Next-generation Sequencing of Nonmuscle Invasive Bladder Cancer Reveals Potential Biomarkers and Rational Therapeutic Targets

Authors

Eugene J. Pietzak

Aditya Bagrodia

Eugene K. Cha

Esther N. Drill

Gopa Iyer

Sumit Isharwal

Irina Ostrovnaya

Priscilla Baez

Qiang Li

Michael F. Berger

Ahmet Zehir

Nikolaus Schultz

Jonathan E. Rosenberg

Dean F. Bajorin

Guido Dalbagni

Hikmat Al-Ahmadie

David B. Solit

Bernard H. Bochner

Doi

10.1016/j.eururo.2017.05.032

PMID: 28583311 · DOI: 10.1016/j.eururo.2017.05.032 · Journal: European Urology (2017)

TL;DR

Pietzak et al. performed targeted exon capture sequencing (MSK-IMPACT 341- or 410-gene panels) on pretreatment index tumors and matched germline DNA from 105 patients with nonmuscle invasive bladder cancer (NMIBC) at Memorial Sloan Kettering — the largest NGS effort focused on NMIBC to date. They found TERT promoter mutations (73%) and chromatin-modifying gene alterations (69%) at high prevalence across all stages and grades, suggesting these are early events in urothelial carcinogenesis. ERBB2 and FGFR3 alterations occurred mutually exclusively in 57% of high-grade tumors. DNA damage repair (DDR) gene alterations were found in 30% of high-grade NMIBC — a rate similar to muscle-invasive disease — and were associated with significantly higher mutational burden (p < 0.001). Most importantly, ARID1A mutations were associated with a 3.14-fold increased risk of recurrence after intravesical bacillus Calmette-Guérin (BCG) therapy (95% CI 1.51–6.51, p = 0.002) PMID:28583311.

Cohort & data

  • Primary cohort: 105 patients with NMIBC, all treatment-naive index tumors with matched germline DNA, profiled on the MSK-IMPACT panel using the 341-gene or updated 410-gene version under an IRB-approved protocol at Memorial Sloan Kettering Cancer Center PMID:28583311.
  • Cohort composition: Low-grade Ta (LGTa, n = 23), high-grade Tis (HGTis, n = 12), high-grade Ta (HGTa, n = 32), and high-grade T1 (HGT1, n = 38). All HGT1 tumors had restaging TUR with confirmation of uninvolved detrusor muscle PMID:28583311.
  • BCG sub-cohort: 62 of the 82 high-grade NMIBC patients received a 6-week induction course of BCG without maintenance therapy and were used for genomic correlates of recurrence PMID:28583311.
  • Comparator MIBC cohorts: 40 pretreatment index muscle-invasive bladder cancer (MIBC) tumors from MSK (“MSK-MIBC”) sequenced on the same MSK-IMPACT protocol, plus 98 MIBC specimens from the TCGA bladder cohort with no prior history of NMIBC PMID:28583311.
  • Cancer type: Nonmuscle invasive bladder urothelial carcinoma — BLCA (note: OncoTree does not have a separate code distinguishing NMIBC from MIBC).
  • Dataset: blca_nmibc_2017 — data made publicly available through the cBioPortal for Cancer Genomics PMID:28583311.
  • Median follow-up: 24.4 months for the 100 patients managed by TUR ± adjuvant intravesical therapy; recurrences occurred in 46 of 100 PMID:28583311.

