January 9, 2017

expert reaction to a study investigating breast cancer screening, tumour size, and over-diagnosis

The findings of a Danish 30-year cohort study was published in the Annals of Internal Medicine, with the authors reporting that breast cancer screening was not associated with a reduction in the incidence of advanced cancers but was associated with over-diagnosis of small tumours.

Prof. Paul Pharoah, Professor of Cancer Epidemiology, University of Cambridge, said:

“This manuscript is an addition to a large body of scientific literature reporting observational science investigating the effects of breast screening mammography.  It does not tell us anything that we did not already know about breast screening and does not add much to the debate about the relative benefits and harms of breast cancer screening by mammography.

“The study showed variation in the incidence of small/early tumours or large/late tumours between regions where there either had or had not previously been population based breast screening mammography.  The authors interpreted these data to say that there is substantial over diagnosis of small tumours (i.e. tumours that not have been diagnosed during a woman’s lifetime in the absence of screening) but no evidence of a reduction in advanced tumours as would be expected. These authors have been very outspoken against screening in the past.

“The main problem with this study is that it is hard to untangle effects of screening from underlying time trends and comparing time trends between regions where there has and has not been screening does not necessarily help.  Additionally, how these changes (whether or not they are due to screening) impact on outcomes/death rates is conjecture.  Other things apart from tumour size affect outcomes – notably lymph node metastases and oestrogen receptor (ER) status and there are no data on this in the paper.  And all of these factors interact with the effectiveness of treatment – e.g. a small reduction in how advanced a tumour is at diagnosis may make no difference if treatment is not very effective, but may be very important with effective treatment (or vice versa we do not know).

“Consequently this study should not influence women who are considering having a screening mammogram on whether or not screening is the right thing for them.   Randomised trials have shown that mammography reduces breast cancer mortality by a small amount, but this comes at the cost of increasing the number of women diagnosed and treated for breast cancer (so called over diagnosis and over treatment) so that some women will have unnecessary surgery and/or radiotherapy.   Whether or not the benefit is worth the harm is an individual decision and different women will make different choices.”

Further information from Prof. Paul Pharoah:

“Observational studies are more prone to bias than randomised controlled trials.  There have been multiple RCTs investigating the benefits of breast screening by mammography – reviewed by the Independent UK Panel on Breast Cancer Screening in 2012.

“If breast cancer screening were effective it would be expected to reduce the incidence of late stage disease and increase the incidence of early stage disease by the same amount, as the primary aim of screening is to reduce mortality by detecting disease earlier

“This study has compared early and late breast cancer incidence trends over time in Denmark in regions where screening was introduced in the 1990’s and regions where it was not introduced until 2007.  They have used tumour size as a proxy for early (< 20 mm tumours) and late (>=20 mm tumours) disease.

“In women in the age group offered screening (50-69 years) there was a substantial increase in the incidence of small/early tumours in regions where there HAD been population based breast screening mammography but a much smaller increase in the incidence of small/early tumours in regions where there had NOT been population based breast screening mammography.  In contrast, the incidence of large/late tumours in regions where there HAD been population based breast screening mammography was stable, whereas the incidence of large/late tumours in regions where there had NOT been population based breast screening mammography increased slightly.

“The findings are similar to those reported for similar studies from Norway and the USA and so they are not really new and do not provide any major new insights into the effectiveness of screening. We know that screening results in over diagnosis.  The magnitude of the over diagnosis can be debated but the results from this study are not that different from other studies.

“In Fig 1 in the paper, the first panel is women aged <50 (non-screening age) but there are big differences in the screened and non-screened regions in the time trends.  The second panel is the screening age group and advanced tumours increased in the non-screened areas, but not in the screened areas.  Thus screening may have prevented an increase (but not resulted in a decline in advanced tumours).  Alternatively something else may have been different between screened and non-screened regions.  The third panel is older women (not screened) and here the pattern is similar in screened and non-screened regions.

“Some of the differences in the age specific graphs in Fig 1 may simply be due to differences in patterns of incidence of ER negative (younger patients) and ER positive (older patients) disease.

“Those who are believers will say that this paper tells us nothing and the randomised trials are key.  The non-believers will say that this confirms that screening is a waste of money.  And people like me who say it is almost impossible to say what the harms and benefits are.”

* ‘Breast Cancer Screening in Denmark: A Cohort Study of Tumor Size and Overdiagnosis’ by Jørgensen et al. will be published in Annals of Internal Medicine at 22:00 UK time on Monday 9th January, which is also when the embargo will lift. 

Declared interests

Prof Paul Pharoah: “I have no conflicts of interest to declare.”