Factors Associated with Mammographic Breast Density among Premenopausal and Menopausal Women in Karachi Pakistan

Background: There are no studies done to evaluate the distribution of mammographic breast density and factors associated with it among Pakistani women. Methods: Participants included 477 women, who had received either diagnostic or screening mammography at two hospitals in Karachi Pakistan. Mammographic breast density was assessed using the Breast Imaging Reporting and Data System. In person interviews were conducted using a detailed questionnaire, to assess risk factors of interest, and venous blood was collected to measure serum vitamin D level at the end of the interview. To determine the association of potential factors with mammographic breast density, multivariable polytomous logistic regression was used. Results: High-density mammographic breast density (heterogeneously and dense categories) was found in 62.4 % of women. There was a signicant association of both heterogeneously dense and dense breasts with women of a younger age group < 45 years (OR= 2.68, 95% CI= 1.60-4.49) and (OR= 4.83, 95% CI= 2.54-9.16) respectively. Women with heterogeneously dense and dense breasts vs. fatty and broglandular breasts had a higher history of benign breast disease (OR = 1.90, 95% CI= 1.14–3.17) and (OR= 3.61, 95%CI= 1.90-6.86) respectively. There was an inverse relationship between breast density and body mass index. Women with dense breasts and heterogeneously dense breasts had lower body mass index (OR= 0.94 95% CI= 0.90-0.99) & (OR= 0.81, 95%CI= 0.76-0.87) respectively. There was no association of mammographic breast density with serum vitamin D levels, diet, and breast cancer. Conclusion: The ndings of a positive association of higher mammographic density with younger age and benign breast and a AH:atypical hyperplasia, AFB:Age at rst birth, BI-RADS:Breast Imaging- Reporting and Data System, BMI:Body mass index, BBD:benign breast disease, CC:craniocaudal, CR:computed radiography, DR:digital radiography, ER:estrogen receptor, FFQ:Food Frequency questionnaire, HER2:human epidermal growth factor 2, MBD:Mammographic breast density, MLO:mediolateral oblique, PR:progesterone receptor, SES:Socioeconomic Status.

has been a lot of interest in the evaluation of factors associated with MBD including the role of environment and genetics and the causal relationship between breast cancer and MBD.
Several factors can affect MBD like age, heredity, parity, ethnicity, diet, hormonal replacement therapy (HRT), and the molecular subtypes of breast cancer (6, 7). An increasing parity status has been inversely associated with MBD and the percent breast density (8). MG del Carmen et al. compared breast density among different races and reported MBD to be lowest in white and African American women and highest in Asian women (9). Race and ethnicity have been explored and identi ed as important determinants of MBD in another study too (10). Though race and ethnicity are relevant to breast density, other factors such as diet, and environmental exposures are also important determinants of mammographic density and hence the risk of breast cancer is different in different ethnic groups (11). There is an inverse association of body weight with the percentage of MBD in both pre and post-menopausal women due to more fat in the breast. It was recently con rmed in another retrospective study showing association of surgical weight loss with a decrease in breast density (12). Few other studies reported a higher intake of fats, protein, alcohol, and Hormonal Replacement Therapy (HRT) with higher MBD (13,14). Tamoxifen, however, reduced MBD in short term, and in a randomized control trial, a 10% reduction in MBD with a 63% breast cancer risk reduction was reported (15). MBD is also reported to be heritable and has shown a correlation coe cient of 0.74 in monozygotic twins in comparison to 0.38 in dizygotic twins (16).
However, the exact in uence of the environment and individual's behavior on this heritable effect is still not clearly known.
Due to limited screening mammography units and screening practices, there is a paucity of epidemiological data on MBD prevalence and factors associated with MBD in Pakistan. Therefore, the joint effects of sociodemographic and other factors associated with MBD among Pakistani women remain unclear. The main objective of this study was to evaluate the factors associated with MBD including serum level of vitamin D, intake of different food categories, physical activity, body mass index, and other risk factors of breast cancer. Another objective was to assess the association of MBD with breast cancer among Pakistani women.

