False-Negative Review from the Mammography Audit: Refining Breast Imaging Practice.

MRI Screening of Women with a Personal History of Breast Cancer.

Introduction: Screening mammography for early detection of breast cancer is subject to both false positive and false negative results. False positive mammograms subject women to unnecessary diagnostic imaging and biopsies that create a burden to the patient and costs to the health care system. False negative mammograms, on the other hand, result in the diagnosis of an “interval” cancer following symptoms and typically at a later stage than if discovered on the prior screen. Prior studies conducted in metropolitan Chicago suggested that differences in both mammogram image quality and quality of the interpretation of the mammogram might contribute to higher false negative rates (FNR) and higher false positive rates (FPR) for ethnic minority patients. We sought to examine predictors of FNR and FPR, and whether ethnic disparities might exist in these outcomes within a single, large healthcare organization. Methods: Screening mammograms during 2001-2009 were linked to incident breast cancer cases for the period 2001-2010 from the Illinois State Cancer Registry (ISCR) using probabilistic linkage methods. Screening mammograms were scored by the interpreting radiologist using the American College of Radiology Breast Imaging Reporting and Data System (BIRADS) and each mammogram was defined as negative (BIRADS 1,2) or positive (BIRADS 0,4,5); BIRADS 3 mammograms were excluded from these analyses. A false positive (FP) mammogram was defined as a screening mammogram with an abnormal interpretation which nonetheless did not result in a breast cancer diagnosis in the subsequent 12 months. A false negative (FN) mammogram was defined as a screening mammogram with a normal interpretation in a woman who nonetheless was diagnosed with breast cancer (“interval cancer”) in the subsequent 12 months. We examined patient factors (race/ethnicity, age at diagnosis, and breast density) and tumor characteristics including hormone (estrogen and progesterone) receptor (ER/PR) status, tumor grade, and behavior (in-situ vs. malignant) and their association with FNR and FPR. Results: A total of 669,222 screens were included in this analysis and 4,058 breast cancer cases were diagnosed in the 12 months subsequent to each screening mammogram (3,071 screen-detected and 988 interval cancers). Overall, the false negative rate was 24.3%. As expected, the FNR decreased with increasing age, and increased with increasing breast density. Also as expected, FNR was higher for ER-/PR- breast cancer than other forms (31% vs. 25%, p=0.01), was higher for high grade/undifferentiated tumors and was higher for malignant than in situ tumors (27% vs. 19%, p <0.001). Contrary to expectation, however, FNR was higher in White than African Americans patients (26% vs. 20%, p<0.001); and FNR was higher for digital than analog mammograms (27% vs. 23%, p=0.001). With regards to false positive rate (FPR), FPR was 12.6% overall. As hypothesized, FPR was slightly higher in African Americans compared to Whites (14% vs. 11%, p=0.014), and higher for mammograms performed on women with heterogeneously/extremely dense versus less dense breasts (14% vs. 11%, p<0.0001). Conclusion: Contrary to expectation, we did not find a racial disparity in the probability of a false negative mammogram, though we did find a modestly increased false positive rate in ethnic minorities. Radiologists in this organization are provided feedback on the quality of their screening interpretations on a regular basis, and this quality control program may be responsible for leveling the quality of screening mammography by race/ethnicity. Citation Format: Firas M. Dabbous, Garth Rauscher, Terry Macarol, Jenna Khan, Therese Dolecek, Sally Friedewald, Teena Francois, Tom Summerfelt. Examining racial/ethnic disparities in mammography screening performance in a single, large healthcare organization. [abstract]. In: Proceedings of the Sixth AACR Conference: The Science of Cancer Health Disparities; Dec 6–9, 2013; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2014;23(11 Suppl):Abstract nr B74. doi:10.1158/1538-7755.DISP13-B74

