Imaging Surveillance Options for Individuals With a Personal History of Breast Cancer: AJR Expert Panel Narrative Review

Abstract

Annual surveillance mammography is recommended for breast cancer survivors based on observational studies and meta-analyses showing reduced breast cancer mortality and improved quality of life. However, breast cancer survivors are at higher risk of subsequent breast cancer and have a 4-fold increased risk of interval breast cancers compared to individuals without a personal history of breast cancer. Supplemental surveillance modalities offer increased cancer detection compared to mammography alone, but utilization is variable, and benefits must be balanced with possible harms of false-positive findings. In this review, we describe the current state of mammographic surveillance, summarize evidence for supplemental surveillance in breast cancer survivors, and explore a risk-based approach to selecting surveillance imaging strategies. Further research identifying predictors associated with increased risk of interval second breast cancers and validated risk prediction tools may help physicians and patients weigh the benefits and harms of surveillance breast imaging and decide on a personalized approach to surveillance for improved breast cancer outcomes.

The goal of surveillance breast imaging in breast cancer survivors is the same as screening mammography in individuals without a breast cancer history—early detection of an in-breast event to reduce morbidity and mortality (Fig. 1). Annual surveillance mammography is recommended for survivors based on observational studies and meta-analyses showing reduced breast cancer mortality and improved quality of life [1–4]. However, surveillance mammography is an imperfect test. Compared to individuals without a breast cancer history, survivors have more than a 7-times higher risk of a second event (in-breast recurrence or new cancer in the same or opposite breast) [5, 6]. The annual risk of recurrence is highest in the first five years (10.4%), peaking in the second year after treatment (15.2%) [7]. Mammography is less sensitive in breast cancer survivors, who also have a 4-times higher interval breast cancer rate (breast cancers diagnosed within 12 months of a negative mammogram, which may be considered a failure of surveillance technology) compared to the overall screening population [8]. This challenge presents an opportunity to improve the approach to surveillance breast imaging by increasing surveillance benefits (e.g., early detection of a second breast cancer leading to reduced advanced disease, less intensive treatment, and increased survival) and decreasing surveillance failures and potential harms (e.g., false-positive recalls and false-positive biopsies). In this review, we discuss the current state of breast cancer surveillance and present a risk-based approach that may improve outcomes of breast cancer surveillance imaging. While we use the term “women” to describe individuals at risk for female breast cancer (i.e., malignant neoplasm of the female breast) when cited references use the term, we recognize that some of these individuals may self-identify as another gender.

Figure 1.

Current state of surveillance mammography and outcomes

Currently, variable approaches to breast cancer surveillance are applied in asymptomatic individuals treated with lumpectomy or unilateral mastectomy. All approaches include annual mammography with variable and opportunistic use of supplemental imaging, typically by breast MRI (the most commonly used supplemental imaging in this context) or breast ultrasound. Observed variation in supplemental MRI use is based as much on factors such as referring physician specialty, access to imaging centers, and social determinants of health as it is based on second breast cancer risk factors [9–11]. Studies suggest supplemental ultrasound also has a variable pattern of use [12–14]. Such variation in breast cancer screening or surveillance is likely not optimal for maximizing surveillance benefits while minimizing potential harms. Furthermore, while supplemental surveillance imaging may increase second breast cancer detection, it may also yield unequal benefits and harms across individuals with varying clinical, demographic, and social determinants of health compared to targeted risk-based use.

Current Surveillance Guidelines

A review of current breast cancer surveillance guidelines demonstrates consensus among national and international medical organizations with recommendations for annual surveillance mammography in individuals with a personal history of breast cancer (PHBC) and residual breast tissue (Table 1) [15–24]. After mammography (Fig. 2), the most recommended surveillance imaging modalities are breast MRI and whole-breast ultrasound, with MRI being preferred [15, 17–19, 23–26]. Societies recommending breast MRI generally endorse its supplemental use in individuals who are younger (e.g., <50 years old at diagnosis), have dense breasts at surveillance mammography, or are otherwise at increased risk for second breast cancers [15, 17, 19, 23]. The American College of Radiology (ACR) and Society of Breast Imaging (SBI) also recommend ultrasound as an alternative in patients for whom MRI is recommended but who cannot undergo MRI for reasons such as pregnancy or gadolinium contrast allergy [17]. The European Society for Medical Oncology notes that ultrasound surveillance can be considered in the follow-up of invasive lobular carcinomas [23]. While some societies have no recommendation for or against breast MRI or ultrasound surveillance, the National Institute for Health and Care Excellence in the United Kingdom recommends against routine use of either modality in the surveillance of breast cancer [22], although individuals otherwise at high risk for breast cancer follow separate protocols that include MRI [25]. National surveillance guidelines frequently have limited or no discussion of less common imaging techniques and modalities, such as abbreviated breast MRI, contrast-enhanced mammography (CEM), and molecular breast imaging (MBI), likely related to a paucity of rigorous studies in the surveillance setting. However, the ACR recommends against the use of MBI, even in high-risk women such as those with a PHBC [17].

