Sex Hormone Imbalance in Keratoconus: Estrogen-Mediated Collagen Remodeling and Biomechanical Weakening

Abstract

Purpose

To investigate the role of sex hormone imbalance in keratoconus (KC) pathogenesis and its effects on corneal collagen metabolism and biomechanics.

Methods

Fifteen newly diagnosed, treatment-naïve KC patients and 12 age- and sex-matched controls were enrolled. Plasma and tear estradiol (E2), testosterone (T), and matrix metalloproteinase-2 (MMP2) concentrations were measured by ELISA. Associations between sex hormone and clinical parameters were analyzed using Pearson correlation. Human corneal stromal cells (HCSCs) were exposed to physiological or supraphysiological E2/T to evaluate cell viability, matrix metalloproteinases (MMPs), tissue inhibitors of metalloproteinases (TIMPs), type I collagen (COL1), and hydroxyproline (HYP). Estrogen receptor-β (ERβ) involvement was assessed using the ERβ selective antagonist PHTPP. A high-estrogen rat model was established to evaluate central corneal thickness (CCT), stromal collagen ultrastructure, and biomechanical properties.

Results

KC patients exhibited significantly elevated plasma and tear E2 but unchanged T levels. Tear E2 concentrations positively correlated with mean keratometry, diopter, and MMP2 levels and negatively correlated with minimum corneal thickness and best-corrected visual acuity (BCVA). Plasma and tear E2, as well as tear MMP2, were significantly increased in male KC patients and showed significant correlations with KC-related parameters. In contrast, no significant differences or associations were observed in female KC patients in this cohort. Hormone treatments showed no cytotoxicity in HCSCs. High estrogen exposure increased MMP2 and decreased TIMP2 mRNA expression, increasing the MMP2/TIMP2 ratio, reducing COL1 protein, and enhancing collagen degradation. Estrogen receptor-β (ERβ) expression was upregulated, and pharmacologic inhibition with PHTPP partially restored MMP/TIMP balance and collagen homeostasis. In vivo, estrogen-treated rats demonstrated reduced CCT, disrupted stromal collagen organization, and decreased Young's modulus.

Conclusions

Elevated estrogen disrupts corneal collagen homeostasis and biomechanical integrity via an E2–ERβ–MMP2 axis, establishing a mechanistic link between sex hormone imbalance and KC pathogenesis.

Keratoconus (KC) is a progressive, bilateral corneal ectatic disorder characterized by stromal thinning, anterior protrusion, and irregular astigmatism, ultimately resulting in progressive visual impairment.^1,2 Once considered rare, recent population-based studies indicate that the pooled global prevalence of KC is approximately 0.1% to 0.3%, with marked geographic variability.³ Higher prevalence rates have been reported in the Middle East, South Asia, and parts of Africa, whereas lower rates are observed in East Asian and Western populations.^3–7 Clinically, KC typically manifests during adolescence, progresses most actively in the second and third decades of life, and often stabilizes thereafter.⁸ Whether biological sex influences disease susceptibility or progression remains unresolved. Although several epidemiological studies, particularly in Korean populations, have reported a female predominance⁹ or no significant sex difference,⁷ findings are inconsistent. For example, Asian population-based studies have reported a higher incidence in males accompanied by earlier onset and more rapid progression in men.¹⁰ These discrepancies highlight the need for further investigation into the potential biological contribution of sex in KC.

Marked inter-ethnic variations, familial aggregation, and the identification of multiple susceptibility loci strongly support a genetic contribution to KC pathogenesis.¹¹ In parallel, environmental and behavioral modifiers, including chronic eye rubbing, atopy, and ocular mechanical stress, substantially modulate disease risk and severity.^12,13 Together, these findings support a multifactorial etiology in which genetic predisposition interacts with environmental and mechanical insults.

At the tissue level, disruption of extracellular matrix (ECM) homeostasis and impaired corneal biomechanics are considered central to pathogenesis.¹⁴ KC corneas exhibit reduced types I and V collagen, disorganized stromal lamellae, and increased expression of matrix metalloproteinases (MMPs) accompanied by decreased tissue inhibitors of metalloproteinases (TIMPs), resulting in excessive ECM degradation.^15–17 In addition, diminished lysyl oxidase (LOX)-mediated collagen cross-linking, elevated inflammatory cytokines, and oxidative stress further weaken stromal structural integrity.^18–20 Although these mechanisms explain progressive corneal thinning and ectasia, they do not fully account for the characteristic epidemiologic patterns, particularly adolescence-predominant onset and male bias, thus suggesting the involvement of additional upstream regulatory factors.