Materials And Methods
The source population of this study is from a multicenter hospital-based case-control study which was conducted at two large tertiary care hospitals of Karachi as previously described (17). 477 women with complete records of MBD were extracted from the main study. The inclusion criteria of 178 breast cancer cases were newly diagnosed primary histologically con rmed cases of breast cancer from Aug 2015 to July 2018. Women who were extremely sick, unable to complete the interview, been living outside Pakistan for more than a year, receiving adjuvant or neoadjuvant chemotherapy or radiation, and diagnosis of breast cancer for greater than 6 months before study enrolment were excluded. A total of 299 controls were enrolled from those attending in-and out-patient services for general medical, and surgical departments of the two participating hospitals and who had no previous diagnosis of breast cancer or any other cancer. There is no organized breast cancer screening program in Pakistan and most of the mammography done is either diagnostic or opportunistic screening (18). Therefore, study participants were women undergoing both diagnostic and screening (annual and biennial) mammograms. All mammograms were taken before breast cancer diagnosis among cases. All subjects completed an interview-based questionnaire that included information on age, education, socioeconomic status, parity, age of mother at rst birth, breastfeeding, age at menarche and menopause, age of mother at rst birth, history of any comorbid or benign breast disease, family history of breast cancer.
Menopausal status was either premenopausal or postmenopausal. Participants also reported the average number of hours per week, engaged in physical activity of different intensities for at least ten minutes, like vigorous exercise or moderate exercise of household activities like mopping, etc., and walking. Body mass index BMI (kg/m2) and tumor characteristics were recorded from medical les and reports. Women with missing information of BI-RADS (Breast Imaging-Reporting and Data System) density were excluded.
All consenting participants were interviewed, after informed consent, in a separate room to ensure privacy using a structured questionnaire Mammography measurement details Two view mammography was performed for all patients comprising of medio-lateral oblique (MLO) and cranio-caudal (CC) views on a computed radiography (CR) system. In the CR system, the X-Rays passing through the breast, grid, and cassette cover are absorbed by the plate reader system that comprises photostimulable storage phosphor (PSP). An electronic latent image is produced on the PSP due to the local absorption of X-ray energy that varies with the anatomical variation of breast parenchyma. The cassette is subsequently placed in the reader that captures the information and converts it to a digital signal that is nally displayed at the workstation (19). The soft copies of the mammographic images were downloaded and reviewed at picture archiving and communication system (PACS) by breast imagers, and qualitative assessment of mammographic density was done by dedicated radiologists with years of experience in interpreting mammography and breast density.

Assessment of Mammographic Breast Density (MBD)
Using the American College of Radiology Breast Imaging Reporting and Data System (BI-RADS 5th edition), the mammographic breast density on mammograms was categorized as; Category 1, Predominantly fatty ( less than 25% glandular); category 2 for scattered broglandular (25 to 50% glandular ); category 3 for heterogeneously dense (51 to 75% glandular) and category 4 for dense breast parenchyma ( more than 75% glandular) (20). Categories 3& 4 both are high MBD. Analyses were restricted to patients with an available BI-RADS measurement. The density measurements of the breast contralateral to the tumor were used to avoid a distortion of measurements due to the tumor itself.
Dietary intake assessment with a Food Frequency Questionnaire (FFQ): Dietary intake assessment was done by using the validated food frequency questionnaire FFQ (21).
Intake frequency was categorized into 7 groups and category for each food item was converted to daily intake. Each participant was also asked about their average portion size/ common serving size of the food. The intake frequencies were multiplied by standard portion size to calculate servings per day of all food items. All the food items were grouped into 6 components including fruits, vegetables, dairy, grains, white meat, red meat, and plant proteins and total servings/day was calculated for each food category.

Measurement of serum 25 (OH)D level
After the interview, blood samples were collected from the study participants and serum 25 (OH)D level was measured using ELISA.

Tumor characteristics
Histopathology and estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor (HER2/ neu) status of breast cancer cases were retrieved from medical records (22).

