Mammography Screening: Keeping Women Alive

Abstract

Evaluation of: Tabár L, Vitak B, Chen TH et al. Swedish Two-County Trial: impact of mammographic screening on breast cancer mortality during 3 decades. Radiology 260(3), 658–663 (2011). In the 1980s, the periodic invitation of women aged 40–69 years for mammographic screening in the Swedish Two-County Trial showed a strong 30% reduction in breast cancer mortality. The result of 2–3-yearly mammographic examinations has persisted throughout the long follow-up of three decades. Through the richness of the collected and verified data, the trial has also demonstrated a substantial and absolute reduction in mortality risk. For each 414 women screened for 7 years (approximately four screening examinations), one breast cancer death was prevented. Transferring these outcomes to, for example, the national program of the UK, for every 1000 women aged 47–73 years attending the 3-yearly screenings (nine screening examinations) at least five to seven breast cancer deaths would be prevented. In recent follow-up papers by the Swedish trial group, the major human cost of screening (false-positive outcome, occurrence of interval cancer, overdiagnosis and radiation exposure) were judged to be in balance with the accurately demonstrated mortality benefit.

In the three continents of Europe, North America and Oceania, large-scale population breast cancer screening programs have been in existence for at least two decades. The launch of these programs was initiated by an enormous amount of evidence regarding breast cancer mortality reduction from randomized trials and observational studies coming from demonstration projects [1–3]. The randomized Swedish Two-County Trial was the first to demonstrate the life-saving effect of screening mammography, which resulted in early detection, and the increased effect of immediate therapy [4].

The radiologist László Tabár, together with a multidisciplinary working group of clinicians, research scientists and the Swedish National Board of Health and Welfare, initiated the Two-County Trial in the 1970s. It is now time, following the stream of publications, updates, independent reviews and overviews of the other Swedish trials, to comprehensively present the full spectrum of absolute numbers of prevented breast cancer deaths with increasing time of follow-up [5,6]. The next step is to yield impact measures. With this resulting information it will be possible for women, professionals in the healthcare community and decision makers to make informed choices regarding participation in screening, and balancing the intended and potential side effects of the screening examination and its sequelae.

Methods

The Swedish Two-County trial started in Kopparberg (currently named Dalarna, in 1977) and Östergötland (1978), two of Sweden's 24 counties. It was a randomized study at the community level; people were invited for a screening examination by a single-oblique mammographic view [4,7]. After exclusion of women with previously diagnosed breast cancer, there were 77,080 women in the active study population (ASP; invited for screening) and 55,985 in the passive study population (PSP; not invited, but receiving usual care). The screening phase lasted 7 years. A randomly selected group of women aged 40–49 years were invited to screening every 24 months, and women aged 50–74 years every 33 months. The first mortality results of the trial were published in 1985 [4], showing a statistically significant 30% reduction in mortality from breast cancer among women invited to screening with an 85% average participation rate.

All the women who were diagnosed with cancer during the 7-year screening phase of the trial were followed up for occurrence of breast cancer death, as have all women participating in the ASP and PSP for vital status. Follow-up occurred up to 31 December 2005 in Dalarna and up to 31 December 2006 in Östergötland (i.e., 28 and 29 years after the start of the trial). Case status and cause of death were determined by local committees consisting of heads of clinical departments. Later on in 1987, the Swedish Cancer Society set up an independent overview committee to review all collected data [8].

In the ASP group, 1426 tumors were diagnosed during the screening phase of the trial and 351 breast cancer deaths resulted from these tumors over a maximum follow-up of 29 years. In the comparative PSP group the numbers were 1042 and 367, respectively. These numbers form the numerator to the denominator of the years of follow-up that were delivered by the 77,080 and 55,985 subjects invited or not invited to screening. Using these data the breast cancer mortality rate in the ASP was calculated as 4.6 per 1000 woman-years in contrast to 6.6 per 1000 woman-years in the control population. The relative risk of these rates yielded a precise and statistically significant decrease in breast cancer mortality in women invited to screening of 0.69 (95% CI: 0.56–0.84), which can be interpreted as a 31% mortality reduction.

The authors converted these estimates to absolute numbers of breast cancer deaths prevented owing to screening. They also calculated the numbers of women needed for screening to prevent one woman from dying from this disease. These impact measures of screening are presented for different follow-up times after randomization.

