Association Between Hormone Levels and Retinopathy in Patients With Type 2 Diabetes: A Cross-Sectional Study

Abstract

Purpose: To explore the relationship between hormone levels and diabetic retinopathy (DR).

Methods: This cross-sectional study recruited 2432 patients with type 2 diabetes from Tianjin Medical University General Hospital between 2016 and 2019. Logistic regression was used to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for the association between hormone levels and DR.

Results: DR was diagnosed in 953 patients (39.19%), with no significant gender difference in prevalence. Among men, DR prevalence increased with testosterone level increasing (Q4 vs. Q1: odds ratio [OR] = 1.64; 95% confidence interval [CI], 1.11–2.41) but decreased with prolactin levels increasing (Q4 vs. Q1: OR = 0.65; 95% CI, 0.44–0.96). Among women, high follicle-stimulating hormone (FSH) (Q4 vs. Q1: OR = 2.22; 95% CI, 1.40–3.52) and luteinizing hormone (LH) levels (Q4 vs. Q1: OR = 1.77; 95% CI, 1.20–2.63) were linked with increased DR prevalence. No associations were found in premenopausal women, but postmenopausal women with high prolactin level had increased DR prevalence (Q2 vs. Q1: OR = 1.56; 95% CI, 1.01–2.19). Dose-response relationships were suggested for FSH (P = 0.087) and testosterone (P = 0.088) with DR prevalence, though caution is advised due to the risk of type I error arising from multiple comparisons. Other risk factors included low body mass index (OR = 0.98), smoking (OR = 1.32), long diabetes duration (OR = 1.02), and high systolic blood pressure (OR = 1.01).

Conclusions: Prolactin (negatively) and testosterone (positively) were correlated with DR risk in men. FSH and testosterone showed suggestive dose-response relationship with DR prevalence in postmenopausal women. Further research is needed for type 1 diabetes.

Translational Relevance: Sex hormones play a crucial role in DR development, affecting men and women differently.

Introduction

Type 2 diabetes mellitus (T2DM) is a chronic metabolic disease. By 2045, an estimated of 700 million people will live with diabetes, and one-third of them will have diabetic retinopathy (DR) worldwide.¹ DR is a specific microvascular complication of diabetes; it remains the leading cause of vision loss and blindness.² 

Extensive research has shown that sex hormones play important roles in the development of T2DM and might influence men and women differently.³ Specifically, men are at a higher risk of diabetic microvascular complications, whereas the impact of diabetes-related macrovascular events tends to be more pronounced in women.⁴ Several epidemiological studies have identified male sex is a risk factor for DR,⁵ while others have reported a higher prevalence of DR in females, although the disease tends to be more severe in males.⁶ Additionally, Klein et al.⁷ noted that females with T2DM and DR exhibited better recovery outcomes compared to males. Studies to date have increasingly suggested potential associations between hormones and retinopathy, particularly in patients with diabetes. A study on prolactin in 40 nondiabetic and 181 diabetic men with DR found that the level of prolactin was lower in patients with proliferative DR (PDR) than that in those without DR, which demonstrated that high levels of prolactin might protect against DR in men.⁸ Sun et al.⁹ reported that men with progesterone levels ≤0.425 nmol/L had an 8.44 times lower risk of PDR compared to those with levels ≥1.27 nmol/L, suggesting that high progesterone levels may predict DR in men. Previous researches proposed that high levels of estradiol and low levels of testosterone were just independent predictors of T2DM risk in men, but not in women.^10,11 However, data from others suggested that elevating levels of estradiol and testosterone were related with an increased risk of T2DM in postmenopausal women,^3,12 and another study demonstrated that FSH, LH, and estradiol were not correlated with DR in women.¹³ 

To date, these studies have primarily focused on a single sex. Thus the specific objective of our study was to simultaneously analyze the relationship between hormones levels (including FSH, LH, prolactin, estradiol, progesterone, and testosterone) and DR risk among men, premenopausal women, and postmenopausal women, respectively. Results from this study may be relevant to the development of prevention and treatment strategies of retinal disorders. 