Table 1.

Society Recommendations for Routine Surveillance Imaging after Breast Cancer Treatment

Society and Year	Mammographya	MRI	Ultrasound
American Cancer Society (ACS) and American Society of Clinical Oncology (ASCO), 2015 [15]	Annual mammography	Recommended against unless patient meets high-risk criteria as per ACS guidelines [100]	Not specified
National Comprehensive Cancer Network (NCCN), 2021 [16]	Annual diagnostic mammography	Not specified	Not specified
American College of Radiology (ACR), 2018 [17], 2019 [18]	Annual mammography, DM ± DBT	Annually for those with dense breast tissue or those diagnosed at age < 50	Consider for those who qualify for but cannot undergo breast MRI
American Society of Breast Surgeons, 2019 [19]	Annual mammography, DBT preferred	Supplemental imaging for those who have either dense breast tissue or were < age 50 at diagnosis when recommended by their physician; MRI preferred	Supplemental imaging for those who have either dense breast tissue or were < age 50 at diagnosis when recommended by their physician
Canadian Cancer Society [20]	Annual mammography	Not specified	Not specified
Canadian Association of Radiologists, 2016 [21]	Diagnostic mammography	Not specified	Not specified
National Institute for Health and Care Excellence (United Kingdom), 2018 [22, 25]	Annual mammography	Recommended against; except in women at high risk for breast cancer previously recommended for MRI screening	Recommended against
European Society for Medical Oncology (ESMO), 2019 [23]	Annual mammography	May be indicated for young patients, especially in patients with dense breast tissue and genetic or familial predispositions	Can be considered in the follow-up of lobular invasive carcinomas
Oncology societies of Korea (KSMO), China (CSCO), India (ISMPO,) Japan (JSMO), Malaysia (MOS), Singapore (SSO), and Taiwan (TOS), 2020 [24]	Annual mammography	When needed, based on ESMO recommendations	When needed, based on ESMO recommendations
Cancer Australia, 2020 [26]	Annual mammography	Consider in patients < 50 years old who are carriers of high-risk gene mutations and who have not undergone risk-reducing mastectomy (annually)	When indicated on clinical or radiological grounds

^aExcept for the ACR and CAR, all organizations specify bilateral mammography for individuals treated with breast conservation and unilateral mammography of the intact breast for individuals treated with unilateral mastectomy. ACS/ASCO, NCCN, NICE, and Cancer Australia guidelines state imaging of the reconstructed breast after mastectomy is not indicated or should not be performed.

CSCO = Chinese Society of Clinical Oncology, DM = Digital mammography, DBT= Digital breast tomosynthesis, ISMPO =lndian Society of Medical and Pediatric Oncology, JSMO = Japanese Society of Medical Oncology, KSMO = Korean Society of Medical Oncology, MOS = Malaysian Oncological Society, SSO = Singapore Society of Oncology, TOS = Taiwan Oncology Society

Figure 2.

Second breast cancer detected by surveillance mammography. 54-year-old woman with a history of left breast ductal carcinoma in situ (DCIS) treated with lumpectomy and whole breast radiation. (A) Mediolateral oblique (left) and craniocaudal (right) views at surveillance mammography 8 years after treatment show new retroareolar microcalcifications. (B) Magnified mediolateral (left) and craniocaudal (right) views confirm suspicious microcalcifications. Pathology from stereotactic guided biopsy showed multiple foci of microinvasive carcinoma in a background of extensive high-grade DCIS (ER+, PR-, HER2-).

Surveillance Mammography Guidelines

Most societies do not comment on a preferred mammographic technology [e.g., screen-film, digital mammography, or digital breast tomosynthesis (DBT)]. Of those that do, the ACR and SBI recommend digital mammography with or without DBT, and the American Society of Breast Surgeons (ASBrS) specifically endorses DBT as the preferred mammographic technology [17, 19]. In current clinical practice in the United States, screen-film mammography has been almost entirely replaced by digital techniques, and since its U.S. FDA approval in 2011, use of DBT has continued to increase. In a study of insurance claims from a national private insurer, 70% of screening mammograms were billed with DBT in 2019 [27]. In a study of 31,906 mammograms in 8170 breast cancer survivors, DBT led to fewer false-positives and higher specificity without a significant decrease in cancer detection rate compared to digital mammography [28].

