Skip to Main Content

In the midst of a pandemic like Covid-19, for which there are no FDA-approved drug treatments, hope is important. That’s one reason why remdesivir, an antiviral drug that Gilead Sciences originally made to fight Ebola, has been propelled into the spotlight with the hope that it can stop, or at least curtail, the ravages of SARS-CoV-2, the virus that causes Covid-19.

Data from the open-label SIMPLE trial, sponsored by Gilead, and the randomized controlled Adaptive Covid-19 Treatment Trial, sponsored by the National Institute of Allergy and Infectious Diseases, show that remdesivir may accelerate recovery rates among patients with advanced Covid-19. The drug’s modest effects are a far cry from the strong antiviral activity it demonstrated in preclinical primate models of coronavirus (both MERS and SARS-CoV-2). Yet that has been enough for the Food and Drug Administration to grant emergency use authorization for remdesivir and for the Japanese Ministry of Health, Labour, and Welfare to approve it for the treatment of Covid-19.

As chemists, we are troubled by the challenges to mass producing remdesivir. We aren’t alone. On the day that results from the two trials emerged, Gilead CEO Daniel O’Day praised the chemists behind the drug, saying he is “proud of the team because this is a complicated chemical process. It takes many, many steps.”


But does it really have to be that complicated? O’Day’s admission is interesting given that Gilead has another compound in its pipeline that is easier to make, has been shown to be effective against coronavirus in animal models, and is potentially as effective as remdesivir, if not more so.

Some background: Remdesivir works by interfering with the cellular machinery that allows viruses to replicate inside a human host. It is a pro-drug, meaning it must be metabolized and undergo a sequence of five bioactivation steps before it becomes GS-441524 triphosphate, the active compound that impedes viral replication.


Remdesivir isn’t Gilead’s only antiviral nucleoside analogue. The company has also developed GS-441524, another pro-drug that, as its name suggests, the body also converts into GS-441524 triphosphate, but in just in three steps. GS-441524 is easier to synthesize than remdesivir, requiring three steps instead of the seven needed for remdesivir.

Researchers initially thought that remdesivir would be activated more quickly than GS-441524 in human cells infected with the SARS and MERS coronaviruses. Yet data from primary human airway epithelial cells — one of the most clinically relevant cell-based models of the human lung — showed no statistically significant difference in potency between the two compounds. These data align with previous reports on the similar effectiveness of remdesivir and GS-441524 in coronavirus-infected cat cells. When GS-441524 was used to treat cats with feline infectious peritonitis, a progressive and usually fatal disease caused by a coronavirus, it displayed remarkable safety and therapeutic efficacy, with 96% of cats recovering after treatment.

Recent research in coronavirus-infected nonhuman primates demonstrated problems with remdesivir that inadvertently showed the antiviral effectiveness of GS-441524. In multiple studies testing remdesivir in coronavirus-infected mice or rhesus macaques, it was rapidly converted to GS-441524 in the bloodstream.

Take the latest controlled study conducted in rhesus macaques infected with SARS-CoV-2: After remdesivir was administered intravenously, GS-441524 was present in serum samples at concentrations 1,000-fold greater than remdesivir. Upon completion of the study, the researchers found that only GS-441524 — not remdesivir — was detected in the macaques’ lungs, yet they exhibited no signs of respiratory disease, significantly reduced viral loads, and a distinct reduction in damage to lung tissue. Such results reinforce those obtained from a prior study, also in macaques, and data from other species that GS-441524 exhibits strong antiviral activity.

Data in cats and primates have pointed to GS-441524’s safety. In the study using GS-441524 to treat feline coronavirus, the researchers noted its “impressive” safety profile when administered at high doses, and reported that no systemic signs of toxicity were observed over 12 to 30 weeks of treatment. In primates, GS-441524 was found to be present at high concentrations in the blood (1,000-times higher than remdesivir) with no apparent adverse effects.

The first step in the bioactivation of GS-441524 is the rate-limiting step, something that remdesivir was designed to avoid. But that doesn’t matter clinically because of remdesivir’s rapid transformation to GS-441524 in the bloodstream.

