Introduction: An unexpected e-mail arrives
One of the consequences of the growing traffic and prominence of this blog over the last few years is that people who would otherwise have probably ignored what I or my partners in blogging write now sometimes actually take notice. Nearly a decade ago, long before I joined this blog as a founding blogger, if I wrote a post criticizing something that a prominent academic said, it was highly unlikely that that person would even become aware of it, much less bother to respond to whatever my criticism was. I was, quite simply, beneath their notice, sometimes happily, sometimes unhappily.
It appears that those days might be over. Last week Dr. Daniel Kopans, a prominent Harvard radiologist and well-known long-time defender of screening mammography, sent me a rather unhappy e-mail complaining about my “attack” on him on this blog, a charge that he repeated in a subsequent e-mail. Before I publish his initial e-mail verbatim (with his permission), I would like to point out that, while it’s true that I did criticize some of Dr. Kopans’ statements rather harshly in my post about the Canadian National Breast Screening Study (CNBSS), even characterizing one statement as a “howler,” I would hardly characterize what I wrote as an “attack.” That to me tends to imply a personal attack. Using Dr. Kopans’ apparent definition, what he has said and written about investigators like those running the CNBSS, as documented in my post, about H. Gilbert Welch, who published a large study in 2012 estimating the extent of overdiagnosis due to mammography, and the U.S. Preventive Services Task Force (USPSTF), the group that in 2009 suggested changing guidelines for routine screening mammography in asymptomatic women to begin at age 50 instead of age 40, would appear to also qualify as “attacks.”
Be that as it may, I also wondered why Dr. Kopans hadn’t noticed my CNBSS post until more than three months after it had originally appeared. Then, the day after I received Dr. Kopans’ e-mail, my Google Alert on mammography popped up an article in the Wall Street Journal by Dr. Kopans entitled “Mammograms Save Lives: Criticism of breast-cancer screenings is more about rationing than rationality.” That’s when I guessed that someone probably had either posted or e-mailed Dr. Kopans a link to my previous post in response to that article. Given the confluence of events, I think it’s a perfect time to discuss both Dr. Kopans’ e-mail and his article, because they cover many of the same issues.
Before I go on, I feel obligated to repeat my mandatory disclaimer for posts about the controversy over screening mammography: What I am talking about is screening of asymptomatic women with mammography to detect breast cancer. If you have a lump or other breast symptoms, none of this discussion applies, and you should have your lump or other symptoms investigated by your doctor and undergo whatever imaging is deemed medically appropriate.
Dr. Kopans is displeased with Dr. Gorski
Here is the text of the e-mail from Dr. Kopans that arrived on Thursday:
Dear Dr. Gorski:
I was astonished and disappointed when I came across your “Posts” concerning breast cancer screening and your attacks on me. In particular the inflammatory statement that I was accusing the Canadian investigators in the Canadian National Breast Screening Study of “fraud”. I have never stated this and it was inappropriate for you to suggest it. There is no evidence of fraud in the CNBSS trial, but there is clear evidence that the fundamental requirements of a randomized, controlled trial were violated. I do not make up the rules for these trials. You are well aware that blind allocation is critical. When there is a statistically significant excess of advanced cancers allocated to the screening arm and the coordinators had knowledge of the clinical examination findings and they allocated the women on open lists, and there was an excess of cancer deaths among the screened women you have to completely suspend scientific thought to not be concerned that there was a problem.
On top of that when there is not a single expert who does not agree that the mammography was poor to unacceptable and the size of the cancers was the same for the mammography group and the non-mammography group why would you even consider this a test of mammography screening? If you were involved in a trial and you felt that you had done a good job, you would defend your work against all criticism. I am unaware of any radiologist or technologist in the CNBSS who has defended the trial. I assume you do breast surgery. Dr. Miller has insisted that the nurses doing the clinical breast examination detected most of the cancers in the mammography arm. How often can you feel the cancers detected by mammography even when you know where to feel? How often do radiologists have to guide you to the lesion?
