M. L. Zierhut, D. J. Nickens, W. Tan; Pfizer, San Diego, CA

BACKGROUND: Data from anaplastic lymphoma kinase (ALK)-positive non-small cell lung cancer (NSCLC) patients enrolled in 2 Phase II/III clinical trials were used to examine the effect of crizotinib (CRZ) exposure on heart rate corrected QT interval (QTc).
METHODS: Patients orally received a recommended starting dose of 250 mg BID CRZ. In total, 405 paired CRZ concentrations (C) and electrocardiograms (ECG) were acquired at pre-specified times on days 1 and 22 from 62 patients. QT was corrected using Bazett’s (B), Fridericia’s (F) and a study specific (S) factor, βx. Using pre-dose ECG data, βS was estimated by: ln(QT) = ln(α) + βS∙ln(RR) + ε; where ε~ N(0,σ), with α and σ also estimated. The relationship between C and RR or QTc was assessed using the following equation: Y = θ1 + η1 + (θ2 + η2)∙C + ε; where Y represents either RR or QTc, θ1 and θ2 are the intercept and slope, respectively, and ηx ~ N(0,ωx).
RESULTS: A significant C-RR relationship was found, with a model predicted (90% confidence interval [CI]) change of 174 ms (142, 206) at the median observed maximum C (Cmax) of 318 ng/mL. The slopes of QTc versus RR for the three factors were: B=-0.057, F=0.031, S=-0.002 ms/ms, presenting a bias in the estimation of drug induced changes in QTc. At Cmax, the estimated biases based on the slopes from B, F and S were -9.9, +5.4 and -0.35 ms, respectively, indicating that QTcS best corrected for the effect of RR on QT. The analysis did not reveal a statistically significant C-QTcS relationship, and the model predicted (90% CI) change in QTcS at Cmax was 2.99 ms (-0.118, 6.09).
CONCLUSION: Higher C was associated with higher RR. As a result, βB and βF led to more notable biases in the estimation of drug induced changes in QTc than βS did. Based on C-QTcS modeling, no clinically relevant QT prolongation is expected at a CRZ dose level of 250 mg BID.