Inherited Germline Cancer Susceptibility Gene Variants in Individuals with Non–Muscle-Invasive Bladder Cancer Open Access

Inherited germline pathogenic and likely pathogenic (P/LP) variants can have profound implications for the development, treatment, and screening of various cancer types (1). Recent investigations have found that a substantial proportion of individuals with bladder cancer have P/LP germline variants in cancer predisposition genes, especially in DNA damage response (DDR) genes (2–4). To date, most of these studies have focused mostly on patients with locally advanced or metastatic urothelial carcinoma and the limited data available for non–muscle-invasive bladder cancer (NMIBC) suggests the presence of P/LP variants may be associated with substantially worse clinical outcomes (5). Yet, the influence of P/LP germline variants in patients treated with Bacillus Calmette-Guerin (BCG) immunotherapy, the most effective and commonly used treatment for high-grade NMIBC, is unknown. Thus, the prevalence and significance of P/LP variants in patients with NMIBC remain poorly defined.

NMIBC is the most common form of urothelial carcinoma accounting for more than 70% to 80% of the estimated 573,278 individuals diagnosed with bladder cancer worldwide each year (6–8). Current risk stratification schemas that rely on clinicopathologic features cannot accurately predict which patients with NMIBC will experience recurrence or progression to muscle-invasive disease (9). To better understand the biology of NMIBC, our group and others have recently reported on the somatic mutational spectrum of NMIBC, identifying a high rate of somatic DDR gene alterations and high tumor mutational burden in high-grade NMIBC (10, 11). While somatic alterations are important determinants of biologic behavior in NMIBC, we hypothesized that P/LP germline variants are also of biologic and clinical importance. We therefore sought to determine the prevalence of P/LP germline variants in patients with NMIBC and to investigate their impact on clinically relevant outcomes.

All patients provided written informed consent to an institutional review board (IRB)–approved prospective protocol (NCT01775072) for tumor and matched normal DNA sequencing via MSK–Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT), a clinical next-generation sequencing platform that is FDA-authorized to identify somatic genetic variants in over 341 cancer-related genes (12). Peripheral blood samples were used for matched germline DNA sequencing. Participating patients consented to receive the results of their somatic mutational profile. This study was performed in accordance with Declaration of Helsinki and following IRB approval [Memorial Sloan Kettering Cancer Center (MSKCC; New York, NY) protocols 16–1249 and 17–420], secondary germline analysis was conducted in anonymous fashion. Briefly, detailed clinical data, including self-reported ancestry, religion, and race, was abstracted for each patient record. Baseline demographics, social history, family cancer history, and personal cancer history were obtained using a structured clinical assessment and manual review of electronic medical records. After clinical annotation, sequence data were assigned a unique study identifier and irretrievably delinked from personal identifiers before germline variant calling and analysis was performed in a permanently anonymized fashion. Patients included in the current study were also included in a previous germline analysis report by our group (2), but limited details on NMIBC treatment and clinical annotation were available in that report.

This cohort comprised the first 99 patients at MSKCC with newly diagnosed NMIBC to undergo clinical MSK-IMPACT tumor genomic profiling within our CLIA-certified lab. These patients’ data were originally analyzed to characterize the somatic genomic landscape of NMIBC (10), and they were unselected for risk of inherited cancer predisposition syndromes. All patients in this cohort underwent evaluation, treatment, and follow-up at MSKCC starting at the time of their initial diagnosis. As previously described, all patients in this cohort underwent evaluation, treatment, and follow-up at MSKCC starting at the time of their initial diagnosis. Restaging TURBT was performed in all high-grade stage T1 (HGT1) tumors to confirm that the detrusor muscle was uninvolved. A board-certified genitourinary pathologist (H. Al-Ahmadie) reviewed representative hematoxylin and eosin slides to confirm grade and stage of all index NMIBC tumors. Treatment and management were at the discretion of the treating urologic oncologist at MSKCC. Patients treated with BCG immunotherapy received six weekly full doses of TICE BCG. Patients were followed at MSKCC every 3 months by cystoscopy and urine cytology for the first year, then every 3 to 6 months.