How can you defend a trial that has major implications and fundamental compromises by saying that we will simply get rid of the prevalence cases? Since it was not blinded, we have no idea how the allocation was compromised. Furthermore, Dr. Miller has acknowledged that as many as 25% of the control women had “diagnostic’ mammograms so they were not really contaminating the control arm. As I am sure you know, diagnostic mammography starts off with a bilateral screening mammogram so this is simply not true. The trial started with insufficient power (I am certain as an expert who is publically critical of a critic, that you have read all of the papers on the CNBSS ??)
Dr. Miller went public in the 1990’s claiming that mammography was squeezing cells into the blood leading to earlier deaths in the mammography arm. After causing major concerns, he withdrew the claim invoking ataxia telangiectasia, but now says it was because removing the primary allowed the metastatic lesions to grow faster. Even in mice where this happens it is a short lived phenomenon and the mice do not die any sooner – a point that Prof. Miller neglects to make.
If all of the problems with the CNBSS had occurred in a surgery or chemotherapy RCT you would be outraged that the results made it through peer review. You seem to have no problem defending a corrupted (in scientific jargon) trial when it comes to mammography screening. I have little doubt that your ego will not allow you to recognize your hypocrisy.
By the way – I am attaching my critique of the Bleyer and Welch paper. Dr. Bleyer has admitted in his rebuttal that they used “best guesses”. He also admitted that since they had no data on mammography, they could not fault mammography – yet they did. He is critical of my use of the data from the Connecticut tumor registry, yet the papers that I can find that looked at breast cancer incidence prior to SEER all used the Connecticut tumor registry. If you use the Connecticut data there is no overdiagnosis of invasive cancers (there were actually fewer than expected)
DCIS is a major unresolved issue, but this is well known. The Bleyer and Welch paper would have never been published had they only raised concerns about DCIS so they combined it with small invasive cancers just the way all experts combine the two – not actually.
I assume that since you defend the Bleyer and Welch paper you have seen a number of invasive cancers disappear on their own. Certainly with 70,000 each year you must have seen many such disappearing cancers.
If you are so concerned about overtreatment (over [sic] course overdiagnosis is the pathologists’ failing) why are you (oncologists) poisoning so many women who do not need it? Among women diagnosed with breast cancer (all palpable) in the 1940’s treated only by mastectomy 30% were alive 30 years later, yet all would be treated with systemic therapy today. Why are you overtreating all these women?
There are numerous legitimate concerns about breast cancer and breast cancer screening, but it is not reasonable to withdraw support for a test that has been shown with greater certainty than any other screening test to save lives, by supporting flawed “science” and misleading physicians and the public.
Daniel B. Kopans, M.D.
Professor of Radiology, Harvard Medical School
Senior Radiologist, Breast Imaging Division
Department of Radiology, Massachusetts General Hospital
Avon Comprehensive Breast Evaluation Center
I must admit that I was disappointed in this response.
Dr. Gorski responds to Dr. Kopans, with trepidation
It’s always a bit of a wake-up call to be so castigated by such a prominent radiologist. Did I err? What makes me think I have what it takes to get into a serious discussion with the likes of Dr. Kopans? After all, what’s a relatively unknown surgeon and translational researcher compared to such an eminent radiologist as Dr. Kopans? Still, one of the greatest things about the Internet (and simultaneously the one of the worst things, as I have acknowledged), is that it lets anyone with an Internet connection voice his opinion. So here goes.
First, however, let me point out that there is actually some common ground between Dr. Kopans and me, in the hopes that he will understand that I am not his enemy, merely a critic of certain statements of his. There is a difference. Also, contrary to how I am sometimes portrayed (and what Dr. Kopans seems to be implying), I am most definitely not in the camp that believes that mammography screening programs should be abandoned, as has been proposed by various researchers and pundits, and I do believe that mammography saves lives. I’m even the team captain for our cancer center’s team in the local Komen Race for the Cure! Where I differ with Dr. Kopans is that I also accept the emerging evidence that overdiagnosis due to mammographic screening is a significant problem that leads to overtreatment. In other words, in some women mammography can cause harm, and more harm than Dr. Kopans would apparently acknowledge, and I believe that our current guidelines should be changed to reflect that emerging science and find the “sweet spot,” where benefits are maximized and harms minimized. Having looked into this for nearly five years now, I understand the acute discomfort of change that comes from questioning a medical intervention that I once accepted and finding that I agree that the guidelines do need to be changed.