Key findings

  • TERT promoter dominates the mutational landscape. TERT promoter mutations were the single most frequent alteration at 73% of NMIBC tumors, present at similar frequency across stage (p = 0.2) and grade (p = 0.15): 61% (14/23) of LGTa, 88% (28/32) of HGTa, 79% (30/38) of HGT1, and 85% (34/40) of MSK-MIBC. The TCGA whole-exome data did not interrogate the TERT promoter region PMID:28583311.
  • Chromatin-modifying gene alterations are nearly universal. Alterations in chromatin-modifying genes occurred in 69% (72/105) of NMIBC tumors. KDM6A was altered in 38% overall (52% LGTa, 38% HGTa, 32% HGT1) and ARID1A in 21% overall. Neither showed a statistically significant association with grade or stage PMID:28583311.
  • Mutually exclusive ERBB2/FGFR3 alteration pattern. 57% (47/82) of high-grade NMIBC tumors harbored alterations in either ERBB2 or FGFR3, and the alterations were mutually exclusive (Supplementary Table 7). FGFR3 alterations decreased with grade/stage (LGTa 83%; HGTa 59%; HGT1 34%; MSK-MIBC 8%; TCGA-MIBC 16%); ERBB2 alterations were almost exclusively in high-grade tumors, with hotspot S310F mutations and amplification observed especially in HGTis (6/12 Tis tumors had ERBB2 alterations) PMID:28583311.
  • FGFR3 fusions identified in NMIBC. Four FGFR3 fusions were identified, including a FGFR3-TNIP2 fusion (predicted activating; in-frame FGFR3 kinase domain) and an FGFR3-TACC3 fusion in an LGTa tumor PMID:28583311.
  • STAG2 mutations enriched in low-grade Ta disease. STAG2 alterations occurred in 23% of NMIBC overall, but truncating STAG2 mutations were significantly enriched in LGTa tumors (39% [9/23] LGTa vs 16% HGTa, 24% HGT1, 5% MSK-MIBC, 15% TCGA-MIBC; p = 0.046) PMID:28583311.
  • TP53/cell-cycle pathway alterations rise with stage. TP53/MDM2 alteration rates increased stepwise with stage (p < 0.001) and grade (p < 0.001) — 9% LGTa, 19% HGTa, 45% HGT1, 75% MSK-MIBC. Cell-cycle gene alterations (RB1, CCND1, CDKN1A, CDKN2A) similarly increased with stage (p = 0.028) and grade (p = 0.009) PMID:28583311.
  • High DDR alteration burden in high-grade NMIBC. Deleterious DDR gene alterations were identified in 30% (25/82) of high-grade NMIBC versus only 4% (1/23) of low-grade tumors (p = 0.012). The high-grade NMIBC rate was similar to the MSK-MIBC cohort (33%, 13/40). ERCC2 missense mutations were the most common DDR alteration (17%, 14/82 high-grade NMIBC; 20% of MSK-MIBC), clustering around the conserved helicase domain. Other affected DDR genes: ATM, BRCA1, BRCA2, ERCC4, PALB2, CHEK2, FANCC, MSH6 PMID:28583311.
  • DDR alterations correlate with elevated mutational burden. Tumors with deleterious DDR gene alterations had significantly higher mutational burden than DDR-intact tumors (median 26, IQR 15–36 vs median 8, IQR 5–12 mutations/Mb; p < 0.001). ERCC2-mutated tumors specifically had median 29 (IQR 23–37) vs 8 (IQR 6–14) for ERCC2 wild-type (p < 0.001). High-grade NMIBC had a mutational burden similar to MIBC (median 9, IQR 6–21 vs 10, IQR 7–19; p = 0.84) and significantly higher than low-grade NMIBC (7, IQR 5–10; p = 0.032) PMID:28583311.
  • ARID1A is the only gene significantly associated with BCG recurrence. Among all genes altered in ≥5 tumors of the 62-patient BCG cohort, only ARID1A mutations were significantly associated with recurrence after BCG (HR = 3.14, 95% CI 1.51–6.51, p = 0.002), surviving multiple-comparison correction (p = 0.04). The association persisted when ARID1A missense VUS were included (HR = 3.08, 95% CI 1.49–6.35, p = 0.002) and in the larger 100-patient TUR cohort (HR = 2.07, 95% CI 1.10–3.88, p = 0.024). 32 of 62 patients (52%) recurred at median 24 mo follow-up PMID:28583311.
  • TP53 status, ERBB2/FGFR3 status, and mutational count did NOT predict BCG recurrence. TP53 alterations alone or in combination with MDM2 showed no association with recurrence (p = 0.94 and p = 0.36). Tumors with ERBB2 and FGFR3 alterations had similar recurrence rates as wild-type (p = 0.3). Mutational count on MSK-IMPACT was not associated with BCG recurrence (HR = 0.96, 95% CI 0.83–1.1, p = 0.3) PMID:28583311.

Genes & alterations

  • TERT — Promoter mutations in 73% of NMIBC; uniformly high across grade and stage; supports use as a noninvasive urinary screening/surveillance biomarker.
  • FGFR3 — Hotspot mutations in 49% overall, more frequent in lower grade/stage. Four fusions identified, including a novel FGFR3-TNIP2 and an FGFR3-TACC3 fusion in a LGTa tumor. Mutually exclusive with ERBB2.
  • ERBB2 — Hotspot S310F missense and amplification, almost exclusively in high-grade NMIBC, including 6/12 Tis specimens. Together with FGFR3 covers 57% of high-grade NMIBC.
  • ARID1A — Mutations in 21% of NMIBC overall; the only single gene significantly associated with recurrence after BCG (HR = 3.14, p = 0.002).
  • KDM6A — Most commonly altered chromatin-modifying gene (38% of NMIBC).
  • STAG2 — 23% overall; truncating mutations enriched in LGTa (p = 0.046), supporting the lower-grade/stage interpretation in the conflicting literature.
  • TP53 / MDM2 — Alterations rise stepwise with stage/grade (p < 0.001).
  • RB1, CCND1, CDKN1A, CDKN2A — Cell-cycle alterations more common at higher stage/grade.
  • PIK3CA — Altered in 26% of NMIBC; part of the broadly altered RTK/PI3K pathway (79% of tumors).
  • ERCC2 — Most common DDR gene alteration (17% of high-grade NMIBC), missense mutations clustering in conserved helicase domains; associated with markedly elevated mutational burden.
  • ATM, BRCA1, BRCA2, ERCC4, PALB2, CHEK2, FANCC, MSH6 — Less frequent DDR alterations contributing to the 30% high-grade NMIBC DDR-altered fraction.
  • POLE — One LGTa harbored a truncating H1901Lfs*15 mutation deemed likely non-functional (3’ end, no hypermutation phenotype).

