Collaborative study from the Bladder Cancer Advocacy Network for the genomic analysis of metastatic urothelial cancer

Damrauer, Jeffrey S.; Beckabir, Wolfgang; Klomp, Jeff; Zhou, Mi; Plimack, Elizabeth R.; Galsky, Matthew D.; Grivas, Petros; Hahn, Noah M.; O’Donnell, Peter H.; Iyer, Gopa; Quinn, David I.; Vincent, Benjamin G.; Quale, Diane Zipursky; Wobker, Sara E.; Hoadley, Katherine A.; Kim, William Y.; Milowsky, Matthew I.

doi:10.1038/s41467-022-33980-9

Download PDF

Article
Open access
Published: 04 November 2022

Collaborative study from the Bladder Cancer Advocacy Network for the genomic analysis of metastatic urothelial cancer

Nature Communications volume 13, Article number: 6658 (2022) Cite this article

5798 Accesses
8 Citations
55 Altmetric
Metrics details

Subjects

Abstract

Urothelial Cancer - Genomic Analysis to Improve Patient Outcomes and Research (NCT02643043), UC-GENOME, is a genomic analysis and biospecimen repository study in 218 patients with metastatic urothelial carcinoma. Here we report on the primary outcome of the UC-GENOME—the proportion of subjects who received next generation sequencing (NGS) with treatment options—and present the initial genomic analyses and clinical correlates. 69.3% of subjects had potential treatment options, however only 5.0% received therapy based on NGS. We found an increased frequency of TP53^E285K mutations as compared to non-metastatic cohorts and identified features associated with benefit to chemotherapy and immune checkpoint inhibition, including: Ba/Sq and Stroma-rich subtypes, APOBEC mutational signature (SBS13), and inflamed tumor immune phenotype. Finally, we derive a computational model incorporating both genomic and clinical features predictive of immune checkpoint inhibitor response. Future work will utilize the biospecimens alongside these foundational analyses toward a better understanding of urothelial carcinoma biology.

PERCEPTION predicts patient response and resistance to treatment using single-cell transcriptomics of their tumors

Article 18 April 2024

KRAS mutation: from undruggable to druggable in cancer

Article Open access 15 November 2021

Spatial transcriptomics reveals discrete tumour microenvironments and autocrine loops within ovarian cancer subclones

Article Open access 03 April 2024

Introduction

Substantial progress has been made in understanding the molecular landscape of urothelial carcinoma (UC) over the past several years. Since The Cancer Genome Atlas’ (TCGA) first molecular characterization of muscle-invasive bladder cancers in 2014¹, additional efforts have identified specific subtypes associated with patient outcomes and response to treatments^{2,3,4,5,6,7,8,9}. At present, in spite of these efforts, there is only one targeted agent that has been FDA approved for the management of patients with advanced disease. Additional efforts, including the current study and ongoing and planned clinical trials, will help to better define the role for targeted therapy in UC. Additionally, the use of biomarkers to select treatment is infrequently used with the exception of programmed death-ligand 1 (PD-L1) protein expression for first line immune checkpoint inhibitors (ICI) and FGFR2/3 alterations for erdafitinib^10,11. There is an unmet need to identify biomarkers with clinical utility.

Most biomarker research has been performed as retrospective efforts in clinical trials. In an attempt to leverage “real world” experience to build a clinical database and biospecimen repository to develop and validate biomarkers, the Bladder Cancer Advocacy Network (BCAN) aligned efforts with academic institutions to design the Urothelial Cancer – Genomic Analysis to Improve Patient Outcomes and Research (UC-GENOME) study. The overarching goal was to provide every patient with metastatic UC the opportunity to become an exceptional responder with molecular analyses for the identification of predictive as well as prognostic biomarkers. The project launched in 2016 with two co-equal aims: (1) to provide a comprehensive next generation sequencing report including DNA mutations, copy number alterations (CNA), fusions, and tumor mutational burden (TMB), with potential therapeutic options including clinical trials, at no cost to patients; and (2) to create a biorepository and clinical database to perform foundational analyses including DNA and RNA sequencing for future collaborative research efforts.

Here, we present the foundational molecular characterization of the UC-GENOME dataset including DNA and RNA sequencing with analyses incorporating the clinical data elements to provide an overarching view of the tumor and the tumor microenvironment in this real-world population of patients with metastatic UC.

Results

Patient characteristics/data description

218 patients with metastatic UC were accrued (Table 1), of which, primary tumors were collected for the majority of patients (87%) with the remaining samples collected from metastatic sites (13%). Most patients were current/former smokers (67.9%) and had a bladder primary tumor at initial diagnosis with high grade and/or invasive disease (83.5%). The most common histologic variants were squamous (15.9%), micropapillary (11%), and adenocarcinoma (5.3%). All other histologic variants occurred at <5% (Supplementary Fig. 1a). Systemic therapy for metastatic UC was administered in 85.8% of patients (chemotherapy (49.1%), ICI (72.9%), targeted therapy (6.9%), and antibody-drug conjugates (11.9%)). With a median follow up from metastatic disease diagnosis of 21.6 months, 127 of 218 (58.3%) patients died (89.8% related to UC). Next generation sequencing (NGS) suggested potential treatment options for 69.3% of patients, however only 5.0% of patients received a targeted therapy based on the results (2.7% received treatment on a clinical trial based on the NGS). DNA and RNA sequencing was successful for 191 and 176 patients, respectively, including 147 samples with complete molecular profiling plus clinical data (Fig. 1a).

Table 1 Demographic, clinical characteristics and NGS-based treatment (n = 218)

Full size table

**Fig. 1: UC-GENOME study design and molecular subtyping shows enrichment of Stroma-rich primary tumors in metastatic datasets.**

Transcriptional profiling reveals enrichment of Stroma-rich primary tumors in metastatic datasets

The Kamoun et al.³ consensus molecular subtypes (Fig. 1b) and the underlying expanded subtyping classifications (Supplementary Fig. 1b) were observed in our UC-GENOME cohort with gene expression patterns prototypical of each subtype. Ba/Sq subtype showed increased expression of basal and squamous markers. The luminal subtypes (Luminal papillary [LumP], Luminal unstable [LumU], and Luminal non-specified [LumNS]) shared expression of luminal markers. The stroma-rich subtype had elevated expression of extracellular matrix and smooth muscle genes as well as epithelial to mesenchymal transition associated genes. Increased expression of immune related genes was observed in Ba/Sq and Stroma-rich tumors. The four NE-like tumors highly expressed neuroendocrine-related genes. As previously described^2,12, females were numerically, but not significantly, enriched in Ba/Sq (17/53, 32%) versus other subtypes (23/117, 19.6%; p = 0.08, Fig. 1c). Of the histologic variants, samples with squamous differentiation were significantly enriched in the Ba/Sq subtype (p < 0.0001; Supplementary Fig. 1c). We next evaluated if either ECOG or age was associated with subtype. There was no significant association between ECOG and age, nor either variable with subtype (Supplementary Fig. 1d–f).

