Volume 129, Issue 16 p. 2456-2468
ORIGINAL ARTICLE
Open Access

Breast cancer risk characteristics of women undergoing whole-breast ultrasound screening versus mammography alone

Brian L. Sprague PhD

Corresponding Author

Brian L. Sprague PhD

Office of Health Promotion Research, Department of Surgery, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

Department of Radiology, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

University of Vermont Cancer Center, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

Correspondence

Brian L. Sprague, UHC Bldg Rm 4425, 1 S Prospect St, Burlington, VT 05405, USA.

Email: [email protected]

Search for more papers by this author
Laura Ichikawa MS

Laura Ichikawa MS

Kaiser Permanente Washington Health Research Institute, Seattle, Washington, USA

Search for more papers by this author
Joanna Eavey MSPH

Joanna Eavey MSPH

Kaiser Permanente Washington Health Research Institute, Seattle, Washington, USA

Search for more papers by this author
Kathryn P. Lowry MD

Kathryn P. Lowry MD

Department of Radiology, University of Washington and Seattle Cancer Care Alliance, Seattle, Washington, USA

Search for more papers by this author
Garth Rauscher PhD

Garth Rauscher PhD

Division of Epidemiology and Biostatistics, School of Public Health, University of Illinois at Chicago, Chicago, Illinois, USA

Search for more papers by this author
Ellen S. O’Meara PhD

Ellen S. O’Meara PhD

Kaiser Permanente Washington Health Research Institute, Seattle, Washington, USA

Search for more papers by this author
Diana L. Miglioretti PhD

Diana L. Miglioretti PhD

Kaiser Permanente Washington Health Research Institute, Seattle, Washington, USA

Division of Biostatistics, Department of Public Health Sciences, University of California Davis, Davis, California, USA

Search for more papers by this author
Shuai Chen PhD

Shuai Chen PhD

Division of Biostatistics, Department of Public Health Sciences, University of California Davis, Davis, California, USA

Search for more papers by this author
Janie M. Lee MD, MSc

Janie M. Lee MD, MSc

Department of Radiology, University of Washington and Seattle Cancer Care Alliance, Seattle, Washington, USA

Search for more papers by this author
Natasha K. Stout PhD

Natasha K. Stout PhD

Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, Massachusetts, USA

Search for more papers by this author
Jeanne S. Mandelblatt MD, MPH

Jeanne S. Mandelblatt MD, MPH

Department of Oncology, Georgetown University Medical Center, Georgetown University, Washington, District of Columbia, USA

Search for more papers by this author
Nila Alsheik MD

Nila Alsheik MD

Advocate Caldwell Breast Center, Advocate Lutheran General Hospital, Park Ridge, Illinois, USA

Search for more papers by this author
Sally D. Herschorn MD

Sally D. Herschorn MD

Department of Radiology, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

University of Vermont Cancer Center, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

Search for more papers by this author
Hannah Perry MD

Hannah Perry MD

Department of Radiology, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

University of Vermont Cancer Center, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

Search for more papers by this author
Donald L. Weaver MD

Donald L. Weaver MD

University of Vermont Cancer Center, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

Department of Pathology & Laboratory Medicine, University of Vermont Larner College of Medicine, Burlington, Vermont, USA

Search for more papers by this author
Karla Kerlikowske MD

Karla Kerlikowske MD

Departments of Medicine and Epidemiology and Biostatistics, University of California, San Francisco, California, USA

General Internal Medicine Section, Department of Veterans Affairs, University of California, San Francisco, California, USA

Search for more papers by this author
First published: 12 June 2023
Citations: 1

Abstract

Background

There are no consensus guidelines for supplemental breast cancer screening with whole-breast ultrasound. However, criteria for women at high risk of mammography screening failures (interval invasive cancer or advanced cancer) have been identified. Mammography screening failure risk was evaluated among women undergoing supplemental ultrasound screening in clinical practice compared with women undergoing mammography alone.

Methods

A total of 38,166 screening ultrasounds and 825,360 screening mammograms without supplemental screening were identified during 2014–2020 within three Breast Cancer Surveillance Consortium (BCSC) registries. Risk of interval invasive cancer and advanced cancer were determined using BCSC prediction models. High interval invasive breast cancer risk was defined as heterogeneously dense breasts and BCSC 5-year breast cancer risk ≥2.5% or extremely dense breasts and BCSC 5-year breast cancer risk ≥1.67%. Intermediate/high advanced cancer risk was defined as BCSC 6-year advanced breast cancer risk ≥0.38%.

Results

A total of 95.3% of 38,166 ultrasounds were among women with heterogeneously or extremely dense breasts, compared with 41.8% of 825,360 screening mammograms without supplemental screening (p < .0001). Among women with dense breasts, high interval invasive breast cancer risk was prevalent in 23.7% of screening ultrasounds compared with 18.5% of screening mammograms without supplemental imaging (adjusted odds ratio, 1.35; 95% CI, 1.30–1.39); intermediate/high advanced cancer risk was prevalent in 32.0% of screening ultrasounds versus 30.5% of screening mammograms without supplemental screening (adjusted odds ratio, 0.91; 95% CI, 0.89–0.94).

Conclusions

Ultrasound screening was highly targeted to women with dense breasts, but only a modest proportion were at high mammography screening failure risk. A clinically significant proportion of women undergoing mammography screening alone were at high mammography screening failure risk.

INTRODUCTION

Several studies have reported increased use of supplemental ultrasound screening following the implementation of US state laws mandating breast density notification to women undergoing screening mammography.1-5 A proposed amendment to the US Mammography Quality Standards Act to require density notification nationally has further increased attention to limitations of mammography in women with dense breasts (heterogeneously dense or extremely dense) and the potential role for supplemental screening to increase early detection of breast cancer.6, 7 However, there remain no consensus guidelines in the United States regarding which women undergoing screening mammography should receive supplemental ultrasound screening,8-10 and uncertainty persists among women, primary care providers, and radiologists.8, 11-16

Supplemental ultrasound screening has the potential to detect cancers missed by mammography, although this gain is accompanied by potential harms, including recall for additional imaging and benign breast biopsy among women who do not have breast cancer (false positives).8 The balance of benefits and harms may potentially be improved by targeting ultrasound screening to women at high risk of mammography screening failure, defined as a symptomatic interval invasive cancer or an advanced breast cancer diagnosis after mammography screening.17, 18 In addition to breast density, other breast cancer risk factors (e.g., family history of breast cancer, prior benign breast disease diagnosis, obesity) have been shown to be associated with likelihood of a mammography screening failure.19 In the current clinical setting of variable referrals for ultrasound screening, it is unclear how closely use of ultrasound screening is associated with a woman's risk of mammography screening failures.

We sought to evaluate the breast cancer risk characteristics of women undergoing ultrasound screening in a large sample of breast imaging facilities within three regional registries of the Breast Cancer Surveillance Consortium (BCSC). We contrasted the examination-level risk characteristics of women undergoing ultrasound screening to those of women undergoing mammography screening alone (no supplemental screening with ultrasound or magnetic resonance imaging [MRI]) at the same facilities and characterized risks of interval invasive cancer and advanced cancer. The results inform understanding of ultrasound screening practice patterns in the United States and provide evidence regarding the extent to which ultrasound screening is used by women at high risk of mammography screening failures.

METHODS

Study setting and design

We analyzed observational clinical data from the Metro Chicago Breast Cancer Registry (2014–2018), San Francisco Mammography Registry (2014–2020), and Vermont Breast Cancer Surveillance System (2014–2020) within the BCSC (https://www.bcsc-research.org/).20 These three registries were chosen because they include a number of facilities performing screening ultrasound, whereas screening ultrasound remains rare within other BCSC registries. Each registry prospectively collected clinical breast imaging data from participating facilities within its catchment area. The registries and a central Statistical Coordinating Center received institutional review board approval to enroll participants, link data, and perform analyses. All procedures were Health Insurance Portability and Accountability Act–compliant and registries and the Statistical Coordinating Center received a Federal Certificate of Confidentiality and other protections for the identities of women, physicians, and facilities.

Study population

This study included screening ultrasound and screening mammography examinations performed at 32 breast imaging facilities with at least 10 screening ultrasounds during the study period, 2014–2020 (40 facilities at the three participating BCSC registries had no screening ultrasounds or fewer than 10 screening ultrasounds and were not included). Examinations among adult women aged ≥18 years with no personal history of breast cancer were eligible for inclusion. Eligible screening ultrasound examinations were included regardless of mammography use before or after the ultrasound examination. As a comparison group, screening mammograms performed in women with no evidence of supplemental screening (ultrasound or MRI) within the next 12 months were identified from the same facilities and years.

Data collection

Participating breast imaging facilities provided type of breast imaging modality (digital mammography, tomosynthesis, ultrasound, MRI), examination indication (screening vs. diagnostic), and mammographic breast density data to BCSC registries using standard nomenclature from the Breast Imaging Reporting and Data System (BI-RADS).21 Breast density was recorded by radiologists during clinical interpretation of mammography using four categories: almost entirely fatty, scattered fibroglandular densities, heterogeneously dense, or extremely dense.21 For screening ultrasounds, breast density was taken from the closest mammogram within 18 months (if available; otherwise, breast density was defined as missing). Demographic, risk factor, and health history information was self-reported by women at the time of breast imaging or extracted from electronic medical records. Previous benign breast biopsy results were abstracted from clinical pathology reports. For the purposes of determining personal history of breast cancer (an exclusion criteria), we used clinical records and breast cancer diagnosis data ascertained by BCSC registries' linkage of women's breast imaging records to pathology databases; regional Surveillance, Epidemiology, and End Results programs; and state tumor registries, as previously described.20

Key measures and definitions

Women's characteristics were defined at each breast imaging examination. Women with heterogeneously or extremely dense breasts on mammography were considered to have dense breasts, consistent with conventional clinical and regulatory definitions.22, 23 We grouped previous benign diagnoses based on risk of developing subsequent breast cancer using published taxonomy24-27 (lobular carcinoma in situ [LCIS] > atypical hyperplasia > proliferative without atypia > nonproliferative) or as unknown if a woman reported a previous biopsy with no available BCSC pathology result. First-degree family history of breast cancer included any breast cancer diagnosis in a mother, sister, or daughter. Body mass index was categorized based on height and weight. Postmenopausal women were those with both ovaries removed, whose periods had stopped naturally, aged 60 or older, current users of postmenopausal hormone therapy, or last menstrual period was more than a year ago. Premenopausal women reported a period within the past 180 days or birth control hormone use. If menopausal status data were missing, women were assumed to be postmenopausal if aged 52 years or older and premenopausal otherwise.

Risk for invasive breast cancer and interval invasive breast cancer was determined using the BCSC 5-year risk model, version 2.0, which is applicable to women aged 35 to 74 years with no history of ductal carcinoma in situ or invasive breast cancer.27 Five-year invasive breast cancer risk is based on age, race and ethnicity, first-degree family history of breast cancer, history of benign breast disease, and breast density.27 Using previously developed definitions,17 we classified women as having high interval invasive breast cancer risk after mammography if they had heterogeneously dense breasts and BCSC 5-year invasive breast cancer risk ≥2.5% or extremely dense breasts and BCSC 5-year invasive breast cancer risk ≥1.67%. Women in these groups have interval invasive cancer rates exceeding one per 1000 examinations after mammography screening.17

We estimated advanced breast cancer risk (defined as prognostic pathologic stage II or higher)28 using the BCSC 6-year advanced breast cancer risk model (https://tools.bcsc-scc.org/AdvBC6yearRisk/#/), which is applicable to women aged 40 to 74 years with no history of LCIS, ductal carcinoma in situ, or invasive breast cancer.29 The model is based on age, race and ethnicity, menopausal status, body mass index, first-degree family history of breast cancer, history of benign breast disease, breast density, and mammography screening interval.29 For all women, we estimated advanced cancer risk under an annual mammography screening regimen in the absence of supplemental screening. We classified women's advanced cancer risk according to previously defined thresholds for low/average (<0.38%) and intermediate/high advanced cancer risk (≥0.38%).29

Statistical analyses

Screening breast imaging examinations were the unit of analysis. We described examination-level demographics and breast cancer risk characteristics of women undergoing ultrasound screening examinations and compared them with those of women undergoing screening mammography alone. We tested the association of demographic and risk characteristics with receipt of screening ultrasound using logistic regression adjusted for BCSC registry and estimated by generalized estimating equations with a working independence correlation structure to account for clustering of multiple examinations per woman. Secondary analyses were restricted to examinations among women with dense breasts. All analyses were conducted using SAS, version 9.4 (SAS Institute, Cary, North Carolina).

RESULTS

The study included 38,166 ultrasound screening examinations among 29,112 women and 825,360 mammography screening examinations among 377,140 women, yielding a ratio of one ultrasound screening examination for every 22 mammography alone screening episodes. Most ultrasound screening examinations (75%) were the woman's first ultrasound screen and 63% of screening ultrasounds occurred within 9 months following a mammogram. Two percent of ultrasound screens occurred among women who had a previous MRI screening examination recorded in the BCSC database, compared with 0.4% among the mammography-alone group. Approximately 69% of ultrasound screening examinations occurred among women aged 40 to 59 years and 25.7% were among women who identified as Asian, Black, or Hispanic (Table 1). A total of 95.3% of ultrasound screening examinations were among women with dense breasts, whereas 21.7% occurred among women with a first-degree family history of breast cancer, 24.3% had a prior benign breast disease diagnosis, and 43.7% were among women who were overweight or obese. There were statistically significant differences in the distribution of demographic and risk characteristics between ultrasound screening examinations and mammography screening-alone examinations (Table 1; p < .0001). In comparison to mammography, ultrasound screening examinations were much more likely to occur in women who were younger than aged 50 years, premenopausal, had dense breasts, and normal body mass index. Small differences were observed in race/ethnicity, family history of breast cancer, and personal history of benign breast disease.

TABLE 1. Examination-level demographic and risk characteristics for women undergoing ultrasound screening examinations or mammography screening alone at 32 breast imaging facilities participating in the Vermont, San Francisco, and Chicago Breast Cancer Surveillance Consortium registries, 2014–2020.
Ultrasound screening (N = 38,166 examinations) Mammography screening alone (N = 825,360 examinations)
n %a n %a p b
Age, years <.0001
<40 1056 2.8 8083 1.0
40–49 13,581 35.6 183,539 22.2
50–59 12,611 33.0 259,759 31.5
60–69 7631 20.0 235,392 28.5
70+ 3287 8.6 138,587 16.8
Race/ethnicityc <.0001
Asian 2573 6.9 109,367 13.7
Black 2978 8.0 86,652 10.9
Hispanic 3981 10.7 54,776 6.9
White 26,820 72.4 530,805 66.5
Other/multiple 685 1.8 16,039 2.0
Unknown 1129 (3.0) 27,721 (3.4)
Menopausal status <.0001
Premenopausal 18,720 49. 250,456 30.3
Postmenopausal 19,446 51.0% 574,904 69.7
BI-RADS breast density <.0001
Almost entirely fat 119 0.3 81,942 10.0
Scattered fibroglandular densities 1647 4. 395,131 48.2
Heterogeneously dense 29,119 77.9 294,442 35.9
Extremely dense 6497 17.4 47,901 5.8
Unknown 784 (2.1) 5944 (0.7)
First-degree family history of breast cancer <.0001
No 29,785 78.3 674,185 82.6
Yes 8249 21.7 142,338 17.4
Unknown 132 (0.3) 8837 (1.1)
History of benign breast disease <.0001
None 28,882 75.7 654,132 79.3
Prior biopsy, diagnosis unknown 4762 12.5 98,506 11.9
Nonproliferative lesion 3099 8.1 49,613 6.0
Proliferative changes without atypia 1036 2.7 18,606 2.3
Proliferative changes with atypia 329 0.9 3853 0.5
Lobular carcinoma in situ 58 0.2 650 0.1
Body mass index (kg/m2) <.0001
Underweight (<18.5) 890 2.5 10,468 1.4
Normal (18.5–<25) 19,484 53.9 273,983 36.7
Overweight (25–<30) 10,120 28.0 225,676 30.2
Obese (30+) 5682 15.7 236,753 31.7
Missing 1990 (5.2) 78,480 (9.5)
  • Abbreviation: BI-RADS, Breast Imaging Reporting and Data System.
  • a Percentages are among nonmissing; percent missing is shown in parentheses.
  • b Comparing the distribution of the risk characteristic among women undergoing ultrasound screening vs. mammography screening, from a logistic regression model using generalized estimating equation with an independent correlation structure to account for multiple examinations per woman.
  • c All race/ethnicity groups except Hispanic are non-Hispanic.

In analyses restricted to examinations among women with dense breasts, there were 35,616 ultrasound screening examinations and 342,343 mammography-alone screening episodes, corresponding to a ratio of 1:10 (Table 2). In this subset of examinations, ultrasound screening was more likely to occur in White women, and slightly more likely to occur among women who were younger than 50 years, premenopausal, had extremely dense breasts, a family history of breast cancer, and a history of benign breast disease. The distribution of body mass index was very similar across groups.

TABLE 2. Examination-level demographic and risk characteristics for women with dense breasts undergoing ultrasound screening examinations or mammography screening alone at 32 breast imaging facilities participating in the Vermont, San Francisco, and Chicago Breast Cancer Surveillance Consortium registries, 2014–2020.
Ultrasound screening (N = 35,616 examinations) Mammography screening alone (N = 342,343 examinations)
n %a n %a p b
Age, years <.0001
<40 839 2.4 5285 1.5
40-49 12,887 36.2 110,495 32.3
50-59 11,696 32.8 112,881 33.0
60-69 7103 19.9 75,861 22.2
70+ 3091 8.7 37,821 11.0
Race/ethnicityc <.0001
Asian 2453 7.1 66,012 20.1
Black 2843 8.2 28,817 8.8
Hispanic 3823 11.0 22,493 6.8
White 24,855 71.8 204,319 62.2
Other/multiple 626 1.8% 6810 2.1
Unknown 1016 (2.9) 13,892 (4.1)
Menopausal status <.0001
Premenopausal 17,489 49.1 146,709 42.9
Postmenopausal 18,127 50.9 195,634 57.1
BI-RADS breast density <.0001
Heterogeneously dense 29,119 81. 294,442 86.0
Extremely dense 6497 18.2 47,901 14.0
First-degree family history of breast cancer <.0001
No 27,871 78.5 280,834 83.2
Yes 7646 21.5 56,846 16.8
Unknown 99 (0.3) 4663 (1.4)
History of benign breast disease <.0001
None 26,866 75.4 266,371 77.8
Prior biopsy, diagnosis unknown 4513 12.7 43,391 12.7
Nonproliferative lesion 2922 8.2 21,937 6.4
Proliferative changes without atypia 968 2.7 8494 2.5
Proliferative changes with atypia 296 0.8 1794 0.5
Lobular carcinoma in situ 51 0.1 356 0.1
Body mass index (kg/m2) .07
Underweight (<18.5) 854 2.5 7986 2.6
Normal (18.5–<25) 18,636 54.8 165,914 54.0
Overweight (25–<30) 9457 27.8 85,831 28.0
Obese (30+) 5076 14.9 47,257 15.4
Missing 1593 (4.5) 35,355 (10.3)
  • Abbreviation: BI-RADS, Breast Imaging Reporting and Data System.
  • a Percentages are among nonmissing; percent missing is shown in parentheses.
  • b Comparing the distribution of the risk characteristic among women undergoing ultrasound screening vs. mammography screening, from a logistic regression model using generalized estimating equation with an independent correlation structure to account for multiple examinations per woman.
  • c All race/ethnicity groups except Hispanic are non-Hispanic.

Among screening ultrasound examinations in women aged 35 to 74 years, 27.2% occurred in women with intermediate BCSC 5-year invasive breast cancer risk (1.67%–2.49%) and 18.1% occurred among women with high or very high BCSC 5-year invasive breast cancer risk (≥2.50%) (Figure 1A). In the comparator group of mammography-alone examinations, 19.6% occurred in women with intermediate BCSC 5-year invasive breast cancer risk and 9.5% occurred among women with high or very high BCSC 5-year invasive breast cancer risk. In analyses restricted to examinations among women aged 35 to 74 years with dense breasts, 46.4% of screening ultrasound examinations occurred in women with intermediate or higher BCSC 5-year invasive breast cancer risk, whereas 40.7% of mammography alone screening examinations occurred in women with intermediate or higher BCSC 5-year invasive breast cancer risk (Figure 1B).

Details are in the caption following the image

Distribution of BCSC 5-year invasive breast cancer risk. Among (A) all women aged 35–74 years and (B) women aged 35–74 years with dense breasts undergoing ultrasound screening examinations or mammography screening alone at 32 breast imaging facilities participating in the Vermont, San Francisco, and Chicago BCSC registries, 2014–2020. The BCSC 5-year invasive breast cancer risk model is not applicable for women younger than age 35 or older than age 74 years. Among women aged 35–74 years, invasive breast cancer risk could not be calculated because missing breast density for 1.8% of eligible ultrasound screening examinations and 0.7% of eligible mammography screening examinations. BCSC indicates Breast Cancer Surveillance Consortium.

The joint distribution of breast density and BCSC 5-year invasive breast cancer risk across screening ultrasound and mammography screening examinations is shown in Figure 2. A total of 22.6% of screening ultrasounds were performed in women at high risk of interval invasive cancer after mammography because of having either heterogeneously dense breasts and BCSC 5-year invasive breast cancer risk of ≥2.5% or extremely dense breasts and BCSC 5-year invasive risk ≥1.67%. Among the mammography-alone comparator group, 8.0% of examinations were in women at high interval invasive cancer risk (adjusted odds ratio [aOR], 3.28; 95% CI, 3.17–3.39 for high interval invasive cancer risk, ultrasound vs. mammography). Among women with dense breasts, 23.7% of screening ultrasounds were performed in women at high risk of interval invasive cancer, whereas 18.5% of mammography-alone examinations were in women at high risk of interval invasive cancer (aOR,1.35; 95% CI, 1.30–1.39).

Details are in the caption following the image

Joint distribution of breast density and BCSC 5-year invasive breast cancer risk. Among women aged 35–74 years undergoing (A) ultrasound screening or (B) mammography screening alone at 32 breast imaging facilities participating in the Vermont, San Francisco, and Chicago BCSC registries, 2014–2020. Diagonally striped bars represent groups with high interval invasive cancer risk based on breast density and estimated BCSC 5-year invasive breast cancer risk.17 The BCSC 5-year invasive breast cancer risk model is not applicable for women younger than age 35 or older than age 74 years. Among women aged 35–74 years, breast density and invasive breast cancer risk were missing for 1.8% of eligible ultrasound screening examinations and 0.7% of eligible mammography screening examinations. BCSC indicates Breast Cancer Surveillance Consortium.

Figure 3A illustrates that 30.7% of ultrasound screens among women aged 40 to 74 years occurred in women with intermediate or high BCSC 6-year advanced breast cancer risk (≥0.38%). Among the mammography-alone comparator group, 18.6% of examinations occurred in women with intermediate or higher advanced breast cancer risk (aOR, 1.60; 95% CI, 1.55–1.64 for intermediate or higher advanced cancer risk, ultrasound vs. mammography). In analyses restricted to examinations among women aged 40 to 74 years with dense breasts, 32.0% of screening ultrasound examinations occurred in women with intermediate or high BCSC 6-year advanced breast cancer risk, whereas 30.5% of mammography screening–alone examinations occurred in women with intermediate or high BCSC 6-year advanced breast cancer risk (aOR, 0.91; 95% CI, 0.89–0.94; Figure 3B).

Details are in the caption following the image

Distribution of BCSC 6-year advanced-stage breast cancer risk. Among (A) all women aged 40–74 and (B) women aged 40–74 years with dense breasts undergoing ultrasound screening examinations and mammography screening examinations at 32 breast imaging facilities participating in the Vermont, San Francisco, and Chicago BCSC registries, 2014–2020. The BCSC advanced cancer risk model is not applicable for women younger than age 40 or older than age 74 years. Among women aged 40–74 years, advanced cancer risk could not be calculated due to missing body mass index or breast density or prior LCIS for 6.2% of ultrasound screening examinations (4.6% for women with dense breasts and 9.4% of mammography screening examinations (10.3% for women with dense breasts). BCSC indicates Breast Cancer Surveillance Consortium; LCIS, lobular carcinoma in situ.

DISCUSSION

Our results from a geographically diverse sample of breast imaging facilities in the United States demonstrate that ultrasound screening was predominantly used by women with dense breasts. Other breast cancer risk factors were also more common, and risk of mammography failures (interval invasive breast cancer and advanced cancer) was higher among ultrasound screening examinations compared with mammography screening alone. Analyses restricted to examinations among women with dense breasts indicated only modest differences in risk of interval or advanced cancer between the ultrasound- and mammography-alone groups. Overall, our findings indicate strong selection of women for ultrasound screening based on breast density alone and moderate selection based on other breast cancer risk factors, corresponding to a wide distribution in risk of mammography screening failure among women undergoing breast ultrasound screening.

The moderate differences in the risk distributions for invasive breast cancer, interval cancer, and advanced breast cancer observed in the full study population between ultrasound screening examinations versus mammography alone examinations narrowed substantially when restricted to examinations among women with dense breasts. Most notably, the prevalence of intermediate or high 6-year advanced breast cancer risk among women with dense breasts was higher among mammography alone–screening examinations compared with supplemental ultrasound screening examinations after adjusting for BCSC registry. This reinforces our conclusion that, aside from breast density, other factors associated with mammography failure risk such as obesity29 are not commonly used to select women for supplemental ultrasound screening.

Our results also demonstrate that a clinically significant proportion of women at high risk of advanced cancer underwent mammography screening alone with no supplemental screening. These findings suggest a potential role for risk assessment at the time of mammography screening or during primary care visits when screening strategies are discussed.

Although consensus guidelines for supplemental ultrasound screening do not exist in the United States, the American College of Radiology's Expert Panel on Breast Imaging recently concluded that ultrasound screening may be appropriate for high-risk women with nondense breasts (almost entirely fatty and scattered fibroglandular densities) and intermediate- or high-risk women with dense breasts, whereas supplemental ultrasound screening is usually not appropriate for average or intermediate risk women with nondense breasts.10 The panel disagreed as to whether ultrasound screening may be appropriate for average-risk women with dense breasts. In our study, mammography failure risk varied widely among women undergoing breast ultrasound screening. Approximately one-half of ultrasound screening examinations occurred among women with dense breasts and low or average breast cancer risk who are not at high risk of screening failures. One-fifth of women undergoing ultrasound screening in our study were classified as having high interval invasive breast cancer risk with mammography based on their breast density and BCSC 5-year risk.17 Approximately one in three women undergoing ultrasound screening had intermediate or higher advanced cancer risk. These observations underscore the need for additional research to determine in what population ultrasound screening reduces the risk of screening failures.

Ultrasound screening was relatively rare at the participating breast imaging facilities (one per 22 mammography alone screening episodes and one per 10 mammography alone examinations among women with dense breasts). Almost 20% of mammography-alone screening examinations (corresponding to more than 126,000 screening episodes) occurred among women with intermediate or high advanced cancer risk, exceeding by far the total number of ultrasound screening episodes (38,166) and demonstrating that a large group of women at high advanced cancer risk did not undergo supplemental ultrasound or MRI screening. Efforts to increase supplemental ultrasound or MRI screening use among women at high risk of mammography failure may improve breast cancer screening outcomes.17, 18, 29 Numerous studies have demonstrated that the addition of supplemental ultrasound to mammography increases cancer detection rates,8, 30 and evidence is also emerging that supplemental ultrasound screening is associated with reduced interval cancer rates compared with mammography alone.31, 32

Although breast density notification laws have focused attention on the limitations of mammography for women with dense breasts, a growing amount of literature has highlighted the importance of additional risk factors in contributing to risk of mammography screening failures.17, 18, 29 Laws and policies that mandate breast density notification without consideration of other risk factors may be inadequate to direct supplemental screening efforts to women at risk of mammography failures. Tools for estimating breast cancer risk and risk of mammography screening failure are available, although consensus on which measures are most appropriate has not been established and further research on implementation strategies in clinical practice will be needed. The BCSC 5-year breast cancer risk model v2.0 calculator is freely available online33 and can be combined with breast density to identify women at elevated risk of mammography interval invasive breast cancer as done in this study.17 The BCSC 6-year advanced cancer risk model was developed to directly estimate advanced cancer risk as a function of age, race/ethnicity, menopausal status, screening interval, BMI, first-degree family history of breast cancer, history of benign breast disease, and breast density.29 To facilitate use, the risk model publication29 includes look-up tables showing advanced cancer risk for combinations of risk factors and is available as an online calculator.

Our results are consistent with previous work showing that ultrasound screening utilization is associated with age, breast density, and family history of breast cancer.4 In the previous BCSC analysis of ultrasound screening in the BCSC during 2006–2013, 74% of ultrasound screening examinations occurred among women with dense breasts and 46% occurred among women with intermediate or higher BCSC 5-year invasive cancer risk.34 Although BI-RADS guidelines for breast density assessment changed in 2013,21, 35 our previous work indicates that density assessment in clinical practice remained consistent within the BCSC36 and thus is unlikely to explain the change in breast density prevalence between the previous BCSC ultrasound study and the current study. Although our current study indicates that ultrasound screening during 2014–2020 was more narrowly targeted to women with dense breasts, the distribution of breast cancer risk among women undergoing ultrasound remained similar between the two studies. Our analyses provide new evidence regarding the distribution of mammography failure risk among women undergoing ultrasound screening.

Interpretation of our results must be tempered by the limited data on ultrasound screening performance according to risk characteristics and uncertainty about the appropriateness of ultrasound screening in population subgroups.10 Although previous work has characterized women at high risk of mammography screening failures,17, 18, 29 evidence on the impact of supplemental ultrasound screening on breast cancer outcomes among women at high mammography screening failure risk is not yet available. Ultrasound screening is not without potential harms in the form of false positives. A previous BCSC study estimated that supplemental ultrasound screening was associated with an approximate doubling of the false-positive biopsy recommendation rate compared with mammography alone.34 Further study is needed to characterize ultrasound screening performance and long-term outcomes according to risk of mammography screening failure. In the current absence of consensus recommendations, the decision to undergo supplemental ultrasound screening should involve a consideration of both potential benefits and harms, personal preferences, and values.11

Strengths of our study include a geographically and racially diverse sample of breast imaging facilities from US community practice, prospectively collected breast imaging and risk factor data via three BCSC registries, and the use of established BCSC risk prediction models. Notably, the three registries were selected because they included several breast imaging facilities offering screening ultrasound; thus, the rate of ultrasound screening is higher than would be observed across all BCSC registries. We did not examine the impact of density notification laws in this study, which were enacted in 2013, 2017, and 2019 for California, Vermont, and Illinois, respectively (Illinois mandated insurance coverage of supplemental screening for women with dense breasts beginning in 2009). However, our study characterizes utilization of screening ultrasound during a contemporary era of increased breast density awareness following national patient advocacy efforts,37-40 which undoubtedly contributed to the observed enrichment of ultrasound screening examinations among women with dense breasts. The BCSC risk models have moderate discrimination, they do not consider potentially useful predictors such as family history of breast cancer in second-degree relatives, genetic polymorphisms, or quantitative image-based mammographic features, and it is unclear how commonly they are used in clinical practice. However, the BCSC invasive breast cancer risk model is well calibrated in comparison to other models and externally validated in three cohorts.41-43 To our knowledge, the BCSC advanced cancer risk model is the only available risk model for advanced breast cancer. Our study does not directly address the degree to which ultrasound screening may be targeted based on other breast cancer risk models commonly used in clinical practice. Finally, we were unable to examine the frequency of high-penetrance genetic mutations (e.g., BRCA1/2). However, the modest frequency of first-degree family history among women undergoing ultrasound screening suggests that these were unlikely to be common, and supplemental MRI screening is recommended for BRCA mutation carriers.44 Further study is needed of sociodemographic predictors of ultrasound screening, barriers to ultrasound screening access, and implementation strategies for delivery of information on risk of breast cancer and mammography failures.

In summary, we found that ultrasound screening in this geographically diverse multisite study was strongly targeted to women with dense breasts. The distributions of breast cancer risk, interval invasive breast cancer risk, and advanced cancer risk varied widely among women undergoing ultrasound screening. Many women at high risk of screening mammography failure did not undergo supplemental screening following mammography. Consideration and further public awareness of other breast cancer risk factors beyond breast density could facilitate identification of women at high risk of mammography screening failures who may be appropriate for supplemental ultrasound screening.

AUTHOR CONTRIBUTIONS

Brian L. Sprague: Conceptualization, data curation, funding acquisition, investigation, and writing - original draft. Laura Ichikawa: Conceptualization, data curation, formal analysis, investigation, methodology, and writing - review and editing. Joanna Eavey: Conceptualization, data curation, formal analysis, investigation, methodology, and writing - review and editing. Kathryn P. Lowry: Conceptualization, investigation, and writing - review and editing. Garth Rauscher: Data curation, funding acquisition, investigation, and writing - review and editing. Ellen S. O’Meara: Data curation, investigation, and writing - review and editing. Diana L. Miglioretti: Data curation, funding acquisition, investigation, and writing - review and editing. Shuai Chen: Investigation, methodology, and writing - review and editing. Janie M. Lee: Investigation and writing - review and editing. Natasha K. Stout: Funding acquisition, investigation, and writing - review and editing. Jeanne S. Mandelblatt: Funding acquisition, investigation, and writing - review and editing. Nila Alsheik: Data curation, investigation, and writing - review and editing. Sally D. Herschorn: Investigation and writing - review and editing. Hannah Perry: Investigation and writing - review and editing. Donald L. Weaver: Investigation and writing - review and editing. Karla Kerlikowske: Conceptualization, data curation, funding acquisition, investigation, and writing - review and editing.

ACKNOWLEDGMENTS

We thank the participating women, mammography facilities, and radiologists for the data they have provided for this study. More can be learned about the BCSC at: http://www.bcsc-research.org/. This work was supported by the National Cancer Institute (R01CA248068). Data collection for this research was additionally supported by the Breast Cancer Surveillance Consortium with funding from the National Cancer Institute (P01CA154292 and U54CA163303), a Patient-Centered Outcomes Research Institute (PCORI) Program Award (PCS-1504-30370), and the Agency for Healthcare Research and Quality (R01 HS018366-01A1). Brian L. Sprague's effort was also supported by the National Institute of General Medical Sciences (P20GM103644). Jeanne S. Mandelblatt's effort was also supported by the National Cancer Institute (U01CA199218 and R35CA197289). The collection of cancer and vital status data was supported in part by several state public health departments and cancer registries throughout the United States (https://www.bcsc-research.org/work/acknowledgement). The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The statements presented in this work are solely the responsibility of the authors and do not necessarily represent the official views of PCORI, its Board of Governors or Methodology Committee, the National Cancer Institute, or the National Institutes of Health.

    CONFLICTS OF INTEREST STATEMENT

    Kathryn P. Lowry receives grant funding from GE Healthcare outside the submitted work. Diana L. Miglioretti received textbook royalties from Elsevier. Janie M. Lee receives research grant funding from GE Healthcare outside the submitted work. Karla Kerlikowske is an unpaid consultant with Grail Inc. for the STRIVE study. No other conflict of interest disclosures were reported.