The infamous hydroxychloroquine

A (limited) in silico perspective.

Image for post
Image for post
image: ChiralJon licensed under CC BY 2.0

The story of hydroxychloroquine during this pandemic is a sad and winding one. I think this article from Tablet does a great job articulating the tensions and misadventures that took place in the quest to either support or refute the case for its efficacy as a coronavirus therapeutic. In any case, while the FDA withdrew its emergency use authorization (EUA) for hydroxychloroquine (hcq for short) use, there are some other jurisdictions that still administer it for coronavirus indication.

When people ask what is our ability at EMSKE to support or refute a given small molecule drug’s therapeutic efficacy for patients diagnosed with COVID-19, I always have to make clear that we restrict our lead-generation strategies to one specific mechanism of managing COVID-19 cases: protease inhibition.

By contrast, there are myriad ways that a given small molecule drug might interact with the infection and immunological response for the better (or the worse) of the patient. They could interfere with the spike RBD site (aka the ‘tip of the spike’) interface with the human’s ACE2 (aka the doorway to the human cells through which the virus enters) enzyme. Rather than touching the virus at all, they could modulate the body’s immunological response to keep it from overreacting. Or a completely different mechanism altogether.

So with that in mind, I thought we would review hcq for SARS-CoV-2 protease inhibition, applying the same toolchain and processing formula we use for all the other small molecule drug candidates (usually natural product ones) we look at.

In past articles we’ve talked so much about what makes a drug candidate perform well, but very little about what it means to be a poor performer. To remind, our scoring system is calibrated to 3rd party biomedical researchers’ published results on inhibitory assays of the original SARS’ protease from the literature. The calibration informs us that the effect of our docking score on actual concentrations required to inhibit protease activity is logarithmic in nature. Every additional kilocalorie per mol required to achieve the 50% inhibition value (‘IC50’) represents approximately a doubling of the concentration required in vitro to achieve that setting.

So if we have a compound with a score of -7.7 kcal/mol , it inhibits _roughly_ at 82 “micromolar” or uM, which is to say 82 millionths of a mole per litre. If you remember your high school chemistry, that’s 82 * 6.02 * 10²³/10⁶ = 494 * 10¹⁷ molecules of the drug per litre. (In practice it could be as low for example as 50 or as high as say 120 uM, but biochemically these values are comparable to each other as order-of-magnitude is what matters).

Anything beyond 100 micromolar is generally not worth looking at, and between 10 uM and 100 uM might be interesting but probably needs optimization of the molecule before it can be useful in humans . Otherwise there’s too strong a chance the drug will be requiring such a high dosage (and therefore concentration) to be effective that it interferes with other normal human biochemical functions — i.e. side effects that nobody wants.

Then without further ado, here is hydroxychloroquine’s result:

hydroxychloroquine: -5.6 kcal/mol.

In hydroxychloroquine’s case therefore, we project a concentration of 344 uM is required to inhibit 50% of the protease’s action in a laboratory setting. That’s a lot. Drug screeners generally wouldn’t even countenance as an antiviral that interacted with its target at an IC50 > 100uM as a lead, much less a trial-ready drug.

And remember, in this discussion, beyond just ignoring other elements of the viral replication pathway than the protease, we’re also ignoring the eminently important role that pharmacokinetics plays in the biological filtering that the body presents to any small molecule drug before it can reach its target.

Here is one recent paper surveying actual efficacy assay results of hcq on SARS-CoV-2 in vitro, from the journal Nature: . IC50 of 4.17 uM . So a much better concentration than our protease-restricted assay would predict.

Perhaps it is attacking a different mechanism in viral replication? Indeed there are many elements of the viral replication mechanism worth mounting an attack on. But for hcq at least, the protease is unlikely to be one of them. And anyway, the Nature paper mentioned that when hcq was applied to infected macaques (a primate), it had no discernable antiviral effect.

Alas, maybe someday software-based models will be sufficiently developed enough with viral replication inhibitory models and pharmacokinetic models to be able to to nip future pandemics in the bud without the need for all the in vitro and clinical trialing work. Perhaps then the scientific establishment could more easily avoid another hcq-like controversy and the collective embarrassment that’s stemmed from it. But that day is not today.

Written by

Coming from a multidisciplinary MIT technical background and startups in San Francisco & SE Asia, Rick leads @EMSKEPhyto in Kenya.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store