Assessing the impact of frequent mammography after breast cancer treatment from a single-institution retrospective analysis

Introduction

Breast cancer screening leads to earlier diagnoses, as evidenced by the more than doubling in early-stage breast cancer incidence within three decades after the incorporation of routine mammography in women over 40 years old. Moreover, the incidence of late-stage breast cancer decreased by 8% with mammography.¹ However, mammography is also associated with a false-positive rate that leads to additional imaging and biopsies.² In one recent study of nearly 450,000 women, half of the women that completed 10 years of annual screening mammography had at least one false-positive result and 11% had one subsequent false-positive biopsy,³ so there is some physical, psychological and financial harm. This risk-benefit tension is best illustrated in women with mammograms reported as BI-RADS 4, for whom biopsy is almost always recommended, although the positive predictive value is only 21.1%.⁴

In the United States alone, there are over 3.8 million breast cancer survivors, and that number continues to grow.⁵ These survivors are at risk for both local and distant recurrence regardless of surgical treatment (lumpectomy or mastectomy). A meta-analysis concluded that earlier detection of recurrence confers a survival advantage.⁶ Consequently, ‘regular’ mammography is recommended for surveillance after breast conserving therapy (BCT). However, the definition of ‘regular’ is subject to some debate, leading to wide variation in the timing of initiation and in the frequency of surveillance mammography.⁷ Consequently, the American Society of Clinical Oncology (ASCO) and the National Comprehensive Cancer Network (NCCN) published breast cancer surveillance guidelines recommending that, “Women treated with breast-conserving therapy should have their first post-treatment mammogram no earlier than 6 months after definitive radiation therapy. Subsequent mammograms should be obtained every 6 to 12 months for surveillance of abnormalities. Mammography should be performed yearly if stability of mammographic findings is achieved after completion of locoregional therapy” (National Comprehensive Cancer Network, NCCN clinical practice guidelines in oncology: Breast cancer. V.3.2020, 2020).⁸ The guidelines attempted, in part, to limit the use of semi-annual mammography after BCT, except when mammographic abnormalities are detected; however, they have not been universally accepted. The American College of Radiology has not taken a position on this policy (although their website posted eight other position statements related to mammography in April 2022). Consequently, numerous institutions continue to recommend and perform semi-annual mammography for two or more years after BCT. In this single institution retrospective study, we investigate the impact of biannual mammography after BCT on breast cancer specific outcomes.

Methods

After approval from the Cedars-Sinai Institutions Review Board (IRB # Pro00057198; April 10, 2019), a retrospective analysis was performed from May 1, 2019, using an institutional database at Cedars-Sinai Medical Center. Patient consent was deemed not necessary and was waived by the Cedars-Sinai Institutions Review Board. Female patients with stage 0-III breast cancer, diagnosed between January 1, 2015, and December 31, 2017 and treated with BCT were included. Patients had to meet the following inclusion criteria: (i) had a new diagnosis of invasive or in situ breast cancer (excluding classic lobular in situ); (ii) completed breast conserving surgery; (iii) were then advised to get screening mammograms every 6 months for two years following surgery; and (iv) had at least one mammogram within 0-12 months, and at least one mammogram within months 13-24 following surgery. Patients with evidence of distant metastases at diagnosis, incomplete records, or a prior history of invasive breast cancer were excluded. Of the 1,056 identified patients, 359 met inclusion criteria.

Clinical data was collected including mammography BI-RADS scores, breast biopsy results, recurrence data, and imaging that led to biopsies. Electronic medical records were independently audited by one of the senior authors (S.K.) to verify that no recurrences were missed.

Results

We identified 359 eligible women. The median age was 65 years (range 25-88 years). A total of 339 patients (94.4%) had unilateral breast cancer and 20 patients (5.6%) had bilateral breast cancer. Breast cancer stage ranged from 0 to IIIC, and multiple histologic subtypes were included. Neoadjuvant chemotherapy was administered in 28 patients (7.8%), adjuvant chemotherapy was administered in 39 patients (10.9%), adjuvant radiation was administered in 305 patients (85.0%), and adjuvant endocrine therapy was administered in 259 patients (72.1%) (Table 1).

