Breast cancer survivors who have residual breast tissue are recommended to undergo mammographic surveillance annually. ASCO, the American Cancer Society (ACS), and NCCN recommend annual mammograms beginning at 1 year after the mammogram that led to the cancer diagnosis, and at least 6 months after the completion of postlumpectomy radiation.1–3 It has been shown that mammographic detection of asymptomatic recurrences is associated with earlier disease stage and increased overall survival.4–7
In contrast to the consistent recommendation for mammography of residual breast tissue in this setting, ACS, ASCO, and NCCN state that there is insufficient evidence to recommend for or against MRI for routine breast cancer surveillance.1–3 According to guidelines, breast MRI should never be used instead of mammography, and only in addition to mammograms for those with a >20% lifetime breast cancer risk based on very strong family history, with a known cancer predisposition syndrome, or who had radiation therapy to the chest between ages 10 and 30 years. Evidence to support MRIs among breast cancer survivors is limited,8,9 but a recent case series study suggested that MRI may be more specific in breast cancer survivors than in women with only genetic risk or a strong family history.10
Real-world adherence to breast imaging guidelines among mixed-age women with breast cancer has been understudied in the United States. In women aged >65 years diagnosed with stage I–II breast cancer between 1992 and 1999, a SEER-Medicare analysis revealed that only 78% underwent mammography during months 7 to 18 after diagnosis, and only 57% had mammography yearly within 3 years.11 Patients who continued to see oncology specialists and who were younger, white (vs black), and living in certain regions were more likely to undergo mammography. Other studies have shown similar findings in older patients diagnosed more than a decade ago, generally with only 3 to 4 years of follow-up.12–17 A recent study using survey responses from 1,040 breast cancer survivors aged >65 years in the National Health Interview Survey found that 78.9% self-reported receipt of a mammogram in the prior 12 months, including only 86% of the 365 who had a life expectancy >10 years.18 We aimed to expand on this work by assessing rates and predictors of breast MRI and mammography in a modern cohort of mixed-age breast cancer survivors, and investigating how rates of imaging changed as time passed after diagnosis.
A retrospective analysis was conducted using the OptumLabs Data Warehouse, a large US database that includes administrative claims data from privately insured patients and Medicare Advantage enrollees across all 50 states and of all ages and ethnic and racial groups. Administrative claims are available on >100 million enrollees and include medical claims for professional (eg, physician), facility (eg, hospital), and outpatient pharmacy claims.19,20 The Mayo Clinic Institutional Review Board deemed this study exempt from review.
We identified all women with newly diagnosed non-metastatic breast cancer treated with breast surgery (lumpectomy or unilateral mastectomy) between January 1, 2005, and May 1, 2015, using previously validated claims-based algorithms.21,22 Patients were required to be age ≥18 years and have at least 12 months of continuous health plan coverage before their first breast cancer diagnosis and for 13 months following the definitive breast surgery (without diagnosis of metastatic breast cancer during that period). Those who received bilateral mastectomy within 18 months of their original diagnosis were excluded from this cohort (n=4,848). All types of systemic therapy (or lack thereof) were allowed.
For each patient, we assessed demographic and clinical characteristics at baseline diagnosis of breast cancer, including age, sex, race/ethnicity, geographic region (characterized as Northeast, Midwest, South, West, and other/unknown), local therapy type (lumpectomy followed by radiation, lumpectomy alone, mastectomy alone, mastectomy followed by radiation), receipt of chemotherapy, whether a primary care provider (PCP) or hematology/oncology visit occurred during the year of follow-up, and total number of medical comorbidities captured by ICD-9 codes on claims occurring within 12 months before breast cancer diagnosis. Comorbid conditions at baseline were identified using previously defined Elixhauser algorithms, which consider 32 specific conditions.23
The primary outcome was women having had at least 1 “diagnostic” or “screening” mammography claim (Current Procedural Terminology [CPT] code 77055, 77056, 77057, 76090, 76091, or 76092; or Healthcare Common Procedure Coding System [HCPCS]
Patient Characteristics During First Follow-Up Year (N=27,212)
We described the baseline characteristics of the cohort by imaging type (mammogram alone, mammogram and MRI, MRI alone, or no breast imaging)during the first 13-month period of follow-up after surgery. Characteristics were presented descriptively. We calculated the percentage of women who had at least one mammogram and/or MRI during each subsequent 13-month period of follow-up. We assessed the proportion who received each of the 4 imaging types during the first 13-month follow-up period for each year of breast surgery from January 2005–May 2015.
