e16212 Background: No established risk factors or toxicity biomarkers exist for HCCs treated with ICI, even though they are commonly used in clinical practice. The outcomes in real-world practice are worse than those reported in clinical trials. This retrospective review aims to identify HCC patients at high risk of failing ICI therapy. Methods: HCC patients who received at least one ICI dose between 1/1/2017 and 5/30/2023 (before tremelimumab/durvalumab approval) at the Ohio State University were included in this study. The patient's baseline (BL) characteristics (at the first dose of ICI) were extracted with immune-related adverse event (irAE) details and survival outcomes. We examined BL characteristics significantly different in poor (progression-free survival (PFS) ≤ median (m)) and worst (PFS ≤ 25% percentile) responders, and poor (overall survival (OS) ≤ m) and worst (OS ≤ 25% percentile) survivors. Chi-square test and logistic regression were performed using JMP Pro 17 (SAS Institute Inc., Cary, NC). Results: Our cohort included 53 HCC patients. The median age was 66 years (range 40-86), with 81% male and 85% White (13% African American and 2% other). The median OS and PFS were 19 months (m) (95% confidence interval [CI], 8-26) and 5m (95% CI, 2-9), respectively. The poor responders (PFS ≤5m, N = 30) had patients with significantly higher alcohol history (AHx, 90% vs. 70%) and mean alpha-fetoprotein (AFP, 44, 155 vs. 811 ng/mL) levels, a higher proportion of Child-Pugh (CP) C (A/B/C %, 53/27/20 vs. 53/47/0), and negligible bevacizumab (bev, 77 vs. 24) use, all (p < 0.05). The worst response group (PFS ≤2m, N = 16), furthermore, had significantly (p < 0.05) higher Albumin-bilirubin (ALBI) grade (G) 3 patients (50% vs. 16%), higher mean total bilirubin (2 vs. 1,2 ng/mL), lower tyrosine-kinase inhibitor (TKI, 7% vs. 46%), and lower mean albumin (Ab, 3 vs. 3.4) levels. A significantly (p < 0.05) higher fraction (fx) of poor survivors (OS ≤19m, N = 33) got ICI In the first or second line (85% vs. 45%) and did not receive any locoregional therapy (LRT, 66% vs. 35%). The worst survivors (OS ≤8m, N = 14) got ICI in the first line (100% vs. 59%, p = 0.04, and significantly (p < 0.05) lower Alb (3 vs. 3,4) and higher AFP (90620 vs. 3104) levels. The irAE group (N = 10, 19%) had higher AFP (115289 vs. 3482, p = 0.01) and a lower distant metastasis rate (20% vs. 58%, p = 0.02), and none of them received bev (0% vs. 30, p = 0.04). Conclusions: Response to ICI depends on BL liver function (AHx, ALBI, CP, Alb) and combination therapy (bev). Survival depends on HCC management up to the administration of ICI (LRT and TKI).