Hey—what’s up with that method from 1998 that was going to cure cancer in 2 years??

Our recent post about the non-existent Los Angeles tunnel reminded me of another bit of hype from the past: a claim by science hero James Watson in 1998 that cancer would be cured in two years. I was curious what was up with that so I googled and came across this news article from 2013 pointing to him writing, “We now have no general of influence, much less power … leading our country’s War on Cancer.”

B-b-b-ut . . . cancer had already been cured 13 years earlier! So what’s the problem? Maybe the “war” analogy is appropriate: the threat from Germany was stopped in 1918 but then twenty years later they had to do it all over again.

And this interview from 2016, where we learn that Watson at age 88 can still serve a tennis ball at 100 miles per hour. Also, he says, “I was pessimistic about curing cancer when gene-targeted drugs began to fail, but now I’m optimistic.”

Ahhh, now he’s optimistic. That’s good! The article continues with this juicy bit:

On what he sees as the best hope for treating and even curing advanced (metastatic) cancer: an experimental drug from Boston Biomedical (for which Watson is a paid consultant):

Papers have identified the gene STAT3, a transcription factor [that turns on other genes], as expressed in most kinds of cancer. It causes cancer cells to become filled with antioxidants [which neutralize many common chemotherapies]. In the presence of the experimental drug that targets STAT3, cancers become sensitive to chemotherapies like paclitaxel and docetaxel again. This is the most important advance in the last 40 years. It really looks like late-stage cancer will be partly stopped by a drug.

Hey, wait a second! In 1998, Watson was talking about a cure for cancer in two years. According to the news article from back then, he said that the developer of this cure “would be remembered along with scientists like Charles Darwin as someone who permanently altered civilization.” And that was less than 40 years previous.

So in what sense is this advance from 2016 “the most important advance in the last 40 years,” if only 18 years earlier there had been the advance that led his pal to be remembered along with Darwin etc etc.?

I’m all for people updating their opinions based on data, but if you’re gonna be hyping things that disappear, shouldn’t you at least acknowledge that you’ve changed your mind. I have the same feeling about this as I do about Stasi-guy and the Nudegelords memory-holing their hype of disgraced food scientist Brian Wansink (work that they earlier referred to as “masterpieces”).

As discussed in a previous thread:

The problem is that Watson was using “letter of recommendation” language rather than science language or journalism language. Letters of recommendation are full of B.S.; it’s practically required. For example, I recall years ago being asked to fill out a recommendation with options that went something like this:
– best student I’ve ever seen
– in top 1% of all students
– top 5%
– top 15%
– top 50%
– bottom 50%.
Top 15% is pretty good, right? But it’s on the bottom half of the scale. So to fill out this form in the way that’s expected of me, I had to do some mental gymnastics, where first I considered some narrow category where this particular student was excellent, and then I could honestly declare the student to be in the top 5%. I don’t remember who the student was, I just remember this annoying form which is pretty much demanding that I do some exaggeration.

The flip side of this is recommendations that are too honest. For example, I remember once receiving a letter of recommendation from an econ professor saying that a certain student was pretty good, not good enough for one of the top 8 programs but ok for anything from 9 through 20. This was just obnoxious, in no small part because of the ridiculous implication that the student could be evaluated at that level of precision, also in the assumption that qualifications could be summarized in a single dimension.

Anyway, the connection to Watson is that he was an academic administrator for many years, so I guess he got in the habit of writing letters that were full of hype so he could justify each hire and promotion he made, and so he could promote his students and postdocs to jobs elsewhere.

The trouble is that when the hype gets reported straight up with no acknowledgement later that it didn’t happen as they claimed it would. Same as with that tunnel story.

25 thoughts on “Hey—what’s up with that method from 1998 that was going to cure cancer in 2 years??

  1. A little more about Watson’s arrogance and academic imperialism comes through in Richard Rhodes’ book on E.O.Wilson, where
    Watson’s hostility to Wilson and any form of biology research other than his is described. Full disclosure: I have read the book but this
    is not a book review.

  2. “a certain student was pretty good, not good enough for one of the top 8 programs but ok for anything from 9 through 20. This was just obnoxious, in no small part because of the ridiculous implication that the student could be evaluated at that level of precision, also in the assumption that qualifications could be summarized in a single dimension.”

