Baseline Screening Mammography: Performance of ...

4 downloads 0 Views 497KB Size Report
Amana L. Akhtar1. Marie B. Synnestvedt3. Mitchell Schnall1. Emily F. Conant1. McDonald ES, McCarthy AM, Akhtar AL,. Synnestvedt MB, Schnall M, Conant EF.
Wo m e n ’s I m a g i n g • O r i g i n a l R e s e a r c h McDonald et al. Baseline Screening FFDM Versus Digital Breast Tomosynthesis

Downloaded from www.ajronline.org by Univ of Penn Libr on 06/01/16 from IP address 170.212.0.64. Copyright ARRS. For personal use only; all rights reserved

Women’s Imaging Original Research

Baseline Screening Mammography: Performance of Full-Field Digital Mammography Versus Digital Breast Tomosynthesis Elizabeth S. McDonald1 Anne Marie McCarthy 2 Amana L. Akhtar 1 Marie B. Synnestvedt 3 Mitchell Schnall1 Emily F. Conant 1 McDonald ES, McCarthy AM, Akhtar AL, ­Synnestvedt MB, Schnall M, Conant EF

Keywords: baseline screening, digital breast ­tomosynthesis, full-field digital mammography, recall, screening harms, screening mammography DOI:10.2214/AJR.15.14406 Received January 20, 2015; accepted after revision April 7, 2015. E. F. Conant is a member of the Hologic Scientific Advisory Board. Based on a presentation at the ARRS 2014 Annual Meeting, San Diego, CA. Supported by grant U54CA163313 from the National Cancer Institute at the National Institutes of Health: Population-Based Research Optimizing Screening through Personalized Regimens (PROSPR) Network. 1 Breast Imaging Division, Department of Radiology, University of Pennsylvania, Perelman School of Medicine, 3400 Spruce St, 1 Silverstein, Philadelphia, PA 19104-4283. Address correspondence to E. S. McDonald ([email protected]). 2 General Medicine Division, Massachusetts General Hospital, Harvard Medical School, Boston, MA. 3 Division of General Internal Medicine, University of Pennsylvania, Perelman School of Medicine, ­Philadelphia, PA.

This article is available for credit. AJR 2015; 205:1143–1148 0361–803X/15/2055–1143 © American Roentgen Ray Society

OBJECTIVE. Baseline mammography studies have significantly higher recall rates than mammography studies with available comparison examinations. Digital breast tomosynthesis reduces recalls when compared with digital mammographic screening alone, but many sites operate in a hybrid environment. To maximize the effect of screening digital breast tomosynthesis with limited resources, choosing which patient populations will benefit most is critical. This study evaluates digital breast tomosynthesis in the baseline screening population. MATERIALS AND METHODS. Outcomes were compared for 10,728 women who underwent digital mammography screening, including 1204 (11.2%) baseline studies, and 15,571 women who underwent digital breast tomosynthesis screening, including 1859 (11.9%) baseline studies. Recall rates, cancer detection rates, and positive predictive values were calculated. Logistic regression estimated the odds ratios of recall for digital mammography versus digital breast tomosynthesis for patients undergoing baseline screening and previously screened patients, adjusted for age, race, and breast density. RESULTS. In the baseline subgroup, recall rates for digital mammography and digital breast tomosynthesis screening were 20.5% and 16.0%, respectively (p = 0.002); digital breast tomosynthesis screening in the baseline subgroup resulted in a 22% reduction in recall compared with digital mammography, or 45 fewer patients recalled per 1000 patients screened. Digital breast tomosynthesis screening in the previously screened patients resulted in recall reduction of 14.3% (p < 0.001; p for interaction = 0.21). The recall rate reduction for baseline screening was especially pronounced in women younger than 50 years (p = 0.005). DBT implementation resulted in an increase in cancer detection in the baseline subgroup of 40.5% versus an increase in the previously screened subgroup of 17.4%. DBT implementation resulted in an increase in PPV1 in the baseline subgroup of 85% versus 35.3% in the previously screened subgroup, although the p-interaction was not significant. CONCLUSION. If resources are limited, women younger than 50 years who are undergoing baseline screening or do not have prior available mammograms may benefit more from digital breast tomosynthesis than from digital mammography alone.