While guidelines are consistent in specifying frequency of surveillance mammography at annual intervals, there is less guidance on the optimal time to initiate and discontinue routine surveillance imaging, due to limited evidence. Table 2 summarizes organization-specific guidelines on surveillance imaging initiation, discontinuation, and frequency. Some organizations (National Cancer Comprehensive Network, Canadian Cancer Society, and Cancer Australia) provide guidance on initiating surveillance mammography, typically advising that it be performed at least 6 months after completion of radiation therapy based on expert consensus [16, 20, 26]. However, most societies with a recommendation for surveillance mammography do not provide guidance on when to discontinue mammography in individuals with a PHBC.

Table 2.

Summary of Guidelines on Initiation, Discontinuation, and Frequency of Surveillance Mammography

Society and Year	Initiation	Discontinuation	Frequency
American Cancer Society (ACS) and American Society of Clinical Oncology (ASCO), 2015 [15]	Not specified	Not specified	Annual
National Comprehensive Cancer Network (NCCN), 2021 [16]	6–12 months after completion of radiation therapy	Not specified	Annual
American College of Radiology (ACR), 2018 [17], 2019 [18],	Not specified	Not specified	Annual
American Society of Breast Surgeons, 2019 [19]	Not specified	Not specified	Annual
Canadian Cancer Society [20]	6 months after completion of treatment	Not specified	Annual
Canadian Association of Radiologists, 2016 [21]	Not specified	Not specified	Not specified
National Institute for Health and Care Excellence (United Kingdom), 2018 [22]	Not specified	Surveillance until age 50 or for 5 years, then return to population screening (non-high-risk women)	Annual for 5 years or until age 50, then return to every 3 years
European Society for Medical Oncology (ESMO), 2019 [23]	Not specified	Not specified	Annual
Oncology societies of Korea (KSMO), China (CSCO), India (ISMPO), Japan (JSMO), Malaysia (MOS), Singapore (SSO), and Taiwan (TOS), 2020 [24]	Not specified	Not specified	Annual
Cancer Australia, 2020 [26]	12 months after diagnosis	Not specified	Annual
International Society of Geriatric Oncology (SIOG), 2021 [29]	Not specified	-When life expectancy is < 5 years, including those with a history of high-risk cancers -Consider when life expectancy is 5 to 10 years	Annual or biennial (based on cancer risk and age)

CSCO = Chinese Society of Clinical Oncology, ISMPO =1 ndian Society of Medical and Pediatric Oncology, JSMO = Japanese Society of Medical Oncology, KSMO = Korean Society of Medical Oncology, MOS = Malaysian Oncological Society, SSO = Singapore Society of Oncology, TOS = Taiwan Oncology Society

The International Society of Geriatric Oncology (SIGO) sought to address this issue in a consensus statement, recommending discontinuation of surveillance when life expectancy is less than 5 years, even for patients whose primary breast cancer had high-risk features [29]. The SIGO consensus statement is based on breast cancer risk in older women with a PHBC and studies on the benefits and harms in women without a history of breast cancer. This recommendation is similar to ACR and American Cancer Society screening guidelines for average-risk women without a PHBC, which recommend consideration of general health and life expectancy in determining the utility of ongoing screening mammography [30, 31]. In practice, discussion of remaining life expectancy with older patients is often a barrier to implementing this recommendation. Efforts to address this in the general screening population include conversation scripts and life expectancy estimations given to providers and decision aids given to patients. Such efforts have resulted in positive perceptions of such conversations, increased shared decision-making about ending screening, and fewer screening mammograms in older patients [32, 33].

