In the News

On Cancer: Recent Study Should Not Change Mammography Guidelines

By Media Staff  |  Wednesday, February 12, 2014
Colin Begg, Chairman of the Department of Epidemiology and Biostatistics Colin Begg, Chairman of the Department of Epidemiology and Biostatistics

A study published on Tuesday by the British Medical Journal questioning the benefit of yearly mammogram screenings has gained momentum in the media, while causing confusion among women and concern among experts in the field, including those here at Memorial Sloan Kettering.

The long-running Canadian study contends that annual screening in women aged 40 to 59 does not lower breast cancer death rates. But experts such as Larry Norton, Memorial Sloan Kettering’s Deputy Physician-in-Chief for Breast Cancer Programs, caution women not to be swayed by the headlines. They say the study has critical weaknesses that invalidate its conclusions and that regular mammography screening continues to be recommended for women in this age group.

“This is further follow-up of a flawed study,” Dr. Norton says. “There are no new data here, so it should not influence any changes in screening recommendations.”

The initial Canadian National Breast Screening Study (CNBSS), which began in 1980, was strongly criticized for two reasons. First, the mammograms performed were judged by outside experts to be of extremely poor quality, and the skills of the imaging technologists were deemed inadequate. Second, the subjects used for the study were not properly randomized into two groups of equal risk for comparison.

Colin Begg, who heads Memorial Sloan Kettering’s Department of Epidemiology & Biostatistics, says the Canadian study has always been an outlier. “It conflicts with all the other large randomized trials conducted a generation or more ago, which showed that mammograms lead to a notable reduction in breast cancer mortality,” he explains.

Study is ‘Flawed and Misleading’

Carol Lee, a diagnostic radiologist at Memorial Sloan Kettering who also chairs the Communications Committee of the Breast Imaging Commission of the American College of Radiology, says, “It is not surprising that the update didn’t show a benefit from screening, as the original report did not.”

The American College of Radiology and the Society of Breast Imaging also called the study “incredibly flawed and misleading.” The organizations released a joint statement saying, “The results of this BMJ study, and others resulting from the CNBSS trial, should not be used to create breast cancer screening policy, as this would place a great many women at increased risk of dying unnecessarily from breast cancer.”

The authors of the BMJ study suggest that annual screening mammograms can lead to “overdiagnosis” of breast cancer, causing women to receive treatments such as surgery for early-stage breast cancers that never would have progressed. But Memorial Sloan Kettering experts assert that it is impossible to determine whether a breast cancer will become dangerous and that fears of overdiagnosis are misplaced.

“Women contemplating screening for breast cancer are overwhelmingly concerned about dying from breast cancer, not the various side effects of screening,” Dr. Begg says.

So what should women do? Bottom line: Doctors at Memorial Sloan Kettering agree that annual mammograms starting at age 40 save lives and decrease the likelihood that women will die of breast cancer by at least 25 to 30 percent. 


I disagree with the above commenter. Dr. Lee articulated the concerns of a physician concerned about saving women's lives from breast cancer. A rehashed discredited study clearly out of date provides misleading information. She merely pointed out the fact that garbage in equals garbage out. Not only does early detection save lives, it also allows for less invasive, toxic or disfiguring treatment protocols. We must continue to do long term studies with uncontaminated randomized control groups and base our decisions on the best information out there. Not only has treatment improved, so has mammography, but I fear the recently published study being toted in the press does not reflect these changes in practice.

Let me be the first to say, I am not one to throw the baby out with the bath water. There may be a role for the routine mammogram in certain populations of women in the US. However, for the vast majority of women, I believe it is only a matter of time before routine mammograms are no longer recommended. I cannot speak to the point about the mammographers being unskilled. Regarding the older studies, many experts have asserted that treatment for breast cancer has improved since these studies were completed and early diagnosis means less (in terms of mortality) given the better treatment. Dr. Begg's quote at the end encapsulates the screening conundrum very well. Unfortunately, there is far too much hysteria regarding breast cancer. Women should be able to make an informed decision before subjecting themselves to any cancer screening test. We are moving into an age of "personalized medicine." Yes, this term is indeed overused, but it is appropriate. Patients need to weigh the benefits, risks, and alternatives of screening with their individual provider before making any decision regarding their health. We are moving beyond this one-size fits all paradigm.

