Overdiagnosis of Breast Cancer Is Real

An Interview With Dr Archie Bleyer

Tricia Ward; Archie Bleyer, MD


August 21, 2015

Editor's Note: Dr Saurabh Jha guest-edited the August 2015 print issue of Academic Radiology devoted to overdiagnosis in radiology. He invited two authors to contribute opposing viewpoints on breast cancer screening. Dr Archie Bleyer, interviewed below, has published analyses showing significant overdiagnosis of breast cancer. To read the interview with Dr Stephen Feig, who does not believe that there is any clinically significant overdiagnosis, click here.

Medscape: You published a controversial analysis in the New England Journal of Medicine[1] with Dr H. Gilbert Welch that brought the subject of overdiagnosis and breast cancer screening to prominence. What prompted you to look at this in the first place?

Dr Bleyer: It started in the 1990s when I was asked to conduct a statewide screening program in Texas to screen young children for neuroblastoma (the scope of the study was broad), and the funding was $1 million. As the principal investigator, I was responsible for making sure that it was successful, but we had to abort the study because we learned that we were overdiagnosing children who didn't have cancer. We thought they had neuroblastoma and were treating them with surgery and, in most instances, chemotherapy; some even had radiation. And we did them harm.

That is when I learned about overdiagnosis. In the late 1990s, I started working with adolescents and young adults (up to age 40 years) with cancer. In fact, this is now my primary professional goal. For the female population, we obviously had to answer questions about screening as they approached their 40th birthday and were being asked to prepare for breast cancer screening. That is how I got from infants to adults, and then women between 40 and 75 years led me to learn about overdiagnosis of that disease— or, as it turns out, nondisease, because that is the problem of overdiagnosis.

Medscape: There have been criticisms of your original paper, some of which were published on Medscape. One was regarding your use of 0.25% as the background annual increase in the incidence of cancer vs 1.0%, Can you address that?

Dr Bleyer: The harshest criticism was that Gil Welch and I were alleged to have ignored the data, which couldn't be further from the case. We were very careful about the background incidence. The estimated background incidence over the past 30 years during which screening was applied in the United States has been assumed by a number of investigators to be high. The argument goes that if we didn't correct for an increasing incidence of breast cancer, our estimates of overdiagnosis would be too great. The real question is, what is the background incidence to make that correction?

Those who have criticized us want to use a rate of about 1% per year; that was based on a trend, between 1940 and 1980, which we looked at very carefully. That trend was based primarily on the state of Connecticut, which represented about 2% of the United States at the time.[2]

There were two problems with going back to the 1940 to 1980 era. One was that it was no longer applicable, as we have subsequently shown.[3] Second, that one state representing 2% of the United States was a very small sample size. We were very careful and tried to use the most accurate background incidence, which was approximately one fourth of what others thought to be the rate.

Depending on the background rate, you can come up with an overdiagnosis rate of zero—ie, it's not a problem—or you can get much higher rates than we projected—ie, it's a major problem.

I think the criticism was unfair, and we tried to address it in the article in Academic Radiology.[4] If you use a much larger sample size than Connecticut, and more contemporary background rates, we would be justified in the estimates we used.

Even so, we did an analysis with the 1% background rate increase in this updated review and still found an overdiagnosis rate of 12% (instead of 30%). That is 12% of women having no breast cancer but being treated for it with radiation, chemotherapy, hormonal therapy, and surgery. We estimated that to be 31,000 women per year. Even if you think we used the wrong incidence background rate, overdiagnosis is still a problem.

I have been personally attacked in the literature. I have been unfairly adjudicated by colleagues.

Medscape: There have been a wide range of estimates published for the rate of overdiagnosis. Critics say that inadequate adjustment for lead time (time from screened earlier diagnosis to survival advantage) and follow-up explains the discrepancies. Do you think you addressed these?

Dr Bleyer: We were very aware of that. The best report published in the past few years on how long you have to wait to adjust and correct for lead time says it's 30 years.[5] We did, in fact, wait 30 years because we covered the era from 1976 to 2009, and I updated it to 2012 recently. That is well over 30 years, so we don't think lead time was ignored. In fact, we more than compensated for it using the most conservative estimate.

In our New England Journal of Medicine article we reported all of the overdiagnosis estimates that were available at the time. There have been a couple published since and there were a total of 12 others. The estimates range from a few percent (0.5% was the lowest) to as much as 75%—huge variability. Our estimate of 30% or 31% was right down the middle of that range. We didn't feel that we were out of line and that we were overestimating or underestimating

In the article in Academic Radiology, I made an effort to explain why there is such a variability and why it is easy to underestimate the overdiagnosis rate.

