Background: Body mass index (BMI) and mammographic density (MD) are established but inversely correlated determinants of breast cancer risk. While higher BMI in post-menopausal women increases risk, greater BMI in early adulthood appears protective. MD, reflecting the proportion of fibroglandular to fatty tissue, is a strong positive risk factor. The extent to which MD mediates the protective association of early-adult BMI with post-menopausal breast cancer remains unclear. Methods: We will perform a mediation analysis within the Predicting Risk of Cancer at Screening (PROCAS) cohort (>57,000 women, UK NHS Breast Screening Programme). Early-adult BMI (age 18–21 years) is self-reported at recruitment; MD is measured at screening using visual analogue scales (VAS) and automated volumetric methods. Post-menopausal breast cancer incidence is determined through national cancer registry linkage. Analyses will estimate natural direct and indirect effects of early-adult BMI on cancer risk mediated by MD, using counterfactual mediation models with bootstrap confidence intervals, guided by an investigator-derived directed acyclic graph. Expected Impact: This study will clarify causal pathways linking adiposity and breast tissue composition to post-menopausal breast cancer. Findings may refine mechanistic understanding, improve individualised risk prediction, and support life-course strategies for breast cancer prevention. Keywords: Body mass index, mammographic density, breast cancer, mediation analysis, PROCAS

Contrast-Enhanced mammography (CEM) has been shown to have higher diagnostic accuracy compared to mammography in several studies. However, comparison between CEM and Digital Breast Tomosynthesis (DBT) has not been widely explored. The aim of this study is to compare the diagnostic accuracy of CEM with DBT plus synthetic 2D (S2D) mammography. Methods: 318 symptomatic women or women recalled from screening mammography were included in this retrospective study. Three experienced breast imaging radiologists blindly read CEM images, followed by DBT+S2D mammography images six months apart. Radiologists reported findings according to the American College of Radiology (ACR) Breast Imaging Reposting and Data System (BI-RADS) lexicon. Lesions were confirmed by histopathology or two-year radiological follow-up (benign and non-cancer cases only). Results: 239 (75.2%) women had malignant lesions while 79 (24.8%) were benign or non-cancer cases. Readers correctly diagnosed 236/239 (sensitivity 98.7%) malignant and 55/79 (specificity 69.6%) benign/non-cancer cases using CEM and 234/239 ( sensitivity 97.9%) malignant and 26/79 (specificity 32.9%) benign/non-cancer cases using DBT+S2D. Positive and negative predictive values (PPV) and (NPV) were: 90.8% vs 81.5% and 94.8% vs 83.8% for CEM and DBT respectively. The area under the receiver operator characteristic (AUROC) curve was 0.84% vs 0.65% for CEM and DBT+S2D respectively. Conclusion: CEM has excellent sensitivity and better specificity, PPV, NPV and accuracy compared to DBT+S2D mammography, offering the potential as an alternative screening/diagnostic tool in Saudi Arabia.

Studies have identified genetic and epidemiologic factors associated with mammographic density (MD) phenotypes. However, MD-associated genetic variants only account for a small proportion of the total estimated heritability. Interrogating interactions between genetic and epidemiologic factors could potentially identify additional MD-associated loci, expand our understanding of the genetic basis of MD phenotypes, and clarify how epidemiologic factors modulate relationships between genetic variants and MD. We conducted six separate genome-wide, gene-environment (GxE) interaction analyses, applying 2 degrees of freedom (df) and 1df interaction tests, for each of three MD phenotypes (percent density, dense area (DA), and nondense area (NDA)). The six epidemiologic factors considered were height, ever parous, parity, ever menopausal hormone therapy, ever breastfeeding, and months of breastfeeding. We included European ancestry participants from multiple studies within the Markers of Density consortium and the Breast Cancer Association Consortium (n = 4895-16 218 depending on specific analyses). We identified 11 loci with genome-wide significant (P < 5 × 10-8) interaction tests including two novel common genetic signals interacting with parity (8p21.2) and ever breastfeeding (19p13.2) for NDA. Our results suggest that epidemiologic risk factors might influence relationships between common genetic variants and MD phenotypes at particular genomic loci.

Background and purpose. The average of two expert assessments of Mammographic Density (MD) using Visual Analogue Scales (VAS) has been related to risk of breast cancer, despite reader variability. Much of the evidence for this method of MD assessment came from a single-centre, single-mammography-vendor setting. We investigate the inter-observer agreement of readers assessing density in images from multiple vendors and its association with mean density and mammography vendor. Methods. We analysed MD assessments from 11 experienced readers who each assessed the cancer-free breast of 50 women with unilateral breast cancer at three time points, with 3-51 mammograms from 6 different vendors. The three vendors with the fewest were grouped. The standard deviation of VAS readings for each mammogram was used as a marker of agreement. Regression analysis was used to investigate agreement with vendor. To further investigate agreement, twenty mammograms showing Low Agreement (LA) between readers and 20 with High Agreement (HA) were selected. Results. Regression analysis found that vendor was not significantly associated with reader agreement (F(3,146) = 1.47, p = 0.225; R-2 = 0.029). Mammogram system vendor accounted for 3% of variability in reader agreement. There was a strong association between reader agreement and VAS density (p < 0.001). Standard deviations were 16.1-21.3% (LA) and 2.2-8.7% (HA). Discussion. Despite the very different appearance of mammograms from different vendors, disagreement was not found to be associated with vendor in this analysis, but it was found to be associated with density. Since reader agreement is low for some cases, averaging multiple visual assessments of breast density may be advisable if resources permit.

