Immune checkpoint inhibitors (ICIs), including anti-cytotoxic T-lymphocyte antigen 4 (CTLA-4), anti-programmed cell death-1 (PD-1), and anti-programmed cell death-1 ligand 1 antibodies, have been applied for several advanced malignancies.1 However, ICIs can cause adverse events, termed immune-related adverse events (irAEs), including pneumonitis, skin toxicities, colitis, and endocrine dysfunction.2 3 There are two different types of pituitary irAEs, that is, isolated adrenocorticotropic hormone (ACTH) deficiency (IAD) without pituitary enlargement and hypophysitis with deficiency of multiple anterior pituitary hormones accompanied by pituitary enlargement.4 5 IAD can be induced by either anti-CTLA-4 or anti-PD-1 antibodies, whereas hypophysitis is mainly induced by anti-CTLA-4 antibodies.4 6 7 Although immune responses to the pituitary glands have been implicated in the pathogenesis of pituitary irAEs,8 9 the specific mechanisms remain unclear.

Anti-pituitary antibodies (APAs), measured by indirect immunofluorescence (IIF), are a surrogate marker of the presence of autoimmunity against pituitary glands10 and are detected with a higher frequency in cases of some pituitary diseases, particularly biopsy proven hypophysitis.11 Although APAs are negative at baseline, one study showed that APAs became positive at the onset of ipilimumab (anti-CTLA-4 antibody)-induced hypophysitis in all patients examined.9 In addition, APAs can be detected at the onset of IAD induced by combination therapy with nivolumab (anti-PD-1 antibody) and ipilimumab.12 Although these data indicate that APAs can become positive after the onset of ICI-induced IAD or hypophysitis, it remains unclear whether APAs are detectable before the onset of pituitary dysfunction.

Specific human leukocyte antigen (HLA) variants can be used as a marker of autoimmune diseases. Recent studies have shown that the frequencies of HLA-DQB1*06:01, HLA-DPB1*09:01, and HLA-DRB5*01:02 are higher in cases of anti-PD-1 antibody-induced IAD compared with those in healthy controls from a database of Japanese individuals.13 Another study reported that the frequencies of HLA-DR15, HLA-B52, and HLA-Cw12 were higher in patients with pituitary irAEs, including both hypophysitis and IAD, compared with those in healthy controls from a database of Japanese individuals.14 However, it is unclear whether the susceptible HLA alleles are different between ICI-induced IAD and hypophysitis.

Because pituitary irAEs are almost always accompanied by ACTH deficiency, which is life-threatening,8 it may be ideal to identify patients with a risk of pituitary irAEs before the initiation of ICI treatment. Accordingly, in this study, we aimed to examine whether APAs were present before the onset of pituitary irAEs and to analyze susceptible HLAs in patients with ICI-induced IAD and hypophysitis.

In this study, we showed that APAs were present at baseline in patients who developed ICI-induced IAD and that APAs became positive during ICI treatment in patients who developed hypophysitis. Our data also showed that susceptible HLA alleles overlapped but were not the same between patients with ICI-induced IAD and hypophysitis.

Anti-PD-1 antibodies, the main causative drugs of ICI-induced IAD (13 of 17) in this study, function mainly in the effector phase in which pre-existing autoreactive T cells are activated.21 22 This may explain the high prevalence of positive APAs at baseline in patients who developed ICI-induced IAD. We previously reported that the incidence of thyroid irAEs was significantly higher in patients who had anti-thyroid antibodies prior to ICI treatment than in those who did not.15 16 23 Similarly, our findings indicated that patients who had positive APAs at baseline were prone to develop ICI-induced IAD. In addition, positive conversion of APAs at the onset of pituitary irAEs was also observed in four cases who developed IAD during anti-PD-1 antibody treatment, suggesting that the humoral or cellular immune system was activated by ICI treatment. While there is no report showing histological features of pituitary glands in patients with ICI-induced IAD, lymphocytic infiltration in the anterior pituitary gland has been reported in patients with idiopathic IAD.24–26 Therefore, cytotoxic T cells specific to ACTH secreting cells may also be involved in the pathogenesis of ICI-induced IAD.

Interestingly, APAs had become positive in three of four patients who developed hypophysitis during ipilimumab administration and in whom sera were collected 2–3 weeks before the onset of hypophysitis. Because we could not detect APAs in sera collected from patients without pituitary irAEs at 12 weeks after the initiation of ipilimumab, our findings suggested that positive conversion of APAs occurred during ipilimumab treatment in patients with ipilimumab-induced hypophysitis. Anti-CTLA-4 antibodies, including ipilimumab, function in the priming phase, in which T cells are activated to recognize new antigens22; thus, this class of ICIs may provoke de novo development of autoimmune reactions against pituitary antigens in some patients, resulting in a positive conversion of APAs in patients with ICI-induced hypophysitis. Moreover, because T-cell infiltration has been observed as an irAE in involved organs, such as the skin, colon, and liver,27 cytotoxic T cells may contribute to the development of pituitary dysfunction. In addition to T-cell infiltration, an autopsy study also reported B-cell infiltration in the pituitary gland from a patient who developed hypophysitis induced by tremelimumab (an anti-CTLA-4 antibody).8 Taken together, it is likely that positive conversion of APAs may reflect the new development of autoimmunity in the pituitary glands.

