Comprehensive molecular characterization of muscle invasive bladder cancer

Authors

A. Gordon Robertson

Jaegil Kim

Hikmat Al-Ahmadie

Joaquim Bellmunt

Guangwu Guo

Andrew D. Cherniack

Toshinori Hinoue

Peter W. Laird

Katherine A. Hoadley

Rehan Akbani

Mauro A. A. Castro

Ewan A. Gibb

Rupa S. Kanchi

Dmitry A. Gordenin

Sachet A. Shukla

Francisco Sanchez-Vega

Donna E. Hansel

Bogdan A. Czerniak

Victor E. Reuter

Xiaoping Su

TCGA Research Network

John N. Weinstein

David J. Kwiatkowski

Seth P. Lerner

Doi

10.1016/j.cell.2017.09.007

PMID: 28988769 · DOI: 10.1016/j.cell.2017.09.007 · Journal: Cell (2017)

TL;DR

Robertson et al. (TCGA bladder analysis working group) report the comprehensive molecular characterization of the full TCGA cohort of 412 chemotherapy-naive muscle-invasive bladder cancers (BLCA) — an expansion of the prior 131-tumor analysis — integrating whole-exome, SNP6 copy-number, RNA-seq, miRNA-seq, DNA methylation, and RPPA profiling. They identify 58 significantly mutated genes (34 not seen in the original 131-tumor cohort), confirm that APOBEC3A/APOBEC3B-driven mutagenesis dominates the mutational landscape (67% of all SNVs) and is paradoxically associated with the best survival, refine the mRNA expression taxonomy into five subtypes — luminal-papillary, luminal-infiltrated, luminal, basal-squamous, and a newly recognized neuronal subtype with the worst prognosis — and propose subtype-stratified therapeutic strategies. The dataset is published as cBioPortal study blca_tcga_pub_2017, superseding the 2014 blca_tcga_pub cohort.

