Drug Development for Pulmonary Arterial Hypertension: Unleashing the Potential of Single-Patient Studies Using Continuous Monitoring

Martin R. Wilkins, Sofia S. Villar, James Wason, Mark Toshner, Alexander M. K. Rothman

https://doi.org/10.1002/pul2.70177

Editorial

Drug Development for Pulmonary Arterial Hypertension: Unleashing the Potential of Single-Patient Studies Using Continuous Monitoring
Martin R. Wilkins, Sofia S. Villar, James Wason, Mark Toshner, Alexander M. K. Rothman
First published: 15 October 2025 https://doi.org/10.1002/pul2.70177
SectionsPDFTools Share
In the late 1940s, the US Air Force recorded an unacceptable number of crashes and emergency landings, prompting a formal enquiry. It was resolved that neither the engine nor the pilot was at fault. Attention turned instead to the design of the cockpit, which was based on the ‘average’ airman from 1926. Could it be that the body proportions of airmen 20 years on had changed? A subsequent survey of over 4000 airmen collected 132 detailed measurements on each pilot [1]. From this list, 10 were chosen that seemed most relevant to cockpit design and the number of airmen who fell into the middle 30% range for all 10 was calculated. The answer was 0. Even with three measures, under 3.5% of airmen were ‘average′. The survey report concluded that “The tendency to think in terms of the ‘average man’ is a pitfall… it is virtually impossible to find an ‘average man’ in the Air Force population” [2]. A contemporaneous competition to find ‘Norma′, the average American woman, based on average body proportions came to the same conclusion for women; the ‘average woman’ is rare [3].

These stories are well known [4]. Yet we continue to rely on average measurements in a variety of situations, including when designing and interpreting the results from clinical trials. In drug development, this may cause us to miss an efficacy signal that exists in a subset of subjects in a study that fails to meet a pre-defined average endpoint: perhaps stopping a promising candidate drug from progressing further. Or it might lead to pursuing a higher dose than is necessary for ‘responders′, to achieve a pre-defined averaged measurement for the whole study group. Moreover, translation of the mean result from a randomized clinical trial (RCT) to the clinic is frequently disappointing. Valid estimates are difficult to come by and numbers will depend upon the indication but less than a half and perhaps lower than a third of patients likely respond to the licensed dose [5, 6]. Relying solely on average patient data to inform a dose and dose regimen to treat an individual patient is a pitfall analogous to that of designing a cockpit for the average US Air Force pilot.

A common approach to addressing variation in drug response is to conduct a subgroup analysis of a RCT and identify ‘responders’ (i.e. a subgroup that appears to derive benefit from the drug) based on dichotomizing the outcome; using a threshold measurement that is viewed as clinically meaningful. But this is predicated on the assumption that the recorded response for each patient is reliable and reproducible; if the response cannot be replicated, then our identification of ‘responders’ is insecure [7].

Read the full editorial