Breast cancer is the most common cancer diagnosed in US women, and the second leading cause of death from cancer in US women.12 ACS guidelines for breast cancer screening in average-risk women were last updated in 2003,2 and screening guidelines for women at very high risk were last updated in 2007.3 Guidelines for women at very high risk are appropriate for women with known or suspected inherited susceptibility to breast cancer, or women who have undergone mantle radiation to the chest at an early age for Hodgkin lymphoma3 (Table 2). Guidelines for the early detection of breast cancer in average-risk women consist of a combination of regular clinical breast examination (CBE), counseling to raise awareness of breast symptoms beginning at age 20 years, and annual mammography beginning at age 40 years (Table 2).
Women should undergo CBE every 3 years between the ages of 20 and 39 years, and annually after age 40 years. This exam should take place during periodic health examinations. When CBE is performed, it is an opportunity for health care professionals to review and update the woman's family history, discuss the importance of early breast cancer detection, and answer any questions she may have about her own risk, new technologies, or other matters related to breast disease. During these discussions, health care professionals should emphasize the importance of awareness and recognition of breast changes, and of contacting their physician promptly if changes are perceived. They should also emphasize the importance of awareness of a family history of breast and ovarian cancers in first- and second-degree relatives on both the maternal and paternal sides of the family. An opportunity to update family history should take place during encounters for other preventive care or screening. Approximately 8.4% of all women report a family history of breast cancer in first-degree relatives, and approximately 2.7% of women between the ages of 20 and 29 years report a family history of breast cancer in first-degree relatives.13 Thus, it is important to take and regularly update a family history at a young age, because some younger women will be candidates for beginning breast cancer screening before age 40 years. In an examination of data from the 2005 Cancer Control Module of the National Health Interview Survey (NHIS), Hall and colleagues estimated that approximately 1.4 million US women (<1%) had a family history of breast cancer that, based on criteria established by the US Preventive Services Task Force (USPSTF),14 warrants a referral for genetic counseling and evaluation for genetic testing.15 However, <2% of respondents who would be candidates for genetic counseling reported having been tested. This low rate of testing should be taken as a clear indication for additional research to better understand both the opportunity structure for counseling and barriers to testing, but almost certainly one factor associated with low rates of testing is suboptimal attention to family history in the primary care setting. Research on the degree to which family history data are gathered, gathered completely, and then used indicated that this simple, initial step in risk assessment generally is not done or is not done competently.16
Although the ACS no longer recommends monthly breast self-examination (BSE), women should be informed about the potential benefits, limitations, and harm (principally the possibility of a false-positive result) associated with BSE. Women may then choose to perform BSE regularly, occasionally, or not at all. If a woman chooses to perform periodic BSE, she should receive instructions in the technique and periodically have her technique reviewed. Although the elimination of the direct recommendation for monthly BSE has seemed counterintuitive, there is little direct evidence to indicate that monthly BSE has a distinct advantage over a heightened sense of breast awareness, ie, an attentiveness to changes in the breast that are noticed while engaging in normal, daily activities or simply perceiving a change.17
The ACS recommends that average-risk women should begin annual mammography at the age of 40 years. Women also should be informed about the scientific evidence demonstrating the value of detecting breast cancer before symptoms develop, and that the balance of benefits to possible harm strongly supports the value of screening and the importance of adhering to a schedule of regular mammograms.18 The benefits of mammography include a reduction in the risk of dying from breast cancer, and if breast cancer is detected early, less aggressive surgery (ie, lumpectomy vs mastectomy), less aggressive adjuvant therapy, and a greater range of treatment options. Women should also be told about the limitations of mammography, specifically that mammography will not detect all breast cancers, and that some breast cancers detected with mammography may still have poor prognosis. Furthermore, women should be informed about the potential for false positives, some of which may not be resolved with additional imaging, and that if not, a biopsy will be required to rule out the possibility of breast cancer.
There is no specific upper age at which mammography screening should be discontinued. Rather, the decision to stop regular mammography screening should be individualized based on the potential benefits and risks of screening in the context of overall health status and estimated longevity.19 As long as a woman is in good health and would be a candidate for breast cancer treatment, she should continue to be screened with mammography.