Progress in deep learning accessibility and ease of use has led to its increased prevalence across research and clinical applications. Often, one of the first steps involves cleaning the training dataset by removing incomplete or erroneous data points. Mammograms collected from the Predicting Risk Of Cancer At Screening (PROCAS) study (n=57,897) include two views of each breast; if any case has more than four images the case is removed from the AI training set (approximate to 5% of participants). This can occur because the breast were too large to fit in a single image, requiring a multiple "mosaic" views, or because of a technical repeat. Other standard exclusion criteria include: presence of breast implants (approximate to 1%), non-standard additional views (approximate to 2%), and previous cancer diagnosis (approximate to 3% of participants). Removal of data for these reasons can inject bias into the training dataset. If any exclusion criteria correlate to specific demographics, then this can result in a systematic misrepresentation in the AI model. This work explores the quality of deep learning models trained on 'cleaned' data (n=33,408) and 'unclean' data containing only data from study participants with more than four mammographic views (n=1,709). Models were trained to predict percentage breast density, a strong indicator of cancer risk, averaged over two expert readers' assessments. The unclean dataset was shown to produce the same quality of model as a model trained on the 20 times larger clean dataset. This indicates that standard data cleaning practices require further assessment, given they contain enough information to produce a high-quality model. Exploring the effect of using this data can improve the amount of data available for AI training and improve the inclusion and equitable representation in technological advancements.

