Urine DNA methylation assay enables early detection and recurrence monitoring for bladder cancer

Abstract

BACKGROUNDCurrent methods for the detection and surveillance of bladder cancer (BCa) are often invasive and/or possess suboptimal sensitivity and specificity, especially in early-stage, minimal, and residual tumors.METHODSWe developed an efficient method, termed utMeMA, for the detection of urine tumor DNA methylation at multiple genomic regions by MassARRAY. We identified the BCa-specific methylation markers by combined analyses of cohorts from Sun Yat-sen Memorial Hospital (SYSMH), The Cancer Genome Atlas (TCGA), and the Gene Expression Omnibus (GEO) database. The BCa diagnostic model was built in a retrospective cohort (n = 313) and validated in a multicenter, prospective cohort (n = 175). The performance of this diagnostic assay was analyzed and compared with urine cytology and FISH.RESULTSWe first discovered 26 significant methylation markers of BCa in combined analyses. We built and validated a 2-marker-based diagnostic model that discriminated among patients with BCa with high accuracy (86.7%), sensitivity (90.0%), and specificity (83.1%). Furthermore, the utMeMA-based assay achieved a great improvement in sensitivity over urine cytology and FISH, especially in the detection of early-stage (stage Ta and low-grade tumor, 64.5% vs. 11.8%, 15.8%), minimal (81.0% vs. 14.8%, 37.9%), residual (93.3% vs. 27.3%, 64.3%), and recurrent (89.5% vs. 31.4%, 52.8%) tumors. The urine diagnostic score from this assay was better associated with tumor malignancy and burden.CONCLUSIONUrine tumor DNA methylation assessment for early diagnosis, minimal, residual tumor detection and surveillance in BCa is a rapid, high-throughput, noninvasive, and promising approach, which may reduce the burden of cystoscopy and blind second surgery.FUNDINGThis study was supported by the National Key Research and Development Program of China and the National Natural Science Foundation of China.

Keywords: Cancer; Genetics; Molecular diagnosis; Oncology; Urology.

Figure 1. Workflow indicating study design.

SYSMH, Sun Yat-sen Memorial Hospital; TCGA, the Cancer Genome Atlas; BCa, bladder cancer; FDR, false discovery rate; LASSO, the least absolute shrinkage and selection operator; RF, random forest.

Figure 2. Discovery of DNA methylation markers to distinguish BCa and normal tissue.

(A) Unsupervised hierarchical clustering of 26 methylation markers differentially methylated between NAT (n = 21) and BCa tumor tissue (n = 412) in the TCGA cohort. (B) Box plot presenting the β-value distribution of cg21472056 among BCa tumor tissue samples (n = 412), NAT (n = 21), and normal blood WBCs (n = 656). A β-value of zero represents no methylation, whereas 1 represents full methylation. The data are shown as median with the interquartile range. Statistical significance was assessed using 1-way ANOVA followed by Dunnett’s tests. (C) Unsupervised hierarchical clustering of 26 methylation markers differentially methylated among NAT (n = 21), BCa tumor tissue (n = 21), and matched urine (n = 18) in the SYSMH cohort. The unavailable value is shown in gray. (D) Box plot presenting the β-value distribution of cg21472056 among BCa tumor tissue (n = 21), matched urine (n = 18), and NAT (n = 21), which was detected by TOF-MS. The data are shown as median with the interquartile range. Statistical significance was assessed using 1-way ANOVA followed by Dunnett’s tests. (E) The Spearman correlation analysis of cg21472056 methylation level between the tumor tissue and matched urine in 18 patients. Pearson’s χ² test was used to analyze statistical significance. **P < 0.01 and ***P < 0.001.

Figure 3. Construction and validation of a urine diagnostic model to detect BCa in 3 cohorts by using 2 markers.

