A comparison of different regression models for the quantitative analysis of the combined effect of P2Y12 and P2Y1 receptor antagonists on ADP-induced platelet activation

The combination index (CI), a common quantitative indicator of the degree of synergy/antagonism, may be determined using different regression methods. However, any analysis with constraints has the potential for underestimating the combined effect of multiple drugs.This in vitro study describes the combined effects of selected platelet antagonists on ADP-induced platelet activation in different regression models.The inhibitory effects of P2Y12 receptor antagonists in combination with P2Y1 receptor antagonists (i.e. cangrelor with MRS 2279, prasugrel metabolite with MRS 2179 and PSB 0739 with MRS 2179) were characterized with the aid of three software packages: CompuSyn (for linear regression with constraints), CISNE (for non-linear regression with constraints) and GraphPad Prism (for non-linear regression without constraints). The synergism between P2Y12 and P2Y1 inhibitors was quantified by CI and synergy area.MRS 2279 and MRS 2179 were found to act synergistically with selected P2Y12 receptor antagonists to potentiate their antiplatelet effect. The models of regression with constraints, linear regression in particular, demonstrated a worse fit for experimental data than non-linear regression without constraints; this resulted in an incorrect estimation of the combined effects of two antiplatelet drugs, i.e., underestimating the CI and overestimating the synergy area. Also, the synergy area was found to better reflect the differences among models than the CI.These findings suggest that non-linear regression without constraints offers more precise quantitative determination of combined effects between two drugs compared to the regression models with constraints.