Background: The possibility to predict the pathologic complete response (pT0) or pT≤1 after neoadjuvant immunotherapy may deeply impact the management of MIBC and orient clinical trials. The PURE01 study (Necchi A et al., J Clin Oncol 2018) evaluated preoperative pembrolizumab before radical cystectomy (RC) in T2-3bN0MO patients (pts). The trial is ongoing after an amendment to increase the sample size and includes cT4a pts.

Methods: Pts received 3 cycles of pembrolizumab, q3 weeks, prior to RC. Tumor samples from baseline tissue (TURB) and post-therapy tissue (RC) have been assayed. The reported analyses included hybrid capture-based comprehensive genomic profiling (CGP) to evaluate all classes of genomic alterations (GA). Tumor mutational burden (TMB) was determined on up to 1.1 Mbp of sequenced DNA; PD-L1 expression was assayed with immunohistochemistry and reported with the combined positive score (CPS, Dako 22C3 mAb). Pre- vs. post-therapy GA frequencies, and their association with pT0 response, have been compared via Fisher's exact tests. Logarithmic probabilities of pT0 were plotted according to the continuously coded value of the TMB. A nomogram against pT0 and pT≤1 response was developed that included clinical T-stage and baseline PD-L1 CPS (dichotomized as < or ≥10) plus TMB. Performance testing of the nomogram is herein reported in terms of discrimination (Harrell's c index).

Results: From 02/17 to 11/18, 87 pts were enrolled and completed treatment with RC. Selected baseline GA more frequently observed in nonresponders (pT≥2) vs. pT0 were: CCND1 (17.1 vs. 6.7%, p=0.27) and TSC1 (20 vs. 3.3%, p=0.06). Of note, GA of KRAS and SMARCA4 were not seen in pT0 pts but in 11.4% and 14.2% of pT≥2 pts instead, respectively. Increasing TMB values featured a linear association with pT0 probabilities (p=0.03). The c-indexes of the nomogram were 0.78 and 0.75 for pT0 and pT≤1 endpoints, respectively. Pts presenting with cT3-4 stage, CPS≥10, and TMB≥20 mut/mb had a median nomogram-derived probability of achieving pT0 = 79.5% and of achieving pT≤1 = 98%. These probabilities substantially decreased with lower TMB values (10 mut/mb cut-off) for cT3-4 stages, while for cT2 pts they were maintained as high as 76.3% and 98.5% for pT0 and pT≤1, respectively.

Conclusion: The first nomogram including key biomarker information, derived from the ongoing PURE-01 study, may be a suitable tool to select patients who deserve single-agent pembrolizumab as preoperative approach. Pending validation with additional prospective data from the ongoing trials, this tool is proposed with the aim of developing new personalized neoadjuvant immunotherapy studies. Key genomic biomarker features may characterize nonresponding residual disease and help refine postoperative targeted therapy strategies.

This abstract is also being presented as Poster B24.

Citation Format: Andrea Necchi, Marco Bandini, Daniele Raggi, Alberto Briganti, Elena Farè, Andrea Gallina, Renzo Colombo, Marco Bianchi, Maurizio Colecchia, Roberta Lucianò, Andrea Salonia, Filippo Pederzoli, Jeffrey S. Ross, Russell Madison, Francesco Montorsi, Siraj M. Ali, Jon H. Chung. Development of a biomarker-based nomogram to predict the pathologic response to neoadjuvant pembrolizumab in muscle-invasive urothelial bladder cancer (MIBC) [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2019 May 18-21; Denver, CO. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(15_Suppl):Abstract nr PR07.