Purpose: Bladder preservation is a viable option for some patients with muscle-invasive bladder cancer (MIBC), but an effective noninvasive biomarker test to accurately identify promising candidates is lacking. Here we present the clinical application of a novel tissue-agnostic, urine-based minimal residual disease (MRD) assay in the neoadjuvant setting for personalized disease surveillance and actionable target identification to facilitate bladder-sparing treatment approaches. Patients and Methods: The urinary tumor DNA (utDNA) analysis was evaluated in an investigator-initiated phase I trial RJBLC-I2N003 in which 20 patients diagnosed with resectable MIBC were treated presurgically with the PD-1 inhibitor toripalimab followed by radical cystectomy (RC). Results: We showed that neoadjuvant toripalimab therapy was feasible, safe, and induced a 40% rate (8/20) of pathologic complete response. Longitudinal utDNA profiling outperformed radiographic assessment and conventional biomarkers to predict the pathologic outcome of immune checkpoint blockade. In addition to detecting 3 exceptional responders with molecular MRD-negative status, we identified 7 other individuals characterized for utDNA response and 4 harboring FGFR3 mutants, all of whom (60%, 12/20) could have postponed or avoided RC. Conclusions: These findings demonstrate the safety and efficacy of neoadjuvant toripalimab, and suggest the immense potential of noninvasive utDNA MRD testing to guide tailored decision-making with regard to bladder preservation and change the current treatment paradigm for patients with MIBC.

Purpose: Neuroendocrine bladder cancer (NEBC) poses a formidable clinical challenge and attracts keen interests to explore immunotherapy as a viable treatment option. However, a comprehensive immunogenomic landscape has yet to be thoroughly investigated. Experimental Design: Leveraging a long-term cohort of natural NEBC cases, we employed a multimodal approach integrating genomic (n = 19), transcriptomic (n = 3), single-cell RNA sequencing (n = 1), and IHC analyses (n = 34) to meticulously characterize the immunogenicity and immunotypes of primary NEBC tumors. Information on clinical, pathologic, medical imaging, and treatment aspects was retrospectively retrieved and analyzed. Results: Our study unveiled that despite a considerable mutational burden, NEBC was typically immunologically inactive, as manifested by the “immune-excluded” or “immune-desert” microenvironment. Interestingly, a subset of mixed NEBC with concurrent urothelial bladder cancer histology displayed an “immune-infiltrated” phenotype with prognostic relevance. When compared with urothelial bladder cancer, NEBC lesions were distinguished by a denser cellular composition and augmented peritumoral extracellular matrix, which might collectively impede lymphatic infiltration. As a result, single-agent immune checkpoint inhibitors demonstrated limited efficacy against NEBC, whereas pharmacologic immunostimulation with combination chemotherapy conferred a more favorable response. Conclusions: These new insights derived from genomic profiling and immune phenotyping pave the way for rational immunotherapeutic interventions in patients with NEBC, with the potential to ultimately reduce mortality from this otherwise fatal disease.

Table S1. Clinicopathological characteristics of 20 patients in our study. Table S2. Representativeness of study participants. Table S3. Adverse events associated with toripalimab. Table S4. SNVs and InDels of therapy-naïve neoplastic tissues detected by whole exome sequencing. Table S5. SNVs and InDels of basal urine samples detected by whole exome sequencing. Table S6. SNVs and InDels detected by personalized MRD panel sequencing. Table S7. SNVs and InDels detected by the fixed actionable/hotspot panel sequencing. Table S8. Mutational measurements of each urine/plasma sample in our study.

Supplementary methods and clinical trial protocol

Purpose: Bladder preservation is a viable option for some muscle-invasive bladder cancer (MIBC) patients, but an effective noninvasive biomarker test to accurately identify promising candidates is lacking. Here we present the clinical application of a novel tissue-agnostic, urine-based minimal residual disease (MRD) assay in the neoadjuvant setting for personalized disease surveillance and actionable target identification to facilitate bladder-sparing treatment approaches. Patients and Methods: The urinary tumor DNA (utDNA) analysis was evaluated in an investigator-initiated phase I trial RJBLC-I2N003 in which 20 patients diagnosed with resectable MIBC were treated presurgically with the PD-1 inhibitor toripalimab followed by radical cystectomy (RC). Results: We showed that neoadjuvant toripalimab therapy was feasible, safe, and induced a 40% rate (8/20) of pathological complete response. Longitudinal utDNA profiling outperformed radiographic assessment and conventional biomarkers to predict the pathologic outcome of immune checkpoint blockade. In addition to detecting 3 exceptional responders with molecular MRD-negative status, we identified 7 other individuals characterized for utDNA clearance and 4 harboring FGFR3 mutants, all of whom (60%, 12/20) could have postponed or avoided RC. Conclusions: These findings demonstrate the safety and efficacy of neoadjuvant toripalimab, and suggest the immense potential of noninvasive utDNA MRD testing to guide tailored decision-making with regard to bladder preservation and change the current treatment paradigm for MIBC patients.