Currently, the ACR/SBI and the Canadian Association of Radiologists support using a screening or diagnostic indication in mammographic evaluation of women with a PHBC, with variability in current clinical practice reflecting these recommendations. In a study of women with a PHBC undergoing routine mammography, over 30% of examinations had a non-screening indication [34] and, in a national survey, 73% of practices recommend diagnostic mammography following breast-conserving treatment prior to returning to screening [35]. A diagnostic evaluation offers the benefits of real-time interpretation, same-day evaluation of abnormalities, and communication of results directly to the patient. However, routine use of a diagnostic indication for individuals with a PHBC has potential downsides. Patients may face higher costs when undergoing diagnostic mammography because the 2010 Patient Protection and Affordable Care Act eliminated out-of-pocket costs for screening mammography but not diagnostic examinations [36]. Additionally, standard audits of screening mammography do not include examinations with diagnostic indications. The observed variability in mammography indication has implications for observational studies aiming to produce generalizable results for individuals with a PHBC (e.g., symptomatic individuals undergoing diagnostic examination vs asymptomatic individuals undergoing diagnostic examinations based on facility or ordering provider protocols) [34, 35, 37]. One approach to reducing indication variability is implementation of screening with real-time interpretation and same-day diagnostic imaging, if needed. This approach would increase the use of screening indications for asymptomatic individuals and reduce out-of-pocket costs, while preserving the ability to perform same-day diagnostic workups. A possible disadvantage to this approach is a higher recall rate compared with batch reading, although evidence remains limited [38]. However, this approach has been reported as a feasible approach to mammographic screening, with similar expected impact if applied to surveillance [39, 40].

Adherence to Surveillance Mammography Guidelines

Studies have also shown that surveillance mammography is underused. Adherence to annual mammography is highest in the first year following breast cancer treatment, ranging from 63–86% [41–44], with adherence rates decreasing by time after initial treatment. Wirtz et al. showed a decline in adherence over the first 6 years of follow-up, plateauing at 63–66% in years 7–10 [42]. Studies evaluating adherence to surveillance also highlight disparities in its use. Age less than 40 years at diagnosis, Black race, Hispanic ethnicity, noncommercial insurance coverage, lower neighborhood median income, and lower attained education levels all predict lower adherence to annual surveillance mammography [42, 45].

Although much of the work on interventions to improve uptake of routine mammography has focused on screening populations, these same tools may be effective in a surveillance population and warrant further investigation. For example, telephone calls and reminder letters from patient navigators increase screening mammography use among individuals from historically marginalized populations (i.e., racial and ethnic minority groups, low-income women, and inner-city residents) [46]. In another study, patient engagement with an interactive computer program followed by ongoing support from a lay health advisor available to help address barriers to mammography was associated with increased uptake of mammography among low-income Black women [47]. These interventions suggest that improving adherence to surveillance imaging may be another worthwhile approach to improving cancer outcomes among survivors, especially for those at risk of health disparities.

Current Limitations of Surveillance Mammography

Regardless of its clear benefits, broad support by society recommendations, and widespread use, the diagnostic accuracy of annual mammography in individuals with a PHBC is lower than in those without this history. Mammography has lower sensitivity (69.9% vs 86.9%) and higher interval cancer rates (3.4 vs 0.8 per 1000 examinations) in breast cancer survivors undergoing surveillance compared to screening benchmarks (95% of examinations from individuals without a history of breast cancer) [8]. Breast density is an important mediator of surveillance mammography performance, with sensitivity lowest and interval cancer rates highest among individuals with extremely dense breasts. In a report of screening mammography in average-risk women using both digital mammography and DBT, women with extremely dense breasts of any age saw no increase in cancer detection rate with DBT [48]. These results reflect the limitations of mammography, regardless of PHBC. Poor prognostic markers associated with interval breast cancers, including larger size, negative hormone receptor status, and positive lymph node status [8], highlight interval breast cancer as a clinically significant adverse outcome of surveillance mammography.

While guidelines consistently recommend annual surveillance intervals, one approach to reducing interval cancers includes more frequent surveillance, specifically mammography at semi-annual intervals. This approach is common in clinical practice with Patel et al. reporting that 73% of surveyed facilities performed diagnostic mammography every 6 months following lumpectomy for 1–5 years, rather than annually as supported by most guidelines [37]. Similarly, in a study of 19,955 women from facilities in the Breast Cancer Surveillance Consortium, one-third of those treated with breast conservation had more than one surveillance imaging examination in the first year following diagnosis [49]. Evaluation of more frequent imaging in women with a PHBC treated with lumpectomy has produced mixed results, with some authors concluding imaging at semi-annual intervals does not increase cancer yield and others finding that recurrences identified at semi-annual (vs annual) intervals were significantly less advanced [50–53]. Further research on optimal imaging frequency and duration of the diagnostic period would help create effective standardized surveillance protocols, as in the ongoing Mammo-50 trial assessing less frequent mammography [54].

Artificial intelligence (AI) technologies may also be explored to improve the future performance of surveillance mammography. While AI has shown promise in improving interpretive performance in the setting of screening mammography [55], a paucity of work has investigated its use in women who have undergone breast conserving surgery [56]. Until new AI tools are validated in this unique population, the relative performance deficiency of annual mammography may be better addressed through supplemental surveillance.

