Addition of Contrast-enhanced Mammography to Tomosynthesis for Breast Cancer Detection in Women with a Personal History of Breast Cancer: Prospective TOCEM Trial Interim Analysis

Abstract

Background

Digital breast tomosynthesis (DBT) is often inadequate for screening women with a personal history of breast cancer (PHBC). The ongoing prospective Tomosynthesis or Contrast-Enhanced Mammography, or TOCEM, trial includes three annual screenings with both DBT and contrast-enhanced mammography (CEM).

Purpose

To perform interim assessment of cancer yield, stage, and recall rate when CEM is added to DBT in women with PHBC.

Materials and Methods

From October 2019 to December 2022, two radiologists interpreted both examinations: Observer 1 reviewed DBT first and then CEM, and observer 2 reviewed CEM first and then DBT. Effects of adding CEM to DBT on incremental cancer detection rate (ICDR), cancer type and node status, recall rate, and other performance characteristics of the primary radiologist decisions were assessed.

Results

Among the participants (mean age at entry, 63.6 years ± 9.6 [SD]), 1273, 819, and 227 women with PHBC completed year 1, 2, and 3 screening, respectively. For observer 1, year 1 cancer yield was 20 of 1273 (15.7 per 1000 screenings) for DBT and 29 of 1273 (22.8 per 1000 screenings; ICDR, 7.1 per 1000 screenings [95% CI: 3.2, 13.4]) for DBT plus CEM (P < .001). Year 2 plus 3 cancer yield was four of 1046 (3.8 per 1000 screenings) for DBT and eight of 1046 (7.6 per 1000 screenings; ICDR, 3.8 per 1000 screenings [95% CI: 1.0, 7.6]) for DBT plus CEM (P = .001). Year 1 recall rate for observer 1 was 103 of 1273 (8.1%) for (incidence) DBT alone and 187 of 1273 (14.7%) for DBT plus CEM (difference = 84 of 1273, 6.6% [95% CI: 5.3, 8.1]; P < .001). Year 2 plus 3 recall rate was 40 of 1046 (3.8%) for DBT and 92 of 1046 (8.8%) for DBT plus CEM (difference = 52 of 1046, 5.0% [95% CI: 3.7, 6.3]; P < .001). In 18 breasts with cancer detected only at CEM after integration of both observers, 13 (72%) cancers were invasive (median tumor size, 0.6 cm) and eight of nine (88%) with staging were N0. Among 1883 screenings with adequate reference standard, there were three interval cancers (one at the scar, two in axillae).

Conclusion

CEM added to DBT increased early breast cancer detection each year in women with PHBC, with an accompanying approximately 5.0%–6.6% recall rate increase.

Clinical trial registration no. NCT04085510

© RSNA, 2024

Supplemental material is available for this article.

Summary

In this prospective trial interim report, adding contrast-enhanced mammography to digital breast tomosynthesis increased breast cancer detection with limited increase in recalls in women with personal history of breast cancer.

Key Results

• ■ In this prospective trial interim report of 1273 women with a personal history of breast cancer, adding annual contrast-enhanced mammography (CEM) to digital breast tomosynthesis (DBT) increased the number of second malignancy detections by 7.1 per 1000 screenings in year 1 and 3.8 per 1000 in years 2 plus 3 (incidence screenings), with corresponding increases in the number of recalls by 6.6% and 5.0%, respectively.

• ■ In 18 breasts with cancer detected only at CEM, 13 (72%) were invasive and eight of nine (88%) with nodal staging were N0.

• ■ To date, interval cancer rate after two-observer integrated assessment was three of 1883 (1.6 per 1000 screenings) including two axillary recurrences missed at DBT plus CEM.

Introduction

Women treated for breast cancer are at increased risk for recurrent and second breast cancers, with a greater risk for women with estrogen receptor–negative disease (1). Early detection of these cancers improves patient outcomes (2). Unfortunately, women with a personal history of breast cancer (PHBC) often present with symptomatic cancer after a normal screening mammogram, that is, "interval cancer." In an analysis of Breast Cancer Surveillance Consortium data, Houssami et al (3) found a higher cancer prevalence (10.8 per 1000 screenings) in women with PHBC than in those without (5.8 per 1000 screenings), lower mammographic sensitivity (65.4% vs 76.5%), and a higher interval cancer rate (3.6 per 1000 vs 1.4 per 1000 screenings) (all P < .001).

