Re: Effectiveness of and overdiagnosis from mammography screening in the Netherlands: population based study
Autier et al (1) analysed outcome of the Dutch mammography screening programme and conclude that screening is ineffective. This is an ecological study, focussing on rates over time of breast cancer incidence, incidence by stage, and mortality. There is a clear and substantial reduction in mortality from the disease in the screening period. Thus one might conclude that the screening is reducing mortality as expected from the randomised trial results. The authors, however, attribute most of this reduction to changes in therapy, on the basis of a small trend in reducing incidence of cancers of stage II or worse within the screening period. However, there are no data on incidence trends prior to the screening period (although it has been observed that breast cancer incidence was rising dramatically throughout the western world in the years immediately prior to the setting up of screening programmes), so we do not know what the incidence would have been, overall and by stage, in the screening period if the screening had not happened. Similarly there are no data on which tumours were exposed to screening and which tumours received which treatment. Thus, the authors' conclusions cannot be verified.
The authors assume that all of the increase in breast cancer incidence from 1989 to 2012 is due to overdiagnosis, and that despite increases in breast cancer incidence in all ages that the expected numbers of advanced stage cancers aged 50-69 years would remain the same. The authors present no actual data on the overall increase in breast cancer incidence in this time period in the unscreened age groups (although data on stage 1 and DCIS are presented). In the UK, breast cancer incidence in women aged <50 years of age increased by 53% between 1989 and 2012 and by 41.5% in women aged 75-85 years populations that still did not undergo regular breast screening in 2012 [2]. In the same time period breast cancer incidence in the 50-69 year age group rose by 67%. If one takes a mid-point estimate of the increase in the unscreened population of 47% then the percentage increase due to overdiagnosis is more likely just 20%.
The assumption that as there was no increase in stage 2-4 breast cancers in the <50-year age group means that none would have been expected in the 50-69 year group is also flawed. There has been much greater awareness of breast cancer at younger ages and women are far less likely to be ignored with a small breast lump in 2012 than in 1989. Also breast screening in women aged 40-49 years with a family history of breast cancer is likely to have downstaged at least some cancers. The inclusion of all stage 1 breast cancers in an assessment of potential overdiagnosis is also clearly incorrect. No data is supplied on the grade of the stage 1 cancers. Grade 3 breast cancers are almost never overdiagnosis, and corroboration of the Dutch results would need to have shown an over-representation of grade 1 invasive cancers.
References
1. Autier P, Boniol M, Koechlin A, Pizot C, Boniol M. Effectiveness of and overdiagnosis from mammography screening in the Netherlands: population based study. BMJ. 2017 Dec 5;359:j5224.
2. Office for National statistics https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/...
D Gareth Evans, Professor of Medical Genetics and Cancer Epidemiology, Anthony Howell, Professor of Medical Oncology; Dr Sacha Howell, Medical Oncologist, Dr Susan Astley Imaging scientist, Professor Anthony Maxwell, Radiologist: All University of Manchester, Oxford Rd Manchester, UK
Stephen W. Duffy; Professor of Cancer Screening, Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, Charterhouse Square, London UK
Competing interests:
No competing interests
19 December 2017
D Gareth Evans
Professor of Medical Genetics and cancer epidemiology
Anthony Howell, Sacha Howell, Susan Astley, Anthony Maxwell, Stephen Duffy
University of Manchester
Genomic Medicine, St Mary's Hospital, University of Manchester, Oxford Rd Manchester, M13 9WL UK
Rapid Response:
Re: Effectiveness of and overdiagnosis from mammography screening in the Netherlands: population based study
Autier et al (1) analysed outcome of the Dutch mammography screening programme and conclude that screening is ineffective. This is an ecological study, focussing on rates over time of breast cancer incidence, incidence by stage, and mortality. There is a clear and substantial reduction in mortality from the disease in the screening period. Thus one might conclude that the screening is reducing mortality as expected from the randomised trial results. The authors, however, attribute most of this reduction to changes in therapy, on the basis of a small trend in reducing incidence of cancers of stage II or worse within the screening period. However, there are no data on incidence trends prior to the screening period (although it has been observed that breast cancer incidence was rising dramatically throughout the western world in the years immediately prior to the setting up of screening programmes), so we do not know what the incidence would have been, overall and by stage, in the screening period if the screening had not happened. Similarly there are no data on which tumours were exposed to screening and which tumours received which treatment. Thus, the authors' conclusions cannot be verified.
The authors assume that all of the increase in breast cancer incidence from 1989 to 2012 is due to overdiagnosis, and that despite increases in breast cancer incidence in all ages that the expected numbers of advanced stage cancers aged 50-69 years would remain the same. The authors present no actual data on the overall increase in breast cancer incidence in this time period in the unscreened age groups (although data on stage 1 and DCIS are presented). In the UK, breast cancer incidence in women aged <50 years of age increased by 53% between 1989 and 2012 and by 41.5% in women aged 75-85 years populations that still did not undergo regular breast screening in 2012 [2]. In the same time period breast cancer incidence in the 50-69 year age group rose by 67%. If one takes a mid-point estimate of the increase in the unscreened population of 47% then the percentage increase due to overdiagnosis is more likely just 20%.
The assumption that as there was no increase in stage 2-4 breast cancers in the <50-year age group means that none would have been expected in the 50-69 year group is also flawed. There has been much greater awareness of breast cancer at younger ages and women are far less likely to be ignored with a small breast lump in 2012 than in 1989. Also breast screening in women aged 40-49 years with a family history of breast cancer is likely to have downstaged at least some cancers. The inclusion of all stage 1 breast cancers in an assessment of potential overdiagnosis is also clearly incorrect. No data is supplied on the grade of the stage 1 cancers. Grade 3 breast cancers are almost never overdiagnosis, and corroboration of the Dutch results would need to have shown an over-representation of grade 1 invasive cancers.
References
1. Autier P, Boniol M, Koechlin A, Pizot C, Boniol M. Effectiveness of and overdiagnosis from mammography screening in the Netherlands: population based study. BMJ. 2017 Dec 5;359:j5224.
2. Office for National statistics https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/...
D Gareth Evans, Professor of Medical Genetics and Cancer Epidemiology, Anthony Howell, Professor of Medical Oncology; Dr Sacha Howell, Medical Oncologist, Dr Susan Astley Imaging scientist, Professor Anthony Maxwell, Radiologist: All University of Manchester, Oxford Rd Manchester, UK
Stephen W. Duffy; Professor of Cancer Screening, Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, Charterhouse Square, London UK
Competing interests: No competing interests