Multimodal Urine Cell-Free DNA-Based Residual Disease Detection and Prognostication in Muscle-Invasive Bladder Cancer Patients

Abstract

<h3>Purpose/Objective(s)</h3> Our group recently showcased a liquid biopsy assay (uCAPP-Seq) based on the detection of single nucleotide variants (SNVs) in urine cell-free DNA (cfDNA) from patients with muscle-invasive bladder cancer (MIBC). We found that assay positivity corresponded with the presence of residual cancer and a higher risk of disease recurrence despite curative-intent therapy (Chauhan et al., PLOS Medicine, 2021). Here, we explored multimodal urine-based genomics by integrating ultra-low-pass whole genome sequencing (ULP-WGS) and uCAPP-Seq for improving the sensitivity of residual disease detection and survival prediction among MIBC patients. <h3>Materials/Methods</h3> Urine samples from 74 MIBC patients were collected preoperatively on the day of radical cystectomy to assess urine tumor DNA (utDNA). We performed ULP-WGS of urine cfDNA from all 74 patients as well as 15 healthy adults. Tumor fraction (TF) based on genome-wide copy number alterations in urine cfDNA was estimated using ichorCNA. Variant allele frequency (VAF) based on SNVs was estimated by uCAPP-Seq. Pathologic complete response (pCR) was determined by surgical pathology. Multivariate logistic regression and random forest models were utilized to predict disease status based on VAF and TF levels, along with clinical variables (smoking, age, sex, race, and histology). Kaplan-Meier (KM) analysis was used to assess overall survival (OS) and progression-free survival (PFS). <h3>Results</h3> Mean utDNA VAF and TF levels were both significantly higher in patients without pCR compared to those who achieved pCR (11.9% vs. 4.0% VAF, p=0.044; 12.7% vs. 5.2% TF, p = 0.019) (Table 1). Our multimodal logistic regression model identified utDNA VAF (p < 0.001) and smoking status (p = 0.034) as significant predictors of disease status. Random forest modeling demonstrated that utDNA VAF and TF were both significant predictors. It achieved a sensitivity of 83% and a specificity of 79% for detecting residual disease. Strikingly, among the 15 patients with negative findings by both ULP-WGS and uCAPP-Seq, none experienced cancer progression (median follow-up time = 7.2 months). In contrast, the 59 patients with positive utDNA by ULP-WGS or uCAPP-Seq had significantly worse PFS (median = 6.0 months; HR = 3.54, p = 0.027). Moreover, utDNA positivity trended towards predicting significantly worse OS (HR = 3.43, p = 0.15). <h3>Conclusion</h3> Integration of ULP-WGS with uCAPP-Seq enabled robust detection of residual urine tumor DNA with high sensitivity and predicted progression-free survival in muscle-invasive bladder cancer patients. In the future, this type of multimodal urine-based genomic analysis may lead to more precise risk stratification and nonoperative clinical decision-making for muscle-invasive bladder cancer patients.