The ethical approval was obtained by the Human Research Ethics Committee of the University of Adelaide and the Ethical Review Committees of two hospitals in Karachi Pakistan: Aga Khan University
Hospital AKUH & Karachi Institute of Radiation and Nuclear Medicine Hospital KIRAN. Patients who were literate read and signed the informed consent form and informed consent was obtained verbally from those who could not read or write.

Statistical analysis
Multinomial logistic regression models were applied to compute odds ratios (ORs) and 95% con dence intervals (CIs) for the MBD categories. The fatty and scattered broglandular tissue categories were merged and used as a reference. Multivariate models were adjusted for variables found to be signi cantly associated with breast density and known risk factors for breast cancer such as age, body mass index, age at menarche & menopause in the postmenopausal group, parity, and family history of breast cancer among rst-degree relatives, BMI, and the food categories.

Results
The mean age of the study participants was 46.2 years (SD 11.7). 180 (37.3 %) women had breast biopsy-proven breast cancer and 297 (62.7%) of the women had normal or benign breast disease. Figure 2 shows that high density MBD (heterogeneously and dense categories) accounted for 62.4 % of all participants. This is higher than the reported percentage among women in the USA (23) and lower than the reported percentage among women in Jordon (24). Table 1 shows that there were signi cant differences in age, education, parity, breastfeeding, body weight, BMI, parity, history of breastfeeding, history of comorbid, menopausal status, among women of different breast densities. In both heterogeneously dense & dense groups, there was a higher percentage of women who were of younger age categories, nullipara, lower parity, higher education, and had a family history of breast cancer.  There was a positive association of MBD with young age, nulliparity & low parity, family history of breast cancer and history of benign breast disease BBD. Signi cant protective association of MBD was observed for women of younger age, menopausal status, younger age at rst birth (< 20 years AFB) and BMI (Table 3). Fatty and scattered glandular combined as the reference group. BI-RADS: Breast Imaging Reporting and Data System (fatty and scattered glandular dense = categories A & B, heterogeneously dense = C, dense = D); OR: odds ratio; 95% CI: 95% con dence interval AFB Age at rst birth* mean (SD) **hours/week

Diet and breast density
There was no association reported between different MBD and food categories of grains, fruits, vegetables, plant proteins, dairy products, white and red meat as shown in Table 4. Fatty and scattered glandular combined as the reference group. OR: odds ratio; 95% CI: 95% con dence interval Multivariable analysis using multinomial logistic regression (table 5) shows that women with dense and heterogeneously dense breasts vs. fatty and broglandular breasts were of a younger age group and had higher benign breast disease BBD. An inverse relationship of BMI with MBD was also observed. There was no association of serum vitamin D levels or breast cancer with breast density.