Background Screening for breast cancer produces benefits through cancers being detected earlier, thereby reducing premature deaths and the need for more intensive treatment. As with all screening, it can also produce harms such as false-positive screening test results. One way to improve the ratio of benefits to harms is through risk stratification. The main potential benefits of risk-stratified screening are via identifying women who are currently unaware that they are at increased risk and who can be offered more frequent screening and medicines for breast cancer prevention. However, it is unclear if these benefits would materialise in routine practice, whether additional harms would materialise and whether risk stratification is cost-effective. Objectives Our aims were to develop a risk-stratification system, BC-Predict, and to evaluate its feasibility when delivered as part of the National Health Service Breast Screening Programme. Specific objectives were to: (1) automate BC-Predict informatics systems and integrate into National Health Service Breast Screening Programme; (2) optimise to be acceptable to women who were offered it and healthcare professionals delivering it; (3) assess feasibility of BC-Predict, including estimates of benefits and harms; (4) identify likely cost-effectiveness and (5) engage key stakeholders to consider how risk stratification should be taken forward. Design and methods The PROCAS2-Collator software was created to control the workflow of BC-Predict, and qualitative methods were used to develop patient-facing materials, care pathways and study procedures. The main feasibility study involved women being offered BC-Predict as part of routine National Health Service Breast Screening Programme, with a comparison group of standard National Health Service Breast Screening Programme. Setting and participants The BC-Predict was offered at seven screening sites (three screening centres), with the comparison standard National Health Service Breast Screening Programme organised by two sites (one screening centre), within North West England. Participants were all women offered the National Health Service Breast Screening Programme at participating sites, with nested qualitative work with healthcare providers from the same screening centres. Intervention The BC-Predict risk-stratification system, offered to women when invited to the National Health Service Breast Screening Programme, calculated the 10-year risk based on the Tyrer–Cuzick model and produced risk feedback letters after negative screening test results were received. Women at high risk (≥ 8% 10-year risk) or moderate risk (≥ 5% to < 8% 10-year risk) were thereby encouraged to make telephone appointments to discuss prevention and early detection options. Main outcome measures Uptake of BC-Predict and subsequent prevention and early detection offers. BC-Predict was costed using a National Health Service perspective, and a decision-analytic model-based cost-effectiveness analysis was technically verified with validation. Results The BC-Predict was offered to 19,464 women, where 14,661 women attended screening (60.7%). Only 2429 women (12.5%) who were eligible took up the offer of BC-Predict. Uptake was substantially higher when women were personally approached at the study site: 137/263 (52.1%). Attendance at the telephone risk appointments offered was also lower than expected: 80/197 (40.6%) of high-risk and 68/379 (17.9%) of moderate-risk women. Of those who took up risk appointments, 105/148 (71%) women received a prescription for preventive medication and 63/80 (79%) accepted additional mammography. The cost-effectiveness analysis indicated that risk-based screening using self-reported risk factors and mammographic density provided 0.004 incremental quality-adjusted life-years per woman screened at an additional cost of £42 when compared to the current NHSBSP using 3-yearly screening (incremental cost-effectiveness ratio of £10,500 per QALY). Comparing this ICER with a cost-effectiveness threshold of £20,000 per QALY, suggests that replacing the current NHSBSP with risk-based screening could be a good use of the NHS budget. A nested questionnaire study found no effects on general anxiety or cancer-related worry for women who were offered BC-Predict. Thematic analyses of qualitative interviews revealed women were positive about BC-Predict, with only transient increases in worry reported by high-risk women. Healthcare professionals who were involved with the implementation were generally enthusiastic about the risk-stratified screening. The agenda-setting meeting identified a consensual view that risk-stratified screening is likely to happen eventually and that there is a need to develop plans to prepare for it. Limitations The study was not randomised and was dramatically impacted by the COVID-19 pandemic, with uptake of the study and of risk appointments in those identified as moderate or high risk almost certainly affected. As such, generalisability of the results will need to be reassessed after the results from the My Personalised Breast Screening trial are available. Conclusions The present work suggests that risk-stratified screening for breast cancer is feasible, acceptable and likely to be a cost-effective use of the healthcare budget. Key stakeholders at all stages viewed risk-stratified screening as generally desirable and inevitable. Future work The My Personalised Breast Screening trial has recruited over 50,000 women to examine the effectiveness of risk-stratified screening at preventing later-stage (2+) breast cancers. It is timely to consider information technology and workforce needs now and how best to engage women, especially those who are currently underserved by the existing National Health Service Breast Screening Programme. Study registration This study is registered as clinicaltrials.gov NCT04359420. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Programme Grants for Applied Research programme (NIHR award ref.: RP-PG-1214-20016) and is published in full in Programme Grants for Applied Research; Vol. 13, No. 13. See the NIHR Funding and Awards website for further award information. Plain language summary The National Health Service Breast Screening Programme invites all women aged 50–70 years to have 3-yearly mammograms. The majority of women in this group who should be offered additional health care (more frequent mammograms and risk-reducing medication) based on their breast cancer risk are unaware of this risk. We created an online breast cancer risk assessment intervention (BC-Predict) to estimate 10-year breast cancer risk, based on a self-report questionnaire, alongside mammogram information and deoxyribonucleic acid from a saliva sample. We produced written risk feedback information that was sent to women who wanted risk feedback after completing BC-Predict in the following categories: below average, average, moderate or high risk. Moderate- and high-risk women were encouraged to make a risk appointment with a healthcare professional. We ran focus groups with healthcare professionals at the National Health Service Breast Screening Programme centres, taking part in our study, to try to minimise concerns they had about BC-Predict. We interviewed British-Pakistani women to try to ensure that BC-Predict was usable by them. The BC-Predict was offered to 19,464 women from three National Health Service Breast Screening Programme centres alongside their 3-yearly mammogram invitation. Overall, 2429 women took part and 148/576 (25.7%) moderate- or high-risk women accepted the invitation for risk appointments. All these numbers were lower than in previous research. Of the women who were at moderate or high risk and made appointments to discuss risk, 105/148 (71%) women received a prescription for preventive medication and 63/80 (79%) accepted additional mammography. Questionnaires found no changes in anxiety or cancer-related worry. Interviews with 40 women who received BC-Predict risk feedback were positive about it, though some had short-term concerns. The economic analysis, comparing the NHS costs and health benefits of Predict-EC with the current national breast screening programme, indicates that risk-based breast screening could be a good use of the NHS budget. It is possible to include breast cancer risk assessment in the National Health Service Breast Screening Programme. Further work should focus on finding out how best to increase uptake, especially in more underserved communities. Scientific summary Background Screening for breast cancer (BC) provides benefits in terms of preventing premature deaths and less invasive treatment being needed due to BCs being detected earlier. In line with all screening, it also produces harms such as overdiagnosis and false-positive screening test results. One way to improve the balance of benefits to harms, that has been proposed, is risk stratification. National Institute for Health and Care Excellence provides guidance related to the management of women at increased risk of BC, which indicates that women at higher risk should be offered more frequent screening and risk-reducing medication. However, the majority of women at increased risk do not have a strong family history of BC and are unaware of their risk. Providing women at screening, with the option of having their risk assessed, would allow higher-risk women to receive these offers. However, it is not clear if women would take up services in routine practice to gain these benefits, or whether additional harms would outweigh any benefits. The present programme of work concerned a risk-stratification system, BC-Predict, to be offered to women when invited to the National Health Service Breast Screening Programme (NHSBSP). It calculates 10-year BC risk based on the validated Tyrer–Cuzick model, which combines information about self-reports of family history and factors that affect lifetime hormone levels with mammographic density from mammography and, in a subsample, single-nucleotide polymorphisms (SNPs) derived from saliva. Risk assessments identified women as belonging to one of the following categories: below average, average, above average (moderate) or high risk. The BC-Predict system then produces risk feedback letters and advice related to preventive options, which are sent to women participants and their general practitioner. Women at high risk (≥ 8% 10-year) or moderate risk (≥ 5% to < 8% 10-year) were thereby invited to make a telephone appointment to discuss prevention and early detection options at a family history, risk and prevention clinic (FHPC). Aims and objectives The aims of the present programme were to develop the BC-Predict system and to evaluate its feasibility when delivered as part of the NHSBSP. Specific objectives were to: (1) ensure that BC-Predict informatics systems functioned as intended; (2) ensure that it was acceptable to women who were offered it and healthcare professionals (HCPs) delivering it; (3) assess multiple aspects of feasibility of BC-Predict when rolled out in real time in screening centres in North West England, including benefits such as uptake of BC-Predict, risk consultations and chemoprevention harms, such as increased anxiety of women taking up BC-Predict; (4) identify key drivers underpinning the relative cost-effectiveness of embedding BC-Predict into the NHSBSP and (5) engage key stakeholders in an agenda-setting meeting to identify how risk stratification should best be taken forward. Methods The PROCAS2-Collator software was created to control the workflow of BC-Predict, including integration of information on risk factors (self-reported information on family history and hormone-related factors, e.g. age at first pregnancy, via questionnaire; mammographic density; and in a subsample, SNPs from saliva). Letters informed participating women of their risk categories and the implications of this. Qualitative methods were used to develop patient-facing materials, and BC-Predict care pathways and study procedures, with (1) women who had taken part in an earlier study where they received BC risk estimates, (2) British-Pakistani women from a deprived location and (3) HCPs from multiple professional backgrounds who were preparing to deliver BC-Predict. The main feasibility study involved the offer of BC-Predict to women as part of routine NHSBSP, with a comparison group of standard NHSBSP, and it was registered with clinicaltrials.gov (NCT04359420). Inclusion criteria for BC-Predict were women born biologically female, invited for 3-yearly mammographic breast screening, able to provide informed consent and complete a self-report risk assessment questionnaire. Uptake of BC-Predict and subsequent offers of risk review appointment, enhanced screening and preventive medication were recorded. A nested questionnaire study with 662 women examined changes in potential psychological harms, relative to women offered standard of care NHSBSP, at baseline (screening appointment) and 3 months and 6 months later. Further nested qualitative studies examined the experiences of women who received each of the four risk results after they had completed the BC-Predict pathway and participating HCPs after their centres finished offering BC-Predict. The cost of delivering BC-Predict in the NHS was calculated, and a decision-analytic model-based cost-effectiveness analysis was technically verified and validation of the model was carried out. A final meeting involved analysis of discussions of key stakeholders in BC screening to develop a risk-stratified breast screening implementation agenda. Results Development of BC-Predict software, materials, care pathways and procedures The PROCAS2-Collator software was developed and tested through an iterative series of pilots, with study team and services users to assess whether it was functioning as intended. A series of think-aloud interviews with 57 women developed and refined materials for use with women within each risk category. Women found these materials to be clear and appropriately pitched. The materials addressed concerns, such as users wanting materials to be framed in terms of the greater majority of women who do not develop cancer, and what women could do to reduce their risk. Interviews were conducted with 19 British-Pakistani women, of whom 14 required a translator. Although there was enthusiasm for risk-stratified screening, a number of misunderstandings were identified, such as believing that breast screening is for those who present with symptoms. The information materials were therefore translated and made available in multiple languages and offered along with all BC-Predict invites. Focus groups with 29 HCPs who would shortly be delivering BC-Predict identified concerns over capacity limitations in a service that was already stretched and concerns about increasing anxiety in some women and exacerbating existing inequalities in screening. This work fed into developing and refining procedures and care pathways for BC-Predict. For example, the creation of a BC-Predict study hotline and centralised risk discussion appointments at a single regional FHPC centre. Main feasibility study: uptake rates The BC-Predict was offered to 19,464 women at seven screening sites organised by three screening centres, of whom 14,661 women attended screening (60.7%). Of the invited cohort, only 2429 women (12.5%) took up the offer of BC-Predict. Of those who attended the screening, 16.6% accepted the offer of BC-Predict. This figure was substantially higher when women were personally approached at the study site. When personally approached, 137/263 (52.0%) women took up the offer of BC-Predict, and 79/125 (63.2%) of women took up BC-Predict when approached and were offered a paper questionnaire instead of requiring completion of an online form. Overall uptake was lower in women living in more deprived locations, as assessed by the Index of Multiple Deprivation. Telephone risk appointments were taken up by 80/197 (40.6%) of high-risk and 68/379 (17.9%) of moderate-risk women. Of those who took up risk appointments, 105/148 (71%) women accepted the offer of a prescription for risk-reducing medication. There was also a high uptake of additional screening in 107/148 (72.3%) women. Main feasibility study: impact on women offered BC-Predict and healthcare professionals There were no changes in the questionnaire ratings of general anxiety and cancer-related worry for women who were offered BC-Predict compared to women who were offered standard NHSBSP. Questionnaire scores were typical of those found in other studies of NHSBSP samples. Within BC-Predict, women, who were told that they were at high risk, rated their own risk higher and reported more cancer worry at 6 months than the women who were informed of being at lower risk, but changes were modest in size. A thematic analysis of 40 interviews with women within each of the four risk results revealed that women were positive about BC-Predict. Women who did not expect to be at high or moderate risk reported that they felt transient increases in worry, when found to be so, but that these adverse effects did not last. Fourteen HCPs working in the NHSBSP who were interviewed after their involvement with BC-Predict were generally more enthusiastic about having implemented the risk-stratified screening. Major increases in workload, which were a concern prior to implementation, did not materialise. Main feasibility study: cost-effectiveness analysis The cost of estimating a woman’s risk of BC was calculated to be relatively small, with an expected cost of £8.46 per woman for the approach used in BC-Predict. The addition of SNPs would result in an appreciably higher cost of £88.87. The cost-effectiveness analysis indicated that risk-based screening, using self-reported risk factors and mammographic density, was a more cost-effective use of resources than the 3-yearly screening, 2-yearly screening or no screening. At higher cost-effectiveness thresholds, the 2-yearly screening (as is routine in many other European countries) was equally as cost-effective as risk-stratified screening, but it was associated with a higher number of screens per woman (8.24 vs. 6.00) and therefore was an additional burden on the NHSBSP. Agenda-setting meeting A meeting at the end of the programme, an agenda-setting meeting involving key stakeholders, including HCPs, academic experts, patient and public involvement contributors and NHSBSP staff, was held. The meeting identified a consensual view that risk-stratified screening would happen eventually, pending results of the My Personalised Breast Screening (MyPeBS) effectiveness trial and the need to develop plans to prepare for it. These stakeholders identified the following recommendations for research: identify procedures for how best to engage women identify how best to promote uptake by women who are currently underserved by existing NHSBSP, notably women from minority ethnic populations and those living in more deprived areas how to organise risk-stratified screening to avoid placing any additional burden on NHS staff and to consider further the role of general practice consider feasibility of extending screening intervals of women at lower risk, for new service to not require additional funding validate risk prediction algorithms for other ethnic groups and over longer follow-up periods consider whether risk assessments that consider other diseases alongside BC would be valuable to women. Conclusions The present work has indicated that risk-stratified screening for BC is feasible as part of the NHSBSP. It was acceptable with no major harms to women who were offered it, and HCPs were involved in its delivery. We originally anticipated offering BC-Predict to 18,700 women, and on the basis of Predicting-Risk-Of-Cancer-At-Screening results, we anticipated that 8000 women would consent to BC-Predict, of whom 1169 would be seriously considered for chemoprevention and 117 women would take it up. Thus, although the uptake of BC-Predict was lower than expected, this was partly attributable to coronavirus disease discovered in 2019 affecting the NHSBSP during the duration of the study. By contrast, risk-stratified screening also resulted in a higher uptake of medicines to reduce the risk of BC than previously reported, so the number taking up medication in the present study (n = 105) was only slightly below the number (n = 117) that was originally expected. Given this, and as chemoprevention produces NHS cost savings, a validated decision-analytic model indicated that a risk-stratified approach to BC screening is likely to be the most cost-effective option or equally as cost-effective as a 2-yearly screening. However, current capacity constraints in the NHSBSP, including the number of available radiographers, and the requirement for significantly more screening appointments in a 2-yearly approach may prove difficult to accommodate. The MyPeBS trial has recruited over 50,000 women to examine the effectiveness of risk-stratified screening for reducing later-stage (2+) BCs, which should report by 2027. It is timely to consider plans for this, including considering information technology and workforce needs to allow roll-out in anticipation of these trial results. There is a need for further research to address how to increase the uptake, promote informed choices and to avoid exacerbating inequalities further. Study registration This study is registered as clinicaltrials.gov NCT04359420. Funding This award was funded by the National Institute for Health and Care Research (NIHR) Programme Grants for Applied Research programme (NIHR award ref.: RP-PG-1214-20016) and is published in full in Programme Grants for Applied Research; Vol. 13, No. 13. See the NIHR Funding and Awards website for further award information.