Cohort & data

Key findings

  • High mutation burden, APOBEC-dominated. Mean 8.2 / median 5.8 non-synonymous mutations per Mb across 412 tumors; 51% C>T transitions and 27% C>G transversions. Bayesian NMF on the SNV catalogue resolved 5 mutation signatures: APOBEC-a, APOBEC-b, C>T at CpG (5-methylcytosine deamination), POLE (one ultra-mutated case with POLE P286R, >4000 SNVs), and an ERCC2 signature. APOBEC-a + APOBEC-b accounted for 67% of all SNVs; mutation load correlated with APOBEC3A and APOBEC3B expression PMID:28988769.
  • APOBEC mutagenesis is mostly clonal and likely early. 64% of all mutations, 62% of APOBEC-a-, and 75% of APOBEC-b-signature mutations were clonal (CCF ≥ 0.9), implying APOBEC mutagenic activity acts early in bladder cancer development PMID:28988769.
  • Mutation-signature clusters predict survival. Four clusters (MSig1–MSig4) showed distinct outcomes (p = 1.4×10⁻⁴). MSig1 (high-APOBEC, high-burden) had 75% 5-year survival; MSig2 (low mutation rate) had the poorest at 22%. MSig4 was enriched in ERCC2-signature mutations and ERCC2 mutations (24/39 cases, p = 10⁻¹³); ERCC2-signature mutations were highest in smokers carrying ERCC2 mutations (p = 6.9×10⁻¹¹) PMID:28988769.
  • Neoantigen load tracks mutation burden and predicts survival independently of age, AJCC stage, and squamous differentiation (p = 8×10⁻⁴; Table S2.12). HLA mutations were detected in 19/412 (4.6%) by Polysolver, enriched in MSig1 (p = 0.039) and in BCG-treated patients (4/35 = 11.4% vs 8/261 = 3.1%, p = 0.04) PMID:28988769.
  • 58 significantly mutated genes (SMGs) by MutSig 2CV (q < 0.1; Table S2.9), 34 of which were new vs the 131-tumor cohort. Seven new SMGs mutated in >10% of samples: KMT2C (18%), ATM (14%), FAT1 (12%), CREBBP (12%), ERBB2 (12%), SPTAN1 (12%), KMT2A (11%) PMID:28988769.
  • Mutual exclusivity / co-occurrence patterns (Tables S2.10–11, q < 0.2): mutually exclusive — CDKN2A–TP53, CDKN2A–RB1, CDKN2A–E2F3, TP53–MDM2, FGFR3–E2F3, FGFR3–RB1; co-occurring — TP53+RB1, TP53+E2F3, FGFR3+CDKN2A. FGFR3+CDKN2A co-occurred in 27 (7%) tumors, suggesting MIBCs that progressed from non-invasive precursors. 3/4 plasmacytoid tumors had nonsense CDH1 mutations (consistent with PMID:26901067) PMID:28988769.
  • Recurrent gene fusions (n=784 total). The most common was an intra-chromosomal FGFR3–TACC3 fusion (n=10). PPARG was involved in 4 TSEN2-PPARG and 2 MKRN2-PPARG fusions; PPARG was overexpressed in fusion-positive tumors (p = 6×10⁻³); 4/6 fusions yielded predicted proteins retaining both DNA- and ligand-binding domains. Bladder cell-line RNA-seq detected additional CASC15-PPARG (5637, 1A6) and NR2C2-PPARG (UC9) fusions also retaining functional PPARG domains PMID:28988769.
  • Copy-number landscape (Affymetrix SNP6 + GISTIC). 34 amplified and 32 deleted focal regions (q < 0.1). Recurrent amplifications targeted AHR, BCL2L1, CCND1, CCNE1, E2F3, EGFR, ERBB2, FGFR3, GATA3, KRAS, MDM2, MYCL, PPARG, NECTIN4 (PVRL4), SOX4, TERT, YWHAZ, ZNF703. The most common (22%) focal deletion targeted CDKN2A (9p21.3); RAD51B (14q24.1) deletions (n=10, 2%) appeared as a new recurrent deletion not detected in the original 131-tumor cohort. 221/412 (54%) tumors had whole-genome doubling PMID:28988769.
  • Five mRNA expression subtypes by NMF consensus clustering (n=408): luminal-papillary (35%), luminal-infiltrated (19%), luminal (6%), basal-squamous (35%), neuronal (5%); subtype-associated overall survival p = 4×10⁻⁴ PMID:28988769.
    • Luminal-papillary — papillary morphology (58% vs 20%, p < 10⁻¹³); enriched FGFR3 mutations (42/57; p < 10⁻⁹), amplification, overexpression, and FGFR3-TACC3 fusions (8/10; p = 4×10⁻³); low CIS scores; retained sonic-hedgehog signaling.
    • Luminal-infiltrated — lowest purity (median 0.46), strong smooth-muscle and myofibroblast signatures; corresponds to prior TCGA Cluster-II; high CD274 (PD-L1) and PDCD1 (PD-1).
    • Luminal — highest expression of uroplakins (UPK1A, UPK2) and umbrella-cell markers (KRT20, SNX31).
    • Basal-squamous — high CD44, KRT5, KRT6A, KRT14; squamous markers; 37/45 squamous-feature tumors (p < 10⁻¹¹); enriched TP53 mutations (p = 5×10⁻³); more common in females (33% vs 21%, p = 0.024); strongest immune signature.