In 2007, the ACS issued new guidelines for women who were known or likely carriers of a BRCA mutation and other rarer high-risk genetic syndromes, or who had been treated with radiation to the chest for Hodgkin disease.3 Annual screening mammography and magnetic resonance imaging (MRI) starting at age 30 years are recommended for women with a known BRCA mutation, women who are untested but have a first-degree relative with a BRCA mutation, and women with an approximately 20% to 25% or greater lifetime risk of breast cancer based on specialized breast cancer risk estimation models capable of pedigree analysis of first- and second-degree relatives on both the maternal and paternal sides. Although the Breast Cancer Risk Assessment Tool, more popularly known as the Gail model, provides a good generalized measure of short- and long-term risk based on a woman's age, ethnicity, history of breast biopsy and breast cancer, age at menarche, parity, and age at first live birth, it does not have the capacity to analyze detailed family histories, including first- and second-degree relatives on both the maternal and paternal sides.20 Thus, although individual lifetime risk estimates generated from the Gail model can exceed the threshold of approximately 20% or greater risk to age 90 years, the elevated risk may be due to risk factors other than family history. To estimate risk of breast cancer in women with a significant family history who have not undergone genetic testing and do not have an affected relative who has tested positive, health professionals should use specialized software that can address family history in first- and second-degree relatives on both the maternal and paternal sides. There are several models that can estimate risk based on complex family histories and assist clinicians to estimate breast cancer risk or the likelihood that a BRCA mutation is present, including the Claus,21 Tirer-Cusick,22 BRCAPRO,23 and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm24 models. Some of these models also can accommodate complex family histories and conventional risk factors, such as reproductive history or a history of prior breast biopsy. A link to supplemental material related to these models is included in the online publication (http://caonline.amcancersoc.org/cgi/data/57/2/75/DC1/1).3
Although MRI may eventually prove to be cost-effective and advantageous for women at elevated risk due to other combinations of risk factors, at this time recommendations for annual screening mammography and MRI are based strictly on known or estimated high-risk mutation carrier status or history of high-dose radiation therapy at a young age. The expert panel concluded that there was insufficient evidence to recommend for or against MRI screening in women with a 15% to 20% lifetime risk as defined by these same family history-based risk estimation models, or women with a history of ductal or lobular carcinoma in situ, a history of biopsy-proven proliferative lesions, a prior history of breast cancer, or extremely dense breasts. MRI is not recommended for women at average risk, although investigations are underway to determine whether MRI should be considered for other higher-risk groups.3, 25
Commentary on Updated Breast Cancer Screening Guidelines From the USPSTF
In late 2009, the USPSTF updated their guidelines for breast cancer screening,26 sparking enormous controversy based on several changes from the previous recommendation.27–29 Whereas the Task Force previously had endorsed mammography screening for women aged 40 to 49 years,29 the new guidelines downgraded mammography screening in the 40s to a C rating, stating, “The USPSTF recommends against routine screening mammography in women aged 40 to 49 years.”26 (The USPSTF uses a rating system for preventive services: A, recommendation for the service, with expectation of high net benefit; B, recommendation for the service, with expectation that the net benefit is moderate; C, recommendation against routinely providing the service; D, recommendation against the service; and I, insufficient evidence to recommend for or against the service.30) For women in their 40s, the Task Force concluded that “the decision to start regular, biennial screening mammography before the age of 50 years should be an individual one and take into account patient context, including the patient's values regarding specific benefits and harms.”26 (The magnitude of the controversy led the USPSTF subsequently to amend their recommendation for women in their 40s by dropping the statement recommending against routine screening, and emphasizing only the value of individual, informed decision making.31) The USPSTF also changed the recommended screening interval for women aged 50 to 74 years from 1 to 2 years to biennial screening, and concluded that there was insufficient evidence (I rating) for or against screening women aged ≥75 years and use of digital mammography or MRI for breast imaging. Finally, the USPSTF concluded that there was insufficient evidence (I rating) to recommend for or against CBE in women aged ≥40 years, and recommended against physicians teaching BSE (D rating).26
