Patterns of hormonal changes in hypophysitis by immune checkpoint inhibitor

Data source and study sample

This single-center retrospective study analyzed electronic records data from patients managed at Asan Medical Center between January 1, 2017, and June 30, 2021. All adult patients (aged ≥ 20 yr) with solid cancer and endocrine immune-related adverse events (irAEs) who started immunotherapy were screened for inclusion in the analysis. The types of ICI used at our hospital included an anti-CTLA-4 inhibitor (ipilimumab), anti-PD-1 inhibitors (nivolumab and pembrolizumab), and anti-PD-L1 inhibitors (atezolizumab, avelumab, and durvalumab). Patients diagnosed with any solid cancer (for example—lung cancer, esophageal/gastric cancer, liver cancer, biliary tract cancer, or kidney cancer) but not hematological malignancies were eligible.

Patients included in the final analysis were those with available data on thyroid and adrenal function tests, as well as on gonadotropin or growth hormone function, following the commencement of ICI therapy. We excluded patients who did not undergo any hormone tests after the initiation of chemotherapy, those who received only one dose of immunotherapy, and patients with underlying pituitary disease. Finally, 21 patients were included in the analysis (Fig. 1). Patients who used a tyrosine kinase inhibitor were excluded only from the thyroid function analysis. Moreover, patients who received glucocorticoid treatment for other reasons–such as immune-related hepatitis, drug-induced pneumonitis, or post-radiotherapy brain edema–before starting ICI therapy were excluded from the adrenal function analysis. We excluded patients whose gonadal dysfunction occurred more than 18 months after beginning ICI therapy, as well as those among whom central hypogonadism could not be determined due to menopause.

Figure 1.

Patient selection flowchart. ICI, immune checkpoint inhibitor; GH, growth hormone.

This study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (IRB) of Asan Medical Center (IRB No. 2021-0965). The requirement for informed consent was waived by the IRB because de-identified data were used in the analyses. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cohort studies.

Clinical and laboratory measurements

Patient data were collected through a review of electronic medical records. Body mass index was calculated as weight in kilograms divided by height in square meters. Blood pressure was measured using an automatic manometer with a cuff placed around the patient’s right upper arm after resting for 5 minutes. Blood samples were drawn into vacuum tubes from antecubital veins and analyzed in Asan Medical Center’s accredited central laboratory. Laboratory serum measurements included fasting plasma glucose, total cholesterol, and C-reactive protein (CRP) levels. Blood samples for the measurement of fasting plasma glucose and total cholesterol levels were collected after overnight fasting (after 8 h of fasting for plasma glucose and after 12 h of fasting for total cholesterol). Fasting plasma glucose and CRP were measured using enzymatic colorimetric and immunoturbidimetric methods, respectively, on a Toshiba 200FR autoanalyzer (Toshiba Medical System Co. Ltd., Tokyo, Japan) [16]. Fasting total cholesterol was measured using enzymatic colorimetric methods on a Toshiba 200FR Neo analyzer [16]. The reference range for each hormone test is listed in Supplementary Table 1.

Diabetes mellitus was defined according to the following criteria: (1) at least one claim per year for the prescription of antidiabetic medication; (2) fasting glucose level ≥ 126 mg/dL; or (3) HbA1c (glycated hemoglobin) ≥ 6.5%. Hypertension was defined as the presence of at least one claim per year for the prescription of antihypertensive medication or a systolic/diastolic blood pressure measurement ≥ 140/90 mmHg. Hypercholesterolemia was defined as a total cholesterol level ≥ 240 mg/dL or treatment with lipid-lowering medication [17].

Definition of irH and non-hypophysitis endocrine irAEs

The definition of hypopituitarism was adapted from the Endocrine Society Clinical Practice Guidelines and the description in the relevant position statement published by the Korean Endocrine Society [8]. Based on biochemical evaluation of the hypothalamic-pituitary axis, hypopituitarism is defined as the presence of one or more of the following hormonal deficiencies: central hypothyroidism, central adrenal insufficiency, central hypogonadism, or growth hormone deficiency [8,18]. Although there is some variation in the details between studies, hypophysitis is generally defined as the presence of at least one of the following: evidence of hypopituitarism as assessed by biochemical testing, clinical symptoms (fatigue, headache, anorexia, nausea, or vomiting), hyponatremia that may be seen in hypophysitis, or pituitary imaging findings suggestive of hypophysitis [7,19-21]. Owing to the retrospective nature of this study, clinical manifestations and imaging findings were nonspecific and unclear as diagnostic criteria. Therefore, irH was defined as a deficiency of one or more hormones following ICI treatment [22]. Patients were classified into the irH group if they exhibited any of the following conditions: central hypothyroidism, central adrenal insufficiency, central hypogonadism, or growth hormone deficiency.