Next, let me deal with what I view to be the most heartening revelation, namely that Dr. Kopans denies accusing the CNBSS investigators of research misconduct. This is a good thing, because such charges should never be made lightly. It is odd to me, however, how clearly puzzled Dr. Kopans appears to be that a reasonable person might have honestly interpreted his rather over-the-top characterization of the CNBSS randomization process that way, which makes me feel reluctantly obligated to explain again how that could have happened and requires that I quote some of Dr. Kopans’ own words again:
In order to be valid, randomized, controlled trials (RCT) require that assignment of the women to the screening group or the unscreened control group is totally random. A fundamental rule for an RCT is that nothing can be known about the participants until they have been randomly assigned so that there is no risk of compromising the random allocation. Furthermore, a system needs to be employed so that the assignment is truly random and cannot be compromised. The CNBSS violated these fundamental rules (6). Every woman first had a clinical breast examination by a trained nurse (or doctor) so that they knew the women who had breast lumps, many of which were cancers, and they knew the women who had large lymph nodes in their axillae indicating advanced cancer. Before assigning the women to be in the group offered screening or the control women they knew who had large incurable cancers. This was a major violation, but it went beyond that. Instead of a random system of assigning the women they used open lists. The study coordinators who were supposed to randomly assign the volunteers, probably with good, but misguided, intentions, could simply skip a line to be certain that the women with lumps and even advanced cancers got assigned to the screening arm to be sure they would get a mammogram. It is indisputable that this happened since there was a statistically significant excess of women with advanced breast cancers who were assigned to the screening arm compared to those assigned to the control arm (7). This guaranteed that there would be more early deaths among the screened women than the control women and this is what occurred in the NBSS. Shifting women from the control arm to the screening arm would increase the cancers in the screening arm and reduce the cancers in the control arm which would also account for what they claim is “overdiagnosis”.
He also went on to accuse the investigators of the CNBSS of intentionally ignoring an issue that would allegedly allow those doing the randomization to subvert the randomization. One notes that the American College of Radiology (ACR) response to the publication of the CNBSS uses very similar language, leading me to believe that Dr. Kopans had a major hand in writing it. If you don’t believe me, compare Dr. Kopans’ criticism of the study in the BMJ that I quoted in my original post and the ACR press release criticizing the study. The wording of the relevant sections about subversion of randomization is almost identical.
Given its virtually-identical charges against the investigators of the CNBSS, I find Dr. Jen Gunter’s discussion of the charges quite informative. She read the ACR press release and found its arguments…wanting. She even entitled her post Did the American College of Radiology accuse the BMJ mammography authors of misconduct? In her post, she looked at it a different way than I did when I wrote about the CNBSS, a way that to me is simultaneously simpler but probably more revealing than how I did it, by pointing out that the CNBSS authors stated in their paper that their randomization was blinded (true, as I pointed out). She then put it in context:
So the authors are saying their randomization was blinded and the ACR’s counter-claim is that is couldn’t have been. Both can’t be right. The ACR is either accusing the author of lying or saying he had rogue study nurses who didn’t follow protocol. The ACR does not provide any references to support this claim.
This last part of the ACR claim sounds a lot like school yard taunt, “You lied. How do I know? Because I said so.” It is the kind of claim one makes when one A) hasn’t thought it all through and is letting emotions rule, B) has an otherwise weak argument that needs to be bolstered, or C) you want to be over the top to get page clicks. This accusation actually leads me to evaluate the ACR’s other two claims with greater scrutiny.
One notes that an accusation that is in essence an accusation that the investigators lied in their scientific papers about the randomization process of their study or accusation that (1) they had rogue nurses who ignored study protocol in order to subvert randomization and (2) that they ignored evidence that their nurses were subverting randomization in that way is an accusation of fraud in my book. I further note that Dr. Gunter wrote both the Canadian and American Colleges of Radiology for evidence to support this accusation and got none. True, the ACR did respond, but provided no evidence. As I pointed out in my original article, I believe the authors did this adequately in both their paper reporting the results of the CNBSS and another article in which they answered charges like those leveled by Dr. Kopans and the ACR, and other reviews of the CNBSS failed to find convincing evidence for subversion. In any case, the vitriol in the debate over mammographic screening, particularly of women in their 40s, is nothing new.
