New 6-factor prognostic model for patients (pts) with advanced urothelial carcinoma (UC) receiving post-platinum atezolizumab.

Abstract

413 Background: Prognostic factors for overall survival (OS) have been identified in pts receiving post-platinum chemotherapy for advanced UC, but it is unknown whether these factors and/or others optimally predict OS for pts treated with PD1/PD-L1 inhibitor therapy. Methods: Pt level data from two UC salvage trials evaluating atezolizumab were used: IMvigor210 (n = 310) for training and PCD4989g (n = 95) for validation. Univariable and multivariable Cox regression analyses were performed to evaluate the association of the prognostic factors recognized in the chemotherapy setting (ECOG performance status [ECOG-PS], liver metastasis (LM), anemia, treatment-free interval, albumin), neutrophil-lymphocyte ratio (NLR), eosinophil count, platelet count (PLT), site of primary/metastases, stage at diagnosis, smoking, LDH, prior therapies and immune cell PD-L1 status by IHC with OS. Clinical factors were dichotomous and lab values normalized by logarithmic transformation as needed. Stepwise selection was employed to propose an optimal model using the training dataset; pts were then categorized by number of risk factors. Concordance, discrimination (c-statistic) and calibration were assessed in the validation dataset using bootstrap analyses. Results: The factors included in the optimal prognostic model for OS were: ECOG-PS 1 vs. 0 (HR 1.64 [95% CI: 1.20, 2.24], p = 0.002), LM (1.45 [1.08, 1.94], p = 0.014), PLT (1.73 [1.14, 2.61], p = 0.010), NLR (1.84 [1.45, 2.34], p < 0.001), LDH (1.54 [1.19, 1.99], p = < 0.001) and anemia (HR = 1.60 [1.17, 2.21] p = 0.004). The c-statistic was 0.690 (95% CI = 0.649-0.715) and 0.759 (0.694-0.795) in the training and validation datasets, respectively. 1-year OS of pts in the training and the validation cohorts were similar. PD-L1 score was statistically significant when adjusted for the optimal model, but did not improve clinical interpretability (c-statistic = 0.698). Conclusions: A new validated 6-factor prognostic model for OS including ECOG-PS, LM, PLT, NLR, LDH and anemia is proposed in the setting of post-platinum atezolizumab for advanced UC. Applicability of the model to other PD1/PD-L1 inhibitors and PD-L1 IHC assays warrant investigation.