This second cohort comprised 115 patients with NMIBC at initial diagnosis. While 41% (47/115) of these patients were evaluated and treated at MSKCC for their initial tumor in the same fashion as the original NMIBC cohort, the other patients were initially managed for their NMIBC at an outside hospital prior to referral to MSKCC upon disease recurrence and/or progression. Thus, the expanded NMIBC cohort is enriched for patients exhibiting poor clinical outcomes, as they were offered clinical sequencing of tumor tissue at the discretion of their treating physician to determine eligibility for targeted therapy clinical trials or following progression to locally advanced or metastatic disease, among other reasons. To reduce the risk of potential under-staging and falsely including a patient who actually had muscle-invasive disease in our NMIBC cohort, only patients with an initially non–muscle-invasive tumor (Tis, Ta, or T1 with uninvolved muscularis propria in the specimen) that was confirmed by a re-staging transurethral resection (TUR) or at least one tumor-free follow-up cystoscopy prior to a progression event were included (13).

All BCG-treated patients included in our outcomes analysis had high-grade NMIBC and received at least five of six weekly doses of BCG with or without maintenance therapy. Two patients received less than three instillations of induction BCG (stopped due to toxicity) and were considered non-BCG treated.

This cohort included 169 patients who initially presented with de novo metastatic urothelial bladder cancer or whose initial bladder tumor invaded into or beyond the muscularis propria (clinical stage ≥T2) and had no history of upper tract urothelial carcinoma (UTUC) or NMIBC (13). These patients were offered clinical sequencing of tumor tissue at their treating physician's discretion for various clinical purposes.

Germline sequencing data (BAM file) generated from the paired tumor/normal MSK-IMPACT clinical assay were analyzed in accordance with the recommendations of the American College of Medical Genetics and Genomics (ACMGG; Bethesda, MD; ref. 14). The analysis focused on germline variants in ≥76 genes with a known hereditary cancer predisposition association, for which all coding regions were sequenced in both germline normal and tumor tissue. As previously described (2), both automated computational pipelines and manual curation were used to ensure optimal germline variant calling. The potential clinical significance of identified variants was assessed by evaluating association with genetic disease as determined by reviewing the literature and online databases in conjunction with established bioinformatics pipelines. Germline variants were initially prioritized using the Pathogenicity of Mutation Analyzer (PathoMAN) classification tool that automates germline genomic variant curation in an unbiased and efficient manner (15). To further increase robustness, germline variants were also manually curated by a member of the MSKCC Niehaus Center for Inherited Cancer Genomics research team with expertise in clinical and molecular genetics according to ACMGG criteria (14). Only P/LP germline variants were included in this analysis; variants of unknown significance were reviewed but were not reported or analyzed. Biallelic inactivation through LOH was determined by the FACETS algorithm (16). P/LP variants were classified at the gene level as having high penetrance [relative risk (RR) of >4], moderate penetrance (RR of 2–4), low penetrance (RR < 2), uncertain penetrance, or association with an autosomal recessive condition (1). Classification was performed at the variant level for CHEK2, APC, and ERCC3: APC p.Ile1307Lys and CHEK2 p.Ile157Thr were considered low penetrance and the ERCC3 p.Arg109X, uncertain penetrance (17).