Sex hormones have emerged as potential modulators of ocular physiology.²¹ Beyond their systemic reproductive functions, sex hormones regulate tissue metabolism and homeostasis in multiple organ systems, including the eye. At the ocular surface, androgens maintain lacrimal and meibomian gland function and exert anti-inflammatory effects,^22,23 whereas estrogens exhibit antioxidant and neuroprotective properties in neural tissues.²⁴ The cornea itself is hormonally responsive; fluctuations in hormonal status during the menstrual cycle and pregnancy are associated with reversible alterations in corneal thickness and curvature.^25,26 Moreover, estrogen receptors (ERs), progesterone receptors (PR), and androgen receptors (ARs) are expressed in corneal epithelial, stromal, and endothelial cells,²⁷ supporting the possibility of direct hormonal regulation of corneal structure and biomechanics.

Clinical observations further implicate hormonal influences in KC progression. Disease exacerbation has been reported during pregnancy and estrogen replacement therapy.^26,28 Experimental studies support a role for sex hormones, showing androgen-dependent KC-like phenotypes,²⁹ whereas clinical investigations have reported altered systemic sex hormone profiles in patients with KC.^30,31 Notably, systemic sex hormone levels peak during puberty,³² and their temporal fluctuations parallel the natural course of KC, reinforcing the hypothesis that hormonal imbalance may contribute to disease onset and progression.

Despite these associations, the precise pathogenic mechanisms remain poorly defined. The molecular and cellular signaling pathways through which sex hormones influence corneal biology have not been fully elucidated. In particular, the regulatory networks linking sex hormones to collagen metabolism, MMP activity, and inflammatory mediators are largely unexplored. A detailed investigation into the causal relationship between sex hormone imbalance and corneal collagen homeostasis is therefore essential to clarify KC pathogenesis and may provide a foundation for the development of novel therapeutic strategies.

Methods

Participants

This observational study enrolled 27 subjects (27 eyes) at Tianjin Medical University Eye Hospital, including 15 newly diagnosed, treatment-naïve patients with KC and 12 age- and sex-matched controls. KC was diagnosed by a corneal specialist based on slit-lamp biomicroscopy, corneal topography, and tomographic imaging. Disease severity was graded according to the Amsler–Krumeich classification,³³ which incorporates mean keratometry, corneal thickness, diopter, and the presence of corneal scarring. Control subjects were recruited from individuals with refractive errors but without clinical or tomographic evidence of corneal ectasia. Exclusion criteria included systemic autoimmune disorders, a history of ocular laser surgery, contact lens wear within 2 weeks, active ocular inflammation, and the use of systemic or topical hormone or anti-inflammatory medications within the preceding 3 months. Individuals with factors known to influence sex hormone levels—including menstrual period, pregnancy or lactation, and systemic endocrine disorders—were excluded, and the same criteria were applied to the control group for consistency. When both eyes met the inclusion criteria, one eye was randomly selected using a computer-generated randomization sequence. The study adhered to the tenets of the Declaration of Helsinki and received approval from the Institutional Review Board of Tianjin Medical University Eye Hospital (protocol no. 2024KY-08). All subjects provided informed consent after they received a comprehensive explanation of the purpose, procedures, potential risks, and anticipated benefits of the study.

Blood and Tear Sample Collection and Preparation

All blood and tear samples were collected between 8:00 and 10:00 AM under fasting conditions. Venous blood samples (3–5 mL) were drawn into ethylenediaminetetraacetic acid (EDTA) anticoagulant tubes and centrifuged at 3000g for 15 minutes at 4°C within 1 hour of collection to obtain plasma. Tear samples were collected without topical anesthesia using sterile Schirmer strips placed in the inferior conjunctival sac for 5 minutes with the eyes gently closed. After collection, the strips were transferred to microcentrifuge tubes containing 200 µL cold phosphate-buffered saline and centrifuged to elute tear fluid. All samples were stored at −80°C until analysis.