Women with a personal history of breast cancer (PHBC) are at risk of developing another breast cancer and are recommended for screening mammography. Few high-quality data exist on screening performance in PHBC women. To examine the accuracy and outcomes of mammography screening in PHBC women relative to screening of similar women without PHBC. Cohort of PHBC women, mammogram matched to non-PHBC women, screened through facilities (1996-2007) affiliated with the Breast Cancer Surveillance Consortium. There were 58,870 screening mammograms in 19,078 women with a history of early-stage (in situ or stage I-II invasive) breast cancer and 58,870 matched (breast density, age group, mammography year, and registry) screening mammograms in 55,315 non-PHBC women. Mammography accuracy based on final assessment, cancer detection rate, interval cancer rate, and stage at diagnosis. Within 1 year after screening, 655 cancers were observed in PHBC women (499 invasive, 156 in situ) and 342 cancers (285 invasive, 57 in situ) in non-PHBC women. Screening accuracy and outcomes in PHBC relative to non-PHBC women were cancer rates of 10.5 per 1000 screens (95% CI, 9.7-11.3) vs 5.8 per 1000 screens (95% CI, 5.2-6.4), cancer detection rate of 6.8 per 1000 screens (95% CI, 6.2-7.5) vs 4.4 per 1000 screens (95% CI, 3.9-5.0), interval cancer rate of 3.6 per 1000 screens (95% CI, 3.2-4.1) vs 1.4 per 1000 screens (95% CI, 1.1-1.7), sensitivity 65.4% (95% CI, 61.5%-69.0%) vs 76.5% (95% CI, 71.7%-80.7%), specificity 98.3% (95% CI, 98.2%-98.4%) vs 99.0% (95% CI, 98.9%-99.1%), abnormal mammogram results in 2.3% (95% CI, 2.2%-2.5%) vs 1.4% (95% CI, 1.3%-1.5%) (all comparisons P < .001). Screening sensitivity in PHBC women was higher for detection of in situ cancer (78.7%; 95% CI, 71.4%-84.5%) than invasive cancer (61.1%; 95% CI, 56.6%-65.4%), P < .001; lower in the initial 5 years (60.2%; 95% CI, 54.7%-65.5%) than after 5 years from first cancer (70.8%; 95% CI, 65.4%-75.6%), P = .006; and was similar for detection of ipsilateral cancer (66.3%; 95% CI, 60.3%-71.8%) and contralateral cancer (66.1%; 95% CI, 60.9%-70.9%), P = .96. Screen-detected and interval cancers in women with and without PHBC were predominantly early stage. Mammography screening in PHBC women detects early-stage second breast cancers but has lower sensitivity and higher interval cancer rate, despite more evaluation and higher underlying cancer rate, relative to that in non-PHBC women.