(A and B) ROC curves and the associated AUCs of the diagnostic prediction model using urine DNA methylation analysis in the training (A) and testing (B) cohorts. (C–E) Unsupervised hierarchical clustering of 2 methylation markers that were differentially methylated between the DNA of BCa and non–cancer subjects in the training (C, n = 222), testing (D, n = 91), and independent prospective validation (E, n = 175) cohorts. Each row represents an individual patient and each column is a CpG marker. The real disease status and prediction status by model are shown ahead. (F) The UD score of healthy participants (n = 12), non–cancer patients (n = 225), and patients with BCa (n = 251) are shown. The dotted line shows the cutoff value (0.3564) to distinguish BCa from non–cancer cases. The data are shown as median with the interquartile range. Statistical significance was assessed using 1-way ANOVA followed by Dunnett’s tests. (G) The sensitivity, specificity, accuracy, PPV, and NPV of this model in the training, testing, and validation cohorts were determined by the cutoff value. ***P < 0.001.

Figure 4. The significantly improved sensitivity of utMeMA in the diagnosis of BCa in comparison with urine cytology and FISH.

(A–C) The UD score of patients with BCa in different grade (A), stage (B), and number (C) of tumors (n = 251). Carcinoma in situ (CIS) means all the cases that include CIS (n = 24). (D) The UD score of patients in 4 types of non–cancer diseases of the urinary system, including benign bladder lesions (BBLs), urolithiasis, benign prostatic hypertrophy (BPH), and other benign diseases (n = 237). The data are shown as median with the interquartile range. Statistical significance was assessed using 1-way ANOVA followed by Dunnett’s tests (A, B, D) and unpaired t test (2-tailed, C). (E) Distribution of predicted diagnostic status using utMeMA across patients with BCa (n = 251) with associated tumor stage, grade, cytology, and FISH results. CIS means the cases which are CIS alone (n = 2). (F–I) The sensitivity of utMeMA in BCa patients with indicated grade (F), stage (G), early-stage (H), and number (I) of tumor, in comparison with urine cytology and FISH. CIS means all the cases that include CIS (n = 24). (J) The specificity of utMeMA in patients with non–cancer diseases in comparison with urine cytology and FISH. (K) The specificity of utMeMA in patients with 4 types of non–cancer diseases. Statistical significance was assessed by χ² test (G–L). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5. Application of utMeMA to detect minimal BCa tumor.

(A and B) The UD score and sensitivity of utMeMA in patients with different BCa tumor sizes, in comparison with urine cytology and FISH. Statistical significance was assessed using 1-way ANOVA followed by Dunnett’s tests (A) and χ² test (B). (C and D) The UD score and sensitivity of utMeMA in patients with BCa with single or multiple small tumors, in comparison with urine cytology and FISH. Statistical significance was assessed using unpaired t test (2-tailed, C) and χ² test (D). The data are shown as median with the interquartile range (A and C). (E) Example of a patient with minimal tumor detected by utMeMA, but missed by cytology, FISH, MR imaging, and ordinary cystoscopy, who was later diagnosed by fluorescence cystoscopy. The pathology of the tumor was Ta and low grade. *P < 0.05, **P < 0.01. The magnifications of the images in fluorescence cystoscopy, urine cytology, FISH and pathology were ×10, ×600, ×3000, ×400, respectively.

Figure 6. Application of utMeMA to detect residual tumor and monitor the recurrence of BCa.

(A) The distribution of UD score in BCa patients with or without residual tumors (n = 47). Statistical significance was assessed using unpaired t test (2-tailed). The data are shown as median with the interquartile range. (B) The landscape of pathological characters and detection results in re-TURBT cohort, including 15 cases with residual tumor and 32 cases without tumor. (C) The sensitivity and specificity of utMeMA in the detection of residual tumor, in comparison with urine cytology and FISH (n = 47). (D) The distribution of UD score in patients with BCa with or without recurrent tumor. The data are shown as median with the interquartile range. Statistical significance was assessed using unpaired t test (2-tailed). (E) The landscape of pathological characteristics and detection results in surveillance cohort, including 38 cases with tumor recurrence and 43 cases without recurrence (n = 81). (F) The sensitivity and specificity of utMeMA in detection of recurrent tumor, in comparison with urine cytology and FISH (n = 81). (G–I) The sensitivity of utMeMA in patients with recurrent BCa with indicated grade (G), stage (H), and size (I) of tumor, in comparison with urine cytology and FISH (n = 38). Statistical significance was assessed using χ² test (C, F, G–I). *P < 0.05, **P < 0.01.