Limitations of proposed novel trial design

Résumé fourni par la source

We read with interest the article by Mani et al.[1], which proposes a novel trial design for new antiretroviral drugs intended to be used in treatment-experienced HIV-infected patients on a failing regimen. The new design is primarily motivated by the small current number of patients with multidrug-resistant virus causing difficulties in recruiting to trials for which this is an inclusion criterion. The new design has two phases. In the first phase, patients experiencing virological failure are randomized to receive the investigational drug or placebo for 10–14 days, while remaining on the failing regimen; the primary assessment of efficacy (virological response) takes place at the end of this phase. In the second phase, the background regimen is reoptimized for all patients, and those who received placebo in the first phase are also switched to the investigational drug, with follow-up continuing to 24 weeks. The rationale for this design is to provide evidence on safety and efficacy, while minimizing the chance of resistance developing to the new drug or additional resistance to the background drugs. We have several concerns about this new trial design and the implication in the article that it could replace traditional phase 3 trials (a control group maintained for a minimum of 48 weeks) currently required for licensing by regulatory agencies [2]. As all participants would receive the new drug after 10–14 days, the randomized safety evidence elicited by the new design is limited to acute adverse reactions. Thereafter, judgement is required in assigning the causality of each adverse reaction to an individual drug, which is recognized to be very difficult in the context of combination therapy [2]. Thus, only highly specific adverse reactions which have been previously unobserved in all of the other background drugs can be reasonably attributed to the new drug. Such reactions, if they exist at all, are likely to be rare, and may well not be observed in the proposed sample size of 200–300 patients over 24 weeks of observation. Another weakness of the proposed design is that it leaves unanswered the critical questions of the durability of efficacy and the emergence of resistance because, again, there is no control group to reliably assess these factors. Furthermore, phase 2 dose-ranging monotherapy studies assessing short-term virological response would also usually be conducted, and the added value of similar information in treatment-experienced patients, who may or may not have viral cross-resistance to the investigational drug, is unclear. The authors acknowledge these limitations of the proposed design and the need for ‘confirmatory’ or ‘other’ trials [1]. However, if approval was granted on the basis of evidence from the proposed trial design, there is no guarantee that larger and longer trials would be conducted. Phase 3 trials should attempt to reflect the way in which the investigational drug will be used in clinical practice, including the drugs with which it is likely to be coadministered, while acknowledging that different considerations may apply in licensing and academic-led trials. In contrast, the proposed trial design assesses the investigational drug in an artificial way, namely adding it to a failing regimen. We are also not persuaded by the arguments in the article about protecting patients from unknown toxicities and losing future treatment options. The stringent monitoring of phase 3 trials, including the regular unblinded assessment of the data by an Independent Data Monitoring Committee, ensures that the minimum number of patients are exposed to a drug which may have suboptimal efficacy or an unacceptably high rate of serious side effects. Indeed, there is strong ethical argument against rolling out new drugs in clinical practice in potentially large numbers of patients, without this being underpinned by a traditional phase 3 trial [3]. Acknowledgements Conflicts of interest None declared.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.