Skip to main content

Academia (vs. Industry)

Targets Versus Drugs

There was a comment on the blog the other day about how there are people in academia who feel that the discovery of a new target or pathway is basically finding a new drug, and that the rest is “technicalities”. I’ve encountered that view of the world before (Donald Light/Rebecca WarburtonMarcia Angell, and similarly Arnold Relman), so it’s not just some aberration, but it’s just amazingly wrong. People keep pointing this out, including people who’ve worked in the industry and people who haven’t, but to no avail.

So here’s an example that illustrates the difference. If you pick one of the most famous drugs in the world (aspirin, first reported in 1899), it can come as a surprise to people outside the field how long it took to find out how it worked. It wasn’t until 1971 that John Vane at the University of London figured out that it (and the other non-steroidal anti-inflammatory drugs) worked by inhibiting the cyclooxygenase enzyme (COX), which is responsible for producing the signaling molecules (prostaglandins) that produce the downstream effects. Aspirin does other things as well, but that’s the big mechanism that everyone had been searching for. During the 1980s, extensive work on this enzyme and its genetic background led to the discovery in 1991 (at Brigham Young, by Dan Simmons and his group) of a second subtype of the enzyme (COX-2).

And that set off quite a chase. Because it looked like if you could inhibit COX-2 and not COX-1, you might be able to get the pain-relieving antiinflammatory effects of aspirin without the gastrointestinal side effects of bleeding, irritation, etc. The idea of an “aspirin 2.0” was immensely appealing, and a great deal of work went into finding such compounds. In 1992, a team from the University of Rochester filed a patent application for an assay to distinguish whether new compounds were binding selectively to COX-2, and this eventually issued as US5,837,479 in 1998. A division of that also issued in 2000 as US6,048,850, the “850 patent”, as it came to be referred to. Then the fireworks display really started.

Immediately after being granted that patent (that very day, actually), the University of Rochester filed suit against Pfizer for marketing two COX-2 inhibitors, which Rochester claimed violated their patent. The G. D. Searle company (part of Monsanto) had discovered Celebrex (celecoxib) and the related drug Bextra (valdecoxib) and Pfizer had taken them over largely to get the rights to the drugs. You’ll also remember Vioxx (rofecoxib), another drug in this class from Merck, who were not party to this case, and these companies were by no means the only ones who had been working on selective COX-2 chemical matter. This wasn’t the only lawsuit, either. Brigham Young, mentioned above, had licensed the Simmons discovery to Monsanto, whose drug research arm was also taken over by Pfizer, and had sued Pfizer over royalties (eventually settling with the company for a payment of $450 million).

Pfizer’s response to the Rochester suit was to ask for summary judgment – basically, saying that the facts of the law were so plain that there was no use in even going to trial. Issued patents start off as presumed valid, of course, but that motion was granted by the District Court of the Western District of New York, whereupon Rochester immediately appealed that decision. It ended up, as these cases do, at the Court of Appeals for the Federal Circuit, which upheld the summary judgment in 2004. And here’s where we talk about targets versus drugs.

The ‘850 patent had several claims, with several key ones coming down to a method of treating human patients with a compound that was a selective inhibitor of COX-2. That’s what Rochester claimed that Pfizer was infringing, by selling such a compound to human patients. Pfizer, meanwhile, claimed that the patent was invalid from the start, because it didn’t meet some key requirements. A patent has to have a complete written description of the invention, and Rochester felt that they’d done that – but Pfizer held that you can’t just say you own the method of treating someone by giving a selective compound without describing what that selective compound is – that description is inadequate. A written description has to be in enough clear and complete detail that another person “skilled in the art” can reproduce the invention, you have to go on to describe how the invention is used, and you also have to have disclosed the “best mode” of doing so.