The Breast Imaging Reporting and Data System (BI-RADS) was developed to standardize breast imaging reporting and facilitate cancer-probability communication and biopsy decision-making. However, the probability of malignancy for BI-RADS 4 designated breast lesions ranges from 2 - 95% and contributes to a high unnecessary biopsy rate. Over the decades, BI-RADS 4 tissue biopsy-proven positive predictive value (PPV3) rate have not improved (currently 21.1%) and translates to high false-positive rates of mammography. Recent advancements in breast cancer evaluation have resulted in digital breast tomosynthesis (DBT) to generate 3D images. Objectives. 1. To investigate the clinical performance of digital mammography (2D) versus DBT (3D) among Houston Methodist's BI-RADS 4 population. 2. To compare both modalities and determine if 3D mammography performs better than 2D mammography especially per cancer detection rates (CDR) and biopsy-derived positive predictive value (PPV3) among BI-RADS 4 population in a health system in Texas. Methods. We extracted retrospective clinical and mammography data for patients who underwent screening or diagnostic examination using 2D (DM) and 3D (DBT), performed between February 1, 2015, to September 30, 2020, from our clinical data warehouse at Houston Methodist. Data elements extracted include patient demographics including age, gender, race, and marital status; mammography modality used (2D vs. 3D); personal history of breast cancer; history of prior mammogram; final BI-RADS category; biopsy type: core needle or surgical; pathology results (malignant or benign) performed within three months after the mammogram; tumor staging; and hormone receptor and growth-promoting protein expression i.e. ER, PR, and HER2 status. Using student t, Fisher's exact, and Chi-squared tests, we evaluated the collected data to determine statistical significance of the difference between modalities in BI-RADS 4 cases including malignancy rate. We calculated the adjusted odds ratio between modalities for cancer detection rate (CDR) and biopsy-derived positive predictive value (PPV3). A p-value of &amp;lt; 0.05 was considered statistically significant. Results. A total of 158,630 encounters (83,905 unique patients) and 185,213 encounters (106,169 unique patients) had 2D and 3D mammography respectively performed across our hospital system. Out of these, 6,887 encounters (6,462 unique patients) in 2D and 6,483 encounters (6151 unique patients) in 3D were assessed as BI-RADS 4 within the period. Using Fisher’s exact test, the results show that the BI-RADS 4 assessed cases in 2D mammography are significantly more than those in 3D mammography by 24%, p-value &amp;lt; 10e-16 (1.24; 95% CI: 1.198 - 1.284). Among the parameters, only racial distribution (P=0.0018), history of breast cancer (P=0.0030), and prior mammogram performed (p&amp;lt;2e-16) were found to be significantly different between the modalities. The CDR among BI-RADS 4 cases were 117.47 for 2D and 122.32 for 3D with an adjusted odds ratio of 0.99 (0.89, 1.10) P=0.8725 adjusted by logistic regression. The PPV3 among BI-RADS 4 cohort were 15.09% (2D) and 16.27% (3D) and again, this difference was not significant with an adjusted odds ratio of 1.03 (0.92, 1.15), P=0.5796 adjusted by logistic regression. Conclusions. While DBT showed statistically significant improvement in performance and sensitivity in assigning BI-RADS 4 cases compared with DM i.e. 24% less assignment, there was no improvement in PPV3 and CDR in BI-RADS 4. Thus, DBT does not provide a reduction to unnecessary biopsies in BI-RADS 4 assessed patients. These findings are based on patient populations of our 9-hospital health system. Further investigation in different health systems is necessary to confirm these findings. Citation Format: Chika Frank Ezeana, Mamta Puppala, Lin Wang, Jenny C. Chang, Stephen T.C. Wong. A clinical study indicates that 3D mammography shows no improvement over 2D mammography in cancer detection rates and biopsy-derived positive predictive value among BI-RADS 4 populations [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-01-11.

