Reproductive risk factors of pterygium in postmenopausal women: a nationwide study in the Republic of Korea

This study is to elucidate the associations between female reproductive factors and pterygium. A total of 1,339,969 postmenopausal women in a retrospective cohort of Korean National Health Insurance Service data on ages 40 and above in 2009 was included. Cox proportional hazards regression was conducted to assess the hazard ratio (HR) for pterygium according to reproductive factors. Late menarche, early menopause, short reproductive period, increasing parity (≥ 2 children), breastfeeding (≥ 6 months), and no use of hormone replacement therapy (HRT) or oral contraceptive (OC) were significantly associated with risk of pterygium. In multivariate analysis, the HR for pterygium was 1.764 (95% confidence interval [CI], 1.529–2.035) for menarche age ≥ 17 years (reference: menarche age < 12 years). The HR of menopause age ≥ 55 years was 0.782 (95% CI, 0.724–0.845) (reference: menopause age < 40 years). The HR of parity ≥ 2 was 1.261 (95% CI, 1.148–1.385) (reference: nulliparity). The HR of breastfeeding ≥ 1 year was 1.663 (95% CI, 1.564–1.768) (reference: no breastfeeding). The HRs of HRT and OC use for any length of time were lower than those for the non-user groups (reference). Reproductive factors that increase estrogen exposure have protective effects against pterygium in females.

Study population. From the NHIS database, we collected data for women above the age of 40 who underwent breast cancer screening from 1 January 2009 to 31 December 2009. Among 3,109,506 female subjects, we identified 1,939,690 eligible postmenopausal women. We first excluded individuals who reported having a hysterectomy procedure (n = 203,854), as most did not know whether they underwent simultaneous oophorectomy. We defined pterygium with diagnostic codes from ICD-10. Individuals who had a diagnosis of pterygium before the health screening date (n = 46,438) were identified from the Korean NHIS medical service claims data and excluded. In addition, we required a one-year lag period and excluded individuals with a pterygium diagnosis within that year (n = 7,226). Individuals who died within one year after the health screening date were also excluded (n = 3,566). We excluded 338,637 individuals with missing data for at least one variable. A total of 1,339,969 individuals was included in the final analyses (Fig. 1).

Reproductive factors.
According to NCSP guidelines, the study subjects completed a questionnaire addressing their age at menarche, age at menopause, and parity. Information regarding total lifetime breastfeeding history, hormone replacement therapy (HRT) history, and use of oral contraceptives (OC) was collected. Age at menarche was categorized as ≤ 12 years, 13-14 years, 15-16 years, and ≥ 17 years, to be consistent with the distribution of age at menarche among Korean women. Age at menopause was categorized as < 40 years, www.nature.com/scientificreports/ 40-44 years, 45-49 years, 50-54 years, and ≥ 55 years. The reproductive period was calculated as the interval between the age at menarche and the age at menopause. Parity was categorized as 0, 1, or ≥ 2 children. Total lifetime breastfeeding history was categorized as never, < 6 months, 6-12 months, or ≥ 12 total months. The duration of HRT was categorized as never, < 2 years, 2-5 years, ≥ 5 years, or unknown. The duration of OC was categorized as never, < 1 year, ≥ 1 year, or unknown 10,14 .
Study outcomes and follow-up. The endpoints of the study were pterygium diagnosis, which was defined with the following diagnostic codes defined by the Korean Standard Classification of Diseases 7th revision , with a few changes specific to Republic of Korea based on the International Classification of Diseases 10th revision: H11.0 (Pterygium), H11.01 (Central pterygium of the eye), H11.02 (Double pterygium of the eye), H11.03 (Peripheral pterygium of the eye), H11.05 (Recurrent pterygium of the eye), or H11.08 (Unspecified pterygium of the eye). The cohort was followed from one year after the health check-up date to the date of incident pterygium or until the end of the study period (December 31, 2018), whichever came first.
Other variables. Several general factors regarding health status and variables regarding economic status were included in the statistical regression model. The selected factors were smoking, alcohol consumption, physical activity, body mass index, hypertension (HTN), diabetes mellitus (DM), dyslipidemia, income grade, and chronic kidney disease (CKD). Glomerular filtration rate (GFR, ml/min/1.73m 2 ) was used as the criterion for CKD and divided into three categories. The definitions of the lifestyle variables are as follows. Smoking status was categorized into three groups: non-smokers, current smokers who had smoked 100 cigarettes or more in their lifetime and is currently smoking or quitted for less than a month, and ex-smokers who had smoked 100 cigarettes or more in the past but had quit for at least one month 15,16 . Alcohol consumption status was categorized into three groups: non-drinkers, mild-drinkers who drank less than 30 g a day on average, and heavy-drinkers who drank more than 30 g a day. Regular exercise was defined as strenuous physical activity performed for at least 30 min at least five times a week.
Body mass index (BMI) was defined as weight in kilograms divided by the square of height in meters (kg/m 2 ). BMI was divided into five categories: < 18.5 kg/m 2 , 18.5 to < 23 kg/m 2 (normal), 23 to < 25 kg/m 2 , 25 to < 30 kg/ m 2 , and ≥ 30 kg/m 214,17 . Waist circumference (WC) was analyzed to figure out metabolic aspects of postmenopausal women. Elevated WC was defined as ≥ 85 cm for women, according to the Korean Society for the study of Obesity's cut-off points for central or abdominal obesity 18 .
Subjects were considered to have HTN, DM, or dyslipidemia if they were on hypertensive, diabetic, or lipidlowering medication, respectively. In addition, subjects with systolic blood pressure) ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg, fasting blood glucose level ≥ 126 mg/dL, or triglyceride ≥ 240 mg/dL were considered to have HTN, DM, or dyslipidemia, respectively.
Household income status was classified into four groups in which Q1 and Q4 represent the lowest and highest income quartiles, respectively.