Remdesivir’s lackluster results in patients with advanced Covid-19 in the NIAID-sponsored trial and the finding that it provided no statistically significant benefit in a clinical trial conducted in China among patients with severe Covid-19 symptoms are likely due to the suboptimal level of active GS-441524 triphosphate in the lungs. Patients with advanced or severe Covid-19 generally have a high viral load in their lungs and would need a high concentration of GS-441524 triphosphate to combat it. The benefit of using GS-441524 over remdesivir is that GS-441524 can almost certainly be given at much higher doses due to its lower toxicity. This would result in more conversion to the active compound, GS-441524 triphosphate, in the lungs.

When viewed through a different lens, the initial results from the NIAID-sponsored trial are more encouraging than they would seem. The active agent, GS-441524 triphosphate, clearly exerts antiviral activity against SARS-CoV-2 in humans, as supported by the accelerated recovery rates in advanced Covid-19 patients enrolled in the trial. Our analysis of preclinical and clinical trial data strongly suggests that early and direct administration of GS-441524 should be considered as a synthetically simpler and potentially more effective alternative to remdesivir, especially as GS-441524’s remarkable safety would enable higher dosing.

We see numerous advantages to using GS-441524 rather than remdesivir as an anti-Covid-19 therapy. GS-441524 is easier to synthesize and dissolves in water, which can speed manufacturing and enable higher dosing. It is a smaller molecule than remdesivir, which would make it easier to produce an aerosolized formulation for inhalable therapeutic and prophylactic treatment — this would be particularly attractive for achieving a high concentration of the drug in lung cells while minimizing systemic toxicity or side effects. And it is also less toxic than remdesivir. For these reasons, we do not see the point of making a significantly more complex drug like remdesivir when what actually reaches infected lungs is GS-441524.

The attractive profile of GS-441524 from both manufacturing and clinical perspectives raises this question: Why hasn’t Gilead opted to advance this compound to the clinic? We would be remiss for not mentioning patents, and thus profits. The first patent on GS-441524 was issued in 2009, while the first patent for remdesivir was issued in 2017.

We aren’t the only ones questioning Gilead’s strategy. We have spoken with a number of chemists, biochemists, veterinarians, and others who are also surprised that GS-441524 has remained out of the spotlight. Veterinarians we spoke to have noted that the strong antiviral activity of GS-441524 has resulted in a “miraculous turn of events” for cats infected with feline coronavirus, which was once considered a death sentence.

Given GS-441524’s optimal properties, we — along with the millions of people awaiting an effective treatment for Covid-19 — are left to wonder why Gilead isn’t giving it the same attention it is giving remdesivir. The world can only hope it isn’t for the sake of protecting its intellectual property.

Victoria C. Yan is a graduate research assistant specializing in phosphonate chemistry at the University of Texas MD Anderson Cancer Center in Houston. Florian L. Muller is an assistant professor specializing in cancer drug development in MD Anderson’s Department of Cancer Systems Imaging.

  • On theoretical grounds, all this makes great sense.

    However, on practical grounds, switching horses in mid-race would be unworkable. Remdesivir already has an extensive use record in humans, and already has an existing emergency use authorization.

    Getting a new drug approved for any initial human use is a long and daunting process. We have a very fast-moving pandemic. We don’t have many months available to roll out a new agent that’s never been approved for human use. It would come too late to help very many people.

    In comparison to the above obstacle, a somewhat more complex chemical synthesis is a trivial issue. Most compounds can be prepared in large batches, and parallel batches are not complicated to implement.

    At this late stage, trying to jump horses in mid-race, just to produce a highly similar agent probably makes little sense, either for Gilead or the rest of the world.

    • The article makes sense in theory. But I completely agree with Steve. The points made are irrelevant currently. In a fast moving pandemic, Remdesivir, which has gone through all phase trials is the only “fast” answer.

      Pure scientific perspective is only useful when there is enough time on hand. Translating a new drug through all trials take time, time which we don’t have. Even assuming that Remdesivir is in theory “lesser” effective, it would stand fastest chance to be approved. We have months, not years to get something which works, even if lesser effective.

      If COVID-19 becomes endemic, I am sure we will see other avenues explored.