Background Women with a personal history of breast cancer (PHBC) have a high rate of subsequent breast malignancy. Annual mammography with selective ultrasound has been the standard method of surveillance after breast cancer for many years, aiming for early detection, treatment and improved survival. In surveillance populations, interval cancers, which can account for ~30% of subsequent breast cancers, are more likely to be larger, hormone receptor negative and lymph node positive than cancers detected by surveillance (Lee et al, Radiology 2021). Interval cancer rates of 3.6 per 1000 mammographic screens have been reported previously in women with PHBC (Houssami et al, JAMA 2011). For women with PHBC a more sensitive surveillance approach may be justified, noting mammography has lower program sensitivity in PHBC surveillance than in screening. Magnetic resonance imaging (MRI) is used selectively but with resource and access limitations. Contrast enhanced mammography (CEM) offers a more sensitive modality than conventional mammography with specificity comparable to MRI, but its utility for surveillance is uncertain. Methods Retrospective study of 1,190 women with PHBC who commenced annual CEM surveillance in an Australian hospital setting between June 2016 and December 2022 (Elder et al, Breast Cancer Res Treat 2023) combining outcomes of initial CEM and any subsequent surveillance imaging over that period, including incident surveillance-detected cancers and interval cancers (cancers diagnosed in the year following a negative surveillance episode), recalls for assessment, the contribution of contrast to recall, and pathology and treatment details for cancer diagnoses. Outcomes were reported using descriptive statistics and hazards modelling. Results There were 3,784 episodes for analysis: 1,190 first CEM surveillance episodes, and 2,594 subsequent surveillance episodes of which &amp;gt;90% were contrast based imaging. 79% of women had at least three annual rounds of surveillance imaging. 186 cases from the total 3784 surveillance episodes were recalled for assessment (recall rate 4.9%). 72 (39%) recalled cases identified malignant lesions (true positives (TP)), with 50 invasive cancers and 22 cases of DCIS. 114 (61%) recalled cases were false positives (FP). 51% of cases were only recalled due to contrast and 35% of these were TP. Invasive cancers were predominantly stage 1 (64%) or stage 2 (32%) and most were grade 2 (44%) or grade 3 (47%). The median invasive cancer size was 16mm (IQR 9-25mm). 62% of invasive cancers were hormone receptor positive HER2 negative, and 24% were triple negative. The median DCIS size was 19mm (IQR 10-26mm) with only a single low grade case. 40% of invasive cancers and 59% of DCIS were recalled due to contrast. Comparison of tumour features indicates similarities between contrast-directed recalls and other diagnoses; conclusive findings would require a larger sample. Five interval cancers were identified, of which three were asymptomatic and detected on surveillance imaging scheduled early for other reasons. Thus, the rate of symptomatic interval cancers was 0.8 per 1000 screens (program sensitivity 96.0%). Surveillance-detected cancer rates differed significantly by index cancer subtype (χ2=11.9, p=0.0026), with highest rates for women with triple negative index cancers. Incidence cancer rates were higher among the 6.9% of women with moderate or marked BPE at first CEM surveillance episode (χ2=8.8, p=0.032), but did not differ significantly by age group (χ2=5.2, p=0.39) nor breast density (χ2=4.7, p=0.19). Conclusions Routine use of CEM in annual surveillance of women with PHBC led to 1.85-fold increase in the detection of clinically significant malignant lesions, with lower interval cancer rates than previous published series of women with PHBC. CEM appears to increase the sensitivity of surveillance programs for women with PHBC, improving on imaging without contrast. Citation Format: Julia Matheson, Carolyn Nickson, Kenneth Elder, Allan Park, Allison Rose, Bruce Mann. Outcomes of Surveillance using Contrast Enhanced Mammography in Women with a Personal History of Breast Cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PS05-06.

A history of breast cancer and older age allow risk stratification of mammographic BI-RADS 3 ratings in the diagnostic setting

