Comparative Genomic Landscape of Urothelial Carcinoma of the Bladder Among Patients of East and South Asian Genomic Ancestry

Abstract

Introduction

Urothelial bladder carcinoma (UCB) is a multifactorial disease driven by environmental exposures—especially cigarette smoking—inherited genetic variants, and an accumulation of somatic genetic events. It is the 10th most common cancer worldwide, with more than 55 000 new cases annually, causing significant morbidity, mortality, and burden in the healthcare systems.¹ Rates vary by regions of the world explained in part by the fact that different ancestries demonstrate varying germline mutations and tend to encounter different environmental exposures in a dynamic interplay. In UCB, the highest incidence is seen in the developed countries of Southern and Western Europe and North America with an incidence of 10-15 cases per 100 000 persons.² Conversely, the lowest incidence is seen in Middle Africa, Central America, and Western Africa.¹ Patients of Asian ancestry have a similarly low incidence of UCB. Specifically, those patients of East Asian ancestry (EAS) from the countries of China, Japan, and Korea have an incidence of 4.5 cases per 100 000 persons. Those of South Asian ancestry (SAS) from India, Pakistan, Afghanistan, Nepal, Bangladesh, Sri Lanka, and Bhutan see an even lower incidence of roughly 2.5 cases per 100 000 persons.² While the incidence is low, these Asian cohorts make up a significant portion of the world’s population and account for over 160 000 cases per year. Unfortunately, much of the large-scale scientific research in the UCB space has disproportionately included less of this population from clinical trials.

Understanding the genomic differences in cancers of diverse populations will become even more important, as we continue to see advances in precision and personalized oncology. The use of targeted agents in advanced UC has only recently been used clinically.³ Unfortunately, much of these data did not include patients of Asian ancestry. Few Asian patients were included in the large clinical trials that define the standard treatment options for patients. A U.S. Food and Drug Administration (FDA) analysis combining data from 7 clinical trials involving immune checkpoint inhibitors (ICI) in UC found that overall, 82% of enrolled patients were White and only 10% were Asian, with 7% of other races or ethnicities.^4-6 This inequality will become increasingly relevant, as targeted agents based on genomic alterations become more important in the treatment of UC. There is a critical need to improve our understanding of the molecular pathogenesis of the disease in the East and South Asian populations to provide insights that could impact future treatment options. To address that, we present a comprehensive genomic profiling (CGP) study of UCB in patients of East and South Asian genomic ancestry and focus on the identification of frequent and potentially targetable genomic alterations.

Methods

A total of 8728 consecutive, centrally reviewed UCB specimens underwent comprehensive genomic profiling (CGP) as per routine clinical care in a Clinical Laboratory Improvement Amendments-certified (CLIA), College of American Pathologists-accredited (CAP), New York State‐regulated reference laboratory (Foundation Medicine, Inc., Cambridge, MA, USA). Briefly, ≥50 ng of DNA per specimen was isolated and sequenced on Illumina HiSeq instruments to high, uniform coverage (median >500×), as previously described.⁷ The DNA extracted from formalin-fixed, paraffin-embedded (FFPE) tumor specimens was analyzed using an adaptor ligation, hybridization capture-based platform (FoundationOne /FoundationOne CDx) which interrogates the entire coding region of at least 324 cancer-related genes and additional select introns from at least 19 genes commonly rearranged in cancer.

Genomic data were analyzed for base substitutions and short insertions/deletions (short variants), amplifications and homozygous deletions (copy number changes), and rearrangements (including gene fusions). Tumor mutational burden (TMB) was determined on at least 0.8 Mbp of targeted sequence and is reported as mutations per megabase (Mb), and microsatellite instability (MSI) status was determined on at least 1500 loci. As self-reported race was not available, predominant patient ancestry was determined for each specimen using a custom SNP-based classifier, as previously described.⁸ Mutational signatures were determined using the decomposition method of Zehir et al using the 96-feature single-base substitution COSMIC reference signatures generated by Alexandrov et al.^9,10 Tumor cell PD-L1 expression was determined by immunohistochemistry (IHC, Dako 22C3) with tumor proportion score (TPS) 0% as negative, 1%-49% as low expression, and TPS ≥50% as high expression. Patient age, biological sex, and site of specimen collection were extracted from accompanying pathology reports. Approval for this study, including a waiver of informed consent and a Health Insurance Portability and Accountability Act (HIPAA) waiver of authorization, was obtained from WCG Institutional Review Board (Protocol No. 20152817).

Categorical variables reported as frequencies were compared between groups via Fisher’s Exact testing while patient age, mean number of pathogenic genomic alterations (GA) per specimen, and mean TMB were compared using 2-sample 2-tailed T testing. False discovery rate (FDR) correction was performed by the Benjamini-Hochberg procedure to correct P values for multiple hypotheses, and the resulting significance values are presented. Statistical significance was defined as a P < .05.