In other words, Rochester was claiming that the important parts were the discovery of COX-2 and especially the discovery of the assay for selective COX-2 compounds. The compounds themselves? A mere technicality – anyone of ordinary skill in the art can come up with compounds and drugs. The Court of Appeals disagreed. The pointed out that those three requirements just mentioned have been found (over and over) to be independent of each other. If you want to claim a chemical compound (a drug, in this case), you have to show what it is and how to make it and use it, not just specify what it does. This means that you can’t say that your invention is giving someone such a drug and then claim every drug of that kind that someone else finds. The court noted that nothing in Rochester’s patent indicated that they had any such COX-2 compounds or knew what they might be. As the court said:

Even with the three-dimensional structures of enzymes such as COX-1 and COX-2 in hand, it may even now not be within the ordinary skill in the art to predict what compounds might bind to and inhibit them, let alone have been within the purview of one of ordinary skill in the art in the 1993-1995 period in which the applications that led to the ′850 patent were filed. Rochester and its experts do not offer any persuasive evidence to the contrary. 

And although this was not an issue in this case, I would also like add that even having a compound in hand from such a selectivity assay does not necessarily give you a drug. In order to be useful, a drug has to be able to be dosed in a human patient, last long enough to have a beneficial effect, and not by itself (or through its breakdown products) cause so many harmful side effects that it’s not worth taking. It also has to be produced on large scale, under very tight tolerances, and be stable enough so that it can be stored and shipped without alteration. A rather large amount of time, effort, and money goes into figuring out these things.

But the main point is that a target, a mechanism is most certainly not a drug. This is true scientifically, and thanks to Univ. of Rochester v. Searle, it’s true legally as well.

Postscript: of course, the history of the COX-2 inhibitors was rocky. Merck’s Vioxx was taken off the market due to cardiovascular side effects, followed by Bextra, and Pfizer ended up paying a $2.3 billion dollar fine for its marketing of the latter. Celebrex is still sold, however – extensive reviews of its effects in human patients have not shown excess cardiac liabilities. The University of Rochester did not, as far as I know, offer to help share in the losses incurred by the withdrawals.



47 comments on “Targets Versus Drugs”

  1. Magrinho says:

    And this is a rare case for which the prophetic claims (U-Rochester) were sort of true.

    There are multitudes of patent applications claiming “a selective inhibitor of X will ameliorate disease state Y” that end up dead wrong. The selective inhibitor of X is made after much effort and does nothing useful in animal models or human trials for disease state Y.

    I guess we have to keep beating this dead horse – drug discovery is full of frightening (and $$$) leaps of faith.

  2. Mad Chemist says:

    I’d never heard that story of U. Rochester. The level of hubris involved on their part is staggering.

    1. Kevin Pels says:

      If you think that’s crazy, you won’t believe how litigious literally any pharma’s legal team is in defense of its market rights.

  3. luysii says:

    Remember aspirin is ACETYL salicylic acid, an acetylating agent, as its mechanism of enzyme inhibition. What else is acetylated in the cell? Just histones, among other things, the acetylation of which changes the expression of all our genes.

    Years ago in practice, it drove me nuts to see people claiming to know just how much aspirin to give to prevent stroke and heart attack, based on studies of COX inhibition with such and such a dose. It was the mechanistic tail wagging the therapeutic dog. There is probably much more data now, but I went with the observational study that showed the greatest stroke reduction risk wasn’t baby aspirin, but 2 adult aspirins twice a day (1,300 milligrams).

    I’d point out that coronary arteries were quite muscular (they have to be to absorb the motion they are subject to as the heart contracts), while brain vessels were not — a nondiseased basilar artery is nearly translucent. All to no avail. But it’s no longer my problem, but yours and your parents.

    1. gator says:

      That’s a heck of a dose difference–weren’t you worried about bleeding events?

      1. luysii says:

        4 adult aspirins a day isn’t that much of a dose. Arthritics take much more for years without getting into trouble. Some people are sensitive to it, but bleeding complications are not common. Of course my mother had one such, but I’ve been taking this dose for decades (a case of a doctor taking his own medicine) without any problem. No one exsanguinates as a complication and it’s easily managed.

