluated the replacement dose of levothyroxine in ICI-induced hypothyroidism. Additionally, there is no biomarker reported to identify individuals susceptible to ICI-related thyroid disorders. The purpose of this study was to evaluate the optimum management of ICI-induced thyroid disorders and to explore the role of baseline TSH as a predictive biomarker of ICI-induced thyroid disorders.

Study design and patient cohorts

This cohort analysis was performed retrospectively by collecting data from chart reviews with removal of individual identifiers. The patients were identified through a Research Patient Data Registry (RPDR), a program developed at our institution to search patients in the electronic database by diagnosis, laboratory values, medication, and/or keyword. Some of the patients with ICI-related thyroid disorders were evaluated clinically in the outpatient endocrinology clinic of Brigham and Women's Hospital and received continued longitudinal clinical care. The period for this study was from December 4, 2009, to August 14, 2017. Institutional Review Board approval was obtained for the study. We identified 65 patients with ICI-induced thyroid disorders and 98 individuals who were exposed to ICIs but did not develop thyroid disorders as a control group (Fig. 1). Patients were excluded if they did not have pretreatment TSH measurements, if pretreatment TSH was greater than up limit of normal range (5.5 mIU/L) by assays in our institutions, if dates of HDG treatment were missing, or if they received HDG prior to the onset of thyrotoxicosis. After exclusions, 53 and 98 patients with and without ICI-induced thyroid disorders, respectively, were analyzed.

Figure 1.

Flow chart of study design and subjects. HDG, high-dose glucocorticoids.

Figure 1.

Flow chart of study design and subjects. HDG, high-dose glucocorticoids.

Close modal

Definition of thyrotoxicosis and hypothyroidism

Serum TSH, thyroxine (T4), free T4 (FT4), and/or free triiodothyronine (FT3) were measured prior to initiation of ICI treatment, and then every 2 to 4 weeks as mandated by clinical trial protocols or clinical practice. Thyrotoxicosis was defined as a decreased TSH with an elevated T4, FT4, or FT3. Hypothyroidism was defined as an elevated TSH with a decreased T4, FT4, or FT3. Subclinical hypothyroidism and hyperthyroidism were defined as normal T4, FT4, and FT3 with elevated or decreased TSH, respectively. Reference ranges of normal laboratory values in our institutions were TSH, 0.5–5.5 mIU/L; T4, 5.0–11.0 μg/dL; FT4, 0.9–1.7 ng/dL; FT3, 2.8–4.4 pg/mL; thyroperoxidase antibodies (TPOAb), < 35 IU/mL; thyroid-stimulating immunoglobulin index (TSII), ≤ 1.3.

Definitions of time to onset of thyroid disorders

Time to onset of thyrotoxicosis was defined as the number of days between the administration of the first dose of ICI and the date of the first documented biochemical evidence of thyrotoxicosis. Duration of thyrotoxicosis was defined as the number of the days between the initial biochemical evidence of thyrotoxicosis and the first subsequent documented biochemical evidence of euthyroidism or hypothyroidism. Time to conversion from thyrotoxicosis to hypothyroidism was defined as the number of days between the initial biochemical evidence of thyrotoxicosis and the first documented biochemical evidence of hypothyroidism. Time to onset of hypothyroidism was defined as the number of days between the administration of the first dose of ICI and the date of the first biochemical documentation of hypothyroidism.

Patient regimens

The patients were divided into subgroups based on the regimens of ICIs received. We analyzed time to onset of thyrotoxicosis and hypothyroidism by different treatments. We evaluated the association of baseline TSH with the development of ICI-induced thyroid disorders. We further divided the patients into two groups: those with and without HDG treatment. Systemic HDG treatment was defined as the administration of glucocorticoids at a dose of more than 7.5 mg prednisone or equivalent daily for more than one week. HDG treatment was given to 15 patients: before the onset of thyrotoxicosis (n = 1), after the onset of thyrotoxicosis (n = 7), before the onset of hypothyroidism (n = 1), or after the onset of hypothyroidism (n = 6). Among these 15 patients, 3 patients received HDG treatment for thyrotoxicosis, 10 for other immune-related adverse events (irAE), 1 for brain metastasis, and 1 for headache. The patients in the HDG group received prednisone (n = 4, at 30–60 mg daily for 2–6 weeks), dexamethasone (n = 2, at 2–6 mg daily for 9 days to 55 weeks), methylprednisolone followed by prednisone (n = 5, methylprednisolone at 40–260 mg daily for 1–3 days; then prednisone at 60–120 mg daily for 2–15 weeks), methylprednisolone followed by dexamethasone and prednisone (n = 1, methylprednisolone at 180 mg daily for 8 days, then dexamethasone at 36 mg tapered over 4 weeks, then prednisone at 200 mg tapered over 9 weeks), methylprednisolone followed by hydrocortisone (n = 1, methylprednisolone at 50 mg daily for 1 day, then hydrocortisone at 50–150 mg daily for 2 weeks), hydrocortisone followed by prednisone (n = 1, hydrocortisone at 50 mg daily for 1 day, then prednisone at 80 mg tapered over 2 weeks), or dexamethasone followed by prednisone (n = 1, dexamethasone at 8 mg daily for 24 days, prednisone at 80 mg tapered for 46 weeks). We analyzed the impact of HDG on the duration of thyrotoxicosis and the time to conversion from thyrotoxicosis to hypothyroidism, including the patients treated with HDG before or after the onset of thyrotoxicosis (n = 8). We also analyzed the time to onset of hypothyroidism (patients treated with HDG before or after the onset of thyrotoxicosis and before the onset of hypothyroidism were included, n = 9). We did not analyze the impact of HDG on the time to onset of thyrotoxicosis, because no patients received HDG prior to initiating ICI treatment and only 1 patient received HDG before the onset of thyrotoxicosis. We analyzed the impact of HDG on the duration of thyrotoxicosis, the time to conversion from thyrotoxicosis to hypothyroidism, and the time to onset of hypothyroidism. We evaluated the impact of HDG on the maintenance dose of levothyroxine.