During the 24 months of follow-up, 47 biopsies were performed on 46 patients (12.8%). 31 of these were performed on the ipsilateral breast and 16 were performed on the contralateral breast. Pathology was benign in 35 of the biopsies (74.5%) and malignant in 12 of the biopsies (25.5%). The imaging studies leading to these biopsies were a combination of surveillance mammography, magnetic resonance imaging (MRI), or diagnostic imaging based on clinical findings (Table 2).

Mammograms were obtained 6 months after BCT in 289 patients (80.5%), with a median time to completion of 6.7 months. In total, 11 (3.8%) of these mammograms led to biopsies, seven (2.4%) of which were in the ipsilateral breast. One detected recurrent invasive ductal carcinoma; the other 10 were benign.

Mammograms were obtained 12 months after BCT in 309 patients (86.1%). Overall, 12 (3.9%) of these mammograms led to biopsies, eight (2.6%) of which were in the ipsilateral breast. This led to the identification of three recurrences (invasive ductal carcinoma in one patient and ductal carcinoma in situ (DCIS) in two) and one new primary (DCIS); seven biopsies were benign.

Mammograms were obtained 18 months after BCT in 282 patients (78.6%), with a median time to completion of 18.6 months. A total of nine (3.2%) of these mammograms led to biopsies, four (1.4%) of which were in the ipsilateral breast. One detected a new focus of DCIS in the contralateral breast; the other eight were benign (Table 3). In these nine patients, the prior mammograms (done at 6 and 12 months) showed no significant abnormalities. The only findings were benign-appearing calcifications (in two patients), a simple cyst a (in one patient) and scarring at the prior surgical site (in one patient).

Mammograms were obtained 24 months after BCT in 275 patients (76.6%). A total of six (2.2%) of these mammograms led to biopsies, four (1.5%) of which were in the ipsilateral breast. Two showed recurrent DCIS, one found new DCIS, and three were benign.

Thus, among the 359 patients included in the study, nine (2.5%) patients had DCIS or cancer identified by screening mammography within 24 months of surgical treatment. This included six loco-regional recurrences in the ipsilateral breast and three new primaries in the contralateral breast. Three additional recurrences were detected by imaging that was ordered because of clinical signs/symptoms. Among the 6-month and 18-month mammograms, only one recurrence was diagnosed (by 6-month mammogram, see below). All recurrences were local without evidence of metastatic disease.

Discussion

In this study, 571 mammograms were completed at 6- and 18-months after BCT among 359 patients, and lead to 20 biopsies, of which two identified malignancy (10%) and 18 were false positives (90%). The added value of interval mammograms is likely limited by the natural history of breast cancer, which has few early recurrences, and by successful drug development which had reduced recurrence rates further. There are very few recurrences that occur within two years.^9,10 By contrast, 15% of local recurrences present more than 20 years after diagnosis.¹¹ Thus, it is difficult to undertake a prospective randomized trial studying the impact of treatment within two years of diagnosis, and this presumably explains why the literature consists of retrospective single institution studies with low recurrence rates.

The largest study supporting semi-annual mammography compared patients that “complied” with their institutional recommendation for five years of semi-annual surveillance to “noncompliant” patients. They found more early-stage cancers in “compliant” patients.¹² As with other studies, this was a single-institution retrospective non-randomized study, and while they enrolled 2,329 patients, it was still limited by small numbers of events (e.g., 4 vs. 6 patients with stage 2 disease at recurrence). In addition, they failed to differentiate new primary breast cancers from recurrent disease, did not delineate whether the recurrences were ipsilateral, and most important, failed to determine whether the recurrences were identified on the standard-of-care annual mammograms or the additional 6-month interim mammograms.

