Developing countries aren’t going to want to wait to access verified COVID-19 therapies
With the coronavirus epidemic still reaching its zenith in the US, I observe friends and family in San Francisco who work in ER wards and nearby county clinics post their scary treatment stories and PPE fundraisers on Facebook. It’s evident that something is very wrong.
Meanwhile, I’m stuck at home in Kenya with a lot of time outside my social venture work to occupy. And as former astronaut Chris Hadfield says, “Become an expert on the thing that threatens you”. So in spite of my layman’s credentials (I’m an engineer, not a biochemist or molecular biologist) I’ve been reviewing the literature on this disease. And as much as it’s out of line to try and inform people outside of one’s domain, these are strange times and I feel compelled to post what I’ve learned about here -I don’t want to wonder in the future whether I lost people I know because I stayed quiet if no risk would be posed for it.
I’ll describe by way of outlining my learning progression, because it should be evident how I got from A to B to C.
In late February as the epidemic in China was starting to wane but picking up elsewhere, my anxiety grew. Seeing reports that an HIV antiviral Kaletra (lopinivir and ritonavir) was showing early anecdotal success in patients in China, I asked my Chinese b-school classmates for links to scholarly papers gaining currency on the topic in China, simply so that I could try to reduce my anxiety. (I figured that papers gaining currency in Chinese media would have a slight lead on what I could find through Western channels)
One of the papers they identified was this preprint from Hong Kong Polytechnic University, whose researchers cleverly use Python tools and a drug database to do a drug-docking study:
https://f1000research.com/articles/9–129 (f1000 is an open research platform supported by the Bill & Melinda Gates Foundation)
(in case the f1000research link is down, this is a permalink to the pdf)
“Prediction of the SARS-CoV-2 (2019-nCoV) 3C-like protease (3CLpro) structure: virtual screening reveals velpatasvir, ledipasvir, and other drug repurposing candidate”
Yu Wai Chen, Chin-Pang Bennu Yiu, Kwok-Yin Wong
Hong Kong Polytechnic University
It describes ‘in-silico’ simulated drug screening against the SARS-nCoV-2 virus.
Proteins are capable of interacting if they have a 3D structural match and complementary chemical bonds at their active sites — like two puzzle pieces fitting together. These matchings, called drug docking, can be simulated on powerful computers, and so are called ‘in silico’, to distinguish themselves from ‘in vitro’ (testing a compound on actual animal cells, typically in a petri dish), and ‘in vivo’, (testing the efficacy of a compound in a live animal or even a live human population).
Reading that paper sensitized me to the proteins that are being targeted by drug designers during this epidemic. For example, one protein makes the virus’s famous spikes. There are other proteins required for the virus to enter a cell and begin replicating itself using the cell’s machinery. Altogether there are between 27 and 29 unique proteins that make up this particular coronavirus.
The protein the researchers from Hong Kong Polytechnical University decided to target is called 3CLpro. And helpfully they compared this protein to its counterpart in the coronavirus from the 2003 outbreak, SARS-CoV.
“All 11 3CLpro sites are highly conserved or identical, inferring that their respective proteases have very similar specificities. The 3CLpro sequence of SARS-CoV-2 has only 12 out of 306 residues different from that of SARS-CoV (identity = 96%).”
In addition to mapping the in silico efficacy of FDA-approved prescription drugs, they also study other non-prescription purchaseable compounds from their 7000+ drug database, including those that are easily plant-derived. They use the binding energy as the figure of merit for the efficacy of a drug or compound to affect the 3CLpro protein. The logic goes, if you can bind more strongly to one of the virus’ protein sites, then you’re probably inhibiting the virus’ performance somewhat.
Since I could see there was already a lot of scholarly work happening on the approved prescription antivirals such as lopinivir/ritonavir and remdesivir, I decided to look up 3CLpro with regards to the non-prescription compounds that also ranked quite highly in the HK Polytechnic U researchers’ study -specifically hesperidin and diosmin. Hesperidin and diosmin are found prevalently in orange peels and orange juice, and hesperidin is also found prevalently in peppermint.
On performing the search, one of the earliest hits was this:
Lin, Tsai, Tsai et al. “Anti-SARS coronavirus 3C-like protease effects of Isatis indigotica root and plant-derived phenolic compounds”
which is a 2005 paper out of China Medical University in Taiwan covering in-vitro drug testing on SARS-nCoV using Vero cells.
This older paper performed in the wake of the original SARS epidemic describes how they tested a series of compounds against the 3CLpro protein in a type of standard animal cell test called “Vero” cells. (These are cells that come from African Green Monkeys, and have been used since the 1950’s as a standard cell test that researchers could easily replicate findings)
The 2005 China Medical University paper also tests compounds that the HK Polytechnic study didn’t even look at, like indigo and sinigrin. As the authors acknowledge, it turns out that hespiritin outperformed in inhibiting 3CLpro’s cleaving action (action that is necessary for the virus to propagate) at low concentrations of 8 micrograms per mL, which was a lower threshold of activity than the other compounds.
Finally, and usefully, the China Medical University researchers looked at what upper limit of quantity of the compounds under test would instead kill the test cells. Lots of compounds kill viruses in a petri dish, but you want to know which of those compounds doesn’t also kill the cells they are intended to protect. As they say after all, “the dose makes the poison.”
