Real world incidence, predictors and outcomes of endocrine immune-related adverse events following immune checkpoint inhibitors

Endocrine dysfunction is one of the most common immune-related adverse events (irAEs) reported in immune checkpoint inhibitors (ICIs) clinical trials. The aim of this research was to thoroughly assess the clinical features of endocrine irAEs, investigate the risk and predictive variables, and provide guidance for clinical therapy. This study performed a retrospective review of clinical data from 269 patients with malignant malignancies who underwent initial immune checkpoint inhibitor treatment at the First Bethune Hospital of Jilin University between May 2021 and October 2022.The incidence of endocrine irAEs was 31.6% (85/269), while the autoimmune polyendocrinopathy syndrome (APS) was observed in 1.1% (3/269). The median time for the first adverse reaction was 46 (33.5, 100.5) d. The occurrence of ICIs-related thyroid injury events was significant, with the incidence at 27.5% (74/269), whereas the frequency of grade 2 and higher adverse reactions was 33.8% (25/74). Female gender (OR = 2.723, 95 % CI: 1.447-5.125) and a history of chemotherapy (OR = 2.716, 95 % CI: 1.079-6.836) were distinct risk factors for thyroid injury associated with ICIs. In the 106/269 who had baseline antibody testing, the presence of anti-thyroglobulin antibodies (TGAb) (AUC = 0.717) and thyroid peroxidase antibody (TPOAb) (AUC = 0.690) could aid in predicting this injury, with TGAb demonstrating greater reliability. The prevalence of ICIs-related diabetes was 3.7% (10/269), with a higher occurrence in male patients compared to female patients, and the rate of grade 3 and above adverse reactions was 10% (3/10). The occurrence of ICIs-related pituitary inflammation was 1.1% (3/269), primarily involving pituitary hormones such as thyroid stimulating hormone (TSH) and adrenocorticotropic hormone (ACTH). The incidence of ICIs-related adrenocortical hypofunction was 0.4% (1/269), with a grade 3 adverse event. The antibody testing was performed in a nonrandom subset and thus the predictive AUC results might be affected by bias.