Clinical implications

  • Targeted therapy opportunity: 57% of high-grade NMIBC harbor alterations in ERBB2 or FGFR3 — a strong rationale for trials of targeted kinase inhibitors as alternatives or adjuncts to BCG, particularly with the development of less toxic, highly selective FGFR3 inhibitors and intravesical delivery PMID:28583311.
  • ARID1A as a candidate predictive biomarker for BCG response. ARID1A mutations associate with significantly worse recurrence-free survival after induction BCG (HR = 3.14, p = 0.002). If confirmed in independent cohorts, ARID1A status could stratify patients for BCG and motivate testing of EZH2 methyltransferase inhibitors, which exhibit synthetic lethality in ARID1A-mutated cancers PMID:28583311.
  • High mutational burden + DDR alterations support checkpoint immunotherapy trials. High-grade NMIBC has a mutational burden comparable to MIBC, and DDR-altered tumors carry markedly elevated burden. Given the established mutational-load → checkpoint-inhibitor response association in metastatic urothelial carcinoma (e.g. atezolizumab), this profile supports ongoing trials of systemic PD-1/PD-L1 inhibitors in NMIBC patients PMID:28583311.
  • Noninvasive urinary biomarkers feasible. The high prevalence and grade/stage independence of TERT promoter mutations and chromatin-modifying gene alterations supports development of urinary screening/surveillance assays that could replace cystoscopy PMID:28583311.
  • Mutational count alone does not predict BCG response. On the MSK-IMPACT panel, no association was observed between mutational count and BCG recurrence (p = 0.3) — discouraging panel-based TMB as a standalone BCG biomarker without further investigation PMID:28583311.
  • Comparator therapeutic context: Treatment of the cohort included intravesical mitomycin-c (39% of LGTa) and intravesical BCG (13% LGTa, 81% HGT) — patients who received perioperative mitomycin or other adjuvant perioperative therapies were excluded from index sequencing PMID:28583311.

Limitations & open questions

  • Sample size by subgroup is small. While 105 NMIBC tumors is the largest such NGS cohort, subgroup analyses by stage/grade (e.g., 12 HGTis, 23 LGTa) are limited. Longer follow-up will be needed for robust survival inference PMID:28583311.
  • Predictive vs prognostic ambiguity for ARID1A. Whether ARID1A mutations are a true predictive biomarker of BCG response or simply prognostic for poor outcome in NMIBC overall is not resolved by this single-arm BCG cohort and warrants validation in an independent cohort with a non-BCG comparator arm PMID:28583311.
  • Tumor heterogeneity caveat. Both intertumor and intratumor heterogeneity may confound results from the single-region targeted-panel approach used PMID:28583311.
  • Carcinoma-in-situ (Tis) sequencing is technically difficult. Nearly two-thirds of attempted Tis specimens had inadequate tumor purity for analysis, biasing the Tis subset and motivating alternative inputs (cytology, urinary cell-free DNA, urinary exosomes, single-cell sequencing) PMID:28583311.
  • TMB inferred from a 341/410-gene panel. Use of MSK-IMPACT mutational count as a TMB surrogate may lack the resolution of whole-exome sequencing for the BCG-response question; the authors call out the need for a larger cohort and WES PMID:28583311.
  • Pathway conflict around STAG2. This study finds STAG2 truncating mutations enriched in LGTa, which is consistent with some prior reports but not others; a unified interpretation across NMIBC and MIBC remains open PMID:28583311.

Citations from this paper used in the wiki

  • “TERT promoter mutations (73%) and chromatin-modifying gene alterations (69%) were highly prevalent across grade and stage, suggesting these events occur early in tumorigenesis.” (Abstract / Results, p. 4–5)
  • “ERBB2 or FGFR3 alterations were present in 57% of high-grade NMIBC tumors in a mutually exclusive pattern.” (Abstract / Results, p. 5)
  • “DNA damage repair (DDR) gene alterations were seen in 30% (25/82) of high-grade NMIBC tumors, a rate similar to MIBC, and were associated with a higher mutational burden compared with tumors with intact DDR genes (p < 0.001).” (Abstract / Results, p. 6)
  • “ARID1A mutations were associated with an increased risk of recurrence after BCG (hazard ratio = 3.14, 95% confidence interval: 1.51–6.51, p = 0.002).” (Abstract / Results, p. 6, Table 2)
  • “Targeted NGS with a 341 or updated 410 cancer-associated gene panel was performed on formalin-fixed paraffin embedded sections of treatment-naive index tumors along with matched germline DNA for 105 patients with NMIBC…” (Methods, p. 3)
  • “We were unable to identify an association between mutational count on MSK-IMPACT and tumor recurrence after BCG (HR = 0.96, 95% CI: 0.83–1.1, p = 0.3).” (Results, p. 6)
  • “…we have made all genomic and clinical data from this study publically available through the cBioPortal for Cancer Genomics.” (Discussion, p. 8)

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