We compared the subtype specific gene expression patterns of UC-GENOME to a cohort of patients with non-metastatic UC, TCGA BLCA¹², and a metastatic cohort, IMvigor210¹³. We observed high correlation of gene expression (UC-GENOME versus IMvigor210, mean correlation 0.79 and UC-GENOME versus TCGA, mean correlation 0.72; Supplementary Fig. 1g). While the subtype expression patterns were similar, the subtype distribution differed between non-metastatic and metastatic cohorts (Fig. 1d). A significant increase in proportion of stroma-rich tumors was observed in the metastatic datasets, IMvigor210 and UC-GENOME, compared to non-metastatic cohorts, TCGA and Kamoun (Fig. 1e, f and Supplementary Data 1).

To determine if this was due to bias from the specimen site (i.e. tissue from a primary tumor -vs- tissue from a metastasis), we performed principle component analysis (PCA; Supplementary Fig. 2a, b) and hierarchical clustering (HC; Supplementary Fig. 2c), pseudo-coloring by consensus subtype either primary/met tissue (PCA) or the collection site (HC). Whereas the PC1 and PC2 separate samples into their indicated subtypes, all the metastatic tissue samples fall within the variance range of the primary tissue. Furthermore, with hierarchical clustering, the metastatic tissue samples do not group by either metastatic tissue nor metastatic site, indicating that tissue origin does not confound the gene expression analysis. Additionally, when performing the analysis with primary samples only (Supplementary Fig. 2d, e), the significant enrichment of Stroma-rich tumors within the metastatic UC-GENOME and IMvigor210 cohorts persists (Supplementary Fig. 2f).

Targeted DNA sequencing demonstrates a high prevalence of TP53 E285K mutations in UC

DNA sequencing of 591 genes was performed on 191 patients including 166 patients with RNA data (Fig. 1a and Supplementary Data 1). At least one non-silent variant was observed in 552 of the 591 genes sequenced. Similar to TCGA¹², we observed a high prevalence of non-silent variants in TP53 and RB1 and the chromatin modifying genes KMT2D, ARID1A, KDM6A, KMT2C, EP300, KMT2A, CREBBP, and SMARCA4, and DNA damage repair genes BRCA2, PRKDC, ATM, ERCC2, and FANCA (Fig. 2a). We observed frequent hotspot variants in genes associated with aberrant kinase signaling including FGFR3 (S249C) and PIK3CA (E545K) (Supplementary Fig. 3a, b). As previously described, FGFR3 mutations were enriched in LumP tumors (OR 7.2, p-val 6.1e-5)^5,12 (Fig. 2a). FGFR3-TACC3 gene fusions were found in 4% of evaluable samples (Fig. 2a) including one tumor which also had a FGFR3 S373C mutation. Analysis of CNVs identified amplifications of CCND1 (10%), MDM2 (9%), ERBB2 (7%), CCNE1 (2%), and EGFR (3%) at similar frequency as in TCGA¹².

**Fig. 2: Targeted DNA sequencing analysis demonstrates a high prevalence of *TP53 E285K* mutations in UC.**

The most frequently mutated gene, TP53, had an increased frequency of mutation in Ba/Sq and LumU tumors with a corresponding increase in an expression-based signature of p53 pathway dysfunction (Supplementary Fig. 3c). The most frequent TP53 variant was E285K (11%) (Fig. 2b and Supplementary Fig. 3d). The E285 variant was observed at a higher frequency in the metastatic cohorts (UC-GENOME, 10.8%, and IMvigor210, 6.9%) than in the primary TCGA cohort (5.9%) (Fig. 2b–d). Of all TP53 mutations across the PanCancer cohort from TCGA, which consists of >10,500 tumors (>4,250 TP53 mutations), the TP53 E285K variant was the 30th most frequently mutated amino acid (0.83%), with bladder cancers accounting for 39% (15/38) of the total E285 variants (Fig. 2e).

Mutational signature profiling reaffirms the importance of APOBEC in UC

Mutational signatures and shared patterns of single nucleotide variants (SNVs) can identify potential sources of mutagenic pressure, either intrinsic (mutations in DNA repair genes) or extrinsic (carcinogen exposure)¹⁴. By performing consensus clustering of the per-sample cosine similarity (CS) to the published COSMICv3 mutation signatures, we identified two distinct sample clusters (Fig. 3a, Supplementary Fig. 4a). Cluster 1 (K1) had high CS with defective DNA mismatch repair (MMR) signatures, while Cluster 2 (K2) had a high CS with APOBEC activity signatures (SBS2 and SBS13) (Fig. 3a). Independently, high TMB tumors were associated with increased APOBEC activity signatures SBS2 and SBS13 (Fig. 3b, c), this was represented in our clustering as K2, which in addition to having increased APOBEC, had increased TMB (p = 0.002). Cluster 2 was numerically, but not statistically enriched for TP53 mutations (K1 = 46/87 [53% mutant], K2 = 46/75 [61% mutant], p = 0.34). TMB (p = 0.002) and both APOBEC mutational signatures (SBS2 [p = 0.02] and SBS13 [p = 0.002]) were also significantly higher in the TP53 mutant samples (Supplementary Fig. 4b–d).

**Fig. 3: Mutational signature clustering and analysis reaffirms the importance of APOBEC in UC.**

The CS for SBS2, but not SBS13, was significantly elevated in micropapillary histology compared to tumors with squamous differentiation (p = 0.002) or all other UC tumors (p = 0.018) (Fig. 3e and Supplementary Fig. 4e). Micropapillary tumors were also significantly associated with increase POLE mutation signature versus squamous (p = 0.026) and other UC tumors (0.02; Supplementary Fig. 4f). In contrast, tumors with squamous differentiation had a significantly increased cosine similarity to mismatch repair defect signatures, such as SBS44, as compared to micropapillary (p = 2.3e-5) but not other UC tumors (Fig. 3f).

To evaluate the relationship between APOBEC activity and survival following either chemotherapy or ICI, samples were subdivided into 3 groups according to the CS to SBS2 and SBS13. Having a low CS to SBS13, but not SBS2, was associated with decreased survival as calculated from the initiation of treatment for both chemotherapy (p = 0.013) and ICIs (p = 0.011) (Fig. 3g, h and Supplementary Fig. 5a, b). Overall survival from the time of diagnosis, however, was not significantly different among the CS group for SBS2/SBS13 (Supplementary Fig. 5c, d), suggesting that CS levels of SBS2 and SBS13 are not merely prognostic.

