High-dimension flow cytometry reveals comprehensive deviations in immunophenotypes associated with non-response to melanoma immunotherapies

Abstract Immunotherapies have revolutionized the treatment of many cancers, but most patients fail to respond. To investigate immune phenotypes associated with patient response, we assessed 68 unique peripheral blood markers using high-dimension flow cytometry. Samples from 41 metastatic melanoma patients treated with αPD1 or αCTLA4 and 12 healthy donors (HD) were evaluated. To avoid a loss of higher order data, we used a novel computational approach, CytoBrute, to analyze the data through semi-comprehensive Boolean gating. Lineages (e.g. CD3+CD4+) were evaluated for all possible combinations for up to 15 markers. In αPD1 treated patients 1,051 immune signatures differentiated patient response at baseline with a p-value <0.05, while 823 differentiated αCTLA4 treated patients. Only eight signatures overlapped between the two therapies. Signatures included both low (e.g. CD4+CD45RO+CD278−) and high-dimension immunophenotypes (e.g. CD14+CD33+CD11C+CD163+PDL2+41BBL+CD86+CD40+PDL1−CD66B−CD80−GAL9−CD15−CD19−OX40L−). A total of 2,895 signatures significantly differentiated HD versus patients, of which 270 significantly differed (p<0.05) based on patient response. Samples from non-responding patients less closely resembled HD relative to responding patient in 205 of these 270 signatures (p<1e-8). Comparisons of all 102,000 assessed phenotypes showed that non-responding patients had a lesser degree of similarity in immunophenotypes in comparison to responders. Our data demonstrate a novel and powerful approach to interrogating complex immunophenotypes and show that comprehensive immunophenotypic deviations are associated with non-response to melanoma immunotherapies.