External validation of a biomarker based pre-cystectomy algorithm to predict nonorgan confined urothelial cancers.

Abstract

The role of neoadjuvant chemotherapy before surgery in patients with muscle invasive bladder cancer remains debated and the need for tools to identify patients who would benefit from chemotherapy is pertinent. We previously published a preoperative algorithm to predict nonorgan confined disease. This algorithm included tumor markers (CEA, CA 125 and CA 19-9) as well as clinical parameters. In this study we validated the accuracy of this algorithm in an independent, external cohort. We used the Toronto Biobank to measure preoperative serum levels of CEA, CA 125 and CA 19-9 in 76 consecutive patients with clinically organ confined bladder cancer (cT2 or less) who underwent radical cystectomy. Clinical parameters were retrieved from our prospective bladder information system database and incorporated into our marker based algorithm. A numerical score was generated for each patient and a previously published cutoff was used to predict the presence of nonorgan confined disease. The accuracy of the model was quantified with the area under the curve, and the positive and negative predictive values were calculated. On pathological evaluation 38 patients (50%) had nonorgan confined tumors. The AUC of the algorithm was 0.79 (95% CI 0.69-0.89). The positive and negative predictive values were 79% (95% CI 71-87) and 74% (95% CI 66-82), respectively. We externally validated a pre-cystectomy model to predict pathological stage. The algorithm may possibly aid in selecting patients who would benefit from neoadjuvant chemotherapy before cystectomy.