    Hilarious! Do you want a recommendation or not? :) Not much room between your ridiculously exaggerated version and the flip side. You don’t want exaggeration but you don’t want an honest evaluation either? What do you write when a poor student requests a recommendation? Do you go for the highest level of sunshine? “Top 5% at Playskool shape bench”? :)

    BTW, here’s the new “equity” version of the Playskool “shape” bench – the shapes are different on top but all the same beneath! :) Can’t go wrong! Every shape fits in every hole! Perfect for testing students – everyone is in the top 5%! Wow, this could really level the playing field for university entrance exams! So amazing to see how innovation is changing education!

    • Chipmunk –

      > BTW, here’s the new “equity” version of the Playskool “shape” bench…

      What will prevent you from projecting your political agenda into any context?

      The basic toy design is decades old. The objects are permanently attached. You completely misunderstand the entire concept of the toy, which is different than those you’re thinking of.

      https://youtu.be/6CgnPnpTWyU

  3. Andrew -.

    Orthogonal to your point, but…a hobby horse of mine.

    > For example, I recall years ago being asked to fill out a recommendation with options that went something like this:

    That is reflective of a problematic paradigm, imo; you’re not being asked to really describe attributes of that student, but to evaluate the student by comparing them to other students.

      • Dace –

        > ”Les Misérables” is literally the second best musical-tragedy I’ve ever seen, just don’t ask me how many musical-tragedies I’ve seen.’

        If you asked me about Les Misérables, would it be more informative if I told you it was the best of 100 musicals I’ve seen, or if I give you details as to what it’s about, and the qualities which made me enjoy it so much?

        Perhaps my taste in musicals is completely different from yours. If so, following the first evaluation you’d waste your money on a musical you didn’t like. With the 2nd, you’d know from my evaluation you wouldn’t want to buy a ticket.

        Those types of evaluations don’t need to be mutually exclusive – but often students are evaluated on a norm-referenced scale with little information on a criterion-referenced scale, which is often much more useful in understanding the student’s strengths and weaknesses.

  4. Writers of LoRs evaluate your competence in various respects, but this evaluation is noisy and only partially influenced by your inherent competence. It’s customary w/ LoRs to ask prospective writers if they’d be willing to write you a “strong” letter, and for a writer to refuse the request if they can’t, thereby thresholding on effect size… unless “letter strength” is purely positional (top 1%, 5%, etc.), relative to the distribution of your peers, in which case you’re thresholding on how far into the right tail of some “null” distribution their evaluation of you lies. Thus, willingness to write a strong letter serves as a 1-sided test, and letter writers’ estimates are exaggerated upwards: errors of magnitude, and perhaps in rare cases errors of sign (when you’re a very poor student, but the letter writer really likes you bc you share a hobby, or something).

    As for estimated timelines of biomedical advances… I can imagine other filtration mechanisms at play, eg news media not interviewing or heavily publicizing more pessimistic timelines, or those takes not spreading virally as far and thus unlikely to be remembered decades hence. I vaguely recall hearing some “rule” way back observing that futurists’ predictions of progress in senescence / anti-aging research tends to place the onset of “actuarial escape velocity” just within each futurist’s baseline life expectancy. Can see there being a filter there too: you wouldn’t care as much to become a “futurist” if you couldn’t at least reap the benefits of whatever technologies you “predict”!

  5. https://pubmed.ncbi.nlm.nih.gov/9390530/
    https://pubmed.ncbi.nlm.nih.gov/9389468/
    https://pubmed.ncbi.nlm.nih.gov/9011734/
    The above articles are from respected journals published in 1997 that certainly portrait very optimistic views of the utility of anti-angiogenesis therapies as proposed by Dr. Judah Folkman. Perhaps Dr Watson was simply swept up in the enthusiasm. I was personally in a very skeptical, almost cynical, frame of mind at that time since the grievous mistakes of high dose breast cancer therapy were around at that time.
    I have met Dr Watson. I saw him lecture and sat with him at some drug company luncheon. My impression was that he was a tightly focused individual but no polymath; I saw him as a nerd, but a supernerd.
    The lesson is that endorsements even by stars is not evidence.
    Academic recommendations are endorsements by their nature.