Statistical analysis

Summaries of patient characteristics are primarily descriptive. Comparisons between groups of characteristics measured on a continuous scale are based on Wilcoxon rank-sum or Kruskal–Wallis tests. Comparisons of characteristics measured categorically are based on Fisher exact tests. The distributions of time to thyrotoxicosis or hypothyroidism are described using the method of Kaplan–Meier; median times are presented with 95% confidence intervals. A Cox proportional hazards regression model was used to identify predictors of time to develop thyrotoxicosis. Statistical analyses were performed using SAS 9.4 (SAS Institute Inc.). Statistical significance is defined as P ≤ 0.05; there were no corrections for multiple comparisons.

Study population

Among 53 patients who developed thyroid disorders, 18 (34%) patients received pembrolizumab (pembro), 9 (17%) patients received nivolumab (nivo), and 26 (49%) patients received ipilimumab (ipi) + nivo therapy (Fig. 1). The demographic data are summarized in Table 1.

Table 1.

Clinical characteristics of patients with ICI-related thyroid disorders

CharacteristicPembroNivoIpi + NivoTotal
Number 18 26 53 
Age, years 
 Median (range) 62 (32–82) 60 (42–81) 58 (34–78) 60 (32–82) 
Gender, n (%) 
 Female 10 (56) 9 (100) 15 (58) 34 (64) 
 Male 8 (44) 0 (0) 11 (42) 19 (36) 
Malignancy, n (%) 
 Melanoma 12 (67) 3 (33) 21 (81) 36 (68) 
 Leiomyosarcoma 1 (6) 0 (0) 0 (0) 1 (2) 
 Breast cancer 3 (17) 0 (0) 0 (0) 3 (6) 
 Lung cancer 1 (6) 0 (0) 0 (0) 1 (2) 
 Glioblastoma 1 (6) 2 (22) 0 (0) 3 (6) 
 Renal cell carcinoma 0 (0) 0 (0) 1 (4) 1 (2) 
 Meningioma 0 (0) 3 (33) 0 (0) 3 (6) 
 Bladder cancer 0 (0) 1 (11) 1 (4) 2 (4) 
 Ovarian cancer 0 (0) 0 (0) 2 (8) 2 (4) 
 Malignant neoplasm of  esophagus 0 (0) 0 (0) 1 (4) 1 (2) 
Preexisting thyroid disorder, n (%) 
 Yes 4 (22) 3 (33) 1 (4) 8 (15) 
 No 14 (78) 6 (67) 25 (96) 45 (85) 
CharacteristicPembroNivoIpi + NivoTotal
Number 18 26 53 
Age, years 
 Median (range) 62 (32–82) 60 (42–81) 58 (34–78) 60 (32–82) 
Gender, n (%) 
 Female 10 (56) 9 (100) 15 (58) 34 (64) 
 Male 8 (44) 0 (0) 11 (42) 19 (36) 
Malignancy, n (%) 
 Melanoma 12 (67) 3 (33) 21 (81) 36 (68) 
 Leiomyosarcoma 1 (6) 0 (0) 0 (0) 1 (2) 
 Breast cancer 3 (17) 0 (0) 0 (0) 3 (6) 
 Lung cancer 1 (6) 0 (0) 0 (0) 1 (2) 
 Glioblastoma 1 (6) 2 (22) 0 (0) 3 (6) 
 Renal cell carcinoma 0 (0) 0 (0) 1 (4) 1 (2) 
 Meningioma 0 (0) 3 (33) 0 (0) 3 (6) 
 Bladder cancer 0 (0) 1 (11) 1 (4) 2 (4) 
 Ovarian cancer 0 (0) 0 (0) 2 (8) 2 (4) 
 Malignant neoplasm of  esophagus 0 (0) 0 (0) 1 (4) 1 (2) 
Preexisting thyroid disorder, n (%) 
 Yes 4 (22) 3 (33) 1 (4) 8 (15) 
 No 14 (78) 6 (67) 25 (96) 45 (85) 