Background:Breast density is an independent risk factor for breast cancer and affects the sensitivity of mammography screening. Therefore, new breast imaging approaches could benefit women with increased breast density in early cancer detection and diagnosis. Objectives:To assess the diagnostic performance of abbreviated breast MRI compared with mammography and other imaging modalities in screening and diagnosing breast cancer among Saudi women with dense breast tissue. Methods:A retrospective diagnostic study was conducted using anonymized medical images and histopathology information from 55 women, aged ≥30 years, who had dense breasts (Breast Imaging and Reporting Data System [BI-RADS] breast density categories C and D) and an abnormal mammogram. The sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were calculated for mammography, digital breast tomosynthesis (DBT), synthetic mammography (SM) derived from DBT, ultrasound, and abbreviated breast MRI (ABMRI). Results:A total of 19 women had pathology-proven breast cancer. Among all methods, ABMRI showed the highest sensitivity (94.7%) and specificity (58.3%), while mammography showed the lowest (84.2% and 44.4%, respectively). AUC for ABMRI was higher than all the methods including mammography (0.751 vs. 0.643; P < 0.05). Conclusion:ABMRI appears to be more accurate in cancer diagnosis than mammography and other modalities for women with dense breast tissue. Further research is advised on a larger sample of Saudi women to confirm the benefit of ABMRI in breast cancer screening and diagnosis for women with increased breast density.

