This study showed that women with higher BD had a reduced risk of death from all causes and breast cancer.
The role of BD as prognostic factor has been addressed in several studies with, however, inconclusive results. A prospective analysis on the Breast Cancer Surveillance Consortium data of over 9000 women with invasive breast carcinoma showed that BI-RADS density (comparing BI-RADS 4 versus BI-RADS 2) was not related to the risk of death from BC (HR 0.92 95% CI 0.71-1.19) or any other cause (HR 0.83 95% CI 0.68-1.02), after accounting for some patient and tumour characteristics (site, age at and year of diagnosis, stage, BMI, mode of detection, treatment and income) [12]. Pre-diagnostic BD, assessed using a computer-assisted method, among 607 BC cases within the Hawaii component of the Multiethnic Cohort was not associated with death from BC (HR 0.95 per 10% 95% CI 0.79-1.15) and from other causes (HR 1.08 per 10% 95% CI 0.98-1.20); the model was adjusted for age at diagnosis, ethnicity, overweight, stage at diagnosis and radiation treatment [13]. Similarly, in a British hospital–based study of 759 women aged 50-69 with primary operable invasive BCs, BC–specific survival was unrelated to BI-RADS mammographic parenchymal pattern. A nonsignificant trend was observed for women with denser breasts who showed a better overall survival than women with fatty breasts [14]. Moreover, a Swedish study observed that high BD, assessed according to Tabar’s classification, was suggestively, but not significantly, associated with poorer long-term survival after adjustments for age, tumour size, node status, grade, and BMI (HR 1.75 95% CI 0.99-3.10) [15]. Focusing on BC related deaths, a cohort study on 22,597 African American and White women with BC enrolled from the Carolina Mammography Registry showed no association with BD (HR 0.91, p= 0.124 dense [BI-RADS 3-4] versus fatty [BI-RADS 1-2] adjusted for age, ethnicity, and tumour stage) [16].
Partially in contrast with our results, an analysis including 619 BC cases selected from a prospective cohort (The Malmö Diet and Cancer Study) showed that, after adjustment for age and other prognostic factors, women aged 50-69 with dense breast (BI-RADS 4), as compared to fatty one (BI-RADS 1), had an increased risk of BC related death (HR 2.56 95% CI 1.07-6.11). In the same study, when deaths from causes other than BC were considered, the HR was 0.74 (95% CI 0.31-1.73) [17]. Moreover, including only women diagnosed with metastatic BC, a Saudi Arabian study observed that moderate/high BD patients (>25% of radio-dense fibroglandular tissue) had a worse median progression-free survival than low density ones (9.3 months, 95% CI 8.51–13.60 vs 18.4 months, 95% CI 14.88–22.15 respectively, p = 0.002) [18].
The association that we observed between lower BD and poorer outcome is consistent with several studies. In a Finnish analysis assessed on 270 patients aged 32 to 86 with newly diagnosed BC, very low BD at the time of diagnosis (percentage of glandular tissue <10% ) was an independent, poor prognostic factor even after correcting for possible confounders (HR 3.28 95% CI 1.75-6.13 compared to the remaining patients) [19]. In a Korean study on 969 patients with primary operable invasive BC, the high density group (BI-RADS 3-4) demonstrated a superior overall survival compared to the lower one (BI-RADS 1-2) after adjustment for 14 factors including nine clinicopathologic factors and five treatment factors (HR 0.38 95% CI 0.21–0.71) [20]. A cohort study on 989 BC patients aged 50-69 identified within the Danish mammography screening program showed that patients with dense breast (BI-RADS 3-4), compared to those with fatty breast (BI-RADS 1-2), had a lower case fatality (case fatality rate ratio 0.60, 95% CI 0.43-0.84), as well as a reduced risk of BC death (age-adjusted RR 0.53, 95% CI 0.34-0.82) [21]. A case-only study including 2233 women diagnosed with invasive breast aged from 38 to 97 showed, after adjustment for age and mode of detection, an association of borderline statistical significance between high BD and BC -specific survival (HR 0.84 95% CI 0.68-1.03 for dense breast, Wolfe scale > 25%, versus fatty breasts, Wolfe scale ≤25%) [22].
Evidence that women with lower BD are at increased risk of BC-related death, irrespective of several well-known risk factors, supports the growing evidence that tumour microenvironment is closely intertwined with the outcome of the disease. Until now, studies have widely explored characteristics of dense breast, while less attention has been given to the fat of breast tissue. Adipose tissue is an effective endocrine organ able to secrete a variety of bioactive molecules. The paracrine support of the tumours by cytokines and growth factors secreted by adipocytes, as well as the chronic low-grade inflammation sustained in surrounding tissue by the peritumoral fat, may, at least in part, enhance malignant progression mechanisms [23,24].
Limitations of this study included BD assessment that relied on BI-RADS classification, a visual and subjective method; held by a single reader, no formal assessment of intra or inter-observer variability was performed. Nevertheless, the BI-RADS reporting system allows to rank BC patients into interpretable descriptors without requiring special software. Further, due to known BD changes with age, we assessed BD at the time of diagnosis. We had no information on mode of detection (i.e., screening and interval cancers), treatment received and cancer recurrence. However, we assessed a comprehensive spectrum of BC related prognostic factors, including detailed tumour characteristics and several patient aspects (e.g., reproductive, sociodemographic, anthropometric, lifestyle factors). Prior studies showed remarkable differences in populations and methodologies, in BD assessment method and classification, as well as in adjustment for confounding factors. These disparities limited comparison between studies and could partly explain a lack of concordance across them.
In conclusion, low BD appears to affect survival in women with invasive BC, even after adjusting for several known prognostic factors. Readily available data on BD at diagnosis may provide useful prognostic information.