Impact of Tumour Stroma-Immune Interactions on Survival Prognosis and Response to Neoadjuvant Chemotherapy in Bladder Cancer
Libo Liu,Longhao Xu,Daqin Wu,Yingying Zhu,Xiaoyang Li,Chunru Xu,Ke Chen,Yi Lin,Jianwen Lao,Peicong Cai,Xuesong Li,Yun Luo,Xiang Li,Jian Huang,Tianxin Lin,Wenlong Zhong
DOI: https://doi.org/10.1016/j.ebiom.2024.105152
IF: 11.205
2024-01-01
EBioMedicine
Abstract:Background: The tumour stroma is associated with unfavourable prognosis in diverse solid tumours, but its prognostic and predictive value in bladder cancer (BCa) is unclear. Methods: In this multicentre, retrospective study, we included 830 patients with BCa from six independent cohorts. Differences in overall survival (OS) and cancer-specific survival (CSS) were investigated between high-tumour stroma ratio (TSR) and low-TSR groups. Multi-omics analyses, including RNA sequencing, immunohistochemistry, and single-cell RNA sequencing, were performed to study stroma-immune interactions. TSR prediction models were developed based on pelvic CT scans, and the best performing model was selected based on receiver operator characteristic analysis. Findings: Compared to low-TSR tumours, high-TSR tumours were significantly associated with worse OS (HR = 1.193, 95% CI: 1.046-1.361, P = 0.008) and CSS (HR = 1.337, 95% CI: 1.139-1.569, P < 0.001), and lower rate of pathological complete response (pCR) to neoadjuvant chemotherapy (NAC). High-TSR tumours exhibited higher infiltration of immunosuppressive cells, including Tregs and tumour-associated neutrophils, while low-TSR tumours exhibited higher infiltration of immune-activating cells such as CD8(+) Teff and XCR1(+) dendritic cells. The TSR prediction model was developed by combining the intra-tumour and tumour base radiomics features, and showed good performance to predict high-TSR, as indicted by area under the curve of 0.871 (95% CI: 0.821-0.921), 0.821 (95% CI: 0.731-0.911), and 0.801 (95% CI: 0.737-0.865) in the training, internal validation, and external validation cohorts, respectively. In patients with low predicted TSR, 92.3% (12/13) achieved pCR, while only 35.3% (6/17) of patients with high predicted TSR achieved pCR. Interpretation: The tumour stroma was found to be significantly associated with clinical outcomes in patients with BCa as a result of tumour stroma-immune interactions. The radiomics prediction model provided non-invasive evaluation of TSR and was able to predict pCR in patients receiving NAC for BCa.