Central hypothyroidism is diagnosed when thyroid-stimulating hormone (TSH) levels are low, normal, or slightly elevated (TSH < 10 mU/L) and free thyroxine (fT4) levels are below the laboratory reference range [18,23]. Primary hypothyroidism was defined as TSH > 10 mU/L or TSH above the upper limit of normal and fT4 below the lower limit of normal [22]. Distinctions were made between immune-related hypothyroidism, thyroiditis, and central hypothyroidism. Immune-related hypothyroidism was characterized by a TSH level > 10 mU/L immediately after ICI administration. Immune-related thyroiditis was indicated by initial hyperthyroidism after ICI administration, followed sequentially by hypothyroidism.

Central adrenal insufficiency was defined as a morning cortisol level < 3 μg/dL between 8:00 and 9:00 AM or a peak cortisol level < 18 μg/dL (< 500 nmol/L, on radioimmunoassays) at 60 minutes on a corticotropin stimulation test when the adrenocorticotropic hormone (ACTH) level did not exceed the normal range [18]. The distinction between central hypothyroidism and sick thyroid syndrome, along with the distinction between central adrenal insufficiency and adrenal insufficiency caused by other factors (opioids or megestrol acetate), was determined by the existence of coexisting pituitary hormone deficiencies, such as central hypogonadism and growth hormone deficiency [7]. Growth hormone deficiency was defined as an age-specific serum insulin-like growth factor 1 level lower than the lower limit of normal [24,25].

Central hypogonadism was defined as low levels of sex hormones (testosterone in men and estradiol in women) and low (or inappropriately normal) gonadotropin (follicle-stimulating hormone [FSH] and luteinizing hormone [LH]) levels in the absence of androgen deprivation therapy or hormone replacement therapy. In postmenopausal women, the absence of elevated serum FSH and LH levels indicates gonadotropin dysfunction [18].

Non-hypophysitis endocrine irAEs were defined in patients who did not meet the criteria for irH but exhibited endocrine abnormalities [10].

Outcome variables

Thirteen patients in the irH group were analyzed in terms of the type of cancer diagnosed, type of ICI therapy, dosage, treatment intervals, cycles, symptoms, affected hormonal axes, and associated non-hypophysitis endocrine irAEs, and magnetic resonance imaging (MRI) findings. Among all participants who were included in the analysis, baseline characteristics were analyzed by stratifying participants into an irH group and a non-irH group. For each group, age, sex, body mass index, fasting glucose, total cholesterol, CRP, and underlying diseases (diabetes mellitus, hypertension, or hyperlipidemia) were investigated. Additionally, the type of cancer diagnosed in patients with irH, the type of ICI used, and the dosing cycle of ICI were investigated.

Within the irH group, the proportions of each type of hormonal deficiency were analyzed across the affected hormonal axes. Furthermore, the cumulative incidence and time to onset of irH were investigated. Twenty-one participants underwent hormone level quantification at baseline prior to ICI treatment initiation, at the onset of endocrine irAEs (including irH), or at the end of follow-up. Notable trends in TSH and fT4 levels were observed from the start of ICI therapy to irH onset, with comparisons between the irH and non-irH groups.

Statistical analysis

Descriptive statistics were used to analyze the patients’ baseline characteristics, cancer types, ICI use, and the frequency of hormonal axis deficiencies. The analysis encompassed all patients from the initiation of chemotherapy, spanning from January 1, 2017, to June 30, 2021. We compared the clinical characteristics of patients treated with ICI, with and without irH. For baseline characteristics, continuous variables are presented as median (interquartile range), and categorical variables are presented as frequency (proportion). Demographic and biochemical characteristics according to irH status were compared using the Mann–Whitney U test for continuous variables and Fisher’s exact test for categorical variables.

Cumulative bar graphs and Venn diagrams were used to depict the proportions of affected hormonal axes in patients with irH. The Kaplan–Meier method was used to analyze the cumulative incidence and time to onset of irH. We evaluated the trend changes in TSH and fT4 levels in relation to time as a line graph by setting the start date of ICI administration for each patient as zero and the duration from this time point to the date of irH onset in each patient on the time axis.

Statistical analyses were performed using SPSS Statistics for Windows, version 28.0 (IBM Corp., Armonk, NY, USA) and R, version 4.3.1 Foundation for Statistical Computing, Vienna, Austria. A p value < 0.05 was considered significant.

Baseline characteristics of patients in the irH and non-irH groups

A total of 2,104 adult patients with solid tumors who received immunochemotherapy during the study period underwent at least one pituitary hormone test. After applying the exclusion criteria and removing those without the necessary hormone test results, 21 patients were included in the final analysis. Of the 2,104 patients screened, 896 received nivolumab (an anti-PD-1 inhibitor) alone, 733 received pembrolizumab, 416 received atezolizumab (anti-PD-L1 inhibitor), 18 received ipilimumab (anti-CTLA4 inhibitor) alone, 46 received nivolumab plus ipilimumab combination therapy, 44 received durvalumab (anti-PD-L1 inhibitor), and 1 received pembrolizumab and ipilimumab combination therapy. Twenty patients received multiple ICI agents sequentially: 10 with atezolizumab, one with ipilimumab, 12 with nivolumab, three with nivolumab plus ipilimumab, and 15 pembrolizumab.