Background: Surveillance is a fundamental tool in the early detection and secondary prevention of many cancers. For women at increased genetic risk of breast cancer, mammography and breast magnetic resonance imaging (MRI) serve as the standard screening modalities. Use of surveillance mammography and MRI has been understudied among women with variant of uncertain significance (VUS) compared to pathogenic and likely pathogenic variants (P/LP). To address this gap, we examined the use of breast cancer surveillance and breast surgery in women who underwent multiple gene sequencing in a multicenter cohort of patients. We also expanded the surveillance literature by assessing correlates of breast MRI and mammography among women with VUS and investigating how rates of imaging changed over time after genetic testing. Methods: Using data from two cancer settings, we calculated use of risk reducing mastectomy (RRM) and surveillance for all women at genetically elevated risk of breast cancer, regardless of their personal history of breast cancer, with VUS or P/LP variants in a breast cancer susceptibility gene of high penetrance (BRCA1, BRCA2, PALB2, PTEN, TP53) and moderate penetrance (ATM, CDH1, CHEK2, NBN, NF1, STK11). The primary outcome was longitudinal use of surveillance mammography and breast MRI for women during the 13-month span after genetic testing, and each subsequent 13-month period up to 6 years afterwards. Results: Of 889 women, those with and without personal history of breast cancer were similar with regards to race/ethnicity, marital status, and high- or average-risk status. However, women with a personal history of breast cancer were on average older (54.1 vs 48.2 years), had longer follow-up time since genetic testing (3.4 vs 3.0 years), and were more likely to have VUS (62.5% vs 37.7%) compared to those without personal history of breast cancer. VUS carriers were less likely to undergo RRM compared to those with P/LP (HR=0.17, p=&amp;lt; 0.001) and high-risk women were more likely to undergo RRM than average-risk women (HR=3.91, p=0.005). Longitudinally, surveillance use among unaffected women decreased from 49.8% in the first year to 31.2% in the sixth year after genetic testing. In comparison, a greater proportion of women with a personal history of breast cancer underwent surveillance, which increased from 59.3% in the first year to 63.6% in the sixth year after genetic testing. Mammography rates did not differ between women with P/LP and VUS within the first 13 months after genetic testing and up to 4 years afterwards. Over the first four years after genetic testing, women with VUS were less likely to undergo annual MRIs compared to P/LP. This observation was true for women without a personal history of breast cancer (OR=0.34, p=0.003; OR=0.37, p=0.03; OR=0.19, p=0.004 for years 1, 2, and 3 respectively) as well as for women with a personal history of breast cancer (OR=0.31, p&amp;lt;=0.001; OR=0.33, p=0.002; OR=0.37, p=0.012; OR=0.3, p=0.14 for years 1, 2, 3, and 4 respectively). Conclusion: In this study of surveillance mammography and breast MRI use among women at elevated risk of breast cancer, we found that women with P/LP variants in breast cancer susceptibility genes are more likely to undergo annual breast MRI compared to those with VUS, whereas there was no difference between the groups in their use of annual surveillance mammography. This study is one of the first to examine maintenance of breast surveillance in a sample of women at elevated risk of breast cancer with non-negative genetic test results in BRCA1/2 as well as non-BRCA1/2 genes, while adjusting for personal and family history of cancer. In addition, we found that VUS, whether in high or moderate penetrance breast cancer susceptibility genes, was associated with lower use of annual breast MRI compared to P/LP variants, and equivalent use of annual mammography. These results add important evidence to dispel the myth of VUS-associated mismanagement of care. Citation Format: Sukh Makhnoon, Minxing Chen, Brooke Levin, Megan Ensinger, Kristin Mattie, Generosa Grana, Sanjay Shete, Banu K. Arun, Susan K. Peterson. Use of breast surveillance between women with pathogenic variants and variants of uncertain significance in breast cancer susceptibility genes [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P6-02-04.

Background There is limited research on supplemental screening breast US in women with a personal history of breast cancer (PHBC). Purpose To compare the performance of supplemental screening breast US in women with and women without a PHBC by using a matched cohort. Materials and Methods Consecutive asymptomatic women who underwent radiologist-performed supplemental breast US and mammography between January 2013 and December 2013 at a tertiary referral university hospital were retrospectively identified. Inclusion criteria were negative or benign findings at mammography, follow-up data for at least 1 year, first cancer stage of 0 to II in women with a PHBC, and incidence screening in women without a PHBC. The two groups were matched 1:1 according to age and breast density. Performance measures were compared with McNemar test, generalized estimating equation, or penalized likelihood logistic regression. Results A total of 3226 women with a PHBC were matched with 3226 women without a PHBC (mean age ± standard deviation, 52 years ± 9; mammographic breast density, fatty in 603 and dense in 2623). Fourteen cancers (six screen-detected, eight interval cancers) were found in women with a PHBC and 13 cancers (12 screen-detected, one interval cancer) in women without a PHBC. Supplemental US in women with a PHBC compared with women without a PHBC showed lower sensitivity (43% [95% confidence interval {CI}: 18%, 71%; six of 14 cancers] vs 92% [95% CI: 64%, 100%; 12 of 13 cancers]; P = .03), higher interval cancer rates (2.5 [95% CI: 1.1, 4.9; eight of 3226 women] vs 0.3 [95% CI: 0, 1.7; one of 3226 women] per 1000; P = .02), and higher specificity (92.8% [95% CI: 91.9%, 93.7%; 2982 of 3212 women] vs 89.3% [95% CI: 88.2%, 90.4%; 2870 of 3213 women]; P < .001), respectively. Conclusion Supplemental US screening in women with a personal history of breast cancer had lower sensitivity and higher interval cancer rate but higher specificity relative to women without a personal history of breast cancer. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Lee and Lee in this issue.