Methods

Study Design and Study Population

This cross-sectional study enrolled 2432 T2DM patients, aged 18 to 90 years, between January 2016 and January 2019 at the Department of Endocrinology, Tianjin Medical University General Hospital. T2DM was defined according to 2019 American Diabetes Association criteria.¹⁴ Patients who were pregnant or had type I or other types of DM, acute complications of diabetes, serious infections, serious heart disease, serious liver disease or cancer were excluded. Patients who had taken drugs affecting hormones levels in the past three months were further excluded. Postmenopausal women were defined as those who had not menstruated for at least 12 months or who had a previous hysterectomy or oophorectomy. This study was approved by Tianjin Medical University General Hospital Ethics Committee (approval number IRB2023-WZ-089). All procedures adhered to the tenets of the Declaration of Helsinki, and all subjects provided written consent. 

Data Collection

Blood biochemistry and clinical assessment were conducted on all patients. Data on demographic and clinical characteristics of patients including age, sex, diabetes duration, the status of smoking and drinking, body weight, height and blood pressure (BP) and medications for diabetes treatment were gathered via medical records. “Current smokers” were defined as people having smoked more than 100 cigarettes in one's lifetime and currently smoking cigarettes. “Current drinkers” were defined as individuals consuming alcohol more than once a month in the past 12 months. A stadiometer and a vertical ruler were used to measure subjects' weight and height while they wore light clothing. Body mass index (BMI) was calculated by dividing weight in kilograms divided by height in meters squared. Patients were divided into four groups based on their BMI status: low weight group (BMI < 18.5), normal group (18.5 ≤ BMI ≤ 23.9), overweight group (24.0 ≤ BMI ≤ 27.9) and obese group (BMI ≥ 28).¹⁵ 

Fasting plasma glucose was measured by a hexokinase method (TBA-200FR, Tokyo, Japan). Hemoglobin A1C (HbA1c) was tested by high-pressure liquid chromatography (MQ-2000PT, Shanghai, China). Triglycerides (TG), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), total cholesterol (TC), and serum creatinine were detected by automatic biochemical immunity analyzer (VITRO 5600; Johnson & Johnson Vision, Jacksonville, FL, USA). Serum FSH, LH, prolactin, estradiol, progesterone and testosterone were measured by electrochemiluminescence immunoassay (Archirect i2000; Abbott Laboratories, Abbott Park, IL, USA). Blood samples were collected after a minimum of eight hours of fasting, between 6:00 and 9:00 AM. 

All patients underwent fundus photography (VISUCAM-524; Carl Zeiss, Dublin, CA, USA), slit lamp microscope (SL 115 Classic; Carl Zeiss Meditec, Jena, Germany) and noninvasive optical coherence tomography (5000; Carl Zeiss, Dublin, CA, USA) examination to assess the retinopathy. Retinopathy was defined as if any of the following lesions was present, such as, microaneurysms, hemorrhages, cotton wool spots, intraretinal microvascular abnormalities, hard exudates, venous beading, new vessels, and retinal detachment.¹⁶ Based on international clinical DR disease severity scale,¹⁶ patients were classified into no-DR group and DR group. The fundus photographs were graded through a double-blind evaluation conducted by two ophthalmologists. In instances of discordance, a third ophthalmologist was consulted to reach a consensus. 

Statistical Analysis

Statistical analyses were performed using SAS software version 9.4 (SAS Institute Inc, Cary, NC, USA). Continuous variables were represented by mean ± standard deviation, and categorical variables were represented by percentage. Continuous variables were compared using Student's t-test, and categorical variables were compared using χ² test between the two groups. Hormone levels (i.e., FSH, LH, prolactin, estradiol, progesterone and testosterone) were classified into four groups based on quartiles of each hormone. In order to determine the risk of DR for each quartile of hormone levels, the odds ratios (ORs) and 95% confidence intervals (CIs) were calculated using multivariable logistic regression models. Potential confounders were selected using a backward elimination process, with a significance threshold of P < 0.05 for inclusion in the final model. The final covariates included age, BMI, smoking status, diabetes duration, systolic BP (SBP), and concentrations of HbA1c and HDL-c. The P value for trend was calculated by treating the median value of each quartile as a continuous variable in the regression models. Potential interactions were assessed through stratified analysis and confirmed using the Wald method by including a multiplicative interaction term in the model and evaluating its significance. A value of P < 0.05 was considered statistically significant. 