Cell Culture and Establishment of Sex Hormone Imbalance Models

Primary human corneal stromal cells (HCSCs; ScienCell Research Laboratories, Carlsbad, CA, USA) were cultured in Fibroblast Medium supplemented with 2% fetal bovine serum (FBS), 1% fibroblast growth supplement, and 1% penicillin–streptomycin (ScienCell Research Laboratories). Cells were incubated at 37°C in 5% CO₂ and used for experiments at passages 2 and 3. All treatments were performed in phenol red-free Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F-12) with charcoal-stripped FBS to eliminate estrogen-mimicking background signals.³⁴ Before treatment, cells were FBS-starved for 16 hours and subsequently exposed for 24 hours to combinations of E2 and T at concentrations approximating physiological and supraphysiological levels.³⁵ Treatment conditions included normal hormone levels (10⁻¹⁰ M E2+10⁻⁹ M T), higher estradiol (10⁻⁸ M E2+10⁻⁹ M T), lower estradiol (10⁻¹² M E2+10⁻⁹ M T), higher testosterone (10⁻¹⁰ M E2+10⁻⁸ M T), and lower testosterone (10⁻¹⁰ M E2+10⁻¹⁰ M T). E2 (Sigma-Aldrich, St. Louis, MO, USA) and T (Solarbio, Beijing, China) were dissolved in dimethyl sulfoxide (DMSO; Solarbio, China), and the final DMSO concentration was kept below 0.1% in all groups, including vehicle controls. For receptor inhibition experiments, the estrogen receptor-β (ERβ) inhibitor PHTPP (MedChemExpress, Monmouth Junction, NJ, USA) was dissolved in DMSO and added at a final concentration of 20 µM 30 minutes before hormone treatment. Untreated control cells received an equivalent volume of DMSO to ensure consistent vehicle exposure across groups.

Cell Viability Assay

Cell viability was evaluated using a Cell Counting Kit-8 (CCK-8; MedChemExpress) according to the manufacturer's protocol. HCSCs were seeded in 96-well plates at a density of 5 × 10⁴ cells/well and exposed to the indicated treatments. Absorbance at 450 nm was measured using a microplate reader (Tecan, Männedorf, Sweden). Background absorbance from blank wells containing medium and reagent only was subtracted, and values were normalized to the control group.

Quantitative Real-Time PCR

Total RNA was isolated from HCSCs using an RNA extraction kit (EZBioscience, Roseville, MN, USA) according to the manufacturer's instructions. RNA concentration and purity were determined spectrophotometrically using a NanoDrop 2000 (Thermo Fisher Scientific, Waltham, MA, USA). Complementary DNA (cDNA) was synthesized from total RNA using a reverse transcription kit (Color Reverse Transcription Kit; EZBioscience). Quantitative real-time PCR (qRT-PCR) was performed using a SYBR Green–based supermix (EZBioscience) on a LightCycler 480II system (Roche Diagnostics, Indianapolis, IN, USA) with gene-specific primers. Relative gene expression levels were normalized to the internal reference gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and calculated using the comparative 2^−ΔΔCt method. Primer sequences are listed in Table 1.

Table 1.

Primer Sequences for qRT-PCR

Gene Name	Forward Primer (5′–3′)	Reverse Primer (5′–3′)
GAPDH	AGAAGGCTGGGGCTCATTTG	AGGGGCCATCCACAGTCTTC
MMP1	TGTGGACCATGCCATTGAGA	TCTGCTTGACCCTCAGAGACC
MMP2	CTGTGTCTTCCCCTTCACTTTC	GTCATCATCGTAGTTGGCTGTG
MMP3	TGATGGACCTGGAAATGTTTTG	AGGGAGTGGCCAATTTCATG
TIMP1	ACCACCTTATACCAGCGTTATGA	GGTGTAGACGAACCGGATGTC
TIMP2	GCTGCGAGTGCAAGATCAC	TGGTGCCCGTTGATGTTCTTC
TIMP3	CAGGTCGCGTCTATGATGGC	AGGTGATACCGATAGTTCAGCC
LOX	GTACAACCTGAGATGCGCGG	CCTCTGGGTGTTGGCATCAA
LUM	GGATTGGTAAACCTGACCTTCAT	GATAAACGCAGATACTGCAATGC

GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MMP1, matrix metalloproteinase-1; MMP2, matrix metalloproteinase-2; MMP3, matrix metalloproteinase-3; TIMP1, tissue inhibitor of metalloproteinase-1; TIMP2, tissue inhibitor of metalloproteinase-2; TIMP3, Tissue inhibitor of metalloproteinase-3; LOX, lysyl oxidase; LUM, lumican.