Time to onset of thyrotoxicosis and hypothyroidism

ICI-related thyroid disorders presented as hypothyroidism only (21%), thyrotoxicosis only (4%), or thyrotoxicosis followed by hypothyroidism (75%; Supplementary Table S1). Onset of thyrotoxicosis and hypothyroidism was analyzed by the method of Kaplan–Meier survival analysis. The follow-up of patients who did not develop thyrotoxicosis is censored at the time of hypothyroidism development. The median time to onset of thyrotoxicosis after the initiation of ICIs was 44 days (range, 19–447), 56 days (range, 13–126), and 21 days (range, 7–64), in the pembro, nivo, and ipi + nivo groups, respectively (Fig. 2A, P = 0.018). The P values were 0.593, 0.004, and 0.158 between pembro and nivo, pembro and ipi + nivo, and nivo and ipi + nivo, respectively. The median time to onset of hypothyroidism after the initiation of ICIs was 84 days (range, 43–544), 84 days (range, 14–154), and 62 days (range, 21–141) in the pembro, nivo, and ipi + nivo groups, respectively (Fig. 2B, P = 0.003). The P values were 0.333, 0.002, and 0.057 between pembro and nivo, pembro and ipi + nivo, and nivo and ipi + nivo, respectively. Patients in the combination ipi + nivo group had significantly shorter times to develop both thyrotoxicosis and hypothyroidism than those in the pembro and nivo monotherapy groups (Fig. 2A and B). Forty of 42 patients (95%) who had thyrotoxicosis converted to hypothyroidism (Supplementary Table S1). There was no difference in the time to onset of hypothyroidism in patients with or without preceding thyrotoxicosis (Supplementary Fig. S1).

Figure 2.

Time to onset of thyrotoxicosis and hypothyroidism by different treatment regimens. Onset of thyrotoxicosis and hypothyroidism was analyzed by the method of Kaplan–Meier. A, Onset of thyrotoxicosis after first treatment with ICI. The follow-up of patients who did not develop thyrotoxicosis is censored at the time of development of hypothyroidism. Median onset time of thyrotoxicosis (median with range, days): pembro: 44 (range, 19–447); nivo: 56 (range, 13–126); ipi + nivo: 21 (range, 7–64). B, Onset of hypothyroidism after first treatment with ICI. Median onset time of hypothyroidism (median with range, days): pembro: 84 (range, 43–544); nivo: 84 (range, 14–154); ipi + nivo: 62 (range, 21–141). Significant difference is defined as P < 0.05.

Figure 2.

Time to onset of thyrotoxicosis and hypothyroidism by different treatment regimens. Onset of thyrotoxicosis and hypothyroidism was analyzed by the method of Kaplan–Meier. A, Onset of thyrotoxicosis after first treatment with ICI. The follow-up of patients who did not develop thyrotoxicosis is censored at the time of development of hypothyroidism. Median onset time of thyrotoxicosis (median with range, days): pembro: 44 (range, 19–447); nivo: 56 (range, 13–126); ipi + nivo: 21 (range, 7–64). B, Onset of hypothyroidism after first treatment with ICI. Median onset time of hypothyroidism (median with range, days): pembro: 84 (range, 43–544); nivo: 84 (range, 14–154); ipi + nivo: 62 (range, 21–141). Significant difference is defined as P < 0.05.

Close modal

Effect of HDG on the outcome of thyroid disorders

There were 11, 2, and 40 patients in this analysis who manifested with hypothyroidism only, thyrotoxicosis only, and thyrotoxicosis followed by hypothyroidism, respectively (Supplementary Table S1). Of the 42 patients with thyrotoxicosis (thyrotoxicosis only + thyrotoxicosis followed by hypothyroidism), 8 received HDG (19%) before or after the onset of thyrotoxicosis. The median number of days from the initiation of immunotherapy to the initiation of HDG treatment was 26 days. Among 40 patients who developed hypothyroidism following thyrotoxicosis, 8 patients received HDG before the onset of hypothyroidism. Among 51 patients who developed hypothyroidism (hypothyroidism only + thyrotoxicosis followed by hypothyroidism), 9 patients received HDG prior to the development of hypothyroidism. Most of the patients in this study did not receive HDG for their thyroid disorders because thyroid-related symptoms were usually mild, even if thyroid function tests showed markedly abnormal TSH, free T4, and free T3. HDG was generally used to treat other irAEs, except in 3 patients who were treated for severe symptomatic thyrotoxicosis.

Figure 3 illustrates the impact of HDG on the outcome of thyroid disorders. The median duration of thyrotoxicosis in the HDG and no HDG groups was 28 days (range, 7–85) and 42 days (range, 14–273), respectively. There was no significant difference in the median duration of thyrotoxicosis between the two groups (Fig. 3A, Wilcoxon rank-sum P = 0.26). The median time to conversion from thyrotoxicosis to hypothyroidism was 39 days (range, 14–169) and 42 days (range, 14–315) in the HDG and no HDG groups, respectively, which were not significantly different (Fig. 3B, Wilcoxon rank-sum P = 0.92). The median time to onset of hypothyroidism was 63 days (range, 21–190) and 63 days (range, 14–489) in the HDG and no HDG groups, with no significant difference (Fig. 3C, Wilcoxon rank-sum P = 0.50). None of the patients had an identified remission of their hypothyroidism during the study period.

Figure 3.

The effect of systemic HDG treatment on ICI-related thyroid disorders. A, Duration of thyrotoxicosis. Lines represent median with range; gray dots represent individual values. The difference between the two groups was not statistically significant (Wilcoxon rank-sum P = 0.26). B, Time to conversion from thyrotoxicosis to hypothyroidism in patients with or without HDG treatment. The difference between two groups was not statistically significant (Wilcoxon rank-sum P = 0.92). C, The time to onset of hypothyroidism after the initial treatment with ICI in patients with or without treatment with HDG. The difference between the two groups was not statistically significant (Wilcoxon rank-sum P = 0.50). D, Dose of levothyroxine (LT4) needed to achieve euthyroidism (μg/kg body weight/day). There was no significant difference among the groups (pembro, nivo, ipi + nivo; Kruskal–Wallis test P = 0.64). E, Impact of HDG on the dose of LT4 needed to achieve euthyroidism. There was no significant difference between the patients with or without HDG treatment (Wilcoxon rank-sum P = 0.46). For each panel, lines represent median with range; gray dots represent individual values.

