**4. Discussion**

Our results demonstrated that it was feasible to incorporate BTD in high-risk assessment for BC. Including BTD in risk assessment optimized supplemental high-risk screening. No patients with non-dense tissue became eligible and 16% (8/50) ineligible, while 12% (10/81) with dense BTD became eligible, only 6% (5/81) (category C) ineligible. More patients with dense BTD were eligible while fewer with non-dense BTD required B-MRI, with an overall reduction of 3.1% (4/131) HROBSP eligible women. Although including density impacted eligibility for 17% women, the overall number of eligible women with vs. without density was not significantly different (*p* = 0.3938). Despite having been referred for a high-risk assessment for BC, 5.8% (8/139) women, and the *BRCA* carrier, had never undergone mammography.

It is well known that sensitivity of mammography is reduced in women with dense breasts, decreasing as BTD increases, 81–93% for fatty (A), 84–90% for scattered fibroglandular densities (B), 69–81% for heterogeneously dense (C) and 57–71% for extremely dense (D) breasts in women 40–74 years of age [14]. The interval cancer rate (BC detected after a normal screening study) is significantly higher in women with the most dense breasts when screened every 2 years versus every year [15]. B-MRI screening every 3–4 years in addition to mammography is cost-effective in average risk women who have the most dense breasts [16]. For women at high-risk, B-MRI is essential to permit early stage detection of BC and to reduce BC mortality. Our study found that more women with dense BTD became eligible for B-MRI while more with non-dense became ineligible, with no overall impact on use of B-MRI. Because of the masking effect of BTD, where BCs are obscured by dense tissue on mammography, contrast-enhanced B-MRI is required for early detection of BC in women with dense BTD [17].

Initial IBIS models did not incorporate BTD (v7,2004) but were updated in 2018 (v8), when BTD was shown to be more accurate in long-term assessment of BC risk [12]. Recently, researchers showed that risk stratification is improved when adding volumetric and visual mammographic density [18]. Destounis found a significantly higher proportion of high-risk women (defined as LTR ≥20%) when incorporating BI-RADS into IBISv8 compared with v7 (11.4% vs. 8.3% *p* < 0.001) [19]. In this study, fewer women with non-dense BTD were included in the high-risk category. A case–control study in 2019 of 474 patient participants and 2243 healthy control participants) of women aged 40–79 years found more women were included in the high-risk category, using IBISv8 instead of IBISv7 (7.1% vs. 4.8%) [20]. The

Brentnall study defined high risk as 8% 10-year or >20% lifetime risk while our study used the OBSP definition of high-risk as ≥25%. It would have been interesting to compare the impact using a similar risk assessment of ≥20%. In their study, BI-RADS assessment was a better predictor of risk than volumetric assessment of mammographic density. Our study however was a prospective evaluation of women already referred for high-risk assessment. IBISv8 is used by the HROBSP but the BTD information had not been widely incorporated as it was not being provided to the GCs performing the assessment. We showed that the HROBSP NN could provide available mammogram reports with the referral for HROBSP genetics' assessment. This occurred with 94.2% (130/138) patients assessed, which allowed for easy incorporation of BTD. Only 13.1% (17/130) of women had a prior mammogram but the report was unavailable to the GC and required radiologist input. In this instance, in took about 5 min for the GC and the radiologist to converse with each other to obtain this information.

Recently, as of October 2021, OBSP has made it mandatory for all mammograms reported through the OBSP to include the BTD according to BI-RADS categories. However, at present there is no mandatory requirement for BTD to be reported in diagnostic mammograms or screening mammogram done outside the OBSP program. BTD reporting should be mandatory for all mammograms done in Canada. In future, given these mandatory reporting requirements, it is expected that input from the radiologist will further decrease over time. In this prospective study, 16.9% (22/130) of women were clinically impacted by the incorporation of BTD with 4 fewer women requiring B-MRI. These results were consistent with the retrospective data (14.4% impacted and 4% fewer eligible women).

It is notable that 8 women from our prospective study and 8 from the retrospective data had never had a mammogram. In Ontario, population screening for BC with biennial mammograms begins at age 50. Women eligible for HROBSP assessment have a strong family history of breast and/or ovarian cancer and it is generally recommended that these women start annual mammograms (outside of OBSP) beginning at age 40 or 5–10 years prior to the earliest BC diagnosis in the family [21]. It should be noted that women considered at high risk are not included in the 2018 Canadian Task Force Preventive Health Care Guidelines [22]. Requiring a baseline mammogram prior to acceptance for HROBSP assessment would help to educate primary care providers and allow all women the opportunity to have BTD incorporated into their HROBSP assessment.

We recognize several limitations in our study. Inter-observer variation is well known in visual categorizing of BTD, and automated BTD was not used in our study. The 2013 BI-RADS 5th ed. density classification is based on the masking effect of BTD, while the 2003 BI-RADS 4th ed. [23] was based on visually estimated percentage BTD; both are associated with high rates of inter- and intra-observer variability [24,25]. Future considerations for OBSP include whether women's BC risk needs to be reassessed at some point, for example after menopause when many women's BTD decreases. There are also many women aged 30–39 y who were referred for HROBSP assessment to our centre. Neither IBISv8 or CanRisk are validated for inclusion of BTD for this age group and it is unknown whether they would benefit from reassessment after age 40. In May 2021, the BOADICEA tool was incorporated into the CanRisk tool which can now be used for unaffected women and includes similar personal risk factors as IBISv8 including BTD. The CanRisk tool provides some advantages to the genetic clinic assessment. For example, It allows for incorporation affected relatives with pancreatic and prostate cancer in addition to breast and ovarian and provides a mutation carrier likelihood for five BC risk genes (*BRCA1*, *BRCA2*, *PALB2*, *ATM*, *CHEK2*), and three ovarian cancer risk genes (*RAD51C*, *RAD51D*, *BRIP1*) compared to *BRCA1* and *BRCA2* only with IBISv8 ([26]). Unlike IBIS, CanRisk, does not however, allow for incorporation of benign breast disease such as atypical ductal hyperplasia and lobular carcinoma in situ, which are significant risk factors for BC. Which tool should be used for calculating HROBSP for unaffected women in our population is an ongoing question. Our study is limited by the small sample size but comparison with our retrospective data showed consistent results that are also in keeping with the recent larger data sets from

Brentnall and Destounis' studies [19,20]. Lastly, our study was limited by the lack of an outcomes audit and information about the stage of breast cancer in women previously assessed for eligibility in HROBSP. Including outcome information might prompt further evaluation of the eligibility threshold.