Results

Study Population

The demographic and clinical data of patients were shown in Table 1. A total of 2432 T2DM patients (1377 women, 1055 men) were included; 953 (534 women, 419 men) patients were diagnosed with DR (DR group), and the other 1479 patients (843 women, 636 men) were without DR (no-DR group). The prevalence of DR was 39.19%. There was no significant difference in DR prevalence between men and women (P = 0.639). Compared to the no-DR group, patients with DR were significantly older, more likely to be smokers, had a longer duration of diabetes, a higher prevalence of insulin use, a lower prevalence of oral hypoglycemic medications use, and higher SBP (P < 0.05). 

Table 1.

 

Demographic and Clinical Characteristics of Study Participants by Diabetic Retinopathy Status

 

Associations Between the Levels of Hormones and DR Risk Among Men and Women

The associations between hormones levels and the risk of DR among men and women were shown in Table 2. Among men, after adjusting for potential confounders, compared with the first quartile, participants in the fourth quartile of prolactin had odds of DR that were significantly decreased by 35% (OR = 0.65, 95% CI, 0.44–0.96); however, participants in the fourth quartile of testosterone had odds of DR that were increased by 64% (OR = 1.64; 95% CI, 1.11–2.41). No significant associations between FSH, LH, estradiol, and progesterone levels and DR prevalence were found among men. Among women, compared with the first quartile, participants in the fourth quartile of FSH and LH had odds of DR that were significantly increased by 122% (OR = 2.22; 95% CI, 1.40–3.52) and 77% (OR = 1.77; 95% CI, 1.20–2.63), respectively. Although there was no markedly overall trend between prolactin and DR prevalence, compared with the first quartile of prolactin, the odds of DR was increased by 43% in the second quartile (OR = 1.43; 95% CI, 1.05–1.96). No significant associations were detected between estradiol, progesterone, and testosterone and DR prevalence in women. 

Table 2.

 

Association Between Levels of Hormones and Diabetic Retinopathy Among Men and Women With Type 2 Diabetes

 

In this study, women patients with DR were further divided into premenopausal groups (35.84%) and postmenopausal groups (64.16%). Table 3 showed the associations between hormones levels and DR prevalence among them. Among premenopausal women patients, multivariable analysis revealed no significant associations between DR prevalence and FSH, LH, prolactin, estradiol, progesterone and testosterone levels. Among postmenopausal women, the ORs for DR were 2.05 (95% CI, 1.03–4.06) in the highest quartile of FSH levels and 1.53 (95% CI, 1.05–2.21) in the third quartile of testosterone levels compared to the lowest quartile. We observed marginally significant dose-response relationships between DR prevalence and FSH (P = 0.087), as well as testosterone levels (P = 0.088), indicating that increased FSH and testosterone concentration were associated with an increased prevalence of DR. Compared with the first quartile of prolactin, the odds of DR was increased by 56% (OR = 1.56; 95% CI, 1.01–2.19) in the second quartile. There were no significant associations between LH, estradiol, and progesterone levels and DR prevalence among postmenopausal women. 

Table 3.

 

Association Between Levels of Sex Hormones and Diabetic Retinopathy Among Premenopausal and Postmenopausal Women With Type 2 Diabetes

 

To assess whether hormones effects varied by participants’ age and age at diabetes onset (early- vs. late-onset T2DM), analyses were stratified by age at enrollment (<40 and ≥40 years) and age at diabetes diagnosis (<40 and ≥40 years). In stratification by age at enrollment, significant interactions were observed between LH and age in women, with a positive association limited to those aged ≥40 years (Q4 vs. Q1: OR = 1.58; 95% CI, 1.02–2.44; P for interaction = 0.032) (Supplementary Table S1). A marginally significant interaction between testosterone and age was also noted in DR, showing a harmful effect only in men under 40 years (Q4 vs. Q1: OR = 4.51; 95% CI, 1.38–14.69; P for interaction = 0.067). Additionally, a dose-response relationship between FSH (Q4 vs. Q1: OR = 1.91; 95% CI, 1.12–3.25; P for trend = 0.028), progesterone (Q4 vs. Q1: OR = 0.60; 95% CI, 0.36–0.97; P for trend = 0.032), and DR was observed in men diagnosed with diabetes at ≥40 years, although interaction testing by age of diabetes onset did not reach significance (Supplementary Table S2). Upon subdividing the outcome into PDR and NPDR (Non-PDR), similar patterns were observed in the NPDR vs. No-DR analysis, whereas fewer associations were detected in the PDR vs. No-DR analysis (Supplementary Table S3). In the latter, only two associations reached significance, with elevated PDR prevalence observed in men in the second quartile of estradiol (OR = 10.14) and women in the second quartile of progesterone (OR = 5.52). However, these findings should be interpreted with caution due to the limited number of PDR cases (n = 53), resulting in reduced statistical power in the PDR versus No-DR analysis. 