Results

Of the cohort of 8728 UCB, 7447 (85.3%) were of European genomic ancestry (EUR), 541 (6.2%) were of African genomic ancestry (AFR), 461 (5.3%) were of Admixed-American genomic ancestry (AMR), 205 (2.3%) were EAS, and 74 (0.85%) were SAS. Clinical features, mean number of pathogenic GA per specimen, and a selected set of GAs and signatures likely associated with systemic therapy response are reported in Table 1.

The biological sex distribution was similar among the cohorts, with 71.2% male in the EAS cohort, 75.7% male in the SAS cohort, and 74.9% male in the remaining global cohort. Median age at time of tissue acquisition was also similar, ranging between 70 and 72 for all ancestry cohorts. There were no significant differences in MSI status, mutational signatures, or PD-L1 status between any cohorts. Across all 3 cohorts, MSI-high was rare, occurring in 0.9% of the global cohort, 1.5% of the EAS cohort, and 0% of the SAS cohort. There were no significant differences identified between the EAS and SAS cohorts when directly compared to each other.

When evaluating the EAS cohort, pathogenic GAs were most frequently detected in TP53, occurring in 59.0% of patients. TERT promoter mutations were found to be significantly less frequent in EAS as compared to the remaining global cohort (54.2% vs. 72.9%; P < .005). This association held true when stratifying the global cohort and comparing only to those of EUR ancestry (54.2% vs. 73.6%; P < .005) (Fig. 1A). Interestingly, PIK3CA alterations were significantly less common in the EAS cohort, as well (12.7% vs. 22.1%; P < .05), which again was seen when comparing only to the EUR group (12.7% vs. 22.1%; P < .05). Rates of putative driver mutations were comparable between EAS and the overall group, as well as to the EUR group. Notably and in relation to targetable mutations, the EAS cohort had a similar rate of alterations in FGFR3 (16.6% vs. 18.4%; NS). When expanding the analysis to potentially less clinically meaningful genes in the disease, EMSY, NSD3, FGFR1, and PAX5 were discovered to be significantly biased toward the EAS group (P < .05), while no other genes with significantly enriched in the non-EAS cohort (Fig. 1C). TMB was not significantly different when examining the prevalence of TMB ≥10 and TMB ≥20 cohorts.

Open in new tabDownload slide

Open in new tabDownload slide

Open in new tabDownload slide

Figure 1.

(A) Paired longtail of genomic alterations found in 205 patients of East Asian (EAS) and 7447 patients of European (EUR) genomic ancestry with UCB. Mutations in TERT promoter (73.60% vs. 54.15%; P = 6.34E−08) and PIK3CA (22.09% vs. 12.68%; P = 6.34E−08) are enriched in the EUR group. The most frequent 25 genes ordered by combined frequency in both cohorts are shown. (B) Tile plot of pathogenic genomic alterations identified in 205 patients with UCB of East Asian genomic ancestry. MSI-high was detected in 1.5% of patients were MSI-high and 7.3% were TMB-high (>20 mut/Mbp). (C) Volcano plot showing enrichment (P < .05) for pathogenic gene alterations between EAS and non-EAS groups. EMSY, NSD3, FGFR1, and PAX5 are significantly enriched in the EAS group while TERT promoter and PIK3CA are enriched in the non-EAS group.

The SAS cohort was found to have statistically similar rates of GAs when compared to the remaining global cohort and to the EUR cohort. However, there was a noticeable decrease in the incidence of TERT promoter alterations in the SAS cohort when comparted to EUR (58.1% vs. 73.6%; P = .06) (Fig. 2A). Whereas TERT was the most common alteration in the global cohort, occurring in approximately 73% of patients, TP53 was the most common alteration in SAS cohort, found in 67.6% of patients. The SAS cohort had fewer alterations in FGFR3 (9.5% vs. 18.5%, P = .30). Interrogating the full set of genes yielded a single additional significant finding—enrichment of MST1R mutations in the SAS group (P < .05) (Fig. 2C). When evaluating the TMB, both the overall and SAS cohorts were similar at 6.25 mut/Mb.

Open in new tabDownload slide

Open in new tabDownload slide

Open in new tabDownload slide

Figure 2.

(A) Paired longtail of genomic alterations found in 75 patients of South Asian and 7447 patients of European genomic ancestry with UCB. No mutations were enriched in either group. The most frequent 25 genes ordered by combined frequency in both cohorts are shown. (B) Tile plot of pathogenic genomic alterations identified in 75 UCB patients of South Asian genomic ancestry. MSI-high was detected in 0% of patients were MSI-high and 12.2% were TMB-high (>20 mut/Mbp). (C) Volcano plot showing enrichment (P < .05) for pathogenic gene alterations between SAS and non-SAS groups. MST1R is significantly enriched in the SAS group while no genes are enriched in the non-SAS group.