Clinical characteristics were compared between P/LP groups using Fisher exact tests or χ² tests for categorical variables, and Wilcoxon rank-sum for continuous variables. As clinical data were analyzed in anonymized fashion, event or censoring times for recurrence-free survival (RFS), biologic progression-free survival (bPFS), clinical PFS (cPFS), and time to development of UTUC were provided as intervals. Kaplan–Meier method was used to estimate probabilities for RFS, bPFS, cPFS, and time to development of UTUC, and log–rank tests were used to test for differences between groups. The icenReg R package was used to fit Cox proportional hazards models for interval censored data and identify univariable associations between P/LP germline variants and outcomes of interest. Median follow-up was estimated as the median minimal follow-up time of the censoring interval for those who did not have a progression event (18). Recurrence was defined as histologically proven urothelial cancer. Biologic progression was defined as the development of secondary muscle-invasive bladder cancer (MIBC; tumor stage ≥T2) or metastasis. As radical cystectomy is a clinically meaningful event that is often performed before biologic progression, we also analyzed cPFS as a composite endpoint of radical cystectomy and/or biologic progression, whichever occurred first. RFS, bPFS, and cPFS were calculated from the time of pretreatment TUR until the event of interest. Analyses on BCG treatment were restricted to patients with BCG-naïve, high-grade NMIBC who received at least five of six weekly doses of BCG with or without maintenance therapy. Time to UTUC development was calculated from initial diagnosis of NMIBC until histologic proven urothelial cancer of the ureter or renal pelvis/calyxes. All analyses were conducted in R version 4.1.0.

The data generated in this study are available within the article and its supplementary data files. Additional data are available upon reasonable request from the corresponding author.

We identified 214 patients who presented with NMIBC as their initial urothelial carcinoma diagnosis for whom germline analysis was performed (Supplementary Fig. S1). Patient and clinicopathologic features for the original and expanded NMIBC cohorts are in included in Table 1. As expected, patients within the original cohort were representative of the demographics and treatment outcomes of a typical NMIBC patient population (9, 10), whereas patients within the expanded cohort were enriched for poor clinical outcomes, with out of 115 patients, 66 progressed to secondary MIBC or metastasis and 31 developed subsequent UTUC (Table 1). The median lower bound of the censoring interval (a proxy for median follow-up) among those who did not progress was 50 months. As there was a similar frequency of P/LP germline variants in the original and expanded cohorts [12% (12/99) vs. 8.7% (10/115), P = 0.4; Table 1], these cohorts were analyzed together to investigate the significance of P/LP germline variants in patients with NMIBC.

Of the 214 patients initially diagnosed with NMIBC, 22 (10%) had a P/LP variant in a gene associated with cancer predisposition (Table 2; Fig. 1A). Similar to prior reports in patients with locally advanced and metastatic urothelial cancers (2–4), we found a high frequency of P/LP DDR germline variants in patients with NMIBC (15 of 22, 68%; Fig. 1A). Within the nucleotide excision repair (NER) pathway, 2 patients had P/LP germline variants in ERCC2 with either LOH or a somatic mutation in the second allele, suggesting that these variants likely contributed to bladder cancer development (Table 2; Fig. 1B). 2 patients carried germline ERCC3 p.R109X alterations, a known founder mutation in the Ashkenazi Jewish population known to be associated with a moderate risk of breast cancer (17). In the homologous recombination repair (HRR) pathway, 3 patients had high-penetrance, potentially actionable BRCA1 variants. No BRCA2 variants were seen in patients with NMIBC despite germline BRCA2 variants being associated with advanced urothelial carcinoma (2–4). Although Lynch syndrome is associated with both bladder cancer and UTUC (19), there were no P/LP variants in mismatch repair genes in our NMIBC cohort.

P/LP germline variants in non-DDR genes included 2 patients with fumarate hydratase (p.Lys477dup) mutations, but this variant is not associated with hereditary leiomyomatosis and renal cell cancer syndrome (Table 2; Fig. 1B; ref. 20). In addition, 4 patients carried a low-penetrance APC p.I1307K Ashkenazi Jewish population founder variant associated with a 1.5- to 2.2-fold increased relative risk of developing colon cancer for which the National Comprehensive Cancer Network recommends an enhanced screening colonoscopy schedule (21). All 3 patients with LOH of APC p.I1307K underwent radical cystectomy following BCG failure (2 progressed to secondary MIBC), but the significance of this observation remains unclear. Furthermore, the tumor sequenced for somatic mutational calling to determine biallelic inactivation in the expanded NMIBC cohort was not always from the index NMIBC tumor so further investigation into these findings are warranted.