We sought to determine how breast cancers that occur within 1 year after a normal mammogram are discovered. Using population-based mammography registry data from 2000-2002, we identified 143 women with interval breast cancers and 481 women with screen-detected breast cancers. We surveyed women's primary care clinicians to assess how the interval breast cancers were found and factors associated with their discovery. Women with interval cancers were twice as likely to have a personal history of breast cancer (30.1%) as women with screen-detected cancers (13.6%). Among women with interval cancers, one half of the invasive tumors (49.5%) were discovered when women initiated a health care visit because of a breast concern, and 16.8% were discovered when a clinician found an area of concern while conducting a routine clinical breast examination. Having a lump and both a personal and a family history of breast cancer was the most common reason why women initiated a health care visit (44%) (P <.01). Women with interval cancers are most likely to initiate a visit to a primary care clinician when they have 2 or more breast concerns. These concerns are most likely to include having a lump and a personal and/or family history of breast cancer. Women at highest risk for breast cancer may need closer surveillance by their primary care clinicians and may benefit from a strong educational message to come for a visit as soon as they find a lump.

Background There are few interval cancer studies of incident screening MRI for women with a personal history of breast cancer (PHBC). Purpose To evaluate the performance measures of screening breast MRI in women with a PHBC across multiple rounds and to identify subgroups who might be more at risk for interval cancer. Materials and Methods Between January 2008 and March 2019, consecutive screening breast MRI studies for women who had undergone breast-conserving surgery because of breast cancer were retrospectively identified. Inclusion criteria were negative or benign findings at mammography with US, availability of at least 1 year of follow-up data, and examinations having been performed within 12 months after the initial cancer surgery. Performance measures were calculated for each round. Multivariable logistic regression analysis was performed to determine factors associated with the risk of interval cancer. Results Among the 6603 MRI examinations for 2809 women (median age, 47 years; interquartile range, 42-53 years), the cancer detection rate was 8.3 per 1000 screening examinations (55 of 6603 examinations) and the interval cancer rate was 1.5 per 1000 screening examinations (10 of 6603 examinations). The sensitivity and specificity were 85% (55 of 65 examinations; 95% CI: 76, 93) and 88.3% (5775 of 6538 examinations; 95% CI: 87.6, 89.1), respectively. At multivariable analysis, interval cancers were associated with a first-degree family history of breast cancer (odds ratio [OR], 5.4; 95% CI: 1.3, 22.5; P = .02), estrogen receptor- and progesterone receptor-negative primary cancers (OR, 3.6; 95% CI: 1.1, 12.2; P = .04), and moderate or marked background parenchymal enhancement (OR, 10.8; 95% CI: 3.3, 35.7; P < .001). Conclusion Performance of screening breast MRI in women with a personal history of breast cancer was sustained across multiple rounds, and a first-degree family history of breast cancer, estrogen receptor- and progesterone receptor-negative primary cancers, and moderate or marked background parenchymal enhancement at MRI were independently associated with the risk of developing interval cancers. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Slanetz in this issue.