Figure 3.

The effect of systemic HDG treatment on ICI-related thyroid disorders. A, Duration of thyrotoxicosis. Lines represent median with range; gray dots represent individual values. The difference between the two groups was not statistically significant (Wilcoxon rank-sum P = 0.26). B, Time to conversion from thyrotoxicosis to hypothyroidism in patients with or without HDG treatment. The difference between two groups was not statistically significant (Wilcoxon rank-sum P = 0.92). C, The time to onset of hypothyroidism after the initial treatment with ICI in patients with or without treatment with HDG. The difference between the two groups was not statistically significant (Wilcoxon rank-sum P = 0.50). D, Dose of levothyroxine (LT4) needed to achieve euthyroidism (μg/kg body weight/day). There was no significant difference among the groups (pembro, nivo, ipi + nivo; Kruskal–Wallis test P = 0.64). E, Impact of HDG on the dose of LT4 needed to achieve euthyroidism. There was no significant difference between the patients with or without HDG treatment (Wilcoxon rank-sum P = 0.46). For each panel, lines represent median with range; gray dots represent individual values.

Close modal

Effect of HDG on the maintenance dose of levothyroxine

We defined the maintenance dose of levothyroxine (LT4) as the dose needed to reach biochemical euthyroidism, i.e., normal TSH with normal T4 or FT4. We calculated the dose of LT4 (μg) relative to body weight (kg). We typically started LT4 at a dose of 0.8 to 1.0 μg/kg/day if patients were younger than 60 years and had no known cardiac disorders. In patients older than 60 years or those with known cardiac disease, we started with a lower dose of LT4 (25–50 μg daily) and increased the dose more slowly, usually taking 2 to 3 months to reach euthyroidism. In this analysis of levothyroxine maintenance dose, we excluded patients whose thyroid function tests did not reach euthyroidism during the study period. Among 38 patients who had data for levothyroxine dose and reached euthyroidism, the median dose of LT4 was 1.4 (range, 0.3–2.5) μg/kg/day. There was no significant difference in levothyroxine maintenance dose among nivo-, pembro-, or combination ipi + nivo-induced hypothyroidism (Fig. 3D; Kruskal–Wallis test, P = 0.64). We evaluated the impact of HDG on the dose of LT4 needed to achieve euthyroidism. The doses of LT4 were 1.5 (range, 0.4–2.3) μg/kg/day and 1.3 (range, 0.3–2.5) μg/kg/day in patients with and without HDG treatment, which were not significantly different (Fig. 3E, Wilcoxon rank-sum P = 0.46).

Baseline TSH concentration in patients with ICI-related thyroid disorders

The median pretreatment TSH for patients who developed ICI-induced thyroid disorders was 2.3 (range, 0.3–5.2 mIU/L) compared with 1.7 (range, 0.5–4.5 mIU/L) for those who did not. Patients who developed ICI-related thyroid disorders had significantly higher baseline TSH than patients who did not (P = 0.05; Table 2). Male patients with ICI-induced thyroid disorders had significantly higher pretreatment TSH than the remaining three subgroups of patients (P = 0.02; Table 2). When we performed further analyses in subgroups of patients whose initial presentation was with hypothyroidism or thyrotoxicosis, significantly higher pretreatment TSH was observed in only the group whose initial presentation was with hypothyroidism but not in those who presented with thyrotoxicosis (Supplementary Fig. S2). Among males, baseline TSH values were significantly higher in those with initial manifestations of either hypothyroidism or thyrotoxicosis. In contrast, in females, there was no significant difference in baseline TSH among any of the groups (Table 2; Supplementary Fig. S2). In patients with ICI-induced thyroid disorders, there was no difference in pretreatment TSH according to ICI regimen or HDG treatment (Supplementary Table S2). To validate the findings, we analyzed baseline TSH from 44 patients (all female) enrolled in a prospective randomized phase II open-label study of eribulin mesylate + pembrolizumab for patients with metastatic hormone receptor (HR)-positive breast cancer. Seven patients were excluded from the analysis, because 6 had pretreatment abnormal TSH and 1 did not have a pretreatment TSH performed. Nine patients developed thyroid disorders. The median pretreatment TSH was 1.64 (range, 0.58–3.40 mIU/L) and 1.64 (range, 0.63–4.60 mIU/L) in patients with or without pembrolizumab-induced thyroid disorders (Supplementary Table S3). There was no significant difference in pretreatment TSH values between the patients with or without pembrolizumab-induced thyroid toxicity. There were no thyroid disorders observed in the arm with eribulin mesylate monotherapy.

Table 2.

Baseline (pretreatment) TSH concentrations (mIU/L) in patients with or without ICI-induced thyroid disorders

Thyroid disordersNMinMedianMax
   TSH (mIU/L)   
Yesa  53 0.28 2.30 5.21 
No  98 0.50 1.70 4.45 
Overall  151 0.28 1.76 5.21 
 Sex     
Yes 34 0.28 1.66 5.21 
 Mb 19 0.51 2.98 4.83 
No 32 0.78 1.49 4.45 
 66 0.50 1.74 3.90 
Thyroid disordersNMinMedianMax
   TSH (mIU/L)   
Yesa  53 0.28 2.30 5.21 
No  98 0.50 1.70 4.45 
Overall  151 0.28 1.76 5.21 
 Sex     
Yes 34 0.28 1.66 5.21 
 Mb 19 0.51 2.98 4.83 
No 32 0.78 1.49 4.45 
 66 0.50 1.74 3.90 

NOTE: All, all patients; yes, patients developed ICI-induced thyroid disorders; No, patient did not develop ICI-induced thyroiditis. F, female; M, male.

aPatients who developed ICI-related disorders had significantly higher baseline TSH compared with patients who did not (Wilcoxon rank-sum P = 0.05).

bMales with ICI disorders had significantly higher pretreatment TSH than the remaining three subgroups of patients (Kruskal–Wallis P = 0.02).

Thyroid antibodies and FT3/FT4 ratio

Nineteen patients in this study were tested for TPOAb, of which 8 patients (42%) were positive. Among those tested, 60% (3/5), 50% (1/2), and 33% (4/12) had positive TPOAb in the pembro, nivo, and ipi + nivo groups, respectively. Seventeen patients were tested for TSII during thyrotoxicosis. Only 1 (6%) had detectable TSII, and this patient subsequently developed hypothyroidism. The calculated FT3/FT4 ratio was 1.8 (range, 0.6–2.7) in 21 patients during the thyrotoxicosis phase.

Thyroid disorders have emerged as one of the most common adverse events associated with anti–PD-1 monotherapy and with combination anti–PD-1 and anti–CTLA-4 therapy (2, 12–14). To explore the proper management of ICI-related thyroid disorders and to identify predictive biomarkers, we investigated the nature of ICI-induced thyroid disorders, the role of HDG on the outcome of thyroid disorder, the maintenance dose of levothyroxine, and the pretreatment TSH in patients with or without ICI-induced thyroid disorders.

Thyroid disorders associated with ICI therapy have been described as hyperthyroidism, hypothyroidism, and thyroiditis in previous reports (2, 3, 15). Consistent with previous work (16), the findings in the current study demonstrated that most ICI–related thyroid disorders are the result of thyroiditis. This conclusion is supported by the findings that 96% of the patients with ICI-related thyroid disorders developed hypothyroidism at the end of this study. The evolution of ICI-induced thyroid disorders follows the typical time course of thyroiditis (17). The onset time of thyrotoxicosis and hypothyroidism between pembro and nivo was not statistically significant. The median FT3/FT4 ratio during thyrotoxicosis was 1.8 (range, 0.6–2.7), consistent with thyroiditis (18). Although 1 patient in this study had positive TSII, this patient eventually developed hypothyroidism, indicating she developed thyroiditis as well.

The role of HDG in the management of ICI-related thyroid disorders remains unclear. High-dose steroid treatment is very effective in shortening the period of thyrotoxicosis in naturally occurring autoimmune thyroiditis (19). HDG is recommended by American Thyroid Association guidelines for the management of thyrotoxicosis induced by destructive thyroiditis (20). Although ICI-induced thyroiditis typically manifests with less severe symptoms, the thyroid function test results can be extremely abnormal. It is therefore relevant to investigate the role of HDG in the management of ICI-induced thyroiditis. We evaluated the impact of HDG treatment on the outcome of ICI-related thyroid disorders. We found that HDG did not significantly alter the duration of thyrotoxicosis or the time to conversion from thyrotoxicosis to hypothyroidism, suggesting that HDG did not alter the evolution of ICI-induced thyroid disorders. There was no significant difference in pre-ICI treatment TSH between the HDG and no HDG groups. The severity of thyroid disorders was therefore not a confounding factor in determining the impact of HDG on the outcome of thyroid disorders. Our results suggest that HDG did not improve the outcome of ICI-related thyroid disorders. We therefore do not recommend routine use of HDG in patients who develop ICI-related thyroid disorders such as thyrotoxicosis.

The average maintenance dose of levothyroxine in this study was 1.29 μg/kg/day, which is lower than the maintenance dose needed in the athyroid patients (1.6–1.8 μg/kg/day; refs. 21–25). The current recommendation for the starting dose of levothyroxine is 1.6 μg/kg/day (11); based on our findings, we suggest revising the guidelines to start with a lower dose, 0.8 to 1.0 μg/kg/day. The maintenance dose of LT4 in the HDG and non-HDG groups was not significantly different, suggesting that HDG did not prevent or reduce ICI-induced thyroid damage.

Our findings that the pretreatment median TSH value was significantly higher in patients who developed ICI-related thyroid disorders than in patients without ICI-related thyroid disorders are unexpected. In subgroup analysis, only male patients who developed ICI-related thyroid disorders had significantly higher pretreatment TSH than control male patients, but there was no significant difference between female patients with or without ICI-related thyroid disorders. In our validation cohort of female patients with breast cancer, there was no significant difference in pre-pembrolizumab–treatment TSH between the patients with or without pembrolizumab-induced thyroid toxicity. Pending prospective studies in both male and female patients, the role of baseline TSH as a predictive biomarker for the development of ICI-related thyroid dysfunction remains to be determined.

Thyroid autoantibodies have been suggested as biomarkers to predict the development of ICI-related thyroiditis, but some studies do not support this suggestion (26–36). In the current study, we found that 42% of tested patients had positive TPOAb. Our study showed a trend of a higher incidence of TPOAb in anti–PD-1 monotherapy–induced thyroid disorders than in combination therapy, but the difference did not reach statistical significance. The role of thyroid autoantibodies in ICI-induced thyroid disorders remains uncertain. Future prospective, randomized clinical studies may help to clarify the uncertainty.