To better define which patients with NMIBC should be considered for germline testing and counseling, we compared patient demographics and initial clinicopathologic tumor characteristics between the 22 patients with P/LP variants and the 192 patients without. The presence of germline alterations overall was associated with having high-grade disease (75% no P/LP variant vs. 100% P/LP variant, P = 0.005) and multiple tumors at initial diagnosis (64% single tumor no P/LP variant, 38% single tumor P/LP variant, P = 0.023, Table 3); the same associations were also true specifically for DDR variants (Supplementary Table S1).

Interestingly, all 22 P/LP germline variants were found in patients with high-grade NMIBC (n = 163), with none of the 46 patients with low-grade Ta (LGTa) tumors having a P/LP germline variant (P = 0.005, Fig. 2). The risk of progression for patients with low-grade NMIBC is usually ≤2% (9), but inclusion of the expanded NMIBC cohort resulted in our study being uniquely enriched for patients with LGTa tumors who experienced progression events. In total, 25 of 46 patients with an initial LGTa tumor eventually progressed to ≥T1 and/or high-grade disease, 13 of whom eventually developed MIBC/metastatic disease [5-year PFS 90%; 95% confidence interval (CI), 80–100]. Therefore, the germline factors analyzed here are unlikely to contribute to disease progression in low-grade NMIBC.

As P/LP variants were only found in patients with high-grade NMIBC, we repeated our analysis comparing patients with and without P/LP variants among this subset to better risk-stratify patients for germline screening. Within this group, we found no differences in patient demographics or clinicopathologic tumor characteristics between those with any P/LP germline variants and those without (Supplementary Table S2), or those with versus without DDR variants (Table 3).

We next sought to determine whether there was any difference in the frequency of P/LP germline variants between those initially diagnosed with high-grade NMIBC (n = 163) versus patients who had muscle-invasive or metastatic bladder cancer at initial diagnosis (n = 169). We found no difference in the proportion of P/LP germline variants overall [13% (22/163) vs. 12% (21/169), P = 0.8] or in DDR genes specifically [9.2% (15/163) vs. 11% (18/169), P = 0.7; Fig. 2].

Because both somatic mutations and germline SNPs in DDR genes have been associated with response to BCG (10, 13, 22, 23), we hypothesized that the presence of P/LP germline variants in DDR genes would correlate with clinical benefit from BCG. However, among patients treated with BCG (n = 148), we found no statistically significant difference between those with and without P/LP germline DDR variants in terms of RFS (HR 0.98 ref: no variant, P > 0.9), bPFS (HR ref: no variant 1.1, P = 0.9), or cPFS (HR ref: no variant 1.14, P = 0.8). One-year RFS was 62% (95% CI, 40–95) in those with a DDR variant versus 67% (95% CI, 49–90) in those without (Fig. 3A), the 2-year bPFS was 77% (95% CI, 57–100) in the variant group and 83% (95% CI, 77–89) in those without (Fig. 3B), and the 2-year cPFS was 77% (95% CI, 57–100) in the variant group and 80% (95% CI, 73–87) in those without (Fig. 3C).

Our group recently reported that patients with initial NMIBC tumors treated with BCG that progressed to secondary MIBC had fewer somatic ERCC2 missense mutations than those presenting with primary MIBC, which may contribute to less clinical benefit from cisplatin-based chemotherapy (13). Therefore, we also compared the frequency of P/LP variants between patients with primary (n = 169) and secondary MIBC after prior BCG (n = 62). We found no difference in the proportion of P/LP germline variants in either DDR genes [11% (18/169) versus 6.5% (4/62), P = 0.3] or any of the assessed germline genes [12% (21/169) vs. 9.7% (6/62), P = 0.6].

As UTUC is associated with Lynch syndrome, we hypothesized that patients with NMIBC who subsequently develop UTUC would have a greater portion of P/LP germline variants in mismatch repair genes. 36 patients of 214 in our cohort eventually developed UTUC, with a 5-year UTUC-free survival rate of 92% (95% CI, 88–96). However, we found no P/LP germline variants in mismatch repair genes (MLH1, MSH2, MSH6, PMS2) in any patient within our entire cohort. Moreover, we found no difference in risk of developing UTUC between patients with a P/LP DDR germline variant versus those without (HR 1.02 ref: no variant, P > 0.9; Fig. 3D).