Shifted apparent diffusion coefficient (sADC) is an alternative to magnetic resonance elastography for calculating liver stiffness virtually, without the use of special elasticity software/hardware. The aim of this work was to investigate the effects of changing repetition time (TR), echo time (TE) and number of excitations (NEX) on sADC quantification using an abdominal phantom. Diffusion weighted images with b-values of 0, 200, and 1500 s/mm(2) were acquired at 1.5 T using a fat-suppressed spin echo sequence with varying TR (1-16 s), TE (75-90 ms) and NEX (4-12). ADC maps were generated by the scanner and sADC maps were calculated from the b = 200 and b = 1500 diffusion weighted images. Regions of interest were selected in the phantom liver parenchyma and lesion. Percent difference was calculated for each variable as [(Maximum-Minimum)/Maximum]*100. The session was performed 5 times and the mean ADC / sADC for each variable across each session was used to calculate the between-session percent difference. At TR>5000 ms the percent difference in ADC and sADC was generally less within a session than between sessions. At TR<5000 ms, ADC increased initially with TR. For NEX and TE, percent difference between sessions was generally greater than percent difference within sessions. This phantom study showed little obvious effect of varying TR (when TR>5000 ms), NEX or TE on ADC or sADC compared to the differences between sessions with the MR protocol used.