The symptoms of ICI-related thyrotoxicosis and hypothyroidism at first presentation were usually mild and nonspecific. Despite their mild symptoms, the biochemical tests usually showed a very high TSH of 31.5 (range, 2.4–187.8) mIU/L in hypothyroidism, and a very high FT4 of 2.8 (range, 1.0–7.7) ng/dL in thyrotoxicosis. As the thyroid receptor is a nuclear receptor that regulates gene expression by binding to hormone response elements (37), acute changes in thyroid hormone may take time to induce genomic change. Indeed, it takes more than a week for patients with hypothyroidism to feel better after initiating levothyroxine (38). During the acute period of rapid and marked changes in thyroid hormone, patients with ICI-induced thyroid disorders might not become symptomatic for a period of time if left uncorrected. If attention is not given to these patients early in the course of their thyroid disorder because of mild symptomatology, serious consequences such as life-threatening thyroid storm (39) or myxedema coma (40) can occur. Close monitoring and treatment with a beta-blocker in patients with thyrotoxicosis, as well as timely replacement with levothyroxine in patients with hypothyroidism, will help to prevent these serious consequences.

In summary, with increasing use of ICI therapy in a broad spectrum of malignancies, it is advantageous for clinicians to learn the diagnosis and management of ICI-induced thyroiditis, one of the most common irAEs. Based on this and previous studies (16, 26, 28, 31, 32, 34, 35), we recommend measuring TSH and FT4 before the initiation of ICI treatment and repeating them before each cycle of ICI, for at least 5 cycles. In our practice, supportive care with or without a beta-blocker is sufficient in most patients in thyrotoxicotic phase. We do not routinely use HDG treatment for thyrotoxicosis. We use HDG in patients who present with symptoms suggestive of thyroid storm such as tachycardia, fever, and mental status change, or in patients with cardiac disease. When patients develop hypothyroidism, we initiate levothyroxine replacement, usually starting with 0.8 to 1 μg/kg/day in young patients without cardiac disease, or with a lower dose of 25 to 50 μg/day in elderly patients (>60 years) or patients with cardiac disease, titrating the levothyroxine dose every 3 to 4 weeks based on the TSH and FT4. Finally, the role of pretreatment TSH as a useful biomarker to predict ICI-induced thyroiditis remains to be determined.

This study has limitations due to its retrospective design. To confirm our findings of the impact of high-dose corticosteroids on the outcome of ICI-induced thyroiditis, the maintenance dose of levothyroxine in the phase of hypothyroidism, as well as the association between pretreatment TSH and development of ICI-induced thyroiditis, future randomized prospective clinical trials will be necessary. Furthermore, the recommendations based on this retrospective study could be strengthened or revised by the future randomized prospective studies.

F.S. Hodi reports receiving commercial research grants from Bristol-Myers Squibb and Novartis, has patent royalties to institution per institutional policy (MICA-related disorders), is a consultant/advisory board member for Bristol-Myers Squibb, Merck, Aduro, Partners Therapeutics, Pfizer, Pionyr, Verastem, Torque, Compass Therapeutics, EMD Serono, Sanofi, Novartis, Takeda, Surface, Genentech, Apricity, and Bayer. R. Barroso-Sousa has received honoraria from the speakers bureau of Roche and is a consultant/advisory board member for Eli Lilly. S.M Tolaney reports receiving commercial research grants from AstraZeneca, Eli Lilly, Nanostring, Cyclacel, Merck, Nektar, Novartis, Pfizer, Genetech/Roche, Exelixis, Bristol-Myers Squibb, and Eisai, is a consultant/advisory board member for AstraZeneca, Eli Lilly, Puma, Sanofi, Tesaro, Celldex, Merck, Nektar, Novartis, Pfizer, Genentech/Roche, Immunomedics, Eisai, and Nanostring. L. Min reports receiving a commercial research grant from Bristol-Myers Squibb. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. Ma, L. Min

Development of methodology: C. Ma, A. Zhang, Y. Zhou, L. Min

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Ma, F.S. Hodi, J. Zhou, A. Zhang, Y. Zhou, C.P. Andrews, R. Barroso-Sousa, S.M. Tolaney, L. Min

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Ma, F.S. Hodi, A. Giobbie-Hurder, X. Wang, J. Hu, S.M. Tolaney, L. Min

Writing, review, and/or revision of the manuscript: C. Ma, F.S. Hodi, A. Giobbie-Hurder, A. Zhang, F. Mao, T.E. Angell, R. Barroso-Sousa, U.B. Kaiser, S.M. Tolaney, L. Min

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F.S. Hodi, C.P. Andrews, L. Min

