The ‘failure’ of clinical trials in the search for COVID-19 treatments
Professor Jean-Pierre Boissel outlines the urgent need for disruptive R&D approaches that can tackle the search for COVID-19 treatments more effectively.
It may not be appropriate to use the word ‘failure’ when summarising the utility of clinical trials in the search for a curative or preventative treatment for disease. For the purposes of this text, let us assume that ‘success’ and ‘failure’ will be used in reference to the objective of clinical trial designers: to obtain sufficient evidence that a treatment under evaluation is of benefit to patients.
SARS-Cov-2, responsible for the COVID-19 infection, rarely kills those who are infected, but unfortunately, the number of deaths caused by the virus now exceeds 1.2 million worldwide and continues to increase despite the global public health measures that have been put in place. As time goes on, the number of individuals requiring access to intensive care facilities exceeds capacity. The main measure available to fight against the spreading of the virus, social distancing, is becoming ever more difficult to bear.
Compared to the number of deaths and the millions of those infected with COVID-19, the number of patients included in randomised clinical trials, the only type of investigation able to provide the necessary evidence for the safety and efficacy of potential new treatments, is ridiculous: less than 10,000 in total for the Recovery and Discovery trials, for which we have initial data, plus an equivalent number for a US NIH trial, a WHO study, and trials launched in China or by the industry.
Despite hundreds of trials on their way there are only a dozen drugs being tested currently with reduction of death as endpoint and yet there are as many as 150 drugs already on the market for other related indications and hundreds more that could potentially have a role to play, targeting the viral proteins, the virus interactions with the target cells or on the human systems that are solicited by the infection.
If, as it is likely, it will be necessary to combine drugs to obtain a cure, and even personalize these combinations to individuals, the number of ongoing or completed trials appears insignificant compared to the scale of the disease. However, in vitro or binding-based efficacy insight needs to be confirmed by reliable clinical evidence, as shown by the outcomes of COVID-19 trials with remdesivir and hydroxychloroquine.
Most of the trials are underpowered to demonstrate efficacy on mortality and long-term consequences of the disease. The scale of the impact of COVID-19 dwarfs the clinical data obtained to date so it is probably not surprising there is currently no recommended drug treatment for COVID-19, except dexamethasone and, in some countries, remdesivir, despite little evidence that this drug reduces mortality.
We have ‘half-successes’ such as dexamethasone for Recovery, which might only be effective in subgroups of patients, and still needs to be confirmed by at least one other trial, or ‘half-failures’ such as Discovery, the complexity of which is poorly suited in the context of caring for severely ill patients hospitalised in intensive care units. Recruitment into trials is difficult, or outcomes are uninformative as with remdesivir, or like the hydroxychloroquine trial in China, inadequately powered. Even Solidarity, the WHO trial, is some way from yielding strong data.
The only relatively well-argued result we have today is the ineffectiveness of hydroxychloroquine, which was predictable. Also, evidence in favour of a reduction of mortality with lopinavir/ritonavir is elusive. Evidence regarding efficacy of remdesivir on mortality is awaited, the NIAID trial results published recently did not show a statistically significant reduction of mortality
At this rate, how many years might it take, unless luck or serendipity strikes, to test all the options and finally find a cure for the most severe patients?
The picture is hardly more promising for vaccines: the number of approaches, difficulty in finding patients in which to carry out pivotal trials, irrelevant choice of endpoints … the outlook is far from promising.
There are many reasons as to why so few eligible patients are being recruited into clinical trials even though infection numbers are high. The exhortation by the late Thomas C Chalmers to “randomise the first patient” which would translate into the best ethical behaviour by doctors has never been put into practice. If it had been applied to COVID-19, at the very least 2,000 molecules or combinations could have been tested so far against mortality: taking 200 as an average number of endpoints per trial (quite a large number indeed!) and the death toll today. However, this is pure speculation since the story of randomised clinical trials which began more than half a century ago shows us that Chalmers’ words are, sadly, wishful thinking.
The traditional approach to R&D, with its high failure rate, is not acceptable when it comes to COVID-19: the expected time to efficient treatment or vaccine (ETETV) is unbearable compared to the rate of growth of the burden of the disease. Repurposing existing drugs with prior evidence of PK and safety for the potential treatment of COVID-19 only marginally improves the inefficient series of decision-making stages in the traditional R&D process
There is therefore an urgent need to turn to a disruptive approach, modeling and simulation (M&S), to make efforts to tackle COVID-19 more efficiently.
The M&S approach should be integrative, from the molecular level to the clinical and population outcomes. It should be conceived both as a proof of concept tool and a MIDD (Model Informed Drug Development, to use the FDA jargon) tool for potential treatments, including vaccines, and combinations of drugs. It will allow inefficacious treatments to be dropped earlier, sparing time, resources and patients. Also, as a knowledge integrator, enabling the assimilation of large amounts of knowledge that scientists will gather on the virus and the disease. Once the model is achieved and validated, simulations to test hundreds of treatments or vaccines can be run in parallel by sponsors in a just couple of days.
For COVID-19, this can be achieved through a series of connected modules:
- the SARS-cov2 virus, its dissemination dynamics, entry into the cell, replication, interactions between viral intracellular activities and energy metabolism, the regulation and phenotype and age-related distribution of ACE2, the infected cell fate;
- the systems level and responses (innate and adaptive immune system, inflammation, complement, vascular system and coagulation;
- alveolar and gas exchange changes and dynamics of organ irrigation changes;
- dynamics of organ failure (heart, gut, kidney, nervous system);
- patient level and clinical outcomes, including chronic lung fibrosis and chronic fatigue syndrome;
- population level (ways and dynamics of viral penetration, role of social connections, social context, Covid-19 prevention measures).
An M&S approach will enable the prediction of major impairments due to COVID-19. Not only death, but also the chronic fatigue syndrome (‘Long COVID’) which is a major burden for some patients.
Eventually, the model could help to individualise treatment for patients.
About the author
Professor Jean-Pierre Boissel is co-founder and chair of the scientific advisory board, Novadiscovery.
Jean-Pierre is a cardiologist by training. He was until recently an Emeritus Professor of Clinical Pharmacology at the University of Claude Bernard Lyon 1, and has been a scientific director at the French Institute for Health and Medical Research (Inserm), advisor to the General Surgeon and is still a member of a number of scientific societies in Europe and the United States.