Western Blot Analysis

Equal amounts of protein (20 µg) were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE; 7.5% gels, Epizyme, Cambridge, MA, USA) and transferred onto polyvinylidene fluoride (PVDF) membranes (MilliporeSigma, Burlington, MA, USA). Membranes were incubated with primary antibodies against type 1 collagen (COL1; 1:1000; Proteintech, Rosemont, IL, USA), ARs (1:1000, Proteintech), ERα (1:1000, Proteintech), ERβ (1:1000, ImmunoWay Biotechnology, San Jose, CA, USA) and β-actin (1:10,000; Proteintech), followed by horseradish peroxidase (HRP)-conjugated secondary antibodies (1:10000; Abcam, Cambridge, UK). Protein bands were visualized using enhanced chemiluminescence (Epizyme). Band intensities were quantified using ImageJ (National Institutes of Health, Bethesda, MD, USA), and target protein expression was normalized to β-actin.

Immunofluorescence Staining

HCSCs grown on coverslips were fixed with 4% paraformaldehyde, permeabilized, and incubated with anti-ERβ primary antibodies (1:100; Cell Signaling Technology, Danvers, MA, USA), followed by Alexa Fluor 647–conjugated secondary antibodies (1:200; Solarbio). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; Solarbio). Fluorescence images were acquired using a confocal laser scanning microscope (LSM 800; Zeiss, Oberkochen, Germany) under identical acquisition settings. Mean fluorescence intensity was quantified using ImageJ.

Hydroxyproline Assay

Hydroxyproline (HYP) levels in conditioned media were quantified using an acid hydrolysis-based³⁶ colorimetric hydroxyproline assay kit (Jiancheng Bioengineering Institute, Nanjing, China) according to the manufacturer's instructions. Absorbance at 550 nm was measured using the microplate reader. HYP concentrations were calculated from a standard curve generated with known standards and normalized to cell number.

Animal Model

Male Sprague Dawley rats (3 weeks old, 80–100 g) were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and housed in the specific pathogen–free (SPF) Laboratory Animal Center of the Eye Hospital of Tianjin Medical University. Based on the clinically observed elevation of E2 in male KC patients, and to minimize confounding from the pronounced endogenous estrogen fluctuations associated with the estrous cycle,³⁷ rats 3 to 4 weeks old were selected to establish a high-estrogen model. After 1 week of acclimatization, rats were randomly divided into three groups (n = 5 per group). Rats in the high-estrogen group received daily subcutaneous injections of 17β-estradiol (100 µg/kg; Sigma-Aldrich) dissolved in sesame oil.³⁸ Vehicle control animals received sesame oil alone, whereas untreated controls received no intervention. The injection volume was standardized to 0.1 mL per 100 g body weight. All procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Animal Protection and Utilization Committee of Tianjin Medical University Eye Hospital (approval no. SYXK-2023-0002).

Corneal Thickness Measurement

Central corneal thickness (CCT) was measured in vivo using anterior segment optical coherence tomography (AS-OCT; Optovue, Fremont, CA, USA). Rats were anesthetized, and one eye per animal was randomly selected and positioned perpendicular to the incident scanning beam. Horizontal B-scans centered on the corneal apex were acquired, and three images without motion artifacts were selected for analysis. CCT was defined as the perpendicular distance between the anterior epithelial surface and the posterior endothelial boundary at the corneal center. Measurements were obtained at baseline and at 2 and 4 weeks after treatment (n = 5 per group).

Transmission Electron Microscope Imaging

For ultrastructural analysis, central corneas were excised immediately after euthanasia and sectioned into approximately 1-mm³ stromal blocks. Samples were fixed in 2.5% glutaraldehyde and 1% osmium tetroxide, dehydrated with ethanol solutions, embedded with LR White ultrathin sections, and observed by a transmission electron microscope (HT7700; Hitachi, Tokyo, Japan).