    • Neuronal (new) — 20 tumors including 3/4 with neuroendocrine histology and 17 without histopathologic NE features; 10/20 (50%) had TP53+RB1 co-mutation or TP53+E2F3 amplification; 17/20 (85%) had p53/cell-cycle pathway alterations; poorest survival of the five subtypes (p = 4×10⁻⁴) PMID:28988769.
  • DNA methylation hypomethylation cluster 4 = luminal-papillary signature. A high-purity hypomethylation group with frequent FGFR3 mutation (p = 6×10⁻⁹), CDKN2A deletion (p = 4×10⁻¹³), no TP53 or RB1 mutations, papillary histology, node-negative disease, younger patients (median age 61 vs 69, p < 4×10⁻³), and better survival (p < 0.05) PMID:28988769.
  • 158 epigenetically silenced genes identified by integrating methylation + expression. CDKN2A, FAT1, and CASP8 were silenced in some tumors and (mutually exclusively) mutated in others. Top silenced novel hits included LXN (latexin, 27%), PARP6 (26%), SPATC1L (19%), NAPRT (13%). No promoter hypermethylation observed for TP53, PTEN, TSC1, TSC2, NF1, NF2, or RB1 PMID:28988769.
  • Pathway burden. p53/cell-cycle pathway inactivated in 89% of tumors (TP53 mutation 48%, MDM2 amplification 6%, MDM2 overexpression 19%); TP53 mutations enriched in genome-doubled tumors (p < 10⁻⁷). DNA repair: ATM 14%, ERCC2 9% (all non-silent ERCC2 missense, frequently in/around the helicase domain), RAD51B deletion 2%. Oncogene hotspots: FGFR3 S249C/Y373C (more frequent in T2 [21%] than T3/T4 [10%], p = 0.003; better survival, p = 0.04), PIK3CA 22% (helical E542/E545 dominant, n=54; kinase M1043/H1047 n=10), ERBB2 S310 in extracellular domain (24/57 = 42% of ERBB2 mutations, also APOBEC-attributable) PMID:28988769.
  • Chromatin-modifier mutations are pervasive and predominantly inactivating. 10 of 39 SMGs with mutation frequency > 5% were chromatin-modifying or chromatin-regulatory: histone demethylase KDM6A; methyltransferases KMT2A, KMT2C, KMT2D; acetylases CREBBP, EP300, KANSL1; SWI/SNF member ARID1A; Polycomb ASXL1, ASXL2. 50% frame-shift/nonsense vs 26% in other SMGs (p = 10⁻³⁰). ARID1A, CREBBP, and KDM6A are also targets of focal deletion (4.2%, 14.2%, 4.9% respectively) PMID:28988769.
  • lncRNA / miRNA expression subtypes refine the mRNA taxonomy. Four lncRNA consensus clusters and four miRNA consensus clusters were associated with purity, EMT score, CIS gene-set scores, and 5-year survival (lncRNA p = 0.015; miRNA p = 1.7×10⁻³). lncRNA cluster 3 (best-survival luminal-papillary subset) was depleted in TP53 mutations and enriched in FGFR3 mutations/fusions; high cancer-associated lncRNAs included MALAT1, NEAT1, UCA1, DANCR, GAS5, NORAD (LINC00657). miRNA subtype 3 had the best survival (low EMT, high miR-200) PMID:28988769.
  • Regulon analysis (VIPER-style on 23 candidate regulators). Independently confirmed in the Sjödahl et al. 308-sample cohort. Luminal-papillary cases activated 11 regulons consistent with GATA3, FOXA1, PPARG as luminal drivers; basal/squamous cases activated 12 regulons with TP63 and EGFR as discriminators between miRNA-defined sub-clusters PMID:28988769.
  • Microbe / virus integration. RNA-seq + WES + WGS evidence: HPV (n=11, 4 with genomic integration with breakpoints near BCL2L1, SLC2A1-AS1, DEC1, SEC16A, CCDC68); HHV4 (n=6, no integration); HHV5 (n=6, no integration); Polyomavirus (n=1, BK integration breakpoints near FIGN and LIMA1). Viral infection contributes to a small percentage of urothelial carcinomas PMID:28988769.
  • RPPA proteomic clusters. Five clusters with distinct outcomes (p = 0.019). Cluster 1 (epithelial/papillary) best survival; clusters 1 & 2 had elevated HER2 (ERBB2) suggesting candidates for anti-HER2 (trastuzumab/T-DM1); cluster 3 (proliferative/low signaling) high EGFR, candidate for EGFR inhibitors; clusters 4 & 5 EMT-high, worst outcomes PMID:28988769.
  • Multivariate survival model (LASSO Cox). Variables retained at |β| > 0.1 included AJCC stages III/IV, mRNA neuronal and luminal subtypes, low-mutation MSig2, miRNA subtype 4, and KLF4 regulon activity (all worse survival); luminal-infiltrated, age, GATA3 regulon activity, and MSig4 retained smaller coefficients. Tertile-thresholded score stratified the cohort into three risk groups (p < 0.001) PMID:28988769.

