Recent data show that mammographic screening of asymptomatic women is effective and essential

Screening mammography is employed widely both in the United States and abroad. Several recent publications, however, have claimed that screening does not reduce breast cancer mortality and causes over-diagnosis resulting in unnecessary mastectomies. Others claim that mammographic screening of asymptomatic women is effective and essential. This controversy is debated in this Point-Counterpoint by four of the world's leading experts in screening mammography. Although screening mammography is hotly debated, recent randomized controlled trial (RCT), case-control (comparing women who had developed breast cancer with matched control subjects who had not), and service screening (women invited to a screening service) data confirm that screening mammography saves women's lives. A 29-year follow-up of the Swedish two-county trial showed that women aged 40–74 “invited to screening” had a statistically significant 31% fewer breast cancer deaths than uninvited women.1 This is consistent with the first published results from this study that showed a significant 30% reduction in breast cancer mortality among women invited to screen.2 Three other population-based trials in Sweden (1976–1990) compared groups of women invited to screening mammography to a usual-care (non-invited but may or may not have received mammograms as part of their normal health care) control group. While all four Swedish trials differed in numbers of rounds of screening (2–5), screen intervals (18–33 months), numbers of mammographic views (1 or 2), and ages of women at entry to the study (39–74 years), the four trials combined showed a statistically significant 23% reduction in breast cancer deaths in the “invited to screen” group compared with uninvited women (95% confidence interval: 12%–33%).3,4 Two other population-based RCTs, the Health Insurance Plan of New York (HIP) and Edinburgh trials, each demonstrated statistically significant benefit from “invitation to screening.”5–8 Recent analysis of RCTs done for the U.S. Preventive Services Task Force yielded a statistically significant mortality benefit from screening mammography in women of each age decade from 40 to 69 years.9,10 RCTs underestimate the true benefit of screening because not all women randomized to “invitation to screening” attend screening, but all breast cancer deaths in this group count toward mortality in the “invited” group, whether they attended mammography or not. Of women invited to screening in the HIP trial, only 67% attended the first round, 53% attended the second round, and less than half attended 3rd and 4th rounds.5–7 Similarly, women randomized to the uninvited group can receive mammography outside the trial, but their breast cancer deaths are counted as mortality among the control group. It was estimated that 20%–30% of control group women in RCTs had at least one mammography exam outside the trial.11 Most RCTs had longer screening intervals than the ACR recommendation of annual screening, some used only a single view per breast rather than two as ACR recommends, and all were performed with old technology. More recent case-control and service screening studies have assessed the benefit of mammography among women who actually received screening compared to women who did not. A case-control study of the Dutch population-based screening program showed that mammography screening reduced breast cancer deaths by 49% compared to unscreened women.12 These results are consistent with service screening programs showing that women screened regularly have 40%–76% fewer breast cancer deaths than unscreened women.13–15 Ironically, increased criticism of screening mammography is occurring just when we are starting to get mammography “right” with high technical quality, improved interpretation, and minimally invasive breast biopsy. The real question is not whether we should be screening U.S. women with mammography, but how we can overcome socio-economic barriers16 and extend the proven benefits of mammography screening to all U.S. women over age 40. Over the past few years, multiple papers have questioned the benefit of mammographic screening17,18 and have shown that screening causes over-diagnosis (diagnosis of disease that would never cause symptoms), leading to increased mastectomy use.19,20 The premise of mass screening is that earlier intervention will prevent advanced disease and therefore reduce breast cancer deaths. However, an overview of seven countries with long availability of breast screening, including the United States, found no reduction in late stage disease since screening started.21 This is not surprising. Screening detects breast cancers that are a little over 1 cm in diameter on average, rather than 2 cm for clinically detected disease.22 This represents a reduction of 1–2 volume doublings of the 32 necessary to reach 2 cm (Ref. 22) and is equivalent to a few months growth for aggressive, fast-growing cancers22 which therefore easily “slip through the screen.” On the contrary, screening effectively detects slow-growing and dormant tumors, resulting in over-diagnosis. Screening advocates usually cite trials from Sweden. But despite high participation since screening started in the mid-1980s, the breast cancer mortality reduction in Sweden has been only 16% from 1989 to 2006 in women aged 50–69 years, much less than in Denmark (−26%) and Norway (−23%), which had only limited screening, e.g., only 20% of women aged 50–69 years were offered screening in Denmark.24 The average reduction in breast cancer mortality in Europe has been almost twice as large in younger, non-screened age groups as in those screened, and equally large in countries with and without screening.24 The primary cause of improvements in mortality has been more effective therapy, not screening.25 Screening's effect on the recorded incidence of ductal carcinoma in situ and early stage invasive breast cancers has been massive, both increasing several-fold since screening began.25 This is due to the detection