Surveillance mammography in older women—should we press on?

Early detection of breast cancer through mammographic screening and advances in breast cancer treatment have led to an increase in the number of women with personal history of breast cancer who are alive and disease-free many years after diagnosis. Currently, there are 4.3 million women in the United States with a previous diagnosis of invasive breast cancer, 67% of whom are aged 65 years or older.¹ Breast cancer survivors are at risk of developing recurrent cancer and are at a higher risk of developing a new primary breast cancer than age-matched controls.^2,3 The goal of surveillance mammography is to improve breast cancer outcomes in these women through early detection. Current guidelines^4-9 recommend annual surveillance mammography in women with personal history of breast cancer, extrapolating evidence from randomized controlled trials of mammographic screening, even though these randomized controlled trials did not include women aged older than 74 years or women with personal history of breast cancer.¹⁰ A randomized controlled trial directly evaluating mortality reduction and other benefits of surveillance mammography in older women or addressing when it should be discontinued is unlikely to ever be undertaken. So, although observational studies are limited by lead time, selection, and other biases, we must rely on them for data to inform recommendations for surveillance mammography in patients with personal history of breast cancer.¹¹

In this issue of the Journal, Richman et al.¹² contribute to our understanding of surveillance mammography performance in older women. They describe short-term outcomes after surveillance mammography in 33 607 women aged 67-100 years in a cohort study using Surveillance, Epidemiology, and End Results–Medicare data, stratified by age at diagnosis and time since diagnosis with a median follow-up time of 12.1 years. Subsequent diagnostic imaging with mammography or ultrasound was performed after 11.5% of surveillance mammograms, which was a proxy for abnormal interpretation rate or recall rate. Diagnostic imaging and biopsy declined with age and time since diagnosis. Cancer diagnosis rate was 7.1 per 1000 mammograms for second primary cancers and 3.9 per 1000 for in-breast recurrent cancers, totaling 11.0 cancers detected per 1000 surveillance mammograms. Cancer recurrence was more likely to be detected in the first 5 years after initial diagnosis than later, whereas second primary cancers were more likely to be detected more than 5 years after initial diagnosis than earlier for all age groups. Positive predictive value among women with subsequent imaging (PPV1) for either recurrence or second primary was 6.0%. Positive predictive value among women who underwent biopsy (PPV3) was 29.5%. Second primary breast cancers diagnosed were stage 0 or I in 80.3%, with stage 0 cancers comprising a larger proportion of cases in the first 5 years after diagnosis compared with the second 5 years. This study did not address long-term outcomes such as the need for intensive treatments or mortality.

Compared with screening mammography benchmarks from the Breast Cancer Surveillance Consortium derived from mammograms performed during a similar period (2007-2013),¹³ Richman and colleagues¹² report mammography performance metrics that are favorable (Table 1), suggesting that performance of surveillance mammography in older women is as good as mammography performed for general routine screening for women of all ages in the United States.

Table 1.

Comparison of mammography performance metrics in older surveillance population versus routine screening population.

Audit metric	Richman et al.12	BCSC13
Abnormal interpretation rate, %	11.5	11.6
Cancer detection rate, number per 1000 examinations	11.0	5.1
Positive predictive value, abnormal interpretations (PPV1), %	6.0	4.4
Positive predictive value, biopsy performed (PPV3), %	29.5	28.6

Abbreviation: BCSC = Breast Cancer Surveillance Consortium.

This may be because mammography performance is better in older women than younger women because of decreased breast density and increased cancer incidence with age, as has been previously demonstrated with data from the National Mammography Database.¹⁴ However, there is a known decrement in mammography performance in age-matched women with personal history of breast cancer, as sensitivity is decreased because of the appearance of posttreatment changes in the breast.^3,15 Richman and colleagues’¹² data suggest that, in the balance, surveillance mammography in older women performs similarly to routine screening mammography in all-comers.

The consensus among national screening guidelines is that weighing benefits and risks supports mammographic screening for women aged 40-74 years in the context of an individual’s life expectancy and comorbid conditions.¹⁶ As Richman et al.¹² demonstrate that mammographic performance characteristics are comparable, the question then becomes one of context for older women with personal history of breast cancer. A major contextual factor to consider is the relative contribution of overdiagnosis in this group. Overdiagnosis is the diagnosis of breast cancer that would not have been detected in a woman’s lifetime in the absence of screening. This is not directly observable, and published estimates vary widely, with those studies that account for a baseline increase in cancer incidence estimating 0%-10% overdiagnosis in the general screening population¹⁷ and others much higher based on differences in modeling assumptions.¹⁸ Given that there is a time lag to realize the benefits of screening,¹⁹ the potential for overdiagnosis is of relatively greater concern in older populations and breast cancer survivors because of increased comorbidity, competing causes of death, and the increased likelihood of diagnosis of relatively more indolent tumors in older women. Another contextual factor to consider is the recent trend toward de-escalation of breast cancer treatment in women aged older than 70 years, which may increase rates of recurrence and therefore the potential benefits of surveillance imaging, particularly in women with higher risk tumor subtypes.