While I do believe that Dr. Kopans does not think he was accusing the CNBSS investigators of research misconduct/fraud when he made those criticisms of the CNBSS randomization process, I can’t help but also note that back in the 1990s he did explicitly accuse other investigators of “manipulat[ing] data in a fashion that borders on scientific fraud” in their article in JAMA. He even sent letters with this accusation to George Lundberg, then editor of JAMA, and Haile Debas, then dean of the UCSF medical school. Around the same time he characterized a recommendation by an NCI panel not to recommend screening for women in their 40s as “fraudulent” and “meant to mislead.” Even Laszlo Tabar, for instance, a radiologist who directs the Swedish Two County Trial of mammography, and a strong defender of mammographic screening programs whom I quoted in my post about the CNBSS, admitted in 1997 in the same article in Science that Dr. Kopans can be overly belligerent.
In any case, I suspect that Dr. Kopans is so convinced that mammography is such an unalloyed good that he does not react well to criticism of it and that Dr. Kopans and the ACR, honestly believing the CNBSS to be flawed beyond any redemption, wanted to show just that, to the point that they let their rhetoric get away from them. The bottom line is that at least two reasonable people (Dr. Gunter and myself), both of whom work in medical fields in which we have every reason to support mammography, were left with the impression that Dr. Kopans’ and the ACR’s criticisms of the CNBSS report in the BMJ were accusations of research misconduct. If there are two of us who interpreted it that way and were willing to blog about it, I’d be willing to bet that there were a lot more of us out there who interpreted his words that way. One hopes that Dr. Kopans will rein in his overheated rhetoric in future discussions and not be surprised how, given his history, his words are interpreted.
But what about the rest?
A lot of it is what has become fairly standard anti-CNBSS boilerplate that’s been repeated before over and over again, such as the claim that the mammography was substandard. This issue has been addressed before, as I discussed. As for the issue of Dr. Miller (the principal investigator of the CNBSS), even if true it’s irrelevant to whether the CNBSS results were valid or not; so I call non sequitur on that one. As for Dr. Kopans’ accusation that I am a hypocrite and such an egotist that I will never admit that I’m wrong, now—as Crocodile Dundee once said about a knife—that’s a personal attack. (True, it’s in an e-mail, but Dr. Kopans gave me permission to post it; so it’s not as though he’d be likely to shy away from saying it in public.) It’s unlike any criticism I wrote in my original post and anything that I’ve written in this post. He’s also wrong. If I saw a surgical study with flaws correctly judged as massive as the flaws portrayed in the CNBSS by Dr. Kopans, I quite likely would comment. It’s just that I don’t agree with Dr. Kopans’s characterization of the magnitude of the flaws of the CNBSS. In any case, I leave the reader to make his or her own judgment reading my original post, Dr. Kopans’ response, and this post, particularly in light of his denigration of my qualifications with respect to his, as I move on to other substantive points.
The Bleyer and Welch paper, as regular readers might remember, was published in the New England Journal of Medicine about a year and a half ago. I discussed it in detail when it first came out. One notes that my assessment was not quite as effusive as Dr. Kopans portrays it. I thought its findings were in line with the emerging body of evidence suggesting that overdiagnosis of breast cancer is more common than previously suspected. A key prediction of the hypothesis that detecting breast cancer earlier will lead to fewer deaths from breast cancer is that a screening test that leads to the detection of more early-stage cancers—and, yes, it is, contrary to Dr. Kopan’s assertion, reasonable to include ductal carcinoma in situ (DCIS) among early-stage breast cancers if one believes that a significant number of DCIS lesions will progress to invasive cancer—should lead to a decrease in the number of late-stage cancers. Bleyer and Welch didn’t find that. They found a large increase in the incidence of early-stage cancers but a much smaller decrease in the incidence of late-stage cancers. That’s it.
