# Integrating single-cell regulatory atlas and multi-omics data for differential treatment response and multimodal predictive modeling in CDK 4/6 inhibitor-treated breast cancer

## TL;DR

This study combines single-cell and multi-omics data to identify a new biomarker for predicting response to CDK4/6 inhibitors in breast cancer patients.

## Contribution

The study introduces TPRI, a novel prognostic index based on five transcription factors for CDK4/6 inhibitor response prediction.

## Key findings

- TPRI stratified patients into high/low-risk groups with significant differences in survival and treatment response.
- High-risk patients showed increased stemness and mTOR pathway activation.
- A multimodal model combining TPRI, DEGs, and miRNAs achieved 0.704 AUC for prognostic accuracy.

## Abstract

CDK4/6 inhibitors are cornerstone therapies for advanced HR+/HER2- breast cancer, yet treatment response heterogeneity remains a major clinical challenge. This study integrates single-cell regulatory landscapes with multi-omics data to decode resistance mechanisms and develop predictive biomarkers for CDK4/6 inhibitor response stratification.

Single-cell RNA-seq data (GSE158724, n=14 samples) and bulk multi-omics profiles (TCGA-BRCA, n=1,059; GSE186901, n=90) were analyzed. Gene regulatory networks were reconstructed using SCENIC to identify resistance-specific regulons. The Tumor Prognostic Regulon Index (TPRI) was derived from five prognostic transcription factors and validated in independent cohorts. Experimental validation including qPCR of core TFs was performed in patient-derived samples. Multimodal predictive models integrating TPRI, differentially expressed genes, and miRNAs were developed using logistic regression, with performance assessed via ROC/AUC analysis.

We identified 86 resistance-associated regulons and established TPRI based on five prognostic TFs (ATF1, TEAD4, NFIL3, FOXO1, ETV3). TPRI significantly stratified patients into high/low-risk groups with differential overall survival and treatment response (Fisher’s exact test P=0.0237). qPCR confirmed elevated expression of these TFs in resistant tumors (P<0.01). High-risk patients exhibited increased stemness indices (mRNAsi, P<2.2e-16) and mTOR pathway activation. The multimodal model (TPRI + top 30 DEGs + top 30 miRNAs) achieved superior prognostic accuracy (95%CI:0.6575-0.75).

This study establishes TPRI as a novel biomarker for CDK4/6 inhibitor response prediction, validated through multi-omics integration and qPCR confirmation. The model provides actionable risk stratification, where high-risk patients may benefit from combinatorial mTOR-targeted therapies. Limitations include sample size constraints for methylation integration. Future studies should validate these findings in prospective clinical trials.