mammograms: 50% false positive rate in 10 yrs

 A recent analysis found that the cumulative probability of a false positive result after 10 years of annual mammography screening was in the 50% range (see mammogram false pos rate 50% in 10yrs JAMA2022 in dropbox, or doi:10.1001/jamanetworkopen.2022.2440) 

 

Details: 

-- observational comparative effectiveness study from screening exams between 2005-2019 in the Breast Cancer Surveillance Consortium 

-- 903,495 individuals aged 40-79, with 2,969,055 non-baseline mammography screening exams, interpreted by 699 radiologists

    -- baseline mammograms were not included because of higher false positive rates, perhaps related to younger women with denser breast tissue 

-- mean age of women at the time of the screening exams 58 years, with 58% in those <60yo and 46% performed in women with dense breasts 

-- 15% of exams used digital tomosynthesis 

-- false positive categories assessed: 

    -- false positive recall: initial mammogram having the radiologists’ BI-RADS attribution of cancer risk (1 and 2 being basically normal), including 0 (needing more imaging), 3 (probably benign), 4 (suspicious abnormality), or 5 (highly suggestive of cancer) 

    -- short-interval follow-up recommendation: typically after BI-RADS 3 after diagnostic imaging workup 

    -- biopsy recommendation: BI-RADS 4 or 5 after diagnostic imaging workup

 

Results: (in attempt to declutter these results, many of the confidence intervals were eliminated)

 

-- 10-year cumulative probability for risk of at least one false positive result: 

    -- annual screening with tomosynthesis versus digital mammography: 

        -- false-positive recall: 49.6% versus 56.3%, difference -6.7 percentage points (-7.4 to -6.1) 

        -- short interval follow-up recommendation: 16.6% versus 17.8%, difference -1.1 percentage points (-1.7 to -0.6) 

        -- biopsy recommendation: 11.2% versus 11.7%, difference -0.5 percentage points (-1.0 to -0.1) 

 

-- 10-year cumulative probability for risk of at least one false positive result: 

    -- biennial screening with tomosynthesis versus digital mammography: 

        -- false-positive recall: 35.7% versus 38.1%, difference -2.4 percentage (-3.4 to -1.5)  

        -- short interval follow-up recommendation: 10.3% versus 10.5% difference -0.1 percentage points, nonsignificant 

        -- biopsy recommendation: 6.6% versus 6.7% difference -0.1 percentage point, nonsignificant

 

-- cumulative probabilities of false positive results were substantially lower for biennial versus annual screening (DM=digital mammography, T=tomosynthesis): 

    -- false-positive recall: T 35.7%/DM 38.1% vs T 49.6%/DM 56.3% 

    -- short interval follow-up: T 10.3%/DM 10.5% vs T16.6%/DM17.8% 

    -- biopsy recommendation T 6.6%/DM 6.7% vs T 11.2%/DM11.7%

-- cumulative probabilities of false positive results were substantially lower in older vs younger women 

        -- false-positive recall: annual screening in women 70-79yo versus 40-49yo: T39.8%/DM 47.0% versus T60.8%/DM 68.0%  

        -- short follow-up: T13.3%/DM14.2% versus T20.7%/DM 20.9% 

        -- biopsy recommendation: T9.1%/DM9.3% versus T13.2%/DM13.4% 

-- see their tables 2, 3 and 4 for detailed cumulative 10-year probability for age groups, breast density, and annual versus biennial screening, and comparing digital breast tomosynthesis versus digital mammography 

    -- as one example for women age 50 to 59: 

        -- annual screening in those with almost entirely fatty breasts 29.1% for tomosynthesis versus 36.3% for digital mammography; for women with extremely dense breasts 60.4% for tomosynthesis versus 58.8% for digital mammography 

        -- biennial screening in women with almost entirely fatty breast tissue: 18.3% with tomosynthesis and 42.6% from mammography; and in women with extremely dense breast tissue 42.2% for tomosynthesis and 36.9% for mammography 

            -- notably a bit worse in this case with tomosynthesis

 

Commentary: 

-- prior studies have found that false-positive results are common, lead to unnecessary additional imaging and biopsy procedures, financial and opportunity cost, and patient anxiety.

-- tomosynthesis is a spreading modality for breast imaging, using multiple 2-dimensional images to construct a 3-dimensional image by computer, with some studies finding increased cancer detection rates, especially for DCIS (ductal carcinoma in-situ) and small invasive cancers (but not for those >2cm), though one study did find an increase in false-positive screenings. 

    -- it should be noted that DCIS is not really a cancer, but a pre-cancer (stage 0); and has a small but significant chance of becoming an invasive cancer if untreated, with 3.3% mortality at 20 years (see https://pubmed.ncbi.nlm.nih.gov/26291673/ ). 

 

-- this study was useful in that it assessed the cumulative false-positive rates of mammographic screening over 10 years and added to the prior data suggesting that biennial exams were just about as sensitive for malignancy, had significant reductions in false-positive recall (2-7% versus annual screening), and had ½ the radiation dose; this study also did fuller assessment by age as well as breast density, as well as extending the analysis to tomosynthesis. 