Statistical analyses.
Continuous variables are presented as mean ± standard deviation and median; interquartile range, and categorical variables are presented as number and percentage. Continuous variables were analyzed using Student t-test and categorial variables were analyzed using Chi-squared test in Table 1. Hazard ratio (HR) and 95% confidence interval (CI) values for pterygium were analyzed using the Cox proportional hazards model for various reproductive factors. The multivariate-adjusted proportional hazards model was applied: (1) Model 1 was not adjusted; (2) Model 2 was adjusted for age, age at menarche, age at menopause, parity, duration of breastfeeding, duration of HRT, duration of OC use, alcohol consumption, smoking, regular exercise, income, BMI, WC, HTN, DM, dyslipidemia, and CKD; and (3) Model 3 was adjusted for all variables as in Model 2, but age at menarche and age at menopause were replaced by reproductive period. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA), and a P-value < 0.05 was considered statistically significant. Multicollinearity among variables were analyzed by Variance Inflation Factor (VIF).

Results
Baseline characteristics of the study population. The Table 1 showed the characteristics of women diagnosed with pterygium and women without pterygium in detail. All factors listed in Table 1 showed a statistically significant difference between the two groups.
Associations between reproductive factors and the risk of pterygium. Late menarche, early menopause, short reproductive period, increasing parity (≥ 2 children), breastfeeding (≥ 6 months), and no use of HRT or OC were significantly associated with risk of pterygium (Table 2).
In model 2, the HR for pterygium was 1.764 (95% CI, 1.529-2.035) in the oldest age at menarche group (≥ 17 years) compared to the youngest age at menarche group (≤ 12 years). The HR was 0.782 (95% CI, 0.724-0.845) in the oldest age at menopause group (≥ 55 years) compared to the youngest age at menopause group (< 40 years). The HR showed a dose-response relationship with age at menarche and age at menopause.

