Skip to main content

Biological News

Precision Medicine Real Soon Now

Here’s a strongly opinionated look at where the “precision medicine” field is these days, and I think that this is just the sort of article that the field (and the journalists covering it) need to see, whether you agree with it or not:

In 1999 Francis Collins published a foundational document of precision medicine entitled “Medical and Societal Consequences of the Human Genome Project,”(1) which made predictions about the ways the human genome would be used to predict, prevent, and treat disease in 2010. In 2000, he suggested that “Over the longer term, perhaps in another 15 or 20 years, you will see a complete transformation in therapeutic medicine” (2).

The vision described in the article became the aspirational template for the precision medicine movement (Figure 1). We have passed the 2010 deadline and are rapidly approaching 2020, yet the “complete transformation in therapeutic medicine” has not occurred. Using the framework of the predictions made nearly 20 years ago, we argue that the foundational assumptions of precision medicine are unsound.

It was indeed clear by 2010 that the expectations from the original human sequencing were not being met, and that part of the reason was that some of those expectations were a bit too grandiose. That doesn’t mean that the whole thing was a bust, by any means – the push towards fast, cheap sequencing has had profound effects throughout the biomedical sciences, certainly up at the R&D end. I would have to think that the drop in sequencing cost/base pair would have taken place anyway, but it surely would have taken longer without the impetus of the Human Genome Project. But talk of “complete transformation” in medicine, yeah – that’s not quite the case. To get our definitions straight, this paper defines “precision medicine” or “personalized medicine” as “the view that incorporating information encoded in the human genome as the dominant factor in the prediction, diagnosis, and treatment of human disease will lead to marked improvements in human health“.

We’ve long known about the genetic contributions to many rare diseases. One of the hopes for the new era was that we’d be able to discover some genetic variants or combinations with relevance to many of the more widespread ones (diabetes, Alzheimer’s, etc.) But that’s not the case. What you find is a cloud of dozens or hundreds of genes that put together explain only some fraction of the disease landscape. And it’s not like the rest is lurking in the genome somewhere – we’re studying the whole genome already, so that excuse, which was very popular at one time, is no longer operative. Nope, the rest of it is in all the things that aren’t in the gene sequences: epigenetic markers, to be sure, but more generally the uncounted effects of environment and development. Now, it’s true that you can do a pretty good prediction of height from gene sequence these days, with a model that blends in a whole heap of gene variants, each of which contribute a bit up or down. But that’s only if you stipulate good nutrition, in that case, since that can override things pretty thoroughly. And for so many other traits, we don’t quite know what nongenetic influences to control for, or how much genetic influence is left once you’ve done that. As this paper puts it, “Extensive analyses of thousands of potential gene-health outcomes often fail to match, let alone exceed, the predictive power of a few simply acquired and readily measured characteristics such as family history, neighborhood, socioeconomic circumstances, or even measurements made with nothing more than a tape measure and a bathroom scale.”

So if you had to boil down the lessons of the “genomic revolution”, I’d suggest “Sequence isn’t everything”. You might think that cancer would be an exception to that, since there are certainly specific molecular mechanisms that tumors use to become (and remain) tumors (although we don’t know them all, to be sure), and since these are often driven by gene mutations. The problem is, as anyone in the field can tell you, that tumor cells also tend to be genetically unstable. Any given solid tumor is almost certainly a mosaic of dozens, hundreds of mutated cell lines, with more being thrown off all the time. So even in the cases where we would expect a benefit, it’s elusive, except for some gene variants with exceptionally high penetrance.

The authors sum up:

However, nearly two decades after the first predictions of dramatic success, we find no impact of the human genome project on the population’s life expectancy or any other public health measure, notwithstanding the vast resources that have been directed at genomics. Exaggerated expectations of how large an impact on disease would be found for genes have been paralleled by unrealistic timelines for success, yet the promotion of precision medicine continues unabated.

The problem is partly with the results, and partly with the predictions. Expectations were raised too high. The general public got the idea that Genes Are Destiny, which viewpoint is being exemplified (and reinforced) by the popularity of the consumer-sequencing companies. Those might be interesting to figure out your ancestry (although there are a few little tiny uncertainties here and there), and some folks have definitely found out more about their immediate ancestry than they realized there was to know. But as far as predicting disease or overall health. . .not so much. That’s a tangle of environment and heredity like you’ve never seen. Unless you hit one of those rare-disease genes with huge penetrance (where genes really are destiny), your gene sequence is just part of that tangle, and good luck unraveling it.

