First things first: congratulations to Mary-Claire King, who this week won the Lasker Prize in Medicine for “bold, imaginative, and diverse contributions to medical science and human rights”. The award acknowledges King’s work in the 1980’s, reuniting families with the biological children taken from them during Argentina’s “Dirty War”, and it highlights her role in the pursuit and identification of genes implicated in familial breast cancer. Myriad’s Mormon database may have won them the patent race, but their Rosie Ruiz victory belied the long history of the chase, which King pursued for more than 20 years, stubbornly refusing to believe that pedigrees strewn with cancer were simply bad luck. What seems like common sense now was iconoclasm at the time, and it begs the conclusion that it took a woman in science to take these women’s stories seriously. King’s role in the discovery of BRCA1 was a triumph of independent thinking, as well as brilliance and persistence. You go, girl scientist!!!
King used the occasion of the Lasker Foundation’s announcement to publish a call for BRCA 1 and 2 testing for all women at age 30. The manifesto, published in JAMA, decries the current state of affairs, where women are frequently tested only after they get cancer. “To identify a woman as a carrier only after she develops cancer is a failure of cancer prevention.” This is a bold challenge, and given the source, one that demands some careful consideration.
There are some negatives to universal screening, many of which King discusses. A diminished but enduring problem with BRCA testing is the tendency to produce variants of uncertain significance (the dreaded VUS!). King suggests that for this purpose, we report only unambiguous loss-of-function mutations. Another concern is that our empiric data about the risk of cancer associated with these genetic findings comes primarily from families with a history of cancer – suggesting a possible ascertainment bias. For that reason, the US Preventative Services Task Force recommends testing only in at-risk populations. In response King cites a new study out of Israel documenting increased cancer risk in women with one of three Ashkenazi Jewish founder mutations even among women with no family history of cancer (and amazingly, 50% of all women identified with a mutation had no history of breast or ovarian cancer that would have brought them to clinical attention).
But, as King herself recognizes, “major questions arise in generalizing from the results of the study in Israel to population-based screening in the United States or any other country.” That study reviewed only 3 very well-characterized mutations in a single ethnic group with an equally well-characterized risk for breast cancer. It is a leap to assume the results are valid for all variants and all populations. Still the likelihood is high that mutations that disable BRCA 1 or 2 will cause at least some significant increase in cancer risk. Should we be advocating for universal screening?
Well, there are a couple of negatives to consider. First, King rightly notes that we can’t hand out results on a huge scale when we aren’t certain what they mean. So that means we are only looking at a subset of BRCA results with the potential to increase cancer risk. Furthermore, her plan reduces the complexities involved in cancer counseling in that it looks only at BRCA 1 and 2, and not at any of the other genes that contribute to overall cancer risk. While this makes universal screening much simpler, it also limits the effectiveness. So while you might make a ‘something is better than nothing’ argument for screening, you have to wonder if this quick and dirty approach will end up replacing a more thorough risk analysis for many women in the higher risk categories. It’s not easy getting women in for cancer counseling now – are they not less likely to seek out genetic counseling if they have already been ‘screened”? Similarly, will physicians be less likely to refer and insurance companies less likely to pay for a more thorough breast cancer risk assessment and panel testing? Not an issue, perhaps, for families with a striking, definitive cancer history– and never an issue for those with their own resources — but beyond this we run the risk of cannibalizing the cancer counseling programs that we have built, which provide a fuller and deeper approach to of risk assessment.
Second, the program King describes will not generate the data we need to improve our ability to interpret results of genetic testing for breast cancer predisposition. One of the goals if universal testing ought to be that an investment now would get us out of the too-much-data-too-little-interpretation hole we are stuck in today. On Twitter, response to King’s essay buzzed about the prospect of a massive database of variants:
But returning only information on variants we have already characterized will not enable us to make any headway on classifying the rest.
Universal BRCA 1 and 2 screening is an enticing idea, and there is no doubt that it will identify tens of thousands of women at increased risk for breast and ovarian cancer, for which we have preventative measures with proven efficacy. But a program at this scale will not provide assessment or counseling at anything like our current standard of care, and the potential harm that could cause for families with mutations in BRCA 1 and 2 as well as the other cancer-related genes that will not be reported must be weighed against what we can accomplish; to that end, a pilot study would put some numbers on the scale. Furthermore, as a project universal screening represents a significant investment in public health. It makes sense for us to consider whether or not such an investment moves the ball forward in terms of improving genomic interpretation, and King’s plan, as envisioned, does not.
Follow me on Twitter!
5 responses to “Mary-Claire King (winner of the 2014 Lasker Prize!) proposes a plan for universal BRCA 1/2 testing that is both TOO MUCH and TOO LITTLE”
I think that while universal screening may be a worthy goal, it should not be limited to breast cancer or cancer in general. Hypertrophic cardiomyopathy (as well as other hereditary cardiac conditions) is as common as HBOC–if it is undetected it can have more deadly consequences than breast cancer and when the predisposition is known it is often manageable. If we are going to screen healthy populations, we need to look at what hereditary conditions are available and what screening would be the most beneficial, and I am not sure that BRCA (although I spend my life talking about it) is the area where we will save the most lives. It should be part of a broader discussion about which conditions should be part of a panel for screening healthy populations.
I didn’t mean to give the impression that this magic database would come for free with lots of sequencing (limitations of Twitter and all that) – obviously this would need to be an effort in which large-scale sequencing was coupled with recording of all variants (regardless of whether or not they were reported back to subjects as actionable) and clinical follow-up to see which individuals ultimately did come down with cancer 5, 10 or 15 years after testing. That way all variants, even VUS, are ultimately linked with clinical annotation.
Perhaps this is just impossible to implement in the US system, but it certainly seems feasible in other populations (Scandinavian nations spring to mind as the most obvious candidates).
Obviously there’s a cost/benefit calculation that needs to be done. But it’s worth emphasizing that if we restrict screening to families affected by cancer we will *never* have accurate penetrance estimates for variants (these will always be inflated by ascertainment bias). The only way to build an accurate database of pathogenicity and penetrance is to do prospective screening of everyone and follow up their clinical outcomes. This seems to me to be a goal worth building towards, even if it’s impossible right now.
I agree with you completely about the value of testing in controls to determine true empiric risk estimates — that is why I don’t want to see us spend tons of money on a universal screening program that would not provide that data (essentially because we cannot or will not include counseling, without which we can’t return uncharacterized variants). I spoke years ago with someone on the Board at Myriad who expressed frustration that they never did studies in families without a cancer history, but of course it was an unnecessary expense to them, and it raised the specter of lowering the generic breast or ovarian cancer risk associated with the genes. Horrors.
I think the most likely scenario is that we don’t get that unbiased collection of variant data until we move to population-wide WES or WGS, at which point it will be easier — of course, at which point we will be wishing we run these experiments earlier, so we had better information to give out. But, it will sort itself out in time. And the incremental costs will be small, as posed to this plan, which is a huge investment to make, and still provides no improvement in our interpretive abilities.
I wonder if the “universal” screen couldn’t be piloted on a relatively small population and thereby give some insight into the challenges–logistic, informatic, clinical, etc–of scaling up? The Prospective Registry of Multiplex Testing (PROMPT) is doing something akin to this retrospectively with other HBOC genes…though of course BRCA1/2 are not included (gee, I wonder why!).
Who would pay for such a venture? I don’t know…how about NCI, NHGRI and a consortium of all the labs that jumped into this space post-SCOTUS?
Presumably that’s just what Craig Venter is going to do with his 100k genomes a year program in San Diego.