These changes in the recommendations were unanticipated. Based on prior evidence reviews, the only new data that the USPSTF was likely to consider was from the results of the UK Age Trial, which randomized women aged 40 to 41 years to a group invited to screening versus usual care to measure the effect of mammography in a group of women in their 40s without any age migration past age 50 years.32 The Age Trial observed a 17% reduction in breast cancer mortality in the group invited to screening compared with the control group (relative risk [RR], 0.83; 95% confidence interval [CI], 0.66-1.04; P = .11), a result that was consistent with results from earlier trials, and also consistent with the less sensitive screening protocol applied in this study,33 ie, double-view mammography on the first exam and single-view mammography on subsequent exams. The USPSTF updated the meta-analysis of all trials excluding the Edinburgh trial, and concluded that an invitation to mammography was associated with a 15% reduction in the risk of dying from breast cancer among women screened in their 40s, and a 14% reduction in risk for women screened in their 50s.34 However, based on the rate of false-positive results experienced in routine screening, and an updated estimate of the number needed to invite to screening to save 1 life, the USPSTF concluded that the additional benefit from beginning screening at age 40 years versus age 50 years was small, and did not sufficiently outweigh the harm associated with screening.26 In their analysis, it was estimated that 1904 women aged 40 to 49 years would have to be invited to mammography to save 1 life, versus 1339 women aged 50 to 59 years. The decision to give mammography for women in their 40s a C rating was principally influenced by a decision analysis concluding that the majority of the benefit from screening women aged 40 to 74 years was due to screening between the ages of 50 and 74 years (17% mortality reduction), and that beginning screening at age 40 years only increased the mortality reduction an additional 3%.35 Extending the screening interval from 1 to 2 years to biennial screening also was unexpected, and was based on statistical modeling. In the decision analysis, screening women aged 50 to 74 years biennially was estimated in the decision analysis to achieve between 70% and 99% of the benefit, and to reduce harm by 50%.26, 35
Although the controversy over the guidelines change was commonly interpreted as either a step toward health care rationing, or an example of the public's distrust of science and expert groups, there was little questioning of whether the USPSTF's methodology actually had provided a sound estimate of the balance of benefits and harm from mammography for women in their 40s. The ACS argued that even without disputing the underlying methodology, the difference between the number needed to invite for women in their 40s and in their 50s (1904 vs 1339) was still acceptable because of the greater years of life gained by preventing a death from a breast cancer diagnosed before age 50 years.36 Furthermore, due to the USPSTF's decision to limit the evidence review to the randomized trials of breast cancer screening,37 the underlying data and methodology used to estimate the benefit of screening underestimate the true value of mammography for women of all ages. In comparison, the ACS includes a broader body of evidence in formulating its own guideline for breast cancer screening, and thus has reached different conclusions about the value of screening women in their 40s, annual versus biennial screening, CBE, and BSE.2 The ACS has not taken a position on digital mammography, other than to regard it as an acceptable alternative to screen-film mammography, and probably superior to screen-film for younger women with heterogeneously dense or very dense breasts.38, 39 The ACS's current position on screening with MRI was noted above.
Although it is common to observe that experts can disagree on the interpretation of data, the USPSTF's new guidelines are the result of relying on older, limited data, using efficacy estimates from meta-analyses as a measure of effectiveness, and then making a judgment call on the relative value of screening in different age groups based on the balance of benefits and harm. Furthermore, some measures of disease burden were chosen over others, and there was a reliance on modeling over empirical data. Together, these methodological decisions applied to limited data resulted in an underestimate of benefit and an overestimate of harm, and then a judgment call that mammography was not recommended before age 50 years, and should only be done every 2 years after age 50 years.
The USPSTF addressed disease burden for all women in terms of incidence and mortality, and noted that incidence increases with increasing age. The decade risk of being diagnosed with breast cancer is 1 in 69 (1.5%) between the ages of 40 and 49 years, 1 in 42 (2.2%) between the ages of 50 and 59 years, and 1 in 29 (3.4%) between the ages of 60 and 69 years.40 It is also worth noting that the risk between ages 45 and 50 years is 1 in 114, and for ages 50 to 55 years it is 1 in 92, more similar over that 10-year period than the contrast between 40 to 49 years versus 50 to 59 years, which also is not very dissimilar.40 Women in their 40s have about two-thirds of the decade risk experienced by women in their 50s. Although the USPSTF focused primarily on the absolute risk of developing breast cancer and the meta-analysis estimate of the reduction in the risk of death, 2 other measures of disease burden are equally if not more relevant, ie, incidence-based mortality and premature mortality.
