Locoregional Radiotherapy Candidates in de Novo Metastatic Nasopharyngeal Carcinoma: Real-World Insights in the Immunotherapy Era.

BACKGROUND

Chemotherapy combined with immunotherapy (CT-IO) has become the first-line treatment for de novo metastatic nasopharyngeal carcinoma (dmNPC). Locoregional radiotherapy (LRRT) following chemotherapy has been shown to significantly improve survival outcomes in patients with dmNPC. However, it remains unclear whether LRRT provides additional benefits in the context of CT-IO. Furthermore, there is no consensus on how to identify the optimal patient population for LRRT after first-line CT-IO.

METHODS

This study included patients with dmNPC who received platinum-based palliative chemotherapy and anti-PD-1 immunotherapy, with or without LRRT. Progression-free survival (PFS) was assessed in LRRT and non-LRRT groups using inverse probability of treatment weighting (IPTW) to mitigate selection bias. Median PFS (mPFS) at the 6-month landmark was estimated using Kaplan-Meier analyses. A novel prognostic nomogram was developed and validated to predict PFS and stratify patients by risk. Using prognostic scores from the nomogram, a model-based tree approach was employed to assess stratified treatment outcomes and identify the ideal candidates for LRRT.

RESULTS

A total of 500 patients were included, with 367 receiving LRRT and 133 not receiving it. At the 6-month conditional landmark, IPTW-adjusted Kaplan-Meier curves demonstrated significantly improved survival in the LRRT group compared with the non-LRRT group (mPFS, not reached vs 21.5 months; P<.001). Patients were randomized into training and validation cohorts in a 7:3 ratio. A prognostic model integrating serum lactate dehydrogenase (LDH) level, posttreatment Epstein-Barr virus DNA level, number of metastatic lesions, and liver metastases status was developed from the training cohort and graphically represented as a nomogram. The model demonstrated favorable discrimination (C-index, 0.721; 95% CI, 0.681-0.761) and predictive accuracy (1-year time-dependent area under the curve [tAUC]), 0.788), and its performance was validated in the internal cohort (C-index, 0.752; 95% CI, 0.698-0.806; 1-year tAUC, 0.778). A tree-based risk stratification derived from the model classified patients into 2 prognostic subgroups. Low-risk patients benefited from additional LRRT (mPFS, not reached vs 23.6 months; P<.001), whereas high-risk patients did not (mPFS, 18.3 vs 16.5 months; P=.210).

CONCLUSIONS

In patients with dmNPC, additional LRRT following first-line CT-IO was associated with improved PFS, particularly among low-risk patients identified using a novel prognostic model.