Biomechanical Testing

After the rats were killed, cornea–sclera complexes were isolated, maintained in saline, and radially cut into rectangular sclera–cornea–sclera strips (6–8 mm) with the cornea centered. Thickness and width were measured to determine cross-sectional area. Uniaxial tensile testing was performed using a mechanical tester equipped with a 10N load cell (Care International, Shanghai, China). Scleral ends were clamped with the cornea positioned between the grips, and the initial gauge length was recorded. Engineering stress and strain were calculated as = F/A and ε = ΔL/L₀, respectively. Stress–strain curves were analyzed using Prism (GraphPad, Boston, MA, USA), and Young's modulus was determined at 2%, 5%, and 10% strain levels (n = 5 per group).

Enzyme-Linked Immunosorbent Assay

Estradiol (E2), testosterone (T), and matrix metalloproteinase-2 (MMP2) concentrations in human plasma and tear samples were quantified using commercially available enzyme-linked immunosorbent assay (ELISA) kits (E-OSEL-H0005, E-OSEL-H0007, and E-EL-H1445; Elabscience, Houston, TX, USA) according to the manufacturer's instructions. For animal experiments, rat blood was collected by cardiac puncture, allowed to clot at room temperature for 30 minutes, and centrifuged at 3000g for 15 minutes at 4°C to obtain serum. Rat E2 levels were measured using a species-specific ELISA kit (E-OSEL-R0001; Elabscience) following the same procedures. For all assays, concentrations were calculated from standard curves, and tear samples were normalized to total protein. All samples were analyzed in parallel using the same kit lot to minimize inter-assay variability.

Statistical Analysis

All quantitative data are presented as mean ± standard deviation (SD). Statistical analyses were conducted using Prism 10.0.1. Normality of continuous variables was evaluated using the Shapiro–Wilk test. Variables with normal distributions were compared using the Student's t-test for two groups or one-way analysis of variance (ANOVA) with Tukey's post hoc test for three or more groups. For non-normally distributed variables, the Mann–Whitney U test was used for two-group comparisons, and the Kruskal–Wallis test followed by Dunn's post hoc test with Bonferroni adjustment was applied for multiple-group comparisons. Correlations among E2, T, and KC-related parameters were analyzed using Pearson's correlation analysis, whereas correlations among E2, T, and MMP2 levels were evaluated using Spearman's rank correlation analysis due to non-normal distribution. Categorical variables were analyzed using Fisher's exact test. A two-tailed P value of less than 0.05 was considered statistically significant.

Results

Plasma and Tear E2, T, and MMP2 Levels in KC Patients

A total of 27 subjects were enrolled in this study, including 15 KC patients and 12 age- and sex-matched controls. Demographic characteristics are summarized in Supplementary Table S1. The male-to-female ratio was 10:5 in the KC group (mean age, 20.53 ± 2.45 years) and 7:5 in the control group (mean age, 19.67 ± 2.19 years), with no significant differences in age or sex distribution. Tear samples were obtained from a subset of the subjects who provided plasma samples. Normal distribution assessments showed that E2 and T levels in both plasma and tear samples followed a normal distribution, whereas tear MMP2 levels did not meet normality assumptions.

Given that E2 and T represent the principal circulating estrogen and androgen, respectively, their plasma concentrations were quantified by ELISA. KC patients exhibited significantly higher plasma E2 levels compared with controls, whereas plasma T levels did not differ (Fig. 1A). Sex-stratified analysis showed elevated plasma E2 in male KC patients relative to male controls, with no difference in plasma T. Female KC patients showed no significant differences in plasma E2 or T compared with female controls (Fig. 1D).

Figure 1.

