Beyond binary: a novel anatomical-geographical-biological framework for perineural invasion risk stratification in oral squamous cell carcinoma – a proof-of-concept study

Source abstract

Background Perineural invasion (PNI) is an adverse pathological feature in oral squamous cell carcinoma (OSCC) and an indication for adjuvant radiotherapy but is conventionally reported as a binary variable. The prognostic relevance of its quantitative morphological features remains incompletely defined. Materials and methods In this single-centre retrospective cohort, 101 consecutive patients with pathologically node-negative OSCC treated by curative-intent resection (January 2023 – March 2024) were analysed. Six block-level morphological PNI parameters were assessed on digitised H&E whole-slide images by two pathologists: largest involved nerve diameter (P1), maximum nerves involved (P2), pattern of invasion (P3; intratumoural, extratumoural or both), maximum circumferential extent (P4), nerves with mitotic figures (P5), and minimum distance from the tumour front to the nearest uninvolved nerve (P6). Interobserver reproducibility was evaluated for all five PNI parameters. Cutpoints were derived by maximally selected rank statistics. Univariate and bivariable Cox regression, bootstrap stability, and a composite 0–6 score were evaluated against disease-free survival (DFS) and locoregional control (LRC). Results PNI was identified in 34 of 100 evaluable patients; binary comparison showed a non-significant trend toward inferior DFS (p = 0.104) and no LRC difference (p = 0.490). Among 30 of the 34 PNI-positive patients with complete morphological data (11 DFS events; median follow-up 20.7 months), nerve diameter >429 µm was the strongest predictor of DFS (HR 3.96, 95% CI 1.18–13.02, p = 0.026), and the most consistently selected variable in LASSO bootstrap resampling (64.8%). Large nerve calibre (P1), an extratumoral component (P3) and intraneural mitoses (P5) co-occurred as an aggressive triad (pairwise φ 0.52–0.61). The composite 0–6 score was associated with DFS (HR 1.34 per point, 95% CI 1.01–1.77, p = 0.044) and stratified patients into low-risk (0–2) and high-risk (3–6) groups for DFS (p = 0.003) and LRC (p = 0.045). Conclusions PNI in node-negative OSCC is biologically significant and prognostically influential even within this favourable-prognosis (pN0, R0) subgroup; quantitative morphological characterisation, particularly of invaded nerve diameter, adds prognostic information beyond binary reporting. A composite morphological score may identify PNI-positive patients for future prospective evaluation of risk-adapted adjuvant strategies; external multicentre validation is required before clinical application.