Neoadjuvant immunotherapy, chemotherapy, and combination therapy in muscle-invasive bladder cancer: A multi-center real-world retrospective study

Abstract

To parallelly compare the efficacy of neoadjuvant immunotherapy (tislelizumab), neoadjuvant chemotherapy (gemcitabine and cisplatin), and neoadjuvant combination therapy (tislelizumab + GC) in patients with muscle-invasive bladder cancer (MIBC) and explore the efficacy predictors, we perform a multi-center, real-world cohort study that enrolls 253 patients treated with neoadjuvant treatments (combination therapy: 98, chemotherapy: 107, and immunotherapy: 48) from 15 tertiary hospitals. We demonstrate that neoadjuvant combination therapy achieves the highest complete response rate and pathological downstaging rate compared with neoadjuvant immunotherapy or chemotherapy. We develop and validate an efficacy prediction model consisting of pretreatment clinical characteristics, which can pinpoint candidates to receive neoadjuvant combination therapy. We also preliminarily reveal that patients who achieve pathological complete response after neoadjuvant treatments plus maximal transurethral resection of the bladder tumor may be safe to receive bladder preservation therapy. Overall, this study highlights the benefit of neoadjuvant combination therapy based on tislelizumab for MIBC.

Graphical abstract

Figure 1

The flow diagram of patient selection NAC.NICB: combination of neoadjuvant chemotherapy and immunotherapy; NAC: neoadjuvant chemotherapy; NICB: neoadjuvant immunotherapy; TURBT: transurethral resection of the bladder tumor.

Figure 2

Developing and validating the pathological response prediction nomogram in the NAC.NICB cohort (A) The spearman correlations between four hematological indicators during the discovery stage. Brown color represents negative correlation, while blue color represents positive correlation. PLR: platelet-to-lymphocyte ratio. (B) The PLR.GHR level in responders and non-responders during the discovery stage. (C) Receiver operating characteristic (ROC) curves of five hematological indicators for predicting pathological response during the discovery stage. (D) The pathological response prediction nomogram during the discovery stage. (E) Bar plots show the distribution of response scores calculated by the nomogram in different response groups during the discovery stage. (F) ROC curves of clinical stage, PLR.GHR, and response scores for predicting pathological response during the discovery stage. (G) The calibration curves show that the predicted response sensitivity was consistent with the actual sensitivity during the discovery stage. (H) The decision curves highlight the highest net benefit of response score compared with clinical stage and PLR.GHR during the discovery stage. (I) ROC curves of clinical stage, PLR.GHR, and response score for predicting pathological response during the validation stage. (J) The calibration curves show that the predicted response sensitivity was consistent with the actual sensitivity during the validation stage. (K) The decision curves highlight the highest net benefit of response score compared with clinical stage and PLR.GHR during the validation stage.

Figure 3

Biomarker analysis of pathological response in the NAC.NICB cohort (A) GO and (B) KEGG analysis of differentially expressed genes between response and resistance samples. The enrichment score distribution patterns of (C) 19 immunotherapy efficacy associated signatures, (D) four stromal signatures, and (E) two NAC efficacy associated signatures between response and resistance samples.

Figure 4

Reclustering of bladder cancer cells in the NICB cohort (A) UMAP plot of subgroups of bladder cancer cells. (B–F) The most enriched functions of clusters 0–5. (G) UMAP plot of four bladder cancer subtypes. (H) Histogram indicating the proportions of four bladder cancer subtypes of NICB resistance and response groups.