mammography, another hit

25-year results fro the Canadian National Breast Screening Study just published in BMJ open access (see mammog canadian study bmj 2014 in dropbox, or doi: 10.1136/bmj.g366), finding no benefit from mammog screening and that 22% of mammog-detected breast cancers were overdiagnosed!! lots of data from the study, will include since allows a more critical understanding.

 

study methodology:

        --15 screening sites in Canada assigned 90K women aged 40-59 randomly to mammog (5 annual mammos) or control, ending in 1988

        --all women had clinical breast exam (CBE) and were taught self-exam

        --women 40-49 received an initial mammo, then randomized (irrespective of clinical exam) to annual (ie, another 4) plus CBE vs no mammog/just usual care

        --women aged 50-59 randomized to 5 mammos and CBE or just CBE (in this group, felt unethical to have a “no screen” arm)

        -- high adherence to groups: 90% completed mammos in the mammo group

 

follow-up (up to 25 years after randomization):

        --1190 breast cancers diagnosed (666 in mammog arm, 524 in control)

        --5193 cancers found during followup period (2584 in mammog arm and 2609 in control)

        --of the 666 cancers dx’d in the mammog arm during the study, 484 were screen-detected, 176 were interval cancers

        --size of cancer in mammog arm was 1.9 cm vs 2.1 cm in control arm

        --in mammog arm: 30.6% of cancers detected were node-positive and 68.2% palpable

        --in control arm: 32.4% node-pos and all were palpable (similar proportions of palpable cancers in each group were node-positive)

        --1005 women died from breast cancer during the 25 years (1.1%), including 29.4% with diagnosis during the screening period

        --25-yr survival 77.1% for women with tumors <2cm and 54.7% if >2cm

        --25-yr survival 70.6% for women with mammog-detected cancer and 62.8% in control arm

        --no diff in 25-yr survival in those with palpable breast ca (66.3% vs 62.8%)

        --and, those nonpalpable mammog-identified tumors had 25-yr survival of 79.6%

       

        --for all-cause mortality, 9477 (10.6%) died in followup period, no diff between mammog and control arms

        --breast-cancer specific deaths (1005), no diff between mammog and control arms

        --361 breast cancer deaths during the screening period: 25-yr survival similar between groups if died from breast cancer during the screening period. and hazard ratio remained similar if screening period extended to 6 or 7 years

        --no diff in hazard ratio if look at women initially aged 40-49 (1.09, nonsignif) or 50-59 (1.02, nonsignif)

        --so, calculated over-diagnosis: excess of  142 cancers in mammog group (666vs 524), and this excess remained constant 15 yrs after enrollment and represents 22% of all screen-detected invasive cancer – ie, 22% of women who had mammo as screen were found to have an “invasive cancer” by pathologic exam, got treated aggressively, but would not have died from breast cancer. and they can say this because there was such a long followup period and the excess deaths remained constant after 15 ys. this overdiagnosis rate suggests that there is one over-diagnosed breast cancer for every 424 women who had mammog in this trial!!!

 

so..... this follows recent emails/blogs suggesting the poor performance of mammog screening. one issue is that some of the reason for lack of efficacy of screening is that there is better therapy (works better, less offensive). but one could argue (though i have not seen mentioned in this debate) that breast cancer therapy is still pretty miserable (and expensive) and that even though mortality is not affected, would be much better for the woman to have a little procedure and mild therapy than much more aggressive therapy later. then, of course, the counterarguments that we are medicalizing women more, treating many women who would never get cancer (though we are unable to identify which they are), creating lots of distress, exposing them to radiation. (by the way, as the editorialist notes, the data here on screening efficacy and over-diagnosis is akin to the data for prostate screening in men!!). also, this study was only a 5 year annual screen, not the much more aggressive protocols we use (USPSTF did downgrade to biennial screen age 50-74, with 40-49 offered in special cases)

 

my own practice has evolved over the past 1-2 years to passively discourage women age 40-49 to screen (ie, if a woman wants it, will screen, but i will add that there is potential harm of radiation exposure and there are no clear clinical data of benefit), and in the 50-70 year range, encourage women to do screening every other year (less radiation, fewer false positives, pretty similar pick-up rates).