Research on breast cancer screening varies by design
There has not been a randomized clinical trial conducted on screening mammography since the days of the analog mammogram. The research that has been conducted is difficult to compare due to variations in screening ages and intervals, technology sensitivity, and patient adherence with recommended screening. Treatments for breast cancer also have changed dramatically over time, so the findings of older studies may no longer be relevant to current breast cancer screening. The kind of analysis that needs to be done is an interrupted time series, where you can look at the trajectory of breast cancer survival and whether screening mammography shifts that survival in any way.
One specific study from Australia measured the impact of newer available breast cancer treatments, including tamoxifen for women with receptor-positive tumors and newer chemotherapy strategies.5 The authors analyzed screening mammography trends in one large province where women aged 50 to 69 were offered biennial screening. Trends from the 1990s showed that more women were being screened over time. Simultaneously, however, advances in therapy were entering clinical practice. The researchers pointed to a substantial decline in mortality from breast cancer from the early 1980s until 2013. But their conclusion was that none of the decline in mortality for breast cancer could be attributed to screening mammography when they looked at time trends; from their perspective all of the important decline in breast cancer mortality resulted from better treatment. They concluded that government programs should not support screening mammography.5
That is a recommendation that we do not support. However, we do recognize the conundrum that mammography is less sensitive among those who have dense breasts. In order to have congruent professional guidelines, we support research funding to determine which types, starting ages, and intervals of screening would be best in various patient populations. The USPSTF cites data from studies performed in the 1980s based on outdated technology; more recent (and relevant) randomized clinical trials have not been performed, and yet this information is critical to provide sufficient evidence to develop appropriate guidelines.
Our recommendations for gathering new data
The kind of data we would find most valuable would assess how different screening strategies impact overall mortality and breast cancer-specific mortality. It would require decades of follow-up—which of course means that screening technology will change over that time. A surrogate for evaluating overall survival is to look at interval cancers, which are all breast cancers diagnosed following negative mammograms and prior to the next screening. These cancers may or may not be biologically active, again focusing us on the need to look at overall survival of the patient. In addition, reducing breast cancer mortality may not reduce overall mortality, because the treatment for breast cancer may cause heart disease, or osteoporosis, or something else that impacts overall survival. These are important considerations for women and physicians who are making choices on treatment. What matters to a patient are 2 overlapping questions:
- Do I have a life-threatening condition or do I not?
- Has screening identified a condition that might lead to treatment that’s unnecessary?
The problem is that with breast cancer we can’t tell the difference. We do not understand the biological potential of a lesion when we evaluate an image on MRI, or computed tomography (CT), or mammography.
A re-look at presented data
A trial conducted by Bakker and colleagues6 was discussed by the authors of the DenseBreast-info.org quiz in which they recommended breast MRI for all women with extremely dense breasts (but no other risk factors for breast cancer) detected on screening mammograms.7 The Bakker study was large and conducted in the Netherlands. The primary outcome of the trial was to compare the incidence of interval breast cancers of women aged 50 to 75 randomly assigned to MRI versus those assigned to continued screening mammography every 2 years. Importantly, among the more than 8,000 women who were assigned to MRI, 59%, or fewer than two-thirds, chose to actually undergo MRI.
Among women randomized to MRI, 20 interval cancers were found—4 were diagnosed in those who actually had MRIs, and 16 were diagnosed among women who were randomized to MRI but didn’t undergo the study. Among women assigned to screening mammography only, 161 interval cancers were diagnosed among more than 32,000 women screened. The primary outcome findings were 2.5 interval cancers per 1,000 screenings among women randomly assigned to MRI, and 5 interval cancers per 1,000 screenings among those randomly assigned to mammography only.6
Because the trial included women aged 50 and older, we can’t apply these results to younger women, who often undergo screening mammography in the United States. In addition, the majority of the population in the Netherlands are of Western European ethnicity, a less-diverse population of women than in the United States. Furthermore, among the tumors that were detected in the MRI group, a larger proportion were DCIS, early-stage tumors, well differentiated, and hormone receptor-positive. This observation supports that many of the MRI-detected tumors were cases of overdiagnosis, or the detection of tumors destined not to cause clinical problems for the patient during her lifetime, or for which earlier diagnosis would impact survival.
We also know that treatment of these small ER-positive tumors carries risks for patients, as we may treat them by depriving a patient of estrogen for the rest of her life, with potential consequences of sexual dysfunction, osteoporosis, and perhaps cardiovascular disease depending on her age at the time of that diagnosis. Weighing the risks and benefits of not only treatment but also use of more sensitive screening techniques such as MRI is extremely important. Although Bakker and colleagues’ study results are interesting, we do not feel they support routinely recommending MRI for women found to have extremely dense breasts with mammography.