Discussion

This study provides the largest analysis of ancestrally associated genomic changes in UCB. Prior to our study, the largest ancestral analysis to date involved use of the comprehensive multi-omics sequencing of the TCGA database.¹¹ In that study, the authors evaluated the ancestry effects on mutation rates DNA methylation, and mRNA, and miRNA expression among 10 678 patients across 33 cancer types. Of the 10 678 patients in this study, there were only 669 patients of East Asian ancestry and 27 patients of South Asian ancestry. Again, that study spanned 33 cancers. In this UCB-specific analysis, we were able to include 205 and 70 patients of EAS and SAS ancestry.

The TERT promoter is the most common mutation that occurs in approximately 60%-80% of patients with bladder cancer.¹²TERT mutation is correlated with increased telomerase activity and shorter but stable telomere length, with growing evidence that this empowers the genetically unstable cells to evade senescence by maintaining telomeres from critically shortening.^12,13TERT mutations can be identified throughout the spectrum of urothelial carcinoma with conflicting reports on association of mutation with grade, stage, and prognosis.¹² Despite the frequency of this mutation, there are no approved agents that specifically target the gene itself or telomerase activity. Unfortunately, compounds that have been evaluated in clinical trials have proven either ineffective in adult malignancies or too toxic when tested in pediatric populations.^14,15 Nevertheless, research has demonstrated efficacy when targeting TERT expression through a different, epigenetic mechanism, inhibition of the BET bromodomain.¹⁶ BET bromodomain proteins occupy super-enhancer loci of oncogenes to increase transcription. By inhibiting these proteins, targeted agents can suppress oncogene transcription and exert anticancer effects. BET inhibitors have shown preclinical efficacy in bladder cancer models and require further testing in clinical trials.¹⁷

In addition to their therapeutic potential, TERT promoter mutations have also been associated with higher TMB and thus might serve as a potential biomarker associated with response to ICI.¹⁷ One group explored the use of tumor-genomic profiling in predicting the response of advanced-stage UC to ICI.¹⁸ In 119 patients with locally advanced or metastatic UC, researchers demonstrated that TERT promoter mutations were an independent factor associated with ICI response, progression-free survival, and overall survival.

FGFR3 gene mutations are common in UC. The FGFR inhibitor erdafitinib, received accelerated FDA approval in April 2019 as salvage therapy in patients with FGFR2 or FGFR3 activating mutation or fusion progressing on platinum-based chemotherapy. In a case, a locally advanced bladder tumor from a 64-year-old SAS man was found to possess an FGFR3-TACC fusion, making erdafitinib a potential therapeutic option (Fig. 3). In this study, we found that like TERT, FGFR3 mutations were less common in SAS patients than in the remainder of the cohort. FGFR3 overexpression occurs in up to 40% of MIBC. Furthermore, initial data suggested that FGFR3 overexpression may be a useful biomarker to identify patients eligible for FGFR inhibitor therapy, but this notion has not been supported by additional data. Interestingly, when evaluating the TCGA bladder cancer database, Asian patients demonstrated significantly higher FGFR3 expression as compared to Caucasians.¹⁹

Figure 3.

Trans urethral resection of bladder tumor (TURBT) of a deeply invasive UCB that presented with metastatic disease in a 64-year-old man of South Asian genomic ancestry. PD-L1 expression by IHC showed CPS score 10. CGP revealed this MSS tumor had TMB 14 mut/Mbp and harbored and FGFR3-TACC3 fusion along with pathogenic short variant mutations in APC, ARID1A, CDKN1A, GATA3, RB1, TERT, TP53, TSC1, and homozygous deletion of RAD51B. The FGFR3-TACC3 fusion involved a chr4 duplication event yielding a 5ʹ-FGFR3(ex1-17 NM_000142)-TACC3(ex8-16 NM_006342) chimera.

Open in new tabDownload slide

PIK3CA alterations, found to be less common in the EAS cohort, have been directly implicated in several solid tumors.²⁰ Studies have shown alterations in 42% to 55% of endometrial cancer, 42% of cervical cancer, and 27% to 36% of breast cancer.^21-24 While clinical trials evaluating PI3K/mTOR pathway inhibitors have largely been disappointing in UC, there have been positive results in metastatic breast cancer with activating mutations treated with the FDA-approved agent, alpelisib.²⁵ Furthermore, preclinical data in bladder cancer models demonstrated that PIK3CA inhibition in combination with ICI has enhanced antitumor effects through increased immune stimulation.²⁶ If PIK3CA inhibition may transition to more clinical trials in UC, it will be important to include East Asian patients to assess safety and efficacy.