Association Between Covariates and DR Risk

Multivariable logistic regression models showed that lower BMI (OR = 0.98; 95% CI, 0.96–1.00), higher SBP (OR = 1.01; 95% CI, 1.00–1.01), longer diabetes duration (OR = 1.02; 95% CI, 1.01–1.04), and smoking (OR = 1.32; 95% CI, 1.05–1.65) were significantly associated with elevated DR risk in patients with T2DM (Table 4). 

Table 4.

 

Association Between Covariates and Diabetic Retinopathy

 

Discussion

Our research examined the relationship between hormones levels and DR risk in 2432 diabetic patients. The DR prevalence in our study sample was 39.2%, which was a little higher than the global rate,¹⁷ and this may be due to different lifestyles (dietary habits and exercise) and economic levels. Another possible reason might be that our participants were hospitalized with a higher rate. In our study, the prevalence of DR in men and women were 39.72% and 38.78%, respectively. No significant sex-difference was found between no-DR and DR group, which was consistent with the previous study.¹⁸ However, in the United Kingdom Prospective Diabetic Study, it showed that men were more prone to develop DR.¹⁹ A retrospective cohort study performed in United States also showed that men had a 122% increased hazard for DR compared to women with T2DM.²⁰ Wisconsin Epidemiological study of Diabetic Retinopathy appeared to point out DR status had better outcomes in women than men equivalent with T2DM.⁷ Those above studies might indicate that hormones to some extent have an influence on the development and progression of DR. 

Our study revealed that high testosterone levels were significantly associated with increased ORs of DR in men, consistent with the prior study that the levels of testosterone were higher in diabetic patients with more severe DR than those with mild retinopathy or no retinopathy.⁹ However, diabetes was characterized by decreasing in bioavailable testosterone levels,²¹ and men with T2DM had a high prevalence of testosterone deficiency that was associated with metabolic syndrome.²² Goto et al.²³ also proposed that testosterone might be negatively correlated with diabetes, which showed that testosterone could inversely predict the risk of diabetes in men. Althoughour study obtained an opposite conclusion, that is, high testosterone levels were positively associated with DR prevalence in men, which suggests further researches are needed. 

An inverse connection between high prolactin levels and DR risk in men was shown in our study, which is accord with the findings in Arnold et al.’s research.⁸ Froland A²⁴ also found that prolactin levels were significantly lower in severe DR in men, compared with those without severe hemorrhagic DR. High levels of prolactin inhibit VEGF-stimulated endothelial cell proliferation, and low levels of prolactin result in decreased VEGF-inhibition, consequently leading to a pro-angiogenic environment.²⁵ Those evidence suggested that diabetic patients with higher PRL levels might have a lower risk for DR. 

Our study found FSH levels were positively corrected with DR prevalence among women, which is in line with the results that FSH levels were higher in the diabetic patients with retinopathy than in those without retinopathy.^26,27 However, Siddiqui et al.¹³ pointed out that no difference was observed in the serum levels of FSH between patients with and without DR in women. In the study, no correlation was observed between DR risk and FSH levels in premenopausal women, whereas DR prevalence increased with elevating FSH levels in postmenopausal women. However, a cross-sectional study in China uncovered that postmenopausal women with FSH levels in the lowest quartile had a threefold increased odds of developing diabetes comparing with those with FSH levels in the highest quartile; that is, patients with high FSH levels halved the risk of developing DM. FSH might be a protective biomarker of glucose metabolism in postmenopausal women.²⁸ Contrary to expectations, our study obtained an opposite conclusion that high levels of FSH were risk predictors of DR. The exact mechanism involved in whether and how FSH affects DR is currently unclear and needs to be elucidated. 