Similarly, Lowry et al (4) reviewed 164 second breast cancer events in women with a history of prior stage I or II breast cancer. Among the 164 events, 125 (76.2%) were detected with screening mammography. Age younger than 50 years at primary diagnosis, initial anatomic stage II disease, and dense breasts (heterogeneously dense or extremely dense) were associated with screening failures, that is, clinically detected (interval) cancers. Yeom et al (5) similarly reported that dense breasts and mammographic failure for primary cancer detection predicted mammographic screening failures in women with PHBC.

The high number of symptomatic interval cancers indicates that routine annual mammography is inadequate for screening women with PHBC, prompting the addition of US or MRI. Contrast-enhanced mammography (CEM) is a lower-cost alternative to MRI (6) that appears to be better tolerated (7,8) and is preferred by approximately 70% of women who have undergone both examinations (9). The ongoing Tomosynthesis or Contrast-Enhanced Mammography, or TOCEM, trial includes three rounds of annual screening with both digital breast tomosynthesis (DBT) and CEM. This prospective clinical trial examines if addition of CEM to DBT substantially improves breast cancer detection compared with DBT alone, with minimal increase in false-positive findings, in women with PHBC. An interim analysis was performed in women with PHBC, examining the impact of adding annual CEM to DBT on cancer yield, stage, recall rate, and other performance characteristics.

Materials and Methods

Study Participants

At the UPMC Health System, in an institutional review board–approved, Health Insurance Portability and Accountability Act–compliant protocol (ClinicalTrials.gov: NCT04085510), a total of 1647 women with PHBC were enrolled (as of January 24, 2024) in this multisite, single-institution prospective study evaluating the addition of annual screening CEM to DBT. This sample size was estimated to provide at least 90% power to show a difference in cancer yields of at least four per 1000 screenings with the addition of annual CEM to DBT, assuming an overall incidence cancer yield of five to seven per 1000 screenings. While there was no intent to stop the trial, this interim analysis of consecutive women enrolled from October 23, 2019, through December 31, 2022, was performed to determine the appropriate next steps in the research program, as accrual completion was nearing and because the first women enrolled were about to complete the study. Women with a personal history of ductal carcinoma in situ (DCIS) or invasive breast cancer who had completed treatment, had at least one remaining breast without implants, were aged 30–85 years, had undergone routine annual mammography, and were asymptomatic, other than nonfocal pain, were eligible to participate. Women with a history of allergic reaction to iodinated contrast media were discouraged from participating (detailed further in the Appendix S1). Women were ineligible if their most recent prior mammogram was obtained more than 3 years earlier or if they were pregnant, breastfeeding, or currently receiving systemic chemotherapy. Women were recruited by mammographic technologists at the time of their routine mammogram. Referrals from breast surgeons and oncologists were also accepted. The research CEM had to be scheduled within 6 weeks of DBT (8-week maximum due to rescheduling).

Imaging Protocols

The methods used for DBT and CEM, as well as the methods used for biopsy, are detailed in Appendix S1 and Table S1.

Interpretation

All radiologists were specialists in breast imaging with 1–30 years of experience at the time of first participation. Of the 19 radiologists (all authors except for J.M.B., A.I.B.), 15 (79%) also interpret breast MRI. All radiologist observers completed training in CEM interpretation (10), with a review of at least 50 cases, and in the American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) atlas for CEM (11). Results are reported for the (randomly assigned and load balanced) first observer, who reviewed DBT first, recorded results, and then immediately reviewed the CEM images. The BI-RADS breast density reported is that which was visually assessed by the first radiologist observer (hereafter, observer 1). Clinical history and prior breast imaging results were available. Also reported are the results after integration with interpretation by a second radiologist (hereafter, observer 2), who reviewed the CEM images first, recorded the results, and then immediately reviewed the DBT images. Final clinical management was decided by observer 1 after observer 2 input consideration, and a clinical report was generated by observer 1. The clinical CEM report and images were available at the time of the next CEM examination. For any findings recommended for additional imaging or biopsy, the observers were asked to record whether the finding was recalled based on DBT or CEM (yes, no, or only in retrospect for each) findings. Recalled findings (excluding technical recalls) were considered test positive and were described and given a tentative BI-RADS final assessment as follows: 3, probably benign; 4A, low suspicion; 4B, moderate suspicion; 4C, high suspicion; or 5, highly suggestive of malignancy. Observers also recorded recommendations. Methods of additional imaging and biopsy are detailed in Appendix S1 and Table S1. For malignant findings, low-energy CEM images were retrospectively reviewed.

