# Predictive modeling of treatment-related thyroid dysfunction and prognostic implication in advanced nasopharyngeal carcinoma with PD-1 inhibitors

## TL;DR

This study shows that thyroid dysfunction during PD-1 inhibitor treatment for advanced nasopharyngeal cancer is common and linked to better survival, with a predictive tool and early biomarker identified.

## Contribution

A predictive nomogram and TSH as an early biomarker for thyroid dysfunction during PD-1 inhibitor therapy in nasopharyngeal cancer.

## Key findings

- TrTD occurred in 49.4% of patients and was associated with improved progression-free survival.
- A predictive nomogram combining clinical factors and thyroid indicators achieved 0.781 AUC for trTD prediction.
- TSH levels increased early, suggesting it as a potential biomarker for trTD onset.

## Abstract

Treatment-related thyroid dysfunction (trTD) frequently occurs in advanced nasopharyngeal carcinoma (NPC) patients treated with PD-1 inhibitors. Its clinical features, incidence, and prognostic value remain unclear. This study aimed to characterize trTD and assess its impact on survival.

We retrospectively analyzed 168 advanced NPC patients who received PD-1 inhibitor therapy between 2019 and 2022. Cox proportional hazards models and Kaplan–Meier analysis were used to assess the prognostic significance of trTD. Clinical factors associated with trTD were identified through logistic regression. A predictive nomogram was constructed and evaluated using receiver operating characteristic (ROC) and calibration curves. Spearman correlation analysis was performed to assess the dynamic relationship between thyroid function indicators and treatment duration.

TrTD developed in 49.4% of patients, mostly mild, including hyperthyroidism, hypothyroidism, and biphasic dysfunction. TrTD was independently associated with improved progression free survival (PFS). Female, elevated ALB, ALT, and TBIL were independent risk factors for trTD. The nomogram constructed by combining these factors and thyroid indicators had good prediction accuracy (AUC = 0.781). Dynamic analysis revealed that TSH levels significantly increased after the fourth treatment cycle, preceding changes in FT3 and FT4, suggesting TSH as a potential early biomarker for trTD onset.

This study identifies trTD as a common event associated with improved survival in advanced NPC patients treated with PD-1 inhibitors. A predictive nomogram and early TSH elevation provide valuable tools for risk stratification and personalized immunotherapy monitoring.

Graphical Abstract