Development of a Screening Nomogram for Regional Lymph Node Metastasis Development in Muscle Invasive Bladder Cancer after Multiagent Systemic Therapy

Abstract

A reliable method of identifying clinically node negative bladder cancer patients at greater risk of developing lymph node metastasis and may require intensified therapy is needed. The purpose of this study was to create a nomogram to quantify the risk for regional lymphatic involvement in non-metastatic muscle invasive bladder cancer. Usingthe National Cancer Database, patients with clinical T2-4N0M0 urothelial carcinoma of the bladder between the years of 2004 - 2020 were selected. All patients completed multiagent chemotherapy followed by surgery for pathologic nodal staging to determine presence or absence of nodal disease. No neoadjuvant radiation therapy was allowed and patients with history of prior malignancies were excluded. Following a 70:30 training to testing data split, baseline variables were assessed using univariate logistic regression. Variables were selected for inclusion in the multivariate logistic regression model using a combination of previously reported findings in the literature and/or if found to have a p-value < 0.05 on univariate analysis. A nomogram was constructed using this final model with assessment using calibration plots and the area under the receiver operating characteristics curve (AUC), respectively. An empiric cut point was chosen at 95% sensitivity to identify patients at "high" and "low" risk for pathologic nodal disease with overall assessment in both cohorts using chi-square. A total of 6194patients were identified for study with a median age of 65 years (IQR = 58 - 71 years). Most patients were male (68.0%) with T2 disease (81.2%). The final multivariate model included age at time of diagnosis (OR = 0.99; 95% CI = 0.99 - 1.00; p = 0.172), time from diagnosis to initiation of chemotherapy (OR = 1.00; 95% CI = 1.00 - 1.01; p = 0.005), papillary histology (OR = 0.85; 95% CI = 0.72 - 1.01; p = 0.068), and clinical T stage (Table 1). Model calibration demonstrated excellent goodness-of-fit with good discrimination (AUC = 0.644). Within the training data, high risk patients were seen to have a twofold increase in pathologic nodal disease (N = 835/3924, 21.2%) when compared to those identified as low risk (N = 38/347, 9.9%) (p < 0.001). Validation within the testing data set demonstrated similar results with pathologic node rates of 22.8% and 7.8% for high- and low-risk patients, respectively (AUC = 0.645, p < 0.001). This study demonstrates a clinically applicable risk stratification tool for identifying patients at risk for developing lymphadenopathy in T2-4 bladder cancer and may help guide future research in selecting patients eligible for escalation of therapy. Future studies should aim to externally validate this tool within prospective cohorts, and seek to determine if this nomogram may provide further prognostic utility.