All those clinics advertising “precision cancer treatment”? Same problem, and worse because of that genomic instability mentioned above. It would be great if we could dial in a personalized treatment for each patient based on tumor sequence, but that’s not where we are yet, outside of very rare cases. Earlier attempts to treat the general run of oncology patients by such techniques have not worked. The current effort in this line is NCI-MATCH, a comprehensive effort to line up targeted therapies versus genetic signature, and it has shown a mixture of potentially promising and not as promising results so far. We do have the first approvals for oncology drugs (larotrectinib  and pembrolizumab) on the basis of gene signature regardless of tissue, so there’s progress. But if you’d told Francis Collins back in 1999 that we’d have the first two approvals of that sort only in the fall of 2018, he would have wondered what the heck went wrong.

The authors of this new paper end by saying that “it is urgent that the biomedical research community reconsider its ongoing obsession with the human genome“, which is strong language. We’ve learned a lot from genomic studies, and we’re still learning more, and it’s not going away. But if by “obsession” they mean trying to apply genomic viewpoints to every problem regardless of suitability, or promising success in some of these programs once we can do just a bit more sequencing – because that’s all they’re lacking – then they have a point. The genome is great, the genome is huge, the genome is important. But it’s not the only great huge important thing out there.

39 comments on “Precision Medicine Real Soon Now”

  1. Earl Boebert says:

    As an outsider, I see a very thin line separating genomic studies from genomic determinism, and an equally thin line separating genomic determinism from eugenics; lines, which if crossed, have been shown to lead to dire results.

  2. Shankaran Kothandaraman says:

    The so called “Normal” genes took millions of years to evolve and perfect by “trial and error” and “hit or miss” methods of mutation, incorporating viral component etc. before settling for the present and is still evolving. After settling, the mistakes thereof also showed up over the period of human evolution and then some. I am not a biologist but when you “squeeze” this time frame with Gene technology (last 2 decades) it is going spring surprises or mistakes with equally truncated space of time! And in a due course of time mistakes pile up in a negative way. Technologies like SiRNA, CRISPR are not panacea or precision medicine for all problems or cure thereof! Humbled by nature!

  3. Red Agent says:

    Genomic medicine is starting to remind me of the old expression about nuclear fusion: It’s the electrical power source of the future. And always will be.

    1. Hap says:

      It’s hard to see how to make it work – trials seems hard to do, for example, and anything that’s bespoke each time is going to be really expensive (and hence not make so much impact, because it can never be made in mass, and so never gets cheaper to make). I know the future is hard to predict (as below), but this seems not to pass a sniff test, or to require more than a handwave.

      1. Ken says:

        Obviously the solution is to (1) sequence the patient’s entire genome, (2) feed that into the sophisticated cellular modeling software to identify the exact cause of the problem, (3) pass that through the AI automated drug design systems which output the required drug structure, and (4) send that off to the nanotech factories to be put together atom-by-atom.

        Except for the cost of step 1 and the non-existence of steps 2-4, we’re practically there.

        1. metaphysician says:

          Hey, that’s how medicine works in the Culture novels. 😉

          I am generally optimistic for the prospects of artificial intelligence, but. . . you really shouldn’t be pursuing a research path if the only way to get it to work requires ASI god-beings. If they get developed, and don’t either kill us or abandon us as a species, then ask *them* to try those solutions. Otherwise, pursue research paths which can be followed by a mere organic mortal brain.

        2. Gollum says:

          You just outlined the entire GSK AI DPU blueprint for reducing drug discovery-to-market time and development costs from 10yrs to 1yr. Minus of course, the realities of steps 2-4. It’s a vision thing.

  4. Hap says:

    1) Predicting the future is hard, and we don’t know what we do that is going to change things most (and whether it is a good idea or not).

    2) As long as big claims get big money (and cost little when they fail), they’ll keep getting made.

    1. John Wayne says:

      #2, +1

  5. Dr. Manhattan says:

    One thing we have learned since 1999 is outside of the protein coding regions, there are an enormous number of RNA species present in humans that influence gene expression and regulation. Add to that alternative RNA splicing of coding genes, epigenetic changes that influence expression and GWAS yielding lots of weak relationships among many genes as risk factors. Clearly just reading the DNA code is not the panacea that was hoped for many diseases. Lots of progress from sequencing, but the basis of disease is far more complex than just reading off a sequence and assigning a risk incidence to one or a few genes.

  6. loupgarous says:

    “Precision medicine”, in the case of my rare neuroendocrine tumors, means “finding out by trial and error what works” because we’re still finding out which DNA sequences code for which tumors, or which diagnostic or therapeutic compound they take up. Until that knowledge is gained, “precision medicine” is aspirational, not real.

  7. anon says:

    Your first link points to reference #12, vs. the top of the main article.

  8. Cameron Pye says:

    Seems like a slightly more pessimistic view than warranted but the gestalt rings true. There’s no doubt that modern -omics research has provided us with a level of biological detail that far exceeds the tools and diagnostics that came before.