The identification of P/LP germline variants can provide guidance for personalized cancer screening, prevention, and treatment. In this large clinically annotated cohort of patients with NMIBC, we found a high rate of P/LP germline variants in patients initially diagnosed with high-grade NMIBC (13.5%), a frequency similar to that in patients with muscle-invasive and metastatic urothelial carcinoma. In contrast, we found no P/LP germline variants in patients initially diagnosed with low-grade NMIBC. While only 46 patients with LGTa were included in our study, they were enriched for adverse clinical outcomes, which suggests that the germline variants typically screened for by current multigene panels do not significantly contribute to disease initiation and progression in patients with LGTa. This distribution of P/LP germline DDR gene variants among bladder cancer grades is similar to that reported for somatic DDR gene alterations (10, 11, 23), suggesting that both somatic and germline alterations in DDR genes are important contributors to the pathogenesis of high-grade bladder cancer, but unlikely to be involved in the development of low-grade NMIBC.

Our data indicate that germline testing and counseling should be considered in any patient with high-grade bladder cancer, whether they are initially diagnosed with high-grade NMIBC, MIBC, or metastatic disease. Identification of P/LP variants in patients with high-grade NMIBC may be even more important, as most patients with NMIBC have a favorable prognosis and do not die from bladder cancer, but from other causes (24, 25). Survivors of bladder cancer have a 19% risk of developing an additional nonurothelial malignancy within 10 years of their bladder cancer diagnosis, and a 34% risk within 20 years (26). This is the highest risk of multiple malignancies among all common cancer types and more than half of those who survive their bladder cancer ultimately die from a subsequent nonurothelial malignancy (26). While shared etiologic exposures such as smoking contribute to this risk (27), P/LP variants in DDR genes may increase susceptibility to tobacco-associated DNA damage and risk of development of additional primary malignancies (28). Germline screening in patients with high grade NMIBC may afford an opportunity for the early detection of subsequent nonurothelial malignancies, along with the benefits of cascade germline testing for at-risk family members (29, 30).

Our findings may also have therapeutic implications in NMIBC. PARP inhibitors confer selective sensitivity against HRR-deficient tumors and are a standard therapy in multiple tumor types for patients with deleterious germline mutations in BRCA1/2 and other DDR genes (31). PARP inhibition for metastatic bladder cancer remains under investigation but may have a role in genomically selected patients (32–35). While toxicity from current systemic PARP inhibitors make them unlikely treatment options for patients with NMIBC, alternative strategies that also exploit tumor vulnerabilities due to impaired DNA repair could be considered. For example, hyperthermia inhibits chemotherapy-induced poly(ADP-ribosyl)ation to a similar degree as pharmacologic PARP inhibition (36), and thus intravesical chemohyperthermia, such as hyperthermic intravesical mitomycin (37), may increase drug activity in patients with deleterious germline and somatic DDR alterations. Moreover, the high prevalence of ERCC2 and ERCC3 germline and somatic alterations in NMIBC suggests that selectively targeting NER-deficient tumors with intravesical cisplatin (38) or the semisynthetic DNA alkylating agent irofulven (39, 40) could provide a precision therapeutic approach for these patients.

Our results partially contrast with those of a group in Shanghai who performed targeted germline sequencing in patients with NMIBC (5). While they found a similar rate of germline DDR variants (11.3%, 8 of 71), they reported that DDR germline variants in patients with NMIBC were associated with “unfavorable outcomes” (5). In that study P/LP DDR germline variants were enriched in patients with secondary MIBC compared with those with “non-progressive” NMIBC [4 of 11 (36%) vs. 4 of 60 (6.7%), P < 0.02; 5]. As we restricted our clinical outcomes analysis to only patients treated with BCG, whereas few, if any, patients in the Shanghai study would have received BCG given the availability issues in China during their study period (5). Thus, our conflicting findings may be the result of DDR germline variants conferring worse outcomes in non–BCG-treated patients compared with those treated with BCG.