    • Remdesivir is an immediate triage to this problem. Multiple studies have documented the strong antiviral effects of early/prophylactic treatment (eg: PrEP for HIV). With some of the properties I mentioned in the article, GS-441524 would seem to fit the bill for an aerosolized prophylactic/treatment. The concern here is that Gilead will focus on remdesivir to the exclusion of GS-441524.

      I mentioned to some other commenters that there is precedence in FDA history of fast-tracking approval of prodrug/drug pairs (see: leflunomide/terifluonomide). Both remdesivir and GS-441524 produce the same active metabolite (GS-441524 triphosphate). GS-441524, while it has not officially gone through phase 1 is thus not quite a “new” compound. A similar fast-track approach for this pair could be looked into as well. It’s something for Gilead to consider.

  • From what I understand, the extra structure on the 5′-carbon of Remdesivir is a quite successful strategy that allows the drug to be more easily absorbed through the intestinal wall, as well as into infected cells. This is called ProTide technology, and has been applied to many other antiviral drugs available on the market. The issue with having a 5′-OH group on the drug is that it is too polar to easily cross cell walls, and therefore relies on transporters to get into infected cells, rather than diffusion. All of these factors reduce its concentration within the infected cells. Also, once Remdesivir makes it into the cell, it undergoes a different metabolic pathway than GS-441524, which omits that rate-limiting monophosphorylation step you mentioned, so it can act a lot quicker. These, I believe, are the reasons why Remdesivir, although more synthetically challenging, is used rather than its precursor. On the other hand, I do agree that GS-441524 could be a good candidate for aerosols, although, again, its lack of lipophilicity could be an issue for it getting into the cells.

    • Yup, ProTide pro-drugs were developed by a talented chemist (Chris McGuigan) to mask the anionic phosphate and enhance cell permeability. ProTides also help nucleotide analogues bypass the first (rate-limiting) phosphorylation step. All of this is contingent upon remdesivir maintaining its ProTide pro-drug moieties when it reaches lung cells (in the case of SARS-CoV-2).

      However, multiple studies (Warren et al. Nature, 2016; Sheahan et al. Sci. Transl. Med, 2017; Williamson et al. bioRxiv 2020) have shown that the intact ProTide is rapidly hydrolyzed in serum (t1/2 ~0.4 h). I mentioned to another commenter below why remdesivir may have been pursued instead of GS-441524. To sum that up: it does, relate to faster bioactivation (in vitro) as you mentioned. However, bear in mind that almost all of these comparisons were conducted in cell culture models, where factors such as serum hydrolases, circulation, etc. are a non-factor. Whatever potency differences are observed between remdesivir and GS-441524 are ultimately irrelevant in vivo if only GS-441524 reaches the lungs.

      As for GS-441524 getting into cells: nucleosides can permeate cell membranes either freely or via nucleoside transporters. Then greater hydrophilicity of GS-441524 would be ideal for aerosolization, which would directly administer it to the lungs.

  • First, Just because a drug works well in cats and primates, does not always translate to humans. GS-441524 seems like it may be a more effective prophylactic inhaler, possibly combined with lerimolab. Also GS-441524 may not work as well or be tolerated in combinatory scenarios, i.e. it might not play well with others. Market dynamics typically favor cheaper more easily produced compounds, so there may be a reason the extra steps were included to produce RMD. Right now we are dealing with a new thing and every Pharma company is throwing everything they can at it to make a therapy vaccine or cure. GLD would be foolish to pass up an opportunity unless it has a reason.

    • As implied in the writer’s article: the patent for GS-441524 is from 2009, while remdesiver is 2017. That means there is more money to be made by Gilead with remdesivir. I am mesmerized by the content in this article, and am quite certain that many Covid-19 patients would eagerly offer to have GS-441524 used and tested on them. The door to that opportunity shoud be open in this pandemic crisis time. And if I get Covid, then count me in.

    • Agreed that efficacy in cats and primates does not necessarily mean that the same effects will be observed in humans. Spot on with GS-441524 being ideal for inhalable prophylaxis—this is my view as well, since it is considerably less hydrophobic than remdesivir, among other things.