Reference Standard

Cancer was defined as invasive breast cancer in the breast or axilla or DCIS. Women with isolated distant metastatic disease that likely originated from an earlier primary breast cancer were excluded from the analysis. Pleomorphic lobular carcinoma in situ is treated like cancer in the study institution and was also considered malignant. The reference standard was considered benign for participants with no breast cancer diagnosis at surgery or follow-up imaging of at least 10.5 months, or, if no subsequent imaging or surgery had been performed, clinical follow-up of at least 12 months as of September 6, 2023. Cancers detected with imaging screening performed at least 10.5 months after the last screening examination were attributed to the next screening, provided that the participant was asymptomatic. The finding recalled was required to be at the site of cancer diagnosis (on review by W.A.B.) to credit a modality with cancer detection; otherwise, the modality result was considered false negative.

Statistical Analysis

Summary statistics of participant demographics were calculated (J.M.B). The cancer yield, recall rate, and positive predictive value of recall, or PPV1, were determined. Biopsy rates were determined at the participant level. Positive predictive values of biopsies performed (PPV3) were calculated at the participant and lesion levels. Sensitivity, specificity, and area under the receiver operating characteristic curve were calculated for participants with the reference standard, and receiver operating characteristic curves were plotted by using seven-category BI-RADS scores. Exact Clopper-Pearson 95% CIs were estimated for individual patient-level proportions. The statistical inference for lesion-level PPV3 and characteristics over multiple years were estimated by using nonparametric bootstrap methods based on 10 000 replicates, with participants as resampling units. For differences between DBT and CEM plus DBT, we also determined P values for nontrivial changes (eg, an increase in p [detection] of more than one per 1000, or p [added detections] = p_{added detections} >.001 or p [added recalls] = p_{added recalls} > .1%) using the appropriate exact tests. The P values are reported as two-sided values; we caution readers against interpreting values greater than .001 as definitive rejections of null hypotheses due to the unplanned nature of this interim report. Analyses were performed using R software (J.M.B.; version 1.1.383, R packages dplyr, boot, propCIs, pROC; R Studio) and/or SAS (A.I.B.; version 9.4, SAS Institute).

Results

Participant Characteristics

Of 1481 women who were scheduled to participate, 122 canceled their appointment (the majority due to logistical concerns). Of 1359 women who provided consent for imaging, 64 (4.7%) were unable to complete CEM due to failure of intravenous access, as were 11 (0.8%) due to estimated glomerular filtration rate of less than 45 mL/min (Fig 1). Four other women were not imaged (two due to immediate allergic reaction, one due to equipment failure, and one did not premedicate for a history of contrast agent allergy). After imaging, two ineligible women were withdrawn (one had only classic lobular carcinoma in situ and one had a malignant phyllodes tumor), and five women died before sufficient elapsed time for inclusion in the year 1 analysis, leaving 1273 women (2322 breasts) who completed year 1 examinations for analysis. Current tomosynthesis had been performed within 8 weeks prior to year 1 CEM for 623 participants and was concurrent with CEM for the remaining participants. Of the 905 women due for and remaining eligible for year 2 examinations, 819 (90.5%) completed year 2 examinations, and 227 of 265 (85.7%) of those eligible completed year 3 examinations. Mild allergic contrast agent reactions were observed in 10 of 1273 (0.8%) participants who completed year 1 imaging, in four of 819 (0.5%) who completed year 2 imaging, and zero of 227 (0%) who completed year 3 imaging. The mean age of the participating women (n = 1273) at study entry was 63.6 years ± 9.6 (SD) (median age, 64 years; IQR, 57–70 years; range, 30.8–84.6 years) (Table 1). Race and ethnicity were consistent with the demographic data in the study region. Nearly half of the participants had dense breasts (614 of 1273 [48.2%] heterogeneously dense; nine of 1273 [0.7%] extremely dense), and the vast majority (1184 of 1273 [93.0%]) were postmenopausal. The majority of women received endocrine therapy, including 115 (9.0%) women currently taking tamoxifen and 450 (35.4%) taking an aromatase inhibitor. The median time since the first cancer diagnosis was 6 years (range, 1–35 years) (Table S2). Demographics remained consistent over time except for participant age.