Although our study is the largest investigation to date focused on germline P/LP variants in patients with NMIBC, the overall low frequency of P/LP variants in each cancer-associated gene limits our ability to make any firm conclusions about their impact on clinical outcomes. Moreover, additional studies are needed to better understand whether germline variants would have the same implications as somatic mutations may have on therapeutic responses in bladder cancer (10, 13, 23, 41). The impact of germline variants is dependent on lineage and penetrance for each specific gene, which requires further investigation in bladder cancer (42). Our findings might also not be generalizable to the wider population of patients with NMIBC, as MSKCC is a specialized cancer referral center in the northeast United States and patients in our study were selected for tumor-normal next-generation sequencing. Our population is also enriched in Ashkenazi Jewish ancestry, a group who tend to harbor more founder germline variants than others (17). In addition, most patients in our study were male and self-reported as White. Taken together, these factors may limit the generalizability of our finding and support the need for validation in larger and more diverse multicenter prospective cohorts.

In conclusion, we found a high rate of P/LP germline variants in patients with high-grade NMIBC (13.5%), a similar frequency to that among patients with locally advanced and metastatic urothelial cancer. If validated in additional NMIBC cohorts, our results support offering germline counseling and testing for patients with high-grade bladder cancer, regardless of clinical stage. This could improve early detection of subsequent nonurothelial malignancies for patients and allow for cascade germline testing for family members potentially at risk for urothelial and nonurothelial malignancies. Finally, the high prevalence of both somatic mutations and germline variants in DDR genes support continued investigation of therapeutic strategies targeting impaired DNA repair in high-grade NMIBC.

E.J. Pietzak reports grants from NIH/NCI, Wofchuck Family Young Investigator Award, and Bladder Cancer Advocacy Network Young Investigator Award during the conduct of the study; and personal fees from UpToDate, Janssen Pharmaceuticals, Merck & Co. Inc., QED Therapeutics, Chugai Pharmaceuticals, UroGen Pharma, and Steba Biotech outside the submitted work. P. Srinivasan reports current employment at Natera Inc. V. Joseph reports a patent for Diagnosis and treatment of ERCC3-mutant cancer; inventors:WO2018170230A1 pending. L. Zhang reports that family members hold leadership positions and ownership interests in Decipher Medicine. K.A. Cadoo reports personal fees from AstraZeneca, MJH Life Sciences, MSD Ireland, GSK Ireland, and Nextcure; nonfinancial support from Roche Ireland and Pfizer; and other support from MSD Ireland and Immunogen outside the submitted work. H. Al-Ahmadie reports personal fees from AstraZeneca and Janssen Biotech; and nonfinancial support from Paige.AI outside the submitted work. J.E. Rosenberg reports personal fees and other support from Seagen, Astellas, AstraZeneca, and Genentech; and personal fees from Mirati, Tyra Biosciences, Alligator, Gilead, QED, Boehringer Ingelheim, Pfizer, EMD Serono, Infinity Pharmaceuticals, BMS, Janssen, Eli Lilly and Company Oncology, and Merck outside the submitted work. G.V. Iyer reports grants and personal fees from Bayer HealthCare, Janssen, and Mirati Therapeutics; personal fees from Basilea Pharmaceutica, Flare Therapeutics, Gilead Sciences and LOXO Oncology at Eli Lilly and Company; and grants from Debiopharm Group, Seattle Genetics, and Aadi Biosciences outside the submitted work. D.B. Solit reports personal fees from Pfizer, Loxo/Eli Lilly and Company Oncology, Scorpion Therapeutics, FORE Therapeutics, Vividion Therapeutics, Fog Pharmaceuticals, and BridgeBio outside the submitted work. M.E. Robson reports personal fees from Research to Practice, Intellisphere, MyMedEd, Change Healthcare, and Physician's Education Resources; consulting (uncompensated) for Artios Pharma, AstraZeneca, Daiichi-Sankyo, Epic Sciences, Merck, Pfizer, Tempus Labs, and Zenith Pharma; grants (institution, clinical trial) from AstraZeneca, Merck, and Pfizer; and other support (editorial services) from AstraZeneca and Pfizer outside the submitted work. Z.K. Stadler reports other support from Genetech/Roche, Adverum, Gyroscope Therapeutics, Neurogene, Optos, Outlook Therapeutics, Regeneron, and RegenexBio outside the submitted work. M.F. Berger reports personal fees from Eli Lilly and Company and AstraZeneca outside the submitted work. D.F. Bajorin reports grants and personal fees from Bristol-Myers Squibb and Merck outside the submitted work. M. Carlo reports grants from Robert Wood Johnson Foundation during the conduct of the study and personal fees from OncLive outside the submitted work. No disclosures were reported by the other authors.