Genes & alterations

  • FGFR3 — Hotspot S249C/Y373C mutations dominate; enriched in luminal-papillary subtype (42/57; p < 10⁻⁹) and in lower-stage T2 tumors (21% vs 10% T3/T4, p = 0.003); associated with better survival (p = 0.04). Also amplified (5/5 in luminal-papillary) and engaged in recurrent FGFR3–TACC3 fusions (n=10). Identified by the authors as the dominant driver in 44% of luminal-papillary tumors and a candidate target for pan-FGFR tyrosine kinase inhibitors PMID:28988769.
  • TACC3 — Fusion partner of FGFR3 in the recurrent intra-chromosomal FGFR3-TACC3 fusion (n=10), an in-frame activating event particularly enriched in the luminal-papillary subtype PMID:28988769.
  • TP53 — Mutated in 48%; enriched in basal-squamous (p = 5×10⁻³) and neuronal subtypes; co-occurs with RB1 and E2F3 amplification, mutually exclusive with MDM2 amplification and with CDKN2A loss; enriched in genome-doubled tumors (p < 10⁻⁷). 10/20 neuronal tumors carry TP53+RB1 co-mutation or TP53+E2F3 PMID:28988769.
  • RB1 — Mostly inactivating mutations (17%), associated with reduced mRNA. Hallmark co-event with TP53 loss in the neuronal subtype PMID:28988769.
  • ERCC2 — Missense mutations in 9% (n=40), preferentially in/around the helicase domain (suggested dominant-negative effect). Defines the MSig4 mutational-signature cluster (24/39 ERCC2-mutant; p = 10⁻¹³); ERCC2-signature mutations are highest in smokers carrying ERCC2 mutations (p = 6.9×10⁻¹¹) PMID:28988769.
  • ATM — Mutated in 14% (n=57); now established as a recurrently mutated DNA-repair gene in MIBC PMID:28988769.
  • ERBB2 — 12% mutated (newly significant SMG); recurrent extracellular-domain S310 mutations (24/57 = 42% of ERBB2 mutations) attributable to APOBEC mutagenesis. Also recurrently amplified. RPPA HER2-high clusters 1 and 2 are candidates for trastuzumab or ado-trastuzumab-emtansine PMID:28988769.
  • PIK3CA — Mutated in 22% (n=100); helical-domain E542/E545 (n=54) more common than kinase M1043/H1047 (n=10); APOBEC-attributable PMID:28988769.
  • KRAS — Recurrently amplified focal SCNA in MIBC PMID:28988769.
  • CDKN2A — 7% mutated, 22% homozygous deletion (most common focal deletion); mutually exclusive with TP53, RB1, and E2F3 alterations; co-occurs with FGFR3 (27 tumors, 7%) — possibly MIBCs that progressed from non-invasive papillary precursors PMID:28988769.
  • CDKN1A — 11% mutations, predominantly inactivating PMID:28988769.
  • MDM2 — Amplified (CN > 4) in 6% and overexpressed (>2× median) in 19%; mutually exclusive with TP53 mutation PMID:28988769.
  • E2F3 — Recurrently amplified; mutually exclusive with CDKN2A loss and FGFR3 mutation; co-occurs with TP53 loss in the neuronal subtype PMID:28988769.
  • PPARG — Six recurrent fusions (4× TSEN2-PPARG, 2× MKRN2-PPARG) with overexpression (p = 6×10⁻³); 4/6 retained both DNA- and ligand-binding domains. Additional cell-line CASC15-PPARG and NR2C2-PPARG fusions identified. Implicated as a luminal driver alongside GATA3 and FOXA1 PMID:28988769.
  • KMT2C — Most frequently mutated newly recognized SMG (18%) PMID:28988769.
  • KMT2A — Newly recognized SMG, mutated in 11% PMID:28988769.
  • KMT2D — Histone methyltransferase chromatin-modifier SMG, predominantly inactivating mutations PMID:28988769.
  • KDM6A — Histone demethylase SMG; recurrent inactivating mutations + 4.9% deletions; loss is implicated in dependence on BRD4–EZH2 signaling and sensitivity to JQ1 and EZH2 inhibition (Ler et al. 2017; Wu et al. 2016, cited) PMID:28988769.
  • CREBBP — 12% mutated (newly recognized SMG), 14.2% deleted; predominantly inactivating PMID:28988769.