Medscape: Your analysis is not a randomized trial and therefore is subject to confounding. Critics point out that the baseline screened population is not necessarily the same patients assessed for the outcome endpoints.

Dr Bleyer: Well, I'm not sure that we didn't evaluate the same population. It was the United States SEER [Surveillance, Epidemiology, and End Results program of the National Cancer Institute] population, which is followed until those women die. They are followed as much as possible annually.

We did not look at death rates. We didn't estimate the overdiagnosis rate on the basis of mortality. The objective was to look at whether women who are diagnosed with breast cancer because of screening are not diagnosed with advanced disease and are spared death, subsequently, but we didn't look at the death rate per se. We evaluated whether or not diagnosis with early-stage breast cancer translated into avoiding the worst kinds of cancers. We couldn't find that there was a benefit in preventing metastatic disease beyond the regional area around the breast. We were concerned with the first step, which was to prevent advanced disease, because that is what kills women.

Medscape: Going back to the criticism that only a randomized controlled trial can assess the question of the benefits of screening—can you speak to that?

Dr Bleyer: I have been reluctant to accept this, but we need a randomized trial because of the value of those in sorting out these kinds of problems. The randomized controlled trials of breast cancer screening were all done between 30 and 50 years ago. Whether or not they were done perfectly, those results were from an era when breast cancer treatment was very different.

In fact, breast cancer wasn't even talked about much in those days. Those trials are just not relevant today. We have had such incredibly successful results in treatment. I have been doing randomized clinical trials in children since 1969, and I know that the randomized trials of the past don't apply, in many instances, to the situation today.

But we still go back to the randomized trials of the past—none of which, by the way, showed an overall mortality benefit [for breast cancer screening]. I wish we could do a randomized controlled trial today, but the problem is that it would take 20 or 30 years to answer the question definitively because of lead time. Additionally, there is such polarity right now that we couldn't agree on a trial that would be required to answer this.

Medscape: Do you mean that women wouldn't agree to be randomly assigned to no screening?

Dr Bleyer: Actually, I meant that the investigators would probably find it difficult to agree on how to do it—at what age, and the intervals of screening and so forth. But I think there is also the problem you brought up, of women questioning whether to enter into this trial. Isn't it well established that screening is beneficial, and am I going to be at risk if I participate?

Being able to accrue women would be difficult. It depends on how we, the medical profession, solve the problem among ourselves before we go to the public. I see that taking many years. This is a highly inflammatory topic.

Medscape: Is your main concern about excess screening, that you would identify smaller cancers that wouldn't amount to anything in the end?

Dr Bleyer: In part. Screening detects early disease when it is successful. Reducing the amount of therapy because the disease is caught earlier is a true benefit.

I want to make sure that you understand that we are not saying that there are no benefits of breast cancer screening. They are real and women's lives are saved, or at least deaths from breast cancer are reduced. The death rates overall have never been shown to be different in the randomized trials; these women don't die of breast cancer but then they die of something else.

What we tried to address is that the benefits have been overestimated and the problem of overdiagnosis has been minimized. It is not fully understood or appreciated, and unfortunately it is not explained to women when they undergo screening, so that they know there is a chance (we can debate what that chance is) that when they are diagnosed with breast cancer as a result of screening that it is not cancer. Even so, it is called cancer and they are treated for it, which is the unfortunate and sometimes tragic result, given all that therapy and the psychological stress that goes with it.

In my community—a private-practice environment, having left the academic medical world after almost 45 years—I see very little effort to explain overdiagnosis to women being screened.

Medscape: You support screening but beginning at age 50 years vs 40 years; is that correct?

Dr Bleyer: Yes. Having worked with the US Preventive Services Task Force on breast cancer screening, I concur with their 50-years start age and to screen every two years or at less frequent intervals, which is what most of the rest of the world does. A lot of independent panels have supported the idea of starting screening at 50 years and doing it every other year.

I definitely want women of average risk to have the chance to have breast cancer detected earlier. I want to make that as clear as possible. In the criticism that we have received, there seems to be an assumption that we are against screening. We are not against screening. We are against the lack of information given to women about the harms and potential risks of screening.

Medscape: How should women be informed of these potential risks? Through educational materials?

Dr Bleyer: It is more than materials. First, it is understanding the concept, which is really difficult. It took me 15 or 20 years. I started the neuroblastoma screening program and it resulted in hurting young children. I didn't realize that that would happen; it only became apparent after many years. This is a difficult concept, which is why the debate continues. Then it is explaining that to the public, to women, to the professional organizations—especially in oncology—so that they can respect the potential harm and, in turn, create the materials and ways in which to inform women of this risk.

There are places in the world (the United Kingdom and Australia), as mentioned in my review, that have developed educational materials, and there is even a randomized trial of some of the materials that were created.[6,7] But the first step is to understand the problem better.