Breast cancer is the leading cause of premature mortality among women due to death from cancer, and a leading cause of premature mortality from all causes of death.41 In the first trial of breast cancer screening (the Health Plan of New York trial), Shapiro and colleagues showed great insight in choosing the age range 40 to 69 years for the study group, because they observed that it included about 75% of the years of potential life lost (YPLL) due to a death from breast cancer before age 80 years. A diagnosis between ages 40 and 49 years accounted for 34% of the YPLL before age 80 years, and a diagnosis between ages 50 and 64 years accounted for 38%.42 Between 2004 and 2006, 18% of deaths from breast cancer were attributable to a diagnosis between ages 40 and 49 years, compared with 22% for women diagnosed between ages 50 and 59 years, a level of disease burden that is not dissimilar.43 Ironically, in the decision analysis, Mandelblatt and colleagues noted that if the goal of the screening program were to efficiently maximize life years gained, then the preferred strategy would be to screen every 2 years beginning at age 40 years.35
As the case can be made that breast cancer is an important health problem for women in their 40s, is the evidence of benefit so small and the magnitude of harm so great that their previous recommendation for screening women in their 40s should be rescinded? With respect to benefit, the randomized trials provided convincing evidence that mammography screening saves lives principally by advancing the lead time and reducing the incidence of advanced disease.18 However, the summary RR from meta-analysis of all the trials is not a good measure of effectiveness for several reasons. First, the trials measured the effectiveness of an invitation to screening, not actually being screened. Second, in a meta-analysis, trials with ineffective protocols are combined with trials that had effective protocols. It is well established that in some of the early trials, women were screened with protocols that were especially limited for women under age 50 years.44 These ineffective protocols are characterized by long screening intervals (≥24 months) and single-view mammography. It is also evident from examining the RR of being diagnosed with an advanced breast cancer in the different trials why some trials showed significant mortality reductions and some did not, because there is a strong association between the magnitude of the risk reduction of being diagnosed with an advanced breast cancer and the eventual observed mortality reduction.45 This was especially the case with 2 second-generation trials, the Gothenburg Trial and the Malmo trial, which screened women at a shorter interval (12 to 18 months) with double-view mammography and observed 44% and 36% mortality reduction, respectively.46, 47 Although meta-analysis has consistently shown a modest mortality reduction,45, 48, 49 the magnitude of the mortality reduction from meta-analysis for women of all age groups is reduced by intention-to-treat estimates and the influence of less effective trials. Because individual trials and meta-analysis have shown that mammography is efficacious for women aged 40 to 69 years, large, long-term observational studies of modern mammography are better sources of data on the effectiveness of mammography screening, including age-specific effects, and they should be used to shape health policy today.