T

here is increasing evidence that the addition of digital breast tomosynthesis to digital mammography results in a decreased recall rate [1–8] and an increase in cancer detection [1, 3, 6–8]. According to the Grading of Recommendations Assessment, Development, and Evaluation working group scale [9], the current quality of the evidence for using digital breast tomosynthesis combined with digital mammography as a useful breast cancer screening tool, compared with digital mammography alone, is high. The evidence for this includes several high-quality

single-site studies with reproducible reductions in recall rates without losses in cancer detection as well as a large multicenter trial involving 454,850 examinations showing similar outcomes [3]. Unfortunately, because of cost constraints and only recent additional reimbursement over that for full-field digital mammography (FFDM), digital breast tomosynthesis remains a limited resource that is not currently offered to all women presenting for screening mammography. A recent survey of the Society of Breast Imaging [10] found that just 27% of its members use digital breast to-

AJR:205, November 2015 1143

Downloaded from www.ajronline.org by Univ of Penn Libr on 06/01/16 from IP address 170.212.0.64. Copyright ARRS. For personal use only; all rights reserved

McDonald et al. mosynthesis clinically. Only 5.5% of digital breast tomosynthesis users were able to offer this examination to all patients, and there were widely varying criteria for stratification of patients to digital breast tomosynthesis or digital mammography screening. Unless there is universal conversion to digital breast tomosynthesis, there is a need for evidencebased guidelines to identify which patients benefit most from this new technology [11]. This is an area of clinical equipoise; clarification and consensus on digital breast tomosynthesis resource utilization have significant potential to positively affect patient care. It is well established that false-positive mammography findings are associated with increased short-term anxiety [12] as well as significant cost to the health care system [13]. Women who present for baseline screening or women presenting with no available prior examinations are more likely to be recalled than patients with comparison images available [14]. The purpose of this study was to evaluate the screening outcomes of digital breast tomosynthesis combined with digital mammography compared with those of conventional digital mammography for the subgroup of patients undergoing baseline screening and patients with no available comparison examinations in a natural experiment incorporating more than 25,000 examinations. This natural experiment involved an audit of population screening data obtained during routine clinical practice. Digital breast tomosynthesis combined with digital mammography imaging was implemented for our entire screening population in the fall of 2011. In this analysis, we compare the screening outcomes of recall rate, cancer detection, and positive predictive values (PPVs) for the specific subgroup of patients with no available comparison examinations over a period of 12 months before and 17 months after digital breast tomosynthesis implementation. Determining whether digital breast tomosynthesis has greater effectiveness in a specific patient population is important for best-practice use of digital breast tomosynthesis in an environment of decreasing reimbursement for radiologic studies and limited resources for implementation of new technologies.

for screening mammography at our institution from September 1, 2010, through August 30, 2011 (n = 10,728), and from October 1, 2011, through February 28, 2013 (n = 15,571). There was a stable group of six dedicated breast imagers across both time periods. Digital breast tomosynthesis was implemented for all mammographic screening on September 19, 2011, consisting of two-view digital breast tomosynthesis and two-view digital mammography of each breast (Dimension, Hologic). Before this time, all women had been screened with FFDM. Patients with no history of breast cancer and no clinical signs of breast cancer presenting for screening comprised both cohorts. At least 12 months of follow-up data were available for all patients in both cohorts. For the purposes of this study, “baseline screening” was defined as either the first screening study for the patient or, if the patient had undergone prior mammographic imaging, no prior study (screening or diagnostic) was obtainable despite attempts to retrieve any prior images. In the digital mammography cohort, 10,728 examinations were performed, of which 1204 were considered baseline studies. In the digital breast tomosynthesis cohort, 15,571 examinations were performed, of which 1859 were baseline studies. Overall, there were 26,299 screening events.