Study supervision: S.M. Tolaney, L. Min

This work was supported by the NIH (K08 HD070957 to L. Min) and the National Natural Science Foundation of China (81300698 to C.J. Ma).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Hodi
FS
,
O'Day
SJ
,
McDermott
DF
,
Weber
RW
,
Sosman
JA
,
Haanen
JB
, et al
Improved survival with ipilimumab in patients with metastatic melanoma
.
N Engl J Med
2010
;
363
:
711
23
.
2.
Larkin
J
,
Chiarion-Sileni
V
,
Gonzalez
R
,
Grob
JJ
,
Cowey
CL
,
Lao
CD
, et al
Combined nivolumab and ipilimumab or monotherapy in untreated melanoma
.
N Engl J Med
2015
;
373
:
23
34
.
3.
Robert
C
,
Schachter
J
,
Long
GV
,
Arance
A
,
Grob
JJ
,
Mortier
L
, et al
Pembrolizumab versus ipilimumab in advanced melanoma
.
N Engl J Med
2015
;
372
:
2521
32
.
4.
Lee
L
,
Gupta
M
,
Sahasranaman
S
. 
Immune Checkpoint inhibitors: an introduction to the next-generation cancer immunotherapy
.
J Clin Pharmacol
2016
;
56
:
157
69
.
5.
Carbone
DP
,
Reck
M
,
Paz-Ares
L
,
Creelan
B
,
Horn
L
,
Steins
M
, et al
First-line nivolumab in stage iv or recurrent non-small-cell lung cancer
.
N Engl J Med
2017
;
376
:
2415
26
.
6.
Bins
S
,
van Meerten
E
,
Mathijssen
RH
. 
Nivolumab for squamous-cell cancer of head and neck
.
N Engl J Med
2017
;
376
:
595
6
.
7.
Naidoo
J
,
Page
DB
,
Li
BT
,
Connell
LC
,
Schindler
K
,
Lacouture
ME
, et al
Toxicities of the anti-PD-1 and anti-PD-L1 immune checkpoint antibodies
.
Ann Oncol
2015
;
26
:
2375
91
.
8.
Weber
JS
,
Yang
JC
,
Atkins
MB
,
Disis
ML
. 
Toxicities of immunotherapy for the practitioner
.
J Clin Oncol
2015
;
33
:
2092
9
.
9.
Barroso-Sousa
R
,
Barry
WT
,
Garrido-Castro
AC
,
Hodi
FS
,
Min
L
,
Krop
IE
, et al
Incidence of endocrine dysfunction following the use of different immune checkpoint inhibitor regimens: a systematic review and meta-analysis
.
JAMA Oncol
2017
;
4
:
173
82
.
10.
Villadolid
J
,
Amin
A.
Immune checkpoint inhibitors in clinical practice: update on management of immune-related toxicities
.
Transl Lung Cancer Res
2015
;
4
:
560
75
.
11.
Brahmer
JR
,
Lacchetti
C
,
Schneider
BJ
,
Atkins
MB
,
Brassil
KJ
,
Caterino
JM
, et al
Management of immune-related adverse events in patients treated with immune checkpoint inhibitor therapy: American Society of Clinical Oncology clinical practice guideline
.
J Clin Oncol
2018
;
36
:
1714
68
.
12.
Topalian
SL
,
Hodi
FS
,
Brahmer
JR
,
Gettinger
SN
,
Smith
DC
,
McDermott
DF
, et al
Safety, activity, and immune correlates of anti-PD-1 antibody in cancer
.
N Engl J Med
2012
;
366
:
2443
54
.
13.
Hamid
O
,
Robert
C
,
Daud
A
,
Hodi
FS
,
Hwu
WJ
,
Kefford
R
, et al
Safety and tumor responses with lambrolizumab (anti-PD-1) in melanoma
.
N Engl J Med
2013
;
369
:
134
44
.
14.
Larkin
J
,
Lao
CD
,
Urba
WJ
,
McDermott
DF
,
Horak
C
,
Jiang
J
, et al
Efficacy and safety of nivolumab in patients with BRAF V600 mutant and BRAF wild-type advanced melanoma: a pooled analysis of 4 clinical trials
.
JAMA Oncol
2015
;
1
:
433
40
.
15.
Ryder
M
,
Callahan
M
,
Postow
MA
,
Wolchok
J
,
Fagin
JA
. 
Endocrine-related adverse events following ipilimumab in patients with advanced melanoma: a comprehensive retrospective review from a single institution
.
Endocr Relat Cancer
2014
;
21
:
371
81
.
16.
Lee
H
,
Hodi
FS
,
Giobbie-Hurder
A
,
Ott
PA
,
Buchbinder
EI
,
Haq
R
, et al
Characterization of thyroid disorders in patients receiving immune checkpoint inhibition therapy
.
Cancer Immunol Res
2017
;
5
:
1133
40
.
17.
Shrestha
RT
,
Hennessey
J
. 
Acute and subacute, and Riedel's thyroiditis
.
In:
De Groot
LJ
,
Chrousos
G
,
Dungan
K
,
Feingold
KR
,
Grossman
A
,
Hershman
JM,
, et al
,
editors
.
Endotext
.
South Dartmouth, MA
:
MDText.com, Inc.
; 
2000
.
18.
Sriphrapradang
C
,
Bhasipol
A.
Differentiating Graves' disease from subacute thyroiditis using ratio of serum free triiodothyronine to free thyroxine
.
Ann Med Surg
2016
;
10
:
69
72
.
19.
Nikolai
TF
,
Coombs
GJ
,
McKenzie
AK
,
Miller
RW
,
Weir
GJ
 Jr
. 
Treatment of lymphocytic thyroiditis with spontaneously resolving hyperthyroidism (silent thyroiditis)
.
Arch Intern Med