Other studies have concluded that annual surveillance is adequate. In one study of 1,425 patients, two recurrences (both noninvasive) were identified on mammograms done less than one year from BCT, and nine recurrences (three invasive, six noninvasive) were identified during the second year. The yield of mammography was thus similar to the general population, suggesting that annual screening should be adequate after BCT.¹³ Other studies have reinforced the conclusion that the yield is simply too low to justify additional screening. One study with 408 patients looked at mammography done within one year of completing radiation therapy (after BCT). Only two cases of DCIS (no invasive cancers) were identified.¹⁴ A third study concluded, after looking at their institutional experience, that the rate of ipsilateral breast cancer recurrence was extremely low during the first two years after BCT (two recurrences out of 375 patients), and the addition of interval ipsilateral mammograms at 6 months and 18 months provided no additional clinical benefit.¹⁵

As another consideration, mammography is less sensitive in women with a history of breast cancer [65.4% (95% CI, 61.5–69.0%)], compared to those with no such history [76.5% (95% CI, 71.7–80.7%)].¹⁶ Sensitivity is also lower within the first five years after primary breast cancer treatment [60.2% (95%CI, 54.7%-65.5%)] compared to after 5 years from treatment [70.8% (95% CI, 65.4%-75.6%)].¹⁶ Often in clinical practice, mammograms are obtained 6 months after BCT to serve as a baseline. However, there may be limited value at this time as the patient heals from surgery and radiation therapy. By one year, mammography provides a more reliable baseline.^17,18

The argument for interval mammography hinges on the benefit of catching a recurrence 6 months earlier. If this is unlikely, then the harm becomes more relevant. Although much of the incurred harm is difficult to quantify (anxiety and pain from biopsies), cost is the one aspect that can be quantified. For example, in one study of 128 patients that had mammography within one year of treatment, there was no benefit, but cost-of-care was markedly increased.¹⁹ Specifically, among the 87 patients who underwent post-operative mammography within 3 to 15 months, 78% of post-operative imaging costs were incurred by tests performed less than 12 months after completion of treatment (USD 32,506/USD 41,571). An additional USD 13,856 was spent on additional imaging and procedures that were clinically inconsequential. Six patients required further intervention at a cost of over USD 2,300 per patient, and three patients required invasive procedures, without clinical benefit.

To fill the void left by the absence of a multi-center prospective randomized trial, we have reviewed our institutional experience and reached similar conclusions. The risk of recurrence within two years of BCT is extremely low. There is limited value from adding mammograms at 6-month and 18-month intervals. These additional imaging studies are associated with false positives and unnecessary biopsies. Consequently, we found that our institutional policy, that recommends semi-annual mammography for two years after BCT is not supported by our institutional data. Specifically, we found minimal benefit despite numerous biopsies.

This does not preclude the possibility that certain sub-groups may benefit from semi-annual surveillance mammography. Perhaps we should be adopting biologically tailored imaging practices as we have for most other aspects of breast cancer care. At a minimum, additional mammography should be limited to patients that have a non-trivial risk of recurrence in two years. This could limit semi-annual screening to selective high-risk populations, such as patients with HER2 positive or triple-negative disease. The 6-month mammograms may also be of value in cases where we are concerned about missing a focus of cancer, e.g., patients with multi-focal disease, extensive DCIS, difficulty obtaining negative margins, or close margins. Failure to complete recommended treatment may be an additional factor.

We acknowledge that our study suffers from many of the same limitations. It is a retrospective single-institution study with a small sample size (359 patients). In addition, we did not consider compliance. All subjects that had an institutional treatment plan to complete 6-month mammograms were included, although there was variable compliance with some patients only completing 2 or 3 mammograms within the 2-year period.

In summary, guidelines vary on recommended frequency and duration of mammographic surveillance following primary breast cancer surgery. In our study of 359 patients, only one breast cancer recurrence was detected by 6-month interval surveillance mammography, and no recurrences were detected by 18-month mammograms. The recurrence was in a patient with high-risk features who was unable to complete recommended treatment. In addition, a significant number of false positive mammograms at 6-months and 18-months led to biopsies. Therefore, our patients would have benefited from the ASCO and NCCN endorsed guidelines for annual surveillance mammography following BCT. While we support this guideline for most patients, we acknowledge that select high-risk patients may benefit from more frequent mammography.