Materials and Methods Study Population

Screening Outcome Measures

This was an institutional review board–approved retrospective analysis of a natural outcomes experiment using methods described elsewhere [6]. The study population consisted of all women presenting

1144

Screening Interpretation All examinations for both cohorts were interpreted by one of six board-certified radiologists with specialization in breast imaging. There was no change in radiologist staffing during the study period. Radiologist experience in breast imaging ranged from 3 to 22 years. Before implementation of digital breast tomosynthesis, all readers received 8 hours of training in digital breast tomosynthesis interpretation, as mandated by the U.S. Food and Drug Administration.

Data Collection All screening mammography studies were reported through a radiology information system (Centricity, GE Healthcare). This reporting system is based on the American College of Radiology BI-RADS categories [15]. Breast density, BI-RADS assessment categories, and demographics were recorded at the time of interpretation, as described elsewhere [6]. There was a subsequent retrospective database query that retrieved population screening volumes and outcomes as well as breast density and other clinical information.

The metrics evaluated were imaging volumes, recall rates (percentages), and cancer detection rates (number of cancers detected per 1000 patients screened). PPVs were obtained: PPV 1 refers to the number of cancers per number of re-

calls, PPV 2 refers to the number of cancers per biopsy recommended, and PPV 3 refers to the number of cancers per biopsy performed [16]. The percentage of patients recalled from screening examinations was assessed as those given a BI-RADS assessment category 0 (additional imaging needed), 4, or 5. Biopsy or surgical outcomes were tracked within 180 days of the screening recall through the electronic medical record, pathology database, and radiology information system. The institutional tumor registry was also queried though September 2013 and the Pennsylvania State Cancer Registry was queried through June 2012 (the most recent available records).

Statistical Analysis Baseline characteristics and screening outcomes in the two cohorts were compared using t tests for continuous variables and the chi-square test for categorical variables. The odds of recall for digital mammography versus digital breast tomosynthesis were compared using logistic regression analysis adjusted for age, race, breast density, and radiologist. Logistic regression models were stratified by whether the patient had prior available mammography studies (i.e., baseline status). The interaction of baseline status with type of mammography was tested using a cross-product term and the Wald test of statistical significance. Logistic regression analysis of the odds of cancer diagnosis was performed among the subgroup of women who were recalled to assess differences in PPV for digital mammography and digital breast tomosynthesis, and results were stratified by baseline status. Because of the smaller sample size in this subgroup, race was not included as an adjustment variable, age was divided into four categories rather than five, and breast density was categorized as dense (heterogeneously dense and extremely dense) or not dense (almost entirely fatty and scattered fibroglandular densities). We did not account for multiple observations per person because we assumed that each person’s outcomes for screening were independent. All statistical tests were two sided and were performed using STATA (version 12, StataCorp) or SAS (version 9.3, SAS Institute) software. A p value of less than 0.05 was considered statistically significant.

Results Multivariate analysis for population differences in age, breast density, prior studies, and race has already been performed on the entire digital breast tomosynthesis and digital mammography cohort and has been published elsewhere [6]. Within the baseline and previously screened subgroups, the characteristics of those screened with digital breast tomosynthesis and digital mammography

AJR:205, November 2015

Baseline Screening FFDM Versus Digital Breast Tomosynthesis TABLE 1: Demographics of the Baseline and Previously Screened Patients Patients Undergoing Baseline Screening (n = 3063)

Downloaded from www.ajronline.org by Univ of Penn Libr on 06/01/16 from IP address 170.212.0.64. Copyright ARRS. For personal use only; all rights reserved

Patient Characteristics Age (y), mean (SD)

Cohort 1: Digital Mammography (n = 1204)a

Cohort 2: Digital Breast Tomosynthesis (n = 1859)b

49.2 (10.1)

48.9 (10.6)

8.7

9.9

Age group

Previously Screened Patients (n = 23,236)

p

Cohort 1: Digital Mammography (n = 9524)

Cohort 2: Digital Breast Tomosynthesis (n = 13,712)

p

0.44

57.9 (10.8)

57.8 (10.6)