We estimated multinomial logistic regression models for the first year imaging tests for everyone and for the fifth year imaging tests for those who were continuously covered for at least 65 months after breast surgery (n=4,790). Via this method, we were able to evaluate each surveillance modality (mammogram, mammogram and MRI, and MRI only) compared with no surveillance. SAS 9.3 (SAS Institute Inc.) was used for all statistical analyses.
The cohort included 27,212 women followed for a median of 2.9 years (IQR, 1.8–4.6). Patient characteristics are displayed in Table 1. We found that the proportion of women excluded from this cohort due to bilateral mastectomies within 18 months after diagnosis increased from 25.9% in 2005 to 48.9% in 2014. A total of 4,790 patients remained part of the cohort through at least 65 months of follow-up.
After surgery from 2005–2015, mammography rates during the first year of follow-up remained relatively stable (Figure 1). Over the entire study period, 86% of women underwent mammography during year 1, 87% during year 2, 85% during year 3, 83% during year 4, and 80% during year 5 after the definitive breast surgery (Figure 2). Predictors of mammography use are shown in Table 2. The average number of mammograms per person per year was 1.04 (SD, 0.51), and the median was 1.00 (IQR, 0.75–1.35).
Over the years, the percentage of patients who underwent MRI in the first year after diagnosis decreased from 8% of those diagnosed in 2005 to 5% of those diagnosed in 2015, with a peak at 15% in 2007 (Figure 1). Over the entire study period, in the first year of follow-up, 2,428 (9%) underwent MRI; this proportion was 10% in year 2, 9% in year 3, 8% in year 4, and 7% in year 5 (Figure 2). Predictors of MRI use during the first year are shown in Table 2.
Longitudinal Surveillance After Treatment
Among the 4,790 patients who had complete follow-up for 5 years postsurgery, 50.2% had a mammogram all 5 years, 1.3% had both a mammogram and an MRI
Summary of Findings
Approximately 6 of every 7 breast cancer survivors underwent a mammogram in their first year of follow-up, regardless of their year of initial surgical treatment. This is higher than the <60% rate of annual mammography reported in women aged 50 to 64 years who had not previously been diagnosed with breast cancer.24 However, as women became long-term survivors, they were less likely to undergo mammography, even in the absence of any change in insurance status, and even in this relatively young cohort. This finding confirms and expands on previous studies showing a significant decline in mammography rates in the fourth or fifth year after treatment.12,13,15–17,25–27 Our study is novel because it assessed MRI use both in combination with mammography and as a standalone breast surveillance strategy. Given the paucity of data to support breast MRI use in cancer survivors without deleterious BRCA mutations,8,9 it is not surprising that most women with residual breast tissue in the OptumLabs database did not undergo annual MRIs to screen for local recurrences or new primary cancers.
Factors Associated With Surveillance Mammography
Determinants of mammography underuse in this population are poorly understood to date. Among 1,304 Italian patients with breast cancer, 80% had a mammogram and/or clinical breast examination during the first year after treatment, but this decreased to 67% at 10 years of follow-up.28 Like us, those investigators identified that patients who had undergone mastectomy and had more comorbidities were more likely to not undergo surveillance mammography. Unlike us, they found that older age was associated with a lower likelihood of mammography. This may reflect national differences in healthcare policies and patterns. We were unable to assess whether poorer finances and longer travel time to the hospital (which were significant predictors the Italian study) were also associated with lower odds of having mammography in the United States. The Italian study did not evaluate the relevance
Adjusteda ORs Compared With No Screening in the First Year of Follow-Up
Breast MRI Trends Over Time
Although most MRIs were performed in conjunction with mammographic surveillance, a small minority
Adjusteda ORs for Screening Methods Compared With No Screening for Women With 5 Years of Continuous Coverage (n=4,790) in the Fifth Year of Follow-Up
Age and Comorbidity
Younger patients may be more likely to have MRIs and less likely to have mammograms due to their denser breast tissue, which reduces sensitivity of mammography. A recent study showed that adding MRIs to mammograms improved sensitivity but reduced specificity in women <50 years of age after breast-conservation therapy.31 Going forward, the increasing availability of 3-dimensional mammography may reduce the use of MRIs in this setting. Our finding that ≥3 comorbidities reduced the chance of both mammography and MRI is not surprising given that the benefits of surveillance are likely greatest in women without a substantial competing risk of mortality from another disease. The lower likelihood of breast imaging during the first 13 months after surgery in those who underwent mastectomy and chemotherapy may be related to lasting toxicities that made it more difficult for patients to return for breast surveillance.