      As for how complicating a synthesis plays in Gilead’s favor: only they can answer that. But with regards to efficacy of remdesivir vs. GS-441524, I return back to your first statement. True, animal models provide a good—but not exact—approximation of the human situation. However, I do believe that animal models are much better approximations of how a drug will behave compared to what occurs in cell culture based models. This is especially true for pro-drugs like remdesivir, which tow a fine line between intracellular and extracellular metabolism. Having spoken to a few biologists on the matter, it seems to me that a major reason why remdesivir was pursued instead of GS-441524 is because remdesivir outperformed GS-441524 in several (but not all) cell-based models (with various viruses). This, of course, is expected—given that permeation of the drug occurs much more quickly and does not have to deal with hydrolyses in serum, etc. That being said—especially for pro-drugs—I believe in vivo data are far more informative than the in vitro situation. And if GS-441524 is the predominant species that reaches the lungs in vivo, then this nullifies any differences in potency that would have otherwise been observed in vitro.

  • This article contains some erroneous assumptions. Most notably, the idea that it was “rapidly converted to GS-441524 in the bloodstream.” True, the cited Nature paper shows blood levels of the nucleotide exceed those of the pro-drug. However, the fraction of the total drug in the blood relative to the fraction in the tissues was not determined. As long as the rate of distribution exceeds the rate of metabolism, tissue accumulation will occur. In fact, the Nature paper does show that the pro-drug drives high concentration of the triphosphate in blood cells. It may be doing the same in other tissues like lung cells. However, since none of these studies directly compared the pro-drug to the free nucleotide at equivalent doses, one cannot draw any conclusions about their relative ability to drive formation of the triphosphate in lung cells, though it remains likely the pro-drug is superior.

    • See Supplementary Figure S1 in Williamson et al. bioRxiv 2020. Note the values on the y-axis (GS-5734 = remdesivir, GS-704277 = L-Ala intermediate metabolite). The paper states that GS-5734 had a “short systemic half life (0.39 h) resulting in transient conversion to the intermediate GS-704277 and persistence of the downstream GS-441524 product at higher plasma levels.”

      The authors then explicitly state that “GS-704277 was not detected in the lung tissue.” From Supplementary Figure S1a, you can see that the detection limit was already quite sensitive (low nM). The fact that they could not even detect the intermediate metabolite in the LUNGS means that intact remdesivir was absent. The Nature paper was looking at triphosphate formation in blood cells but for SARS-CoV-2, the critical organ is the lungs (specifically type II alveolar cells in advanced cases).

      I agree with that the pro-drug WOULD BE superior to GS-441524 for generating triphosphate in the lungs because remdesivir is designed to bypass the rate-limiting first phosphorylation step. HOWEVER, because remdesivir does not actually reach the lungs, this becomes a moot point.

    • Two points-
      1. The L-ala intermediate is just that, an intermediate. Just because this is not detectable in the lung does not mean that the triphosphate is not present. Especially after 12 hours the intermediate would not be expected to be present.
      2. The very title of the section you quote is “Remdesivir is distributed to the main target tissue of SARS-CoV-2, the lungs”

    • Remdesivir is a pro-drug. The sequence of bioactivation is as follows: Remdesivir (GS-5734) —> GS-704277 (L-Ala intermediate) —> GS-441524. If there is no (or very, very little) L-Ala intermediate present, then there is no remdesivir present in the lungs.

      I am not saying that there is not triphosphate in the lungs. In fact, there almost certainly is triphosphate in the lungs due to the antiviral effects observed. The title of the section is, in my view, incorrectly worded—especially when the authors explicitly state “Serum levels of the prodrug and the downstream metabolites were consistent with previously published plasma levels of these compounds in healthy rhesus macaques, which showed a short systemic half-life for GS-5734 (0.39 hrs) resulting in transient conversion to the intermediate GS-704277 and persistence of the downstream GS-441524 product at higher plasma levels.” Hopefully this will get corrected in the peer-reviewed final version of the manuscript. Anyway, given the sequence of bioactivation steps, saying that remdesivir is in the lungs is entirely incompatible with their finding that “GS-704277 was not detected in the lung tissue.”

