Myeloid-Adaptive Response to Atezolizumab (MARTA): A Two-Gene CXCL13–CXCL8 Transcriptomic Biomarker Across Renal Cell and Urothelial Carcinoma

Transcriptomic response biomarkers for immune-checkpoint blockade remain poorly defined across cancers. We developed the two-gene Myeloid-Adaptive Response to Atezolizumab (MARTA) score, integrating CXCL13 and CXCL8 expression. MARTA was developed in 81 atezolizumab-treated patients in IMmotion150, evaluated across treatment groups, and independently assessed in IMvigor210. In IMmotion150, higher MARTA was associated with higher objective response rate (ORR) with atezolizumab alone (odds ratio [OR], 3.20; 95% confidence interval [CI], 1.42–7.19) and atezolizumab plus bevacizumab (OR, 2.61; 95% CI, 1.43–4.76), but not with sunitinib. Higher MARTA was associated with longer progression-free survival (PFS) in both atezolizumab-containing groups (hazard ratio [HR], 0.60 and 0.65), but not with sunitinib (interaction p = 0.01). In IMvigor210, higher MARTA was associated with higher ORR (OR per 1-standard-deviation increase, 1.38; 95% CI, 1.05–1.81) and longer overall survival (OS) (HR, 0.76; 95% CI, 0.67–0.86). The OS association remained after multivariate adjustment (HR, 0.81; 95% CI, 0.69–0.95). Discrimination for ORR and OS, assessed by the area under the curve (AUC) and C-index, respectively, was modest for MARTA, tGE8, and the IFN-γ signature. These findings support further evaluation of MARTA, while prospective validation and cross-platform analytical reproducibility remain necessary before clinical application.