Sylvie Retout (1), Sébastien Jolivet (1), Stefan de Buck (1), Vincent Duval (2), Valérie Cosson (1)
(1) Roche Pharma Research and Early Development, Pharmaceutical Sciences, Roche Innovation Center, Basel, Switzerland, (2) Certara Integrated Drug Development Basel, Switzerland
Introduction: Baloxavir marboxil (BXM) (otherwise known as XOFLUZA™, RO7191686, or S-033188) is an anti-influenza virus drug with a novel mechanism of action. Its active form, baloxavir (BXA), selectively inhibits cap-dependent endonuclease activity, an influenza virus-specific enzyme, which thereby inhibits influenza virus replication. BXM has already been approved in several countries including Japan and US for the treatment of influenza types A or B (ItAB) in patients aged ≥12 who are otherwise healthy (OwH) or at high risk of developing influenza-related complications (HR).
A population PK analysis conducted on phase 1 to phase 3 clinical data identified race and body weight (BWT) as significant covariates on BXA PK, with a 48% lower apparent clearance (CL/F) in Asians compared with non-Asians, and a CL/F increase with increasing BWT. In Asian and non-Asian OwH or HR patients with ItAB, the recommended BXM dosing (40 mg for patients with BWT of 40 kg to < 80 kg, 80 mg for patients with BWT ≥80 kg) was demonstrated to be significantly superior to placebo in the time to alleviation of symptoms (TTAS), a widely accepted and historically used clinical endpoint for anti-influenza drugs.
A new drug application was planned in South Korea (SK) based only on a PK bridging study, avoiding a local phase 3 study. Investigations of any ethnic difference on BXA PK and TTAS that would justify a dose adaptation in SK patients were therefore undertaken to support the bridging strategy.
Objectives: To develop a BXA PK-TTAS model and to quantify the potential influence of covariates, including ethnicity effect, on the BXA PK-efficacy relationship.
Methods:BXA PK-TTAS model was developed using data from 1781 patients with ItAB, included in one OwH phase 2 and two pivotal OwH and HR phase 3 studies. A placebo model describing the natural distribution of TTAS events in the absence of treatment was first developed using a parametric time to event approach [1] on the data from the placebo arms (N=689). Several parametric distributions were tested to describe the instantaneous probability of symptom alleviation at a time t (hazard function): exponential, log-normal, Weibull, Gompertz and log-logistic distributions. Covariates that may influence this baseline hazard were investigated: age, BWT, race (Asian or non-Asian), gender, patient status (OwH or HR patients), vaccination status, smoking status, time from symptom onset to study screening, and composite symptom scores at baseline (TSS0).
In a second step, PK and TTAS data from 1092 patients included in the BXA arms and with available BXA PK data were pooled with the placebo data; the BXA effect on the TTAS distribution was then investigated. A joint sequential analysis was undertaken [2], investigating the impact of the individual PK time courses, derived from the Bayesian estimated individual PK parameters, on the TTAS distribution. Drug effect models (such as linear, power, Emax) were tested on the baseline hazard and the impact of the covariates (listed above) on the drug effect was investigated.
The predictive performance of the developed PK-TTAS model was assessed using Kaplan–Meier visual predictive checks (KM VPCs) [1].
Results: A log-logistic distribution best described the placebo TTAS data. Race, gender, patient status and TSS0 significantly influenced the placebo TTAS distribution, with for instance a longer TTAS in Non-Asian compared to Asian patients.
The hazard increased proportionally with BXA concentrations, indicating a clear BXA PK-TTAS relationship. Importantly, once the covariates on the natural TTAS distribution were accounted for, no other remaining covariates were found to influence the intrinsic BXA exposure effect on TTAS, implying similar BXA exposure effect on TTAS between Asian and non-Asian patients.
KM VPCs showed good predictive performance of the PK-TTAS model, regardless of race, gender, or TSS0, indicating all relevant covariates were well accounted for.
Conclusion: A clear BXA PK – TTAS relationship has been established. While various covariates influence the natural TTAS distribution, no ethnic sensitivity with respect to the intrinsic effect of BXM was shown and therefore no meaningful differences in the efficacy of BXM in SK patients compared to other Asian patients is anticipated. This, together with the demonstration of PK similarities between Asian and SK patients, allowed the BXM approval in South Korea without conducting any local phase 3 study.
References:
[1] Holford N. CPT:PSP (2013) May; 2(5): e43.
[2] Desmée S et al. AAPS J. 2015 May;17(3):691-9.
Reference: PAGE () Abstr 9427 [www.page-meeting.org/?abstract=9427]
Poster: Drug/Disease Modelling - Infection