Richman et al.¹² showed that a mean of 4.8 surveillance mammograms per woman were performed over the duration of follow-up, decreasing with increasing age at diagnosis and time since diagnosis. Other studies show that use of surveillance mammography decreases with age²⁰ and decreasing life expectancy, as one would hope, but in one study, 28.5% of women with personal history of breast cancer had a mammogram in their last year of life,²¹ whereas in other studies, 14% of patients with a life expectancy of more than 10 years did not undergo mammography,²² potentially forgoing opportunities to prevent advanced cancer (with its attendant morbidity and costs) and death.^11,23-28

To direct mammographic use to those most likely to benefit, the International Society of Geriatric Oncology²⁹ developed consensus guidelines for surveillance mammography in women aged 75 years and older, considering age, life expectancy, and breast cancer subtype and treatment. The guidelines recommend continuation when life expectancy is more than 10 years, discontinuation when life expectancy is less than 5 years, and individualized shared decision making for life expectancy between 5 and 10 years, which is a sensible approach. In practice, however, risk stratifying older women with personal history of breast cancer and determining when the risks of performing surveillance mammography outweigh the benefits remain a challenge for health-care providers.³⁰ Clinical breast cancer risk assessment models do not perform well in older women³¹ and do not apply to women who have already had breast cancer, and tools to accurately estimate life expectancy are limited. Additionally, life expectancy is difficult to discuss with patients, especially in primary care settings during brief visits.^29,32

Incongruously, the International Society of Geriatric Oncology multidisciplinary expert panel concluded that physical examination was an important part of survivorship care and should continue indefinitely “as a key way to show patients they are still being cared for,”²⁹ despite the fact that ongoing clinical breast examinations generate false positives and additional testing and are not known to improve outcomes in women with personal history of breast cancer or even in the screening setting in younger women.^29,33,34 This is an area that requires further investigation.

Patients’ values and personal preferences also play a role in determining how long to continue surveillance mammography. Although anxiety and financial concerns associated with mammography and subsequent testing are widely viewed as harms, Karam et al.²⁰ found that older breast cancer survivors expressed strongly positive opinions about surveillance mammography and a preference to continue on an annual schedule without related anxiety, pain, insomnia, or financial toxicity. A large majority (78%) of women reported that annual surveillance mammography gave them peace of mind, whereas 15% preferred having surveillance mammograms less frequently, and 2% reported that they would prefer to have mammograms only if they developed a breast problem. Women who experience false-positive mammograms are also more likely to engage in future screening, suggesting that this is not entirely perceived as a negative experience for patients.³⁵ Another factor to consider is that patients must desire to undergo treatment if cancer is detected to justify continued surveillance.

Mammography is a polarizing topic.³⁶ Yet most would likely agree that future research to clarify its role should address improvements in breast cancer risk prediction, alternative screening tools, optimal screening intervals and cessation, and lower morbidity treatments for patients with breast cancer, especially for older women. Developing future applications of artificial intelligence in mammography and risk stratification and using biomarkers such as circulating tumor DNA may help overcome some of our current challenges.^37-39

All surveillance imaging protocols involve certain risks and benefits. The overall goal is to find breast cancers at early stages to enable less invasive and more tolerable treatment that is effective and decreases mortality. Women, clinicians, and policy makers must weigh the evidence and consider whether and for how long to perform surveillance mammography for breast cancer survivors. Recommendations should be tailored to individual circumstances such as primary tumor characteristics, age, general health, life expectancy, future cancer risk, personal preferences for screening, and intention to seek and ability to tolerate treatment if cancer is detected.

Author contributions

Kimberly Feigin (Conceptualization, Writing—original draft, Writing—review & editing)

Funding

This research was supported in part by the National Cancer Institute at the National Institutes of Health Cancer Center Support Grant (grant number P30 CA008748).

Conflicts of interest

Kimberly Feigin has no disclosures to report.

Data availability

No new data were generated or analyzed for this editorial.