Ba/Sq tumors are T cell inflamed but luminal tumors are enriched in memory B cells and plasma cells

Prior studies have described differential levels of T cell inflammation associated with the UC molecular subtypes^3,9,15. The tumor microenvironment (TME) of UC-GENOME was characterized through immune gene expression signatures (IGS) curated from the literature^{13,16,17,18,19}. As a whole, IGS scores were elevated in Ba/Sq and Stroma-rich subtypes as compared to the luminal subtypes (LumP, LumU, LumNS; Fig. 4a). The Ba/Sq subtype had significantly higher expression of genes within the T cell inflamed and IFNG signatures (Fig. 4b) while LumP and LumU had the lowest median expression. The Stroma-rich subtype had significantly higher scores, as compared to the other subtypes, for IGS related to the presence of stromal infiltration, namely, Fibroblast-TGF-β Response Signature [FTBRS] and EMT_Stroma_core_18 (Fig. 4c). These two signatures have previously been associated with resistance to ICI^18,19.

We next used MiXCR to infer TCR clonality of tumor infiltrating T cells (Fig. 4d). We identified high abundance of TCR-alpha (TRA) and TCR-beta (TRB), while TCR-gamma (TRG) and TCR-delta (TRD) chains did not have adequate abundance and were excluded from subsequent analyses. There were no differences in evenness (inverse of clonality) across consensus subtypes for TRA or TRB chains (Fig. 4d).

Immune cell proportions were estimated by CIBERSORTx. Using the UNC two subtype classification, most immune cell types were enriched in the UNC basal subtype as previously described¹⁵, with the exception that luminal tumors were high for plasma cells, activated mast cells, memory B cells, and activated dendritic cells (Fig. 4e). Using the consensus subtypes, memory B cells and activated dendritic cells were also significantly differentially enriched in LumP (Fig. 4f). No correlations were observed between IGS and TMB²⁰ or TMB and subtype (Supplementary Fig. 6a–d).

We next performed histologic immune phenotyping by dual staining FFPE slides with an antibody against CD8 and Masson’s Trichrome. Tumors were either called “Desert” if lacking CD8 + cells, “Excluded” if CD8 + cells were restricted to the stroma, or “Inflamed” if CD8 + cells were seen in both the stroma and infiltrating the tumor (see Methods section, Fig. 5a). 155 tumors were suitable to make phenotype calls, of which 93 (60%) were determined to be Excluded, 61 (39.4%) were labeled as Inflamed, and 1 (0.6%) called Desert (Fig. 5b). Inflamed tumors had significantly elevated IGS scores for multiple T cell signatures, including: Ayers T cell inflamed (p = 9.3e-13), Bindea T cell (p = 1.8e-7) and Ayers IFNG (p = 1.5e-12) (Fig. 5c–e). Since we had seen these aforementioned signatures are elevated in the Ba/Sq subtype (Fig. 4a, b), we asked whether there was an immune phenotype bias in any of the consensus subtypes. Indeed, the Ba/Sq subtype was significantly enriched in the Inflamed phenotype as compared to the Stroma-rich (p = 0.005), LumP (p = 0.01), and LumNS (p = 0.0009; Fig. 5f). This Ba/Sq enrichment was also present within the IMvigor210 dataset (Supplementary Fig. 6e). We next assessed the relationship between the histologic immune phenotype and clinical benefit (CR + PR + stable disease [SD]). Patients who experience benefit from ICI were split, 28/27 between Excluded and Inflamed, however patients who did not have any clinical benefit were significantly enriched for the Excluded phenotype (p = 0.0126) (Fig. 5g). Among Excluded tumors, those with response (complete response [CR] + partial response [PR]) to ICI also had, albeit not significantly, increased levels of Ayers T cell inflamed signature score (p = 0.19; Fig. 5h). Our analysis confirms the important role of stroma in promoting resistance to ICI^13,18 in addition to the association between IHC subtype and molecular subtype, the enrichment of IHC Inflamed and IHC Excluded tumors in the Ba/Sq, and Stroma-rich tumors respectively.

**Fig. 5: Immune phenotyping demonstrates enrichment of Inflamed tumors in Ba/Sq subtype.**

Clinical response to chemotherapy and ICI: an exploratory analysis

Previous studies have interrogated the predictive and prognostic value of UC subtypes⁴. UC-GENOME patients, stratified by consensus subtype were assessed for correlations to response and clinical benefit to chemotherapy or ICI. Overall, 55% of patients (33/60) responded to chemotherapy, however, within subtype analysis identified Stroma-rich (17/26) and LumU (7/9) as having >50% response (Supplementary Fig. 6f). In contrast, clinical benefit with chemotherapy was seen in 75% of tumors, with only LumNS (1/3) and NE-like (0/2) having a minority of patients derive clinical benefit (Fig. 6a). The response rate to ICI within UC-GENOME was 36% (34/93), with no subtype having greater than a 50% response (Supplementary Fig. 6g). In contrast to response, at least 50% of patients within all subtypes derived clinical benefit to ICI (with the exception of LumNS), with an overall clinical benefit rate of 65% (60/93; Fig. 6b).

**Fig. 6: Molecular associations with clinical benefit and response.**

ICI clinical benefit was associated with increased TMB (Fig. 6c), T cell inflamed, and IFNG IGS scores (Fig. 6d). Increased B cell abundance and signaling (Fig. 6e) along with an increase in the percentage of T follicular helper (Tfh) and M1 macrophage and decreased percentage of naïve CD4 T cells, were also associated with increased clinical benefit to ICI (Fig. 6e, f). However, immune infiltration signatures, FTBRS and EMT Stroma, were not associated with ICI response either within the cohort as a whole or solely within IHC Excluded tumors as documented previously¹³ (Supplementary Fig. 6h, i).

Somatic mutations in ERRC2 or a mutation in FANCC, ATM, or RB1 have been previously shown to predict response to neoadjuvant cisplatin-based chemotherapy^21,22. In UC-GENOME, ERCC2 mutations were associated with a significantly higher response to chemotherapy, validating this biomarker in a metastatic setting; however, it was not associated with ICI response nor improved overall survival with chemotherapy or ICI (Fig. 6g, h). There was no association with clinical benefit or response between chemotherapy and RB1, ATM, or FANCC mutations. However, patients with these alterations did have an improvement in survival from the time of initiation of treatment as compared to patients without these alterations. Patients with RB1, ATM, or FANCC mutations who received ICI showed an increased clinical benefit, but not response, along with an improved overall survival (Fig. 6i, j).

An integrated model of clinical and genomic features predicts response to ICI

The potential to better predict benefit from ICI has important implications for optimizing patient care. While several potential biomarkers have been examined in isolation, few studies have integrated clinical and genomic features to develop predictive models^{23,24,25,26,27}. We used clinical and immunogenomic data (Supplementary Data 2) from a discovery set of ICI treated patients with metastatic UC (IMvigor210) and applied elastic net (EN) logistic regression modeling to develop an integrated predictive model of ICI response. From 50-fold cross-validation on the discovery set, features of worse response included: B cell gene signature, stroma-rich consensus subtype, and claudin signature (Supplementary Fig. 7a), while features of better response were: baseline ECOG performance score 0, TMB, and M1 macrophage signature. The final EN model consisted of 25 predictors, similar to the cross-validation results (Fig. 7a).