0.60

1.6

1.3

0.70

< 40 y

0.16

40–49 y

51.6

51.7

24.2

24.8

50–59 y

23.5

21.8

34.4

33.8

60–69

12.5

12.5

25.1

25.9

≥ 70 y

3.7

4.1

14.7

14.2

30.5 (8.3)

30.3 (8.4)

29.5 (7.2)

29.6 (7.4)

BMI, mean (SD) Race

0.58 0.01

0.05

White

26.8

30.1

42.4

42.1

Black

61.2

57.6

49.7

49.2

Hispanic

1.5

1.4

0.8

0.9

Asian

5.8

4.3

3.0

2.9

Other or unknown

4.7

6.7

4.1

4.9

1

12.4

13.0

10.5

11.8

2

53.9

53.0

57.1

55.9

3

31.5

31.3

30.5

30.4

4

2.2

2.7

1.8

1.9

BI-RADS breast density category

0.27

0.83

0.02

Note—Except where noted otherwise, data are percentage of patients. BMI = body mass index (weight in kilograms divided by the square of height in meters). aThis cohort represents 11.2% of 10,728 women who underwent digital mammography screening. bThis cohort represents 11.9% of 15,571 women who underwent digital breast tomosynthesis screening.

were similar (Table 1). As expected, patients undergoing baseline screening were younger than previously screened patients. Recall rates, cancer detection, and PPVs 1–3 were compared between the baseline and previously screened patients (Table 2). The recall rates for the entire digital mammography and digital breast tomosynthesis cohorts were 10.4% (1112/10,728) and 8.8% (1366/15,571), respectively (p < 0.001) [6]. The recall rates for the baseline subgroup were 20.5% (247/1204) for digital mammography and 16.0% (298/1859) for digital breast tomosynthesis (p = 0.002). The recall rates for the previously screened patients were 9.1% (865/9524) for digital mammography and 7.8% (1068/13,712) for digital breast tomosynthesis (p < 0.001). Digital breast tomosynthesis use in the baseline subgroup resulted in a reduction in recall rate of 22% (p for interaction = 0.21); digital breast tomosynthesis use in the previously screened patients resulted in a reduction in recall rate of 14.3%. Interactions between digital breast tomosynthesis use and baseline status were ex-

plored but were not statistically significant. Even though the p value for interaction between digital breast tomosynthesis implementation among baseline and previously screened patients did not reach statistical significance, digital breast tomosynthesis resulted in 45 fewer recalls/1000 women among the baseline patients and 13 fewer recalls/1000 women among the previously screened patients. For women younger than 50 years, the recall rate for baseline digital mammography was 21.2% versus 16.1% for digital breast tomosynthesis, a reduction of 24.1% (p = 0.005). There was a trend toward a statistically significant additional reduction in recall when women younger than 50 years in the baseline subgroup were offered digital breast tomosynthesis screening, compared with previously screened women younger than 50 years who were offered digital breast tomosynthesis (p for interaction = 0.08). For women age 50 years and older, the recall rate for baseline digital mammography was 19.5% versus 16.0% for digital breast tomosynthesis, a reduction of 17.9% (p = 0.12).

Among women with nondense breasts, the recall rate for baseline digital mammography was 20.3% compared with 15.4% for baseline digital breast tomosynthesis, a reduction of 24.1% (p = 0.004). For women with dense breasts the recall rate for baseline digital mammography was 20.9% compared with 17.3% for baseline digital breast tomosynthesis, a reduction of 17.2% (p = 0.14). In the baseline subgroup, five malignancies were found with digital mammography and 11 were found with digital breast tomosynthesis. In the previously screened patients, 44 malignancies were found with digital mammography and 74 were found with digital breast tomosynthesis. The cancer detection rate of the baseline subgroup was 4.2 cancers/1000 women for digital mammography and 5.9 cancers/1000 women for digital breast tomosynthesis (p = 0.51); the cancer detection rate of the previously screened patients was 4.6 cancers/1000 women for digital mammography and 5.4 cancers/1000 women for digital breast tomosynthesis (p = 0.41). Digital breast tomosynthesis use re-

