Electronic nose-based urinary volatilome profiling accurately discriminates bladder cancer from healthy controls: A prospective diagnostic study.

BACKGROUND

Bladder cancer (BCa) lacks reliable non-invasive diagnostic tools. Urinary volatile organic compounds (VOCs) have emerged as potential discriminative biomarkers. We evaluated the diagnostic performance of an electronic nose (eNose) platform for BCa detection.

METHODS

In this prospective single-center study, 192 patients with histologically confirmed BCa and 143 healthy controls were included. Consecutive patients with suspected primary or recurrent BCa were prospectively screened, and first-morning urine samples were collected in a sterile fashion before Trans-urethral resection of bladder tumor (TURBT) or radical cystectomy. Healthy controls were recruited at the institutional outpatient clinic after exclusion of neoplasia and urinary infection. Patients with ongoing gross hematuria or concomitant urinary infection were not enrolled. Urine samples were analyzed using a 32-sensor eNose (Cyranose C320). Principal component analysis (PCA) and canonical discriminant analysis (CDA) were applied. Cross-validated accuracy (CVA) was calculated using the leave-one-out method. Diagnostic performance was assessed using receiver operating characteristic (ROC) analysis.

RESULTS

The eNose achieved a cross-validated accuracy of 87.5% (P < 0.001). Sensitivity and specificity were 81.8% and 95.1%, respectively. The area under the ROC curve (AUC) was 0.926. Reproducibility was confirmed in serial urine samples obtained from the same patients (CVA 87.2%, AUC 0.90). Positive and negative likelihood ratios were 16.69 and 0.19, respectively.

CONCLUSIONS

Urinary volatilome profiling using an eNose demonstrates high diagnostic accuracy in discriminating BCa patients from healthy individuals. These findings support eNose-based VOC analysis as a promising non-invasive adjunct tool for BCa detection and clinical triage in selected risk-enriched settings, warranting external validation in multicenter studies.