3576 Background: Studies of immune checkpoint blockade therapy (ICT) outcomes have been largely performed in melanoma and lung cancer patients, both of which are enriched for White patients. For example, a National Cancer Database study found that 97% of first-line ICT treatments in melanoma have been administered to White recipients (Patel, ASCO-SITC 2020). Given expanding indication in tumor types affecting more diverse populations, we sought to study whether minority populations might be projected to have differing checkpoint blockade response rates. Methods: Ethnicity information and RNAseq expression profiles and primary site information were obtained for 7087 patients from TCGA. Ethnicity was tested for association with RNA expression of targetable checkpoint genes ( PD1, PDL1, PDL2, CTLA4, IDO1, LAG3, TIM3, TIGIT, OX40, VISTA, and GITR) in 5 tumor histology types by Wilcoxon methods with Benjamini-Hochberg correction for multiple hypothesis testing. A dataset of > 2700 cases was obtained from NantHealth, with paired whole exome/RNAseq data. Ethnicity for 579 patients was assigned using allele-fraction from ~250 single nucleotide polymorphic sites found exclusively in 6 populations within the 1000 Genomes project. Ethnicity/checkpoint associations found in TCGA were tested within this dataset. Results: Within the TCGA cohort, ethnicity was not a factor in differential expression of checkpoint molecules in lung cancer. Within melanomas, in Asian patients PDL1, CTLA4, and IDO1 were expressed at lower levels than in White patients (each p = 0.04). These associations did not remain significant after correction for multiple hypothesis testing. Breast cancers in Black patients had significantly higher PD1, CTLA4, IDO1, LAG3, GITR, and OX40 expression compared to White patients, all remaining significant after correction (adj. p 3.7e-5 to 6.4e-3). Among White patients, colon cancers showed higher expression of PDL1/2, IDO1, LAG3, TIM3, and GITR (p 0.04 to 0.0017). IDO1 was significantly higher in White patients even after correction (adj. p = 0.03), and lower in Black patients (adj. p = 0.03). Conclusions: Ethnicity may represent a significant factor for efficacy checkpoint blockade therapies. White breast cancer patients might be anticipated to exhibit reduced sensitivity to PD1/CTLA4 blockade, while Black colon cancer patients may exhibit reduced sensitivity to IDO1 therapies such as epacadostat. A biomarker-driven approach to patient selection may ameliorate ethnic disparities in ICT outcomes.