AJR:205, November 2015 1145

McDonald et al. TABLE 2: Screening Outcomes for Patients Undergoing Baseline Screening and Previously Screened Patients: ­Recall Rate, Cancer Detection, and Positive Predictive Values (PPVs) for Digital Mammography and Digital Breast Tomosynthesis Cohort 1: Digital Mammography

Cohort 2: Digital Breast Tomosynthesis

Recall rate

247 (20.5)

298 (16.0)

0.002

−22.0

Age < 50 y

154 (21.2)

184 (16.1)

0.005

−24.1

Downloaded from www.ajronline.org by Univ of Penn Libr on 06/01/16 from IP address 170.212.0.64. Copyright ARRS. For personal use only; all rights reserved

Patient Subgroup

p

Change Between Digital Mammography and Digital Breast Tomosynthesis (%)

Patients undergoing baseline screeninga

Age ≥ 50 y

93 (19.5)

114 (16.0)

0.12

−17.9

Nondense breast tissue

162 (20.3)

189 (15.4)

0.004

−24.1

Dense breast tissue

85 (20.9)

109 (17.3)

0.14

−17.2

5 (4.2)

11 (5.9)

0.51

40.5

Cancer detection rate/1000 patients PPVs (%) No. of cancers per no. of recalls

2.0

3.7

0.25

85

No. of cancers per biopsy recommended

12.8

14.5

0.81

13.3

No. of cancers per biopsy performed

15.6

14.1

0.84

−9.6

Recall rate

865 (9.1)

1068 (7.8)

< 0.001

−14.3

Age < 50 y

291 (11.9)

396 (11.0)

0.32

−7.6

Age ≥ 50 y

574 (8.1)

672 (6.6)

< 0.001

−18.5

Previously screened patientsb

Nondense breast tissue

505 (7.8)

630 (6.8)

0.01

−13.1

Dense breast tissue

360 (11.7)

438 (9.9)

0.01

−15.4

44 (4.6)

74 (5.4)

0.41

17.4

No. of cancers detected (rate/1000 patients) PPVs (%) No. of cancers per no. of recalls

5.1

6.9

0.09

35.3

No. of cancers per biopsy recommended

24.6

27.6

0.48

12.2

No. of cancers per biopsy performed

26.6

28.7

0.65

7.9

Note—Except where noted otherwise, data are number (%) of patients. aThere were 1204 patients in cohort 1 and 1859 patients in cohort 2. bThere were 9524 patients in cohort 1 and 13,712 patients in cohort 2.

sulted in increased cancer detection of 17.4% over digital mammography screening in the previously screened patients and 40.5% over digital mammography screening in the baseline subgroup (p for interaction = 0.74). The PPV 1 for baseline screening was 2.0% for digital mammography and 3.7% for digital breast tomosynthesis (p = 0. 25). The PPV 1 for nonbaseline screening was higher for digital mammography and digital breast tomosynthesis, 5.1% and 6.9%, respectively (p = 0.09). Although there was an 85% increase in PPV 1 with digital breast tomosynthesis in the baseline group and 35.3% increase in PPV 1 with digital breast tomosynthesis in the previously screened patients, this difference was not statistically significant (p interaction = 0.62). The cancer yield of biopsies recommended (PPV 2) and biopsies performed (PPV 3) was lower for patients undergoing baseline screening than

1146

for previously screened patients overall, and there was no statistically significant difference in PPV 2 or PPV 3 for digital mammography and digital breast tomosynthesis in either subgroup (Table 2). The absence of prior mammography results was associated with increased odds of recall when digital breast tomosynthesis– screened patients were compared with digital mammography–screened patients, although the p value for interaction was not statistically significant (p = 0.29). In the previously screened subgroup, women screened with digital breast tomosynthesis had 16% lower odds of recall than did women screened with digital mammography, after adjusting for breast density, age, and race (odds ratio [OR], 0.84; 95% CI, 0.76–0.92; p < 0.001) (Table 3). In the baseline subgroup, women screened with digital breast tomosynthesis had 26% lower odds of recall than women

