# A multimodal biological margin risk index predicts recurrence after neoadjuvant immunochemotherapy in head and neck squamous cell carcinoma

## TL;DR

This study introduces a new Margin Risk Index that better predicts cancer recurrence after treatment by combining multiple biological factors in head and neck cancer patients.

## Contribution

A novel multimodal Margin Risk Index is developed to redefine surgical margin status after neoadjuvant immunochemotherapy in HNSCC.

## Key findings

- The Margin Risk Index outperformed traditional margin assessment in predicting locoregional control and distant metastasis-free survival.
- High-risk patients had significantly worse outcomes, with hazard ratios of 2.95 and 3.22 for recurrence and metastasis.
- The MRIx was validated in an independent cohort, confirming its prognostic accuracy and risk stratification ability.

## Abstract

Conventional classification of surgical margins is inadequate for head and neck squamous cell carcinoma (HNSCC) treated with neoadjuvant immunochemotherapy (NICT), as it fails to capture the complex biological changes in the tumor microenvironment. This study aimed to develop a novel definition of a negative margin.

We conducted a retrospective analysis of treatment-naïve, HPV-negative HNSCC patients who completed NICT followed by surgery. Surgical margins underwent multi-modal assessment, including histopathology (tertiary lymphoid structures), tumor burden (Pan-CK, Ki-67), molecular profiling (driver mutations, PD-L1 RNA), and immune contexture (CD8+/FoxP3+ ratio, Granzyme B). A Margin Risk Index (MRIx) was developed by weighting these domains based on their prognostic impact for locoregional control (LRC) and distant metastasis-free survival (DMFS). The MRIx was externally validated in an independent cohort.

The study included a training cohort of 144 patients and an independent validation cohort of 100 patients. The MRIx integrated four domains into a continuous score, stratifying patients into low, intermediate, and high-risk categories. The MRIx significantly outperformed traditional margin assessment, with superior discrimination for both LRC (C-index=0.72) and DMFS (C-index=0.75). External validation confirmed its prognostic power, demonstrating significant risk stratification (log-rank p<0.001 for both LRC and DMFS) and an independent hazard ratio for high-risk patients (HR = 2.95 for LRC; HR = 3.22 for DMFS, both p<0.001).

The proposed MRIx provides a biologically-grounded tool that redefines margin status following NICT. It accurately identifies patients at high risk of recurrence who may benefit from treatment intensification and those with low-risk margins suitable for de-escalation, enabling personalized adjuvant therapy.