In the present study, among women, DR prevalence significantly increased with elevated LH levels, which is in line with the results that patients with DR had higher levels of LH compared with those without DR.²⁹ The possible mechanism that LH correlated to DR might be explained by LHs activating the LH receptors in the retina and inducing VEGFs expression, which could lead to DR.³⁰ A notable finding is the interaction between LH levels and age at enrollment in DR. Although the underlying mechanisms remain unclear, the higher LH levels in women aged ≥40 years (mean 21.9 IU/L) compared to those <40 years (mean 6.1 IU/L) may contribute to the stronger association observed in the older group. Although prior studies have not specifically examined this interaction, it may help explain the conflicting results regarding the LH-DR association. For instance, recent research found no significant difference in LH levels between women with and without DR but did not account for participant age. Future studies should investigate the LH-DR relationship in women while considering potential age-related effect modification. 

In the study, no significant association was found between testosterone levels and DR prevalence in women patients. However, among postmenopausal women, there was a borderline significant dose relationship between testosterone levels and DR prevalence (P = 0.088), that is, the risk of DR in the third quartile of testosterone was 1.53-fold of that in the first quartile. Women after menopause exhibited a change in hormone imbalance, decreasing estrogen levels and increasing testosterones levels, and elevated bioavailable testosterone levels in the absence of estrogen could aggregate the state of chronic inflammation, which might be a culprit for the occurrence and development of DR in postmenopausal women.³¹ 

Last, logistic regression analysis also showed that lower BMI, smoking, longer DM duration and higher BP were risk factors of DR, which is in line with those findings from previous studies that had reported that duration of diabetes, poor blood pressure control, and dyslipidemia accelerated the development of DR.³² BMI is an established risk factor for diabetes. However, the relationship between BMI and DR has been inconclusive. Some studies have reported a positive association between high BMI and DR risk,³³ while others found no correlation³⁴ or even an inverse relationship.^35,36 Our findings, indicating that lower BMI is a risk factor for DR, align with the “obesity paradox,” which suggests a reduced rate of complications, including retinopathy, in individuals with higher obesity levels.³⁷ This paradox has been observed in both adults^38,39 and youth with T2DM.⁴⁰ 

There were some limitations in our study. First, because of the observational nature of this cross-sectional study, it was not possible to establish a causal relationship between hormones and DR. Second, because all patients were diabetic cases, the results might not be applicable to the general population. Further studies with longitudinal follow-up are needed to verify the relationship and mechanistic role of hormones in the development of DR in all the population. Third, the NPDR group was not subdivided into mild, moderate, and severe categories because of insufficient data on disease severity. Given the progressive and regressive nature of DR, understanding disease severity within the population is crucial. Future research would explore the relationship between DR and sex hormones, accounting for different severities of the disease. Fourth, although systolic BP and HDL-c levels were adjusted in the multivariable models, the absence of data on lipid-lowering medications and antihypertensives limited our ability to fully assess their impact on the association between sex hormones and DR. Last, in the current study, hormones were measured using immunoassays, which are less sensitive, particularly for testosterone and estradiol, compared to mass spectrometry. Future studies utilizing more rigorous or calibrated methods for measuring sex steroids are warranted. 

In conclusion, after adjusting for possible confounders, this study has identified high testosterone levels were significantly associated with increased DR risk, and an inverse connection between high prolactin levels and DR risk in men. Increased FSH and testosterone levels were associated with an increased prevalence of DR. Those links between gonadal hormones and DR may open new therapeutic horizons, although the underlying mechanisms are still not well understood. 

Acknowledgments

Supported by National Key Research and Development Program of China (Grant Number 2021YFC2401404), and National Natural Science Foundation of China (Grant Numbers 82020108007, 81830026). 

Disclosure: Y. Meng, None; X. Yang, None; Y. Fan, None; M. Liu, None; F. Zhou, None; Q. Wang, None; Y. Lu, None; Y. Zhu, None; H. Yan, None