In 30 states, women who have had screening mammography are informed of their breast density on the basis of Breast Imaging Reporting and Data System (BI-RADS) density categories estimated subjectively by radiologists. Variation in these clinical categories across and within radiologists has led to discussion about whether automated BI-RADS density should be reported instead. To determine whether breast cancer risk and detection are similar for automated and clinical BI-RADS density measures. Case-control. San Francisco Mammography Registry and Mayo Clinic. 1609 women with screen-detected cancer, 351 women with interval invasive cancer, and 4409 matched control participants. Automated and clinical BI-RADS density assessed on digital mammography at 2 time points from September 2006 to October 2014, interval and screen-detected breast cancer risk, and mammography sensitivity. Of women whose breast density was categorized by automated BI-RADS more than 6 months to 5 years before diagnosis, those with extremely dense breasts had a 5.65-fold higher interval cancer risk (95% CI, 3.33 to 9.60) and a 1.43-fold higher screen-detected risk (CI, 1.14 to 1.79) than those with scattered fibroglandular densities. Associations of interval and screen-detected cancer with clinical BI-RADS density were similar to those with automated BI-RADS density, regardless of whether density was measured more than 6 months to less than 2 years or 2 to 5 years before diagnosis. Automated and clinical BI-RADS density measures had similar discriminatory accuracy, which was higher for interval than screen-detected cancer (c-statistics: 0.70 vs. 0.62 [P < 0.001] and 0.72 vs. 0.62 [P < 0.001], respectively). Mammography sensitivity was similar for automated and clinical BI-RADS categories: fatty, 93% versus 92%; scattered fibroglandular densities, 90% versus 90%; heterogeneously dense, 82% versus 78%; and extremely dense, 63% versus 64%, respectively. Neither automated nor clinical BI-RADS density was assessed on tomosynthesis, an emerging breast screening method. Automated and clinical BI-RADS density similarly predict interval and screen-detected cancer risk, suggesting that either measure may be used to inform women of their breast density. National Cancer Institute.

Background The ability of deep learning (DL) models to classify women as at risk for either screening mammography-detected or interval cancer (not detected at mammography) has not yet been explored in the literature. Purpose To examine the ability of DL models to estimate the risk of interval and screening-detected breast cancers with and without clinical risk factors. Materials and Methods This study was performed on 25 096 digital screening mammograms obtained from January 2006 to December 2013. The mammograms were obtained in 6369 women without breast cancer, 1609 of whom developed screening-detected breast cancer and 351 of whom developed interval invasive breast cancer. A DL model was trained on the negative mammograms to classify women into those who did not develop cancer and those who developed screening-detected cancer or interval invasive cancer. Model effectiveness was evaluated as a matched concordance statistic (C statistic) in a held-out 26% (1669 of 6369) test set of the mammograms. Results The C statistics and odds ratios for comparing patients with screening-detected cancer versus matched controls were 0.66 (95% CI: 0.63, 0.69) and 1.25 (95% CI: 1.17, 1.33), respectively, for the DL model, 0.62 (95% CI: 0.59, 0.65) and 2.14 (95% CI: 1.32, 3.45) for the clinical risk factors with the Breast Imaging Reporting and Data System (BI-RADS) density model, and 0.66 (95% CI: 0.63, 0.69) and 1.21 (95% CI: 1.13, 1.30) for the combined DL and clinical risk factors model. For comparing patients with interval cancer versus controls, the C statistics and odds ratios were 0.64 (95% CI: 0.58, 0.71) and 1.26 (95% CI: 1.10, 1.45), respectively, for the DL model, 0.71 (95% CI: 0.65, 0.77) and 7.25 (95% CI: 2.94, 17.9) for the risk factors with BI-RADS density (b rated vs non-b rated) model, and 0.72 (95% CI: 0.66, 0.78) and 1.10 (95% CI: 0.94, 1.29) for the combined DL and clinical risk factors model. The P values between the DL, BI-RADS, and combined model's ability to detect screen and interval cancer were .99, .002, and .03, respectively. Conclusion The deep learning model outperformed in determining screening-detected cancer risk but underperformed for interval cancer risk when compared with clinical risk factors including breast density. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Bae and Kim in this issue.