Introduction: The aim of this study is to measure reader agreement for i) lesion classification and ii) breast density in Contrast Enhanced Mammography (CEM). Methods: Two experienced and two inexperienced CEM readers reported 60 examinations. Kappa was used to assess inter-reader agreement between experienced and inexperienced readers for lesion classification (benign/malignant) and breast density (dense/non-dense). Weighted kappa was used to assess agreement for BI-RADS categories (1-5) and BI-RADS density (A-D). Intraclass correlation coefficient (ICC) measured agreement for breast density using Visual Analog scale (VAS). Intra-reader agreement for one experienced and one inexperienced reader was measured after a three month interval. Results: Agreement between experienced readers was substantial (kappa=0.66) for benign/malignant, and moderate (kappa=0.57) for BI-RADS categories. Agreement for inexperienced readers was moderate for benign/malignant and BI-RADS categories (kappa=0.52, kappa=0.47 respectively). Breast density (dense/non-dense) agreement was almost perfect for experienced readers (kappa=0.83) and substantial for BI-RADS (kappa=0.70). Inexperienced reader agreement was moderate for dense/non-dense (kappa=0.50) and BI-RADS (kappa=0.49). ICC for VAS was moderate for experienced (ICC=0.60) and good (ICC=0.84) for inexperienced readers. Intra-reader agreement for benign/malignant classification was almost perfect for both experienced and inexperienced readers respectively (kappa=0.83, kappa=0.91). Conclusion: Experienced readers showed substantial agreement for lesion classification and almost perfect agreement for breast density. While inexperienced reader agreement was moderate for both lesion classification and breast density, their agreement for VAS was higher than experienced readers, suggesting that CEM may have a short learning curve and that radiologists could potentially be trained for CEM interpretation, which would help its implementation in other clinical practices in Saudi Arabia.

Introduction: Breast cancer is the most common female cancer worldwide; however ethnic differences have been observed in both prevalence and prognosis, with Black women often having less favorable outcomes. Increased breast density is an independent risk factor for breast cancer and reduces the efficacy of mammographic screening. We investigate how it relates to ethnicity, to facilitate the provision of appropriate screening and advice to all women. Method: We use data from the UK Predicting Risk of Cancer at Screening (PROCAS) study. This involved completion of a questionnaire to obtain personal risk factor information during routine breast screening. Mammographic density was assessed using Visual Analogue Scales (VAS), and these scores were used to train an AI-based density measure, pVAS, which we applied to raw mammographic data from 41,241 women in PROCAS. Analysis of covariance was used to assess the relationship between ethnicity and breast density after adjusting for age, body mass index (BMI), menopausal status, hormone replacement therapy (HRT) use, parity, alcohol consumption, and family history of breast cancer. Pairwise comparisons for each ethnic group were performed using a Bonferroni correction. Results: 91.0% of the study population were white, 1.6% Asian, 1.1% Black and 1.0% Jewish. Jewish women had higher breast density than all other ethnic groups studied (p<0.001), with a mean pVAS of 34.8% (95% CI 33.6-36.1). Asian women had a mean density of 31.4% (95% CI 30.4-32.4) and significantly denser breasts than White women who had a mean pVAS density of 28.6% (95% CI 28.4-28.7). Conclusion: Previous research has reported mixed results. The relationship between risk factors for breast cancer are complex, and data not always complete, making this a challenging area of research. Our results support published evidence that some groups have increased density, and this relationship should be considered to ensure equity in screening and diagnosis.

Objectives: High mammographic density (MD) and excess weight are associated with increased risk of breast cancer. Weight loss interventions could reduce risk, but classically defined percentage density measures may not reflect this due to disproportionate loss of breast fat. We investigate an artificial intelligence-based density method, reporting density changes in 46 women enrolled in a weight-loss study in a family history breast cancer clinic, using a volumetric density method as a comparison. Methods: We analysed data from women who had weight recorded and mammograms taken at the start and end of the 12-month weight intervention study. MD was assessed at both time points using a deep learning model, pVAS, trained on expert estimates of percent density, and VolparaTM density software. Results: The Spearman rank correlation between reduction in weight and change in density was 0.17 (-0.13 to 0.43) for pVAS and 0.59 (0.36 to 0.75) for Volpara volumetric percent density. Conclusions: pVAS percent density measurements were not significantly affected by change in weight. Percent density measured with Volpara increased as weight decreased, driven by changes in fat volume. Advances in knowledge: The effect of weight change on pVAS mammographic density predictions has not previously been published.### Competing Interest StatementThe authors have declared no competing interest.### Clinical TrialISRCTN16431108### Funding StatementThe project was funded by Prevent Breast Cancer (registered charity number 1109839) and supported by the NIHR Manchester Biomedical Research Centre (IS-BRC-1215-20007) infrastructure. The funders had no role in the design, conduct, analysis or write up of the study. MH, DGE, AH , SA are supported by the NIHR Manchester Biomedical Research Centre (IS-BRC-1215-20007). ### Author DeclarationsI confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.YesThe details of the IRB/oversight body that provided approval or exemption for the research described are given below:The study was reviewed by the North-West - Preston Research Ethics Committee reference 17/NW/0440. Written informed consent was obtained from all participants and the study was performed in accordance with the Declaration of Helsinki.I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.YesI understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).YesI have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.YesAll datasets used and analysed during the current study are available on reasonable request.