**Fig. 7: An integrated elastic net model of clinical and genomic features predicts response to ICI.**

The performance of the final EN ICI predictor model was tested on the validation portion of the IMvigor210 dataset (n = 51), UC-GENOME, and an additional independent dataset (UNC-108)²⁸ (Fig. 7b). This model accurately predicted ICI response in the IMvigor210 validation set (AUC = 0.84) and in UNC-108 (AUC = 0.82) datasets. While the model did not perform as well in the UC-GENOME cohort, it did perform significantly better then random chance (AUC = 0.65, p = 0.009). Elevated model prediction scores were significantly associated with ICI response in the IMvigor210 validation set (Supplementary Fig. 7b), and in UNC-108 (Supplementary Fig. 7c). Importantly, the final EN ICI predictor was significantly better at predicting ICI response than a model using TMB alone (AUC = 0.84 vs. 0.68, p = 0.038) as well as other modeling approaches (Supplementary Fig. 7d, e). The final EN model was not predictive of response to chemotherapy (AUC = 0.536, p = 0.345) (Supplementary Fig. 7f) demonstrating its specificity to ICI as well as arguing that it does not merely model for favorable prognosis.

Discussion

We present the results of the foundational analyses of the BCAN-led UC-GENOME project including the integration of DNA and RNA sequencing data with clinical variables to describe the molecular landscape of UC. The UC-GENOME cohort and associated data allowed us to uncover a number of metastatic related features as we compared our results with non-metastatic datasets (TCGA and Kamoun et al.)^3,9 such as: a higher proportion of Stroma-rich molecular subtype and enrichment of the TP53 E285K hotspot mutation in primary tumors from metastatic patients. Additionally, we present and validate an integrated model of clinical and genomic factors associated with ICI outcome in this “real world” population of patients with metastatic UC.

Of the >4,000 TP53 mutations in Pan-TCGA tumors, the TP53 E285K mutation accounted for only 0.83% (n = 38). However, of the TP53 E285K mutations, 39% (n = 15) were from bladder tumors, even though the bladder samples only account for 5.6% of all TP53 mutations. Notably, this mutation is an APOBEC-attributable hotspot mutation²⁹ and was enriched in metastatic UC cohorts. One possible explanation is that upregulated APOBEC activity promotes tumor progression and TP53 E285K mutations are merely a manifestation of elevated APOBEC activity but do not promote metastases. Arguing against this, however, is that APOBEC-high tumors have longer overall survival. Moreover, of the top 10 most frequent APOBEC-associated hotspot mutations²⁹ only the TP53 E285K mutation was enriched in our metastatic UC-GENOME cohort relative to the non-metastatic cohort (TCGA-BLCA). These findings are consistent with a putative model whereby elevated APOBEC activity cultivates the APOBEC-associated TP53 E285K hotspot mutation, which actively promotes tumor progression. This model would account for the enrichment of TP53 E285K in patients with metastatic UC.

We examined previously reported putative biomarkers (ERCC2, ATM, FANCC, or RB1) associated with response to neoadjuvant chemotherapy (NAC)^21,22. In UC-GENOME, ERCC2 mutations were significantly enriched in chemotherapy responders while mutations in ATM, FANCC, and RB1 were not; neither biomarker was associated with prolonged OS. The lack of association between ATM, FANCC, or RB1 mutation and response may be secondary to the use of carboplatin (rather than cisplatin) in a significant proportion of patients as these biomarkers were developed in cohorts uniformly treated with cisplatin-based NAC. Alternatively, ATM, FANCC, and RB1 mutations may have different biological consequences in different disease states (MIBC -vs- metastatic). While Teo and colleagues have shown that patients with mutations in DNA damage and repair (DDR) genes have enhanced response to platinum-based chemotherapy³⁰, we present that ERCC2 mutations specifically may serve as a biomarker associated with platinum-based chemotherapy response in the metastatic setting.

We and others have documented subtype-specific differences in the tumor microenvironment and in general have concluded that tumors of the Ba/Sq subtype express higher levels of IGS. Other subtypes with relatively high IGS expression include UNC Claudin-low¹⁵ and Consensus Stroma-rich tumors³. We were therefore struck by the increased proportion of CIBERSORTx cell types: plasma cells, memory B cells, and activated dendritic cells in UNC Luminal tumors. These cell types are associated with tertiary lymphoid structures (TLS) and accumulating evidence correlates the presence of TLS with ICI response³¹. In this context, the correlation between the proportion of CIBERSORTx defined CD4 + Tfh cells and B cell IGS with ICI response are also consistent with TLS playing an important role in ICI response in the UC-GENOME cohort as multiple reports in breast cancer have shown that CD4 + Tfh cells lie in close proximity to germinal center B cells and are important regulators of antigen specific B cell responses^19,32.

Recent observations have underscored the importance of spatial biology of tumors as it relates to ICI^13,33. Additionally in bladder cancer, the importance of tumor stroma (i.e. cancer associated fibroblasts [CAFs]) as a negative predictor of ICI response has been demonstrated by multiple groups^13,18,28. Herein, we performed a similar analysis, staining for CD8 and trichome on FFPE sections with assignment of immune phenotype (Inflamed, Excluded, and Desert) by a GU pathologist. We report that the relationship between immune phenotype and molecular subtype finding that Ba/Sq tumors are enriched in the Inflamed immune phenotype while Stroma-rich, LumP, and LumU tumors have a higher proportion of Excluded tumors. Moreover, our work validates the negative impact of stroma on ICI response as there was a significant enrichment of Excluded tumors in UC-GENOME patients without clinical benefit to ICI. Nonetheless, much work on the spatial biology of bladder cancer remains to be performed.

We developed an elastic net model that integrates both clinical and immunogenomic variables to predict ICI response. Our EN ICI predictive model performs better then random chance or TMB alone in all validation datasets examined but had a higher degree of accuracy in IMvigor210 and UNC-108 than UC-GENOME (Fig. 7b). Drawing from prior published work from Jialu et al., one potential factor for the performance disparity between the cohorts could relate to the different methods used for generating the transcriptomic data (used in downstream IGS)³⁴. While IMvigor210 and UNC-108 were generated through capture-based RNA sequencing, the UC-GENOME transcriptome data were generated using total RNAseq. Indeed, when we calculated univariate regressions of each immune signature versus ICI response for each data set, we found that the regression of β-coefficients were strongly correlated between IMvigor210 and UNC-108 (R = 0.73) but weakly correlated between UC-GENOME and IMvigor210 (R = 0.36) and between UC-GENOME and UNC-108 (R = 0.36; Supplementary Fig. 7g–i).