screened with digital mammography after multivariate adjustment (OR, 0.74; 95% CI, 0.61–0.89; p = 0.002) (Table 3). Digital breast tomosynthesis was associated with statistically significant increased PPV 1 (57%) compared with digital mammography in the nonbaseline group after adjusting for age and density (OR, 1.57; 95% CI, 1.05–2.33; p = 0.03) (Table 4). The PPV 1 was also higher for digital breast tomosynthesis than digital mammography in the baseline group, although the OR was not statistically significant (OR, 1.81; 95% CI, 0.61–5.43; p = 0.29). Discussion Mammographic screening has shown clinical effectiveness, but it has been criticized for false-positive examinations, leading to anxiety and additional testing, including additional mammographic studies, ultrasound, and un-

AJR:205, November 2015

Baseline Screening FFDM Versus Digital Breast Tomosynthesis TABLE 3: Logistic Regression Analysis and Odds Ratios (ORs) of Recall Among Baseline and Previously Screened Patients

Downloaded from www.ajronline.org by Univ of Penn Libr on 06/01/16 from IP address 170.212.0.64. Copyright ARRS. For personal use only; all rights reserved

Previously Screened Patients

Patients Undergoing Baseline Screening

Characteristic

OR (95% CI)

p

OR (95% CI)

Digital breast tomosynthesis vs digital mammography

0.84 (0.76–0.92)

< 0.001

0.74 (0.61–0.89)

p 0.002

Age group < 40 y

1.00 (Reference)

40–49 y

1.39 (0.94–2.06)

1.00 (Reference)

50–59 y

0.97 (0.66–1.44)

60–69 y

0.86 (0.58–1.28)

≥ 70 y

0.75 (0.50–1.13)

0.17

0.10

1.50 (1.03–2.17)

0.03

0.89

1.77 (1.18–2.64)

0.005

0.46

1.31 (0.83–2.07)

0.24

0.88 (0.44–1.74)

0.71

Race White

1.00 (Reference)

1.00 (Reference)

Black or African American

1.07 (0.97–1.19)

0.17

1.44 (1.15–1.82)

0.002

Other

0.92 (0.77–1.10)

0.37

1.20 (0.86–1.68)

0.28

BI-RADS breast density category 1

1.00 (Reference)

1.00 (Reference)

2

1.52 (1.26–1.85)

< 0.001

2.20 (1.53–3.17)

< 0.001

3

2.01 (1.64–2.46)

< 0.001

2.20 (1.49–3.25)

< 0.001

4

1.31 (0.87–1.97)

0.19

1.73 (0.82–3.63)

0.15

Note—Multivariate models were adjusted for breast density, age, race, and radiologist.

necessary biopsies. Critiques of digital breast tomosynthesis have ranged from a lack of access for some patients to concerns over early adoption and implementation without the collection of appropriate clinical effectiveness data to justify the increased cost [11]. In addition, digital breast tomosynthesis combined with digital mammography imaging has a higher radiation dose than does FFDM imaging alone, with the dose increasing according to the thickness of the breast [17]. Although increased radiation dose has historically been a significant concern, early data have shown that the implementation of synthetic 2D images, which are reconstructed from the tomosynthesis acquisition may supplant the need for the 2D exposures, significantly reducing the dose

of tomosynthesis imaging [18, 19]. Still, digital breast tomosynthesis implementation requires capital investment and results in increased time for radiologist interpretation [8] with little additional reimbursement. Health care resources are limited, and preventive care should optimally be personalized on the basis of evidence specific to each patient population [11]. There has been much discussion surrounding whether digital breast tomosynthesis screening might benefit specific patient populations more than others, and there has been strong interest in determining whether this screening technique might offer greater benefit in women without prior mammography studies for comparison. In this natural experiment involving more than 25,000 screening mammography exam-

inations, there was a statistically significant lower baseline screening recall rate for digital breast tomosynthesis compared with digital mammography (16.0% vs 20.5%; p = 0.002). There may be an even greater effect of screening with digital breast tomosynthesis in the baseline subgroup, with 45 fewer recalls/1000 women versus 13 fewer recalls/1000 women for previously screened patients. The decrease in recall with digital breast tomosynthesis implementation was more pronounced among women with less dense breasts in the baseline and previously screened populations. Although these findings are surprising, prospective data have not supported the original supposition that digital breast tomosynthesis would have greater benefit in women with dense breasts [4,