    • Yes, cancer is a tough one and I expect we’re still some way from making any sort of decisive headway. On the other hand the value of anti-angiogenesis therapies in treating wet macular degeneration should be mentioned – I know several people (including my mum) who’s quality of life in old/oldish age has hugely benefitted. Don’t know BTW whether these therapies arose via development of anti-angiogenesis treatments as tumor-suppressants or from a parallel line of development focused on the eye.

      • Yes, cancer is a tough one and I expect we’re still some way from making any sort of decisive headway.

        We are probably further from curing cancer than when the “war on cancer” began in 1971. Too much misinformation has been generated (as evidenced by the 25% replication rate reported by the cancer reproducibility project) and wrong assumptions ingrained.

        Science makes things easier to understand, but the conclusion on cancer is “it is more complex than we thought”. Actually, their problem is that its impossible to figure out what is going on using NHST.

        Go back to 1954. I would bet most cancer researchers do not even realize why they think cancer is the result of accumulated mutations, but it is based on the shape of the age-specific incidence/mortality curves. If we look at Armitage and Doll’s original data we see already there that the rates for many cancers peak at a certain age: https://pubmed.ncbi.nlm.nih.gov/15599380/

        And that in deriving their equation (which does not fit this data), they hard-code the assumption:

        In this case and so long as the occurrence of each mutation is a relatively rare event-the incidence rate of cancer at age t will be proportional to the probabilities of occurrence of each mutation per unit time and the sixth power of the age

        […]

        This result will be valid for large values of t (of the order of a human lifetime) provided that p_1*t, p_2*t, …, p_r*t are all sufficiently small (as could be assumed in an application of this theory to human cancer).

        Ie, to simplify the calculation, they use the approximation p*t ~ 1 – (1 – p)^t when p << 1. If we get rid of this simplifying assumption and allow p to take values of ~ 0.01, then the rederived equation works. Ie, the cumulative incidence is (1 – (1 – p)^t)^n, where n is the number of mutations that must accumulate on average.

        From that we conclude carcinogenesis cannot be explained by mutations that occur at rates of p < 10^-6 per cell division as has been observed. It is something that happens far more often, probably chromosomal missegregation. And further, that carcinogenesis is common but most of the time the lineage dies off or is cleared by the immune system. In other words, it is not the rate limiting step.

        Or, of course, the accumulation of errors idea is wrong altogether.

        • Anoneuoid –

          >… are probably further from curing cancer than when the “war on cancer” began in 1971.

          Seems to me (I’m certainly no expert) that life expectancy post a given cancer diagnosis at a given stage, has significantly increased. Am I wrong?

          If not, how do you reconcile widespread improvements in treatment with getting farther away from a “cure?” I recognize that it’s theoretically possible that all the improvements with treatment don’t lead towards a “cure,” but seems like a heavy lift to me as it would suggest some improvement in understanding the disease mechanism.

        • This is just the first link that came up explaining it, I didn’t read the paper but you should get the idea:

          Although 5-year survival is a valid measure for comparing cancer therapies in a randomized trial, our analysis shows that changes in 5-year survival over time bear little relationship to changes in cancer mortality. Instead, they appear primarily related to changing patterns of diagnosis.

          https://pubmed.ncbi.nlm.nih.gov/10865276/

        • I’ll read it – I mean sure, longer survival could just mean earlier diagnosis – but I have a hard time believing (at least with some kinds of cancer) that across the board for a given type of cancer diagnosis at a given stage life expectancies haven’t improved. But I’m open to accepting my I impression is wrong.

        • Joshua, you are correct–life expectancy after certain organ-specific cancer diagnosis has improved over time.

          It’s trivial to get around the survival bias introduced by early detection (known as lead-time bias) by comparing survival from diagnosis to death to expected survival and calculating relative survival.

        • Unanon –

          Ya, thank, and we’ll, that paper anyway basically limits to treatments from 30 years ago, and I’d have to wonder if the 5 year-survival comparison would hide a whole lot of shorter-term increased life expectancy, and anyway the paper wouldn’t substantiate Anoneuoid’s assertion that we’re further from a cure, as at most it would support that we’ve been treading water.