1982
;
142
:
2281
3
.
20.
Ross
DS
,
Burch
HB
,
Cooper
DS
,
Greenlee
MC
,
Laurberg
P
,
Maia
AL
, et al
2016 American thyroid association guidelines for diagnosis and management of hyperthyroidism and other causes of thyrotoxicosis
.
Thyroid
2016
;
26
:
1343
421
.
21.
Mandel
SJ
,
Brent
GA
,
Larsen
PR
. 
Levothyroxine therapy in patients with thyroid disease
.
Ann Intern Med
1993
;
119
:
492
502
.
22.
Jonklaas
J
. 
Sex and age differences in levothyroxine dosage requirement
.
Endocr Pract
2010
;
16
:
71
9
.
23.
Olubowale
O
,
Chadwick
DR
. 
Optimization of thyroxine replacement therapy after total or near-total thyroidectomy for benign thyroid disease
.
Br J Surg
2006
;
93
:
57
60
.
24.
Gordon
MB
,
Gordon
MS
. 
Variations in adequate levothyroxine replacement therapy in patients with different causes of hypothyroidism
.
Endocr Pract
1999
;
5
:
233
8
.
25.
Roos
A
,
Linn-Rasker
SP
,
van Domburg
RT
,
Tijssen
JP
,
Berghout
A
. 
The starting dose of levothyroxine in primary hypothyroidism treatment: a prospective, randomized, double-blind trial
.
Arch Inter Med
2005
;
165
:
1714
20
.
26.
Morganstein
DL
,
Lai
Z
,
Spain
L
,
Diem
S
,
Levine
D
,
Mace
C
, et al
Thyroid abnormalities following the use of cytotoxic T-lymphocyte antigen-4 and programmed death receptor protein-1 inhibitors in the treatment of melanoma
.
Clin Endocrinol (Oxf)
2017
;
86
:
614
20
.
27.
Scott
E
,
Long
GV
,
Guminski
A
,
Clifton-Bligh
RJ
,
Menzies
AM
,
Tsang
VH
. 
The spectrum, incidence, kinetics, and management of endocrinopathies with immune checkpoint inhibitors for metastatic melanoma
.
Eur J Endocrinol
2017
;
178
:
173
80
.
28.
de Filette
J
,
Jansen
Y
,
Schreuer
M
,
Everaert
H
,
Velkeniers
B
,
Neyns
B
, et al
Incidence of thyroid-related adverse events in melanoma patients treated with pembrolizumab
.
J Clin Endocrinol Metab
2016
;
101
:
4431
9
.
29.
Guaraldi
F
,
La Selva
R
,
Sama
MT
,
D'Angelo
V
,
Gori
D
,
Fava
P
, et al
Characterization and implications of thyroid dysfunction induced by immune checkpoint inhibitors in real-life clinical practice: a long-term prospective study from a referral institution
.
J Endocrinol Invest
2017
;
41
:
549
56
.
30.
Yamazaki
H
,
Iwasaki
H
,
Yamashita
T
,
Yoshida
T
,
Suganuma
N
,
Yamanaka
T
, et al
Potential risk factors for nivolumab-induced thyroid dysfunction
.
In Vivo
2017
;
31
:
1225
8
.
31.
Delivanis
DA
,
Gustafson
MP
,
Bornschlegl
S
,
Merten
MM
,
Kottschade
L
,
Withers
S
, et al
Pembrolizumab-induced thyroiditis: comprehensive clinical review and insights into underlying involved mechanisms
.
J Clin Endocrinol Metab
2017
;
102
:
2770
80
.
32.
Orlov
S
,
Salari
F
,
Kashat
L
,
Walfish
PG
. 
Induction of painless thyroiditis in patients receiving programmed death 1 receptor immunotherapy for metastatic malignancies
.
J Clin Endocrinol Metab
2015
;
100
:
1738
41
.
33.
O'Malley
G
,
Lee
HJ
,
Parekh
S
,
Galsky
MD
,
Smith
CB
,
Friedlander
P
, et al
Rapid evolution of thyroid dysfunction in patients treated with nivolumab
.
Endocr Pract
2017
;
23
:
1223
31
.
34.
Yamauchi
I
,
Sakane
Y
,
Fukuda
Y
,
Fujii
T
,
Taura
D
,
Hirata
M
, et al
Clinical features of nivolumab-induced thyroiditis: a case series study
.
Thyroid
2017
;
27
:
894
901
.
35.
Alhusseini
M
,
Samantray
J.
Hypothyroidism in cancer patients on immune checkpoint inhibitors with anti-PD1 agents: insights on underlying mechanisms
.
Exp Clin Endocrinol Diabetes
2017
;
125
:
267
9
.
36.
Maekura
T
,
Naito
M
,
Tahara
M
,
Ikegami
N
,
Kimura
Y
,
Sonobe
S
, et al
Predictive factors of nivolumab-induced hypothyroidism in patients with non-small cell lung cancer
.
In Vivo
2017
;
31
:
1035
9
.
37.
Yen
PM
. 
Physiological and molecular basis of thyroid hormone action
.
Physiol Rev
2001
;
81
:
1097
142
.
38.
Vaidya
B
,
Pearce
SH
. 
Management of hypothyroidism in adults
.
BMJ
2008
;
337
:
a801
.
39.
Burch
HB
,
Wartofsky
L
. 
Life-threatening thyrotoxicosis. Thyroid storm
.
Endocrinol Metab Clin North Am
1993
;
22
:
263
77
.
40.
Fliers
E
,
Wiersinga
WM
. 
Myxedema coma
.
Rev Endocrine Metab Disord
2003
;
4
:
137
41
.