Last month, I wrote a blog post that discussed the value of asymptomatic breast cancer screening to public health. Essentially, several studies looking at all cause and breast cancer specific mortality outcomes from breast cancer screening programs – including meta-analysis from the Cochrane Group and the US Preventative Services Task Force – have identified generalized mammography screening programs as ineffective at reducing breast cancer mortality.
On Monday, the BMJ released a follow-up on the Canadian National Breast Screening Study. Data from this study has previously been published at earlier cut-off points and has been the subject of much debate already. The Cochrane Review identifies much of this research in the references to that report and should be read by anyone interested in the details of the debate. The new publication now includes follow-up to 25 years with the cohort population and once again has essentially come to the same conclusion:
Conclusion: Annual mammography in women aged 40-59 does not reduce mortality from breast cancer beyond that of physical examination or usual care when adjuvant therapy for breast cancer is freely available. Overall, 22% (106/484) of screen detected invasive breast cancers were over-diagnosed, representing one over-diagnosed breast cancer for every 424 women who received mammography screening in the trial.
A few details on the study are relevant to further discussion. The trial was a randomized control trial and took place early in the modern era of mammography screening. Study participants were randomized into four groups based on age (40 – 49 & 50 – 59) and intervention (annual mammography for five years vs standard of care/physical exam). 89,835 women participated in the trial and the primary period of the study was between 1980 and 1988 when the study-specific screening centers closed. Since 1988, the cohort of women have been followed to report on their health outcomes in relation to breast cancer. Randomization remained intact in terms of women either regularly attending or not attending mammography screening in the community and through their family physicians.
For the 25 year follow-up, the study combines the age groups and reports only on the mammography vs control group results. During the 5 years of the active screening portion of the study, 666 breast cancers were identified in the mammography group and 524 were identified in the control group. Of these women, 180 and 171 died of breast cancer during the 25 years of follow-up and the difference was not statistically significant. For the 25 year study period, 3,250 women in the mammography arm and 3,133 women in the control arm were diagnosed with breast cancer, of whom 500 and 505 died of breast cancer. Again, the difference was not statistically significant. In addition, the authors argue that the mammography screening program resulted in 22% of screen-detected breast cancers are over-diagnoses – meaning that the women were properly identified as having breast cancer, but the cancers were not life threatening. These women then underwent treatment (surgical or radiological) with all the risks and harms associated with such treatment. Critically, it is impossible to identify precisely who these 22% of women were. This is a statistical assessment rather than a clinical one. At present, there’s no diagnostically certain way to ascertain whether a cancer will develop into a life threatening cancer or go into remission without treatment.
It should go without saying that these results are utterly depressing. For decades now, the marketing of the medical community (including the public health community) has been that mammography screening is effective and vital. However, effective and vital programs should be able to show statistically significant results on a population basis. As the community of people concerned about breast cancer, we should be livid that there are not more effective screening and diagnostic tools available to reduce breast cancer mortality. It’s worth considering whether faith in the effectiveness of mammography screening has calmed some of this anger about the lack of better technologies. Unfortunately, the technology isn’t netting us the results that we should expect.
The study has now come under fire from a number of angles. Primarily, these relate to study design and the mammography technology used in the study. A couple of these criticisms are worth noting. The Diane Rehm show on NPR also discussed this study for a full hour yesterday and a podcast of that show is available here. (As an aside, I also believe that Dr. Kopans – despite having valid areas of concern – put on a clinic for how NOT to be an effective advocate for your cause by being overly dismissive of the data put forth in the study, willfully incorrect on the concept of over-diagnosis, and blind to the distinction between being paid to conduct research in a scientific manner and having a financial conflict of interest in the results of a study. I believe he undermined his credibility by doing so. Importantly, none of that necessarily reflects on the validity of his critiques and is merely a comment on effective advocacy.) Mammography technology used in the study
Commentators have raised the concern that the mammography technology used in the study was old or the mammograms improperly conducted. This argument is a bit paradoxical and self-defeating. Certainly, mammography technology has improved since the early to mid-1980s when the primary study data was collected, but that is and will always be the case when dealing with a study that has 25 years of follow-up. To be valid, the argument must present a reason to actually question the validity of the conclusions. Thus, the argument must be that more sensitive mammography would have significantly decreased the staging/size of the breast cancers detected amongst women who were diagnosed and significantly improved their treatment outcomes as a result.
This argument faces several problems. The study identified 22% of cancers in the mammography arm as over-diagnosis/over-treatment/non-lethal breast cancers. The charge that utilizing significantly more sensitive mammography would result in better outcomes dismisses the validity of these concerns without comment. The more sensitive the mammography screening technology, the more it will detect non-life threatening breast cancers and subject those women to over-treatment (which carries with it higher risk of CVD and lung cancer as discussed in the Cochrane Review).
Secondly, even if the argument is accurate that the mammography was flawed in the sensitivity of the technology, it still doesn’t resolve the issue. The fact of the matter is that 142 additional women were diagnosed in the mammography arm during the active phase of the study, yet these additional detections had no statistically significant effect on the mortality in the mammography group.
Likewise, there’s a timing issue as well. The average size of tumors detected in the mammography arm was 1.9mm vs 2.1mm in the control arm. Of breast cancers detected by mammography alone (i.e. not palpable during a physical exam) the average size was 1.4mm. Since these cancers were detected during annual mammograms, the argument must suggest that they were present but missed in the previous mammogram a year earlier – denying those women earlier treatment interventions. The problem with this argument is that it essentially relies on the idea that the missed cancers were all slow-growing enough that during a year gap, they still only reached an average size below that of the control group. And again, the mortality outcomes didn’t change. Improper randomization of the cohort population
Study participants were initially given a physical exam prior to randomization. As the study states, “The examiners had no role in the randomisation that followed; this was performed by the study coordinators in each centre. Randomisation was individual and stratified by centre and five year age group. Irrespective of the findings on physical examination, women aged 40-49 were independently and blindly assigned randomly to receive mammography or no mammography. . . . Women aged 50-59 were randomised to receive mammography or no mammography”
The suggestion has been that there was some bias introduced to the randomization process either by some exposure of the results of the physical exam to the randomizers or through some other error in randomization and that women with detectable lumps in the physical exam were preferentially allocated to the mammography arm.
Here again, to be valid, there has to be evidence that there were effects on outcomes. However, when women with detectable lumps at enrollment were excluded from the study, the difference between the intervention and the control arms remained the same – no statistically significant difference.
At the end of the day, what is desperately needed is a much better ability to differentiate between progressive and non-progressive cancers. There’s some work being put into the genetics of cancer that will hopefully yield results in this way. But mammography in the absence of the this ability is – tragically – not getting the job done.
Finally, the value of fighting over the results of the study doesn’t advance the ball significantly. The study does not in any way suggest that there is a valid reason to restrict access to mammography for women who desire to be screened. It does, however, suggest that there should be a much more accurate description of the value of that screening.
The views reflected in this blog are those of the individual authors and do not necessarily represent those of the O’Neill Institute for National and Global Health Law or Georgetown University. This blog is solely informational in nature, and not intended as a substitute for competent legal advice from a licensed and retained attorney in your state or country.