Igor Locatelli (1), Maja Petrovič (1,2)
(1) University of Ljubljana Faculty of Pharmacy, Slovenia (2) Krka d.d. Novo Mesto, Slovenia
Introduction:
Pulmonary arterial hypertension (PAH) is chronic and rapidly progressive disease classified as a group 1 pulmonary hypertension [1]. If untreated, it leads to right heart failure and death. Patients are classified according to disease progression into 4 World Health Organization Functional Classes; WHO FC from I (least severe) to IV (most severe). The most widely used test for disease progression is the 6-minute walking test. An improvement of the measured 6-minute walking distance (6MWD) above 33m is a valuable goal in patients with PAH for its prognostic features of increased survival.
Currently, there are three main pathways for vasodilatation used for PAH specific treatment:
- the prostacyclin pathway with prostacyclin analogues (epoprostenol, treprostinil, iloprost, and beraprost),
- the nitric oxide pathway with phosphodiesterase-5 inhibitors (sildenafil and tadalafil) or cyclic guanylate cyclase stimulators (riociguat),
- the endothelin pathway with endothelin receptor antagonists (ambrisentan, bosentan).
Several meta-analyses on this topic have been published recently [2-6]. However, none of these have focused neither on different drug dosage regimen used in clinical trials nor on patient characteristic (e.g. treatment-naïve patients or functional class of the PAH).
Objectives:
To conduct a systematic review of randomized controlled trials and to develop a Bayesian mixed treatment comparison meta-analysis (MTC-MA) model based on 6MWD data in order to estimate mutual comparative efficacy of all PAH treatments tested in trials on treatment-naïve patients.
Methods:
A systematic review protocol was registered in PROSPERO database (CRD42019110832), with detailed description of study identification, eligibility criteria, data extraction, and quality assessment[7]. For each study and each study arm the mean change in 6MWD from the baseline together with the corresponding standard errors were collected and used as observed data for the MTC-MA model. The MTC-MA model allows evaluation of direct (i.e. head-to-head) and indirect comparison among several treatments.
The MTC-MA model was developed under Bayesian framework using Openbugs version 3.2.3. Random effect models, adjusted to allow comparison of trials with more than 2 arms, were utilized for this analysis and the parameter estimation was performed using noninformative prior distributions [8]. It was assumed that the observed data arose from a normal distribution. Additionally, patients’ functional class of the PAH was used as covariate in the MTC meta-regression model. The treatment effect with 95% credible intervals (CrI) between all the observed treatments was reported. Rank ordering of each treatment was also estimated.
Models were evaluated using the Deviance Information Criterion, overall residual deviance, and the degree of between-study heterogeneity. The initial 10,000 iterations of models were discarded as ‘burn-ins’, and models were run for 100,000 iterations to ensure the convergence. For evaluation of potential inconsistency between indirect and direct treatment effect, a node-splitting analysis was used on a specific node with a closed loop in the network [9].
Results:
In total 21 RCTs (44 data points) with 10 different PAH-specific drugs and 13 different treatments (including placebo) were obtained. The observed data resulted in star like network pattern with some triangular loops.
Based on the developed MTC-MA model, the treatments can be rank ordered as follows: tadalafil and ambrisentan combined (treatment effect = 68m, 95% CrI: 50-87) > epoprostenol = i.v. treprostinil (high dose) > 250mg bosentan = ambrisentan = sildenafil = 40mg tadalafil > riociguat = iloprost > oral treprostinil = beraprost = 125mg bosentan > s.c. treprostinil (for the 8-fold lower dose compared to i.v.). No loop-specific inconsistency was observed. Between-study standard deviation was low (2.3m).
Based on MTC meta-regression model we observed that studies with severe patients (WHO FC III or IV) obtained 29m (95% CrI: 1-57) higher treatment effect of the PAH treatment over placebo than studies with low risk patients.
Conclusion:
The combination of ambrisentan and tadalafil showed the biggest improvement in 6MWD, followed by epoprostenol and parenteral treprostinil. Additionally, bosentan (250mg dose), ambrisentan, sildenafil, and tadalafil (40mg dose) also reached clinically significant effect on the 6MWD.
References:
[1] Galie N et al. Eur Heart J, 37(1):67-119, 2016.
[2] Duo-Ji MM et al. Int J Cardiol, 234:90-98, 2017.
[3] Fox BD et al. Can J Cardiol, 32(12): 1520-1530, 2016.
[4] Gao XF et al. Patient Prefer Adherence, 11:871-885, 2017.
[5] Jain S et al. Chest, 151(1): 90-105, 2017.
[6] Lajoie AC et al. Lancet Respir Med, 4(4):291-305, 2016.
[7] https://www.crd.york.ac.uk/PROSPERO/
[8] Jonas DE et al. 2013. Findings of Bayesian Mixed Treatment Comparison Meta-Analyses, Rockville, Maryland, USA.
[9] Dias S et al. Stat Med, 29(7-8):932-944, 2010.
Reference: PAGE 30 (2022) Abstr 10004 [www.page-meeting.org/?abstract=10004]
Poster: Drug/Disease Modelling - Other Topics