Updated 5th June 2020
Shekhar Gupta of ThePrint has a good Cut the Clutter episode on YouTube if you prefer video and don’t mind an Indian accent.
There are a few moving parts here, so let’s break it down. We, in the US tend to associate HCQ with POTUS, but for developing countries (and for those that travel to developing countries), Hyroxychloroquine or HCQ, is an established anti-malarial drug. It is also used for “chronic discoid lupus erythematosus, systemic lupus erythematosus, acute rheumatoid arthritis, and chronic rheumatoid arthritis”1.
The WHO has put its Solidarity Trial on hold given a paper that was published in The Lancet on the 22nd of May 2020. The paper is a retrospective analysis of HCQ as a treatment option for hospitalized patients.
Subsequently, there have been questions raised, on the data sources of the study, along with the statistical methods used. Additionally, an interesting advisory published by the Indian Council of Medical Research (ICMR) seems to contradict The Lancet paper.
What to make of the questions being raised about the study published in the Lancet?
The Lancet study is a retrospective study trying to ascertain the effect of HCQ treatment on patient outcomes (patients that have been admitted into a hospital). In a retrospective study, one gathers data and performs statistical analyses to generate evidence for further research. Usually, one develops a hypothesis, which is then used for a prospective study such as the one being conducted by WHO. In a prospective study, study designs (such as randomization, as an example) allow us to generate evidence which can be widely disseminated. The better, and more robust, a prospective study is, the more chances of it being adopted by the medical community, and hence better outcomes for patients. The intention behind the Solidarity Trial was just that, hence probably why they named it such. The Lancet study has found a statistically significant increase in death rates among the group of patients that were administered HCQ ie HCQ was causing more harm (as measured by the rate of deaths), than good. From this perspective, the WHO pausing its HCQ arm isn’t surprising.
So, what’s the brouhaha about? It’s mainly about the data ie how it was collected and the fact that the data was obtained from a company that warehouses medical information from different systems across the globe. The company in question is Surgisphere and its CEO is an author on the paper. There seems to be some minor discrepancies between the data as it is stored at the hospital level, and the value those data held, on the servers, at the time of the study. There are also questions being raised on the statistical methods used. The authors (read company) have not made the data available for open review. These questions are important to ask, especially when the scripts and formulas used to make analyses are not transparent.
The Lancet has published the paper and nowhere does it mention the paper is in pre-print. We should the paper has been peer reviewed. Believing a peer reviewed article requires us to trust the reviewers (usually reputable) chosen by The Lancet (reputable). It’s unclear if the data and the code for the statistical methods was made available to the reviewers, but it would be unusual if they weren’t. In other words, as far as retrospective studies are concerned, this is how the future of these types of studies will happen when commercial interests are at play. The alternative would have been an open review ie data and code made publicly available. Anyone can attempt to recreate the results. This is a more robust method of review, but again, data is money. I personally am a proponent of reproducible research and this YouTube video goes into details of why it is very important. However, while there may be minor discrepancies, it would be unusual for them to be material in this particular case unless the data warehousing is done without proper quality control.
Bottom line? There is a statistical correlation between poorer outcomes for patients being treated with the dosages of HCQ + Macrolide administered in the hospital settings under study. Note, this combination is assumed to potentially cause cardio-toxicity because both HCQ and Macrolides individually cause cardiotoxicity.
Update: It seems like the data was not made available to the reviewers, and in an attempt to audit the data by an additional Lancet review, the data could not be verfied. In fact, the study has been retracted on June 5th causing even more confusion, in my opinion.
What to make of WHO’s pausing of the Solidarity trial?
The WHO states on their website, that they are pausing the trial because of the observational study published in The Lancet. To be precise, they are pausing the HCQ arm of the trial because they want a review of the data collected, so far, by the Data Safety Monitoring Board (DSMB).They want to do this because they want to make sure the study isn’t causing more harm, than good. This is how trials work, and this review is why trials have DSMBs. Other arms of the study will continue. When WHO pauses a large trial, other similar but smaller, trials have to consider if they, too, are unintentionally causing harm. Hence, the Australian group has paused their trial for internal review, as well. This pause is a good breather because the cardio-toxicity of HCQ is a major concern (Google Scholar Search). Note, The Lancet study authors declare that they aren’t attributing the increased death rates to any potential cardio-toxicity mediated by HCQ. However, the data suggests that may be the case, and hence why everyone wants to look at their own data to confirm. Being cautious is a good thing.
Update: It seems as if the WHO could not find any reasons to discontinue the HCQ arm of the Solidarity trial ie the DSMB did not find anything in the current data that was a cause for halting the HCQ arm of the trial.
Why did we think HCQ could treat Covid19 in the first place?