Regarding the concerns brought up by Dr. Tarone in his Cancer article -- his claims/suggestions were considered, but re-evaluation did not find them valid.

The Mammography Follies: Sense And Sensibility In The Current Controversy
The study exhibits significant methodological compromises and in addition stands against the weight of the systematically reviewed and critically appraised cumulative evidence to date, that in sum, it cannot support its own conclusions, and is decisively contradicted by the most robust of data. This is a brief review undertaken for breast cancer survivors (at the No Surrender Breast Cancer Survivor Forum, where I serve as Director of Medical Research) to show the current irrelevance of the CNBSS study to any practice-based change in the authoritative guidelines affecting mammographic screening.

Methodological Flaws And Incoherences
Consider two critical facts: (1) the detection rate differential, and (2) the mean detection size. The CNBSS's mammography breast cancer detection rate was approximately 3 per 1000 screened woman in the second and later screening rounds, but digital mammography currently delivers a detection rate of 6 per 1000 screened woman, and with tomosynthesis the results are even higher, at 8 per 1000 [Skaane, Radiology 2013). Furthermore, the technology used was substandard in quality, admitted by the CNBSS's own reference physicist who observed that the "quality [in the NBSS] was far below state of the art, even for that time (early 1980's) " [Kopans, AJR 1990; Yaffe, JNCI 1993]. In addition, widespread and image distorting scatter was not controlled and mitigated by use of gridding [Baines, AJR 1990], and more centrally to the issue, the mammographic technology of the 1980s when the CNBSS trial is simply not comparable to our current advanced technology and so conclusions from the earlier era are illicit projected the facts of today. And we have additional confirmation of the study's use of poor quality mammography, given that only 32% of the cancers were detected by mammography alone, an extremely low and unexpected number (where we would typically expect more than double that amount, at least ~66% [Mathis, J Am Coll Surg. 2010; Kopans, AJR 1993]). Indeed, the study authors themselves admit that "The lack of mortality benefit is also biologically plausible because the mean tumor size was 19 mm in the screening group and 21 mm in the control group....a 2 mm difference", and this further confirms that we cannot reasonably expect poor quality mammography to discover breast cancers at smaller sizes or in earlier stages when exhibiting a detection differential as miniscule as 2 mm, and consequently any associated expectation of a mortality reduction would be beyond the pale. In the real world, data support that among women age 40-49, mammography detected tumors exhibit a mean size of at least 9 times as large, at 18 mm (1.8 cm), with a patient (breast self-exam/BSE) or physician (clinical breast exam/CBE) detected (palpable) tumor mean size of 3 cm [Malmgren, Radiology 2012].

Here again we have some additional incoherence: the study shows that even those patients with the smallest non-palpable tumors detected in the screening arm had a 20% death rate from breast cancer, a number that appears irreconcilable with reality: the Swedish Medical Center (WA) RCT found that for cases from 1990 to 2008 for women in the 40 - 49 year age range, (1) the observed death rate from breast cancer among mammography detected cancers was 4% and among patient or physician detected breast cancers was 11, (2) while the breast cancer death rate ranged from 20 to 33%; (2) The shift to more mammography-detected breast cancer cases was accompanied by a shift to diagnosis at a lower disease stage, with diagnosis at stage 0 increasing 66% and diagnosis at stage III decreasing 66%), and concomitantly to breast cancer disease requiring less treatment. Thus, RCT data from this study data demonstrates significant and clinically relevant reduction in morbidity associated with mammographic-detected breast cancer in women in 40 - 49 year age group which includes (a) substantial reduction in number of mastectomies, (b) reduction in chemotherapeutic intervention, and (c) decreased recurrence, with higher rates of disease-free survival (DFS), making for an exceptionally strong case for the benefit of screening in this age group.