Several challenges and limitations of the current study deserve attention. Although targeted DNA sequencing was employed rather than whole exome or whole genome sequencing, the significantly mutated genes detected were consistent with prior sequencing studies. While the study achieved its goal of providing NGS at no cost to the patient with reported treatment options including potential clinical trials for most patients (69%), only 5.0% of patients received a targeted therapy and 2.7% receiving targeted therapy on a clinical trial based on the NGS results. The likely reasons for the low utilization of NGS results to guide treatment include the absence of FDA approved targeted therapies along with the development and FDA approval of immune checkpoint inhibitors during the accrual period (July 2016–May 2019). As the goal of the study was to collect samples within a real-world experience, response was reported based on investigator assessment without the use of formal response criteria, this also limited the number of patients with fully annotated response data. Regular updates to cBioPortal will occur to provide the research community with the most up to date information for future analyses. In terms of the predictive modeling, the EN model can only be applied to the BLCA cohorts since it incorporates UC specific molecular subtypes (i.e. Consensus Stroma-rich). Finally, our analyses are correlative in nature and functional validation will be required to impart causality.

The UC-GENOME project provides a rich biobank (FFPE, plasma, PBMCs) of clinically annotated specimens and their associated molecular data, as described here. The overarching goal of this study was to provide researchers a resource for future collaborative translational research efforts, which can utilize the clinical data, foundational analyses, and bio-banked specimens of this project to advance the development of biomarkers and new treatments for patients with UC. To achieve this goal, The Bladder Cancer Advocacy Network (BCAN) will engage the research community by creating a novel request for proposal mechanism to utilize the UC-GENOME data and specimens toward its mission to advance bladder cancer research and support those impacted by the disease.

Methods

IRB approvals

UC-GENOME (ClinicalTrials.gov identifier: NCT02643043) was supported by the Bladder Cancer Advocacy Network (BCAN) and conducted at University of North Carolina at Chapel Hill, Fox Chase Cancer Center, Icahn School of Medicine at Mount Sinai, University of Washington/ Fred Hutchinson Cancer Research Center, Johns Hopkins University, University of Chicago, Memorial Sloan Kettering Cancer Center, and University of Southern California. The study was coordinated by the Hoosier Cancer Research Network (HCRN). The protocol was approved by all of the participating institutional review boards: University of North Carolina Institutional Review Board, Fox Chase Cancer Center Institutional Review Board, Icahn School of Medicine at Mount Sinai Institutional Review Board, Fred Hutch Institutional Review Board, Johns Hopkins Medicine Institutional Review Board, University of Chicago Institutional Review Board, Memorial Sloan Kettering Institutional Review Board, and USC Health Sciences Campus Institutional Review Board. All patients provided written informed consent. The study protocol is available as part of the Source Data file.

Eligible patients had histologically confirmed UC with metastatic disease at the time of registration and tumor tissue available/suitable for molecular analyses. Tumor tissue was obtained from the primary tumor site or in a minority of cases a metastatic site (Table 1). Tumor tissue and blood was collected and banked for future research (Fig. 1a).

As detailed in the trial protocol, the studies primary objectives included: (1) Estimate the proportion of subjects with metastatic UC enrolled who receive NGS and have a personalized report generated with potential treatment options. (2) Create a biospecimen and data repository by collecting and storing blood and archival tumor tissue (biospecimens) from subjects and linking molecular and biological information from those biospecimens to clinical data in order to promote future translational research in metastatic UC. Secondary objectives included (1) Estimate the proportion of subjects whose personalized report includes targeted therapy options (approved or investigational drugs). (2) Estimate the proportion of subjects who enroll in a clinical trial of targeted therapy based on NGS results. (3) Estimate the proportion of subjects who receive targeted therapy (outside of a clinical trial) based on NGS results. (4) Describe the demographics, treatment history and outcomes for subjects enrolled in the study. (5) Document the number and type of clinical trials and basic/translational science or other research projects based on the biospecimen and data repository.

RNA/DNA

A single block per patient was selected and 4–25 slides (5 and 10 um) from that corresponding block were submitted for processing along with an H&E stained slide. The H&E slide was reviewed, tumor circled, and an estimate of percentage tumor was made by a research pathologist. The pathologist determined how many slides for assay input and requested either full scrape of slide or macrodissection for each slide.

RNA sequencing

Total RNA sequencing libraries were generated from 500 ng of total RNA isolated from formalin-fixed paraffin-embedded (FFPE) tissue, using the Illumina TruSeq RiboZero Gold protocol (https://support.illumina.com/) and sequenced 75 bp paired-end on an Illumina HiSeq 4000. Reads were aligned to the hg38 genome using STAR³⁵ and genes were quantified using Salmon³⁶.

Molecular subtyping was performed on log2 transformed upper-quartile normalized expression data using the BLCAsubtyping³ and consensusMIBC³ R package. Gene annotation conversions were made using biomaRt^37,38. RNA markers were extracted from Robertson et al.⁹. and visualized using ComplexHeatmaps³⁹. To calculate the inter-cohort gene expression correlation, the TCGA BLCA dataset was filtered, retaining genes that were highly (median log2 expression ≥ 5) and variably expressed (STDEV ≥ 1; n = 4357 genes); this gene list was then used to filter the UC-GENOME, IMvigo210 and Kamoun datasets. The Spearman correlation was calculated using the median expression (within the indicated subtype/cohort) for each gene retained after filtering.

DNA sequencing

Tumor only targeted DNA sequencing was performed by Caris Life Sciences (Irving, Texas) on genomic DNA isolated from FFPE tumor samples using a minimum of 50 ng of DNA, of which >20% was required to be of tumor origin, as assessed by H&E staining. The DNA was sequenced via the Agilent custom designed SureSelect XT assay (Caris MI TumorSeek 592-Gene NGS Panel, details at www.carislifesciences.com) on the Illumina NextSeq platform. Copy number variation was determined by comparing the depth of sequencing of genomic loci to a diploid control as well as the known performance of these genomic loci. Gene fusion and variant transcript detection were performed on mRNA isolated from FFPE tumor samples using the Archer FusionPlex Solid Tumor Panel and sequenced on Illumina MiSeq.

Variants were analyzed with maftools⁴⁰. For visualization, only the most disruptive variant based on SIFT and PolyPhen classifications was chosen when a sample had multiple variants per gene. TCGA variant data was obtained from cBioPortal⁴¹.

Mutational signatures

The R package SomaticSignatures⁴² was used to identify mutational signatures in the UC-GENOME cohort (n = 191). Signatures were compared to 49 Single Base Substitution (SBS) signatures from COSMIC version 3¹⁴ with cosine similarity (CS). To determine the optimal number of somatic signature clusters we chose COSMIC signatures in which 10% of samples have greater than >0.25 cosine similarity (CS) and clustered samples using Consensus Cluster Plus (CCP)⁴³ for all samples. The CS signature heatmap was visualized with Complex Heatmaps³⁹. For boxplots and survival curves using the mutational signatures, the dataset was reduced to samples with both RNA and DNA sequencing data available (n = 176). Samples were binned into APOBEC High/Mid/Low by rank ordering the samples by the CS to the indicated APOBEC signature and splitting by tertiles.