This is where things get interesting. For example, a CDC affiliated lead author published research in 2005 titled Chloroquine Is a Potent Inhibitor of SARS Coronovirus Infection and Spread. Note, this is SARS and not SARS2, but the spike protein for both binds to the ACE2 receptor, and HCQ inhibits the glycosylation of the ACE2 receptor ie makes it unsuitable for the virus to bind to it. This is very useful when prevention is a consideration. In fact ICMR has released an advisory advocating for the use of HCQ, as prevention, following an internal review of various data (from both large retrospective and smaller prospective studies). Here’s the link to the advisory. When it comes to HCQ and infectious diseases, having trained in India, I tend to believe the Indians know what they are talking about.
However, of interest here, is the fact, that the dosage being advocated by ICMR is lower (reduced frequency) than those being used by WHO for its trial and lower than the mean dose studied by Mehra et al (The Lancet article). ICMR in fact, has written to WHO regarding the same, in response to the pause on the HCQ arm. ICMR suggests 400 mg BD (two times a day) on Day 1, followed by 400 mg once a week, for 7 weeks for healthcare workers. It suggests asymptomatic household contacts take HCQ for 3 weeks. Note, Mehra et al, do not perform a dose dependent analysis of their data. it is unclear why they did not do this, particularly given that a month prior Dr. Mehra was an author on a paper which alludes to an approach of treating the infection not with an anti-viral, but with drugs that stabilize vascular membranes (an excellent idea to consider). This widely applauded Medium post has quotes from Dr. Mehra regarding his previous paper.
Treatment or Prevention? This is where things get complicated. In medicine, one has to draw the distinction between the prevention of, or reduction of, symptoms from a disease agent vs treatment of a disease vs treatment and management of severer forms of a disease. This nuance is often not properly understood by non-medical people, and the science behind both can sometimes produce, seemingly, contradictory evidence.
Drug dosages and routes of administration have consequence. For example, Aspirin has a well known dose dependent profile of pharmacological action. In low doses it acts as a blood thinner and in higher doses, as a pain killer. It has topical applications too.
What about HCQ? We have some - yet to be peer reviewed - prospective clinical data on HCQ in the US. The VA trial suggests that HCQ was causing more harm, than benefit, and considering the fact that there is evidence SARS2 causes cardio-vascular damage, that isn’t surprising. Note the age of the patient cohort here. The VA population is Male and above the age of 65 yrs; Median BMI of ~30 (Obese); and practically all individuals have Hypertension. This was basically a cohort of patients at risk prone to the severer form of the SARS2 infection.
Frustratingly, it is unclear what the administered dosage of HCQ was. And there’s the rub. The authors allude to the in-vitro (in the laboratory) anti-viral effect of HCQ and note that those dosages, if translated into humans would certainly be toxic. So this raises the question. Should we really be studying the anti-viral effects of HCQ? If so, how can we study the anti-viral effects, if the anticipated dosages cannot be safely administered?
Conclusion
Based on what we know about ACE2 receptors and the role they play in the cardio-vascular system, any exploration of HCQ should be related to its effects as an Immuno-modulator, as an anti-inflammatory compound, and as a drug that potentially prevents the binding of the virus to the ACE2 receptor 2, and not as an anti-viral, unless those doses can be safely administered because we know HCQ causes cardio-toxicity.
Additionally, given that there are relatively unhealthy individuals who are more prone to the severer forms of the disease; those that are known to have cardio-vascular risk factors, administering a drug with known cardio-toxicity was likely not to have worked. There are, however, relatively healthy individuals who aren’t at risk to the cardio-toxic effects. One should consider these groups separately. This is what is known as statification in research. ICMR and the Oxford Group are focused on preventing the severer form of the disease; WHO and Mehra et al focused on patients admitted in hospital. There is a difference.
This is why it is important to understand drug mechanisms by those that report on medical information. It is important to be able to link pathogen -> pathophysiology -> drug mechanism of action -> prevention and/or treatment (read recovery). If one looks at the receptor targets and drug interactions of HCQ (see the Drugbank link below), it’s clear that its use isn’t a straight forward as it might seem. Drug re-purposing, while clever and an efficient use of time, needs clear systems level thinking so as to not only treat the appropriate patient population, but avoid confusion within the general public.
Update: The retraction of the study causes more confusion in my opinion. Does it mean that the study was fraudulent? If so, how could The Lancet publish the data in the first place? Secondly, what does this mean for those entities who are collecting medical data? Its use for research is a powerful incentive, however, the reliability of the data is very important. While journals for public consumption may not allow studies like this to be published in the future without private data audits, what about the use of these types of data-sets by private companies for internal R&D purposes? This is an interesting intersection of data privacy, contracts, business models, and research.