Discussion
Our study demonstrated lifetime estrogen exposure as a protective effect against pterygium in females. In this nationwide cohort study investigating the association between pterygium and reproductive factors, adjusted multivariate analysis revealed statistically significant protective effect with factors of early age at menarche, later age at menopause, longer reproductive period, parity (single childbirth, reference: nulliparous), no breastfeeding, HRT, and OC use. Such associations were independent of possible confounding factors of smoking, alcohol consumption, regular physical activity, BMI, HTN, DM, dyslipidemia, GFR, and income. This study is the first nationwide population-based study on the effects of reproductive factors on pterygium in the Republic of Korea. The associations between reproductive factors and risk of pterygium have been discussed in the few literature. HRT in postmenopausal women lowered the prevalence of fresh pterygium, and premature menopause increased the risk of grade 2 or higher pterygium in females 2,8 . Such findings correspond with our results. However, previous researchs were conducted with limited population and revealed the effects of only few female reproductive factors. In this study, we had largest study cohort regarding this subject and showed effects of multiple female www.nature.com/scientificreports/ reproductive factors on pterygium. Early menopause, late menarche, longer reproductive period, single childbirth, shorter breastfeeding period, HRT, and OC use had a protective effect against pterygium in females.
Clues that lead to mechanism of estrogen on pteygium could be found in literatures. Previous studies have shown the presence of sex steroid hormone receptors (SSHR) in various ocular tissues, such as the lens, retina, choroid, cornea, iris, ciliary body, lacrimal gland, meibomian gland, lid, and palpebral and bulbar conjunctiva 19,20 . Estrogen, which combines with SSHR, has various functions. First, estrogen has been shown to decrease oxidative stress in tissues by preserving mitochondrial function, cell viability, and ATP level 21,22 . Oxidative stress caused by increased production of radical oxygen species (ROS) is thought to be the main pathogenesis of pterygium. Studies on immunohistochemical analysis showed staining of the oxidative stress marker 8-hydroxy-2'-deoxyguanosine in pterygium compared to normal conjunctival tissue 23,24 . Also, pterygium tissue had increased level of nitric oxide, increased total antioxidant status, and decreased antioxidant enzymes 25 . This indicates that high estrogen level prevents development of pterygium by reducing oxidative stress.
Furthermore, estrogen provides cell resistance to transforming growth factor β (TGF β) and promotes normal function of the cell membrane 9 . Growth factors such as TGF β show strong immunohistochemical staining in pterygium tissue and play an important role in cell proliferation, inflammation, connective tissue remodeling, and angiogenesis 26,27 . Estrogen reduces the risk of pterygium by hindering the action of growth factors.
Considering previous research and the results of our study, estrogen exposure, either exogenous or endogenous, seems to have antioxidative effects and plays a protective role against pterygium. Such protective effect of lifetime estrogen exposure on systemic diseases had been discussed in the literatures before. Several studies argued that more lifetime estrogen exposure in postmenopausal women was associated with lower mortality, less cardiovascular disease, and better cognitive function in later life 28,29 .
On the other hand, the association may not be causal relationship, as genetic or environmental factors could also determine lifetime estrogen exposure. There are some evidences that onset of age at menopause is affected by genetic factors, as evidenced by different menopausal age among races, associated gene loci 30 , and association between family history and premature menopause 31 . Moreover, environmental factors such as smoking, lower education, low occupation levels and lower family income are associated with premature or early menopause [31][32][33] . Therefore, there is a possibility that genetic or environmental factors determined the level of estrogen exposure and pterygium development. Also, estrogen might act as surrogator or mediator of genetic and environmental factors in pterygium development.
While single childbirth decreased the HR of pterygium, multiple births increased the HR. This seemingly opposite result should be interpreted with caution, considering the physiological plausibility. It is well known that estrogen level increases dramatically during pregnancy 34 . However, estradiol level decreases in consecutive pregnancies 35,36 . Moreover, in lactating women, prolactin level is increased and results in hypo-estrogenization 37 . A multiparous group would have a longer total lactation period compared to a single childbirth group, resulting in a longer hypo-estrogenization period. These results could explain the increased HR of pterygium in the multiparous group. Furthermore, VIF was analyzed to verify multicollinearity among variables statistically, and none of the variables showed VIF more than 10 to presume multicollinearity.
A preceding study on BMI and pterygium revealed that obesity in females had a significant association with pterygium in multivariate analysis 38 . This supports our result of increased HR in the group with BMI of 23 to 30 kg/m 2 . The study indicated that obesity is a chronic inflammatory disease and responsible for overproduction of ROS and nitrogen species, which cause oxidative stress 38 . Such oxidative stress is a well-known cause of pterygium.
The strength of our study lies in the use of a large population based on a national database. Traditionally, female sex was left out of the discussion on risk factors of pterygium and male sex was emphasized connected with environmental factors. This article focused solely on effects of female reproductive factors on pterygium. Secondly, unlike previous articles that suggested only a few reproductive factors to be associated with risk of pterygium, our study revealed that overall reproductive factors that increase estrogen level have a protective effect against the disease.
There are several limitations to our study. First, we could not obtain information on ultraviolet light exposure and dry eye in NHIS data, although these factors are usually included in a discussion of risk of pterygium 2,3,6,39 . Second, all the information on reproductive factors was based on self-reported histories; therefore, the possibility of bias caused by inaccurate recall cannot be excluded. Third, we were unable to retrieve more detailed data regarding age of initial HRT and OC use, the duration between menopause and HRT or OC initiation, the composition and concentration of HRT and OC (e.g., estrogen alone or combined estrogen-progesterone) and the dose, as these questions were not included in the questionnaires. Further detailed information about female hormone use and user potential risk factors for pterygium would be needed to clarify this possible association between exogenous female hormone use and risk of pterygium in future study. Lastly, as Korea has a homogenous ethnic population, further studies on other ethnic groups should follow.
In conclusion, based on a nationwide large population cohort, this study revealed that reproductive factors that increase female exposure to estrogen had a protective effect against pterygium.