Bias and variance evaluation of QT interval correction methods

QTc interval prolongation and torsade de pointes associated with second-generation antipsychotics and antidepressants: a comprehensive review

We comprehensively reviewed published literature to determine whether it supported the link between corrected QT (QTc) interval prolongation and torsade de pointes (TdP) for the 11 second-generation antipsychotics and seven second-generation antidepressants commonly implicated in these complications. Using PubMed and EMBASE, we identified four thorough QT studies (one each for iloperidone, ziprasidone, citalopram, and escitalopram), 40 studies specifically designed to assess QTc interval prolongation or TdP, 58 publications based on data from efficacy and safety trials, 18 toxicology studies, and 102 case reports. Thorough QT studies, QTc prolongation-specific studies, and studies based on efficacy and safety trials did not link drug-associated QTc interval prolongation with TdP. They only showed that the drugs reviewed caused varying degrees of QTc interval prolongation, and even that information was not clear and consistent enough to stratify individual drugs for this risk. The few toxicology studies provided valuable information but their findings are pertinent only to situations of drug overdose. Case reports were most informative about the drug-QTc interval prolongation-TdP link. At least one additional well established risk factor for QTc prolongation was present in 92.2 % of case reports. Of the 28 cases of TdP, six (21.4 %) experienced it with QTc interval <500 ms; 75 % of TdP cases occurred at therapeutic doses. There is little evidence that drug-associated QTc interval prolongation by itself is sufficient to predict TdP. Future research needs to improve its precision and broaden its scope to better understand the factors that facilitate or attenuate progression of drug-associated QTc interval prolongation to TdP.

The DPP-4 inhibitor linagliptin does not prolong the QT interval at therapeutic and supratherapeutic doses

AIM

To evaluate the potential effects of therapeutic and supratherapeutic doses of linagliptin (BI 1356) on the QT/QT(c) interval in healthy subjects.

METHODS

The study was a randomized, double-blind, placebo-controlled, four-period crossover study using single oral doses of linagliptin (5 mg and 100 mg), moxifloxacin (400 mg) and placebo. Electrocardiogram (ECG) profiles using triplicates of 12-lead 10-s ECGs were digitally recorded pre-dose and after drug administration. The mean change from baseline (MCfB) of the individually heart rate corrected QT interval (QT(c) I) between 1 and 4 h postdrug administration was the primary end point. Blood samples to measure plasma concentrations of linagliptin and its main metabolite were also obtained.

RESULTS

Forty-four Caucasian subjects (26 male) entered the study and 43 subjects completed the study as planned in the protocol. Linagliptin was not associated with an increase in the baseline-adjusted mean QT(c) I, at any time point. The placebo-corrected MCfB of QT(c) I was -1.1 (90% CI -2.7, 0.5) ms and -2.5 (-4.1, -0.9) ms for linagliptin 5 mg and 100 mg, respectively, thus within the non-inferiority margin of 10 ms according to ICH E14. Linagliptin was well tolerated; the assessment of ECGs and other safety parameters gave no clinically relevant findings at either dose tested. Maximum plasma concentrations after administration of 100-mg linagliptin were ∼24-fold higher than those observed previously for chronic treatment with the therapeutic 5-mg dose. Assay sensitivity was confirmed by a placebo-corrected MCfB of QT(c) I with moxifloxacin of 6.9 (90% CI 5.4, 8.5) ms.

CONCLUSIONS

Therapeutic and significantly supratherapeutic exposure to linagliptin is not associated with QT interval prolongation.

Mixed models for data from thorough QT studies: part 2. One-step assessment of conditional QT prolongation

We investigate mixed analysis of covariance models for the 'one-step' assessment of conditional QT prolongation. Initially, we consider three different covariance structures for the data, where between-treatment covariance of repeated measures is modelled respectively through random effects, random coefficients, and through a combination of random effects and random coefficients. In all three of those models, an unstructured covariance pattern is used to model within-treatment covariance. In a fourth model, proposed earlier in the literature, between-treatment covariance is modelled through random coefficients but the residuals are assumed to be independent identically distributed (i.i.d.). Finally, we consider a mixed model with saturated covariance structure. We investigate the precision and robustness of those models by fitting them to a large group of real data sets from thorough QT studies. Our findings suggest: (i) Point estimates of treatment contrasts from all five models are similar. (ii) The random coefficients model with i.i.d. residuals is not robust; the model potentially leads to both under- and overestimation of standard errors of treatment contrasts and therefore cannot be recommended for the analysis of conditional QT prolongation. (iii) The combined random effects/random coefficients model does not always converge; in the cases where it converges, its precision is generally inferior to the other models considered. (iv) Both the random effects and the random coefficients model are robust. (v) The random effects, the random coefficients, and the saturated model have similar precision and all three models are suitable for the one-step assessment of conditional QT prolongation.

Statistical characterization of QT prolongation

The appropriate assessment of QT prolongation remains controversial. We suggest that before the relative merits of various methods can be evaluated, we must state what we assume an assessment of QT prolongation should be about. As a general framework for the assessment of QT prolongation we propose that an assessment of "absolute" or "uncorrected" QT prolongation is properly carried out through a between-treatment (active versus placebo) comparison of the marginal distributions of QT data; an assessment of "heart rate corrected" QT prolongation is carried out through a between-treatment comparison of the conditional distributions of QT data (conditional on RR interval or heart rate). Under this general framework, conditional QT prolongation is, in general, a function of RR interval, and we discuss three possible summary characteristics for that function. We show how current procedures for the assessment of QT prolongation relate to the general approach (that is, to between-treatment contrasts of the marginal and conditional expectation of the QT interval), and to each other. It transpires that only the so-called "one-step procedure" can provide a complete characterization of conditional QT prolongation. We show that the "two-step procedure" with data-driven correction provides an unbiased estimate of expected conditional QT prolongation, which may, from a clinical point of view, be a more satisfactory characteristic than the conventional characteristic, QT prolongation at the reference RR interval. We strongly suggest that two-step procedures with fixed correction be abandoned in the analysis of thorough QT/QTc studies: Fixed correction is either redundant (when a drug has no effect on average RR interval), or systematically biased (when a drug does affect average RR interval).

Statistical models for heart rate correction of the QT interval

The analysis of QT interval data is now an essential part of the assessment of drug safety. As the QT interval is inversely associated with heart rate, an appropriate correction must be applied in order to evaluate QT data in clinical trials. The aim is to characterize changes in QT interval at a standard heart rate, taking into account the correlation between these two variables to adjust for heart rate changes during the course of the trial. It has been shown that the relationship between the RR interval (=1/heart rate) and the QT interval is highly variable between individuals but stable over time within each individual.Many mathematical models have been developed to describe the QT-RR relationship. However, there has been less emphasis on the derivation of suitable statistical models that account for the multilevel structure of the ECG data.An important example is the interpretation of the so-called population-specific heart rate corrections, which are based on data pooled from different subjects. Often, simple regression techniques are used to quantify the population correction, disregarding the subject level and leading to biased parameter estimates. Instead, population-based corrections that account for individual intercepts should be used, in order to distinguish within-subject-effects from between-subject effects. Therefore, population-specific corrections cannot be derived solely from the cross-sectional data. The impact of the different statistical models is illustrated by data from the baseline periods of six clinical QT studies.