Supplemental Surveillance Imaging Options

Given the relatively lower performance of mammography and the generally higher risk of a second cancer in individuals with a PHBC, supplemental surveillance imaging may play an important role in reducing symptomatic interval cancers in this population. In addition to the more commonly recommended supplemental modalities (i.e., MRI and ultrasound), abbreviated breast MRI, CEM, and MBI with positron emission mammography (PEM) or breast-specific gamma imaging (BSGI) are additional modalities available for supplemental surveillance. Until studies of adequate sample size and rigor are available to support inclusion in surveillance imaging guidelines, selection of individuals for surveillance with these modalities should consider the likelihood of improved cancer detection rates and reduction of symptomatic breast cancers while also minimizing the impact of false-positive results.

MRI

National guidelines’ lack of consensus on the use of surveillance breast MRI reflects in part the available evidence on the performance and outcomes of this modality (Fig. 3). Studies evaluating breast MRI have yielded variable results and have often been limited to single-site academic institutions, reducing generalizability. In a study of 13,266 women undergoing 33,938 surveillance mammograms and 2506 breast MRI examinations within the Breast Cancer Surveillance Consortium, surveillance MRI increased cancer detection rates in women with a PHBC, with no statistically significant difference in sensitivity between mammography and MRI [57]. However, in a single-institution study of 607 women who underwent 932 MRI examinations, the sensitivity of breast MRI in women with PHBC was 91.7%, and MRI detected an additional 18.1 cancers per 1000 women who had a negative mammogram and ultrasound [58]. In the same study, most cancers were detected in women younger than 50 years old. Similarly, a multicenter prospective study of 762 women found that the addition of MRI to mammography increased surveillance sensitivity and the cancer detection rate in women diagnosed with breast cancer at age 50 years or younger [59]. Additional evidence can be gleaned from studies evaluating supplemental breast MRI in women with dense breasts, with data supporting its utility in women with dense breasts and a PHBC [60].

Figure 3.

Second breast cancer detected by surveillance MRI. 53-year-old woman with a history of right breast invasive ductal carcinoma, ER+, PR+, and HER2-, treated with partial mastectomy, chemotherapy, radiation, and 5 years of tamoxifen. (A) Mediolateral oblique (left) and craniocaudal (right) views at surveillance mammography 10 years after treatment were given a benign assessment. (B) Axial and (C) sagittal contrast-enhanced T1-weighted MR images from a surveillance MRI performed seven months later show a 4 mm irregular enhancing right breast mass with rapid initial and delayed washout kinetic features (arrow). (D) Radial and (E) antiradial views show a sonographic correlate (arrow). (F) Mediolateral oblique (left) and craniocaudal (right) views from a post-biopsy mammogram demonstrate the biopsy marker clips appropriately positioned (arrow). Pathology showed new primary breast cancer (invasive lobular carcinoma; ER+, PR+, HER2-).

Given the promising performance of breast MRI in individuals with a PHBC, other studies have sought to determine the optimal frequency of MRI in the setting of a PHBC. Park et al. found that in women with a PHBC and negative initial MRI, 90% of second breast cancers were detected with MRI after a 2-year surveillance interval [61], and DeBruhl et al. found that no contralateral breast cancers were detected on MRI performed during a second year of MRI surveillance [62]. Together, these studies suggest that MRI surveillance could be performed at 2-year intervals rather than annually. Although the addition of MRI to mammography has been shown to increase surveillance sensitivity and cancer detection rate, barriers to its widespread implementation must be considered. Patients’ examination tolerance (e.g., claustrophobia and positional discomfort), need for IV gadolinium-based contrast agents, false-positives leading to additional imaging and biopsies (Fig. 4), and high costs associated with breast MRI are important factors to consider [63, 64].

Figure 4.

False-positive finding on surveillance MRI. 36-year-old woman with a history of right breast invasive ductal carcinoma with ductal carcinoma in situ (ER+, PR+, HER2 -), treated with mastectomy and endocrine therapy and who presented for high-risk screening MRI 4 months after surgery. (A) Axial and (B) sagittal contrast-enhanced T1-weighted MR images of the left breast show non-mass enhancement at 1 o’clock, measuring 11 mm and having rapid initial kinetics. The finding was biopsied using MRI guidance (images not shown). (C) Mediolateral oblique (left) and craniocaudal (right) views from post-biopsy mammogram show clip in the expected location (arrow). Pathology results revealed benign stromal fibrosis and focal apocrine metaplasia, concordant with the imaging findings.