There are other compromising incoherences and methodological flaws. One of the most disturbing is the conclusion that mammography is not assistive of significant mortality reduction, and is directly contradicted by the studies own published results (see the Breast Cancer Survival subsection of the Results section), where it is explicitly stated that “The 25 year survival was 70.6% for women with breast cancer detected in the mammography arm and 62.8% for women with cancers diagnosed in the control arm”, which the authors themselves acknowledge to be significant, hence representing a real survival benefit from mammographic screening. In addition, the reported observation of more early deaths in the screened cohort than in the control is wholly to be expected, since the screening group was assigned more women with advanced breast cancers from which of course a higher mortality would be expected. And pre-invasive cancers were ignored in the study, although these are associated with higher survival rates and can generally be readily detected by mammography, so if we consider how the outcomes would change upon their inclusion, the expectation would be that we would see substantially higher survival rates than those reported, although we note in fairness that this is not assured and must be demonstrated by robust data, which we do not presently have.

Studies Show Benefit
The conclusions and recommendations of the CBBSS trial are against the weight of the aggregated data to date. The Mammography Screening of Young Women (SCRY) cohort study, the largest ever conducted (620,620 women) and the Swedish Two-County Trial which has the longest follow-up of any breast screening trial (it began in 1977), also confirmed a dramatic (approx one third (RR = 0.69)) reduction breast cancer deaths in all women ages 40 and over (including women ages 40-49) from regular mammography screening, and this trial showed that follow-up times of greater than 20 years are needed in such trials given that the observed number of breast cancer deaths prevented increases with increasing time of follow-up (most of the prevented breast cancer deaths were those that would have occurred more than 10 years after inception of screening). The ~30% reduction in breast cancer mortality has been consistent throughout the long follow-up period. And the recent NETBCS study demonstrated the benefit from early mammography screening to be about a 50% reduction (OR = 0.51) in the risk of dying from breast cancer. Moreover, a recent failure analysis [15] found that most deaths, more than 70% of the women who died of breast cancer, occur in unscreened women in their 40's, concluding that in order therefore to maximize mortality reduction and life-years gained, initiation of regular screening before age 50 years should be encouraged. Thus, the most rigorous mammography screening trials have shown that the most lives are saved by screening beginning at age 40. These collective robust results are further supported by recent case control studies that find that the death rate from breast cancer is lower among women screened compared to those not screened, with the lower death rate among screened women due to screening, and not attributable to treatment differences [Nickson, Cancer Epidemiol Biomarkers Prev 2012].

The Issue Of Overdiagnosis: Myths And Facts
A major contention of the CNBSS trial is that of the fact, and the adverse impact, of overdiagnosis. While it is undisputed that mammographic screening can detect cancers that might never evolve to become clinically evident or acquire have the potential for lethality, essentially the problem of overdiagnosis), the most robust data finds this to be modest at worst, and likely to be less than 10 percent [Puliti, J Med Screen 2012] and with the overdiagnosis primarily occurring among women over the age of 60 years, with the putative higher estimates of overdiagnosis found in the literature being largely due to the lack of adjustment for breast cancer risk and/or lead time. Furthermore, the Marshfield Clinic study [Onitilo, AJR Am J Roentgenol 2013] demonstrated decisively that missing a mammogram, even for a single year (even in the year before a breast cancer diagnosis), increases the chance of a cancer diagnosis at a later stage - suggesting that interventions to encourage use of mammographic screening may be of particular benefit to women most likely to miss mammograms, including those with no family history of breast cancer - and that in direct contradiction to the USPSTF screening guidelines, these findings were true whether women were older or younger than 50 years. Indeed, women without a mammogram in the 1 year before a breast cancer diagnosis had a 12% increase in the risk of diagnosis of late-stage breast cancer compared with those who underwent annual mammography, regardless of being older or younger than 50, while women with no mammogram in the 5 years before the breast cancer diagnosis had an 18% increase in late-stage breast cancer at diagnosis. Moreover, annual mammographic screening appears to be even more important to women younger than 50 years, as suggested by Marshfield Clinic study since breast cancer diagnosed in younger women is often more aggressive [Barnadas, Breast Cancer Res Treat 2010], so that as few as only 1–2 years of missed mammograms may result in rapid progression to late-stage cancer at presentation. And other studies confirm that mammographic screening for breast cancer be performed annually starting at age 40 years results in the greatest reduction in breast cancer morbidity and mortality [Cady, Ann Surg Oncol 2011; Cady, ASCO 2009; Spencer, Breast J 2004]. Collectively these results strongly motivate encourage women to follow the guidelines for mammographic screening set forth by ACS, NCCN, ACR, and ACOG (recommend against deferment of screening until after 50 years of age and advocate annual mammography beginning at age 40 years.