Purpose:To assess the contribution of germline pathogenic variants (PVs) in population-based series of breast cancers and the best strategy to improve detection rates. Methods:Three cohort studies were utilized, including a hospital-based series identified from new UK mainstream testing criteria (group-1), offering testing to all women (group-2-BReast CAncer [BRCA]-DIRECT), and a Greater Manchester cohort study recruited from the mammography screening population (group-3-Predicting Risk of Cancer at Screening). DNA samples from women with breast cancer were sequenced for PVs in BRCA1, BRCA2, and Partner and Localiser of BRCA2 (PALB2). The Manchester score (MS) was used at different points thresholds. Current mainstream criteria include women diagnosed <40 years and all triple negative <60 years or an MS ≥15. Results:Thirty-six PVs (BRCA1 = 9, BRCA2 = 18, PALB2 = 9) were identified among 1061 women with breast cancer (3.4%). Mainstreaming criteria identified 21 of 36 (58%) of PVs by testing 190 women; detection rate (8.4%), specificity = 83.5%. A better detection rate was found using an MS threshold of 12-points with 66.7% (24/36) sensitivity and 85.7% specificity in 171 women. No PVs were identified in 158 women with grade-1 invasive cancers. The best strategy to detect all PVs was an MS ≥3 with specificity of 32.6%. Conclusion:In order to detect higher PV rates on a population basis the best strategy is to reduce the MS threshold for genetic testing.

Breast density is an important factor in assessing individual breast cancer risk. We aim to identify women at increased risk of developing breast cancer before they enter routine screening, using mammography in combination with known risk factors. This will enable targeting of preventive therapies and personalised screening. To reduce radiation risk, this paper examines whether density measurements in one breast or mammographic view could be used to accurately reflect individual risk. We analysed breast cancer risk using breast density in a 1:3 case-control dataset of mammograms from the Predicting Risk of Cancer at Screening Study (PROCAS). Breast density was measured using pVAS, an AI-based approach. Cancer risk in low and high breast density groups was compared using conditional logistic regression. High breast density was independently associated with increased breast cancer risk. Women in the highest breast density quintile averaged across all views had an Odds Ratio (OR) of 4.16 (95% CI 2.90-5.97) compared to those in the lowest. A similar OR was found in both the left 3.77 (95% CI 2.68-5.31) and right 4.52 (95% CI 3.12-6.55) breasts individually. ORs were also significant for each individual view: right mediolateral oblique (MLO) 4.19 (2.92-6.00), right craniocaudal (CC) 4.40 (3.09-6.27), left MLO 3.27 (2.34-4.56) and left CC 3.65 (2.60-5.11). The ability to predict breast cancer risk due to increased breast density was achieved using one breast and even one mammographic view. This provides the possibility of a pre-screening risk assessment using fewer images and therefore less radiation.

Purpose. To improve breast cancer risk prediction for young women, we have developed deep learning methods to estimate mammographic density from low dose mammograms taken at approximately 1/10th of the usual dose. We investigate the quality and reliability of the density scores produced on low dose mammograms focussing on how image resolution and levels of training affect the low dose predictions. Methods. Deep learning models are developed and tested, with two feature extraction methods and an end-to-end trained method, on five different resolutions of 15,290 standard dose and simulated low dose mammograms with known labels. The models are further tested on a dataset with 296 matching standard and real low dose images allowing performance on the low dose images to be ascertained. Results. Prediction quality on standard and simulated low dose images compared to labels is similar for all equivalent model training and image resolution versions. Increasing resolution results in improved performance of both feature extraction methods for standard and simulated low dose images, while the trained models show high performance across the resolutions. For the trained models the Spearman rank correlation coefficient between predictions of standard and low dose images at low resolution is 0.951 (0.937 to 0.960) and at the highest resolution 0.956 (0.942 to 0.965). If pairs of model predictions are averaged, similarity increases. Conclusions. Deep learning mammographic density predictions on low dose mammograms are highly correlated with standard dose equivalents for feature extraction and end-to-end approaches across multiple image resolutions. Deep learning models can reliably make high quality mammographic density predictions on low dose mammograms.