Estradiol, testosterone, and MMP2 levels in plasma and tears of patients with KC. (A) Plasma levels of E2 and T in patients with KC (n = 15) and controls (n = 12). (B) Tear levels of E2 and T in patients with KC (n = 12) and controls (n = 10). (C) Tear MMP2 levels in patients with KC (n = 12) and controls (n = 10). (D) Sex-specific plasma levels of E2 and T in patients with KC (10 men, 5 women) and controls (7 men, 5 women). (E) Sex-specific tear levels of estradiol and testosterone in patients with KC (7 men, 5 women) and controls (6 men, 4 women). (F) Sex-specific tear MMP2 levels in patients with KC (7 men, 5 women) and controls (6 men, 4 women). Data are represented as mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

To further characterize the local hormonal microenvironment of the ocular surface, tear levels of E2, T, and MMP2 were also assessed. Tear E2 and MMP2 levels were also significantly higher in KC patients, whereas tear T levels were comparable between groups (Figs. 1B, 1C). Consistent with the plasma findings, sex-stratified analysis demonstrated increased tear E2 and MMP2 in male KC patients relative to male controls, whereas tear T levels remained similar. No significant differences in tear E2, T, or MMP2 levels were observed in females (Figs. 1E, 1F).

Correlations Between Tear Estradiol Levels and KC-Related Parameters

The clinical classification of KC patients in the tear cohort is presented in Table 2. With increasing KC severity, mean corneal keratometry (K) increased, minimum corneal thickness decreased, and best-corrected visual acuity (BCVA) progressively worsened. All interstage differences were statistically significant. Correlation analysis demonstrated that tear E2 levels were positively correlated with mean corneal K (r = 0.6764, P = 0.0005) (Fig. 2A) and diopter (r = −0.64427, P = 0.0013) (Fig. 2C) and were negatively correlated with minimum corneal thickness (r = −0.6487, P = 0.0011) (Fig. 2B) and BCVA (r = 0.5361, P = 0.0101) (Fig. 2D). In addition, tear E2 levels showed a positive correlation with MMP2 expression (r = 0.5178, P = 0.0136) (Fig. 2E). In contrast, no significant correlations were observed between tear T levels and mean corneal K, minimum corneal thickness, diopter, BCVA, or MMP2 (all P > 0.05) (Supplementary Figs. S1A–S1E).

Table 2.

Topographic Characteristics of the Cornea for Different Grades of KC (Determined by ANOVA)

		Mean ± SD
		BCVA (logMAR)	Diopter (D)	Mean Central K (D)	Minimum Corneal Thickness (µm)
Control	Total (n = 10)	0.00 ± 0.00	−3.65 ± 0.78	42.13 ± 0.57	543.40 ± 30.70
	Male (n = 6)	0.00 ± 0.00	−3.50 ± 0.69	42.17 ± 0.69	550.17 ± 34.43
	Female (n = 4)	0.00 ± 0.00	−3.88 ± 1.01	42.08 ± 0.40	533.25 ± 24.94
Grade 1	Total (n = 3)	0.03 ± 0.06	−4.00 ± 0.90	45.80 ± 1.08	479.00 ± 13.11
	Male (n = 2)	0.05 ± 0.07	−4.13 ± 1.24	46.25 ± 1.06	480.00 ± 18.38
	Female (n = 1)	0.00	−3.75	44.9	447
Grade 2	Total (n = 6)	0.27 ± 0.09	−5.50 ± 2.62	48.22 ± 2.52	458.67 ± 37.41
	Male (n = 2)	0.23 ± 0.11	−6.38 ± 1.94	45.75 ± 2.76	485.50 ± 20.51
	Female (n = 4)	0.29 ± 0.09	−5.06 ± 3.07	49.45 ± 1.40	445.25 ± 38.38
Grade 3	Total (n = 3)	0.84 ± 0.77	−6.33 ± 1.38	60.50 ± 10.04	408.33 ± 63.54
	Male (n = 3)	0.84 ± 0.77	−6.33 ± 1.38	60.50 ± 10.04	408.33 ± 63.54
	Female (n = 0)	—	—	—	—
P	Total	<0.001	0.049	<0.001	<0.001

Figure 2.

Correlations between tear E2 levels and KC-related parameters. Shown are correlation analyses between tear E2 concentrations and mean central keratometry (A), minimum corneal thickness (B), diopter (C), best-correct visual acuity (D), or MMP2 (E) (n = 22). ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

Sex-stratified analysis showed that male KC patients exhibited significant correlations between tear E2 and mean central K, minimum corneal thickness, diopter, BCVA, and MMP2 (r = 0.7219, r = −0.7703, r = −0.7928, r = −0.5628, and r = 0.7637, respectively), whereas no significant correlations were observed in the female subgroup (all P > 0.05) (Supplementary Fig. S2).