Characterization of the tumor microenvironment and immune signatures

Z-scores were calculated based on all genes within previously published immune gene signatures on a per sample basis^{13,16,17,18,19}. Immune cell fractions were calculated by CIBERSORTx using the default parameters⁴⁴.

TCR clonality was analyzed via MiXCR from RNAseq data to generate TCR clonotype expression matrices for each sample. Data was analyzed with custom scripts as well as the tcR package.

CD8 (Cell Marque, catalog #108R-16, clone: SP16, dilution: 1:400)/Masson’s trichrome (Fisher Scientific, catalog #22-110-648) stained slides were reviewed by a fellowship trained genitourinary pathologist to determine immune phenotype. Slides were scanned at ×100 magnification to evaluate for overall tumor cellularity and staining quality. In areas of invasive tumor, slides were then evaluated at ×200 for CD8 + immune cell infiltrates (brown chromogen). As per Mariathasan¹³, “Desert” categorization required less than or equal to 10 CD8 + T-cells identified in the specimen, averaged over 20 fields (i.e., less than 100 CD8 + in 10 fields). “Excluded” categorization required localization of immune infiltrates to the stroma without extension into the tumor nests. These immune infiltrates were present in both the stroma located between nests in an invasive focus (intra-tumoral), and at the interface between invasive tumor and deeper stroma (peri-tumoral). “Inflamed” categorization required infiltration of CD8 + T cells into the tumor nests with direct contact of immune cells and tumor cells. Scattered CD8 + T cells seen within tumor, without overt stroma infiltrates, were identified in some cases and were not considered sufficient for “Inflamed” and were called “Exluded.

Elastic net

Eight clinical variables (TMB, TMB high [>10 mutations/Mb], ECOG performance status, age, sex, tobacco use, prior platinum, and prior BCG), consensus subtype, and 61 immune gene signature scores were evaluated as potential predictors of response to ICI. Three datasets, IMvigor210¹³, UNC-108²⁸, and UC-GENOME were used for discovery and validation. Within each dataset, continuous variables were standardized (mean = 0.5, SD = 0.5) and samples with any NA values were omitted. The IMvigor210 dataset was divided into discovery (n = 104) and validation (n = 51) sets with balanced immunotherapy response. In the discovery set, elastic net regression with 50-fold cross-validation was used to build an optimal logistic model for response using the R package caret (tuneLength = 15). The β-coefficient mean and 95% confidence interval for each predictor were calculated separately using 50-fold cross-validation with the tuning parameters from the optimal logistic model. Performance of the final model was evaluated in the validation portion of IMvigor210, UNC-108, and UC-GENOME. Performance of the final model in IMvigor210 was compared with the performance of random forest and gradient tree boosting using the R package caret.

Statistical analysis

Categorical variables were compared using Fisher’s exact or chi-square test. Continuous variable comparisons were made using t-test or Wilcoxon rank sum (in cases of non-normal distribution). All boxplots are shown with boxes representing the IQR and midline at the median. Error bars represent Q1/Q3 ± 1.5 × IQR. Correlations were performed using Pearson correlation unless otherwise stated. Multiple comparison correction was performed using Bonferroni correction. Survival analyses were performed using Kaplan–Meier with log-rank tests. Statistical analyses were performed using R unless otherwise noted.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

The UC-GENOME mutation annotation file (MAF), gene expression matrix and clinical data generated in this study have been deposited in the cBioPortal database under “Urothelial Carcinoma, Nature Communications (BCAN/HCRN 2022)” [https://www.cbioportal.org/study/summary?id=blca_bcan_hcrn_2022]. All RNA and DNA FASTQ files are available on dbGaP, phs003066.v1.p1 [https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs003066.v1.p1] under restricted access for disease specific research, with the exception of samples obtained from Johns Hopkins University (JHU). The JHU samples are only available to investigators from non-profit entities with an IRB conducting disease specific research, under dbGaP study ID phs003094,v1.p1 [https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs003094.v1.p1]. The data used to generate the figures in this study as well as the code for the elastic net classifier are provided in the Supplementary Information/Source Data file and also via FigShare.com [https://figshare.com/articles/journal_contribution/UC-GENOME/19491287]. Upper quartile normalized RSEM gene expression data and DNA sequencing data for TCGA were downloaded from the GDC legacy archive [https://portal.gdc.cancer.gov/legacy-archive/]. IMvigor210 data were downloaded from the European Genome-Phenome Archive, EGAS00001004343 [https://ega-archive.org/studies/EGAS00001004343]. UNC-108 RNA sequencing data were obtained from GEO, GSE176307 [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE176307] and DNA sequencing data were obtained through request of the authors. Source data are provided as a Source Data file.

Code availability

The elastic net modeling code has been provided in the Source Data file and via .

