Abstract
OBJECTIVE: Hematological indicators in the early stage of PD-1 inhibitor treatment may show superior predictive ability of the occurrence of immune related adverse event (irAE) compared to the pre-treatment indicators, as the immune response is modulated during the PD-1 inhibitor treatment. The objective of this study was to investigate the predictive capabilities of biomarkers in the early treatment stage for immune related thyroid dysfunction (irTD), and explore the potential predictive cytokines. METHODS: Medical records and blood test results of cancer patients treated with PD-1 inhibitor at a certain medical institution were collected. Logistic regression analysis was utilized to identify the predictive factors of irTD, ROC curves were plotted and the area under the curves (AUC) was calculated. Serum samples were collected before and during early treatment phase, cytokine detection was performed to explore potential predictive cytokines. RESULTS: A total of 264 patients were enrolled, 58 developed irTD (21.97%), including 31 patients with thyrotoxicosis and 27 with hypothyroidism. There were no significant differences in demographic characteristics, tumor types and PD-1 inhibitors between patients with and without irTD. Multivariate logistic analysis showed that anti-thyroglobulin antibody (TgAb) (OR=2.831, 95%CI: 1.077-7.443, P=0.035) and anti-thyroperoxidase antibody (TPOAb) (OR=9.565, 95%CI: 3.399-26.921, P=0.000) in the early treatment phase were independent predictive factors for irTD, the AUC of early-stage biomarkers was larger than that of pre-treatment (0.655 vs 0.571); low level of TSH at the early stage (OR=0.162, 95%CI: 0.077-0.341, P=0.000) was significantly correlated with thyrotoxicosis; female (OR=3.889, 95%CI: 1.457-10.380, P=0.007) and positive TPOAb (OR=8.678, 95%CI: 2.656-28.357, P=0.000) at the early stage were significantly correlated with hypothyroidism. The AUCs of early-stage biomarkers were larger than that of pre-treatment both in thyrotoxicosis (0.812 vs 0.637) and hypothyroidism patients (0.728 vs 0.710). The increase of IL-16 (adjusted P=0.004), IL-12p70 (adjusted P=0.014), IL-17 (adjusted P=0.014), CCL-15 (adjusted P=0.014) and IL-1a (adjusted P=0.021) in the early treatment phase were positively correlated with irTD. CONCLUSIONS: Biomarkers at the early stage of PD-1 inhibitor treatment could predict irTD, and demonstrated stronger predictive ability compared to pre-treatment biomarkers. IL-16, IL-12p70, IL-17, CCL-15 and IL-1a could serve as potential predictive biomarkers for irTD.