Overdiagnosis: A difficult concept
One reason overdiagnosis is so challenging to understand is that it can’t be directly measured, which makes comprehending it that much more problematic for clinicians and our patients.
One way to help grasp the overall issue is to compare screening mammography with cervical and colon cancer screening.
We are well aware that cervical cancer screening has reduced the incidence of mortality from invasive cervical cancer.8 We can argue very validly that the biggest success in any cancer screening program in history and globally has been cervical cancer screening. Our specialty, in particular, should feel proud about this. Screening colonoscopy also has repeatedly been found to reduce colon cancer mortality.9 For breast cancer, decades of media messaging have emphasized the benefits of screening mammograms; however, in contrast with cervical cancer screening and colonoscopy, screening mammography has not reduced the incidence of breast cancer presenting with metastatic or advanced disease. Danish authors pointed out in 2017 that screening mammography has not achieved the hoped for or the promised reduction in breast cancer mortality.10
A report published in the March 2022, issue of Annals of Internal Medicine used modeling techniques to estimate the incidence of overdiagnosis and concluded that, among women aged 50-74 years receiving biennial screening mammograms (consistent with USPSTF recommendations), more than 15% of screen-detected breast cancers would represent cases of overdiagnosis. Of note, the study authors found that, among screen-detected cancers, the proportion representing overdiagnosis among women in their 60s (16.7%) and early 70s (23.6%) was higher than among women in their 50s-60s (11.5%-11.6%).11
The former Chief Medical and Scientific Officer for the American Cancer Society Otis Brawley, MD, has stated that, at the same time that breast cancer screening should not be abandoned, “We must acknowledge that overdiagnosis is common. The benefits of screening have been overstated, and some patients considered as ‘cured’ from breast cancer have, in fact, been harmed by unneeded treatment.”12
“Everybody loves early detection,” said Donald Berry, PhD, from MD Anderson Cancer Center, “but it comes with harms.” He points out that mortality rates have improved for breast cancer, but he attributes it to improved treatment. “The harms [of screening] we know, but the benefits of screening are very uncertain.”13
Inequities in breast cancer screening and outcomes
The importance of health equity is receiving more attention. When examining equity according to breast cancer mortality, ethnic minority populations have worse cancer survival outcomes than White women; the mortality rate is 40% higher among Black women than among White women.1 Lower survival rates are also noted among lower socioeconomic groups and among women who live in rural areas. Lower survival rates among ethnic minority women are also noted for cervical and colorectal cancers.2
In the past, these disparities in mortality were attributed to the historically lower breast cancer screening rates among Black women compared with White women. However, decades of efforts to increase mammography rates have effectively addressed much of the racial/ethnic gap in screening rates.1 In fact, a 2021 study showed Black and Hispanic women to have 6% to 10% higher rates of breast, cervical, and colorectal cancer screening than White women according to US Preventive Services Task Force guidelines.2 The study authors point out that other national data have demonstrated similar results and conclude that “higher cancer mortality among racial/ethnic minority groups will not be reduced solely by increasing rates of cancer screening. Although preventive screenings and timely diagnosis are important elements of prognosis, they are just 2 elements of many along the cancer care continuum that need to be addressed to eliminate disparities in cancer mortality.”
Unfortunately, the randomized trials that have been conducted on mammography have been conducted overwhelmingly in White populations. National registry studies from the Netherlands and Sweden are not representative patient populations for the United States. Recently, the US government proposed an ambitious plan to cut cancer mortality rates and has promised vast amounts of research funding to achieve that goal.3 Hopefully, this funding will support studies which enroll diverse patient populations. We hope to gain knowledge on what elements along the cancer care continuum can be addressed to better reduce or eliminate cancer mortality inequities.
References
1. National Cancer Institute. SEER Explorer. https://seer.cancer.gov/explorer/. Accessed February 9, 2022.
2. Benavidez GA, Zgodic A, Zahnd WE, Eberth JM. Disparities in Meeting USPSTF Breast, Cervical, and Colorectal Cancer Screening Guidelines Among Women in the United States. Prev Chronic Dis. 2021;18:200315. doi: http://dx.doi.org/10.5888/pcd18.200315.
3. Stohlberg SG, Kolata G. Biden presents ambitious plan to cut cancer death rate in half. The New York Times. February 2, 2022.
Continue to: Limitations of breast MRI...