References

The Cancer Genome Atlas Research Network. Comprehensive molecular characterization of urothelial bladder carcinoma. Nature 507, 315–322 (2014).
Article ADS Google Scholar
Damrauer, J. S. et al. Intrinsic subtypes of high-grade bladder cancer reflect the hallmarks of breast cancer biology. Proc. Natl Acad. Sci. USA 111, 3110–3115 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Kamoun, A. et al. A consensus molecular classification of muscle-invasive bladder cancer. Eur. Urol. 77, 420–433 (2020).
Article PubMed Google Scholar
Choi, W. et al. Identification of distinct basal and luminal subtypes of muscle-invasive bladder cancer with different sensitivities to frontline chemotherapy. Cancer Cell 25, 152–165 (2014).
Article CAS PubMed PubMed Central Google Scholar
Marzouka, N. A. et al. A validation and extended description of the Lund taxonomy for urothelial carcinoma using the TCGA cohort. Sci. Rep.-UK 8, 3737 (2018).
Article ADS Google Scholar
Seiler, R. et al. Impact of molecular subtypes in muscle-invasive bladder cancer on predicting response and survival after neoadjuvant chemotherapy. Eur. Urol. 72, 544–554 (2017).
Article CAS PubMed Google Scholar
Rebouissou, S. et al. EGFR as a potential therapeutic target for a subset of muscle-invasive bladder cancers presenting a basal-like phenotype. Sci. Transl. Med. 6, 244ra91 (2014).
Article PubMed Google Scholar
Volkmer, J. P. et al. Three differentiation states risk-stratify bladder cancer into distinct subtypes. Proc. Natl Acad. Sci. USA 109, 2078–2083 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Robertson, A. G. et al. Comprehensive molecular characterization of muscle-invasive bladder cancer. Cell 174, 1033 (2018).
Article CAS PubMed Google Scholar
FDA. FDA limits the use of Tecentriq and Keytruda for some urothelial cancer patients. https://www.fda.gov/drugs/resources-information-approved-drugs/fda-limits-use-tecentriq-and-keytruda-some-urothelial-cancer-patients (2018).
Loriot, Y. et al. Erdafitinib in locally advanced or metastatic urothelial carcinoma. N. Engl. J. Med. 381, 338–348 (2019).
Article CAS PubMed Google Scholar
Robertson, A. G. et al. Comprehensive molecular characterization of muscle-invasive bladder cancer. Cell 171, 540–556.e25 (2017).
Article CAS PubMed PubMed Central Google Scholar
Mariathasan, S. et al. TGFbeta attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells. Nature 554, 544–548 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Alexandrov, L. B. et al. The repertoire of mutational signatures in human cancer. Nature 578, 94–101 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Kardos, J. et al. Claudin-low bladder tumors are immune infiltrated and actively immune suppressed. JCI Insight 1, e85902 (2016).
Article PubMed PubMed Central Google Scholar
Saito, R. et al. Molecular subtype-specific immunocompetent models of high-grade urothelial carcinoma reveal differential neoantigen expression and response to immunotherapy. Cancer Res. 78, 3954–3968 (2018).
Article CAS PubMed PubMed Central Google Scholar
Ayers, M. et al. IFN-gamma-related mRNA profile predicts clinical response to PD-1 blockade. J. Clin. Invest. 127, 2930–2940 (2017).
Article PubMed PubMed Central Google Scholar
Wang, L. et al. EMT- and stroma-related gene expression and resistance to PD-1 blockade in urothelial cancer. Nat. Commun. 9, 3503 (2018).
Article ADS PubMed PubMed Central Google Scholar
Hollern, D. P. et al. B cells and T follicular helper cells mediate response to checkpoint inhibitors in high mutation burden mouse models of breast cancer. Cell 179, 1191–1206.e21 (2019).
Article CAS PubMed PubMed Central Google Scholar
Spranger, S. et al. Density of immunogenic antigens does not explain the presence or absence of the T-cell-inflamed tumor microenvironment in melanoma. Proc. Natl Acad. Sci. USA 113, E7759–E7768 (2016).
Article CAS PubMed PubMed Central Google Scholar
Plimack, E. R. et al. Defects in DNA repair genes predict response to neoadjuvant cisplatin-based chemotherapy in muscle-invasive bladder cancer. Eur. Urol. 68, 959–967 (2015).
Article CAS PubMed PubMed Central Google Scholar
Allen, E. M. V. et al. Somatic ERCC2 mutations correlate with cisplatin sensitivity in muscle-invasive urothelial carcinoma. Cancer Discov. 4, 1140–1153 (2014).
Article PubMed PubMed Central Google Scholar
Johannet, P. et al. Using machine learning algorithms to predict immunotherapy response in patients with advanced melanoma. Clin. Cancer Res. 27, 131–140 (2021).
Article CAS PubMed Google Scholar
Polano, M. et al. A pan-cancer approach to predict responsiveness to immune checkpoint inhibitors by machine learning. Cancers 11, 1562 (2019).
Article CAS PubMed Central Google Scholar
Siah, K. W., Khozin, S., Wong, C. H. & Lo, A. W. Machine-learning and stochastic tumor growth models for predicting outcomes in patients with advanced non-small-cell lung cancer. JCO Clin. Cancer Inform. 3, 1–11 (2019).
Article PubMed Google Scholar
Lu, M. et al. A genomic signature for accurate classification and prediction of clinical outcomes in cancer patients treated with immune checkpoint blockade immunotherapy. Sci. Rep.-UK 10, 20575 (2020).
Article CAS Google Scholar
Chen, S. et al. Biomarkers of the response to immune checkpoint inhibitors in metastatic urothelial carcinoma. Front. Immunol. 11, 1900 (2020).
Article CAS PubMed PubMed Central Google Scholar
Rose, T. L. et al. Fibroblast growth factor receptor 3 alterations and response to immune checkpoint inhibition in metastatic urothelial cancer: a real world experience. Brit. J. Cancer 125, 1251–1260 (2021).
Article CAS PubMed PubMed Central Google Scholar
Shi, M. J. et al. Identification of new driver and passenger mutations within APOBEC-induced hotspot mutations in bladder cancer. Genome Med. 12, 85 (2020).
Article CAS PubMed PubMed Central Google Scholar
Teo, M. Y. et al. DNA damage response and repair gene alterations are associated with improved survival in patients with platinum-treated advanced urothelial carcinoma. Clin. Cancer Res. 23, 3610–3618 (2017).
Article CAS PubMed PubMed Central Google Scholar
Sautes-Fridman, C., Petitprez, F., Calderaro, J. & Fridman, W. H. Tertiary lymphoid structures in the era of cancer immunotherapy. Nat. Rev. Cancer 19, 307–325 (2019).
Article CAS PubMed Google Scholar
Gu-Trantien, C. et al. CXCL13-producing TFH cells link immune suppression and adaptive memory in human breast cancer. JCI Insight 2, e91487 (2017).
Article PubMed Central Google Scholar
Tumeh, P. C. et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 515, 568–571 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Li, J., Fu, C., Speed, T. P., Wang, W. & Symmans, W. F. Accurate RNA sequencing from formalin-fixed cancer tissue to represent high-quality transcriptome from frozen tissue. Jco Precis Oncol. 2018, 1–9 (2018).
Google Scholar
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Article CAS PubMed Google Scholar
Patro, R., Duggal, G., Love, M. I., Irizarry, R. A. & Kingsford, C. Salmon provides fast and bias-aware quantification of transcript expression. Nat. Methods 14, 417–419 (2017).
Article CAS PubMed PubMed Central Google Scholar
Durinck, S. et al. BioMart and bioconductor: a powerful link between biological databases and microarray data analysis. Bioinformatics 21, 3439–3440 (2005).
Article CAS PubMed Google Scholar
Durinck, S., Spellman, P. T., Birney, E. & Huber, W. Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt. Nat. Protoc. 4, 1184–1191 (2009).
Article CAS PubMed PubMed Central Google Scholar
Gu, Z., Eils, R. & Schlesner, M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinformatics 32, 2847–2849 (2016).
Article CAS PubMed Google Scholar
Mayakonda, A., Lin, D. C., Assenov, Y., Plass, C. & Koeffler, H. P. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 28, 1747–1756 (2018).
Article CAS PubMed PubMed Central Google Scholar
Cerami, E. et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2, 401–404 (2012).
Article PubMed Google Scholar
Gehring, J. S., Fischer, B., Lawrence, M. & Huber, W. SomaticSignatures: inferring mutational signatures from single-nucleotide variants. Bioinformatics 31, 3673–3675 (2015).
CAS PubMed PubMed Central Google Scholar
Wilkerson, M. D. & Hayes, D. N. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics 26, 1572–1573 (2010).
Article CAS PubMed PubMed Central Google Scholar
Newman, A. M. et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat. Biotechnol. 37, 773–782 (2019).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

