Covid has changed how we do research

Coronavirus is resetting the balance of risk and benefit for how we perform research.

Even last year it seemed to take an interminable amount of time to get even simple innovations approved — now it happens in the blink of an eye. In just the last few days the U.S. Food and Drug Administration has approved a new COVID diagnostic using the CRISPR-cas9 method with an emergency use authorisation. Perhaps we can maintain this sense of urgency when this is all over.

The fact that we had no pre-existing clinical tools meant that even relatively ineffective diagnostics and treatments have a place. Regulators and governments also now realise that there is a cost to caution and delay.

This is somewhat reminiscent of the early years of HIV-AIDS. In 1980 the first patient in the epidemic was recognised – albeit retrospectively. There are now known to be earlier victims. The virus was not identified until 1982 with the first diagnostic in 1985. It took until 1987 for the first drug — AZT — to be developed by the British company Burroughs Wellcome based on a discovery in 1962. Fascinatingly, many drugs were tried in a multi-parallel approach, and activists succeeded in getting early access in 1986 to the experimental drugs before they were approved. Compared to COVID this seems glacial - a diagnostic was developed within a few short days and vaccine candidates are already entering the clinic.

Researchers are testing a myriad of different hypotheses about COVID biology and therapeutics. Two have struck my eye this week:

Firstly, Vitamin D. People with low vitamin D seem to be worse affected than those with normal levels. Vitamin D is produced by the body in response to sun exposure which could, therefore, explain why equatorial countries and southern hemisphere countries coming out of summer — such as Australia and New Zealand have been less affected than northern hemisphere countries coming out of winter. There are potential holes in the theory — such as the impact of COVID in South America — and the relationship may not be causative, but nonetheless interesting.

Secondly, the GenOMICC study on the genetics of mortality in critical care has been extended to look for genetic markers of susceptibility. Some of these may turn out to have plausible causes of the disease, raising the possibility of new therapeutic targets to prevent people from getting unwell.

As an aside, we must be very careful about drawing linkages between different gene loci and ethnic grouping since we know that the genetic differences between ethnic groups are relatively small when compared to genetic differences within ethnicities. There is no genetic definition of ethnicity or race, and will likely never be.

These research threads indicate the sheer breadth of thinking that is sitting alongside clinical trials of existing drugs. Much of this work is not being coordinated by a central authority but is instead responding to the incentives being provided guided by Adam Smith’s invisible hand.

This provides an interesting contrast with the centralised governmental response in the UK and elsewhere. These responses have been characterised by ponderous, non-transparent thinking by politicians with no expertise. Scientific research is rapidly and brutally challenged on a time scale that could not be more different than governments which effectively change about every four years or so in most countries.

What is also noticeable is the much more pragmatic nature of the research. Previously much of the focus has been on testing specific biologic mechanisms. Only about three in 100 drugs developed this way that are put into humans survive the rigorous testing process to be approved as medicines. Almost all of those 100 drugs have a plausible mechanism yet most fail even though they apparently get to the right biological receptors in a reasonable concentration. Biology is not as logical or predictable as some of the adherents of the scientific method might suggest.

We are now seeing a more open ended approach to research. The starting point is searching within large, real-world datasets looking for associations between outcomes and specific interventions or patient characteristics. Many of these associations will be noise or could be correlated but not causative — however, the value of this starting point is the knowledge that there is a real world relationship that can then be expanded using our knowledge of biological systems.

Although I do not hold out hopes for a treatment or vaccine in the next few months, my prediction is that combining flexible and sensible medicines regulation with this ‘crowd sourced’ model of research could be effective. Focusing on what works — rather than what we can explain — will show impressive speed compared to current drug discovery paradigms.