The potential interaction of hereditary and reproductive factors in the etiology of mammary gland hyperplasia: a case-control study

Background: Family history of breast cancer and female reproductive factors may work together to influence hyperplasia of mammary gland (HMG) risk. However, the association with HMG risk is poorly characterized and might be important to understand the causation of HMG. Methods: A total of 1881 newly diagnosed HMG cases and 1900 controls were recruited between 2012 and 2017. We collected each participant's demographic characteristics, female reproductive factors and family history of breast cancer. A multi-analytic strategy combining unconditional logistic regression, multifactor dimensionality reduction (MDR) and crossover approaches were applied to systematically identify the interactions of family history of breast cancer and reproductive factors on HMG susceptibility. Results: In MDR analysis, high-order interactions among education level, breastfeeding duration and family history of breast cancer were identified among women (OR=7.069, 95%CI: 6.080-8.219). Similarly, in crossover analysis, compared with individuals having low education level and no family history of breast cancer, HMG risk increased significantly for those having high education level and family history of breast cancer (OR=36.389, 95%CI: 11.469-115.451), similar additive interaction effect was observed among short breastfeeding duration women (OR=27.699, 95%CI: 3.730-205.699). Conclusion: This study suggests high-order interactions of high education level, short breastfeeding duration and family history of breast cancer may synergistically increased HMG risk.


Introduction
Hyperplasia of mammary gland (HMG), a multi-factorial complicated disease, accounts for more than 70% of all breast disease that occurs frequently among middle-aged women and is highly associated with breast cancer 1 . Perhaps due to the quickening pace of life, with increasing work-related pressure, the prevalence of HMG is high in China 2 . Understanding indicators of HMG in middle-aged women play an important part in public health and clinical implications. Researches have identified multiple risk factors for HMG 3 , such as late age at menopause 4,5 , nulliparity 4,5 , lacking breastfeeding 6 , high-level education 7 and family history of breast cancer 8 . Nevertheless, etiology of HMG remains largely unknown. Family history of breast cancer is an important indicator for a woman's future risk of developing breast cancer 9 . Part of the genetic information can be explained by hereditary history 3,10 .
There is growing recognition that large sample sizes are needed in order to identify heredity variants that have effects modified by the environment as well 11 . Hereditary-environment interactions have the potential to illustrate the biological causes of disease, distinguish individuals for whom risk factors are most related 12 and develop precision medicine essentially 13 . However, few researches have considered interactions between family history of breast cancer and potentially modifiable risk factors for HMG. Besides, existing studies only take single statistical method to study the interactions between family history of breast cancer and potential modifiable risk factors which are lack of internal validation and decreased statistical power to identify underlying hereditary-environment interactions efficiently 14 .
Using data collected in a large community-based case-control study, we assessed the correlation of HMG with reproductive factors in women self-reporting first and second-degree relatives. We hypothesize that interactions between female reproductive factors and family history of breast cancer will confer an even greater risk of HMG. In this research, we adopted multi-analytic strategies to scientifically examine the interactions between hereditary and female reproductive factors. We used several statistical approaches, including traditional multiple logistic regression, multifactor dimensionality reduction (MDR) in addition to crossover analysis to explore high-order hereditaryenvironment interactions in HMG susceptibility.

1.Study subjects
This study is based on the National Basic Public Health Service Project which provided free of charge for urban and rural residents by Chinese government. From October 2012 to December 2017December , 1966 patients that newly diagnosed by color doppler ultrasonography confirmed hyperplasia of mammary glands (HMG) were selected as cases. Contemporaneously, 1993 HMG-free controls (imaging showed no abnormality) were chosen as controls from the community health service center of Harbin.
Inclusion criteria was female subjects newly diagnosed, aged above 35 years, living in Harbin for at least 6 months, agreed with color doppler ultrasonography examination. Patients with mastitis, angiosarcoma, tumor of mammary glands, breast cancer or other cancers were excluded. 85 cases (4.4%) and 93 controls (4.7%) were excluded because data of female reproductive information or family history of breast cancer appeared abnormal value or missing. Finally, a total of 1881 cases and 1900 controls were enrolled (Fig. 1). All pertinent clinical results were reviewed by two general practitioners to ensure the diagnosis. All participants provided informed consents and the study was approved by the Ethical Committee of Harbin Center for Disease Control and Prevention.

Data collection
Basic characteristics of demographic information ( age, ethnicity, education level, marriage, occupation, and so on), female reproductive factors including menopausal status, age at menopause, parturition and age at first delivery, age at menarche, breastfeeding and its duration, family history of breast cancer were obtained using a structured questionnaire administered by trained interviewers face-to-face. In this research, history of breast cancer was defined as breast malignancy in firstdegree or second-degree relatives (such as mother, sister, grandmothers or aunts). Regular menstruation was considered as menstrual time with 2-7 days and menstrual cycle with 24-35 days.
Menopause referred to the specific period from the appearance of endocrine, biological and clinical characteristics related to menopause to the postmenopausal period. Gland fibrous or cystic or single type of HMG was included in our study. According to the fifth editions of the American College of Radiology (ACR) Breast Imaging Reporting and Data System (BI-RADS) guidelines, the cases classified into stages Ⅱ-V were involved in our study. The details as follows: BI-RADS 2-Benign found, BI-RADS 3-Benign lesions, BI-RADS 4-Suspicion for malignancy and BI-RADS 5-Highly suggestive of malignancy.
The higher grade of lesion site (left breast, right or bilateral site of breast) was taken as BI-RADS grade.