This project was sponsored by the Bladder Cancer Advocacy Network (BCAN) through generous donations from the James Family Foundation, The JPB Foundation, and The Gerald C. McNamara and Renee K. Petrofes Charitable Fund. The sponsoring entity, BCAN, was involved in the study conception, planning, and manuscript review. The authors thank the Hoosier Cancer Research Network (HCRN) for their work compiling the data and specimens, the UNC Office of Genomic Research and cBioPortal for facilitating the sharing of genomic and clinical data, the UNC Pathology Services Core for help with the immune phenotyping and Caris Life Science for assistance compiling the genomic data. The authors would also like to acknowledge all of the patients and their families who participated in the UC-GENOME study. Additional funding was provided through the Cancer Center Core Support Grant (P30-CA016086) and R01-CA241810 (W.Y.K. and B.G.V.).

Author information

These authors contributed equally: Jeffrey S. Damrauer, Wolfgang Beckabir.
These authors jointly supervised this work: William Y. Kim, Matthew I. Milowsky.

Authors and Affiliations

Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC, USA
Jeffrey S. Damrauer, Wolfgang Beckabir, Jeff Klomp, Mi Zhou, Benjamin G. Vincent, Sara E. Wobker, Katherine A. Hoadley, William Y. Kim & Matthew I. Milowsky
Department of Microbiology and Immunology, University of North Carolina, Chapel Hill, NC, USA
Wolfgang Beckabir & Benjamin G. Vincent
Department of Pharmacology, University of North Carolina, Chapel Hill, NC, USA
Jeff Klomp & William Y. Kim
Department of Hematology and Oncology, Fox Chase Cancer Center, Temple Health, Philadelphia, PA, USA
Elizabeth R. Plimack
Division of Hematology and Medical Oncology, Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, New York, NY, USA
Matthew D. Galsky
Department of Medicine, Division of Medical Oncology, University of Washington, Seattle, USA
Petros Grivas
Clinical Research Division, Fred Hutchinson Cancer Center, Seattle, USA
Petros Grivas
Johns Hopkins University School of Medicine, Baltimore, MD, USA
Noah M. Hahn
Section of Hematology/Oncology, Department of Medicine, University of Chicago, Chicago, IL, USA
Peter H. O’Donnell
Memorial Sloan Kettering Cancer Center, New York, NY, USA
Gopa Iyer
University of Southern California Norris Comprehensive Cancer Center, Los Angeles, CA, USA
David I. Quinn
Division of Hematology, University of North Carolina, Chapel Hill, NC, USA
Benjamin G. Vincent
Department of Medicine, University of North Carolina, Chapel Hill, NC, USA
Benjamin G. Vincent, William Y. Kim & Matthew I. Milowsky
Curriculum in Bioinformatics and Computational Biology, Computational Medicine Program, University of North Carolina, Chapel Hill, USA
Benjamin G. Vincent
Bladder Cancer Advocacy Network, Bethesda, MD, USA
Diane Zipursky Quale
Department of Pathology and Laboratory Medicine, University of North Carolina, Chapel Hill, NC, USA
Sara E. Wobker
Department of Genetics, University of North Carolina, Chapel Hill, NC, USA
Katherine A. Hoadley & William Y. Kim
Division of Oncology, University of North Carolina, Chapel Hill, NC, USA
William Y. Kim & Matthew I. Milowsky

Authors

Jeffrey S. Damrauer
View author publications
You can also search for this author in PubMed Google Scholar
Wolfgang Beckabir
View author publications
You can also search for this author in PubMed Google Scholar
Jeff Klomp
View author publications
You can also search for this author in PubMed Google Scholar
Mi Zhou
View author publications
You can also search for this author in PubMed Google Scholar
Elizabeth R. Plimack
View author publications
You can also search for this author in PubMed Google Scholar
Matthew D. Galsky
View author publications
You can also search for this author in PubMed Google Scholar
Petros Grivas
View author publications
You can also search for this author in PubMed Google Scholar
Noah M. Hahn
View author publications
You can also search for this author in PubMed Google Scholar
Peter H. O’Donnell
View author publications
You can also search for this author in PubMed Google Scholar
Gopa Iyer
View author publications
You can also search for this author in PubMed Google Scholar
David I. Quinn
View author publications
You can also search for this author in PubMed Google Scholar
Benjamin G. Vincent
View author publications
You can also search for this author in PubMed Google Scholar
Diane Zipursky Quale
View author publications
You can also search for this author in PubMed Google Scholar
Sara E. Wobker
View author publications
You can also search for this author in PubMed Google Scholar
Katherine A. Hoadley
View author publications
You can also search for this author in PubMed Google Scholar
William Y. Kim
View author publications
You can also search for this author in PubMed Google Scholar
Matthew I. Milowsky
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Study conception, planning, and sample acquisition: E.R.P., M.G., P.G., N.M.H., P.O., G.I., D.I.Q., W. K., D.Q., and M.I.M. Data analysis, interpretation, and manuscript writing and review: J.S.D., W.B., J.K., M.Z., B.V., E.R.P., M.G., P.G., N.M.H., P.O., G.I., D.I.Q., K.A.H., D.Q., S.W. W.K., and M.I.M.

Corresponding authors

Correspondence to William Y. Kim or Matthew I. Milowsky.

Ethics declarations

Competing interests

Research support to the author’s institution from Caris Life Sciences (E.R.P.). Payment for advisory board involvement from Astellas, BMS, Merck, Pfizer, Roche/Genentech, and Seattle Genetics, employment Abbvie, Stock options Abbvie (D.I.Q.). No other authors have competing interests relating to this study.

Peer review

Peer review information

Nature Communications thanks Mototsugu Oya, Zlatko Trajanoski and the Other anonymous Reviewer(s) to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Description to Additional Supplementary Information

Supplementary Data 1

Supplementary Data 2

Reporting Summary

Source_Data_Files

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Damrauer, J.S., Beckabir, W., Klomp, J. et al. Collaborative study from the Bladder Cancer Advocacy Network for the genomic analysis of metastatic urothelial cancer. Nat Commun 13, 6658 (2022). https://doi.org/10.1038/s41467-022-33980-9

Download citation

Received: 03 September 2021
Accepted: 10 October 2022
Published: 04 November 2022
DOI: https://doi.org/10.1038/s41467-022-33980-9

This article is cited by

Optimizing identification of consensus molecular subtypes in muscle-invasive bladder cancer: a comparison of two sequencing methods and gene sets using FFPE specimens
- Florestan J. Koll
- Claudia Döring
- Peter J. Wild
BMC Cancer (2023)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.