          Methinks binary thinking might be afoot. If it ain’t perfect it’s gotta be worse seems to be a pattern.

        • We are probably further from curing cancer than when the “war on cancer” began in 1971. Too much misinformation has been generated (as evidenced by the 25% replication rate reported by the cancer reproducibility project) and wrong assumptions ingrained.

          It’s not obvious that we’re likely to “cure” cancer within the current scientific paradigm, but we’re obviously closer than we were in 1971. The best we can do for now (maybe next 20 years) is to make incremental advances in “treatment”, much as has been done with some success in the last 20 years. Since cancer is a disease largely comprising accumulation of somatic mutations, it’s not obvious how cancer can be “cured” – apart from inherited or viral-induced cancer susceptibility, it’s pretty much a natural consequence of ageing – and any prospective “cure” will require intervention at the level of the (somatic) gene/genome which is at least a couple of decades away I expect. It will require reversing or attenuating the effects of driver mutations in oncogenes, tumor suppressor genes, DNA repair genes and apopotic genes – it’s a tough problem and I doubt the problems lie in “too much misinformation..generated”.

          BTW the Cancer reproducibility Project reported 46% replication rate ( https://elifesciences.org/articles/71601 ) :

          For positive effects, 40% of replications (39/97) succeeded according to three or more of these five methods, and for null effects 80% of replications (12/15) were successful on this basis; combining positive and null effects, the success rate was 46% (51/112). A successful replication does not definitively confirm an original finding or its theoretical interpretation. Equally, a failure to replicate does not disconfirm a finding, but it does suggest that additional investigation is needed to establish its reliability.

          Or, of course, the accumulation of errors idea is wrong altogether.

          The evidence is pretty strong that cancer is the result of acumulation of mutations. This is a good account:
          https://pubmed.ncbi.nlm.nih.gov/23539594/

        • BTW the Cancer reproducibility Project reported 46% replication rate

          Correct, that is ~50% out of the only half the studies where it was possible to get enough info to actually do the replications. The other half were not even reproducible in principle after reading the published reports and contacting the authors. This gives a total of ~25%.

          If you agree after looking into it then we can address your other points.

    • I have heard that Crick was the real brains of the pair, as he made notable scientific advances even apart from Watson, whereas Watson just moved into management.

      • Wonks:

        I don’t know that Watson ever claimed to be a brainiac. I got the impression that he presented himself as a smart ambitious guy who was in the right place at the right time, which seems pretty accurate. As I wrote above, I think the “cancer cure in two years” thing was a bit of sloppiness arising from Watson having spent decades fundraising and writing letters of recommendation, two activities where exaggeration and hype are not just the norm but are pretty much required.

        • Perhaps he didn’t even say it:

          Watson, in a letter to the newspaper published on 7 May, said “my recollection of the conversation [with the New York Times reporter] is quite different”. But in the letter, he calls the work “the most exciting cancer research of my lifetime”.

          Wendy Goldstein, a spokeswoman for the Cold Spring Harbor Laboratory on Long Island, where Watson is president, said Watson was contesting the quotation attributed to him “because he feels a statement from him has offered what could very well prove to be false hope to a great many people”.

          The New York Times says it stands by the Watson quotation. But on 8 May it admitted it had imprecisely paraphrased Klausner by using the words “The National Cancer Institute has made the drugs their top priority” to describe his statement “I am putting nothing on higher priority”

          https://www.nature.com/articles/30072

  6. It looks like no one ever explained how they managed to get such promising results w/ angiostatin in mice. In which case this same mistake is probably being repeated.

    The most obvious issue* is only taking a snapshot of tumor count/size. If the cells in control divide twice per day for two weeks you’d have N0*2^28 cells at the end. A treatment that slows growth by 10% would yield N0*2^25 cells. So a 10% effective treatment is transformed into a 87.5% effective one. And of course it gets worse since growth will eventually plateau.

    Its possible the drug behaves the same in humans, they just don’t sacrifce the patients after 2 weeks.

    * Ignoring lack of blinding and independent replication.

Leave a Reply to chipmunk Cancel reply

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