It’s not as simple as Dr. Kopans makes it out to be
What I’ve found since I first took a more intense interest in this topic is that many of the criticisms of such studies boil down to not believing that overdiagnosis is a real phenomenon in breast cancer. Yet, as I’ve explained before, translating earlier detection of breast cancer—or any cancer, for that matter—into better survival is far more complicated than most people, even physicians, think. Besides overdiagnosis, there are the complicating factors of lead-time bias (explained by yours truly here, here, and here) and length bias, in which the tumors detected by screening tests tend to be the more indolent and less deadly (explained by yours truly here and here). In the past, I have pointed out that the prevalence will increase for virtually any disease that is intensely screened for, using autism as my main example, as well as hypercholesterolemia and hypertension, and comparing it to the prevalence of DCIS, the latter of which has increased by 32-fold since the 1970s. Given that autopsy series routinely show that large percentages of people who die of other causes have foci of cancers such as prostate, breast, and thyroid cancer that never become clinically evident, why is it a surprise that mammography might lead to at least some degree of overdiagnosis?
Finally, I can’t help but note that a couple of observations Dr. Kopans makes are just silly, disappointingly so. (No, that’s not a personal “attack,” simply a characterization of my opinion of what he wrote. If Dr. Kopans can call me a hypocritical egotist, then the least concession he can give me is to let me refer to something he writes as “silly.”) First, there’s the part about how I must have seen so many tumors disappear on their own. I’m sorry, but there’s no polite way to say it: That statement is based on a complete misunderstanding of the concept of overdiagnosis. Those studies only include screen-detected tumors, but more importantly the current standard of care would be to treat all such tumors detected, meaning that every woman I see who has a screen-detected breast cancer will receive some combination of surgery, radiation therapy, chemotherapy, and/or hormonal therapy. I do not simply watch women with invasive breast cancer to see how many of them might regress. Because it would be completely unethical to propose doing so in a randomized clinical trial because of the lack of clinical equipoise in such a design, overdiagnosis must be inferred from population-level data, and it has been. Moreover, it’s important to note that it is currently not possible to identify which woman has been overdiagnosed at the time of diagnosis. I wish it were, because then we could obviate the major harm in mammography screening, and virtually eliminate overdiagnosis. Figuring out strategies to do so for DCIS has been a recent interest of mine as well.
Let’s go back to something that Dr. Kopans also wrote:
If you are so concerned about overtreatment (over course overdiagnosis is the pathologists’ failing) why are you (oncologists) poisoning so many women who do not need it? Among women diagnosed with breast cancer (all palpable) in the 1940’s treated only by mastectomy 30% were alive 30 years later, yet all would be treated with systemic therapy today. Why are you overtreating all these women?
In fact, I have discussed many times how it is important to provide women with the absolute benefits due to chemotherapy as well as the relative benefits. I like to use the example of a woman with a stage I cancer. I tend to round the numbers a bit for simplicity’s sake, but they’re close enough to make the point to a lay audience (and medical students, for that matter, when I’m teaching). So I tend to give an example of a woman with a stage I cancer who has a 90% ten-year survival with local therapy alone. If chemotherapy produces a 30% relative survival benefit (roughly correct, perhaps slightly too high), that would translate only to a 3% absolute survival benefit. In other words, 90 out of 100 women would have survived anyway but got treated, in retrospect “unnecessarily,” while seven women out of 100 would have died regardless but got treated, in retrospect “unnecessarily.” Only three out of a hundred who would have died were saved by chemotherapy. In other words, the number needed to treat (NNT) to save one life is 33. This is the main reason why a hot area of research is predictive testing, to identify who will and won’t benefit from adjuvant chemotherapy, and why we in the field view the OncoType DX test, as imperfect as its predictive power is, as a major advance that has saved thousands of women with early stage estrogen-responsive cancers from chemotherapy who wouldn’t benefit from it.
Of course, it should be noted that the higher the stage, the higher the risk of relapse and death, and thus the larger the absolute benefit of adjuvant chemotherapy. That’s why I’ve done a number of posts in which I take hypothetical cases and run them through Adjuvant!Online to illustrate my points. But I sure don’t sugarcoat problems with adjuvant chemotherapy, contrary to Dr. Kopans’ question. In any case, I’ve pointed out that chemotherapy has contributed to a major improvement in survival of breast cancer patients, even while lamenting how frustratingly…incremental…improvements in chemotherapy drugs have been. I obviously don’t expect Dr. Kopans to have read all my old stuff (that would be ridiculous), but he does seem rather eager to leap to conclusions about me.