    Making that data actionable seems to be the cross for precision medicine to bear now. Between the hype of AI lies some empowering and potentially transformational technologies to help pull insight out of the data deluge.

    However, patients only realize the utility of any of that when physicians have therapies to deploy against these newly (or not so newly in some cases: KRas, MYC, etc) elucidated targets. Sorry folks, we still need new chemistries. See you in the fume hood.

  9. John Hasler says:

    Seems to me that an early application of genomics might be helping physicians decide what *won’t* work. Being warned off adalimumab would have saved me a lot of misery.

  10. Anonymous says:

    Some general comments. In “pure” chemistry, there are catchy concepts that, basically, created self-sustaining, self-fulfilling funding streams. A simple proposal to do “Basic chemistry” or “Good Science” on a new or interesting topic is not enough. The NSF and NIH must have catchy programs! I heard FA Cotton lament this at an ACS meeting (and in a C&EN Letter, too). He said that TAMU could not hire who he thought were the brightest and most creative candidates. They would only hire those with proposals in established areas. If you want to get funding, he said, you have to be working in an established buzzword field: a buzzfield (new word?). Dendrimers! Fullerenes! and so on. There are, I think, similarities in medicine: Immunotherapy! Gene Therapy! Everything is oversold and a small group of powerful PIs and “thought leaders” soon break away from the pack.

    Rosenberg is one of the fathers of Immunotherapy, going back to the 1970s. He was a phenom and had a rapid rise from Harvard – Dana Farber to the NIH. The promise of immunotherapy only took a couple of decades and several hundreds of millions of NIH dollars, but there have been some successes. (I think there was a LOT of good science and knowledge expansion, even if the therapeutic promise was kind of slow to fulfill.) And parallel stories in gene therapy and stem cell therapy.

    Some of these guys shamelessly make HUGE promises to the NIH, to Congress, to VCs, to Boards of Directors (private, public, and not for profits) and to the public … and fail to deliver while growing their programs and reputations and careers. (I, obviously, have always had trouble with that. Many times, the scientific outcomes do not justify continuation of certain studies but some of these leaders must protect their programs and funding through misrepresentation of their results.)

    As discussed here In The Pipeline, “thought leaders” were able to concentrate AD research funding on amyloid to the exclusion of competing ideas.

    I am not saying that he was the first, but historically we can go back to Paul Ehrlich and his Magic Bullet (Zauberkugel) Buzz Word Program first espoused in 1900. By 1909, he had developed Salvarsan for the treatment of syphilis and ushered in the entire field of small molecule chemotherapy.

    1. loupgarous says:

      Useful to note that salvarsan’s original name was “compound 606” – it being the 606th compound studied for that indication. The “magic bullet” came after 605 misfires.

      Ehrlich wasn’t unique for his tolerance of failure in research – Edison’s comment “Genius is one percent inspiration and ninety-nine percent perspiration” is largely forgotten today, but medchem people have always dealt with the reality of it.

    2. anon says:

      Steve Rosenberg and Carl June are physicians. They can support themselves by seeing patients and Medical Schools are literally begging them to do research. Their livelihood are much less impacted by impact factors, tenures and NIH grants. They have the luxury of leisure of trial and error in medical science that most of the Ph.D. scientist can not afford.

  11. Earl says:

    Adding to the ethical dilemmas of DNA testing for ancestry, there was recently a mother who made use of 23andMe to learn more about her 5 year old daughter, and happened upon a relative that is likely to be the child’s father, an anonymous sperm donor. The man reacted very poorly to being contacted by the mother, naturally, and the mother was eventually served a cease and desist letter from the sperm bank as well as having her spare sperm samples confiscated. I suppose there’s always a chance when you donate sperm that you will be contacted one day by the recipient of your sperm; but the odds of that happening increase dramatically with this technology and every day that the database of genetic information grows.

    1. Polynices says:

      Arguably anonymous sperm donation is already an outdated concept exactly because of advances in genomics. If darn near everyone can get fully sequenced for cheap with their genomes available online for heredity searching then there’s essentially no anonymity possible. If we can solve decades old cold case murders we can definitely uncover every last “anonymous donor”.

  12. luysii says:

    The first genetic disease to have its molecular cause determined (Linus Pauling) is Sickle Cell anemia. We currently have no treatment. The ’cause’ has now been known for SEVENTY years. Hemoglobin and its variants have been the E. Coli of the protein chemist. If we really understood proteins we’d have been able to design a nonToxic small molecule to prevent sickling. Some humility is in order.

      1. luysii says:

        True, but let’s wait to see if it works. There have been multiple disasters with gene therapy.

