Skip to Content

Near-Animal, Near-Human: The Future of Drug Testing?

Here’s a good article on animal models in drug discovery, and their many limitations.

We have moved away from studying human disease in humans,” (Elias) Zerhouni lamented to the NIH’s Scientific Review Management Board meeting. “We all drank the Kool-Aid on that one, me included.”
“The problem is that it hasn’t worked, and it’s time we stopped dancing around the problem,” he continued, suggesting researchers have become too reliant on questionable animal data. “We need to refocus and adapt new methodologies for use in humans to understand disease biology in humans.”

The article notes the controversy over mouse inflammation models, among others. I’d add pain as an area where something is clearly off kilter with the traditional animal models, and no one has ever been happy with xenograft models in cancer. (An entire post on “worst animal models” is here).
Well, just use human cell cultures you say, or at least you say if you’ve never tried it. The problem, of course, is that primary human cell cultures are often hard to keep going, and the things you have to do to keep them going often skew them away from being the sorts of cells you were hoping to study. There’s nothing like a real in vivo system, unfortunately. Stem cells hold out a lot of promise for generating human cells of many types, but (as this article explains), the problem is that those cells are fresh, newly minted ones. And those don’t always recapitulate the sorts of changes that you see in (for example) aging neurons.
It looks like we’re moving towards re-creating real human organ-like tissue in vitro, as much as we can. That’s not the least bit easy – there are so many factors that influence cellular physiology, from the obvious ones (constantly changing signals from blood chemistry) to the nonobvious (mechanical forces from nearby muscle contractions). The other way to do it (not discussed in the linked article) is to humanize the animal models as much as possible. One could imagine a rather unnerving “mouse” consisting of mostly (or completely) human tissues out in the periphery, for example.
I’ve defended animal models many times on this site, but my strongest arguments for them are (and remain) that there have to be some intermediate steps on the way to human trials, and that we don’t have anything better that mice et al.. If something better comes along – and we do need something better – then out they go. The lack of really predictive models before human trials is the reason that 90% of all trials fail, and having 90% of all our trials fail is gradually squeezing drug research into a very tight corner indeed. Breaking out of it would be a real accomplishment.

23 comments on “Near-Animal, Near-Human: The Future of Drug Testing?”

  1. Anonymous says:

    90% failure is optimistic, these days it’s more like 2-3%

  2. Steve says:

    This sounds cruel to animals. How about testing on humans? I’ll volunteer for a price.

  3. NJBiologist says:

    “The article notes the controversy over mouse inflammation models, among others.”
    Using the genomic response in a sepsis model as a measure of pain and inflammation models is painting with a broad brush–if you look at clinical efficacy, you find more predictive validity (e.g., Whiteside et al. 2008 Neuropharmacology v. 54 p. 767).
    The idea that models need to be compared to human trial data is sound. We just need to remember that, no matter how cool RNA analyses might look, they’re not the primary endpoint for a clinical trial.

  4. @samadamsthedog says:

    Of course, i have, umm, skin in this game; but couldn’t you humans go back to using prisoners, or orphans?

  5. bacillus says:

    The artificiality of many of the animal models is a major problem. Stroke models in rats springs to mind and schizophrenia models in mice. I have a son with the latter disease, and it is both naive and insulting to think you’re going to model it in rodents in any meaningful way anytime soon. On the other hand, diseases that are natural to both mice and humans, seem to enjoy better success. I’m thinking especially about infectious disease models where the organism in question is a natural pathogen of many mammalian species including humans. Of course, even these can get messed up by researchers giving pulmonary pathogens intraperitoneally, or oral pathogens intravenously etc. Given enough time, many animals might naturally develop their own equivalents of human non-infectious diseases, but the expense would be ridiculous.

  6. kjk says:

    How much of this failure is animal vs human and how much of it is

  7. ex says:

    Even better, why we use lawyers, they are almost human…

  8. Morten G says:

    Rabbits for syphilis, for those of you that remember your medicinal chemistry history.

  9. steve says:

    Like everything else, there are pluses and minuses to animal models. The main issue is to not over-interpret the results. The entire field of immunotherapy for cancer, which is the most promising advance in cancer in the last 20 years, would have been impossible without animal models. Inbred mice are what allowed for the dissection of histocompatibility loci, the dissection of tumor-mediated immune suppression mechanisms, etc., etc. There are many examples of the value of animal models in understanding basic mechanisms that have led to medical breakthroughs. You can’t just test everything in people (even lawyers) without some understanding from in vivo models. The

  10. pharma scientist says:

    @9 Yes! Don’t over-interpet the results. Are they perfect? No. But they can be very useful and we really don’t have better alternatives right now. And don’t be married to one model – sometimes rats are better than mice (they aren’t just bigger versions!). I do like to see the stem cell work progressing. It may become another tool to help the preclinical scientists.

