Reply to McShane

To the  Editor—We are pleased to see that our recent Perspective entitled, “An aspiration to radically shorten phase 3 tuberculosis vaccine trials” [1] was accompanied by a commentary by experienced tuberculosis vaccine researcher Professor Helen McShane [2]. We hope that these pieces will stimulate debate about this important topic. As part of this, we would like to address some of the statements made by Professor McShane.

First, it is argued that key aspects of our suggested approach would be unlikely to be acceptable in today's ethical and regulatory environment. Our view is that the opinion of regulators should not be presupposed and engagement of them in these discussions is crucial, so more creative and efficient trial designs might be fully considered in vaccine development.

Second, at the heart of the issues raised are the safety data from the trial and case ascertainment. With respect to the safety data set, it does not make epidemiological sense that the sample size required for efficacy must be the same, or even similar, as that for safety. Our short-duration vaccine trial design proposes to enroll a larger number of participants to measure efficacy. However, for a disease with relatively low incidence of cases over follow-up, such as tuberculosis, whether all participants need to be subject to all safety evaluations for regulatory purposes is especially important to consider. We propose that safety follow-up for regulatory purposes should have its own independent sample size requirements and that this may well result in a smaller number of participants needing stringent safety assessment compared to efficacy evaluation. We note that such an approach would not be new. For example, only 58% of participants in a phase 3 trial of a herpes zoster subunit vaccine underwent stringent safety follow-up [3].

With respect to case ascertainment, it is important not to conflate stringency of end point ascertainment with completeness. Stringency of end point ascertainment is about specificity, while completeness is about sensitivity. Our key point is that in a vaccine trial specificity should be maximized to prevent bias in the estimation of vaccine efficacy, but that sensitivity can be less than 100% because it will only reduce statistical power without affecting the validity of the efficacy estimate. The decision relates to efficiency, not vaccine efficacy estimation. As such, the effort needed to reach 100% sensitivity should be balanced against the effort needed to enroll the number of trial participants to make up for the loss in power when sensitivity to detect a case is lower than that.

Finally, we want to thank Professor McShane and the journal for stimulating discussions on strategies to come up with a more effective tuberculosis vaccine in the shortest amount of time. We hope that fellow researchers, regulators, and the other relevant stakeholders engage to find creative alternatives to the traditionally slow and extremely expensive pathway for tuberculosis vaccine development.

Notes

Financial support . No financial support was received for this work.