of cancers which would never become clinically evident, let alone lethal, without screening (over-diagnosis). Screening participation turns thousands of healthy women into breast cancer patients, whom we treat with surgery, radiotherapy, and possibly chemotherapy, as we cannot tell which cancers are over-diagnosed. If screening reduces breast cancer mortality by 15%, which is optimistic, ten times as many women receive unnecessary treatment as will benefit.25 The U.S. status quo, aggressive annual screening starting at age 40 that our colleagues continue to advocate,26 likely costs close to one million dollars for every life-year saved, 10–20 times more than we accept for other interventions.27 Independent panels in both the United States9 and Canada28 found that screening women in their 40s is questionable and that biennial screening starting at age 50 is preferred, in agreement with policies in Europe. Hubbard et al. have shown that annual screening does not reduce the proportion of advanced cancers. Furthermore, the cumulative false-positive rate after 10 years for women ages 40–59 was 61% with annual versus 42% with biennial screening.29 Certain professional groups benefit from the status quo,30 while many doctors are reluctant to change established beliefs about cancer biology. Currently, the United Kingdom is reviewing the rationale for breast screening because of the new evidence.31 We question if screening mammography is justifiable at all, but especially the aggressive annual screening of younger women. Screening mammography saves lives.1–10,12–15 Our colleagues distract readers from this fact by raising issues such as over-diagnosis, irrelevant population mortality data, and misstated RCT data. They confuse over-diagnosis with overtreatment (mastectomy). The great majority of cancers, including high-grade DCIS (the DCIS predominately detected by mammography), will progress to metastatic breast cancer (distant metastases) if untreated. Recent estimates of the rate of over-diagnosis by screening mammography range from less than 1% to 10%.9,10 Since no marker exists to identify the few cancers that would not progress, it would be irresponsible to advise against treating a diagnosed breast cancer. Breast cancer treatment in the United States is now highly personalized, taking into account cancer biology, patient age, and health. The vast majority of women with mammographically detected breast cancers are offered breast conserving therapy. Mastectomy is reserved for women who are not candidates for, or decline, breast conservation. In the United States, at diagnosis less than 5% of breast cancers are stage IV (distant metastases) and less than 6% are stage III (locally advanced) (SEER Cancer Statistics Review, 1975–2008, and SEER Survival Monograph, 2007)). It is faulty logic to criticize screening mammography by pointing to a lack of reduction in late stage cancers in a population where more than one-third of eligible women are not screened. In truth, mammography screening yields a significant decrease in late stage breast cancers, decreasing morbidity as well as mortality, and empowering more women to benefit from breast-conserving treatment. Our colleagues' statement “If screening reduces breast cancer mortality by 15%, which is optimistic, ten times as many women receive unnecessary treatment as will benefit” is incorrect. They inappropriately apply RCT results for women 40–49 to all women age 40 and over, underestimating lives saved by screening by as much as two-fold.1–10 Case control and service screening results show that women attending screening have mortality reductions ranging from 28% to 65%.11–15 The consequences of less screening are later stage breast cancers and increased breast cancer mortality. Failure to provide screening mammography puts women's lives at risk. Evidence-based medicine is about relying on the best available evidence and discarding that which is flawed. Our opponents seem to prefer the most optimistic results without assessing their validity, and to disregard unwelcome ones. The Edinburgh trial is widely recognized as untrustworthy. It randomized 87 general practices, which led to substantial baseline imbalances. Twenty-six percent of the controls belonged to the highest social group versus 53% of the intervention group, leading to a spurious 26% reduction in cardiovascular mortality in those screened, which obviously cannot be a screening effect. The Swedish Two-County trial randomized only half as many clusters, but we cannot know if this led to similar baseline imbalances, as only the age distribution has been published, where there was a difference between groups. The lead investigator has not accommodated requests for supplementary baseline data. Long follow-up cannot compensate for fundamental flaws, and trials of high methodological quality (the Malmö and Canadian trials) did not find an effect of screening.28 Reports have repeatedly pointed out that case-control studies should not be used to assess cancer screening because the small possible effect is prone to substantial bias that favors screening.32 These studies compare breast cancer mortality in attendees with non-attendees, but the attendees are predominantly the affluent and healthy. When the Malmö randomized trial was analyzed as a case-control study, it showed a 58% reduction in breast cancer mortality after 9 years when in fact the trial found only a 4%, non-significant, reduction.32 Benefits should be presented in context. A 15% relative risk reduction for ages 40–49 is 0.047% in absolute terms.28 Increased screening participation also means proportional increased over-diagnosis, while increased screening frequency does not provide further benefit.28 Flawed studies have determined our current screening policy. We must stop uncritically accepting results we like and face the facts: comparisons between screened and non-screened countries and age groups clearly show that better treatment, not screening, has caused the decline in breast cancer mortality.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.