-- tomosynthesis overall had significantly lower false-positive recall rate than digital mammography, a slightly lower false-positive short-term follow-up recommendation; but not much difference with biopsy recommendation between the two modalities 

-- one major concern is that there is a somewhat higher radiation dose with tomosynthesis, since “more penetrating x-ray photons are used in order to secure sufficient signal-to-noise ratio to the image detector to produce high quality reconstructed slices”, both overall and for individual views (see https://pubmed.ncbi.nlm.nih.gov/28819862/#:~:text=There%20is%20some%20concern%20about%20the%20dose%20increase%20with%20tomosynthesis.&text=Clinical%20data%20show%20a%20small%20increase%20in%20radiation%20dose%20with%20tomosynthesis.&text=Synthetic%202D%20images%20from%20tomosynthesis%20at%20zero%20dose%20reduce%20potential%20harm.&text=The%20small%20dose%20increase%20should,barrier%20to%20use%20of%20tomosynthesis, a study of almost 5000 images from 1,208 women finding a 38% increased radiation exposure with tomosynthesis) 

-- and, radiation may well induce cancers (see http://gmodestmedblogs.blogspot.com/2018/06/low-dose-radiation-and-subsequent.html , which reviews several of the studies. Of note, much of the information on radiation and cancer comes from old nuclear bomb data (eg Hiroshima and Nagasaki, see radiation exposure hiroshima radiatres2007 in dropbox, or doi.org/10.1667/RR0763.1), which likely reflects radiation exposure in the average Japanese female population. One might posit the multiple hit concept of cancer: those with underlying breast abnormalities (eg DCIS, history of atypia) might well have an increased risk of cancer by radiation. Same for those with COPD who have underlying lung pathology getting multiple low-dose chest CTs (eg see http://gmodestmedblogs.blogspot.com/2014/10/uspstf-lung-cancer-screening-revisited.html ). also, there was a specific assessment of radiation and breast cancer, finding a linear dose-response curve but excess rates in those with a history of mastitis and benign breast disease, reinforcing a differential effect by radiation exposure: see https://pubmed.ncbi.nlm.nih.gov/12105993/  ). ie: this all suggests that radiation is potentially harmful and may be more so in those with abnormal breasts (who, unfortunately, tend to get more xray evaluations....)

-- all of this reinforces the utility of decreasing radiation exposure. And the option of every 2 year screening is pretty enticing 

-- the frequent recommendation for short-interval follow-up, of course affects the amount of radiation exposure, though it might decrease unnecessary biopsies.

-- Some prior studies suggest women with denser breasts may need annual mammograms to have effective screening, given the lower sensitivity to detecting early cancers (again, though with more radiation exposure)

    -- one concerning issue is radiologist interpretation of breast density, which varies dramatically: one study found a median of 38.7% mammograms were interpreted as "dense breasts", but the range was the rather large 6.3% to 84.5% (see http://gmodestmedblogs.blogspot.com/2016/10/radiologist-variability-in-mammography.html ); another study found that there was pretty wide variability in the accuracy of radiologist mammogram interpretation for detecting breast cancer, with range between them exceeding 40% (see https://pubmed.ncbi.nlm.nih.gov/8546556/#:~:text=As%20indicated%20by%20receiver%20operating,the%20population%20of%20US%20radiologists. )

-- a major concern regarding the 50% false-positive rate in annual mammograms (which is still pretty high even with biennial ones) is that these false-positives have real, human effects: breast cancer is really scary and hearing that one needs further studies because their mammogram was not entirely normal does raise huge anxiety and fears, anxiety that may or may not be fully relieved by further diagnostic mammographic exams in the near future. And, of course, some abnormalities lead to biopsies, which are interventions that have potential adverse effects: infections, etc as short-term ones, and potentially long-term changes in women’s heightened sense of their mortality, which conceivably is reinforced by the life-long remaining scars. Other collateral issues include costs of getting mammograms (not always covered by insurance) and loss of work (and potentially income) to get the exams

 

Limitations: 

-- we have pretty limited data on the true effect on radiation in promoting cancer, lacking well-designed large human studies. and most of the studies we have do not take into account underlying organ pathology, or the many other potential promoters of cancer (environmental/occupational exposures, chemicals in food/water, levels of stress/lifestyle/BMI, smoking/alcohol, etc)

-- there is considerable variability in radiation exposure depending on the machines used: older ones have significantly more radiation. and, this study did span 14 years with (presumably) improved xray machines

-- there is likely considerable variability in radiologist interpretation of the exams, as noted above, especially in light of the 699 of them involved in the above study 

-- there was no clinical outcome assessment in defining the “false-positives” in this study. there was 10 year follow-up, but no assessment of individual data on which women did develop invasive cancer stratified by annual vs biennial exams, or by digital mammography vs tomosynthesis

-- they did not assess results of mammography by breast size, important since those with larger breasts usually require higher radiation dose, and more likelihood of creating a cancer 

 

So, given the overall appropriateness of mammograms, one approach to the quite common BIRADS 0 finding (which leads to more diagnostic views) is to reinforce that half the women do have a mammogram at some time that does not allow for a complete view of their breasts, and the likelihood of a breast cancer is still very, very low. It is also important to be able to do the diagnostic mammograms as quickly as possible after the inadequate screening one in order to help relieve some of the anxiety for the majority of women with normal subsequent exams. One other discussion point with women is around age of screening: those at average risk who are 40-49yo do have a much higher rate of false positive mammograms, leading to more studies/more radiation in people who have a longer life expectancy (and more time to develop radiation-induced cancer)...

geoff

 

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@uphams.org

 

to get access to all of the blogs (2 options):

1. go to http://gmodestmedblogs.blogspot.com/ to see them in reverse chronological order

2. click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

3. or you can just click on the magnifying glass on top right, then  type in a name in the search box and get all the blogs with that name in them

 

or: go to https://www.bucommunitymedicine.org/ , a website from the Community Medicine section at Boston Medical Center.  This site does have a very searchable and accessible list of my blogs (though there have been a few that did not upload over the last year or two). but overall it is much easier to view blogs and displays more at a time.

 

 

please feel free to circulate this to others. also, if you send me their emails, i can add them to the list