  • EP300 — Histone acetylase chromatin-modifier SMG; predominantly inactivating mutations PMID:28988769.
  • KANSL1 — Histone acetylase SMG PMID:28988769.
  • ARID1A — SWI/SNF chromatin remodeler; recurrent inactivating mutations + 4.2% deletions PMID:28988769.
  • ASXL1, ASXL2 — Polycomb-group chromatin-modifier SMGs with predominantly inactivating mutations PMID:28988769.
  • STAG2 — Recurrent SMG; one of four MutCN cluster definers (alongside FGFR3 and KDM6A) PMID:28988769.
  • FAT1 — 12% mutated newly recognized SMG; also subject to epigenetic silencing in mutually exclusive subset of tumors PMID:28988769.
  • SPTAN1 — 12% mutated newly recognized SMG PMID:28988769.
  • CASP8 — Mutated in some tumors, epigenetically silenced (mutually exclusively) in others PMID:28988769.
  • CDH1 — Nonsense mutations in 3 of 4 plasmacytoid-histology tumors, consistent with prior plasmacytoid-variant findings PMID:26901067 PMID:28988769.
  • POLE — P286R in a single ultra-mutated tumor (>4000 SNVs), defining the POLE mutational-signature subset PMID:28988769.
  • RAD51B — Recurrent focal deletions (n=10, 2%) at 14q24.1; new finding not seen in the original 131-tumor cohort PMID:28988769.
  • APOBEC3A, APOBEC3B — Expression correlates with APOBEC-signature mutation load; the dominant mutagenic source in MIBC accounting for 67% of all SNVs PMID:28988769.
  • GATA3 — Recurrently amplified; differentially expressed across mRNA subtypes; key luminal-driver transcription factor by regulon analysis; GATA3 regulon activity associated with better survival in the multivariate model PMID:28988769.
  • FOXA1 — Strongly expressed in luminal subtypes; identified as a luminal driver by regulon analysis PMID:28988769.
  • EGFR — Recurrently amplified focal SCNA; high protein expression in basal-squamous subtype and RPPA cluster 3 (proliferative/low-signaling), proposed candidates for EGFR inhibitors PMID:28988769.
  • CD274 (PD-L1) and PDCD1 (PD-1) — Strongly expressed in luminal-infiltrated subtype, the proposed checkpoint-immunotherapy-responsive subset; expression also high in basal-squamous PMID:28988769.
  • CTLA4 — Elevated in luminal-infiltrated and basal-squamous subtypes PMID:28988769.
  • TP63 — Discriminating regulon between miRNA clusters within the basal/squamous mRNA subtype (activated in miR-4, repressed in miR-2) PMID:28988769.
  • KLF4 — Regulon activity retained in the LASSO Cox model, associated with worse survival PMID:28988769.
  • BRD4, EZH2 — Identified by cited preclinical work (Ler et al. 2017; Wu et al. 2016) as a chromatin-modification growth pathway whose inhibition (e.g. JQ1, EZH2 inhibitors) is therapeutically tractable in KDM6A-loss bladder cancer PMID:28988769.
  • NECTIN4 (PVRL4) — Recurrent focal SCNA target; relevant to enfortumab-vedotin biology though the paper itself does not test ADCs PMID:28988769.
  • Epigenetically silenced novel hits — LXN, PARP6, NAPRT, SPATC1L — silenced in 27%, 26%, 13%, 19% of tumors respectively; candidate functional contributors and potential therapeutic-vulnerability biomarkers PMID:28988769.
  • lncRNAs MALAT1, NEAT1, UCA1 — Among the cancer-associated lncRNAs differentially abundant across the bladder cancer subtypes; high in lncRNA cluster 3 (best-survival luminal-papillary subset) PMID:28988769.
  • Uroplakins UPK1A, UPK2 and umbrella-cell markers KRT20, SNX31 — highest expression in the new luminal subtype, suggesting umbrella-cell-like differentiation PMID:28988769.
  • Basal/stem markers CD44, KRT5, KRT6A, KRT14 — defining markers of the basal-squamous subtype PMID:28988769.