In 2005, Gabe and Duffy described the recent history of evaluating mammography service screening in Annals of Oncology.50 They examined studies published between 1990 and 2004 that compared breast cancer mortality trends before and after the introduction of screening, or in screened and unscreened cohorts. They identified 38 nonrandomized studies, including case series, case-control studies, cohort studies, nonrandomized clinical trial (non-RCT) comparison studies, and studies that attempted to measure time/trend descriptive epidemiologic results. These studies were carried out in Australia, Italy, the Netherlands, Finland, Denmark, Sweden, the United Kingdom, and the United States. Many more studies have been published since their review of the literature, and although many were captured in the USPSTF's search of the literature, all were excluded from consideration in favor of examining only the randomized trials.37
Compared with the results of individual RCTs and meta-analyses, there was a range of estimated benefits observed in the service screening programs associated with an invitation to screening and actual exposure to screening, some that significantly exceeded the results from the RCTs and some that showed weaker benefits.50 Results from case-control studies showed mortality reductions associated with exposure to screening ranging from 25% to 58%. A wider range of breast cancer mortality reductions was observed in cohort studies and non-RCT comparative studies, depending on whether the analysis was based on an invitation to screening or actual exposure to screening. Cohort study breast cancer mortality reductions associated with an invitation to screening ranged from 8% to 41%, whereas mortality reductions associated with actual exposure to screening ranged from 37% to 44%. In general, however, the results are at least as good, or better, when compared with the trials that demonstrated the largest reductions in breast cancer mortality. For example, Coldman et al evaluated outcomes in the Screening Mammography Program of British Columbia, comparing outcomes among participants who had at least 1 mammogram with the expected incidence and survival rates among British Columbian women who had not participated in the program.51 During the period of evaluation (1988–2003), 598,690 women underwent 2,196,441 mammograms (average 3.7 screening exams per woman), and 14,247 program participants were diagnosed with invasive breast cancer. Among nonparticipants, 19,913 invasive breast cancers were diagnosed. A breast cancer mortality ratio was calculated as the ratio of observed to expected mortality. The results for women of all ages showed a mortality ratio of 0.60 (95% CI, 0.55–0.65; P < .0001); the mortality ratio was 0.61 (95% CI, 0.52–0.71) for women aged 40 to 49 years, and 0.63 (95% CI, 0.52–0.77) when cases diagnosed after age 50 years were censored. There was no significant difference between the mortality ratios for women aged 40 to 49 years versus ≥50 years.51 In another example, an analysis of service screening in 2 Swedish counties (Dalarna and Ostergotland) compared the mortality rates from breast cancer in the 20 years after the introduction of screening (the screening epoch) with the corresponding rates in the 20 years preceding the introduction (the prescreening epoch).52 After adjustment for changes in incidence, among women aged 40 to 69 years exposed to screening, there was a statistically significant 44% (RR, 0.56; 95% CI, 0.49–0.64) reduction in mortality in the screening epoch compared with the prescreening epoch, and 16% (RR, 0.84; 95% CI, 0.71–0.99) fewer deaths in women not exposed to screening, which can be attributed to improvements in therapy and awareness. Among women aged 40 to 49 years at the time of diagnosis, the investigators observed a statistically significant 48% (RR, 0.52; 95% CI, 0.4–0.67) breast cancer mortality reduction among women exposed to screening, and a nonsignificant 19% reduction in women who were not screened. For women aged 40 to 49 years, the number needed to screen 5 to 6 times over a 10-year period to prevent 1 breast cancer death after 20 years of follow-up was 726.52
It is worth noting that the mortality reductions observed in these service screening evaluations are similar for women aged 40 to 49 years compared with women aged ≥50 years, and significantly greater than what is estimated by a meta-analysis of the randomized trials. Furthermore, as shown in another Swedish study that examined the impact of screening on the risk of being diagnosed with an advanced breast cancer, a similar pattern is observed between the magnitude of the rate of reduction in the risk of being diagnosed with advanced disease and the observed mortality reduction. Based on a total of 10,177,113 person-years of observation and 23,092 breast cancers among women aged 40 to 49 years exposed to screening, there was a significant 45% reduction in the risk of being diagnosed with a tumor >2 cm compared with tumor sizes in the prescreening epoch (RR, 0.55; 95% CI, 0.46–0.66), and among women aged 50 to 69 years, there was a 33% reduction in being diagnosed with a tumor >2 cm in the population exposed to screening compared with the population in the prescreening epoch (RR, 0.67; 95% CI, 0.62–0.72).53
