A major medical journal, the Annals of Internal Medicine, has published a fresh cluster of 12 papers on breast cancer screening. Although I don’t agree with some of the editorials and conclusions of particular reports, I am encouraged by the big picture.
Overall, the series advances women’s health. It supports that breast cancer screening is beneficial, but that the method could be improved. It confirms that mammography lowers deaths from breast cancer in women over age 40, and lessens the occurrence of late-stage (advanced) breast cancer in those over 50 years. It identifies key evidence gaps, and raises questions for research.
Along with the controversial recommendations of a government panel, the journal provides the most thorough review of accuracy of digital mammography I’ve read. This new study finds that false positives–when women get called back for additional tests, but don't have cancer–turn out to be not so common: just 12% per screen. And most of those are for additional images; only 1.6% led to biopsy. What’s more, in this large analysis of modern screening, radiologists missed less than 0.15% of cancers.
What's at the heart of the issue is, or should be, is that each year 41,000 U.S. women die from breast cancer, and if many or most of those deaths could be averted by screening. Some prominent physician-writers have suggested that mammography causes net harm, or fails to catch the kinds of tumors that are likely to kill. Yet a recent large study reviewed here (among others) finds early detection positively influences survival in a major way.
There’s no doubt that the death rate from breast cancer has dropped by a third in recent decades in the United States, despite a rising number of invasive cases. Part of the debate concerns how much of improved survival, among people who have breast cancer, is due to increased screening or due to better treatment. I’d say the answer is both. (And if the proportion by which screening helps is just 30% or 40%, or 60%, is it worth so much argument?)
The second question, equally important, is whether the quality of life for over 230,000 U.S. people diagnosed each year is improved, over the long run, by early detection upon screening, as compared to waiting until an obvious mass or symptoms arise. As I've considered, early diagnosis affords some patients the opportunity of lesser treatment and a finite course of potentially curative therapy. Neither quality-of-life issues, nor costs of care, are addressed in any of the current papers.
At center of the new reports is the final U.S. Preventive Services Task Force (USPSTF) report which is, in itself, damaging. As I’ve written, the expert-less government panel draws conclusions from archaic studies of mammography. It scares women about the potential harms of screening by recounting exaggerated reports of false positives and by assigning too much weight to concern over women’s anxiety. It fails to provide clear advice for those at average risk for breast cancer between 40 and 50 years, who are the very women with the most years of life to gain by participating in high-quality breast cancer screening.
By placing the USPSTF recommendations in context of additional and new findings from a variety of sources, the Annals takes the issue from behind closed panels, away from gatherings of “less is more” enthusiasts who have little experience with breast cancer but write lots about it, and from meetings of radiologists, to an un-siloed discussion of how to make screening better, to reduce avoidable deaths.
The series invites women and doctors and journalists to step out of their “mammography works” or “it harms” trenches, take a look at the modern screening landscape, scrutinize everything and re-assess all options. I think no advocate for women’s health, or for screening, can take this non-dismissive concern for granted.
Rather, I welcome a no-holds-barred approach to examining current data. This is the best way forward. Admitting uncertainty, being open to examining new facts, again, and again, is necessary for progress.
What’s also crucial to inform research plans and interpretation of any findings, is willingness for all involved to hear from experts who are most familiar with current methods of diagnosis and with the disease. That would include radiologists, pathologists, surgical and medical oncologists, and patients. Yes, radiologists have conflicts of interests, but so do charity leaders and academics who have a lot–donations, grants, and reputation–vested in their opinions.
While I can’t review everything in the 12 articles, I’ll outline what I think most important.
The USPSTF Guidelines
The panel advises most women between the ages of 50 and 74 years to be screened. The recommendation is for mammography every other year. Given the legitimate goal of minimizing radiation exposure, I think this particular bit of advice, for biennial screening, makes sense.
The lack of a clear recommendation for women under age 50 years is problematic. Although age is a risk factor for breast cancer, half of diagnoses occur before the age of 61 years. Just last week, the American Cancer Society reported that breast cancer is the most common malignant cause of death for women between the ages of 20 and 60 years. A woman’s chances of developing invasive breast cancer before she completes her 49th year is 1 in 53; the condition is far from rare in middle age.
The USPSTF report essentially ignores modern radiology and oncology practice. While it may be true that prospective randomized studies are lacking to evaluate breast ultrasound, 3D mammography (tomosynthesis), MRI and other diagnostic tools, there’s no reason (or excuse) to draw guidelines from studies from the pre-digital era.
Now the USPSTF is on the defensive because the “C” grade it assigned to evidence for screening women in their forties has the potential to eliminate insurance payment for mammograms, which would render it harder for low-income women to get screened. The loss of coverage would affect many African American women at increased risk for the most lethal form (triple-negative disease) and whose death rates from breast cancer already outpace those of other U.S. demographic groups.
Unfortunately but not surprisingly, the USPSTF recommendations have been politicized. Language mandating a delay of possible implementation of the guidelines, until 2018, appears in the December 2015 U.S. federal budget deal.
I don’t think the USPSTF guidelines are the most significant component in the Annals series. Rather, they’d be harmful if considered alone. So I’ll move on–
Evaluation of False Positives and False Negatives in Digital Mammography
An excellent paper reviews false positive and false negative results after screening by digital mammography. This analysis rests on a huge amount of data from the U.S. Breast Cancer Surveillance Consortium (BCSC) involving 405,191 women ages 40 to 89 years screened between 2003 and 2011. The study was funded by the Agency for Healthcare Research and Quality (AHRQ) and the NCI. Sounds exciting? Maybe not. But everyone should read it.