Purpose: Breast density is associated with the risk of developing cancer and can be automatically estimated using deep learning models from digital mammograms. Our aim is to evaluate the capacity and reliability of such models to predict density from low-dose mammograms taken to enable risk estimates for younger women. Approach: We trained deep learning models on standard-dose and simulated low-dose mammograms. The models were then tested on a mammography dataset with paired standard- and low-dose images. The effect of different factors (including age, density, and dose ratio) on the differences between predictions on standard and low doses is analyzed. Methods to improve performance are assessed, and factors that reduce the model quality are demonstrated. Results: We showed that, although many factors have no significant effect on the quality of low-dose density prediction, both density and breast area have an impact. The correlation between density predictions on low- and standard-dose images of breasts with the largest breast area is 0.985 (0.949 to 0.995), whereas that with the smallest is 0.882 (0.697 to 0.961). We also demonstrated that averaging across craniocaudal-mediolateral oblique (CC-MLO) images and across repeatedly trained models can improve predictive performance. Conclusions: Low-dose mammography can be used to produce density and risk estimates that are comparable to standard-dose images. Averaging across CC-MLO and model predictions should improve this performance. The model quality is reduced when making predictions on denser and smaller breasts.

Purpose:Mammographic breast density is one of the strongest risk factors for cancer. Density assessed by radiologists using visual analogue scales has been shown to provide better risk predictions than other methods. Our purpose is to build automated models using deep learning and train on radiologist scores to make accurate and consistent predictions.Approach:We used a dataset of almost 160,000 mammograms, each with two independent density scores made by expert medical practitioners. We used two pretrained deep networks and adapted them to produce feature vectors, which were then used for both linear and nonlinear regression to make density predictions. We also simulated an "optimal method," which allowed us to compare the quality of our results with a simulated upper bound on performance.Results:Our deep learning method produced estimates with a root mean squared error (RMSE) of 8.79 ± 0.21 . The model estimates of cancer risk perform at a similar level to human experts, within uncertainty bounds. We made comparisons between different model variants and demonstrated the high level of consistency of the model predictions. Our modeled "optimal method" produced image predictions with a RMSE of between 7.98 and 8.90 for cranial caudal images.Conclusion:We demonstrated a deep learning framework based upon a transfer learning approach to make density estimates based on radiologists' visual scores. Our approach requires modest computational resources and has the potential to be trained with limited quantities of data.

Background Risk stratification as a routine part of the NHS Breast Screening Programme (NHSBSP) could provide a better balance of benefits and harms. We developed BC-Predict, to offer women when invited to the NHSBSP, which collects standard risk factor information; mammographic density; and in a sub-sample, a Polygenic Risk Score (PRS). Methods Risk prediction was estimated primarily from self-reported questionnaires and mammographic density using the Tyrer–Cuzick risk model. Women eligible for NHSBSP were recruited. BC-Predict produced risk feedback letters, inviting women at high risk (≥8% 10-year) or moderate risk (≥5–<8% 10-year) to have appointments to discuss prevention and additional screening. Results Overall uptake of BC-Predict in screening attendees was 16.9% with 2472 consenting to the study; 76.8% of those received risk feedback within the 8-week timeframe. Recruitment was 63.2% with an onsite recruiter and paper questionnaire compared to <10% with BC-Predict only ( P < 0.0001). Risk appointment attendance was highest for those at high risk (40.6%); 77.5% of those opted for preventive medication. Discussion We have shown that a real-time offer of breast cancer risk information (including both mammographic density and PRS) is feasible and can be delivered in reasonable time, although uptake requires personal contact. Preventive medication uptake in women newly identified at high risk is high and could improve the cost-effectiveness of risk stratification. Trial registration Retrospectively registered with clinicaltrials.gov (NCT04359420).

The performance of commercially available segmentation tools (deep learning and atlas-based) were assessed for breast contouring of young lymphoma patients on CT. Dice similarity coefficient, mean distance to agreement and Hausdorff distance were used to analyse performance. Deep learning segmentation performed better on more patients (6/10) but atlas-based segmentation performed best on 2/10. The variation of breast densities and arm positions likely affects auto-contouring performance in young lymphoma patients, as atlas libraries struggle to encompass the wide variation, and deep learning training data is typically of older breast cancer patients.

Background Mammographic density (MD) phenotypes, including percent density (PMD), area of dense tissue (DA), and area of non-dense tissue (NDA), are associated with breast cancer risk. Twin studies suggest that MD phenotypes are highly heritable. However, only a small proportion of their variance is explained by identified genetic variants. Methods We conducted a genome-wide association study, as well as a transcriptome-wide association study (TWAS), of age- and BMI-adjusted DA, NDA, and PMD in up to 27,900 European-ancestry women from the MODE/BCAC consortia. Results We identified 28 genome-wide significant loci for MD phenotypes, including nine novel signals (5q11.2, 5q14.1, 5q31.1, 5q33.3, 5q35.1, 7p11.2, 8q24.13, 12p11.2, 16q12.2). Further, 45% of all known breast cancer SNPs were associated with at least one MD phenotype at p < 0.05. TWAS further identified two novel genes ( SHOX2 and CRISPLD2 ) whose genetically predicted expression was significantly associated with MD phenotypes. Conclusions Our findings provided novel insight into the genetic background of MD phenotypes, and further demonstrated their shared genetic basis with breast cancer.