Skip to Content

A Look at Antibody Therapies

Since we were just talking about antibody therapies in immuno-oncology, here’s a timely column by Bruce Booth at LifeSciVC on antibody therapies in general. It’s well worth a read if, like many small-molecule drug discovery folks, you haven’t had to keep up with that area. I’ve written a few times over the years about how antibodies have been gradually taking up spots on the lists of best-selling drugs, and Bruce’s piece continues in that same way. Did you know, for example, that five antibodies have already been approved by the FDA this year? Or that mAb projects have just about double the chances, when entering the clinic, of making it through to an eventual approval versus small-molecule ones? (In case you’re wondering, those percentages actually go up outside of oncology).

There’s also a good discussion of humanized versus fully human antibodies, and how the latter are gradually taking over the market (at present trends). Here’s something that you may not have realized:

In particular, it’s worth highlighting the transgenic UltiMab technology developed at Medarex.  With ten product approvals*, Medarex’ UltiMab platform has produced more approved drugs than any other human antibody platform in the industry. As an aside, BMS’s acquisition of Medarex for $2.6B in 2009, viewed with some skepticism at the time, is probably one of the best biotech acquisitions of all time. Nivolumab (Opdivo), discovered out of the Medarex platform, had sales of $2.6B in the US alone last year, and is forecast to have sales north of $12B within five years. Funny to reflect on this wonderful lead sentence about Medarex in an article in the spring of 2009: “If there’s a pharma company that rewards executives for delivering absolutely nothing, thy name is Medarex.” Delivering nothing? How about one of the most productive antibody platforms ever created.

In all fairness, a lot of people had complaints about Medarex’s management, but the science was indeed solid, and you can make the case that BMS did the acquisition to keep the latter and ditch the former. At any rate, Bruce is right that the platform itself has been very impressive.

Read the rest of his post as well for an interesting discussion of the attempts to use biophysical properties and in vitro assays to sort out which antibodies have a better chance of success in the clinic. Medicinal chemists will immediately recognize the impulse to start “Rule-of-Fiving” things in order to (ostensibly) improve the quality of lead drug candidates, but (just as in small molecules) the evidence, while there is some, is still not as compelling as the ideas behind it. The various attempts to score antibodies before they get to the clinic, while they do flag some interesting differences between the various modes of producing them, so far don’t seem to correlate well with eventual clinical success. And that’s the whole point of the exercise – otherwise, a difference that makes no difference is no difference. This doesn’t mean that no one will ever be able to do this sort of thing, but it does mean that it’s not necessarily far enough along to be a real actionable step in the process.

And Bruce’s final point stands, and just as strongly for antibodies as it does for small molecules. In the end, success depends on our knowledge of disease biology. That is still the big rate-limiting step. You can march into Phase II trials with a potent, selective small molecule or with a beautifully targeted non-immunogenic human antibody, and if you’re wrong on the disease mechanism or the importance of the target you’ve selected, then you’re going to go down in flames either way.

22 comments on “A Look at Antibody Therapies”

  1. Marcin says:

    Antibodies are present snd the future

    1. Barry says:

      People at Genentech expended a great deal of work after the successes of Avastin to improve its transport properties. They came up with Lucentis which retains the affinity and selectivity, but has 1/3 the molecular. Did they get better transport properties for their effort?

  2. Tom says:

    Something I have wondered about monoclonal antibodies … does it matter if an antibody is humanized or fully human? Is there any evidence to suggest this impacts the rate of anti-drug antibody formation, maintainance of response or safety outcomes etc?

  3. steve says:

    The PD-1/PDL1 antibodies are the most successful cancer drugs in history. Still, they only work really well for 20-30% of patients. Lots of work to be done.

    1. In Vivo Veritas says:

      I’d argue that this says more about our understanding of cancer, rather than the quality of the antibodies.

      1. John Wayne says:

        It is hard to separate the variables here, but if I had to bet I would agree with you – biology is hard.

  4. Thoryke says:

    Some successes with antibodies may provoke counterproductive immune responses:

    1. JIA says:

      This New York Times Article is about anti-drug antibodies (or anti-therapeutic antibodies), which indeed can be a problem for mAb drugs (thus Pfizer’s bococizumab failure). However ADAs tend to be more of a problem with non-mAb drugs, typically therapeutic proteins that replace a genetically missing or defective protein. The body recognizes these as “foreign” proteins because they are not naturally made since birth and so self tolerance was not generated. These ADAs are a serious and well-studied problem in hemophilia A and B for example (lack of coagulation factors FVIII and FIX, respectively).