Figure 1:

Flowchart of study participants. Adequate follow-up was biopsy diagnosis of cancer or no breast cancer diagnosis at surgery or follow-up imaging of at least 10.5 months, or, if no subsequent imaging or surgery had been performed, clinical follow-up of at least 12 months. Four women had mild contrast agent reactions in year 2 and planned to premedicate and continue participation, but one failed to premedicate and was excluded in year 3. eGFR = estimated glomerular filtration rate, IV = intravenous.

Table 1:

Participant Demographics for Women with a Personal History of Breast Cancer Who Completed Respective Years of Screening CEM and DBT

As of September 6, 2023, reference standard was available for 1140 examinations at year 1, 637 at year 2, and 106 at year 3 (total of 1883 examinations). Sensitivity, specificity, and area under the receiver operating characteristic curve are reported in Appendix S1, Table S3, and Figure S1 for this subset of participants.

Cancer Yield

Overall, after consideration of observer 2 findings (Table S4) and interval cancers, cancer was found in 45 participants (36 in year 1, seven in year 2, and two in year 3); this encompassed 49 breasts and 57 malignant lesions. In the first screening round of 1273 participants, where all examinations were incidence screening DBT examinations (with prior DBT available), the DBT cancer yield for observer 1 was 20 of 1273 (15.7 per 1000 screenings [95% CI: 9.6, 24.2]). The combination of DBT plus CEM (prevalence CEM) produced a cancer yield of 29 of 1273 (22.8 per 1000 screenings [95% CI: 15.3, 32.6]) (Table 2). The difference in yields (incremental cancer detection rate due to the addition of CEM) was 7.1 per 1000 screenings [95% CI: 3.2, 13.4]; P < .001 for increase beyond one per 1000). In years 2 plus 3, the cancer yield of DBT interpretations was four of 1046 (3.8 per 1000 screenings [95% CI: 1.0, 7.6]) and that of DBT plus CEM was eight of 1046 (7.6 per 1000 screenings [95% CI: 2.9, 13.4]), for an incremental cancer detection rate of incidence screening CEM of 3.8 per 1000 screenings (95% CI: 1.0, 7.6; P = .01 for increase beyond one per 1000) (Figs 2, 3). We note that cancers in five women were seen only by observer 2: In year 1, three invasive ductal carcinomas were seen only at CEM, and one invasive ductal carcinoma was seen at both DBT and CEM; in year 2, one invasive lobular carcinoma was seen only at CEM.

Table 2:

Performance of DBT Alone or in Combination with CEM Across 2319 Screenings in 1273 Women with a Personal History of Breast Cancer According to Observer 1

Figure 2:

Images in a 67-year-old woman with triple receptor–negative invasive ductal carcinoma (IDC) seen only at contrast-enhanced mammography (CEM) at year 2. (A) Left craniocaudal (CC) (left) and mediolateral oblique (MLO) (right) low-energy images show scattered fibroglandular density and postsurgical scarring, with clips in the lower inner quadrant at the site of lumpectomy for a 2.1-cm grade 3 IDC, estrogen receptor– and progesterone receptor–positive and human epidermal growth factor receptor 2 (HER2) (ERBB2 gene)–negative lesion 11 years prior. Scattered benign-appearing calcifications are noted. The participant also completed radiation therapy and adjuvant chemotherapy and was treated with tamoxifen for 7 years and then with an aromatase inhibitor for 3 years, with last use 1 year prior to study entry. (B) Recombined CC (left) and MLO (right) CEM images obtained in year 2 show moderately conspicuous enhancement of an oval mass in the upper outer left breast (arrows), which was new from the prior CEM examination (not shown). This lesion was assessed as Breast Imaging Reporting and Data System (BI-RADS) 4B, moderately suspicious, by observer 1 and as BI-RADS 3, probably benign, but recommended for additional evaluation, by observer 2. At the time, CEM-guided biopsy was not available, so the participant underwent MRI and MRI-guided biopsy. (C) Axial maximum intensity projection from T1-weighted fat-suppressed MRI (left) shows moderately intense enhancement of the same mass (arrow), with plateau and washout kinetics (arrow) on axial post-contrast fat-suppressed T1-weighted image with kinetic overlay (right). MRI-guided biopsy and excision revealed a 0.5-cm grade 3 IDC, triple receptor–negative lesion (Ki-67 proliferation index of 55%). Three sentinel nodes were negative for metastasis.