E.J. Pietzak: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. K. Whiting: Conceptualization, formal analysis, validation, investigation, writing–original draft, writing–review and editing. P. Srinivasan: Data curation, formal analysis, methodology, writing–review and editing. C. Bandlamudi: Data curation, software, methodology, writing–review and editing. A. Khurram: Data curation, software, methodology, writing–review and editing. V. Joseph: Resources, data curation, software, formal analysis, writing–review and editing. A. Walasek: Data curation, writing–review and editing. E. Bochner: Data curation, formal analysis, methodology, writing–review and editing. T. Clinton: Data curation, writing–review and editing. N. Almassi: Data curation, methodology, writing–review and editing. H. Truong: Data curation, writing–review and editing. M.R. de Jesus Escano: Data curation, software, methodology, writing–review and editing. M. Wiseman: Data curation, writing–review and editing. D. Mandelker: Software, methodology, writing–review and editing. Y. Kemel: Software, methodology, writing–review and editing. L. Zhang: Data curation, methodology, writing–review and editing. M.F. Walsh: Resources, software, methodology, writing–review and editing. K.A. Cadoo: Data curation, methodology, writing–review and editing. J.A. Coleman: Resources, writing–review and editing. H. Al-Ahmadie: Resources, writing–review and editing. J.E. Rosenberg: Resources, writing–review and editing. G.V. Iyer: Resources, supervision, funding acquisition, writing–review and editing. D.B. Solit: Conceptualization, resources, formal analysis, supervision, funding acquisition, validation, investigation, project administration, writing–review and editing. I. Ostrovnaya: Conceptualization, software, formal analysis, supervision, validation, investigation, methodology, project administration, writing–review and editing. K. Offit: Software, supervision, methodology, writing–review and editing. M.E. Robson: Software, supervision, methodology, writing–review and editing. Z.K. Stadler: Software, supervision, methodology, writing–review and editing. M.F. Berger: Resources, software, supervision, funding acquisition, methodology, writing–review and editing. D.F. Bajorin: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, methodology, project administration, writing–review and editing. M. Carlo: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, methodology, project administration, writing–review and editing. B.H. Bochner: Conceptualization, resources, software, supervision, funding acquisition, methodology, writing–review and editing.

This work was supported in part by the Sidney Kimmel Center for Prostate and Urologic Cancers at MSKCC, the NIH/NCI with a Cancer Center Support Grant to MSKCC (grant no. P30 CA008748), the Marie-Josée and Henry R. Kravis Center for Molecular Oncology, Cycle for Survival, NCI Specialized Programs of Research Excellence (SPORE) in Bladder Cancer (grant no. P50 CA221745), NIH/NCI K12 Paul Calabresi Career Development Award for Clinical Oncology (award no. K12 CA184746), MSKCC Department of Surgery Research Award, Bochner-Fleisher Scholars Award, MSKCC/Weill Medical College of Cornell University Clinical and Translational Science Methodology NIH/NCATS (grant no. UL1 TR002384), Wofchuck Family Young Investigator Award, and the Bladder Cancer Advocacy Network Young Investigator Award.

The authors thank Jessica Moore for editorial assistance with this manuscript.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.