        1. John Wayne says:

          At the moment, injecting somebody with transfection agents is … optimistic. Hopefully we will figure out the delivery issues and these therapies will get a chance to move the needle.

          1. luysii says:

            I certainly hope you’re right. I’ve taken care of sickle cell crises (they aren’t pretty), and had a patient die from an intracranial aneurysm associated with the disease. We must worry about cytokine storm, and integration in front of an oncogene, with later development of cancer.

      2. Anonymous says:

        Neither Kolata’s NYT article nor any of the (mostly well informed) comments on gene therapy mention that “the first US regulatory approval of a gene therapy” came in January 2018. Spark Therapeutics’ Luxterna, an RPE65 gene therapy for Leber Congenital Amaurosis (a kind of retinitis pigmentosa) was approved. It is also approved in the EU. It is also $425,000 per eye. LCA is a childhood-onset irreversible blinding disease so you kind of want to get treatment before it’s too late.

        Just one story linked in my handle.

    1. NJBiologist says:

      To play devil’s advocate: CFTR was cloned in 1989, and Vertex got their approval in 2012. Still slow, but nowhere near 70 years.

      1. jbosch says:

        And things are developing rapidly to introduce WT into ∆F508 by CRISPR/Cas9 into humans.

  13. Todd says:

    What I find interesting is how just as the hype from the genome is dying down, the real fruit of the genomic revolution is coming to harvest. There’s no way you could have nailed down all the new Alzheimer’s research without some heavy duty sequencing. If you tried it with PCR, you would have had to nail down all possible cross contamination in a way that would be plausible to a journal. Good luck.

    Also, you’re getting a bunch of readouts from liquid biopsies that are made easier by sequencing. Clinical laboratories are finally getting into this field for real. There’s something there but, as usual, it isn’t what one originally expected.

  14. SAS says:

    Will be interesting to see if George Church’s anonymized whole-genome marketplace “Nebula” picks up any steam. Looks like it’s been online for about 2 months. Personally I think it is a great idea and hope it becomes popular.

  15. Cb says:

    For Cystic Fibrosis the Vertex drugs are directed to certain CF recepor mutants: I think the ivacaftor story and later drugs show that personalized medicine may work if number of mutations is relatively small

  16. J Tyson says:

    Speaking of old predictions, back in the 1976 presidential debate Jimmy Carter lectured Gerald Ford, his opponent, that the USA should immediately switch to coal and nuclear power. The reason? The oil would run out by 2011.

    1. Isidore says:

      As the great American philosopher Lawrence Peter Berra pointed out, “It’s difficult to make predictions, especially about the future.”

  17. Katherine says:

    We’re not even making effective use of what information we do have. How many people know their CYP2D6 phenotype? I’m on a drug that’s life-changing for me, but my doctor was ready to give up on it at a lower dose until I reminded her of the test she herself had ordered showing that I’m an ultrarapid metabolizer.

  18. Tomas says:

    Next up, AI. Though as others commented, big claims will be made as long as there is money in it

    1. loupgarous says:

      As someone said about Christianity long ago, AI hasn’t been tried and found wanting as much as it’s not been tried. AI’s real world applications are a few decades old.

      At that stage in its evolution, aviation still was useful mainly for military reconnaissance (unless you count air-air actions) and rapid mail delivery, wings and fuselages were still mostly made of varnished cloth and wood, and navigation was as much by examination of landmarks as by compass.

      The big claims made for aviation at the time now seem modest in hindsight. Military competition drove technical advances in aviation, because that’s where the money was then. Medicine and the military are still driving AI advances (along with materials science developments such as graphene for nanostructure and circuitry and repurposing CdZnS, CdSe and other molecules as “quantum dots”) because the money is there for anything that could make the practice of medicine and war more efficient.

  19. Barry says:

    The difference between Iressa (clinical finding of non-efficacy) and Tarceva (clinical finding of efficacy) was in the (genomic) choice of the right clinical population.
    Cancer may still be the field in which this is most important because we still don’t know how many diseases “cancer” is. Sequencing your tumors is still the best way to know what you’re trying to treat, and therefore what you should treat them with.

  20. Bob says:

    This feels like a “its just too hard, so lets not try” kind of article. There are a lot of proven use cases that effect a large percentage of the population that could be done now, including pharmacogenetics. Not embracing proven scientific evidence seems like mal-practice to me.

    1. Big Freddie says:

      But we are also have lack of well designed clinical trials, approvals based on one positive trial, n=1 clinical treatment plans with no Pretrial registration etc etc. With historical controls (old data with folks who would be reclassified if put into modern diagnostic regimens) and an FDA fully captured by industry anything we try will be malpractice my friend. If everything in medicine becomes a guessing game it is all malpractice.

Comments are closed.