Figure 3:

Images in a 77-year-old woman with invasive ductal carcinoma (IDC) seen only at contrast-enhanced mammography (CEM) and ductal carcinoma in situ (DCIS) seen only on low-energy (LE) images and digital breast tomosynthesis images at year 2. (A) Bilateral craniocaudal (CC) (left) and mediolateral oblique (MLO) (right) LE images show heterogeneously dense parenchyma and postsurgical changes with dystrophic calcifications in the upper right breast and clips in the right axilla from breast-conserving therapy for a grade 1 IDC, estrogen receptor (ER)– and progesterone receptor (PR)–positive, human epidermal growth factor receptor 2 (HER2) (ERBB2 gene)–negative, 19 years earlier, for which she had taken Anastrozole for 4 years. Only observer 1 recalled the participant for linear calcifications in the left breast (arrows), which are better seen on (B) close-up CC (left) and MLO (right) LE images (arrows) and (C) a spot magnification CC view of the central left breast. In addition to typically benign calcifications more laterally and fine linear calcifications (arrow in C), seen only on C more anteriorly, there is a group of amorphous calcifications (circle in C). Both areas of calcification were recommended for biopsy, despite lack of enhancement on (D) recombined CC (left) and MLO (right) CEM images. A moderately conspicuous enhancing mass in the retroareolar left breast was newly seen, which was only evident at CC CEM (arrow in D) and was assessed as Breast Imaging Reporting and Data System (BI-RADS) 4A, low suspicion, by observer 1 and as BI-RADS 4B, moderate suspicion, by observer 2. (E) CEM-directed US images (left = transverse plane, right = longitudinal plane) of the retroareolar left breast show an irregular, hypoechoic mass (arrows) with an echogenic rim, highly suggestive of malignancy (BI-RADS 5). US-guided core biopsy and mastectomy revealed a 2.1-cm grade 3 ER and PR-positive, ERBB2-negative IDC. Stereotactic biopsy of the linear calcifications revealed high-grade ER and PR-positive DCIS, and biopsy of the amorphous calcifications yielded atypical ductal hyperplasia. At mastectomy, 5.5 cm of high-nuclear-grade DCIS was found, discontinuous with the retroareolar IDC. One of two sentinel nodes showed isolated tumor cells (N0).

Interval Cancers

To date, there have been three symptomatic interval cancers (undetected by either of the two observers using both modalities), all in year 1. Two women experienced a palpable lump in the axilla within 6 and 4 months of the initial CEM examination due to axillary recurrences not included on the CEM, or either of CEM or DBT, images, respectively. A third woman noted a lump at the surgical scar 8 months after her first screening CEM examination due to recurrent invasive lobular carcinoma. At retrospective review, there was minimal enhancement of the scar (Fig 4). Further details of five other false-negative findings at DBT plus CEM only at the lesion level are given in Appendix S1.

Figure 4:

Images in a 47-year-old woman with recurrent invasive lobular carcinoma (ILC) at the scar detected clinically 6 months after imaging (interval cancer). (A) Close-up craniocaudal (CC) (left) and mediolateral oblique (MLO) (right) digital breast tomosynthesis 6-mm slab images of the right breast show the area of scarring from breast-conserving therapy 2 years earlier for multifocal estrogen receptor (ER)/progesterone receptor (PR)-positive, human epidermal growth factor receptor 2 (HER2) (ERBB2 gene)–positive ILC, interpreted as benign, Breast Imaging Reporting and Data System (BI-RADS) 2, by both observers. The participant completed both neoadjuvant and adjuvant chemotherapy and was taking tamoxifen. (B) Close-up CC (left) and MLO (right) recombined contrast-enhanced mammography (CEM) images show nonmass enhancement at the scar (arrows), interpreted as benign, BI-RADS 2, by both observers. (C) US scan in longitudinal plane (left) obtained 6 months later, when the participant reported feeling a lump at the scar, shows a superficial irregular, parallel, hypoechoic mass (arrow) at the scar; the mass shows internal vascularity on transverse power Doppler scan (right, arrow). US-guided core biopsy and mastectomy revealed a 2.4-cm grade 3 ILC, ER-positive, PR-negative, HER2-positive lesion (Ki-67 proliferation index of 90%), with one of two sentinel nodes showing isolated tumor cells (N0).

