Regression tree analysis to identify the best candidates for neoadjuvant chemotherapy (NAC) in patients with clinically lymph node-positive penile squamous cell carcinoma (PSCC).

Abstract

2 Background: PSCC patients (pts) with palpable inguinal lymph node (ILN) disease have a poor overall survival (OS). For them, use of neoadjuvant chemotherapy (NAC) is recommended in guidelines, but limited data are available to inform pt selection. We aim to identify predictors of poor OS in clinically ILN-positive (ILN+) pts to define the optimal candidates for NAC. Methods: Within an international, multicenter database of 924 PSCC pts, we identified 334 ILN+ pts with available data. ILN involvement was defined either with the presence of palpable ILN disease or based on preoperative CT-scan. FDG-PET/CT-scan was performed based on clinical judgment of the treating physician. Univariable and multivariable analyses (MVA) assessed predictors of overall mortality (OM). A regression-tree method for censored data was used to generate a risk stratification tool for prediction of 24m OM after diagnosis of ILN+ PSCC. Kaplan-Meier method was used to define the OS benefit related to the use of NAC according to the regression-tree stratified sub-groups. Results: Median age at diagnosis was 58yrs, and 105 (31.4%) had ECOG≥1. Of all, 120 (35.9%), 152 (45.5%), and 62 (18.6%) harbored cN1, cN2, and cN3 disease. Additionally, 152 (45.5%) and 117 (35%) had monolateral vs bilateral clinical ILN involvement. FDG-PET/CT was adopted in 42 (12.2%) pts, and 16 (4.8%) had pelvic LN uptake. Median OS was 107m, with 24m OS = 66%. At MVA, cN2 (HR: 2.28, p=0.006), cN3 (HR: 2.15, p=0.02), PET/CT scan-detected pelvic and ILN involvement (HR: 2.57, p=0.007) were independently associated with higher OM, whether bilateral clinical ILN+ was only univariably significant (HR: 1.56, p=0.02). At regression-tree analysis (AUC 70%), pts with cN3 and cN2 with PET/CT-detected pelvic and ILN involvement had the higher risk of 24m OM (>54%). NAC was associated with improved 24m OS rates (54 vs. 33%) only in this pt subgroup (p=0.002). Conclusions: NAC improves OS in pts with cN3 or cN2 and pelvic FDG-PET/CT scan-detected disease. Our regression-tree may serve to screen ILN+ pts, to identify the optimal NAC candidates before radical treatments.