Effects of Sex Hormone Imbalance on MMP/TIMP Homeostasis and Collagen Metabolism in HCSCs

To evaluate the impact of sex hormone imbalance on corneal collagen metabolism, sex hormone–imbalance models were established in HCSCs, followed by evaluation of MMP/TIMP expression, COL1, and collagen degradation products. CCK-8 assays showed that cell viability remained above 100% in all treatment groups at 24 hours, with no evidence of cytotoxicity compared with controls (Fig. 3A). This indicated that subsequent changes in collagen metabolism were not attributable to reduced cell survival. qRT-PCR analysis demonstrated that MMP2 messenger RNA (mRNA) expression was significantly increased in the lower testosterone treatment (L-T) group relative to controls. In the higher estradiol treatment (H-E) group, both MMP2 and MMP3 mRNA expression were upregulated, whereas TIMP2 expression exhibited a decreasing trend, resulting in an increased MMP2/TIMP2 ratio (Figs. 3B–D). In parallel, H-E exposure downregulated ECM-associated genes, including LOX and lumican (LUM) (Figs. 3E, 3F). Consistent with these transcriptional changes, western blotting revealed reduced COL1 protein levels in H-E–treated cells (Figs. 3G, 3H). Furthermore, collagen degradation product, HYP, was significantly elevated under H-E conditions (Fig. 3I), supporting enhanced collagen degradation.

Figure 3.

High E2 exposure disrupted MMP/TIMP homeostasis and promoted collagen degradation in HCSCs. (A) Cell viability curves assessed by CCK-8 at 0, 6, 12, and 24 hours stimulated by six conditions (n = 5). (B–D) qRT-PCR analysis of MMP and TIMP mRNA expression and the MMP2/TIMP2 ratios in HCSCs under different hormone treatments (n = 3–5). (E, F) qRT-PCR analysis of LOX and LUM mRNA expression in HCSCs under different hormone treatments (n = 3 or 4). (G, H) Representative western blots of COL1 protein expression in HCSCs under two high-estrogen treatments, along with fold changes in the band gray level (n = 4). (I) HYP content in conditioned medium from HCSCs treated under two high-estrogen conditions (n = 6). Data are presented as mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

ERβ Mediates High Estrogen–Induced Collagen Degradation in HCSCs

As steroid hormones, the biological effects of sex hormones depend on the mediation of corresponding specific receptors.³⁹ The expression of estrogen receptor-α (ERα), ERβ, and AR in HCSCs was evaluated following H-E and L-T treatments. Western blotting showed increased ERβ protein levels in H-E–treated cells compared with controls, whereas ERα and AR remained unchanged. No significant changes in ERα, ERβ, or AR expressions were observed in L-T–treated cells (Figs. 4A–D). Immunofluorescence staining further assessed ERβ intracellular localization and relative abundance. ERβ was predominantly localized in the nuclear of HCSCs, with increased fluorescence intensity observed in H-E–treated cells (Figs. 4E, 4F). These results indicated a selective activation of ERβ in HCSCs in response to high estrogen levels.

Figure 4.

Expression of sex hormone receptors under high-estrogen conditions. (A–D) Representative western blots of AR, ERα, and ERβ protein expression in HCSCs under high-estrogen conditions, along with fold changes in the band gray level (n = 3–5). (E, F) Immunofluorescent staining for ERβ in HCSCs of the two high-estrogen treatments, along with fold changes in mean intensity (n = 4). Scale bars: 20 µm. Data are presented as mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

To further investigate the role of ERβ in HCSCs, a selective ERβ antagonist, PHTPP, was treated in HCSCs. Both immunofluorescence staining and western blotting demonstrated a significant reduction in ERβ protein levels following PHTPP treatment, confirming effective receptor inhibition (Figs. 5A, 5G, 5H). At the transcriptional level, MMP2 mRNA expression was decreased, whereas TIMP2 expression was increased, resulting in restoration of the MMP2/TIMP2 ratio toward control levels (Figs. 5D–F). Consistently, COL1 expression was elevated after ERβ inhibition (Figs. 5A, 5C). In addition, HYP content was reduced compared with the H-E group (Fig. 5I). These results showed that inhibition of ERβ activity attenuated high-estrogen–induced collagen degradation and partially restored collagen homeostasis in HCSCs.