TABLE 4: Logistic Regression Analysis of Cancer Diagnosis Among Baseline and Previously Screened Patients Who Were Recalled Previously Screened Patients (n = 1933)

Patients Undergoing Baseline Screening (n = 545)

Characteristic

OR (95% CI)

p

OR (95% CI)

p

Digital breast tomosynthesis vs digital mammography

1.57 (1.05–2.33)

0.03

1.81 (0.61–5.43)

0.29

Age group < 40 y

1.00 (Reference)

40–49 y

0.53 (0.12–2.42)

1.00 (Reference) 0.41

0.44 (0.08–2.44)

0.35 0.67

50–59 y

0.76 (0.17–3.42)

0.72

0.67 (0.10–4.28)

≥ 60 y

1.81 (0.41–8.07)

0.43

1.12 (0.17–7.27)

0.91

1.26 (0.84–1.89)

0.27

1.74 (0.58–5.21)

0.32

BI-RADS density, categories 3–4 vs 1–2

Note—Multivariate models were adjusted for breast density, age, race, and radiologist.

AJR:205, November 2015 1147

Downloaded from www.ajronline.org by Univ of Penn Libr on 06/01/16 from IP address 170.212.0.64. Copyright ARRS. For personal use only; all rights reserved

6, 7]. Additional studies with a greater number of women with dense breasts may clarify the benefit in this subgroup. There was also a trend toward a significant incremental recall reduction with digital breast tomosynthesis use in patients younger than 50 years undergoing baseline screening, which might be confirmed with multicenter study analysis using a larger patient population. There are limitations to our study: the subgroup of patients undergoing baseline screening included some patients who had previously undergone mammography, but their images were not available; therefore, they are not true baseline patients. This was not a randomized trial, and therefore it is subject to bias. However, the conversion to tomosynthesis at our site was not incremental, and patients were not selected on the basis of perceived cancer risk or ability to pay for digital breast tomosynthesis. Rather, once implemented, digital breast tomosynthesis was performed for every screening patient. As an urban academic center, we did not advertise the change in our practice, limiting the possibility of attracting higher-risk patients. This reduces selection bias based on patient characteristics such as breast density, age, or family history of breast cancer. Furthermore, multivariate analysis showed that the risk in both populations was similar [6]. In addition, there was no change in radiology interpreters over the study time period. Although true-positive and false-positive examinations are documented, our false-negative rate is currently being determined with tristate cancer registry analysis. Finally, although a reduction in false-positive examinations may reduce harms from screening, this study was not designed to analyze a clinically significant morbidity or mortality benefit or cost savings from digital breast tomosynthesis implementation. Despite these limitations, to our knowledge, this is the largest baseline subgroup studied with digital breast tomosynthesis combined with digital mammography and represents an important step in determining whether digital breast tomosynthesis might be especially helpful in a subgroup of the screening population. This study shows that digital breast tomosynthesis screening significantly reduces the recall rate for patients undergoing baseline