  • You are correct that the SD are large, but visual inspection of the box and whisker plots suggests a meaningful difference in the effectiveness of the two drugs.

    Your second statement also concedes that remdesivir is more potent than GS-441524. Since remdesivir is metabolized to GS-441524, then the higher hydrophobicity of remdesivir must have improved absorption to the active site, inside the cells.

    I do agree that someone should be following up with tests of the in vivo potency of GS-441524 against SARS-CoV-2. Rhesus monkey trails could easily and quickly be done.

    I have seen evaluations that the cost of producing a 10-day treatment course of remdesivir would cost around $10. Thus the manufacturing cost should not be a problem. The speed with which rendesivir can be produced vs GS-441524 would seem to be a more interesting question. How much quicker can GS-441524 be produced?

    In you column you state that GS-441524 is safer and less toxic than remdesivir. I did not see any evidence of that in the articles that you cited. You state that GS-441524 has remarkable safety record, but the Cho et al. 2012 paper that this statement is linked to is not about safety of the drugs. I did not see anything in that paper about safety. It does seem that GS-441524 has a good safety record on the black market with cats, but where is the human data?

  • What is the dosing per kg of the drug used to treat feline coronavirus relative to the remdivisir dosing in people? Could people tolerate the same dosing of the feline coronavirus drug? Also what is the in vitro effectiveness of the drug for covid-19 relative to it’s effectiveness for feline coronavirus?

  • This is a bad piece of opinion writing. The paper cited (Agostini et al. 2018) in this opinion piece show that the EC50 of GS-441524 was 10 times higher than for GS-5734 in human airway epithelial cells against both SARS-CoV and MERS-CoV. Agostini et al. also states that there was no measurable cellular toxicity for either compound.

    • That is partially correct. If you look at the standard deviations (Table 1, Figure 2a), you will see that the reported EC50 is clearly not statistically significant; the +/- values are huge.

      Also consider the bigger PK picture: it does not seem to matter that remdesivir is more potent than GS-441524 in vitro if the predominant species in the lungs in vivo is GS-441524.

  • The reason is just because of the speed they can get it to market. Remdesivir had already gone through Phase 1 and 2 trials for Ebola. So, they just needed a positive phase 3 trial to have it on the market. This new compound would have to start at phase 1, and there are a lot of drugs starting at phase 1. The market for this drug started 2 months ago.

    Think about AZT. It hardly helped with HIV back in the 90’s, but it did show some antiviral activity, and derivatives of it were used for some time after that (might be still used today, I honestly don’t know). While it sucked as a drug, it sold very well because it was the only thing they had. The same applies here. Remdesivir can be sold now in two countries now, and it will be starting next month when they start charging for it. This compound you speak of would be starting at phase 1.

    It doesn’t even make sense for the company to invest in the compound because, by the time it made it to the market, it would have to compete with a lot of other drugs that will likely be out by then (perhaps Avigan?) and maybe even a vaccine.

    • Fair. But a couple of things.

      First, remdesivir is a pro-drug of GS-441524, which itself is a pro-drug (they both give the active triphosphate species). There is precedence in FDA history for accelerating approval for pro-drug/drug pairs (see: lefluonomide/terifluonomide). Seeing that things seem to be moving a lot more quickly in the drug development world these days, it does seem farfetched that the FDA would do something similar here as well.

      Second, I think it’s worth noting the synthetic complexity and challenge of mass producing remdesivir–something that Gilead has been struggling with. There is strong pre-clinical and clinical evidence that early intervention is much more effective than advanced stage treatment (see: negative Phase III Chinese clinical trial with remdesivir; Wang et al. Lancet 2020).

      In their press release a couple weeks ago, they mentioned that they were interested in developing an inhalable anti-Covid-19 prophylactic/therapeutic. GS-441524 would seem to fit the role here, given its synthetic simplicity and improved water solubility (easier to mass produce and aerosolize).

    • I am wondering, doesn’t the fact the remdesivir is converted to the GF441524 mean the GS441524 is not likely to be MORE toxic than the remdesivir? In this emergency, wouldn’t that be a good theoretical basis for foregoing the Phases 1, 2, 3 and approving the drug for some patients?

Comments are closed.