Sven van Dijkman 1,2, Mathieu Félices 3, Bhaskar Pandurangavittal 4, Sanman Ghorpade 5, Caroline Easterbrook 6, Marcin Zabielski 7, Oscar Della Pasqua 1,2
1 CPMS, GSK (London, United Kingdom), 2 Clinical Pharmacology & Therapeutics Group, UCL (London, United Kingdom), 3 Phinc Development (Massy, France), 4 ICON plc (Bangalore, India), 5 Global Medical Affairs, GSK (Mumbai, India), 6 SERM, GSK (London, United Kingdom), 7 SERM, GSK (Warsaw, Poland)
Objectives: Paroxetine is an established SSRI with extensive post-marketing safety data over 30 years of use, from which some QT prolongation cases were identified with unclear causal relationship. While a TQT study is a preferred option for new medicines, here we show the design optimisation and implementation of a concentration-QT study. A concentration-QT (C-QT) study in healthy volunteers (18-65 years) was designed to determine whether paroxetine at clinical doses leads to clinically relevant changes in QT.
Methods: Clinical trial simulations (CTS) were used to determine feasibility and sample size, and to refine the proposed study protocol. These CTS were performed using a model based on historical ECG data, natural diurnal variation in QTc, as well as historical information regarding drop-out. Three CTS scenarios of a hypothetical true effect of 0, 5 and 10 ms were simulated and re-estimated for varying numbers of participants to determine statistical power and false positive rates. Participants received ascending paroxetine doses of 20, 40 and 60 mg once daily for one week each to ensure steady-state. At the end of each week, as well as on Day -1 (pre-exposure), paroxetine concentrations, cardiac parameters (PR, QTcF, QRS, HR) and safety were measured at timepoints -0.25, 1, 2, 3, 4, 4.5, 5, 5.5, 6, 8, 10, 12h relative to dose. To reduce burden on participants, and in light of the fact that it’s extremely rare for positive controls not to show a signal, it was decided to not include this arm. In addition, each participant served as their own reference using their pre-exposure (baseline) full QTcF profiles over the day, obviating the need for a placebo arm. All measured endpoints were summarised statistically, and a linear mixed-effects model was fit to the PK and ΔQTcF data, taking into account baseline, time-wise variation, and any potential correlation between paroxetine concentration and ΔQTcF [1]. The resulting model was used to predict the ΔQTcF (90% CI) at the mean Cmax,ss of the highest dose (60 mg). An upper 90% CI limit above 10 msec is considered to be a sign of potential QT prolongation requiring further investigation.
Results: CTS showed that an inclusion target of 36 participants should result in a statistical power of at least 80% with false-positive rates below 5% to determine a relevant effect on QTc. Geometric mean Cmax was 36.2 (29.4 – 44.6), 128.2 (108.3 – 151.8) and 221.4 (179.6 – 272.8) ng/mL at doses of 20, 40 and 60 mg once daily respectively. ΔQTcF over time showed a clear diurnal variation as expected from the CTS, with peaks not corresponding to paroxetine peak concentrations, and could be characterised equally well by the time-wise fixed effect parameters or a cosine function. The C-QT correlation showed a weak slope of 0.0108 ms/ng/mL (90% CI: 0.01, 0.03). At the highest dose, the predicted ΔQTcF at the Cmax,ss corresponding to the highest dose (60 mg) was 0.42 (90% CI: -2.68, 3.55) ms.
Conclusions: Based on the 10 ms limit, it was concluded that administration of paroxetine did not cause a clinically relevant QTc prolongation up to and including a dose of 60 mg in healthy volunteers. The use of CTS helped to optimise protocol design and limit burden on participants while maximising study efficiency. Within the context of this evaluation, where substantial safety data are available, the lack of positive control and placebo arms was not found to be a limiting factor.
References:
Garnett C, Bonate PL, Dang Q, et al. Scientific white paper on concentration-QTc modeling. J Pharmacokinet Pharmacodyn. 2018;45:383-397
Reference: PAGE 34 (2026) Abstr 11908 [www.page-meeting.org/?abstract=11908]
Poster: Drug/Disease Modelling - Other Topics