        One problem was getting people to take it. “I spent all this money to come see you, and all you do is say take aspirin? , etc. etc.”

  4. Anon says:

    I think the issuance of the Rochester patent (‘850) started the USPTO’s “quality” initiatives. I also don’t think the Primary Examiner who issued the ‘850 patent ever suffered any consequences.

    1. Chrispy says:

      It seems like the USPTO keeps a pretty low bar for issuing patents, with the understanding that people will hash it out in the courts. There is all kinds of patented nonsense.

      1. Some idiot says:

        On the other hand, I have also had a deal of interaction with the corresponding assessors from the Chinese PO, and (generally speaking) they are very, very good and on the ball.

        1. ein says:

          Interesting. If you want to characterize the behaviors of various national patent offices I’d be fascinated.

  5. 10 Fingers says:

    One interesting big exception to the idea that targets do not equal drugs: biologics.

    We now live in a time where it is possible (assuming the target has the right properties as a protein) to lay out, in detail, all of the steps required for a practitioner of the art to create an antibody to said target – prophetically.

    It is a different world of IP for large and small molecules, in this respect.

    1. JIA says:

      I disagree. While a discovery of an antibody target perhaps gets you closer to a drug than discovery of a small-molecule target, it does NOT get you all the way. Starting from the top — There are many many epitopes on a typical extracelluar protein receptor — where do you want your antibody to bind? It matters. I worked on a series of antibodies to one tyrosine kinase receptor where half the candidates caused the receptor to be “locked” on the cell surface and stop internalizing, while the other half drove internalization. Which type do you want? Or check out recent reviews about Type I binders (rituximab, a best selling drug, now one of the first to face generic competition) vs Type II binders (obinutuzumab) against CD20 on B cell lymphomas. They have totally different effects on cross-linking the receptor (within vs between tetramers). And I haven’t even touched on half life and other pharmacology, effector Fc properties, manufacturability, or anything else necessary to get into humans and then to approval.

      You assert that for monoclonal antibodies, target = drug automatically. But it’s just not true.

      1. BioObserver says:

        Not to mention, even if you know what epitope you want to hit, no one is able to then sit down with pad and paper and draw up the amino acid sequence that will do so. Trial and error arrives at a desired compound.

        1. Duncan Bayne says:

          > Not to mention, even if you know what epitope you want to hit, no one is able to then sit down with pad and paper and draw up the amino acid sequence that will do so.

          As someone with only a layperson’s interest in biology – why is that so? I mean, not by hand, but by simulation? Model the epitope, then test simulated amino acids against it.

          I suspect the answer _may_ be that it’s not _just_ the epitope, you have to consider the behaviour of the entire cell, or possibly organism?

          1. Chrispy says:

            There may come a day when this is possible, and certainly David Baker’s group is getting closer and closer to engineered binders. But for now there are just too many computational loose ends.

      2. 10 Fingers says:

        In retrospect, I think my point was poorly stated and unclear. I do not disagree with your comments, however, they are not directly on the point I was making – which was largely around IP.

        In anticipation of an antibody against a specific target, it is possible to get broad claims applicable to the use of said antibody therapeutically (the kind that, in the COX2 example described above one cannot get for a small molecule generically, and because enablement of an antibody is considered straightforward to a “practitioner of the art”). It is still possible to patent a specific antibody to the target (particularly with novel and unexpected properties – and that’s not always a low bar, particularly if two antibodies bind to the “same epitope” from a patent examiner’s perspective), but in some cases the *use* of that antibody may be constrained by claims granted prior to the time that *any* antibody existed.

