RESUMEN
BACKGROUND: Numerous phase I trials testing immune checkpoint inhibitors (CPI)-based combinations are currently being conducted to improve response rates observed with single agents. However, methodology varies across studies, especially regarding the use of dose escalation. MATERIALS AND METHODS: A literature search was conducted in Pubmed and major oncology meetings libraries for phase I trials reported between 2011 and 2018, containing at least one CPI [CLTA-4 blocking antibody or a PD(L)1 blocking antibody] plus at least one second agent (e.g. tyrosine kinase inhibitor, chemotherapy). Dose escalation schemes, target doses and recommended phase II doses (RP2D) were captured in our database for each study. Combination RP2D (combo-RP2D) was compared with target dose. RESULTS: We identified 113 different studies comprising a total of 120 individual cohorts. The backbone was an anti- cytotoxic T-lymphocyte antigen 4 (CTLA-4) in 40 cohorts and an anti-PD(L)1 in 80 cohorts. Dose escalation was used for the CPI in 29 (24%) cohorts [11% for anti-PD(L)1 and 50% for anti-CTLA-4] and for the second agent in 55 cohorts (46%). For 31 s agents (26%), the combo-RP2D was significantly lower than the expected target dose. Failure to reach the target dose was explained by the type of second agent form (e.g. small molecules versus monoclonal antibodies) (P < 0.001) and the choice of trial design for the second agent by investigators. CONCLUSION: Design of studies investigating new CPI-based combinations must consider the type of second agent. Dose escalation is required for combinations with small molecules but is unnecessary with vaccine/virus/dendritic therapies and monoclonal antibodies.