Recall Rate and PPV3 of Biopsies

The recall rates in year 1 were 103 of 1273 (8.1%) for (incidence) DBT alone for observer 1 and 187 of 1273 (14.7%) for DBT plus CEM (prevalence CEM), a difference of 84 of 1273 (6.6% [95% CI: 5.3, 8.1]; P < .001 for increase beyond 0.1%). The recall rates in years 2 and 3 were 40 of 1046 (3.8%) for DBT alone and 92 of 1046 (8.8%) for DBT plus CEM, a difference of 52 of 1046 (5.0% [95% CI: 3.7, 6.3]; P < .001 for increase beyond 0.1%). The lesion-level PPV3 was 22 of 50 (44%) for DBT in year 1 and 35 of 102 (34%) for DBT plus CEM (difference of −9.7% [95% CI: −19.9, −0.2], P = .045), reflecting a PPV3 of added CEM-prompted biopsies of 13 of 52 (25% [95% CI: 13, 38]). PPV3 decreased to four of 20 (20%) for DBT in years 2 and 3 and nine of 52 (17%) for DBT plus CEM (difference of −2.7% [95% CI: −16.5, 9.5]; P = .69), for a PPV3 of added CEM-prompted biopsies of five of 32 (16% [95% CI: 4, 30]).

Types of Malignancies Detected and Relationship to Imaging Method

Four malignant lesions were seen at DBT but not at CEM; three were calcifications due to DCIS, and one was calcifications due to a 0.4-cm grade 2 invasive ductal carcinoma with associated DCIS. One grade 2 DCIS lesion and the invasive ductal carcinoma with associated DCIS lesion were not evident on low-energy two-dimensional mammograms obtained at the time of CEM, and one each of grade 2 and grade 3 DCIS calcifications was visible on low-energy images.

At the breast level, of the 18 cancers seen only at CEM, nine (50%) were in the 49% of participants with dense breasts (Table 3). Of these 18 cancers, 13 (72%) were invasive, and eight of nine (89%) of those with pathologic nodal staging were node negative. The median invasive tumor size of the largest cancer for each breast was 6 mm (range, 1–21 mm) for those seen only at CEM and 9 mm (range, 1–18 mm) for those seen at both DBT and CEM. Eight of the invasive cancers seen only at CEM were 1 cm or smaller and node negative (five of these had pathologic nodal staging, and three were clinically node negative), representing a 73% increase in the detection of such tumors from 11 seen at DBT to 19 after adding CEM.

Table 3:

Method of Cancer Detection and Details of Staging 49 Breasts with Cancer in 45 Women

Discussion

Prior studies have shown high rates of symptomatic breast cancer after a normal screening mammogram in women with a personal history of breast cancer (PHBC), prompting recommendations for supplemental screening MRI (12,13). Not all women can tolerate MRI; we studied the use of contrast-enhanced mammography (CEM) to improve screening in such women. Adding annual CEM to digital breast tomosynthesis (DBT) in women with PHBC substantially increased early cancer detection: The added cancer yield at CEM after DBT was 7.1 per 1000 screenings in year 1 (P < .001) and averaged 3.8 per 1000 screenings per year in years 2 and 3 (P = .011). The low cancer yield at DBT, which was only 3.8 per 1000 in incident years (years 2 and 3), suggested that some of the cancers that usually would have been seen were detected in the prior year at CEM.

There are many approaches to compensate for the reduced performance of annual mammography in women with PHBC. Semiannual mammography was assessed by Arasu et al (14), who reported that recurrences diagnosed after semiannual mammography were of a lower stage than those found at annual mammography. In the American College of Radiology Imaging Network, or ACRIN, 6666 trial (15), across 4010 annual screenings at year 1, 2, and 3 in women with PHBC, physician-performed US depicted an additional 17 cancers (4.2 per 1000 screenings) not seen at mammography. Across seven series encompassing 6839 screening MRI scans of women with PHBC (15–21), 102 (14.9 per 1000) cancers were seen only with MRI among the examinations of 655 women recalled for additional testing; the positive predictive value of MRI-prompted biopsies was 97 of 337 (29%). Similar results have been observed across eight studies of screening abbreviated MRI in 6838 women with PHBC, with 106 cancers detected (15.5 per 1000 screenings), and similar specificity as for full protocol MRI, as reviewed by Kuhl (22). Based on these and other results, in 2018, the American College of Radiology issued guidance recommending adding annual screening MRI to annual mammography in women with PHBC diagnosed by 50 years of age or with dense breasts (12). In one analysis of 2950 women with PHBC, 1805 (61%) met these criteria for supplemental MRI (23). Women with PHBC without one of these factors should "consider" adding annual MRI, as restated in 2023 (13). Unfortunately, there is limited availability of MRI screening, even when using abbreviated techniques, and the examination is not always well tolerated (24,25).