Clinical implications

The authors propose a five-subtype framework for stratifying treatment in MIBC, explicitly hypothesis-generating and not yet ready for clinical decision-making (Figure 7) PMID:28988769:

  • Luminal-papillary (35%) — Characterized by FGFR3 mutations / FGFR3-TACC3 fusions / FGFR3 amplification, papillary histology, active SHH signaling, low CIS scores. Low risk of progression; preliminary data suggests low likelihood of response to cisplatin-based neoadjuvant chemotherapy (cisplatin, Seiler et al. 2017). Pan-FGFR tyrosine kinase inhibitors (e.g. erdafitinib, referenced via Karkera et al. 2017 / Nogova et al. 2017) are proposed as a rational targeted strategy PMID:28988769.
  • Luminal-infiltrated (19%) — High EMT/myofibroblast markers, high miR-200, intermediate CD274/CTLA4. Corresponds to TCGA Cluster-II reported to respond to anti-PD-L1 atezolizumab (Rosenberg et al. 2016). May be resistant to cisplatin-based chemotherapy; trials should validate this subtype as a negative predictive biomarker for chemo response and a positive biomarker for checkpoint blockade PMID:28988769.
  • Luminal (6%) — Optimal therapy not defined; future trials may compare neoadjuvant chemotherapy vs mutation-targeted therapy PMID:28988769.
  • Basal-squamous (35%) — Squamous differentiation, basal keratins, high CD274 and CTLA4, strong immune infiltrate. Both cisplatin-based neoadjuvant chemotherapy and immune checkpoint therapy (Sharma et al. 2016) are appropriate; head-to-head trials warranted PMID:28988769.
  • Neuronal (5%) — Neuroendocrine/neuronal expression program with high cell-cycle signature, often without the histopathologic features of NE tumors. Worst survival; identification depends on mRNA-seq or IHC. Etoposide–cisplatin regimens (as for NE tumors elsewhere) are proposed for neoadjuvant or metastatic settings, pending prospective testing PMID:28988769.

Additional therapeutic considerations explicitly raised by the authors:

  • MSig1 (high APOBEC, high mutation burden, neoantigen-rich) cancers should be recognized in checkpoint-inhibitor trials as having a much better baseline prognosis (75% 5-year survival), to avoid attributing host-immune-mediated benefit to therapy PMID:28988769.
  • p53/Rb-pathway-altered tumors (CDKN2A loss, retained Rb) are being targeted with palbociclib (PD-0332991) in NCT02334527 PMID:28988769.
  • HDAC inhibition with mocetinostat has completed Phase 2 accrual in metastatic urothelial carcinoma (NCT02236195); chromatin-modifier mutations are pervasive enough to motivate this class PMID:28988769.
  • BRD4–EZH2 axis in KDM6A-loss bladder cancer — preclinical benefit of BET inhibitor JQ1 and EZH2 inhibition (Ler et al. 2017; Wu et al. 2016) suggests a chromatin-modifier-rationale combination strategy PMID:28988769.
  • HER2-high RPPA clusters 1 and 2 are candidates for trastuzumab or ado-trastuzumab-emtansine (T-DM1) PMID:28988769.
  • EGFR-high RPPA cluster 3 is a candidate for EGFR inhibitors PMID:28988769.

Limitations & open questions

  • Limited follow-up. Median follow-up of 20.9 months for living patients limits long-term survival inference, though event rate (182 deaths) was high enough to be informative for the analyses presented PMID:28988769.
  • Subtype-stratified treatment recommendations are hypothesis-generating only. The authors explicitly state these are “not ready to be used for clinical decision making”; subsequent therapy was not included in the integrated analysis PMID:28988769.
  • APOBEC paradox unexplained mechanistically. The hypothesis that high APOBEC-driven mutation burden generates a host neo-antigen response that improves outcome is plausible but requires testing in independent cohorts and direct immunological assays PMID:28988769.
  • Origin and regulation of APOBEC expression in normal bladder is unknown — preventive strategies would require this knowledge PMID:28988769.
  • Functional significance of many recurrent fusions (e.g. ITGB6-LOC100505984, n=9) is uncertain PMID:28988769.
  • The neuronal subtype is defined by expression, not histology, in 17 of 20 cases; reproducible clinical-grade detection (mRNA assay or IHC panel) needs validation before therapy stratification PMID:28988769.
  • Many of the 158 epigenetically silenced genes may reflect background epigenetic noise rather than functional silencing; only a few (CDKN2A, FAT1, CASP8) were validated against the mutually-exclusive-with-mutation criterion PMID:28988769.
  • No validation cohort for the LASSO Cox multivariate risk model — internal tertile-stratified survival separation (p < 0.001) is not equivalent to external validation PMID:28988769.
  • Cross-cohort generalizability. Regulon analysis was confirmed in the Sjödahl 308-sample cohort, but mRNA subtype prevalences and survival associations should be re-tested in non-TCGA datasets — particularly in non-Western and cisplatin-treated populations PMID:28988769.
  • The 12 patients receiving neoadjuvant chemotherapy after tumor acquisition are too few to test the proposed subtype-by-treatment interactions internally PMID:28988769.

