Molecular residual disease detection in resected, muscle-invasive urothelial cancer with a tissue-based comprehensive genomic profiling-informed personalized monitoring assay

Abstract

Introduction: Circulating tumor DNA (ctDNA) detection postoperatively may identify patients with urothelial cancer at a high risk of relapse. Pragmatic tools building off clinical tumor next-generation sequencing (NGS) platforms could have the potential to increase assay accessibility.

Methods: We evaluated the widely available Foundation Medicine comprehensive genomic profiling (CGP) platform as a source of variants for tracking of ctDNA when analyzing residual samples from IMvigor010 (ClinicalTrials.gov identifier NCT02450331), a randomized adjuvant study comparing atezolizumab with observation after bladder cancer surgery. Current methods often involve germline sampling, which is not always feasible or practical. Rather than performing white blood cell sequencing to filter germline and clonal hematopoiesis (CH) variants, we applied a bioinformatic approach to select tumor (non-germline/CH) variants for molecular residual disease detection. Tissue-informed personalized multiplex polymerase chain reaction-NGS assay was used to detect ctDNA postsurgically (Natera).

Results: Across 396 analyzed patients, prevalence of potentially actionable alterations was comparable with the expected prevalence in advanced disease (13% FGFR2/3, 20% PIK3CA, 13% ERBB2, and 37% with elevated tumor mutational burden ≥10 mutations/megabase). In the observation arm, 66 of the 184 (36%) ctDNA-positive patients had shorter disease-free survival [DFS; hazard ratio (HR) = 5.77; 95% confidence interval (CI), 3.84-8.67; P < 0.0001] and overall survival (OS; HR = 5.81; 95% CI, 3.41-9.91; P < 0.0001) compared with ctDNA-negative patients. ctDNA-positive patients had improved DFS and OS with atezolizumab compared with those in observation (DFS HR = 0.56; 95% CI, 0.38-0.83; P = 0.003; OS HR = 0.66; 95% CI, 0.42-1.05). Clinical sensitivity and specificity for detection of postsurgical recurrence were 58% (60/103) and 93% (75/81), respectively.

Conclusion: We present a personalized ctDNA monitoring assay utilizing tissue-based FoundationOne^® CDx CGP, which is a pragmatic and potentially clinically scalable method that can detect low levels of residual ctDNA in patients with resected, muscle-invasive bladder cancer without germline sampling.

Keywords: CtDNA; MRD; bladder cancer; comprehensive genomic profiling; immunotherapy; monitoring assay; next-generation sequencing.

Figure 1

FoundationOne^® Tracker, a tissue-informed personalized ctDNA monitoring assay. CGP, comprehensive genomic profiling; ctDNA, circulating tumor DNA; FMI, Foundation Medicine, Inc.; mPCR, multiplex polymerase chain reaction; SNV, single-nucleotide variant.

Figure 2

CONSORT diagram. BEP, biomarker-evaluable population; cfDNA, cell-free DNA; CGP, comprehensive genomic profiling; ctDNA, circulating tumor DNA; QC, quality control.

Figure 3

Comprehensive genomic profiling at baseline identifies potentially actionable alterations in patients with bladder cancer. (A, B) Top 40 most commonly altered genes in IMvigor010, including potentially actionable alterations and high TMB (A), as well as other genomic alterations (B). (C) X–Y plot comparing prevalence of genomic alterations in MIBC from IMvigor010 versus advanced-stage bladder cancer from the FoundationCORE^™ database. MIBC, muscle-invasive bladder cancer; TMB, tumor mutational burden.

Figure 4

Characterization of monitorable variant types, as well as their distribution in patient cohorts. (A) Distribution and number of monitorable variants in patients in the observation and atezolizumab arms of IMvigor010 (pathogenic and VUS coding are coding cancer-associated alterations; non-coding and synonymous are intronic or synonymous alterations). (B) Percentage of patients with disease progression in ctDNA populations by arm in the IMvigor010 study. (C) VAF distribution of each monitored alteration in ctDNA-positive samples by arm. ctDNA, circulating tumor DNA; ND, not detected; VAF, variant allele frequency.

Figure 5

MRD detection is prognostic for DFS and OS and predictive of atezolizumab response in patients with MIBC. Kaplan–Meier estimates of DFS (A) and OS (B) for patients monitored and stratified by ctDNA detection (MRD) at the postsurgical time point (univariable analysis). Kaplan–Meier estimates of DFS for ctDNA-positive patients monitored and stratified by TMB (C). High TMB was defined as ≥10 mutations/Mb. Low TMB was defined as <10 mutations/Mb. CI, confidence interval; ctDNA, circulating tumor DNA; DFS, disease-free survival; Mb, megabase; MIBC, muscle-invasive bladder cancer; MRD, molecular residual disease; OS, overall survival; TMB, tumor mutational burden.

Figure 6

Multivariable exploratory analyses^a of clinical factors associated with (A) DFS and (B) OS in the observation arm. (A) DFS where number of events were 103, global P-value (log rank), 2.46e^–16; AIC, 920.88; concordance index, 0.76. (B) OS where number of events were 62; global P-value (log rank), 3.60e⁻¹¹; AIC, 564.55; concordance index, 0.75. ^aModel includes variables with P < 0.05 in univariable models. AIC, Akaike information criterion; CI, confidence interval; ctDNA, circulating tumor DNA; DFS, disease-free survival; IC, immune cell; PD-L1, programmed cell death-ligand 1; pT, primary tumor; OS, overall survival; TMB, tumor mutational burden.