        It is a long (perhaps worthwhile) discussion around why this is the case, but my point is that the IP world is quite different in this regard for large and small molecules. It is an oversimplification, but I think fair to say, that is because the US PTO considers the process of raising an antibody to an antigen, screening it for appropriate properties and turning it into a humanized therapeutic ready for dosing in disease X to be a largely mechanically enabled process (the method of the “actual” dosing may be its own subsequent creative invention).

        Since most of the low hanging fruit has been harvested in the biologics space, the details for antibody targeting and profiles matter more and more. But this difference (“in favor” of large molecule “forward patenting”) was compelling enough that companies like Genentech once devoted huge amounts of their Research budget to trying to “high-throughput patent” antibodies targeting the entire secreted human proteome.

        1. JIA says:

          @10Fingers – Thanks for a thoughtful and informative reply!

  6. Passerby says:

    One point that gets left out from this whole discussion regarding whether academia should share the profits that industry makes on drug is that of academic freedom. Most of the basic biomedical discoveries about targets, mechanisms and pathways from academia have come about because academic scientists were free to work on pretty much whatever they wanted (this was much more so before, not so much now). Industrial scientists are almost completely at the mercy of the dictates of upper management.

    I understand the need to ask whether industry owes something to academia for laying the groundwork, but this discussion should be seen through the light of something that academia has that industry will never have – the freedom to pursue their ideas wherever they may lead.

    1. zero says:

      Academia is funded largely by grants and other federal funds.
      Those funds are derived from tax revenue*.
      Industry pays taxes**.
      There is an existing mechanism for industry to support blue-sky research. We just need to make sure it stays viable, which at this point means way more federal investment in research.

      * This is a dangerous simplification, but it works well in short ‘sound byte’ conversations. The US dollar is a fiat currency; taxes are largely a convenient way to control the supply of money rather than a real necessity. US politics hasn’t caught up with this yet, so we still talk about paying for things as if the federal government were a household. Hence, ‘taxes pay for research’.

      ** [citation needed], more or less. Major international entities are quite skilled at minimizing their tax liabilities. The US is a special case as the sovereign issuer of dollars, but other nations competing for dollars have strong incentives to race to the bottom and capture multinational income.

    2. anon says:

      the freedom of research is no longer available in academia. Most of people are simply clinging to what they had and become a slave of the publishing industry.

      1. T says:

        I think you mean a slave to the funders and their lazy misuse of journal impact factors to judge the quality of a published work. Which is nonsense on so many levels.

        For example, the journal IF is an average across for the journal. It says nothing about an individual paper (most of which are cited fewer times that the IF). So even if you assume that citation of an article is an indication of impact (which is problematic in itself), using the journal IF is ridiculous. If you have a paper in a journal with IF 20 that is cited 5 times in 3 years (by the authors themselves) 4 years later, and another in a journal with IF 2, that has been cited 50 times in 3 years, which would you say has had the most impact on the field?

        This causes publishers as many headaches as researchers, e.g., having to waste large amounts of time and money chasing impact factors because that is what their customers (researchers) demand and look for as one of the key determinants of where they publish.

    3. pv=nrt says:

      I’m in industry, and I have not had much trouble at all working on what I want to work on that will lead to a drug. You have to justify it to upper management, but as long as it will lead to a drug, they’ll love to hear about it and will enthusiastically say yes. Even new people with great ideas get an ear as long as the ideas are great.

      This is not what I hear at all from academic buddies and collaborators. Even a big shot MD dept head who have a talk here was complaining that he couldn’t get more than 1 RO1, and one isn’t enough to do much. Furthermore, he had a great idea for his pet target to be used in a different NIH section, and couldn’t get funding because he wasn’t well known among people in the other study section. We have also had academics apply for jobs here saying that they can’t get a second grant approved and they felt locked in to this project they’ve been working on since their post doc.

      1. anon says:

        NIH requires all perfect grant applications ( a.k.a preliminary results). High impact journals require a perfect story ( the supplementary is often longer than the article itself). The authors list becomes longer and longer each year. All consumes a lot resource to labor for the benefit of the publishers.