Of the 21 patients included in the analysis, 13 had hormone test results suggesting the possibility of irH, and eight did not show any evidence of hypophysitis. Thus, irH was diagnosed in 13 of the 21 patients (61.9%) in our study. Table 1 lists the characteristics of all patients with irH. In the irH group, 11 patients exhibited at least one manifestation of hypophysitis, such as anorexia, nausea, vomiting, fatigue, dizziness, weight loss, headache, or general weakness. Two patients exhibited hyponatremia, which is a potential indicator of hypophysitis. On MRI, two patients had abnormal pituitary gland findings (Supplementary Fig. 1).

Table 1.

Patients with immune-related hypophysitis

The median ages in the irH and non-irH groups were 59.0 and 57.0 years, respectively. The irH group included nine males (69.2%), while the non-irH group included five males (62.5%). Of the irH patients, 12 (92.3%) were treated with anti-PD-1 inhibitors: nivolumab in seven patients (53.8%) and pembrolizumab in five patients (38.5%). One patient (7.7%) received nivolumab and ipilimumab combination therapy, and none received atezolizumab. In the non-irH group, seven patients (87.5%) used anti-PD-1 inhibitors (n = 4, 50% for nivolumab and n = 3, 37.5% for pembrolizumab). None of the patients used a combination therapy of nivolumab and ipilimumab, and only one patient (12.5%) used atezolizumab.

Hormonal deficiencies in irH patients: analysis of affected axes, cumulative incidence, and diagnostic timing

Table 2 lists the laboratory hormone test values at the initiation of ICI therapy and at the time of irH diagnosis for all subjects, assessing deficiencies across each hormonal axis based on the reference ranges.

Table 2.

Hormone laboratory values at baseline and at the onset of endocrine irAEs (including irH) or the end of follow-up

Table 3 shows the proportion of patients who were deficient in each hormone type. Among the 13 patients with irH, three had central hypothyroidism, whereas five had primary hypothyroidism. ACTH levels and morning cortisol or peak cortisol levels at 60 minutes on a corticotropin stimulation test were measured in all irH patients to evaluate the function of the ACTH axis. Among these, 10 (76.9%) had central adrenal insufficiency. No patients with primary adrenal insufficiency were identified. Eight patients (61.5%) had central hypogonadism, and two (15.4%) had growth hormone deficiency. Panhypopituitarism was observed in two patients (15.4%).

Table 3.

Proportions of hormonal deficiencies by axis (TSH, ACTH, gonadotropin, and GH) among patients with hypophysitis (irH group)

During a follow-up period of 15.2 person-years, 13 patients developed irH. The overall incidence of irH was approximately 0.9 per person-year. Kaplan–Meier analysis indicated that the cumulative incidence of irH was 38.8% at 6 months and 57.1% at 1 year. The median time from the first ICI dose to the diagnosis of irH was 7.7 months (Fig. 2).

Figure 2.

Kaplan–Meier plot depicting the cumulative incidence of immune-related hypophysitis over time, with a median onset of 7.7 months after the initiation of immune checkpoint inhibitor therapy.

TSH axis trends from immunotherapy start to hypothyroidism onset

Trends in TSH and fT4 values from ICI initiation to hypothyroidism onset were reviewed in participants whose TSH and fT4 levels were measured. One patient in the non-irH group had overt hypothyroidism associated with sunitinib administration before ICI treatment initiation and a history of levothyroxine treatment; therefore, his thyroid function test results were excluded from the analysis.

Hormonal abnormalities appeared earlier in the irH group than in the non-irH group. In the irH group, eight of the 13 patients received levothyroxine replacement. For these eight patients, the median duration for levothyroxine replacement initiation after ICI commencement was 2.7 months. Specifically, the median time to begin levothyroxine replacement for central hypothyroidism was 3.8 months among these patients. Of the seven patients in the non-irH group, two received levothyroxine replacement for primary hypothyroidism, and the median time until replacement initiation was 7.1 months (Fig. 3). The cumulative incidence of levothyroxine replacement events was higher in the irH group than in the non-irH group, but the difference was not statistically significant (p = 0.01) (Supplementary Fig. 2).

Figure 3.

Changes in TSH and free T4 levels and timing of levothyroxine replacement according to the immune checkpoint inhibitor administration period (irH group [A] vs. non-irH group [B]). TSH, thyroid-stimulating hormone; T4, thyroxine; irH, immune-related hypophysitis; non-irH, non-immune-related hypophysitis.