Discussion
The higher percentage of high MBD in Pakistani women is important information and will be helpful in planning or advising for an organized screening mammography program in Pakistan in future. High density breast, on one hand, increases the rates of false-negative diagnosis and on the other hand, there is a requirement of additional tests. The need for additional tests like ultrasound, & MRI increases the sensitivity of screening programs (25) but the cost then becomes too high for the majority of the asymptomatic women to opt for regular screening as there is no health insurance and high poverty in Pakistan.
Our study nding of signi cant association of younger age with high density is important because there is no clear screening guidelines established for Pakistani women where mean age at the time of breast cancer diagnosis reported is also much younger (46.1 years ± SD 10.1) (26). Mammographic breast density was also found to be positively associated with BBD and negatively associated with BMI. In univariate analysis, lower parity, breastfeeding, and family history of breast cancer also showed a signi cant protective association with dense breast. However, in a multivariate analysis performed to adjust for confounding factors, only younger age, BBD, and BMI remained as signi cant factors associated with high breast density.
Our nding of younger age as an important determinant of a high MBD is similar to other studies done in different populations (27)(28)(29)(30)(31). In another study, there was a decline in density reported with age in women with and without breast cancer (32). It supports the hypothesis that breast density declines with age as younger women have a higher proportion of dense breast tissue compared to older women (6). In a study done on data from the San Francisco Mammography Registry, breast density decreased with age on an annual basis over the perimenopausal stage (33).
In our study, the inverse relationship between MBD and higher BMI remained the same in both menopausal and premenopausal women. The inverse relationship between BMI and MBD has been reported in previous studies among both menopausal and premenopausal women (27,34,35). In the Minnesota Breast Cancer Family Study, higher BMI was associated with lower breast density among postmenopausal women only (36). BMI was also inversely related to MBD in Chinese women (37). An inverse relationship of greater body weight with percentage mammographic density was also reported in premenopausal women, with mammogram showing a larger area of fatty tissue caused by an increased quantity of fat in the breast (38) Similar to our study ndings, a study among Japanese women reported that younger age and BMI were inversely related to high MBD (39). In another breast cancer family study of 426 families, younger age and lower body mass index were associated with increased MBD in both premenopausal and postmenopausal women. Hormone replacement therapy among postmenopausal women was also associated with MBD but this association of MBD with hormone replacement therapy was not seen in the current study as HRT use is very small among Pakistani women. In a study in Norway, volumetric mammographic density (VMD) was inversely associated with age and BMI similar to our study ndings (40,41). A negative association of high MBD with age, low parity and BMI, was similarly reported by other studies (42,43).
Our ndings provide important data on the association of benign breast disease with higher mammographic breast density and are similar to other studies ( (41,44,45). However, this study observed no association between MBD and breast cancer, and this lack of association has also been reported in a study done in the USA (46) and one in China (46). A nested case-control study in Ontario, Canada also reported no association of MBD with breast cancer among women with BRCA mutation (47). Another study at the Massachusetts General Hospital showed that while Asians had the highest breast density, the incidence of breast cancer among them was lesser than that among white women (13). An observational study of the Mayo Clinic Benign Breast Disease (BBD) cohort failed to nd any evidence of an association between MBD and breast cancer risk in women diagnosed with BBD of atypical hyperplasia (AH) type (48). Similarly, another study done at Johns Hopkins showed that breast cancer was not associated with MRI breast density (49). The study ndings did not show any association of MBD with vitamin D levels, similar to the Nurses' Health Study (50). There was also no association observed between MBD and diet.

Limitations
There was no facility available for digital radiography (DR) which has been installed just recently after the completion of the data collection period of the study. Given the current role of breast density in determining breast cancer screening protocols, public health policy, and future research directions, it is important to validate our ndings in a larger scale investigation with the advanced technology of the DR system and assess if it is better than screen-lm mammography in women with dense breasts. We could not calculate percentage density and absolute dense area due to the lack of availability of the software.
There is still a lack of standardization in MBD assessment with high density de nitions varying widely from 25-75% of dense tissues on mammograms in different studies. Though we tried our best to evaluate MBD with standard mammographic procedures, breast density classi cation as well as a standardized de nition of MBD. Still, some misclassi cation of MBD could have affected our results since we used visual classi cation using BI RADS by two experienced radiologists. However, a study reported very good agreement between automatic assessment software of breast density based on arti cial intelligence (AI) and visual assessment by a senior and a junior radiologist (51). The strengths of the study are a good quality of data collection by medical doctors, fair sample size, and comprehensive assessment of all factors associated with MBD.
Conclusion and recommendations To our knowledge, this study is the rst to report unique distribution of MBD and identify factors associated with MBD in Pakistani women. The ndings of positive association of higher mammographic density younger age and BBD and negative association between BMI and breast density are consistent with predictors of mammographic density observed in other populations; however, certain risk factors were not signi cantly associated with BMD. These are important ndings which may be helpful to develop screening guidelines for Pakistani population. Pie chart depicting the distribution of MBD among Pakistani women.

Figure 4
Women age and MBD as de ned by ACR BI-RADS. ACR BI-RADS indicates American College of Radiology Breast Imaging-Reporting and Data System; MBD, mammographic breast density.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.