      With all due respect to the doctor profiled by the Times, using Rituximab and other B cell or plasma cell killing drugs to dampen anti-drug antibody responses is not new. It’s been done in hemophilia for decades.

  5. Kelvin Stott says:

    “In the end, success depends on our knowledge of disease biology. That is still the big rate-limiting step.”

    It doesn’t need to be that way. And I’m not saying that increasing our knowledge is the solution.

    I’m actually suggesting to develop evolution-based systems that don’t depend on knowledge at all!

    1. Derek Lowe says:

      I have no problem with that, in principle. I’m a fan of agnostic, evolution-driven approaches, but I just can’t think (offhand) of a good way to apply those to particular diseases. High-content targeted phenotypic assays can be good, but that approach doesn’t take up the evolutionary advantages.

      1. Kelvin Stott says:

        You only need 4 elements for evolution to work automatically:

        1. Molecular diversity – combichem
        2. Selective pressure – phenotypic screening
        3. A reproducible information code that can be translated automatically – DNA
        4. An integrated system (information flow) that links all these components together in the right sequence.

        I have actually designed such a system. Have just submitted a proposal to our internal open innovation program. Now under review, will let you know if it gets selected for development.

        1. Derek Lowe says:

          Good luck to you!

        2. Chris Phoenix says:

          You need a fifth component: a link between “genotype” and “phenotype” that allows hill-climbing. In other words, incrementally different genomes need to produce incrementally different results.

          And a sixth component: At least a little bit of efficacy at the start. If you start on a level plain (zero effectiveness) then evolution can’t tell you which direction to go to find the hills.

          1. tangent says:

            Yep. Your genome space needs to show some hills, rather than scrambling them around the place into an incoherent hash.

            Machine learning maxim: the encoding is everything.

          2. Anon says:

            That clearly falls under points 3 (information code) and/or (information flow).

    2. Pennpenn says:

      You’re still going to require knowledge to properly screen your approaches, otherwise you’d just end up with a bunch of white noise, wouldn’t you? I mean, no one has the time or biological processing capacity to just throw everything at the wall to see what sticks.

      I might be misunderstanding what you’re going for here, though. Either way, good luck with your proposal.

      1. Kelvin says:

        As long as you can distinguish between what’s good (supposed to happen), and what’s bad (not supposed to happen), even if you don’t understand why it’s good or bad, then you should be OK. At least that’s how normal evolution works.

        1. Toby says:

          I think your evolutionary approach may increase the hit rate or efficiency in phenotypic screening, but nothing more than that. I don’t think it is an “evolutionary” approach to solve what we don’t know about biology. I think you are presupposing that hitting in a phenotypic screening assay will increase the probability of success in the clinic. I haven’t seen a formal comparison, but anecdotally I don’t think there’s a strong correlation between clinical success, and origin of the molecule from phenotypic screening vs. target based approaches. I imagine that there are likely relevant phenotypic assays that have never had a hit, and I imagine your approach could contribute something here. But there is also a lot of unmet medical need where it’s unclear (at best) what assays we would use in phenotypic screening (e.g. Alzheimer’s). There are also a lot of hits from disease areas where phenotypic assays are relatively tractable and scalable (e.g. selective growth inhibition for tumour cell lines in culture), but which never worked out in the clinic – the likely problem being the relevance of the assay. In my opinion the big unsolved problem is how to define the assays for phenotypic screening, not how to hit against them.

          1. Kelvin says:

            Thankfully, evolution doesn’t seem to care about theory or opinions. It just gets on with running billions of real experiments. 😉

          2. Peter S. Shenkin says:

            @Kevin “Thankfully, evolution doesn’t seem to care about theory or opinions. It just gets on with running billions of real experiments.”

            Yes, but evolution doesn’t care about killing subjects when some of those experiments fail.

    3. Dionysius Rex says:

      Hopefully this isn’t the same thing that you pitched to your previous employers!

  6. Anon says:

    “If there’s a pharma company that rewards executives for delivering absolutely nothing, thy name is Medarex.” Delivering nothing? How about one of the most productive antibody platforms ever created.

    The executives didn’t deliver the antibody platform, the scientists did. I bet they didn’t get highly rewarded either. Just like the PCR story.

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.