It might be instructive to compare the NNT for chemotherapy, which at its worst in stage I breast cancer, is around 33 to 50, to the number needed to screen (NNS) to save one life for mammography. For instance, Esserman et al. estimate that to avert one death from breast cancer with mammographic screening for women between the ages of 50-70, an age range intentionally chosen because it is the age range for which mammographic screening is the least controversial, 838 women need to be screened over 6 years for a total of 5,866 screening visits, to detect 18 invasive cancers and 6 instances of DCIS. The additional price of this was estimated to be 90 biopsies and 535 recalls for additional imaging, as well as “many cancers treated as if they were life threatening when they are not.”
It’s also been noted that the younger the patient group, the higher the NNI (number needed to invite to screening) is, ranging from 377 for women aged 60-69, resulting in an absolute risk reduction of dying from breast cancer in the next ten years of approximately 0.3%, to 1,904 for women aged 40-49, translating to an absolute risk reduction of 0.05%. Of course, mammography is not chemotherapy, but I could still turn Dr. Kopans’ argument around and ask him, why we are screening all these women who do not need it, all to produce an absolute reduction in the risk of dying from breast cancer that is so small? (I realize that chemotherapy is “poison” and mammograms are not, but it is still not an unreasonable point.) I would point out that it’s a value judgment to say how small a survival benefit is too small, but there must be a cut-off. If there were not, there’s no reason not to start screening women as soon as they hit age 21, but not even Dr. Kopans proposes doing that, as, below a certain age, the benefit becomes either too small or nonexistent. The risk of breast cancer with age is a smoothly increasing curve; there is no sharp cutoff. Where we choose the cutoff to begin screening, age 40 or age 50, is a value judgment based on an assessment of the risk-benefit ratio and, yes, to some extent cost. That’s reality. I happen to believe that it’s likely worth it for women over age 50; I’m no longer so sure it is for women under age 50, hence my endorsement of a recommendation for personalized screening below age 50, a concept that is anathema to Dr. Kopans. Either way, his choice of age 40 as the time to start screening is no less arbitrary a cut-off than is age 50, and the observation that cancer mortality in Europe is not higher than it is in the US reassures me that I’m at least not entirely mistaken in my conclusions.
In his e-mails and his WSJ article, Dr. Kopans portrays any questioning of the value of mammography as “a disconcerting effort afoot to reduce a woman’s access to mammography screening for breast cancer by making it seem useless or even harmful,” or similar language in which he ascribes dire motives to those with whom he disagrees. To Dr. Kopans, it seems, there can’t be an honest scientific and medical disagreement that leads some researchers and clinicians to question the “one size fits all” design of the current mammographic screening guidelines. Oh, no. It has to be rationing, and that’s how he portrays it in his WSJ article. When Bleyer and Welch estimate a large amount of overdiagnosis from screening mammography, it can’t be because they go where the science leads them or have a legitimate academic disagreement with Dr. Kopans. Oh, no. It has to be yet another effort to “reduce access to lifesaving screening tests.” Dr. Kopans dismisses as a “new effort to limit access to screening” the reasonable suggestion, based on analysis that screening produces more overdiagnosis the less prevalent a disease is in a population screened (which is the reason why even Dr. Kopans doesn’t propose screening women below the age of 40—breast cancer incidence is too low) that developing risk-based screening could improve the efficacy of mammographic screening programs, reducing overdiagnosis and therefore overtreatment. He even concludes his op-ed, “It would be unfortunate to deny women access to screening based on flawed analyses or rationing in the guise of altruism.”
Disappointingly, that is a profoundly insulting characterization of researchers on the other side of a scientific disagreement, one that is beneath a professor of radiology as eminent as Dr. Kopans.
Those of us who take care of women with breast cancer all want what’s best for them. I accept that Dr. Kopans, pit bull that he is defending mammography against all attacks, does so out of a desire to help women to avoid dying from breast cancer. I just wish he’d grant his critics the same consideration. It’s understandable that he’s upset that I would have the temerity to criticize his statements. He should think of it as a dose of his own medicine—and a comparatively mild dose at that.