Since 2022, the National Comprehensive Cancer Network, or NCCN, guidelines have suggested screening CEM for those recommended for screening MRI but who are not able to tolerate it (26), and 2023 American College of Radiology recommendations also include CEM as an alternative to MRI (13). The sensitivity of CEM appears comparable to or slightly lower than that of MRI, but there appear to be fewer false-positive findings with CEM. In a recent meta-analysis of 13 studies comparing CEM with MRI, Xiang et al (27) reported overall sensitivity of CEM and MRI were each 97%, but specificity was better for CEM (0.66 [95% CI: 0.59, 0.71]) than for MRI (0.52 [95% CI: 0.46, 0.58]). Pötsch et al (28) reported 97% sensitivity for MRI and 91% sensitivity for CEM across seven studies encompassing 1137 lesions, with specificity of 69% for MRI and 74% for CEM. CEM has additional benefits compared with MRI, including lower capital equipment and contrast agent costs, a shorter acquisition time, and potentially increased access. Conversely, in women with PHBC, axillary recurrence can be missed at CEM and may be better seen at MRI; to date, two of the three interval cancers in our series were axillary recurrences.

There have been few prior reports of screening CEM in women with PHBC. In 2020, Gluskin et al (29) reported on 971 standalone CEM examinations in 541 asymptomatic women with PHBC, overlapping with the prior report of Sung et al (30). Of the 21 cancers, six were seen on low-energy images, and nine were observed only with contrast material (an added cancer detection rate of 9.3 per 1000 screenings). There were six interval cancers, but only three were symptomatic. The recall rate for CEM was 93 of 971 (9.6%). Elder et al (31) reported on 1191 prevalence screening CEM examinations performed simultaneously with DBT in women with PHBC, with 18 cancers seen at both CEM and DBT and 17 cancers seen only at CEM, for an added cancer detection rate of 17 of 1191 (14.3 per 1000 screenings). The recall rate for CEM was 64 of 1191 (5.4%).

Our interim results, while preliminary, support the use of CEM for annual supplemental screening in women with PHBC. Based on the very small size of cancers observed in this analysis, we are now commencing a study of biennial supplemental CEM. We expect a slight decrease in false-positive findings and hope to maintain the benefits of early cancer detection. We look forward to results from the recently opened Contrast Mammography Imaging Screening Trial, or CMIST, which will evaluate two rounds of annual screening CEM in a broader population of women with dense breasts (ClinicalTrials.gov: NCT05625659). Interestingly, in our study we found detection benefits from CEM were similar in women with dense and women with nondense breasts. CMIST will exclude from participation women who have undergone MRI in the prior 3 years or who plan to undergo MRI between screening CEM rounds. We did not restrict MRI use in this study, but we did observe a lower rate of cancer detection among women who had previously undergone MRI, particularly among those who underwent MRI in the prior 3 years, as detailed in Appendix S1.

There are several limitations to our work. First, we did not uniquely record low-energy findings prospectively nor did we assess recalls prompted only by those findings; therefore, we are not able to report the specificity of CEM as a standalone examination. Additional studies are in progress to address this issue. Second, some of the cancers found at CEM were seen only by a second radiologist; more detailed analysis of individual reader performance and the use of DBT after CEM is planned at the conclusion of the study. Third, the very high cancer yield in year 1 may be due to participant selection bias; women were mostly informed of the study at the time of their routine annual mammography and may have been more inclined to participate if they had been recalled for additional testing (due to heightened concern and logistical issues of already having to return). Fourth, paper case report forms were used, and it is possible that enhancing findings were sometimes seen by the first observer prior to recording DBT interpretations. Finally, our work is performed at multiple centers within one parent institution, and this may limit generalizability of our results. As this is an interim analysis, final estimates of the effects of adding annual CEM to DBT may differ somewhat.

In conclusion, in this interim analysis of an ongoing prospective trial, adding contrast-enhanced mammography (CEM) to digital breast tomosynthesis substantially improved the detection of early breast cancer in women with a personal history of breast cancer, and this benefit appears to persist each year. Cancers found with CEM were nearly all early breast cancers (ductal carcinoma in situ or node-negative invasive breast cancers).