Ultrasound

Sonographic surveillance imaging in individuals with a PHBC may offer an opportunity to detect mammographically occult cancers without the drawbacks of MRI (Fig. 5). Whole-breast ultrasound (WBUS), either handheld or automated, has been explored most extensively for supplemental screening of women at high risk of cancer or with dense breasts [65–67]. In 754 women with a PHBC, Cho et al. found that the cancer detection rate of mammography with handheld WBUS was significantly higher than that of mammography alone (6.8 vs 4.4 per 1000 women), but the increased sensitivity of supplemental ultrasound was not statistically significant [59]. Given the heterogeneity of the studies evaluating supplemental WBUS, the performance of ultrasound in individuals with a PHBC may not be comparable to those without a PHBC. Kim et al. showed performance limitations of mammography with supplemental handheld ultrasound in a matched cohort study and concluded this surveillance strategy in women with a PHBC had lower sensitivity (43% vs 92%) and higher rate of interval cancers (2.5 vs 0.3 per 1000) compared to women without a PHBC [68]. However, in the largest U.S. trial of supplemental ultrasound in addition to mammography for 2662 women with elevated risk of breast cancer and dense breasts (American College of Radiology Imaging Network 6666), the supplemental yield of ultrasound and absolute increase in sensitivity compared to mammography alone was the same in women with a PHBC as in women without a PHBC (i.e., 4.2 vs 4.3 per 1000, respectively, and absolute increase in sensitivity of 28.8% for both groups) [13]. In a study of 15,318 women with dense breasts (3.6% with a PHBC), 3D automated WBUS detected an additional 1.9 cancers per 1000 women screened [69]. Nevertheless, the benefits of supplemental ultrasound are offset by increased recall rates and increased number of biopsies [13, 59, 60]. Further investigation is needed to better assess the benefit of ultrasound in individuals with PHBC.

Figure 5.

Second breast cancer detected by surveillance ultrasound. 42-year-old woman with a history of left breast invasive ductal carcinoma (IDC) (ER +, PR+, and HER2-), treated with lumpectomy and radiation. (A) Mediolateral oblique (left) and craniocaudal (right) views from baseline posttreatment mammogram were given a benign assessment. (B) Antiradial image from a surveillance ultrasound performed 6 months later revealed an irregular mass adjacent to the lumpectomy bed measuring 9 × 7 × 6 mm. (C) Mediolateral (left) and craniocaudal (right) views from a post-biopsy mammogram show the biopsy marker clip adjacent to the lumpectomy bed. Pathology was recurrent IDC (grade 2; ER +, PR+, HER2 equivocal).

Abbreviated Breast MRI

Given the long acquisition and interpretation times, high cost, and limited availability of full diagnostic breast MRI, abbreviated breast MRI protocols have been introduced for screening in women with dense breasts or at high risk of breast cancer [70, 71]. As with other supplemental imaging modalities, most studies evaluating abbreviated breast MRI have primarily focused on women with dense breasts [60, 72]. In general, abbreviated breast MRI has better performance compared to mammography and ultrasound and is comparable to a full MRI protocol. For example, the ECOG-ACRIN EA1141 study compared the performance of abbreviated breast MRI to DBT in a sample of 1444 average-risk women with dense breasts and showed that abbreviated breast MRI detected an additional 7 invasive cancers per 1000 women (11.8 vs 4.8 per 1000 women) and had a much higher sensitivity (95.7% vs 39.1%) [73]. Improved cancer detection rate and sensitivity of abbreviated breast MRI were obtained at the expense of lower specificity (86.7% vs 97.4%) [73]. While women with a PHBC were eligible, only 8 of the 1444 women included in the EA1141 study had a PHBC and a subgroup analysis was not performed. In a retrospective study assessing the performance of abbreviated breast MRI in 725 women with PHBC, the sensitivity of abbreviated breast MRI was 100%, and 7 of 12 second cancers were occult on mammogram and ultrasound [74]. In a retrospective study directly comparing abbreviated breast MRI with mammography and ultrasound in 710 women with PHBC, Baek et al. found that the AUC and sensitivity were highest for abbreviated breast MRI, and the cancer detection rate was 11.7 for abbreviated breast MRI compared to 3.2 for mammography and 4.3 for ultrasound [75]. In a study of 1312 surveillance MRI examinations performed in 1045 women with a PHBC, Park et al. showed that abbreviated breast MRI had better specificity, PPV, and accuracy but lower sensitivity compared to a full diagnostic MRI protocol, although these differences were not statistically significant [76]. While these results are promising, additional multi-institutional studies with larger sample sizes are likely necessary before insurance coverage and widespread adoption of abbreviated breast MRI as a supplemental surveillance tool can be attained.