Our finding that black breast cancer survivors were less likely to undergo surveillance breast mammogram or MRI may be a contributor to the elevated rates of breast cancer mortality that have been identified in black women.32 Given that locoregional recurrences appear to be one of the major drivers of poor prognosis in black women,33,34 lack of imaging may be a substantial problem. Reasons for this are unclear, but socioeconomic status, comorbidities, lack of social support, and differences in healthcare beliefs may be relevant. Tammemagi35 found that comorbidities were present in 86% of black patients with breast cancer and only 65.7% of white patients (P<.001) at a large Detroit medical center, and therefore an inability to completely control for comorbidities using claims data may have heightened the racial disparities we identified. In our cohort, all patients were insured, but we were unable to control for socioeconomic status, which could affect access to breast imaging due to both transportation difficulties, particularly in rural areas, and out-of-pocket costs. A deleterious BRCA mutation in a survivor who decides not to undergo bilateral mastectomy is a strong indication for breast MRI; thus, it is possible that limited access to genetic testing also contributes to the racial disparities in MRI use seen here. However, genetic testing disparities should not impact mammography rates in this cohort.
Strengths and Weaknesses
Strengths of this study include its large size, regional and racial diversity, mixed-age population, and novel and important focus. Limitations include its claims-based methodology, precluding a full understanding of patient and tumor factors that may impact imaging choices. For example, reasons for greater MRI use in certain regions are unknown and will require further study. Because coding for “diagnostic” versus “screening” imaging is not always accurate, we are also unable to clearly discern how imaging rates may have been impacted by the Affordable Care Act mandate that screening, but not diagnostic mammograms, be fully covered. Additionally, our study may have been underpowered to detect certain disparities (eg, between Hispanic vs white patients). Our inability to include uninsured women prevents generalization to the entire US population. Moreover, changes over >3 years could only be assessed in the minority. In addition, we were unable to discern which of the mammograms/MRIs were performed as part of a routine surveillance strategy versus in response to a symptom, an abnormality detected during examination, or a finding on another imaging test (ie, some MRIs may have been used to workup an abnormality detected on a routine mammogram). During the first year, some imaging tests may have been performed postoperatively to assess for residual abnormalities rather than as surveillance for new abnormalities. Furthermore, some MRIs may have been performed due to a known deleterious BRCA mutation.
Implications for Healthcare Professionals and Patients
Use of MRI-based surveillance (which is not indicated for most survivors) is relatively rare, particularly in recent years. However, it is concerning that even in an insured cohort, many breast cancer survivors do not undergo their annual recommended surveillance mammography, especially as more time passes after breast cancer diagnosis. This may reflect that the data supporting annual surveillance mammography (rather than mammography every 18–24 months, for example) are sparse, perhaps leading some clinicians to recommend less frequent imaging for patient convenience.36 Still, our finding will be important for healthcare professionals who wish to encourage annual mammography in this setting. It implies that certain patients may need additional supports and encouragement to adhere to national guidelines. Universal implementation of survivorship care plans that include clear follow-up recommendations may be helpful.
Dr. Mougalian has disclosed that she has stock or other ownership interest with Gilead Services Inc. and Roche Holdings Ltd., has a consulting or advisory role with Eisai Pharmaceuticals and Hylapharm LLC, and has received research funding with NCCN/Pfizer. Dr. Lemaine has disclosed that she has stock or other ownership interest with Exact Sciences, has received honoraria from ACEL RX, has a consulting or advisory role with ACESE RX and Lifecell, has received research funding from Allergan, and has had travel expenses associated with activities listed herein paid by Bonti, ACEL RX, and Lifecell. Dr. Vachon has disclosed that she has a leadership role with and has received research funding from Grail Inc. The remaining authors have disclosed that they have no financial interests, arrangements, affiliations, or commercial interests with the manufacturers of any products discussed in this article or their competitors.
Dr. Ruddy and Ms. Sangaralingham were supported by an NCCN Young Investigator Award (PI: Ruddy).