Abstract
Mechanistic studies are designed to investigate the patient or a disease, and not a medicinal product. They can be descriptive or experimental and often involve a medical product to help understand why human physiology works in the way that it does and not to establish that a medical product has certain clinically observable effects, which can be both positive and negative. As MacAllister and Veighey point out in this issue, these studies are not covered by the Medicines for Human Use (Clinical Trials) Regulations 2004 in the UK, so there is no legal obligation on the investigator to register a protocol or indeed to report its results.
We might imagine that mechanistic studies are subject to publication ‘bias’ towards positive findings in a similar way from which clinical trials suffered before they were more fully regulated. Indeed, a very brief exploratory search of PubMed Central (one sunny Monday lunchtime in September) suggested there may be more positive findings of mechanistic studies published in medical journals than negative ones. Before we get too excited and call for instant regulatory change in response to this interesting yet largely unsubstantiated inference, we must ask whether systematic publication of positive and not negative findings is as much a moral problem for mechanistic studies as it was for clinical trials, and, if it is, whether regulation is needed or best placed to solve it.
The main case for publishing all clinical trial results is to inform clinical practice and clinical research without delay. Mechanistic studies are not designed to discover any information about risks and expected benefits of medicines and so a similar argument seems not to apply. Quite apart from the additional and equally important desire not to duplicate studies unnecessarily and waste the finite resource of willing and healthy humans, there is increasing recognition that, at least in some cases, the results of mechanistic studies can inform the risk profile of medical products in future clinical trials designed now to evaluate their safety and efficacy. For example, Mazzone et al. (2008) described three areas in which pathophysiological considerations in type 2 diabetes could be particularly informative in clinical assessments of cardiovascular risk, namely the roles of hyperglycaemia, diabetic dyslipidaemia (other than the control of LDL-cholesterol concentrations), and inflammation (including that in adipose tissue) in the acceleration of vascular injury.
There is a strong incentive for investigators of mechanistic studies not to measure or record clinically relevant outcomes even when this is not their primary intention, for fear of falling within the burdensome and bureaucratic regulatory framework for clinical trials. Clearly, some of the above results may have immediate and direct clinical relevance. Being at increased risk of cardiovascular disease, patients with type 2 diabetes may take appropriate preventative treatments. 1 The results of mechanistic studies may also be relevant to proposed clinical trials of investigational medical products. Patients with diabetes may be excluded on grounds of safety or may constitute a sub-group for subsequent risk analysis. As a principal investigator of a proposed clinical trial, there is a legal obligation to disclose all available preclinical and clinical evidence to the regulator (in the UK this is the Medicines and Healthcare products Regulatory Agency (MHRA). However, the results of mechanistic studies may not be available to the investigator if they are not published.
While there is a clear prima facie reason to require the publication of all mechanistic studies, which by itself is entirely consistent with the investigator’s primary ambition to pursue science and contribute to medical advances, there may be overriding concerns about regulation per se which might make us hesitate before concluding that the regulation is needed to ensure they are so published or which suggest that it may in practice not be effective in achieving publication.
Firstly, it hardly needs saying that if mechanistic studies were regulated to the same extent as clinical trials, the number of studies would decrease dramatically. These studies are often very small scale in any case, and investigators may not then be prepared to invest the considerable time and effort to negotiate the paperwork. This is unlike a series of clinical trials which may even be seamless, which are mounted on an increasingly large scale to gain statistical power and to reach the standards of evidence required for a license to market a new medicine.
Secondly, regulation of clinical trials has not seemed to have resulted in publication of all results as was initially supposed. Publication bias has long been recognized as a problem in medical science and has been addressed in Europe with the introduction of the EudraCT database. However, the problem is seemingly not solved, as it is accessible only to public authorities (unlike in the US where ClinicalTrials.gov is publicly accessible) and is not complete (Ross et al., 2009).
Thirdly, there is an existing albeit quasi-legal framework under which publication of mechanistic studies can be encouraged or even required (and under which human subjects are protected). As MacAllister and Veighey observe, university ethics committees may now take the lead on governing mechanistic studies. They may need to become more concerned with scientific quality and reporting results than ethics committees recognized by the regulator for reviewing clinical trials would need to be. But what is clear is that the career ambitions of investigators are often consistent with or driven by the desire to publish, and we would think that little pressure would be necessary from outside to promote the practice.
In this issue
In addition to raising the issue of mechanistic studies in the Topics section, there are two original articles in this issue. The first picks up an issue which is often faced in practice but on which there is little analysis. Kirsten Bell and Amy Salmon examine some of the assumptions members of ethics committees make about people addicted to illicit drugs, and so raise the more general question of how independent we should expect the process of ethics review to be, given their inevitable background beliefs and, yes, even their prejudices. The second article, by Jacqueline Crowther and Mari Lloyd-Williams, asks how carers of people with dementia view research on sensitive and emotive topics, given they are vulnerable to the extent they are caring for another. While these articles examine different types of research, they raise different dimensions of vulnerability. As a concept, vulnerability seems interestingly elusive. In some sense, all research subjects are vulnerable to abuse by researchers; otherwise they would not require the protection of ethics committees. There are some who seem especially vulnerable either by virtue of being mental non-competent or by virtue of being socially or politically powerless.