# Multi-omics prediction of axillary treatment response and tumour microenvironment alterations in lymph node-positive luminal breast cancer

## TL;DR

This study develops a multi-omics model to predict treatment response in lymph node-positive breast cancer, potentially reducing unnecessary surgery.

## Contribution

A novel multi-omics model is introduced to predict pathological lymph node complete response after chemotherapy in luminal breast cancer.

## Key findings

- A multi-omic model with 12 radiomic and four clinicopathological features achieved 0.853 and 0.805 AUC in training and validation cohorts.
- Single-cell RNA sequencing revealed immunosuppressive features in patients with poor neoadjuvant chemotherapy response.
- The model could help reduce unnecessary axillary surgery by predicting treatment response in metastatic lymph nodes.

## Abstract

Luminal breast cancer (BC) with axillary lymph node (ALN) metastasis is typically treated with neoadjuvant chemotherapy (NAC). Theoretically, patients who achieve pathological lymph node complete response after NAC can be exempted from ALN dissection and even have the possibility of being spared axillary surgery. However, there is no effective way to preoperatively assess whether a metastatic ALN achieved pathological lymph node complete response (pLCR) after NAC. Therefore, we retrospectively collected imaging, clinical, and pathological data from two centres, built a multi-omic model to predict pLCR, and validated its accuracy and clinical applicability. We identified 12 radiomic and four clinicopathological features for model construction; the areas under the curve for training and validation cohorts were 0.853 and 0.805, respectively. Subsequently, single-cell RNA sequencing analysis was conducted on patients with different efficacy and its association with the tumour immune microenvironment was investigated. Eleven cell clusters in 14 samples from five patients were identified with differing NAC responses; comparative analysis indicated that those with poor responses had immunosuppressive features, which provided a theoretical basis for elucidating the resistance mechanism of NAC in axillary metastatic lymph nodes. The multi-omics prediction model demonstrated good performance in predicting ALN status after NAC, offering the possibility of reducing unnecessary axillary surgery.