Genomic heterogeneity as a barrier to precision oncology in urothelial cancer

Authors

Clinton TN

Chen Z

Wise H

Lenis AT

Chavan S

Donoghue MTA

Almassi N

Chu CE

Dason S

Rao P

Rodrigues JA

Vasani NB

Ridouani F

Rosenberg JE

Bajorin DF

Teo MY

Bochner BH

Berger MF

Ostrovnaya I

Pietzak EJ

Iyer G

Gao SP

Hu W

Al-Ahmadie HA

Solit DB

Doi

10.1016/j.celrep.2022.111859

PMID: 36543146 · DOI: 10.1016/j.celrep.2022.111859 · Journal: Cell Reports (2022)

TL;DR

This study from Memorial Sloan Kettering analyzed 1,313 patients with bladder urothelial carcinoma profiled by MSK-IMPACT, including 119 matched primary-metastasis tumor pairs and 123 tumor-cfDNA pairs, to define the concordance of actionable genomic alterations between primary and metastatic disease sites. They found 23% discordance of actionable mutations (FGFR3, PIK3CA, TSC1, ERBB2) between paired sites, with ARID1A mutations exclusively appearing in metastatic samples when discordant (16% of ARID1A-mutant cases). Plasma cfDNA identified 17% of targetable alterations missed by tissue sequencing, suggesting cfDNA complements tissue-based profiling for treatment selection.

Cohort & data

  • 1,313 patients with bladder urothelial carcinoma prospectively sequenced at MSK (2014-2021) using MSK-IMPACT
  • 22 primary-metastasis pairs analyzed by whole-exome sequencing (tumor purity >=25% by FACETS)
  • 119 matched primary-metastasis pairs analyzed by MSK-IMPACT targeted sequencing (after excluding 2 MSI-H, 3 separate primaries, 4 reverse-order pairs)
  • 45 patients with primary tumor, metastasis, and plasma cfDNA via MSK-ACCESS (~20,000x raw coverage, 1,000x duplex)
  • 123 patients with at least one tumor sample and cfDNA for expanded tumor-cfDNA concordance analysis
  • Cancer types: BLCA, UTUC
  • Dataset: paired_bladder_2022 (cBioPortal)

Key findings

  • Mean mutational concordance by WES was only 42% (range 6%-84%) between primary and metastatic tumors, consistent with early branched evolution
  • TMB was not significantly different between primary and metastatic sites (one-sided Wilcoxon signed rank, p = 0.1; MSK-IMPACT mean 11.7 vs 11.2)
  • 23% of potentially actionable gene mutations (FGFR3, PIK3CA, TSC1, ERBB2) were discordant between primary-metastasis pairs
  • Discordance rates for individual targetable genes: ERBB2 38%, PIK3CA 27%, TSC1 17%, FGFR3 9%
  • ARID1A mutations, when discordant, were always exclusive to the metastatic sample (16% of ARID1A-mutant patients); resampling of primary tumors confirmed absence
  • Among 45 patients with tumor + cfDNA: 20% of OncoKB-defined targetable mutations were identified only by cfDNA
  • In expanded 123-patient cfDNA analysis: 60% of targetable alterations concordant, 17% exclusive to cfDNA, 23% exclusive to tumor tissue
  • One patient showed convergent evolution with 4 distinct oncogenic FGFR3 alterations (3 hotspot mutations + FGFR3-TACC3 fusion) detected in cfDNA but only 1 in the metastatic biopsy
  • TP53 (q < 0.001) and RB1 mutations more frequent in higher-grade/stage tumors; FGFR3 (q < 0.001) and PIK3CA (q = 0.04) more frequent in lower-grade/stage

Genes & alterations

  • FGFR3 — mutations and FGFR3-TACC3 fusions; more common in low-grade/stage; 9% discordance in primary-met pairs; cfDNA detected resistance mutations (N540S, K650E, V553M) during erdafitinib therapy
  • ARID1A — truncating mutations (e.g., E1783*); enriched in metastatic samples (28%) vs low-grade non-invasive (14%); when discordant, always exclusive to metastasis
  • TP53 — mutations (e.g., V157F) more frequent in high-grade invasive and metastatic tumors (q < 0.001)
  • RB1 — mutations enriched in higher-grade/stage tumors (q < 0.001)
  • ERBB2 — mutations and amplifications; more frequent in advanced disease (q = 0.001); 38% discordance in matched pairs
  • PIK3CA — activating mutations (e.g., E545K); more common in low-grade tumors; 27% discordance rate
  • TSC1 — 17% discordance between primary and metastatic sites
  • KDM6A — chromatin modifier commonly mutated in bladder cancer; shared trunk mutation in phylogenetic analyses
  • CREBBP — shared trunk mutation identified in phylogenetic analyses
  • CDKN1A — shared trunk mutation in representative phylogenetic example
  • TERT — promoter mutations shared between primary and metastatic tumors

Clinical implications

  • Archival primary tumor tissue may fail to detect actionable mutations in up to 35% of patients with metastatic bladder cancer, arguing against sole reliance on primary tumor profiling for treatment selection
  • cfDNA analysis is complementary to tissue sequencing and may be preferred for patients with metastatic disease who have received extensive prior therapy
  • FGFR3 discordance between primary and metastatic sites has direct implications for erdafitinib eligibility testing; both tissue and cfDNA may be required
  • ARID1A as a late-arising metastasis-associated alteration may serve as a biomarker for immune checkpoint inhibitor sensitivity and EZH2 inhibitor sensitivity
  • The subclonality and discordance of actionable mutations in bladder cancer may partly explain lower targeted therapy efficacy compared with lung cancer, where driver mutations are typically clonal

Limitations & open questions

  • Variability in time between primary and metastatic specimen collection (median 10.5 months, IQR 2.6-25.5) and heterogeneity in intervening treatments
  • 41% of patients received systemic therapy between primary and metastatic sample collection, which may confound concordance estimates
  • Analysis limited to DNA-level alterations; epigenetic and tumor microenvironmental differences not assessed
  • Variable interval between metastatic biopsy and cfDNA collection may contribute to observed discordance
  • Whether ARID1A mutations directly promote metastasis or are passengers in metastatic clones remains to be established
  • Clinical impact of subclonal targetable mutations on actual treatment response was not evaluated prospectively

Citations from this paper used in the wiki

  • “23% of potentially actionable gene mutations (FGFR3, PIK3CA, TSC1, ERBB2) were discordant in the primary-metastasis pairs”
  • “ARID1A mutations were present in the metastases of three patients but absent in their corresponding primary tumors” (WES cohort)
  • “ARID1A mutations… were exclusive to the metastatic samples of 16% of patients in which an ARID1A mutation was detected in either” (MSK-IMPACT cohort)
  • “20% of targetable mutations as defined by the OncoKB knowledgebase were identified only by the plasma cfDNA analysis”
  • “60% of targetable alterations were concordant between tumor and plasma cfDNA, 17% were exclusive to cfDNA only, and 23% were exclusive to tumor samples”
  • “analysis of archival tumor tissue collected from the primary bladder tumor site will fail to detect clinically actionable mutations in as many as 35% of patients with metastatic bladder cancer”

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