Statistical analyses
The Odds Ratios (ORs) and corresponding 95% Confidence Intervals (95% CI) were summarized to estimate the associations between reproductive factors and HMG risk by univariate and manual stepwise multivariate logistic regression. All references values for exposure were the lower level of

Basic characteristics of HMG patients and controls
The female reproductive factors and ORs for HMG are presented in Table 1. Of the 1881 cases, 1627 were Han ethnicity and 1203 Han ethnicity accounted for the 1900 controls. The difference of mean (SD) ages between cases (51.27 ± 6.62) and controls (51.21 ± 6.62) were not significant (t = 0.27, P = 0.89). Among the study population, distributions of BMI, education, marriage and occupation type were significantly different between cases and controls (P < 0.05). Cases tended to be higher educated; more married people, mental worker and familial tendency; older age at first delivery, shorter breastfeeding duration; less breastfeeding, regular menstruation and parturition than controls. Age at menopause was also significantly different between cases and controls in the 50-54 groups. According to Breast Imaging Reporting and Data System (BI-RADS), 1646 cases were classified as stage II, 192 as stage III, 38 as stage IV, 5 as stage V HMG. BMI and occupation were treated as potential confounders and adjusted in crossover, multiplier interaction and additive interaction analyses.  Therefore, this model was regarded as the best among all the interaction models calculated by MDR.

Female reproduction and family history of breast cancer interactions and the risk of HMG
Compared with the "low-risk" combinations, participants classified as "high-risk" combinations significantly increasing HMG risk by 7.07-folds (95%CI: 6.08-8.22).       Our study is the largest case-control study up to now on HMG and provides ample evidence that the risk of HMG does not only apply to family history of breast cancer alone, but may also extend to breastfeeding duration and education level in Harbin, China. Possible interactions between hereditary and reproductive factors of HMG were noted.
We applied a variety of algorithms to explore the interactions between family history of breast cancer and female reproductive factors for which an interaction might be possible. Firstly, MDR was used to analyse the interactions between six environmental factors that were statistically significant in multiple logistic regression. High-dimensional interactions including education level, breastfeeding duration and family history of breast cancer were detected. Secondly, our results seemed biologically plausible. We found a strong synergistic effect between family history of breast cancer and high education level after adjusted for BMI and occupation. Therefore, more attention should be paid to enhance the awareness and health education among HMG women with high education level and family history of breast cancer 22 . Additionally, antagonistic effect between family history of breast cancer and breastfeeding duration was also observed which was consistent with the published literature 23 . Based on these results, women with a family history of breast cancer may reduce their excess risk of HMG through adjustments in reproductive choices 23 . Thirdly, because the additive model might be better able to explain the biologic interaction, we also estimated the RERI, API and SI by additive model, whereas we did not find statistical difference. There maybe sample sizes decreased after strata that lead to the reducing statistical power 24 . Although we did not find effect of education level or breastfeeding duration combined with family history of breast cancer, several lines of evidence suggest that our findings are biologically plausible. In consist with other studies, our research also find HMG individuals with a family history of breast cancer are easier to develop neoplasia 25,26 . The activation of Akt 1 peaked in lactation by hormone-and anchorage-mediated pathways which regulates survival of epithelial cells. Shorter time breastfeeding duration decreased Akt 1 significantly that contributes to HMG 27 . High education often accompanies with high stress is thought to be connected with increasing risk of breast disease 28 . Normal growth of the mammary gland involves endocrine signals from the Hypothalamic-Pituitary-Gonadal (HPG) axis 29 . Stress has been shown to disrupt the function of endocrine system and increase susceptibility to HMG 30 .
Additionally, increasing level of inflammatory burden and Hypothalamic-Pituitary-Adrenocortical (HPA) axis dysregulation subsequent to stress may also impair HMG 31 . These observations indicated that hereditary-environment interactions might be especially important for HMG because their effects on disease susceptibility were strongly decided by exposure to education and other female reproductive factors. Therefore, it is suitable to consider that HMG prevention strategies should be individualized according to individual's exposure to risk factor profiles 32 .
A dominating superiority of our research is that hereditary-environment interactions are consistently distinguished by both nonparametric and parametric statistical models. Logistic regression has the advantage of analysing for main effect. When high-order interactions involving multidimensional elements are taken into account, they may be limited to deal with simultaneous factors 33 . MDR can identify putative high-order interactions but limit in analyzing main effects in many diseases 34 .
Crossover analysis can evaluate the independent and joint roles of genetic and exposure on disease hazard 24  To the best of our knowledge, our research is the first study to estimate the interaction of family history of breast cancer and female reproductive factors on HMG risk, and we believe that it provides some clues for research into the related mechanisms.

Conclusion
High-order interactions of higher education level, shorter breastfeeding duration and family history of breast cancer might synergistically increased HMG risk. HMG prevention strategies should be individualized according to individual's exposure to risk factor profiles.

Consent for publication
Not applicable.

Availability of data and material
The analyzed data can be accessed from the corresponding author on reasonable request.

Author's contributions
HG constructed statistical analysis strategy and drafted the manuscript. CY organized and coordinated epidemiological investigations. JF supervised the analysis. LL performed the statistical analysis. DP conceived of the study, and participated in its design and helped to review the manuscript. All authors read and approved the final manuscript.

Ethics approval and consent to participate
This study was approved by the ethic committee of Harbin Center for Disease Control and Prevention (No.01-2012). All the participants provided written informed consent before they were interviewed for this study. Flowchart of participant inclusion in the case-control study