  7. tlp says:

    For similar reasons I think grad students and postdocs should be on academic patents if they performed the experiment. “Go do the experiment” rarely works by itself. It’s resolving those pesky ‘technicalities’ overlooked by big picture view that completes a valid invention. And two different scientists are unlikely to end up with the same outcome when trying to discover something new.

    1. Anon says:

      Inventorship is a complex legal question. It is not assigned willy nilly and if not done correctly can invalidate the issued patent.

  8. georgey tenety says:

    send this blog description to congress so they can learn

  9. anon says:

    Here’s to the great John Talley. Drink up buddy….

  10. cb says:

    As we know already for ages the RAS family of proteins is among the most frequently mutated in human cancer. There are many crystal structures available e.g. co-crystallized with GDP/GTP analogues and assume KRAS is a validated target. Why are medicinal chemists not applying their “technicalities” to make a GDP/GTP competitor drug: should be easy starting with a fragment based approach, screening 10M pharma compounds, or 100M DNA encoded library in combination with structure based design and AI and a self learning machine at Google of course; are medicinal chemists opportunustic people who go for the ATP binding site of kinases or are they cowards that fear the strong binding of GDP/GTP with KRAS . Come on, show the world how easy it is to find a drug if you have a validated target.

    1. Wavefunction says:

      Your comment was presumably intended to indicate the ‘undruggability’ of RAS with which I agree, but given recent advances in the field, I am placing my bets on some kind of RAS inhibitor (covalent, fragment-based, DEL-based) to emerge in the next five years or at least a decade. This inhibitor may not be ideal, it may not likely be oral, and its best profile will likely be as a combination therapy, but it will be at least something.

    2. Anonymous says:

      RAS: druggable or not. — Can’t access lit, but it is my understanding that there are plenty of potent RAS inhibitors. One big problem is the difference between in vitro and in vivo. In the cases I’m thinking of, RAS was inhibited but there are so many compensatory pathways that cell growth / proliferation is not inhibited. Until I update my reading on the subject and learn otherwise, I don’t think RAS is a good drug target for that and other reasons. Maybe a RAS inhibitor would be helpful in some sort of multi-drug (multi-target) therapy.

  11. somebody says:

    I think this is a great illustration of an issue I’ve seen with large project teams with many different areas of expertise. Everyone assumes everyone else’s job is straightforward. Everyone in the business agrees that biology is where the science-shattering innovation happens, but, like Lowe points, everything else is hard too. Biologists think if they come up with the target and assay then making a drug is just a matter of time. Medicinal chemists are convinced they make the drug, and any drug they make will be easy enough for the process chemists to make on scale and formulation will definitely figure out a way to formulate it. Process chemists are convinced they are charged with saving the company lots of money, but they think the jobs of the analytical chemists and scientists in commercial are no problem at all. All of these issues are solved when scientists actually talk to each other, so you can see how separating people into silos exacerbates this problem. Silo-ing is a classic symptom of academia.

  12. Scott says:

    Ah, yes, Vioxx… I was given that shortly after I broke my back (L1 burst fracture, no spinal damage). It *really* worked well on keeping the pain down to a reasonable level (I’m talking 2-3 on the 0-10 scale, not the 6-8 I’m lurking on with the current mix of acetaminophen, tramadol, meloxicam, etc.). Horrible shame about the heart issues.

    But yeah, just finding a potential method of action doesn’t count as discovering the drug. You gotta get it into the patient!

    1. Vader says:

      It’s a shame Merck and the current liability climate couldn’t give you the option to risk the cardiovascular problems to remain functional. You may be in a lot of pain most of the time, but at least you”ll get to suffer for a long time!