Figure S1. (A) Representative MRI images for radiographic evaluation of neoadjuvant toripalimab. (B) Representative hematoxylin and eosin staining for histopathological evaluation of neoadjuvant toripalimab. CR = complete response; PR = partial response; SD = stable disease; PD = progressive disease. A modifier “p” refers to pathologic staging after cystectomy. Figure S2. (A) Stacked bar plot shows the percentage of patients with negative or positive PD-L1, low or high TMB, and negative or positive TLS. (B) Oncoprint chart for the mutational landscape of tDNA in patients with ypCR or non-ypCR. Samples were analyzed by whole exome sequencing, and the mutation frequencies of each gene are shown on the right. (C) The line plot illustrates tumor size changes measured by MRI imaging before and after neoadjuvant toripalimab. PD-L1 = programmed death ligand 1; TMB = tumor mutation burden; TLS = tertiary lymphoid structure; ypCR = pathological complete response; pre-tx = pre-treatment; post-tx = post-treatment; MRI = magnetic resonance imaging. Figure S3. (A) Oncoprint chart for the mutational landscape of tDNA and utDNA. Samples were analyzed by whole exome sequencing, and the mutation frequencies of each gene are shown on the right. (B) TMB correlation between tDNA and utDNA as assessed by whole exome sequencing. Red shading indicates 95% confidence interval. Spearman correlation coefficient (r) and P value are shown. tDNA = tumor DNA; utDNA = urinary tumor DNA; TMB = tumor mutation burden. Figure S4. (A) Box plot compares TFsm in utDNA versus ctDNA samples collected at baseline. (B) Venn plots show the number of shared and unique variants in matched utDNA and ctDNA samples. (C) Box plot compares TFcn in utDNA versus ctDNA samples at baseline. (D) Copy number gain (red) and loss (blue) of utDNA and ctDNA, as identified by GISTIC2.0. TFsm = tumor fraction estimate based on somatic mutations; utDNA = urinary tumor DNA; ctDNA = circulating tumor DNA; TFcn = tumor fraction estimate based on copy numbers. Figure S5. (A) The line plot illustrates TFsm changes in utDNA samples upon neoadjuvant toripalimab. (B) The line plot illustrates TFcn changes in utDNA samples before and after neoadjuvant toripalimab. (C) Stacked bar plot showed the percentage of patients with low or high pre-treatment TFsm, TFcn, and MRI measurements according to the optimal cutoff points defined by ROC analysis. (D) Stacked bar plot showed the percentage of patients with low or high post-treatment TFsm, TFcn, and MRI measurements according to the optimal cutoff points defined by ROC analysis. (E) Waterfall plot for the best change of target lesions in 20 patients. The post-treatment urinary MRD status of each patient are arranged along the x-axis. Bar color indicates the pathologic outcome of neoadjuvant toripalimab. TFsm = tumor fraction estimate based on somatic mutations; TFcn = tumor fraction estimate based on copy numbers; MRI = magnetic resonance imaging; utDNA = urinary tumor DNA; utDNA-pre = pre-treatment utDNA; utDNA-post =post-treatment utDNA; ypCR = pathological complete response; pre-tx = pre-treatment; post-tx = post-treatment; MRD = minimal residual disease. Figure S6. (A) Spider plot indicating dynamic changes of TFsm, TFcn, and MRI measurements for each patient during neoadjuvant toripalimab. (B) Box plot illustrates utDNA reduction (defined by TFsm + TFcn < 10%) in utDNA-pre versus utDNA-post samples. (C) IGV plot showing the FGFR3 S249C mutation in patient RZ12 detected by MRD panel sequencing or whole exome sequencing. (D) VAF changes of the FGFR3 S249C mutation in patient RZ12 with progressive disease. TFsm = tumor fraction estimate based on somatic mutations; TFcn = tumor fraction estimate based on copy numbers; MRI = magnetic resonance imaging; C1 = cycle 1; C2 = cycle 2; C3 = cycle 3; C4 = cycle 4; RC = radical cystectomy; ypCR = pathological complete response; MRD = minimal residual disease; WES = whole exome sequencing; tDNA = tumor DNA; utDNA = urinary tumor DNA; utDNA-pre = pretreatment utDNA; utDNA-post = post-treatment utDNA; C2D1 = cycle 2 day 1; VAF = variant allele frequency; PD = progressive disease.