So hespiritin starts to look interesting, but so does sinigrin, because ostensibly it is so benign to the test cells that it is beyond the top of the researchers’ scale in terms of ability to harm the cells. Sinigrin is common in many green vegetables such as brussels sprouts and collard greens. It’s also what gives horseradish its unique test.
Also at this point, note the slightly different spelling hesperidin <-> hesperitin. This will become relevant later, but essentially hesperidin is the naturally-occurring compound, and hesperitin is what the human body metabolizes hesperidin into.
So I started to think about possible administration routes of these compounds. The two that stand out are 1) ingestion, because it’s the easiest and most popular form of taking medication, and 2) inhalation (since this manifests as a respiratory and pulmonary infection).
But then I felt stuck. Things are looking up, but I don’t know what to do next, or even if my approach to reviewing the research is valid. On the advice of a friend who has a PhD in Systems Biology and practicing in the pharmaceutical industry, he suggested I look up the safety of these administration methods with regards to these compounds.
After a lot of citing online research later, I learned that sinigrin all-too-easily becomes toxic at high ingested (and inhalation (!) ) doses! Anyone who’s ever taken a deep whiff of horseradish or sushi wasabi can speak to this. So this literature review on safety helpfully enabled ruling it out, leaving hesperidin as the only remaining candidate to show high in vitro efficacy at accessibly low concentrations.
Then we naturally want to learn how can this compound be had in therapeutically effective doses? For this I did a bit of math based on the China Medical University paper. These researchers identify the effective concentration for hesperitin to be 2.5 micrograms (mcg or μg) per mL (presumably aqueous). Taking simple orange juice as the most available vessel for this, and treating the human body as having the same density as water (1 kg per litre) and assuming uniform concentration of the ingested compound in the body, then a 100-kg person, (say a heavier male) would need 250000 μg, or 250 mg, of hesperitin maintained over some period relevant to the virus’ generation cycle.
According to this really helpful database of concentrations of various compounds in foods, we find that production orange juice contains 26 milligrams / 100ml of hesperidin. Hesperitin is about half the molar weight of hesperidin, so we need to treat this as an effective 13 milligrams / 100ml of hesperitin. So how much OJ does a 100-kg person have to retain without excreting any through urine, in order to reach the dose level indicated by the China Medical University study? The simple math answer is 1.9 litres. The average human’s weight, 62-kg, would need about 1.2 litres.
There are many questions remaining that, absent further expert advice, could affect the efficacy of orange juice on either being prophylactic to coronavirus infection, or winding down an existing infection. But considering that the safety of orange juice consumption in general is universally-accepted, then for the individual, the choice to drink orange juice is essentially a zero-cost option. It doesn’t pose a health-risk anywhere near these quantities (the equivalent of having passed phase-1 clinical trials for pharmaceutical drugs), its affordable, and in all likelihood is sitting inside your fridge as I write.
Don’t like drinking all that orange juice? (Or diabetic?) Well hesperidin (paired with diosmin, which is anyway also naturally prevalent in the same orange juice) is manufactured in pill form, with the hesperidin component typically dosed at 100 mg a day. Diosmin comes along with it at 900mg a day, and although it wasn’t studied in the China Medical University paper, the Hong Kong Polytechnic paper does show it having the same efficacy as hesperidin in the ‘in silico’ simulation.
In the US, while formulations containing these can’t be advertised for treatment of a disease, instead the FDA regards these constituent compounds as GRAS (“Generally Regarded As Safe”) via FDA’s published GRAS certifications on Orange Pomace and Orange Extract.
One note — no one should adopt a false sense of security over their health just because they’re drinking juice containing a compound that is still being investigated. All we know for certain right now is that it doesn’t hurt, and it *may* help. For reference, less than 1 drug or compound in a thousand succeeds from this stage to human effectiveness.
Next I would like to learn, which researchers are pursuing this line of inquiry? I can see a lot of work on the existing FDA-approved antivirals, but not on plant-derived compounds especially those that are considered safe through different ingestion routes at common doses. The motivation is more than just academic; when the right antivirals are determined and verified, it will take months for the manufacturers to ramp up production to the quantities required for a worldwide population of patients. Developing countries like Kenya where I reside will be later in line to receive those medications. But rural developing countries can grow their way to supply if an effective plant-derived compound can be identified.
Update 31 March -
So thanks to some tips from the HK Polytechnic researchers, I’ve learned that at least one other group of researchers from hard-hit Italy also made the connection to the in vitro efficacy quite some time before me. And on the (still quite slim!) chance that it makes it into in vivo trials, they’re preparing to bring online capital equipment to produce 36,000 doses a day from waste orange peels. So I had the chance to speak with them and it was great to see that initiative starting to be underway. But of perhaps greatest value to the drug design & pharamacology community is finding through these researchers that the researcher who has become most well-known for his coronavirus publications, Eric de Clercq, actually highlighted the role of hesperidin from the same China Medical University in vitro efficacy study back in 2006 after the first SARS coronavirus outbreak! He hasn’t written anything I can find on hesperidin since then — — is there any way that 2020 de Clerq can be re-introduced to 2006 de Clercq . . . ?