  11. pharma scientist says:

    One question about Derek’s comments. Does anyone have data to support the “90% of all trials fail” because they lack predictive models? I read a review once that pegged it at about 1/3 failed due to lack of efficacy, 1/3 to toxicity, and 1/3 due to general business issues (too slow to market, ran out of money, changed direction, etc.). Obviously it varies by therapeutic area, but it’s closer to my personal experience. I’ve only had one project die due to lack of efficacy in the clinic (of course I’ve never worked in Alzheimer’s either!).

  12. Derek Lowe says:

    #11, I meant that the overall clinical failure rate was around 90%. Better animal models would immediately address the Phase II efficacy failures, but my hope is that they might also provide some warning about the Phase III tox failures as well. I may be too optimistic there, though.

  13. TX raven says:

    Human a good model for human?
    Why then some drugs, for example antidepressants, work only in 30% of the patient population?

  14. Anonymous says:

    #13, FDA trials are more for safety than efficacy.
    If the drugs does not harm most of the people taking it, it is easy enough to try it and see if it works, and then try another one if that fails. But for some diseases and illnesses, the trial and error approach would be much harder than trying a different drug if the first one does not work. For some really bad illnesses, like Ebola and some cancers, the risk of dying is high enough to try nearly anything to treat it.
    Not all diseases, especially depression, are from one cause or one pathway, so not every drug will work in every case, just like antibiotics only work for certain infections. But there are tests for determining the cause of an infection, but no clear tests for exactly why people get depressed.

  15. dr. moreau says:

    would you like to bring some drugs to test here on my island? I’ll send you directions. (don’t let anyone follow you.)

  16. steve says:

    Looks like my comments aren’t conveying. The study suggesting that mice are not good models for human sepsis got a lot of internet play but the followup studies that showed they weren’t true didn’t, of course. I tried to add the links but the post got held up pending review. Try googling “mice human sepsis” and you’ll see that the paper doesn’t hold up.

  17. dave w says:

    #i6: I’ve had that happen too… apparently the commenting engine has a thing about web links in the comment text!

  18. Cellbio says:

    Steve, I am not convinced the follow up discussion really debunked the claim that mouse models are not useful, if one defines useful as being predictive for efficacy in humans. Are you aware of data that says so? Seems like Jackson labs, a vendor of mice, argued hard that mice are valuable models of biology, fine, but did not really say that they serve a highly valuable tool for finding drugs which work for humans. Maybe I missed something?

  19. Anonymous says:

    #13, I guess that personalized medicine for all will make most of the drugs work for the genetically fit patients. In cancer, it is already happening, e.g. Plexxikon’s B-raf inhibitor drug was tested in phase-III trial only for the genetically-mutated subjects. Also, the new third-generation EGFR and ALK inhibitors for nsclc are being tested/developed only for subjects who have the T790M mutation (EGFR). Ceritinib and Crizotinib (ALK inhibitors) have been approved only for the genetically mutated nsclc cases.

  20. simpl says:

    Humans are fine in the acute phase: meningitis, bird flu, ebola are recent examples where medical urgency flattened scientific hurdles.
    Chronic diseases fare less well: depression, a nice one mentioned by #13, or arthritis.

  21. anon says:

    #13, 14: FDA looks at cumulative safety from all 3 phases and efficacy is demonstrated in Phase 3 (hopefully by mutual pre-agreement on clinically meaningful outcomes). No drug can be FDA-approved without providing a benefit. Another way to look, if a drug has no benefit, no risk is acceptable.
    As others mentioned: if not the appropriate disease diagnosis or not taken according to label, the drug is not going to be optimal.

  22. Anon says:

    How about this “liver on a chip” technology? Anybody using it successfully?

  23. Slicer says:

    It appears that the only real way to test how any given drug will work for any given patient is, in fact, to grow the organ itself from the patient’s own stem cells. It needs to be as good as possible of a copy.
    Then a surgeon can transplant the completed new organ into the patient, and you can keep your drugs! Wololololo
    (note: not advisable for brains)

Comments are closed.