Determination of the falloff constant (k(f)) from modeling biochemical marker release: a new variable for discriminating therapies

A new variable termed the falloff constant (k(f)) was derived from the curve fitting of serial CK-MB measurements. k(f) represents the rate constant of maximal decline in serum CK-MB and is determined from the slope of the lognormal curve at the inflection point. Physiologically, kf's magnitude reflects the balance between CK-MB's rate of release from tissue and the rate of elimination. We examined k(f) in two myocardial infarction (MI) patient sets. The first set was homogeneous and taken from the TAMI 7 study (n = 147) and included 111 patients having TIMI 2-3 flow after thrombolytic therapy and 36 patients who initially had TIMI 0-1 flow. The TIMI 0-1 patients were opened to TIMI 3 by angioplasty within 3 h. The second set consisted of 196 patients enrolled in the IMPACT-AMI study that demonstrated the efficacy of the glycoprotein (GP) IIb/IIIa antagonist eptifibatide. This second set consisted of 93 patients in the GP IIb/IIIa treatment group and 103 in the placebo group. Log-normal curve-fitting parameters including peak maximum and curve area were also compared to k(f) in the GP IIb/IIIa versus placebo set. The Wilcoxon test showed no difference between the two groups of TAMI 7 patients (p = 0.22). However, there was a highly significant difference in kf between the GP IIb/IIIa treatment group versus the placebo group (p = 0.0014). Both k(f) and peak maximum from curve fitting showed significant differences between the GP IIb/IIIa treatment group and the placebo group; however, k(f) showed a substantially lower p-value (p = 0.0014 and p = 0.023, respectively). As expected, k(f) showed no difference between the TAMI 7 groups because this was a homogeneous patient set in that they all had TIMI 3 patency status within 3 h of treatment. However, in the patient set having very different treatments, GP IIb/IIIa versus placebo, there was a highly significant difference in the kf variable. These data suggest that differences in reperfusion are reflected by kf and that this variable may represent a valuable new nonmortality end point derived from curve fitting analysis.