Mammographic density and risk of breast cancer recurrence and mortality.

To the Editor: Breast cancer is the most commonly diagnosed cancer and the leading cause of cancer death among females. Due to the advantages of low pain, high resolution, and good reproducibility, mammography is widely used in early breast cancer clinical screening. Mammography density (MD) is the proportion of dense breast tissue in mammograms and is one of the strongest risk factors for breast cancer. Studies have shown that high MD increases breast cancer risk by 4 to 6 times. A previous meta-analysis of 13 case-control studies showed a positive correlation (odds ratio [OR]: 1.52, 95% confidence interval [CI]: 1.39-1.66) between percentage dense area and breast cancer risk,[ but it did not assess the relationship between MD and breast cancer prognosis.

Several studies have suggested that MD is associated with the prognosis of breast cancer, but the results are inconsistent. A review published in 2014 described the association between MD and breast cancer prognosis, and it concluded that MD is associated with an increased risk of local recurrence but is not related to survival.[ However, this review only extracted the data and did not conduct a systematic evaluation. Therefore, we conducted a meta–analysis and systematic review to evaluate the association between MD and breast cancer outcomes.

In the present meta-analysis, the literature was obtained by searching the PubMed, ISI Web of Knowledge, and Embase databases (up to April 2019). The search terms were as follows: (1) mammographic breast density, mammographic density, and breast density; (2) prognosis, survival, death, mortality, outcome, recurrence, and metastasis; and (3) breast cancer, breast carcinoma, breast tumor, breast neoplasm, mammary cancer, mammary carcinoma, mammary tumor, and mammary neoplasm. Studies from the reference lists of related reviews and original papers were also searched manually. The inclusion criteria for these studies were as follows: (1) studies evaluating the association between MD and breast cancer prognosis; (2) studies containing harzard ratio (HR) and 95% CI; (3) studies meeting the classification standard (≥25% for high MD and <25% for low MD); and (4) studies written in English. The exclusion criteria were as follows: (1) studies that did not include MD or breast cancer prognosis and (2) reviews, reports, or meeting abstracts.

The methods for breast density definition are diverse, including Breast Imaging Reporting and Data System (BI-RADS), Wolfe, Tabár, percent density, and volumetric breast density. The classifications of BI-RADS, Wolfe, and Tabár are shown in Supplementary Table 1. In our study, the cutoff point of high and low MD was 25%, which was consistent with a previous meta-analysis of MD and breast cancer risk.[ Mammary gland density is <25%, and mammary glands generally appear almost entirely as fat in mammograms, which plays an important role in relieving the masking effect and diagnosing cancer. Higher breast density indicates a higher risk of breast cancer, and it promotes the occurrence and development of breast cancer. The literature search was assessed by two reviewers (Liu and Sun). We first filtered according to the title or abstract if it met the inclusion criteria, and we then further screened the full text and extracted the information. The HRs and 95% CIs were intuitively evaluated by forest plots. The Cochrane Q statistic (P > 0.10 was considered statistical heterogeneity) and the I2 statistic (25%, 50%, and 75% were considered to represent low, moderate, and high heterogeneity, respectively) were used to assess heterogeneity of HRs. A fixed-effect model was used if there was no heterogeneity, otherwise the random-effects model. All analyses were assessed with STATA version 13.0 (Stata- Corp LP, College Station, TX, USA), and P values <0.05 were considered significant.

Initially, a total of 1156 articles were found. After filtering according to titles and abstracts, 49 articles were selected for further evaluation based on the full text, and 27 articles were excluded. Eight studies were excluded due to incomplete data or did not meet the classification standard. Finally, 14 articles were included in the meta-analysis [Supplementary Figure 1]. In the selected articles, a study was retrieved from the reference list. The 14 included studies reported on the following topics: 7 studies reported on MD and breast cancer mortality; 4 studies reported on MD and breast cancer recurrence; 1 study simultaneously reported on MD, breast cancer mortality, and breast cancer recurrence; and 4 studies reported on mammographic density reduction (MDR) and breast cancer outcome.

The characteristics of the seven selected articles about MD and breast cancer mortality are listed in Supplementary Table 2 (upper section). The total number of participants was 14,384 with 1462 reported breast cancer deaths. Most of the studies were conducted in the United States and European countries. Except for one study that had a case-only design, the other studies had cohort designs. The mean follow-up duration was longer than 6.4 years. The mean age at baseline was >56.7 years, ranging from 32 to 86 years. Three studies used BI-RADS for classification; three studies used percent density for classification; and one study used Tabár for classification. The studies on MD and breast cancer recurrence are shown in Supplementary Table 2 (middle section). The meta-analysis included 2305 participants and 151 breast cancer recurrence patients. Two studies were from North America, one in Saudi Arabia, and one in Sweden. The median follow-up duration was longer than 18 months. Two studies had cohort designs; one study had a case-only design; and one study had a nested case-control design. The mean age at baseline was >40 years, ranging from 33 years to 87 years. For mammographic features, two studies used Wolfe for classification; and two studies used percent density for classification. Supplementary Table 2 lists the studies that reported on MDR and breast cancer outcomes. Two studies reported on breast cancer death, and two studies reported on recurrence. The numbers of participants and events were 3454 and 365, respectively. Two studies were conducted in Korea; one study in the United States; and one study in Sweden. The mean follow-up time was >18 months. Three studies had cohort designs, and one study had a case-only design. The mean age at baseline was over 45.3 years and ranged from 24 to 78 years. Three studies used percent density for classification, and one study used BI-RADS for classification. The study that reported on MDR included two mammograms as follows: one was a baseline mammogram, and the other was a follow-up mammogram after adjuvant endocrine treatment (tamoxifen). MDR is the difference between the follow-up mammogram and baseline mammogram. The magnitude of MD reduction in the four articles was >0%, >0.5%, and >10%.