Furthermore, the Case Western Reserve retrospective review [Plecha, AJR Am J Roentgenol 2014], confirming an earlier study from the same group examined whether there were significant differences with respect to treatment recommendations, stage at diagnosis, and high risk lesion identification for women 40–49 years old undergoing screening mammography (screened) compared to women with a symptom needing a diagnostic evaluation (nonscreened), finding that in addition to the benefits of receiving a diagnosis at earlier stages of disease, with smaller tumors and node-negativity, patients with breast cancer undergoing screening mammography aged 40–49 years were less likely to require chemotherapy and its associated morbidities (the majority of high-risk lesions were diagnosed in the screened group).

And if we restrict out attention to only those studies that explicitly account and control for (1) cancer incidence during screening and (2) for lead time, given that length-time bias can engender overdiagnosis if there is unnecessary treatment of detected tumors that are indolent or slow-growing, these collectively show an overdiagnosis rate range at only 1% to 10%.

It should finally be pointed out that although abnormal screening mammographic results are often reported as significantly associated with anxiety, and hence constitute a “harm” associated with screening, both the reported degree of, and the significance, of this psychological distress show wide variation varies across studies [Brewer, Ann Int Med. 2007], while acceptance of false-positives among the public appears substantially high, with in one survey of American women 63 % judging that 500 false positives were reasonable to save one life, and 37 % deeming 10,000 or more to be tolerable [Schwartz, BMJ. 2000].

Thus, along with overwhelming number of studies that have convincingly argued against significant overdiagnosis, the argument of harm presented is substantially undermined. Widening the context then to include the strong evidence of benefit from mammographic screening in this age group, this therefore in the absence of sufficiently countervailing harm (overdiagnosis, anxiety), along with the several methodological compromises infecting the CBBSS trial, we must conclude that no compelling case has been made against the currently proactive mammographic screening guidelines inclusive of the 40 - 49 year age group.

I realize it is only a quote, but the statement from Dr. Begg that “[w]omen contemplating screening for breast cancer are overwhelmingly concerned about dying from breast cancer, not the various side effects of screening,” seems incorrect to me. I am equally concerned about the side effects of screening and the possible risks overdiagnosis and burdens of overtreatment to which it may lead, out of the very same concern for my overall health that motivates me to seek screening.

Thanks so much for sharing your experience, Anne.

While I am grateful to have access to this well presented article, I am concerned about what may be overuse of available screening. The data on 3-D is minimal and I just had a 3-D with additional standard screening and Ultasound and have been asked to go for MRI. I understand that MRI will give a better indication, so I will do so. However, I have always had dense breast tissue. Perhaps I should only have the MRI. Will I continue to need all these tests every year? Also, would it be better if the same radiologist would evaluate the tests each year? My mother did have breast cancer in her 80's and I am now 75.

Judith, unfortuntately we are unable to answer specific medical questions such as this on our blog. You should discuss these types of medical decisions with your doctor. If you would like to make an appointment with a Memorial Sloan Kettering physician, please call our Physician Referral Service at 800-525-2225 or go to Thanks for your comment.

my wife is 64 years old.
under OBACARE would it be best to just have theMRI for her yearly detection.
i would rather pay for one decisive exam than go for multiple pre-exams of lesser specificity

Add a Comment

We welcome your questions and comments. Because this is a public forum, please do not include contact information or other personal details. Also, keep in mind that while we can provide general information and resources, we cannot offer personal medical advice. To make an appointment with one of our experts, contact our Physician Referral Service at 800-525-2225 or online.
Your e-mail address is kept private and will not be shown publicly.

More information about formatting options