screening, particularly in women younger than McDonald et al. 50 years, for whom the recall rates are typically high. Although the differences were not statistically significant in this population, likely because of the sample size, cancer detection and PPV 1 were higher for digital breast tomosynthesis than for digital mammography among patients undergoing baseline screening. Analysis of the entire screening cohort also found an additional benefit for digital breast tomosynthesis implementation in women younger than 50 years, with a statistically significant increase in cancer detection [6]. Therefore, if there are limited available resources, patients younger than 50 years who are undergoing baseline screening or who have no prior mammographic images available may benefit more than other women from screening with digital breast tomosynthesis. Larger prospective trials are needed to validate our findings in the baseline patient subgroup. References 1. Ciatto S, Houssami N, Bernardi D, et al. Integration of 3D digital mammography with tomosynthesis for population breast-cancer screening (STORM): a prospective comparison study. L ­ ancet Oncol 2013; 14:583–589 2. Durand MA, Haas BM, Yao X, et al. Early clinical experience with digital breast tomosynthesis for screening mammography. Radiology 2015; 274:85–92 3. Friedewald SM, Rafferty EA, Rose SL, et al. Breast cancer screening using tomosynthesis in combination with digital mammography. JAMA 2014; 311:2499–2507 4. Haas BM, Kalra V, Geisel J, Raghu M, Durand M, Philpotts LE. Comparison of tomosynthesis plus digital mammography and digital mammography alone for breast cancer screening. Radiology 2013; 269:694–700 5. Lourenco AP, Barry-Brooks M, Baird G, Tuttle A, Mainiero MB. Changes in recall type and patient treatment following implementation of screening digital breast tomosynthesis. Radiology 2015; 274:337–342 6. McCarthy AM, Kontos D, Synnestvedt M, et al. Screening outcomes following implementation of digital breast tomosynthesis in a general-population screening program. J Natl Cancer Inst 2014; 106:pii: dju316 7. Rose SL, Tidwell AL, Bujnoch LJ, Kushwaha AC, Nordmann AS, Sexton R Jr. Implementation of breast tomosynthesis in a routine screening practice: an ob-

servational study. AJR 2013; 200:1401–1408 8. Skaane P, Bandos AI, Gullien R, et al. Comparison of digital mammography alone and digital mammography plus tomosynthesis in a population-based screening program. Radiology 2013; 267:47–56 9. Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008; 336:924–926 10. Hardesty LA, Kreidler SM, Glueck DH. Digital breast tomosynthesis utilization in the United States: a survey of physician members of the Society of Breast Imaging. J Am Coll Radiol 2014; 11:594–599 11. Lee CI, Lehman CD. Digital breast tomosynthesis and the challenges of implementing an emerging breast cancer screening technology into clinical practice. J Am Coll Radiol 2013; 10:913–917 12. Tosteson AN, Fryback DG, Hammond CS, et al. Consequences of false-positive screening mammograms. JAMA Intern Med 2014; 174:954–961 13. U.S. Preventative Services Task Force Website. Screening for breast cancer. http://www.uspreventativeservicestaskforce.org/uspstf/uspsbrca. htm. Accessed September 29, 2015. 14. Schell MJ, Yankaskas BC, Ballard-Barbash R, et al. Evidence-based target recall rates for screening mammography. Radiology 2007; 243:681–689 15. Sickles E, D’Orsi CJ, Bassett LW, et al. ACR BI-RADS mammography. In: D’Orsi CJ, Sickles EA, Mendelson EB, et al., eds. ACR BI-RADS ­Atlas, Breast Imaging Reporting and Data System. Reston, VA: American College of Radiology, 2013 16. D’Orsi CJ, Bassett LW, Berg WA, et al. BI-RADS: mammography, 4th ed. In: D’Orsi CJ, Mendelson EB, Ikeda DM, et al., eds. Breast Imaging ­Reporting and Data System: ACR BI-RADS—breast imaging atlas. Reston, VA: American College of Radiology, 2003 17. Feng SS, Sechopoulos I. Clinical digital breast tomosynthesis system: dosimetric characterization. Radiology 2012; 263:35–42 18. Skaane P, Bandos AI, Eben EB, et al. Two-view digital breast tomosynthesis screening with synthetically reconstructed projection images: comparison with digital breast tomosynthesis with fullfield digital mammographic images. ­ Radiology 2014; 271:655–663 19. Zuley ML, Guo B, Catullo VJ, et al. Comparison of two-dimensional synthesized mammograms versus original digital mammograms alone and in combination with tomosynthesis images. ­Radiology 2014; 271:664–671

F O R YO U R I N F O R M AT I O N

This article is available for CME and Self-Assessment (SA-CME) credit that satisfies Part II requirements for maintenance of certification (MOC). To access the examination for this article, follow the prompts associated with the online version of the article.

1148

AJR:205, November 2015