Table 7. Results of regression analysis for predictors of VF cycle length and conduction block (DPI) incidence.

Stepwise Multiple Regression (for predicting Cycle Length and incidence of Conduction Block)
	Model R2	Regression Coefficient	p-Value
Predictor Variables* (For Cycle Length)
Cx43	0.57	0.415	<0.001
Na+/K+ ATPase β1	0.79	−0.428	0.090
Kir2.1	0.86	8.82	0.044
hERG	0.95	−7.95	0.004
Predictor Variables** (For Conduction Block)
Cx45	0.41	−0.59	0.001
Kir3.1	0.63	2.24	0.068
Cx43	0.72	−0.03	0.006
SUR2	0.79	0.0197	0.014
Kir2.3	0.88	0.19	0.049

Stepwise multiple regression analysis was performed to evaluate the contribution of ion channels to Cycle length or incidence of conduction block. Expression level of each ion channel was added to the regression model and those with significant contribution to dependent variables were retained in the model. Data presented in the table demonstrates the most appropriate fit, capable of predicting the incidence of CL and conduction block.

Among all ion channels assessed, expression of Cx43, Na/K ATPase β1, Kir2.1 and hERG were the most significant predictors of Cycle length.

Among all ion channels assessed, expression of Cx45, Kir3.1, Cx43, SUR2 and Kir2.3 significantly correlated with incidence of conduction block.