Contrast Enhanced Mammography

As with MRI, the basis of cancer detection for contrast-enhanced mammography is visualization of tumor angiogenesis [77]. This examination uses dual-energy technique after administering IV iodinated contrast material to produce low-energy, high-energy, and recombined images of each breast in the standard craniocaudal and mediolateral oblique projections [78]. The low-energy image is comparable to a conventional full field-digital mammogram, and the recombined imaging results in an iodine-only image, highlighting suspicious enhancing abnormalities (Fig. 6). In a single-institution retrospective study of 904 women at increased risk of breast cancer (approximately 40% having a PHBC), the sensitivity and specificity of CEM was 87.5% and 93.7% compared to 50% and 97.1% for the low-energy images [79]. In another retrospective cohort study of 611 women at intermediate risk for breast cancer and dense breasts, the sensitivity and cancer detection rate of CEM were higher than DM (90.5% and 31.1 per 1000 vs 52.4% and 18 per 1000, respectively) but specificity decreased to 76.1% compared to 90.5% [80]. Although CEM is not a dynamic examination, it provides additional physiologic information and may prove to be a useful alternative to MRI, particularly in individuals who have contraindications to breast MRI. Limitations of CEM include need for supervision of contrast material administration, higher contrast reaction rate compared to gadolinium-based contrast, and current lack of specific reimbursement for CEM [78]. An additional barrier to implementing CEM in clinical practice is the lack of standardized reporting lexicon for CEM in the BI-RADS Atlas, although this is currently in development [81].

Figure 6.

Second breast cancer detected by contrast-enhanced mammography (CEM). 70-year-old woman with a history of left breast invasive ductal carcinoma (IDC) (ER+, PR+, HER2-), treated with lumpectomy, radiation, and 7 years of endocrine therapy. (A) Mediolateral oblique (left) and craniocaudal (right) recombined images from surveillance CEM 11 years after treatment show non-mass enhancement in the upper outer quadrant of the left breast (arrow) (low-energy images not shown). Pathology from MRI-guided biopsy showed in-breast recurrence (IDC and ductal carcinoma in situ; ER+, PR+, HER2-).

Molecular Breast Imaging

Molecular breast imaging, particularly breast-specific gamma imaging, is a recent modality of interest as technology improvements have led to increased spatial resolution and ability to reduce radiotracer dose [82]. Following the IV administration of Tc99m-sestamibi, a dedicated breast gamma camera is used to image radiotracer uptake in tissues with increased blood flow and metabolic activity [82]. BSGI studies have focused on women with dense breasts because increased breast density does not decrease diagnostic performance [82], with no studies focusing specifically on surveillance imaging in women with a PHBC [83]. In two studies of women with dense breasts (8% and 10% of the cohorts had a PHBC), BSGI resulted in an incremental cancer detection rate of up to 8.8 per 1000 women screened compared with mammography alone [83, 84]. Positron emission mammography has also shown promise in diagnostic imaging; however, the relatively high whole-body radiation dose limits its use in either screening or surveillance settings [82].

Risk-Based stratification

Although supplemental surveillance imaging may help to increase cancer detection and reduce interval cancers, consensus is lacking for selecting individuals who would most benefit [85, 86]. One approach to address this issue might include identifying individuals at increased risk of an interval breast cancer who would most benefit from annual mammography with supplemental surveillance. Risk-stratification tools combined with patient preferences for specific screening strategies as well as associated benefits and harms could guide shared-decision making in the selection of a surveillance protocol. Such a risk stratification tool could be based on information known at the time of diagnosis and treatment (e.g., individual-level characteristics, biological and imaging prognostic markers of the treated cancer, and initial treatment received) and include data available during follow-up (e.g., time since diagnosis and breast density on surveillance mammography).

Currently available risk models—Contralateral Breast Cancer Risk (CBCRisk), Predict Contralateral Breast Cancer (PredictCBC), and INFLUENCE [87–89]—predict the risk of second breast cancer. While a few predictors are used in all models (e.g., age at diagnosis, estrogen receptor status, and endocrine therapy), history of a high-risk lesion and BRCA mutation status are unique to CBCRisk and PredictCBC, respectively. However, none evaluate risk of surveillance outcomes such as interval cancer, which combines both breast cancer risk and the diagnostic performance of mammography at the individual level. Clinical factors that can predict interval breast cancer risk have been identified (e.g., lumpectomy without radiation, dense breasts, and interval primary breast cancer) [5, 90–92], and a prediction model of interval second breast cancer risk is being developed [93].