What about the harm of screening? The harm of mammography includes radiation risk, short- and long-term anxiety associated with false-positive results, biopsy for benign lesions, and the possibility that some breast neoplasms detected on mammography are nonprogressive and thus overtreated. The risk of a radiation-induced cancer from low-dose mammography is so low that although theoretically possible, it would be impossible to measure empirically.54, 55 The possibility that some breast cancers are nonprogressive has been a source of considerable interest, but the weight of evidence from long-term studies suggests that it is a small problem and mostly confined to ductal carcinoma in situ.56, 57 Studies that have concluded that the magnitude of overdiagnosis is large have commonly examined population data over a period of limited duration and confused overdiagnosis with background increases in incidence and increased incidence associated with the lead time gained from screening.58 The more common and more directly measurable harm associated with mammography includes the inconvenience from additional imaging resulting from false positives, benign biopsy for abnormal findings, and short- and long-term anxiety resulting from false-positive results. The USPSTF cited modern US data from the Breast Cancer Surveillance Consortium (BCSC), observing that for every breast cancer detected in women in their 40s, there will be 556 screening mammograms, 47 diagnostic mammograms, and 5 biopsies.26 Data from the BCSC also demonstrate what has been observed for years, ie, that the accuracy and effectiveness of mammography increases incrementally with increasing age, and is not measurably different between 5- and 10-year age groups.59 With respect to short- and long-term anxiety, most studies observe that there is some anxiety associated with screening, and as would be expected, greater anxiety associated with false-positive results.60, 61 However, for most women, anxiety is of limited duration and without lasing effects, and it also has been shown that clear communication about imaging results is associated with reduced anxiety.62 Clearly, methods for improved communication and strategies to reduce avoidable anxiety are a priority area for applied research, and a priority area for adoption of successful strategies. However, it is misleading in these discussions to describe the totality of negative aspects of screening mammography as harm, because they do not all constitute injury, and some are clearly amenable to interventions that can reduce the adverse experiences and consequences. Moreover, a study by Schwartz et al63 revealed that women are aware of the likelihood of a false-positive result, accept false-positive results as a part of screening, and do not regard false positives as significant harm in the context of the underlying goal of early breast cancer detection.
The USPSTF's guideline update was informed by 6 models of breast cancer incidence and mortality that were applied to estimate outcomes of different screening scenarios for specific age groups.35 Although each of the models shares some common inputs, other inputs such as assumptions about age-specific benefits of screening, the duration of the potential lead time, overdiagnosis, and mammography performance may vary between models. The models examined the estimated difference in mortality reduction, life years saved, and harm with annual and biennial screening for different combinations of age groups of women. The models predicted that biennial screening achieved an average of 81% of the benefits of annual screening, with individual model estimates ranging from 67% to 99%.35 Starting screening at age 40 years versus age 50 years resulted in only an estimated median 3% improvement in the reduction in breast cancer mortality, with no difference in the additional reduction in deaths based on annual or biennial screening. However, if life years gained was the outcome measure of interest versus lives saved, half of the models favored extending annual screening to women in their 40s versus extending annual screening to women up to age 79 years. The authors also observed that the rate of false positives was approximately twice as high with annual screening compared with biennial screening. The latter outcome is at variance with evidence from screening programs, and likely is due to the linear nature of the models.64
It has been clear since the earliest analyses of the randomized trials that the duration of the detectable preclinical period is shorter in younger women compared with older women, and increases as a woman ages.18, 65 The early indication that screening intervals of 24+ months were inefficient for women younger than 50 years was apparent from an interval cancer rate that was twice that of women aged ≥50 years.65 Moreover, randomized trial data and more recent evaluations of service screening46, 47, 66 have shown better outcomes associated with 12- to 18-month screening intervals for women in their 40s, and recent observational studies have shown better outcomes from annual versus biennial screening in both women aged 40 to 49 years and those aged ≥50 years.53, 67, 68
Although modeling has many advantages for shaping health policy when large prospective studies are not feasible, it is not a substitute for empirical data. Furthermore, the logic for combining all model outcomes into a meta-analysis to produce the best summary endpoint, as described by Mandelblatt et al,35 is even less persuasive than the idea that a meta-analysis of existing randomized trial data provides the best and most unbiased estimate of the true benefit of mammography, or constitutes a sufficient systematic review of the evidence.69 However, although the importance of annual screening is especially evident for women younger than 50 years, the advantages of annual versus biennial screening diminish as a woman ages,67, 68 and in a program of high quality, the potential for identifying an age after menopause when some women could be screened at a longer than annual interval is an important area for continued research. Whether it is reasonable to ask postmenopausal women to forgo as much as 33% of the benefit that would be achieved with annual screening on the basis of the estimate that biennial screening would reduce the rate of false positives by half35 is not self-evident.