Patients (n = 316) treated with ICIs (Tislelizumab, Pembroliumab, and Xindiliumab) at Qingdao Chengyang District People’s Hospital (2018–2023) were retrospectively analyzed. Based on thyroid irAEs post-ICI, they were stratified into euthyroid and irAEs groups, with the latter further subclassified into thyrotoxicosis (subclinical/clinical) and hypothyroidism (subclinical/clinical). The required sample size was estimated using Kendall’s rule (10 times the variable count). With 20 variables, the baseline sample size was 200; after a 10% buffer for invalid responses, the final target was ≥ 220 cases. Therefore, the inclusion of 316 participants was sufficient for the sample size. Inclusion criteria: The investigator received at least 1 ICIs treatment, and at least 2 or more thyroid hormone levels and TPOAb, TRAb and TgAb tests after initial treatment, with an interval of more than 2 weeks between each test. Exclusion criteria: Prior ICI use, pre-existing thyroid dysfunction, central hypothyroidism, incomplete data, or confounding factors (e.g., tyrosine kinase inhibitors, non-thyroidal illness syndrome, radiation thyroiditis). Ethical approval was obtained (No. 2023100812); informed consent was waived for retrospective data.

Statistical analyses were performed using SPSS 27.0 (IBM Corp.). Continuous variables were assessed for normality using Shapiro-Wilk tests. Normally distributed data were presented as mean ± standard deviation (SD) and compared using Student’s t-test, while non-normally distributed data were expressed as median (interquartile range, Q1-Q3) and analyzed with Mann-Whitney U test. Categorical variables were compared using χ ² test or Fisher’s exact test, as appropriate. To address potential type I error inflation in multiple comparisons, we applied Bonferroni correction when appropriate, with adjusted P-values reported for primary outcomes. Binary logistic regression analysis was conducted to identify independent risk factors for immune-related adverse events (irAEs), incorporating clinically relevant covariates and variables showing P < 0.10 in univariate analyses. Model assumptions were verified including absence of multicollinearity (variance inflation factors < 5) and adequate goodness-of-fit (Hosmer-Lemeshow test P > 0.05). A two-tailed P-value < 0.05 was considered statistically significant, except where multiple comparison adjustments were applied.

Table 1 presented a comparative analysis of baseline characteristics between the euthyroid group (n = 158) and the thyroid irAEs group (n = 158) in 316 cancer patients treated with immune checkpoint inhibitors (ICIs). No significant differences were observed in sex distribution (47.48% vs. 49.37% male, P = 0.282) or median age (66.0 vs. 62.5 years, P = 0.095) between groups. The majority received PD-1/PD-L1 inhibitors (100% in euthyroid vs. 92.41% in irAEs group), with combination CTLA-4/PD-1/PD-L1 therapy exclusively administered to the irAEs group (7.59%, P = 0.513 by Fisher’s exact test). Tumor types differed significantly (P = 0.032), with lung cancer predominating in the euthyroid group (78.48% vs. 58.23%) and gynecologic cancers more frequent in the irAEs group (12.02% vs. 1.90%).

Key thyroid-related parameters revealed lower baseline TSH levels in the irAEs group (median 1.40 vs. 1.92 mU/L, P = 0.023) and markedly higher thyroid autoantibody positivity (28.48% vs. 5.70%, P<0.001). Notably, combined TPOAb-TgAb positivity was prevalent in the irAEs group (14.56% vs. 1.90%). Among irAEs subtypes, hypothyroidism (55.06%) was more common than thyrotoxicosis (44.93%), with clinical hypothyroidism constituting 43.04% of cases.

Table 2 compared thyroid hormone profiles and autoantibody levels between the thyrotoxicosis (n = 71) and hypothyroidism (n = 87) subgroups of patients who developed thyroid immune-related adverse events (irAEs) following immune checkpoint inhibitor (ICI) therapy. The data reveal distinct immunological patterns between these two manifestations of thyroid dysfunction.

Patients with thyrotoxicosis demonstrated significantly lower thyroid peroxidase antibody (TPOAb) levels (11.78 ± 4.72 IU/mL vs 32.14 ± 7.97 IU/mL, P = 0.035) but markedly higher thyrotropin receptor antibody (TRAb) titers (2.56 ± 0.46 IU/L vs 0.24 ± 0.12 IU/L, P = 0.019) compared to the hypothyroidism group. These findings suggest different underlying pathophysiological mechanisms, with TRAb potentially driving the thyrotoxicosis phenotype while TPOAb predominates in hypothyroidism. Notably, thyroglobulin antibody (TgAb) levels showed no significant intergroup difference (56.23 ± 25.05 IU/mL vs 54.57 ± 26.48 IU/mL, P = 0.815).

As expected, thyroid-stimulating hormone (TSH) levels were profoundly suppressed in thyrotoxicosis (0.11 ± 0.02 mIU/L) versus elevated in hypothyroidism (4.21 ± 1.26 mIU/L, P < 0.001). Free thyroid hormones (FT3 and FT4) showed no statistically significant differences between groups, though the numerical values align with the clinical definitions of these conditions.

These results highlighted the importance of differentiating between thyrotoxicosis and hypothyroidism in ICI-related thyroid irAEs, as they exhibit distinct autoantibody profiles that may inform both pathogenesis and clinical management.

Table 3 presented the logistic regression analysis of risk factors for developing thyroid immune-related adverse events (irAEs) following immune checkpoint inhibitor therapy. The analysis identified pre-treatment thyroid autoantibodies as the most significant predictors of thyroid dysfunction, while demographic and other clinical factors showed no significant association.

Notably, thyroid peroxidase antibody (TPOAb) positivity demonstrated the strongest association with thyroid irAEs (OR=8.30, 95% CI: 1.88-37.23, P = 0.001), followed by thyrotropin receptor antibody (TRAb) positivity (OR=6.98, 95% CI: 1.82-32.72, P = 0.002). These findings suggest that pre-existing thyroid autoimmunity substantially increases susceptibility to ICI-induced thyroid dysfunction. In contrast, thyroglobulin antibody (TgAb) status (OR = 0.68, 95% CI: 0.32-1.52, P = 0.503), age (OR = 0.67, 95% CI: 0.42-1.46, P = 0.318), gender (OR = 0.81, 95% CI: 0.46-1.29, P = 0.612), and baseline TSH levels (OR = 0.58, 95% CI: 0.26-1.22, P = 0.471) were not statistically significant risk factors in this cohort.

These results emphasize the critical importance of pre-treatment thyroid autoantibody screening, particularly for TPOAb and TRAb, as these markers may help identify patients at highest risk for developing thyroid irAEs during ICI therapy. The differential predictive values of specific autoantibodies suggest distinct pathophysiological mechanisms underlying different forms of ICI-related thyroid dysfunction.