Results

Following the 7-year screening period, the number of deaths from breast cancer in the ASP group increased from 206 following 10-year follow-up randomization to 351 after 29 years of follow-up. If no invitation to screening had occurred, the expected numbers of death would have been higher: 299 after 10 years and 509 after 29 years. Accordingly, the numbers of deaths prevented in the ASP increases from 71 to 158 over time. It is worth noting that after a 7-year screening phase with on average three screening episodes, most prevented deaths would have occurred, in the absence of screening, after the first 10 years of follow-up.

These results have a counterpart in the number of women needed for screening in order to prevent one breast cancer death. Taking into account the factual numbers of screening examinations and a sufficiently long period of follow-up, roughly speaking, one woman among every 400–500 women participating in a screening program will be prevented from a future breast cancer death.

Significance

The study provides vital information on the effectiveness of breast cancer screening, both at societal level and a personal level. The prevented numbers of deaths and the numbers needed to screen allow for a meaningful interpretation of the impact of mammographic screening: this form of representation clearly demonstrates the impact of screening. Screening has brought about an estimated average annual reduction in breast cancer mortality of 30% (95% CI: 15–44). Given an average annual mortality due to breast cancer among controls in the Swedish counties of 60 per 100,000 [9], this represents a reduction of 18 deaths per 100,000 women per year. All data were reviewed by the independent Swedish Overview Committee. Using the reviewed data, a somewhat different picture arises, but it is still highly significant: a 25% reduction yielding a reduction of 15 deaths per 100,000 women per year. On the other hand, all these estimates are likely to be somewhat diluted. Participation in each screening episode was relatively high (89, 83 and 84% for the first three rounds, respectively), and 85% completed all screening rounds. A substantial proportion (13%) of the control group had screening mammograms. Taking this into account, the efficacy must be higher than the 25–30% benefit presented in the paper, and the absolute numbers accordingly.

In Tabár's report, two aspects further demonstrate the benefits of periodic mammographic screening. First, by evaluating long-term longitudinal data and absolute rates rather than relative rates, a substantial and significant increased prevention of breast cancer death by mammographic screening was noted. Based on these absolute numbers, extrapolations could easily be made, for example to the UK: for every 1000 women aged 47–73 years attending 3-yearly screenings (nine screening episodes), five to seven breast cancer deaths would be prevented (compared with 15 deaths occurring in 1000 women if not screened). The second innovative analysis comprises the timing and maximalization of the follow-up data, now up to 29 years. These data clearly indicate that the effect of screening does not become fully apparent until 10 years or longer after the onset of screening. This is a paramount finding as recent literature are erroneously concluding a smaller effect of breast cancer screening on breast cancer mortality than previously reported [10,11]. Unfortunately, they did not provide comprehensive information on the distribution of the follow-up time in their study [11]. Judging from the average follow-up of 2.2 years, it is likely that this is simply too short a period to correctly estimate a mortality reduction due to screening.

Future perspective

The advantages of screening, for example the prevention of death, must be weighed against the disadvantages, for example the human costs of mammographic screening. Further research on radiation exposure, the physical and psychological effects of further investigation of suspicious mammographic findings in women who are ultimately found not to have breast cancer, the occurrence of interval cancers and the issue of overdiagnosis is required. Although the authors argued in their 2010 article [12] that these side effects were not significant, at least in the Two-County Study, antiscreening sentiments are (re-) emerging.

Currently, we are still debating the value of screening in terms of the modest benefit of screening and the numerous negative outcomes [13]. The debate on the relevancy of screening seems to be the focus of negative discussions, as shown in the extraordinarily high amount of attention paid to recent papers claiming minor effects of screening, which ignore five decades of research [14,15]. As we move forward, we must remember that this debate will not be resolved in the near future. Different views on the methodology used in screening research will maintain this debate.

Conclusion

The paper by Tabár et al. shows strong evidence for revising Peto et al.'s comment in their paper on cervical cancer, that screening has prevented the breast cancer epidemic in the Western world [16].

Executive summary

• The relative survival of breast cancer when diagnosed and treated symptomatically is approximately 75%. Mammographic screening enables better detection of impalpable and asymptomatic breast cancer.

• The Two-County Trial with 2–3-yearly mammographic screening has clearly demonstrated the life-saving effect of early breast cancer detection and therapy.

• The article innovatively quantifies the ultimate benefits of participating in a screening program in terms of absolute impact measures following more than 20 years of follow-up.

• Of 1000 women who underwent nine mammographic screening examinations between the age of 47–73 years, five to seven breast cancer deaths would be prevented.