Among the seven reports on breast cancer mortality, four studies reported a positive association (HR > 1.00) with two of them being statistically significant. Three studies reported HRs of <1.00, but only one study was statistically significant. High heterogeneity was detected with an I2 = 77.8% (Cochrane Q statistic = 26.98, P < 0.001), and the pooled HR from the random-effects model was 1.21 (95% CI:0.83–1.77) [Supplementary Figure2A]. Associations between MD and breast cancer recurrence were available from four studies with a pooled HR of 2.84 (95% CI: 1.89–4.25) from a fixed- effect model [Supplementary Figure 2B]. Moderate heterogeneity was detected with an I2 = 39.0% (Cochrane Q statistic = 4.92, P = 0.178). Of the four articles on MDR and breast cancer outcome, two studies reported results on breast cancer mortality, and two studies reported on recurrence. The pooled Hr from the fixed-effect model was 0.50 (95% CI: 0.36–0.68) [Supplementary Figure 2C]. Low heterogeneity was found with an I2 = 0.00% (Cochrane Q statistic = 1.59, P = 0.662). The sensitivity analysis showed that the main results were not affected after omitting one study in each round. We used a funnel plot to visually inspect the asymmetry [Supplementary Figure 3], and the Begg's test was not significant (MD and breast cancer mortality [z = 0.00; P = 1.000]; MD and breast cancer recurrence [z = 1.70; P = 0.09]; or MDR and breast cancer outcome [z = −0.34; P = 1.000]).

The exact mechanism of mammographic density and breast cancer prognosis remains unclear, whereas genetic factors may play a role. A study has uncovered an association between genetic polymorphisms and mammographic density.[ In addition, mammographic density can be altered by endogenous and exogenous hormonal factors, which may influence epithelial and/or stroma- related processes. It is assumed that such factors may also influence breast cancer prognosis. Furthermore, high MD indicates less adipose tissue. Studies have reported the role of lipids and lipogenic pathway regulation in breast cancer, which may be of potential clinical implications for prognosis. The increased mortality and local recurrence are likely due to delays in diagnosis in dense breasts, and studies have reported that women with BI-RADS 1 density are slightly more likely to be diagnosed with AJCC stage IV breast cancer. However, the relationship between MD and breast cancer prognosis needs further investigation.

Although epidemiological research has revealed that reduced breast density may improve the prognosis of breast cancer, molecular mechanisms still need to be explored. Our meta-analysis included breast cancer patients receiving adjuvant endocrine therapy primarily with tamoxifen. Tamoxifen competitively binds to estrogen receptors, which antagonizes estrogen effects, thereby suppressing the progression of breast cancer. Previous studies have found that MD is influenced by estrogenic activity and that it is responsive to tamoxifen antiestrogen effects.[ In addition, tamoxifen may play an essential role in the density decrease by regulating the expression of tumor growth factor-β1 and insulin-like growth factor. Further analysis is still needed to clarify the relationship among tamoxifen, density changes, and prognosis.

The advantage of our study is that we analyzed not only the association between MD and breast cancer outcome but also the association between MDR and breast cancer outcome, which may have greater clinical significance. Another advantage of our study is that the articles included in the meta-analysis had no potential publication bias. However, there were several limitations. First, the number of studies included in the meta-analysis was limited, especially with three different outcomes. Second, several studies included in the meta-analysis failed to extract complete information, such as follow-up time and baseline age. Third, we found that significant heterogeneity may be caused by the different breast density assessments, but we attempted to align the measurement methods. Fourth, we assessed the relationship between mammographic density and breast cancer prognosis, but the influence of breast cancer treatment and subtypes was not well addressed.

This meta-analysis is not sufficient to support an inverse correlation between mammographic density and breast cancer survival. However, we cannot simply conclude that there is no association between MD and breast cancer survival based on current evidence. In addition, we found a positive association between high mammographic density and breast cancer recurrence. Moreover, we found a relationship between MDR and improvement of breast cancer prognosis, thereby providing clinical significance. Early mammography screening and reducing MD may contribute to improving the prognosis and quality of life of patients with breast cancer.

Funding

This work was supported by a grant from the Hebei Key Technology Research and Development Program (No. 19277739D).

Conflicts of interest

None.