  13. Barry says:

    I was startled to learn that some academic biologists don’t know the difference between a
    “biomarker” and a “drug target”. But they know that “drug targets” get press and funding. So everything is billed as a “target”

  14. Andre Brandli says:

    Derek, thanks for sharing this story behind the 850 patents. It’s an excellent case study to be thought to all students studying medicinal chemistry, pharmaceutical sciences or patent law. It highlights that patents describing composition of matter are the really valuable ones!

  15. Isidore says:

    This is tangentially related, are chemistry or other science graduate students have access to training, whether formally through classes or informally through seminars and such, on patents and patent law? Some companies, at least ones where I have worked, offered such training.

    1. Anon says:

      There’s no formal or even informal training in graduate school on IP law. Just the opposite, a disdain for it. You see it in the disparaging of the experimental section of patents and the ready trading of PDF’s of copyrighted books.

      1. Me says:

        In my graduate training in the late ’90’s it was a part of the “Business of Science” course required by my program.

    2. Jamil says:

      Yes, in UK universities this training is compulsory and we are encouraged to liase with the department that sets up spin off companies while completing a PhD.

  16. GATC says:

    And int he absence of knowledge how quickly they pull the demo card (in this case gender):

    1. b says:

      “Citing the exchange, another industry expert, Dr. Derek Lowe, who has worked for several major pharmaceutical companies since 1989, took aim at Ocasio-Cortez in a commentary piece, calling the congresswoman’s knowledge of Big Parma ‘inadequate.'”

      Derek, you occasionally include recipes on here, but I didn’t know you took to criticizing other’s cooking publicly.

      Mmm… Big Parma. I’m hungry.

      1. Kent G. Budge says:

        Is it good with Russian sauerkraut?

        I still make that recipe from time to time, using Derek’s instructions. Fermented food good.

  17. Robert says:

    Derek — I see you were quoted in an article in the business section of today’s NY Times (link: The article seems relevant to the discussion in this issue of your blog on the general topic of targets vs. drugs (or, related, ligands vs. drugs). The Times article links to a blog post by Mohammed AlQuarishi, identified as a biologist at Harvard Medical School ( who has some interesting comments on, among other things, the level of research in the pharmaceutical industry. One of his comments about pharma executives: “… they’re clueless, rudderless, and asleep at the helm.” Having worked as a medicinal chemist in the industry for 35 years (retired now), I am not going to argue that pharma executives are without fault (far from it). But the basis of his argument here seems to be related to the notion (perhaps common in academia) that biological targets and ligands are equivalent to drugs (a focus of this issue of your blog). Having worked on projects where we sometimes (not always) knew the biological target, and sometimes (not always) had a structure of the target, but were generally able to come up with potent compounds (but not always with drugs), it would seem by this late date that academics would be well acquainted with the difference. But, as you point out, that isn’t always the case.

    Your comments and those of your readers would be of interest regarding this Times article.

  18. Sympa says:

    Patent for self-steering car
    Car crashes count numerous lives. This can be prevented by steering the car in the correct direction.
    I have invented that this can be done by turning the wheels, preferably the front wheels, but it can be embodied in any other wheels or steering device, in an appropriate direction to counteract the undesired motion towards an obstacle.
    The turning using a motor, which can be electric but can be any other kind of motor, is generallty nown in automobile engineering.

  19. Stanislav Radl says:

    As an example, I can hardly imagine that discovery of the proton pump (H+/K+ ATPase) in the secretory membrane of the parietal cell can directly lead to the structure of PPIs (prazoles) that are prodrugs with quite unrelated structure regarding the PP.

    1. Cb says:

      Or fingolimod as prodrug for S1P GPCR

  20. RicoTorpe says:

    This brings to mind the nonsense patent given for any BRCA1/BRCA2 test, rather than a specific test. I do feel your pain. I’m a “computer guy” with a programming background, and I remember being infuriated by Amazon’s “1-Click” patent.

    I can see reasonable minds debte on something like the RSA public-key encryption algorithm. That took a lot of creative brainpower. But the 1-Click? That would be a weekly assignment for a second programming class.

Comments are closed.