552 Background: Previously, we reported a 3-factor model including tumor mutation burden (TMB), peripheral blood neutrophil/lymphocyte ratio (NLR) and visceral metastasis status to predict tumor regression (c-statistic=0.90) when using ICIs for mUC regardless of line of therapy. We conducted an external validation of this model to determine its ability to predict response. Methods: We utilized data from a phase II trial, IMvigor210, which evaluated atezolizumab (atezo) for post-platinum pts or as first-line therapy for cisplatin-ineligible pts with mUC. Pts with TMB, NLR, visceral metastasis and RECIST1.1 response status available were evaluable. Tumor genomic analysis was performed using FoundationOne, which sequences 324 cancer-related genes. Multivariable regression studied the association of TMB (<10 vs. ≥10 mutations/megabase), NLR (<5 vs. ≥5) and visceral metastasis status (yes vs.no) with RECIST 1.1 response status (≥partial response [PR] vs no PR). Model discrimination performance was assessed using the concordance (c)-index. Results: A total of 234 pts were evaluable for analysis, of whom 185 (79%) were men. 164 (70.1%) pts received post-platinum therapy and 70 (29.9%) received first-line therapy with atezo. 60 pts exhibited RECIST 1.1 ≥PR. On multivariate analysis, all 3 factors were independently and significantly associated with ≥PR: TMB ≥10 mutations/megabase (HR -4.254, p=2.1 x 10−5), low NLR (HR=2.591, p=0.00958) and absence of visceral metastasis status (HR=4.105, p=4.04 x 10−5). The c-index of this 3-factor model to predict ≥PR was 0.77. Presence of all 3 favorable factors (TMBhigh, NLRlow, no visceral metastases) was associated with ≥PR in 15/21 pts (71.4%), while the absence of all 3 factors (TMBlow, NLRhigh, visceral metastases) was associated with ≥PR in 2/37 pts (5.4%). Conclusions: A 3-factor model combining readily available clinical and genomic variables including TMB, NLR and visceral metastasis status was validated to robustly predict RECIST 1.1 response in pts with mUC receiving atezo as first-line or post-platinum therapy. This model may improve the therapeutic index and cost efficacy of atezo.