Clinical biomarkers for thyroid immune-related adverse events in patients with stage III and IV gastrointestinal tumors.

ABSTRACT

Background: Thyroid immune-related adverse events (irAEs) associated with immune checkpoint inhibitor (ICI) treatment appear to correlate with a better prognosis. We aimed to investigate clinical biomarkers associated with thyroid irAEs. Methods: We retrospectively analyzed data from 129 patients receiving programmed cell death protein 1 (PD-1) inhibitors for stage III and IV gastrointestinal tumors. Patients were divided into two groups "thyroid irAEs" group and "no thyroid irAEs" group. We compared continuous variables using Mann-Whitney U and Kruskal-Wallis tests and categorical variables using Pearson's chi-square test. Survival curves were generated using the Kaplan-Meier method, and associations between clinical features and thyroid irAEs were assessed using univariate and multivariate logistic regression models. Associations for thyroid irAEs and outcomes [progression-free survival (PFS), overall survival (OS)] of the patients were performed with a Cox proportional hazard model. Results: A total of 129 patients, including 66 gastric cancer, 30 esophageal squamous cell carcinoma, and 33 hepatocellular carcinoma (HCC), were involved in this analysis with 47 cases of thyroid irAEs occurrence. The Cox proportional hazard model analysis confirmed the extended PFS [hazard rate (HR) = 0.447, 95% confidence interval (CI) 0.215 to 0.931, p = 0.031] and OS (HR = 0.424, 95% CI 0.201 to 0.893, p = 0.024) for thyroid irAEs group when compared with those of the no thyroid irAEs group. Association between thyroid irAEs and clinical characteristics at baseline was analyzed subsequently by univariate analysis. Higher body mass index (p = 0.005), increased eosinophil count (p = 0.014), increased lactate dehydrogenase (p = 0.008), higher baseline thyroid stimulating hormone (TSH) (p = 0.001), HCC (p = 0.001) and increased adenosine deaminase (ADA) (p = 0.001) were linked with thyroid irAEs occurrence. The multivariable logistic regression model indicated that ADA [odds rate (OR) = 4.756, 95% CI 1.147 to 19.729, p = 0.032] was independently associated with thyroid irAEs occurrence. Conclusions: Increased baseline level of ADA was associated with thyroid irAEs occurrence in patients with advanced gastrointestinal tumors who received ICI treatment. In the case of abnormal ADA, attention should be paid to the risk of thyroid irAEs.