In current clinical practice, this risk-based approach to supplemental surveillance in individual with a PHBC is implicitly supported by guidelines from ACR/SBI and the ASBrS, which are based on expert consensus and recommend supplemental MRI surveillance of those who were less than 50 years old at the time of primary breast cancer diagnosis or who have dense breasts at surveillance [17–19, 59, 90, 92]. In a new analysis of data from a previously published cohort study of women from the Breast Cancer Surveillance Consortium [8], 62% of women with a PHBC had at least one of these two characteristics and would be eligible for supplemental MRI surveillance based on ACR/SBI and ASBrS recommendations (Fig. 7).

Figure 7.

Breast cancer survivors eligible for supplemental surveillance with MRI. In a cohort of 30,954 women receiving surveillance mammography in U.S. community practice [8], a total of 61.8% (those in the three cells with darker shading and thicker border) would be eligible for supplemental surveillance MRI based on American College of Radiology criteria of age less than 50 years at first breast cancer diagnosis or dense breasts on surveillance mammography.

Knowledge Gaps and Barriers to Risk-Based Surveillance

Further development of accurate risk prediction models validated in diverse cohorts of breast cancer survivors and risk calculators may prove to be valuable for supporting increased adoption and implementation of risk-based imaging surveillance for individuals with a PHBC. In developing a risk calculator for interval second breast cancers, identifying risk factors beyond the commonly included clinical risk factors may prove to be beneficial. Social determinants of health including income, education, neighborhood, and structural racism are associated with higher rates of second breast cancers and mortality independent of differences in breast cancer treatment [94]. Research to date has not focused on whether these factors might be associated with interval second breast cancer, nor on whether inclusion of these factors might improve the accuracy of risk prediction. Risk model developments will also likely incorporate AI. While to our knowledge no current AI algorithm for predicting risk in a surveillance population is available, AI has been investigated in a general screening population. For example, a deep learning algorithm relying on a combination of clinical data and mammography images was shown to better predict breast cancer risk in a general screening population compared to the widely used Tyrer-Cuzick model [95]. Additionally, a model combining an AI algorithm and breast density assessment predicted 50.9% of interval breast cancers in a Dutch general screening population [96].

Adopting risk-based surveillance also requires buy-in from patients and referring physicians [97]. A survey of primary care physicians, oncologists, surgeons, and radiologists about surveillance imaging and supplemental breast MRI found adherence to guideline recommendations for mammography. In the current absence of consistent guidelines for supplemental surveillance, physicians consider a range of patient risk factors, supporting acceptance of a risk-based approach to surveillance [98]. A survey of women’s surveillance preferences indicated high trust in physicians, fears of cancer recurrence, and distrust of surveillance imaging, particularly if mammography missed the first cancer [99]. Further research is needed to identify desired attributes of a risk model for shared decision-making and other facilitators and barriers to clinical adoption.

Consensus Statements.

• Surveillance mammography in individuals with a PHBC is associated with reduced mortality, and medical societies consistently recommend annual intervals. Adherence to surveillance mammography decreases with time from diagnosis.

• Individuals with a PHBC are at increased risk of second breast cancers, particularly interval cancers, which offers an opportunity to improve early second breast cancer detection with more intensive multimodality surveillance. Currently observed patterns of multimodality surveillance use vary substantially and reflect variable endorsement by clinical guidelines.

• While breast MRI has higher sensitivity and improves cancer detection in individuals with a PHBC compared to mammography, these benefits must be balanced with harms of increased recall rate and additional biopsies. The ACR/SBI and the ASBrS recommend breast MRI in women with a PHBC who were diagnosed before age 50 years or who have dense breasts.

• Whole-breast ultrasound increases cancer detection rate at the expense of increased recalls. The ACR/SBI and ASBrS recommend its use as an alternative to breast MRI in individuals who cannot undergo MRI.

• Abbreviated breast MRI and CEM have demonstrated promising results, but further validation in individuals with PHBC is needed to support clinical adoption.

• Selecting individuals at increased risk of interval cancers with mammography alone using a risk prediction model may support more consistent multimodality surveillance imaging use and improved outcomes. Validation in diverse populations is critical for successful adoption and implementation.