This report indicates that false positive results, from digital mammograms, happen in just over 1 in 10 screens (121 per 1,000 women ages 40-49 years), and are even less frequent among older women. Almost all of the false alarms involved requests for additional images, such as sonograms. The rate of biopsy requests was extremely low, 15.6-17.5 per 1,000 screenings, or approximately 1.6%.
These are excellent results, and might lead future investigators to examine the consequences of screening volume and modern-ness of equipment in radiology practices, to see what can be done to ensure that such a degree of accuracy is the norm at all facilities. Of course there are different ways to spin numbers. A mention-worthy aspect of these calculations is that they’re not cumulative over a decade, but per screen; this is one reason why the results appear better than usual.
If your goal is to scare women about mammography, you might max out on the false positive calculations. You may have read papers reporting that most women who get screened suffer from false positives. Analyses like these stack the alarms first by imagining (or modeling) that all women get screened every year (and not biennially), and by considering false positives warranting images just like those warranting a biopsy. Although I don’t think a needle biopsy is that big a deal, it shouldn’t be equated with a sonogram.
For many women contemplating the procedure now, what they’d want to know is that for one digital mammogram of both breasts, the chances of a false positive requiring an additional picture is about 12%. The chances of getting a call for a biopsy is under 1.6%, as determined in the latest government-funded study. Great.
Even better, this study suggests that current technology yields a terrifically low number of false negatives. These false negatives include the “horror stories” you might hear, of women who have supposedly clear screening mammograms and then find out, months later, they’ve got metastatic breast cancer. In the comprehensive new report, the incidence of false negatives was ranged between 1.0 and 1.5 per 1,000 screens, i.e. 0.10-0.15% for digital mammograms with follow-up imaging, as described in the large modern study.
That is fantastically low percent of missed tumors. This finding suggests that people (including me) tend to disproportionately recall–and share stories of–the bad outcomes. The study suggests that radiologists are getting better at accurately detecting breast cancer. My concern about the low number of false negatives reported is that the interval was only one year. If you’re going to recommend screening every two years–as most of the papers in these series support–the false negatives should be checked over two-year intervals.
It’s an excellent result, one that should reassure women that if they get screened in a modern facility with digital imaging, the chances of finding a breast tumor, if present, is extremely high.
Despite these new findings, one study in the collection, based on older data, is titled: “Harms of Breast Cancer Screening: Systematic Review to Update the 2009 U.S. Preventive Services Task Force Recommendation.” And in that paper’s summary lies the statement: “False-positive results are common and are higher for annual screening, younger women, and women with dense breasts.” This point, that “false positives are common,” betrays the accuracy of a message that instead might be communicated to the public, based on the large and newer analysis of digital mammograms, above.
The fact is, in the government-funded analysis of digital mammography involving almost half a million women in the modern era, for women ages 40 to 49 years, the false positive rate was only 121 per 1,000 women (12.1%) for additional images and the rate of call-backs for biopsy was 1.6%.
Common? I’d say not.
Study and Editorial On Supplemental Imaging of Women With Dense Breasts
An essential paper, not to be overlooked, reviews supplemental screening for breast cancer in women with dense breasts. The investigators mention that about 27.6 million (43%) U.S. women between the ages of 40 and 74 years have dense breasts. As I have considered previously, having dense or fibroglandular breasts is a common and benign condition. Although many reports suggest it’s a risk factor for breast cancer, in itself, I am not convinced. The data are correlative and weak. What is clear is that high breast density obscures tumors, so they are missed by mammography.
Dense breasts are frequent in younger women, such as those who are premenopausal and, typically, under age 50 years. Last year I wrote that whole breast ultrasound can be very helpful–and may be excellent, with experienced eyes–in picking up early-stage, invasive breast cancer in women with dense breasts. I questioned whether the procedure should supplant mammography as the standard screening method for younger demographics.
The Annals review notes that hand-held ultrasound or automated whole breast ultrasound can detect extra breast cancer cases, after mammography, but that the data are insufficient to demonstrate a benefit at the population level, such as by a survival benefit. The investigators state that ultrasound tends to pick up early-stage invasive breast cancers. They review what evidence is available, also, for MRIs and 3-D mammography (tomosynthesis), and find none of it conclusive.
In an accompanying editorial, Dr. Wendie Berg of the University of Pittsburgh Medical Center writes: “Although mammographic screening reduces breast cancer mortality, all women do not share this benefit equally.” She reviews the evidence, including the recently-published J-START (Japan Strategic Anti-cancer Randomized Trial) evaluating the addition of ultrasound to mammography in screening, with favorable results in terms of ultrasound’s detection capability. Berg supports offering supplemental imaging to women with dense breasts, preferably with collection of information on molecular subtypes of the disease, risk factors, staging and outcomes.
I couldn’t agree more, that research is needed in this area. I would add that the studies should be performed with input by radiologists, but not only by radiologists as they, like other doctors, may have conflicts of interests. Breast ultrasound doesn’t pay much, as compared to MRIs, 3D exams and new molecular breast imaging methods. Few radiologists will question, out loud, if mammography is the best way to find breast cancer, even in young women, below the age of 50 years, with dense breasts.
Thanks to these new articles, that conversation–about how best to check women for breast cancer, with no assumptions regarding mammography as the gold standard–is more likely to happen.
Meeting room with blank board (via Wikimedia Commons)
