Transcriptome-informed multi-modal AI for predicting neoadjuvant therapy response from breast cancer biopsies

2026-10-02

AI Summary

This study addresses the scarcity of annotated oncological data that constrains deep learning-based biomarker development by proposing a two-stage multimodal AI framework. The approach first infers transcriptomic features from histopathology images and subsequently integrates clinical variables to predict pathological complete response to neoadjuvant therapy in breast cancer. Its core innovation lies in introducing a biologically informed compression strategy that overcomes the target selection constraints of conventional genomic assays, enabling robust generalization under data-sparse conditions while ensuring reliability through spatial consistency validation. Experimental results demonstrate that the model achieves a mixed AUROC of 0.79, outperforming traditional pathology-based biomarkers. Furthermore, it exhibits strong discriminative capacity across molecular subtypes and exceptional sampling robustness, highlighting its potential for precision oncology applications where labeled training data remain limited.

Abstract

Scarcity of labeled data limits development of deep learning biomarkers in oncology. We develop a two-stage AI model predicting pathological complete response (pCR) to neoadjuvant therapy in breast cancer. The first stage learns the transcriptome from histopathology using 8,742 patients across 32 cancer types, corroborated by pathologist review and spatial agreement with measured expression. This simplifies the second stage to predicting pCR from inferred expression and clinical variables. Developed using 1,080 patients (five cohorts) and evaluated in 1,412 patients (nine cohorts), the model achieves a pooled AUROC of 0.79 (95% CI, 0.73-0.85), discriminating responders within molecular subtypes. It outperforms histopathological biomarkers, remaining stable across intratumoral sampling and with minimal biopsy tissue. Ablations show transcriptome-wide inference improves discrimination over clinical variables alone or one-stage pathology models, and robustness by avoiding genomic assays' gene selection constraints. These results indicate that biologically informed compression may generalize to data-sparse applications in precision oncology.

Problem

• breast cancer
• neoadjuvant therapy
• pathological complete response
• labeled data scarcity
• deep learning biomarkers

Innovation

• multi-modal AI
• transcriptome inference
• pathological complete response
• two-stage model
• precision oncology