Medscape: At what stage do you see this discussion taking place? Is it about whether to undergo screening or is it about whether to undergo treatment for something that may not be a real cancer?

Dr Bleyer: It is appropriate at both stages, starting with screening or not and at what age, and along with that how often to be screened. Why is the United States the only country in the world that still does annual breast screening, even though the US Preventive Services Task Force has recommended every other year for women of average risk? And then when a diagnosis of breast cancer is established because the woman had a positive scan and underwent a biopsy, the report from the pathologist says carcinoma, whether it is ductal carcinoma in situ (DCIS) or invasive carcinoma. At that point, additional consideration should be given as to whether that is true cancer.

Medscape: How would you like to see DCIS managed?

Dr Bleyer: The problem that I see with DCIS is its name. We could change the name, to eliminate the idea that it is cancer by not putting "carcinoma" in the diagnostic term and calling it something else, like "indolent lesion of epithelial origin," with the emphasis on "indolent." There is even an acronym for that—IDLE. This way the woman isn't scared immediately by seeing "carcinoma" in the name.

That hasn't occurred in the United States despite the recommendation being several years old, starting with Dr Laura Esserman at the University of California-San Francisco.[8] So the idea that that may not be cancer has still not been achieved within the naming of the entity.

The British National Health System asks all physicians to inform women with that diagnosis that there is a threefold higher chance of being overdiagnosed with cancer than having their life saved. The figure is probably higher in the United States because we screen much more intensively. In the United Kingdom, they screen every 2-3 years starting at the age of 50, and we do it annually starting at the age of 40; we look much harder for breast cancer. We find much more DCIS. However, I am sorry to say that that discussion is not occurring.

Medscape: What about the argument that a lot of DCIS will turn into aggressive cancer with long enough follow-up?

Dr Bleyer: There is a common belief that DCIS will develop into cancer, but that is not correct if you go back and look at the original data. You have to go back many years, several decades, because since then all DCIS has been treated. I have done surveys nationally of experts, surgeons, medical oncologists, and oncologists, and all of the DCIS is treated.[3] There is this fear that if they don't treat it, it will progress. So we don't have any recent data.

When you look at the original data, there is something very interesting that I don't think is appreciated sufficiently. Almost all of the cases that were said to progress to cancer were in women who had symptoms. They had a lump, discharge, discomfort, pain. Those were women who were diagnosed on the basis of symptoms and signs. They were not of average risk.

So the notion that DCIS progresses to cancer in almost every case is based on two reports from the '70s or '80s, before we started treating everybody who has DCIS.

These old data didn't evaluate asymptomatic women at average risk and who had no evidence of cancer at the time. On the contrary, they evaluated women who had DCIS because they had an indication for a biopsy. They had a lump, in most cases. It is a very different population, and to continue to this day to say that all DCIS progresses to cancer is incorrect.

There is another line of evidence for this and it is fascinating: Women who die of other causes and who had autopsies (again, this goes back many years because we do autopsies infrequently nowadays) were found to have DCIS in many cases. These could not all have progressed to breast cancer because the proportion was too high.[9]

I learned the same thing with neuroblastoma in children. Many children have neuroblastoma if they are autopsied after having died of sudden infant death syndrome or some other cause, and that "neuroblastoma" is not true neuroblastoma. In those cases, it disappears spontaneously.

The notion that some breast cancer disappears on its own without therapy is in the early data, whether it is autopsies or clinical studies. The question is, what proportion progresses to cancer? That is where the hope of the future is. If we can find a way of predicting which DCIS will go away and never hurt the woman in her lifetime, either by molecular testing or genomic testing, that is the real hope. I truly believe, at this point, that we will find a way of avoiding having to treat everybody with DCIS. Dr Esserman herself told me it would take 5 years to solve this problem, and that was 3 or 4 years ago.

If we can't avoid treating DCIS completely, at least we can cut back to a minimum therapy that won't hurt the woman very much because her DCIS is unlikely to progress to cancer. Sorry for the long answer, but the criticisms have been pretty hard-hitting.

Medscape: Dr Jha, who guest-edited the issue of Academic Radiology, said, "Speaking about overdiagnosis in breast cancer is like saying 'Voldemort.'"

Dr Bleyer: I refer to it as the third rail. I had no idea how dangerous this topic was, coming from pediatric oncology where the most controversial topics were discussed openly, respectfully. I can remember many controversies that still exist today, but they were dealt with in a gentle and considerate fashion by my colleagues.

When I touched this third rail I had no idea what it would be like. I have been personally attacked in the literature. I have been unfairly adjudicated by colleagues. In my 50 years of medical research this is a new phenomenon for me.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: