Blog

We are so proud of our Chief Scientific Advisor, Dr. Wendie Berg, who was awarded the SBI Gold Medal at this year’s Society of Breast Imaging Annual Meeting — the organization’s highest honor. We are honored to share her remarks in full and congratulate her on an incredible career.

 

---

Receiving the SBI Gold Medal is both very gratifying and very humbling, and the words “gratifying” and “humbling” really sum up a career in breast imaging.  We are blessed by being able to truly make a difference in the lives of our patients every day, and mostly for the better.  I have been super fortunate to have had some impact on the field more broadly, and that gives me a deep sense of meaning.  But we all know that there is much still to be done.  I want to share some thoughts about a good friend of mine.  Susanne was a very successful research geneticist, very involved in her community, public health, and also in supporting political candidates who were interested in public health, communities, and scientific research.  She had dense breasts and knew to get screening MRI starting in 2007.  In 2009, the MRI showed what proved to be a 9-mm invasive lobular carcinoma.  Her nodes were negative and she had bilateral mastectomy.  She took aromatase inhibitors for ten years.  Still, despite all the favorable prognostic factors, in 2023, she was having abdominal discomfort, and was found to have carcinomatosis, and, soon thereafter, orbital metastases.  She signed up for many clinical trials and offered her tissue to researchers, recognizing that she was unlikely to personally benefit, but she felt it might help others in the future.  Sadly, Susanne passed almost exactly a year ago.  We have so much more to learn, and that sentiment is our shared legacy.

Today is an opportunity for me to publicly thank the research organizations, including NIH, the former Avon Foundation, the Breast Cancer Research Foundation, the Pennsylvania Breast Cancer Coalition, Komen Foundation, the Shear Family Foundation, and the many companies who have supported my research and other efforts.  I particularly want to thank the many, many national and international individuals who have helped me and supported me in this endeavor, and you know who you are, but especially Etta Pisano, Ellen Mendelson, Phil Evans, Daniel Lehrer, the many other investigators in ACRIN 6666 (which I just heard referred to as the “demon study”), and those behind the scenes (including Carl D’Orsi and the late Steve Feig).  Thank you to all those involved in DenseBreast-info.org, especially advocate and honorary SBI Fellow JoAnn Pushkin, and to the investigators in the trials comparing positron emission mammography and MRI, screening ultrasound and tomosynthesis, ultrasound AI, and my colleagues at the University of Pittsburgh now helping in studies of CEM.  I especially want to recognize the many women who have given their time and trust to participate in the many research studies I have led in the hopes that they will help others in the future. And today, I can say with confidence, that they have.  

Finally, I want to close with a bit more about Susanne.  She was a devoted and accomplished gardener, and one of her favorite sayings was “Friends are the flowers in the garden of life.”  I am blessed to have had many blooms in a rich garden of friends at the SBI.  We have accomplished so much together, and we have much more growing to do.  Thank you all.

– Dr. Wendie Berg, DenseBreast-info.org Chief Scientific Advisor

New Survey: Supplemental Screening ABUS Better Tolerated than MRI or CEM

DBI European Educator for Albania, and attending radiologist at the University of Cambridge and Cambridge University Hospitals, Dr. Iris Allajbeu, and colleagues, published a survey of 159 BRAID trial participants assessing patient acceptance, experience, and feedback for different methods of supplemental screening in women with dense breasts and negative mammography. Each of automated breast ultrasound (ABUS), abbreviated MRI (AB-MRI) and contrast-enhanced mammography (CEM) were generally well accepted by patients.

 

• When compared to standard mammography, nearly all participants rated ABUS as the same or better; nearly 1 in 6 women rated AB-MRI as worse and 1 in 10 rated CEM as worse.
• Mild or severe discomfort was common with ABUS and CEM.¹
• Anxiety before or during the imaging was least common with ABUS.
• AB-MRI was well-received, with several women highlighting benefits, such as reduced pain due to less breast compression and improved detection accuracy. However, AB-MRI was associated with a higher proportion of negative feedback, suggesting that for some patients, its benefits may be outweighed by procedural drawbacks.

• Withdrawal from the study was due to adverse patient experience for 151 women overall and was much more common with contrast-enhanced methods than with ABUS,² making ABUS the most well tolerated imaging method.

Previously, we summarized the greater cancer detection rates from CEM and AB-MRI compared to ABUS in the first round of screening in the BRAID trial.³  Cancer detection rates were similar at 7-9.5/1000 across all three modalities in the second round of screening.³  False alarm rates were higher for CEM and AB-MRI in the first round and comparable across modalities in the second round.⁴

¹ 40/44 (90.1%) of ABUS participants reported discomfort as did 38/48 (79.2%) of CEM participants, compared to only 14/38 (36.8%) of MRI participants.

² Of 151 withdrawals due to adverse patient experience, 69 were in women having CEM, 66 in women having AB-MRI, and only 12 in women having ABUS.

³ Cancer detection rates were 15/1572 (9.5/1000) for ABUS+MG; 9/1051 (8.6/1000) for AB-MRI+MG; and 6/870 (7.0/1000) for CEM in the second round. (By comparison, cancer detection rates in the first round were 4.2/1000 for ABUS; 17.4/1000 for AB-MRI; and 19.2/1000 for CEM after a negative mammogram.)

⁴ Recall rates in the second round were 4.3% for ABUS; 3.5% for AB-MRI; and 4.7% for CEM.  In the first round, recall rates were 4.0% for ABUS; 9.7% for AB-MRI; and 9.7% for CEM.

 

Updated Overview: International Atomic Energy Agency Panel Endorses Molecular Breast Imaging

An IAEA international expert panel has provided an updated overview of MBI using dual head detectors and ^99mTc-sestamibi as a complement to mammography, especially in women with dense breasts or inconclusive conventional imaging. They conclude that “it remains a valuable modality in breast imaging, offering high sensitivity for the detection of small, node negative disease and detection of large, advanced cancers that are undetected by mammographic screening. MBI is performed at a safe radiation dose, is well-tolerated by patients, and has exceedingly low risk of adverse events.”

The ACR 2021 position is that MBI is “usually not appropriate” for screening in dense breasts due to whole body radiation exposure, though the radiation dose has decreased and additional promising prospective multicenter results have come out since that time.

New Insurance Laws

Alabama and Wisconsin are the latest U.S. states to enact insurance laws for expanded no-cost breast imaging coverage. Based on DBI legislative analysis, there are now 38 states  (plus DC) that have enacted laws for expanded coverage (Minnesota’s is for diagnostic imaging only). The laws vary from state to state, and some still include out-of-pocket costs. Visit DBI’s legislation MAP for details by state.

Efforts to expand insurance coverage for extra testing are active at state and federal levels. To learn more about how state insurance laws intersect with federal initiatives like the new HRSA guidelines and the Find It Early Act, read our new content on the legislation MAP page (located below the map).

Screening Decision Support Flowchart
Resource reminder: Our popular decision support tool Who Needs More Screening? details evidence-based screening recommendations designed to optimize cancer detection in individuals in different risk categories, including dense breasts.

 

 

UK News
Cheryl Cruwys, DBI’s European Education Coordinator and founder of Breast Density Matters UK, continues to lead organized lobbying efforts to the UK government. In a letter signed by 66 MPs to the Secretary of State for Health and Social Care, urgent action was requested for “density inform” measures and needed policy changes.

 

Ireland News
The Health Information and Quality Authority has published a protocol to assess potential modifications to the BreastCheck program to consider ways in which breast density could be incorporated into the mammography screening pathway.

State Insurance Bills

Bills in both Alabama and Wisconsin have advanced to the Governor’s desk. Stay tuned for more details if/when the bills are signed into law.

J-START TRIAL FOLLOW-UP

The J-START, a prospective Japanese randomized trial that compared Mammography plus Ultrasound (US, intervention group) to Mammography alone (control group) in women age 40 to 49 years of age previously showed, as primary endpoints, that the intervention of adding US twice in two years improved detection of early stage (0 or 1) breast cancer and reduced interval cancers, but lowered specificity. In a subset analysis, US improved early breast cancer detection in both nondense and dense breasts.

Longterm follow-up (median exceeding 11 years) is now available, showing significant reduction in advanced stage cancers (stage II or higher: node positive or T2 or both) in the intervention group.

• 234/894 (26%) carcinomas were advanced stage at diagnosis in the intervention group
• 277/843 (33%) carcinomas were advanced in the control group.
• Hazard ratio for advanced stage at diagnosis was 0·83, 95·6% CI 0·70–0·98; p=0·026)
• Divergence between groups emerged around year four, widened until year eight, and remained stable thereafter.

In an accompanying editorial, DBI Medical Advisory Board members Drs Jean Seely and Paula Gordon point out that the magnitude of reduction in advanced cancers due to addition of US in women 40-49 is similar to the magnitude of mortality reduction attributable to screening mammography in that age group across randomized trials. These results are quite compelling for Asian women, though may not generalize to all populations. In many other studies in other patient populations, US has been shown to substantially improve early breast cancer detection when added to annual or biennial 2D mammography in women with dense breasts. The benefits of screening US are very small, however, in women having annual tomosynthesis.

Multi-lingual Patient Flyers

Our patient flyer, Why a Conversation About Your Breast Density Matters is available in over 30 languages and dialects. The one-page printable flyer is an ideal resource for clinics or health fairs. Share confidently, all DBI translated content is meticulously crafted by native speakers specializing in breast radiology.

 

Lecture: When is Screening Mammography Not Enough?

DBI’s Chief Scientific Advisor, Dr. Wendie Berg, was honored to give the Philip Strax, MD Breast Imaging and Saving Lives Lecture for the Texas Radiological Society in Houston on February 14^th, entitled “When is Screening Mammography Not Enough?” Dr. Strax is the radiologist who, together with statistician Sam Shapiro, led the pivotal and pioneering Health Insurance Plan of New York randomized trial of screening mammography in the 1960s.  Dr. Richard Strax, also a radiologist, has honored his father’s legacy with this lectureship. Dr. Daniel Kopans, Dr. Edward Sickles, and Dr. Debra Monticciolo are previous presenters.

Highlights from Dr. Berg’s presentation included:

• Participation in screening is the biggest issue needing improvement, including identifying women at higher risk who should start screening earlier than age 40 (and usually with MRI).  To accomplish this, all women ideally should have risk assessment by age 25, especially those of Black, Hispanic, Asian, or Ashkenazi Jewish descent or with family history of breast or ovarian cancer.
• Breast density, methods to assess risk, and supplemental screening options, with emphasis on when MRI is indicated, and gaps in current insurance coverage that would be addressed by the Find It Early Act.
• The “flip side” of risk assessment of possibly reducing screening of women based on low risk, including the WISDOM trial results. Such approaches are limited by concerns that current methods (including AI) are inadequate to identify all women who will develop cancer in the next 2-5 years.
• A look to the future may include liquid biopsy as a screening method, with the potential that breast imaging would become more focused on diagnostic imaging and biopsy

 

Newly Appointed

DBI’s European Educator and supporter for France, Professor Isabelle Thomassin-Naggara, MD, has been appointed as ESR Quality, Safety & Standards Committee (QSSC) Chair.  This is another distinguished appointment for Prof. Thomassin-Naggara, who also holds the post as Women’s Imaging Editor at the European Journal of Radiology. We extend our congratulations to, Isabelle!

Find It Early Act – Updated Statement of Support

DBI has played a pivotal role in drafting and advocating for this landmark legislation. The bill aims to eliminate patient costs for supplemental and diagnostic breast imaging for women with dense breasts or at increased risk, and close gaps in current state laws and federal guidelines. Crucially, it extends coverage across private and public plans, including Medicare, TRICARE, and the VHA. DBI has coordinated the latest update of the FIEA Statement of Support, now backed by a coalition of 50 leading organizations. To contact your federal representative to ask for their support of this bill, please visit FindItEarlyAct.org for lobbying tools, the Statement of Support, and FIEA Fact Sheet .

Canadian Society of Breast Imaging – Position Statement

The Canadian Society of Breast Imaging published their updated position statement on mammographic breast density and supplemental screening in the Journal of Breast Imaging.  The statement, co-authored by DBI Medical Advisory Board members Paula Gordon, MD and Jean Seely MD, includes the following key recommendations for Canadian women:

• All women should begin regular screening mammography with digital breast tomosynthesis (DBT) (or standard 2D mammography if DBT is unavailable) every 1-2 years beginning at age 40, unless individual risk assessment indicates earlier screening.
• Because vascular-based imaging (MRI or CEM) offers the highest sensitivity and specificity, women at average risk with heterogeneously or extremely dense breasts (ACR density categories C or D) should be offered supplemental screening in the following order of preference: MRI every 2 years, contrast-enhanced mammography every 2 years, or ultrasound annually.
• If availability of supplemental screening is limited, access should be prioritized for women with extremely dense breasts (ACR density category D).
• Women considered high risk for breast cancer (lifetime risk 20%-25% or greater) should have both annual mammography beginning at age 30 years and annual MRI screening from ages 25-69 years. Mammography alone is recommended to continue at age 70 and beyond.

Europe / UK Education

At the invitation of the University of Exeter, Cheryl Cruwys—DBI’s European Education Coordinator—will deliver a lecture to diagnostic radiography students on February 17th. Her presentation will cover the complexities of breast density and provide an overview of the professional healthcare resources available on the DBI website.

New State Insurance Bills

Two new state bills for expanded coverage for breast imaging: Hawaii HB2366, Utah HB0468. For existing state insurance laws, see our MAP.

MASAI Trial

A large, randomized, prospective, controlled, single-blinded non-inferiority trial including nearly 106,000 women in the Swedish breast cancer screening program was conducted to compare interval cancer rates between women screened with standard double reading versus Transpara AI-supported mammography to triage mammograms that should undergo double reading. The study included a mix of tomosynthesis and 2D mammography. Interval cancers were defined as primary breast cancers diagnosed between two screening rounds (typically every 1.5 – 2 years in average-risk women) or within 2 years after the last scheduled screening that were not detected at screening.

• The interval cancer rate was lower among women who received AI-supported mammography compared to those who had standard double reading.¹
• This effect was consistent across age and breast density.
• Sensitivity was higher in the AI-supported mammography group² and specificities were equal among both groups.³

AI-supported mammography improved cancer detection, reduced interval cancer rates and reduced radiologist workload by decreasing the need for double reading.  In the United States, double reading is not routinely performed and isolated AI interpretation is not yet accepted: there are barriers to implementation. Further, AI is not generally covered by insurance so facilities in the USA may opt to charge patients an out-of-pocket fee when offering AI.

¹ Interval cancer rates were 1.55/1000 (82/53043) (95% CI 1·23–1·92) and 1·76/1000 (93/52872) (95% CI 1·42–2·15) in the intervention and control groups respectively, with a non-inferior proportion ratio of 0·88 (95% CI 0·65–1·18; p=0·41).

² Sensitivities were 80.5% (338/420) and 73.8% (262/355) in the AI-supported and standard double reading groups, respectively.

³ Specificity was 98.5% in both groups (51852/52623 in the AI supported group vs. 51752/52517 in the double reading group).

New State Insurance Bills

State legislative initiatives for expanded coverage for breast imaging continue. This week’s bill introductions:  Kansas SB409, New Jersey A1462, New Jersey A1945, New Jersey A2152, New Jersey A2775.

For details on existing state insurance laws, visit the DenseBreast-info.org MAP.

 

New Study

A large study led by Ariel S. Kniss, MD, PhD evaluated the impact of breast density on performance of screening mammography with combination digital 2D mammography and tomosynthesis (DBT) from over 300,000 digital breast tomosynthesis (DBT) exams from over 111,000 women. All screening DBT exams were performed from January 2013 to June 2019 at one institution. Density was categorized using visual BI-RADS categories (A, almost entirely fatty, B, scattered areas of fibroglandular density, C, heterogeneous dense tissue, or D extremely dense tissue).

• Sensitivity of screening DBT decreased with increasing density¹
• Specificity of screening DBT was lower in women with category D density than those with category A, B, or C, respectively. ²

Performance of screening DBT decreases with increasing breast density, reinforcing the importance of supplemental screening considerations in women with dense breasts, even when DBT is used for screening mammography.

¹ Sensitivities were 92.8%, 90.1%, 81.0%, and 61.8% for categories A, B, C, and D, respectively (P < 0.001 for category D vs. A, B, or C, respectively).

² Specificities were 96.7%, 94.4%, 92.5%, and 93.3% for categories A, B, C, and D (P < 0.001 for category D vs. A).

New State Insurance Bills

Three new state bills were introduced this week for expanded coverage for breast imaging: Alabama SB117, West Virginia SB518 and SB519.

For details on existing state insurance laws, visit the DenseBreast-info.org MAP.

Federal Guideline for Insurance Coverage

Effective Jan 1, 2026, Affordable Care Act compliant plans are now required to cover, for women at average risk, “additional” breast imaging in the Health Resources Services Administration (HRSA)-supported Women’s Preventive Services Initiative (WPSI) recommendations. The new guidelines state,  “Women may require additional imaging to complete the screening process or to address findings on the initial screening mammography. If additional imaging (e.g., magnetic resonance imaging (MRI), ultrasound, mammography) and pathology evaluation are indicated, these services are also recommended to complete the screening process for malignancies.”

More details:

• “While there are currently no randomized controlled trials to support separate recommendations for women with dense breasts, the updated clinical recommendation supports additional testing to complete initial screening, if needed, which may be more common for women with dense breasts.”
• “Non-grandfathered group health plans and health insurance issuers offering group or individual health insurance coverage must cover without cost-sharing the services and screenings listed on the updated Women’s Preventive Services Guidelines.”  For most plans, this update will take effect in 2026.

Our input:

• Federal plans such as Medicare, VHA, and TRICARE and grandfathered plans appear to be exempt. Furthermore, while these updates focus on women at ‘average’ risk, they do not address the supplemental imaging requirements for those at increased or high risk. In contrast, the federal Find It Early Act addresses both of these gaps; we remain committed to advancing that legislation.
• Since federal guidelines are frequently codified through state-level legislation, we anticipate a continued influx of state insurance bills designed to mirror or expand on these federal requirements.

State Insurance Bills

New legislation for insurance coverage of expanded breast imaging:

Arizona  SB1165

California AB1570

Florida  SB1494
Rhode Island  S2032

South Carolina S0820

Insurer Breast Imaging Expansion

Blue Cross Blue Shield affiliated organization Highmark Inc. has announced plan members across 4 states will receive 100% coverage for diagnostic breast studies and MRIs. More HERE.

U.S. State Insurance Bills

New 2026 insurance bill introductions for expanded breast imaging include: Indiana HB1112 and Kentucky HB135

BRAID RSNA Presentations and Automated US Study

Additional BRAID Trial Results

Several sessions at RSNA 2025 showed second round results from the UK BRAID trial led by Dr. Fiona Gilbert.  The initial trial enrolled 9361 women age 50-70 from October 2019-March 2024 with dense breasts and a negative mammogram who were randomized to receive supplemental screening with automated breast ultrasound (ABUS), abbreviated MRI (AB-MRI), or contrast-enhanced mammography (CEM), with results published in The Lancet in May 2025.  Women with a negative result from the initial supplemental screen who enrolled by 3/31/23 were invited to have a second round of screening 18 months later that included mammography (MG).

• Cancer detection rates in the second round of screening were similar for each of ABUS, AB-MRI, and CEM arms at about 8/1000.¹
• Recall rates decreased for both AB-MRI and CEM in the second round but were similar to the first round for ABUS.²

¹ Cancer detection rates were 15/1572 (9.5/1000) for ABUS+MG; 9/1051 (8.6/1000) for AB-MRI+MG; and 6/870 (7.0/1000) for CEM in the second round. [By comparison, cancer detection rates in the first round were 4.2/1000 for ABUS; 17.4/1000 for AB-MRI; and 19.2/1000 for CEM after a negative mammogram.]

² Recall rates in the second round were 4.3% for ABUS; 3.5% for AB-MRI; and 4.7% for CEM.  In the first round, recall rates were 4.0% for ABUS; 9.7% for AB-MRI; and 9.7% for CEM.

 

ABUS in Non-dense Breasts

Two referral centers in Korea retrospectively reviewed results of screening women with non-dense breasts and compared performance of digital mammography (DM) alone vs DM + automated breast ultrasound (ABUS). They reviewed 2904 paired screening examinations among 1683 women with non-dense breasts.

• Adding ABUS to DM slightly increased cancer detection in women with scattered fibroglandular density, but not in women with fatty breasts.¹
• The small increase in cancer detection was accompanied by a small increase in false positive recalls.²

• Women diagnosed with breast cancer were older (63 ± 9 years vs. 59 ± 10 years, P = 0.04) and tended to have risk factors (19.2% vs. 4.2%, P < 0.001) compared to those without breast cancer.

¹ Cancer detection rates were 9.0/1000 for DM+ABUS and 7.9/1000 for DM alone (P<.001), with 3/26 (11%) cancers seen only on ABUS, all invasive T1, node-negative tumors; the added yield of 1.1/1000 was all in the 1569 women with scattered fibroglandular density [cancer detection rate 8.5/1000 for DM+ABUS and 7.4/1000 for DM alone (P<.001)].  Six cancers were visible only on DM, including 2 invasive cancers and 4 DCIS.

² False-positive recall rate increased from 61/2878 (2.1%) for DM alone to 143/2878 (5.0%) for DM+ABUS (P<.001).

New State Insurance Law

A new insurance law in Pennsylvania expands no-cost breast imaging coverage. Please see the Pennsylvania legislative analysis from DenseBreast-info HERE.

New Study

Esserman et al published results from the WISDOM (Women Informed to Screen Depending on Measures of Risk) trial in JAMA. The authors sought to demonstrate that reduced screening based on lower risk was safe.  Due to low adherence to recommended screening (with contamination in the annual arm), as indicated in the accompanying editorial, the effectiveness of such an approach remains uncertain. Breast density and other characteristics of women with symptomatic or advanced cancers were not specified. The approach of not screening “low-risk” women in their 40s is particularly problematic for Black, Hispanic, and Asian women who have earlier peak onset of breast cancer and more frequent distant disease at presentation.  See also the commentary by the Society of Breast Imaging.

A total of 28,732 women age 40-74 from all 50 states were randomized to risk-based screening or annual mammography.¹  Women were followed a median of 5.1 years (up to 10) and cancers, biopsies, and screening were self-reported.

• There was no significant difference in rate of stage IIB or higher breast cancers.²  Similar numbers of stage 0, stage I, and stage IIA cancers were seen in both groups (see eTable 5 in Supplement 1).
• The rate of breast biopsies was not different.³
• Reported adherence to recommended screening was less than 50% even in the first year for all participants and declined each year.
• Women in the annual arm were given risk assessment and told they were at elevated risk if they had BCSC 5-year risk > 5% but were not changed screening assignments.
  • While MRI use was more frequent in the risk-based arm (but was lower than expected), a substantial number of MRI examinations were performed in the annual arm.⁴

Women in the elevated and high-risk groups were counseled on risk-reducing medications, but use remained very low (5.6% at baseline, increasing to 10.3% with intervention, in the highest-risk group and 3.5% increasing to 4% in the elevated risk group).

¹ A total of 14212 women were randomized to risk-based screening based on the BCSC Risk Assessment Tool (which includes breast density) and polygenic risk score genetic testing, with 2% (291) “high risk” (≥6% 5-year risk or carrier of known pathogenic variant) and recommended to have alternating 6-month mammography and MRI; 8% (1121) “elevated risk” (top 2.5% of risk by age or extremely dense breasts, or intermediate genetic risk variant) assigned to annual mammography; 63% (8796) “average risk” assigned biennial mammography beginning at age 50; and 27% (3716) “low risk” (women less than 50 with <1.3% 5-year risk).  The other 14160 women were randomized to annual mammography screening.

² There were 41 stage IIB or higher cancers in the risk-based arm (63.1/100,000 person-years) and 43 (68.2/100,000 person-years) in the annual arm (rate ratio 0.92, 95%CI 0.60 to 1.43).

³ There were 1029 biopsies in the risk-based arm (1371/100,000 person-years) and 943 in the annual arm (1272/100,000 person-years, rate ratio 1.08 (95%CI 0.99 to 1.18).

⁴ Rate of MRI per 100,000 person-years was 1187 in the risk-based arm and 795 in the annual arm (rate ratio 1.49, 95%CI 1.34 to 1.66).

Announced at SABCS

At the San Antonio Breast Cancer Symposium last week, Astrin Biosciences presented encouraging results from an AI-proteomics-based liquid biopsy to detect breast cancer.  Specifically:

• High sensitivity and specificity of over 92% each were found in the validation set of 202 cancer cases and 195 controls, with AUC of 0.975.¹
• Sensitivity was maintained across all stages² and subtypes.³
• Patient demographics did not significantly affect results, though further study is warranted.
• They confirmed they were indeed identifying cancer-associated proteins and pathways versus inflammatory markers.

Sources of false positives will be important to understand and mitigate.

¹ A training set of 379 cancer cases and 466 controls was used, and all cases were treatment naïve.  Training used 5-fold cross validation and showed AUC of 0.961.

² 85% (11/13) for stage 0; 88% (61/69) for stage I; 97% (95/98) for stage II; 88% (14/16) for stage III; and 100% (2/2) for stage IV (no significant differences, but relatively small subsets).

³ 100% (20/20) for invasive lobular carcinoma; 94.8% (145/153) for invasive ductal carcinoma; 95.9% (116/121) for hormone receptor +/HER2-; 94% (15/16) for triple negative; 67% (4/6) for triple positive; and 100% (3/3) hormone receptor -/HER2+ (small numbers in most subsets).