Citations from this paper used in the wiki

  • “We report a comprehensive analysis of 412 muscle-invasive bladder cancers characterized by multiple TCGA analytical platforms. Fifty-eight genes were significantly mutated, and the overall mutational load was associated with APOBEC-signature mutagenesis.” (Summary).
  • “Clustering by mutation signature identified a high-mutation subset with 75% 5-year survival.” (Summary).
  • “mRNA expression clustering refined prior clustering analyses and identified a poor-survival ‘neuronal’ subtype in which the majority of tumors lacked small cell or neuroendocrine histology.” (Summary).
  • “Many of the focal SCNAs involved genes known to be amplified in bladder cancer, including AHR, BCL2L1, CCND1, CCNE1, E2F3, EGFR, ERBB2, FGFR3, GATA3, KRAS, MDM2, MYCL1, PPARG, PVRL4, SOX4, TERT, YWHAZ and ZNF703.” (Somatic DNA Alterations).
  • “The APOBEC-a and -b signatures accounted for 67% of all SNVs.” (Somatic DNA Alterations).
  • “Patients with MSig1 cancers (high APOBEC-signature mutagenesis and high mutation burden) showed an exceptional 75% 5-year survival probability.” (Somatic DNA Alterations).
  • “MutSig 2CV identified 58 significantly mutated genes (SMGs) (q < 0.1).” (Somatic DNA Alterations).
  • “7 of the 34 genes were mutated in > 10% of samples: KMT2C (18%), ATM (14%), FAT1 (12%), CREBBP (12%), ERBB2 (12%), SPTAN1 (12%), and KMT2A (11%).” (Somatic DNA Alterations).
  • “The most common was an intra-chromosomal FGFR3-TACC3 fusion (n = 10).” (Somatic DNA Alterations).
  • “PPARG was involved in 4 TSEN2-PPARG and 2 MKRN2-PPARG fusions, and PPARG expression levels were higher than in samples without such fusions (p = 6 × 10−3).” (Somatic DNA Alterations).
  • “Unbiased NMF consensus clustering of RNA-seq data (n = 408) identified five expression subtypes: luminal-papillary (n = 142, 35%), luminal-infiltrated (n = 78, 19%), luminal (n = 26, 6%), basal-squamous (n = 142, 35%), and neuronal (n = 20, 5%).” (mRNA Expression-Based Molecular Subtypes).
  • “the neuronal subtype … had the poorest survival (p = 4·10−4, log-rank test), consistent with the known aggressive phenotype of NE bladder cancers.” (mRNA Expression-Based Molecular Subtypes).
  • “The p53/Cell Cycle pathway was inactivated in 89% of tumors, with TP53 mutations in 48%, MDM2 amplification (copy number > 4) in 6%, and MDM2 overexpression (>2-fold above the median) in 19%.” (Altered Pathways).
  • “Ten of the 39 SMGs with mutation frequency >5% were in chromatin-modifying or chromatin-regulatory genes.” (Altered Pathways).
  • “Etoposide-cisplatin therapy is recommended in neoadjuvant and metastatic settings, as for neuroendocrine neoplasms arising in other sites, but this should also be tested in prospective clinical trials.” (Discussion).
  • “The luminal-infiltrated subtype … has been reported to respond to immune checkpoint therapy with atezolizumab in patients with metastatic or unresectable bladder cancer (Rosenberg et al., 2016).” (Discussion).
  • “Recently, a Phase 2 study of Mocetinostat, a histone deacetylase inhibitor, in patients with locally advanced or metastatic urothelial carcinoma has completed accrual and results are awaited (NCT02236195).” (Discussion).
  • “the p53/Rb pathway is being targeted in a multicenter phase II trial evaluating palbociclib (PD-0332991) in patients with metastatic urothelial carcinoma who have cyclin-dependent kinase inhibitor 2A (CDKN2A) loss and retained retinoblastoma (Rb) expression (NCT02334527).” (Discussion).

This page was processed by crosslinker on 2026-05-15.