Our data showed that ACTH-secreting cells were recognized by APAs in all patients with positive APAs not only at baseline but also at the onset of ICI-induced IAD. However, pituitary cells secreting other hormones (eg, TSH, FSH, LH, GH, and/or PRL) were also recognized by APAs at the onset of ICI-induced IAD in some patients. Because the functions of pituitary hormones other than ACTH were not clinically impaired in ICI-induced IAD, our findings may suggest that cytotoxic T cells rather than humoral immunity disrupted the ACTH-secreting cells.

Several autoantigen candidates have been reported by analyses using APAs as probes in pituitary diseases28–31; therefore, it is important to identify pituitary autoantigens targeted by APAs in patients with pituitary irAEs. These studies may lead to the establishment of a system to evaluate cytotoxic T-cell responses and an ELISA system to quantitatively measure APAs. Because it is not possible in current clinical practice to identify patients with a high risk of pituitary irAEs at baseline or after the initiation of ICI treatment, or alternatively when adrenal insufficiency is suspected, it is necessary to regularly measure ACTH and cortisol levels. Biomarkers for pituitary irAEs, once established, will allow identification of patients who had a high risk of pituitary irAEs before the onset of pituitary irAEs. While ACTH deficiency is life-threatening, we and other researchers have shown that the development of pituitary irAEs is associated with better overall survival if patients are treated appropriately with hormone substitution therapy.4 32 Therefore, a biomarker of pituitary irAEs could not only contribute to prediction of pituitary dysfunction but also lead to improved management of patients with cancer and a better prognosis following ICI treatment.

In addition to the presence of APAs, the association of specific HLA alleles with pituitary irAEs also strongly suggested the involvement of autoimmunity in the pathogenesis of pituitary irAEs. In this study, we found that susceptible HLA alleles overlapped but were not exactly the same between patients with ICI-induced IAD and hypophysitis. Among the susceptible HLA alleles evaluated in this study (HLA-Cw12, HLA-DR15, HLA-DQ7, and HLA-DPw9 for ICI-induced IAD and HLA-Cw12 and HLA-DR15 for ICI-induced hypophysitis), HLA-Cw12, HLA-DR15,14 and HLA-DPw913 were previously reported to be associated with pituitary irAEs. By analyzing the association of HLAs with the development of ICI-induced IAD and hypophysitis separately, we demonstrated that HLA-DQ7 was associated with the development of ICI-induced IAD, but not with hypophysitis. Notably, analyses from the previous two studies used a database of HLAs from a Japanese population as controls. Because some HLA alleles have been shown to be associated with the development of malignancies,33 the distribution of HLA frequencies may not be the same between ICI-treated patients with malignancies and healthy controls in the Japanese database. This study evaluated HLA frequencies in patients with malignancies who were treated with ICIs.

Clinical characteristics of ICI-induced hypophysitis are similar to those of lymphocytic hypophysitis (autoimmune hypophysitis).34 However, associations of different HLA alleles (HLA-DQ8 and HLA-DR53) with lymphocytic hypophysitis have been reported,35 suggesting that the pathogenesis may also be different between lymphocytic hypophysitis and ICI-induced hypophysitis.

This study had some limitations. First, the sample size of patients with ICI-induced hypophysitis was small. However, there was minimal selection bias because all patients who developed pituitary irAEs during the observation period were enrolled in the study, while patients in the control group were selected so that the ratios of malignancy type and drug usage were similar to those of the patients with pituitary irAEs. Second, the method for measuring APAs in this study is only available in a few laboratories owing to the requirement for normal human pituitary glands as the substrate. Accordingly, it is important to identify the autoantigens related to ICI-induced pituitary dysfunction in order to establish an ELISA system that quantitatively assesses APAs, thereby allowing measurement of APAs in a larger patient cohort in future studies.

Pituitary irAEs are not only life-threatening but also associated with better overall survival.4 32 Therefore, it is important to identify patients who may develop pituitary irAEs before initiation of treatment with ICIs. Our data showed distinct and overlapped patterns of APAs and HLA alleles between patients with ICI-induced IAD and hypophysitis and demonstrated that positive APAs at baseline and after treatment, together with susceptible HLA alleles, could be a predictive biomarker for ICI-induced IAD and hypophysitis, respectively.