External validation of the PATH classification in patients with recurrent head and neck squamous cell carcinoma treated with salvage surgery.

Abstract

Background and objective: The PATH classification, proposed in 2024, stratifies patients with recurrent head and neck squamous cell carcinoma (HNSCC) treated with salvage surgery into 4 stages using pathological variables routinely reported. The aim was to externally validate it in an independent cohort and to compare its prognostic performance with that of rpTNM staging.

Patients and methods: Retrospective observational study of 87 patients with recurrent HNSCC treated with salvage surgery with radical intent between 2014 and 2024. Disease-specific survival was estimated using the Kaplan-Meier method. Discrimination of both systems was compared using Harrell's concordance index (C-index) and the Akaike information criterion (AIC).

Results: Five-year disease-specific survival was 65.9% (PATH I), 92.3% (PATH II), 37.4% (PATH III) and 13.5% (PATH IV), with significant differences between stages (p = 0.003), although without an ordered gradient. The rpTNM system did not discriminate between its stages (p = 0.674). The PATH classification showed greater discrimination (C-index 0.670 vs. 0.572; AIC 291.0 vs. 304.1). The 48 patients in rpTNM stage IV were redistributed across all 4 PATH categories, with survival ranging from 85.7% to 15.2%. Margin status was the only variable with independent prognostic impact (70.0% vs. 17.8%; p < 0.001). In a post hoc analysis, grouping PATH I-II versus III-IV showed robust separation (74.4% vs. 31.7%; HR 3.57 in the adjusted multivariable model).

Conclusions: The PATH classification outperforms rpTNM staging in prognostic discrimination and reduces the heterogeneity of stage IV. The gradient across its 4 stages was not reproduced, its discriminative value being concentrated in the dichotomization defined by its first 2 hierarchical criteria and, predominantly, in margin status.