Despite the FDA approval of erdafitinib and ICI in UC, its use in patients of South Asian ancestry may not be widespread. This is largely due to real-world access barriers, for example, availability and cost. In a study from India, investigators evaluated single center use of ICI in solid tumors.²⁷ They found that of 9610 patients who had indications for ICI, only 155 (1.6%) went on to receive it, listing financial constraint as the most common limiting factor. Equitable access to tumor next generation sequencing, clinical trials and safe, effective therapies is a very important priority to eliminate disparities.

Understanding genomic differences at a population level is only a first step. We must Incorporate ancestral data in clinical trial design, specifically by performing these trials in regions of the world from where this data are derived. Indeed, while learning mutational frequencies of SAS is useful, doing so by evaluating a SAS patient living in the US does not translate into real-world success in treating a patient in India. Developing regions of South and East Asia possess different disease burdens that may affect the response rates and toxicity profiles of therapies. Furthermore, use of newly approved agents in these developing regions is not widespread. This is largely due to real-world cost barriers. In one study from India, investigators evaluated single center use of immunotherapy in solid tumors.²⁸ They found that of 9610 patients who had indications for ICI, only 155 (1.6%) went on to receive therapy, listing financial constraint as the most common limiting factor. Therefore, even if data would suggest greater clinical efficacy in this population, there are barriers to treatment access in these resource limited countries.

In this study, the greatest limitation is the lack of clinical, therapy response, and outcomes data that impairs any analysis associated those with genomic data. Moreover, we could not ascertain race and ethnicity based on patient reports. The retrospective and descriptive design of this study, as well as the presence of possible selection bias and confounding factors, may impact interpretation. Furthermore, CGP was performed on a single representative block as part of the clinical workup; therefore, investigation of potential genomic heterogeneity that might correlate with morphologic heterogeneity was outside the scope of this study. In that context, we also did not have adequate plasma samples for ctDNA analysis.

Conclusion

Despite the limitations noted, our series represents one the largest efforts to characterize molecular alterations in advanced UCB based on patient ancestry. The in-depth genomic analysis of South Asian and East Asian patients provides hypothesis-generating insights into potential differences in a population level. Our study should further motivate and support the inclusion of more diverse patient populations in clinical trials in UC and across cancer types.

Funding

The author indicated no financial relationships.

Conflict of Interest

Roger Li: Predicine, Veracyte, CG Oncology, Valar Labs (research support), CG Oncology (clinical trial protocol committee), BMS, Merck, Fergene, Arquer Diagnostics, Urogen Pharma, Lucence (scientific advisor/consultant). Petros Grivas: Aadi Bioscience, AstraZeneca, Astellas Pharma, Boston Gene, Bristol Myers Squibb, CG Oncology Inc., Dyania Health, EMD Serono, Exelixis, Fresenius Kabi, Genentech/Roche, Gilead Sciences, Guardant Health, Infinity Pharmaceuticals, Janssen, Lucence Health, MSD, Mirati Therapeutics, Pfizer, PureTech, QED Therapeutics, Regeneron Pharmaceuticals, Seattle Genetics, Silverback Therapeutics, 4D Pharma PLC, UroGen (consulting), Bavarian Nordic, Bristol Myers Squibb, Clovis Oncology, Debiopharm, EMD Serono, G1 Therapeutics, Gilead Sciences, GlaxoSmithKline, MSD, Mirati Therapeutics, Pfizer, QED Therapeutics (institutional research funding). Andrea Necchi: Merck, AstraZeneca, Janssen, Incyte, Roche, Rainier Therapeutics, Clovis Oncology, Bayer, Astellas/Seattle Genetics, Ferring, Immunomedics (consulting/advisory relationships), Merck, Ipsen, AstraZeneca (research funding). Dean Pavlick, Richard S.P. Huang, Douglas Lin, Natalie Danziger, Jeffrey S. Ross: Foundation Medicine Inc., a wholly owned subsidiary of Roche (employment) and Roche (ownership interest). Joseph M. Jacob: Janssen and Urogen (consulting). Taylor Peak, Philippe E. Spiess, and Gennady Bratslavsky indicated no financial relationships.

Author Contributions

Conception/design: All authors. Provision of study material or patients: D.P., J.R. Collection and/or assembly of data: D.P., R.S.P.H., D.L., N.D., J.R. Data analysis and interpretation: D.P., R.S.P.H., D.L., N.D., J.R. Manuscript writing: T.P., P.E.S., D.P., J.R. Final approval of manuscript: All authors.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References