AimThis systematic review and meta-analysis investigate the added value of structured integration of Breast Imaging Reporting and Data System (BI-RADS) criteria using the Kaiser score (KS) to avoid unnecessary biopsies in BI-RADS 4 lesions.Material and methodsA systematic review and meta-analysis were conducted using predefined criteria. Eligible articles, published in English until May 2024, dealt with KS in the context of BI-RADS 4 MRI. Two reviewers extracted study characteristics, including true positives (TP), false positives (FP), true negatives (TN), and false negatives (FN). Sensitivity, specificity, negative likelihood ratio, and positive likelihood ratio were calculated using bivariate random effects. Fagan nomograms identified the maximum pre-test probability at which post-test probabilities of a negative MRI aligned with the 2% malignancy rate benchmark for downgrading BI-RADS 4 to BI-RADS 3. I² statistics and meta-regression explored sources of heterogeneity. p-values < 0.05 were considered significant.ResultsSeven studies with 1877 lesions (833 malignant, 1044 benign) were included. The average breast cancer prevalence was 47.3%. Pooled sensitivity was 94.3% (95%-CI 88.9%–97.1%), and pooled specificity was 68.1% (95%-CI 56.6%–77.7%) using a random effects model. Overall, 52/833 cases were FNs (6.2%). Fagan nomograms showed that KS could rule out breast cancer in BI-RADS 4 lesions at a pre-test probability of 20.3% for all lesions, 25.4% for masses, and 15.2% for non-mass lesions.ConclusionsIn MRI-assessed BI-RADS 4 lesions, applying structured BI-RADS criteria with the KS reduces unnecessary biopsies by 70% with a 6.2% FN rate. Breast cancer can be ruled out up to pre-test probabilities of 20.3%.Key PointsQuestionWhat, if any, value is added by structured integration of BI-RADS criteria using the Kaiser Score (KS) to avoid unnecessary biopsies in BI-RADS 4 lesions?FindingsThe structured integration of BI-RADS criteria using the Kaiser Score (KS) reduces unnecessary biopsies in BI-RADS 4 lesions by 70%.Clinical relevanceThe structured approach offered by the Kaiser Score (KS) avoids unnecessary recalls, potentially reducing patient anxiety, lessening the burden on medical personnel, and the need for further imaging and biopsies due to more objective and thus efficient clinical decision-making in evaluating BI-RADS 4 lesions.

Early Diagnosis of Hereditary Breast Cancer by Magnetic Resonance Imaging: What Is Realistic?

Purpose To (a) evaluate the frequency of Breast Imaging Reporting and Data System (BI-RADS) category 3 assessment in screening and diagnostic breast magnetic resonance (MR) imaging, (b) review findings considered indicative of BI-RADS category 3, and (c) determine outcomes of BI-RADS category 3 lesions, including upgrades, downgrades, and malignancy rates. Materials and Methods This retrospective study was approved by the institutional review board and compliant with HIPAA. The authors retrospectively reviewed the breast MR imaging database (2009-2011) to identify breast MR images classified as showing BI-RADS category 3 lesions. There were 9216 BI-RADS assessments in 5778 examinations (3360 women). Of the 9216 assessments, 567 (6%) in 483 women (average age, 47.2 years; median age, 47.0 years) were assigned BI-RADS category 3. In women with more than one BI-RADS category 3 lesion, the first lesion reported in the impression was used for data analysis. Outcomes data were available for 435 of the 483 women (90.1%). These women comprised the study cohort. Medical records from January 1, 2009, to May 31, 2015, were reviewed to obtain demographic characteristics and outcomes. χ(2) statistics and 95% exact confidence intervals (CIs) were constructed. Results MR imaging was performed for high-risk screening in 240 of the 435 patients (55.2%) and for diagnostic purposes in 195 (44.8%). Findings included mass (n = 125, 28.7%), focus (n = 111, 25.5%), nonmass enhancement (n = 80, 18.3%), moderate or marked background parenchymal enhancement (BPE) (n = 91, 20.9%), posttreatment changes (n = 16, 3.8%), and other findings (n = 12, 2.8%). Outcomes were as follows: 339 of the 435 patients (78%) did not have evidence of malignancy at more than 24 months, 28 (6.4%) underwent mastectomy (all benign), and 68 (15.6%) had lesion upgrades, with 11 cancers (2.5%). All 11 cancers were diagnosed in women with a genetic mutation or a personal history of breast cancer. No cancer was detected in cases of moderate or marked BPE. Conclusion Six percent of all breast MR imaging assessments were categorized as BI-RADS category 3, with a cancer rate of 2.5% (95% CI: 1.3%, 4.5%). All cancers were in women with a genetic mutation or personal history of breast cancer. Marked BPE does not necessitate a BI-RADS 3 assessment. (©) RSNA, 2016.