Figure 5.

PHTPP partially reversed ERβ-associated type I collagen degradation in HCSCs. (A–C) Representative western blots of ERβ and COL1 protein expressions in HCSCs following PHTPP treatment, along with fold changes in the band gray level (n = 4). (D–F) qRT-PCR analysis of MMP2, TIMP2, and MMP2/TIMP2 mRNA expression in HCSCs following PHTPP treatment (n = 5). (G, H) Immunofluorescent staining for ERβ in HCSCs following PHTPP treatment, along with fold changes in mean intensity (n = 4). Scale bars: 20 µm. (I) Hydroxyproline content in conditioned medium from HCSCs following PHTPP treatment (n = 6). Data are presented as mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

Effects of Estradiol Exposure on Corneal Thickness, Collagen Arrangement, and Biomechanical Properties in Rats

Analysis of clinical samples demonstrated an imbalance in circulating sex hormones in patients with KC, characterized by elevated estrogen levels. Serum E2 concentrations were significantly associated with clinical parameters reflecting KC severity. In vitro experiments further showed that high-estrogen exposure induced a series of molecular alterations related to collagen metabolism in HCSCs. However, the direct effects of elevated estrogen on corneal tissue structure and biomechanical properties remained unclear. To address this, a high-estrogen animal model was established to evaluate the structural and mechanical effects of E2 on the cornea. CCT, collagen organization, and biomechanical properties were subsequently assessed (Fig. 6A).

Figure 6.

Effects of 4-week E2 exposure on CCT, collagen arrangement, and biomechanical properties in rats. (A) Schematic of experimental design. (B) Representative AS-OCT images of corneas at 0 and 4 weeks. (C) Serum estradiol levels at 4 weeks (n = 5). (D) TEM images showing layer-specific collagen fibril arrangement in anterior, middle, and posterior stroma. Red arrows indicate longitudinally oriented CFs; blue arrows indicate transversely oriented CFs. Scale bars: 1 µm and 2  µm. (E) Quantitative analysis of CCT at 0, 2, and 4 weeks (n = 5). (F) Representative stress–strain curves of corneas. (G) Quantitative analysis of Young's modulus under 2%, 5%, and 10% strain (n = 5). Data are presented as mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001.

Serum E2 levels were significantly elevated in the rat high-estradiol group (R-HE) group compared with controls after 4-week exposure (Fig. 6C), indicating successful establishment of the high-estrogen model. AS-OCT measurements showed no differences in CCT between the untreated normal control group (NC) and control groups at any time point. In contrast, R-HE rats exhibited a significant reduction in CCT at 4 weeks compared with both controls (Figs. 6B, 6E). Transmission electron microscopy (TEM) analysis revealed layer-specific alterations in stromal corneal collagen fibrils (CFs) after 4-week estradiol exposure (Fig. 6D). In controls, anterior stromal CFs were densely packed in lamellae, and middle and posterior stromal CFs were regularly aligned with intersecting lamellae. R-HE corneas showed looser anterior stromal lamellae with reduced fibril uniformity and decreased parallelism and compactness in middle and posterior stroma with local disorganization, indicating disruption of the ordered corneal collagen arrangement.

Corneal biomechanical properties were further assessed by uniaxial tensile testing. Stress–strain curves showed lower stress responses across increasing strain levels in R-HE corneas (Fig. 6F). Quantitative analysis of Young's modulus under 2%, 5%, and 10% strain conditions is presented in Figure 6G. Although no significant difference was observed at 2% strain, the Young's modulus in R-HE corneas was significantly reduced at 5% and 10% strain. These results indicated a significant decrease in corneal stiffness and mechanical stability following 4-week estradiol exposure.

Discussion

KC is widely recognized as a multifactorial disorder involving genetic susceptibility, chronic eye rubbing–related mechanical stress, and persistent inflammation, with long-term repetitive eye rubbing consistently identified as a major risk factor.^12,13 Age-related physiologic corneal crosslinking and nonenzymatic glycation progressively increase stromal stiffness through accumulation of advanced glycation end products, theoretically conferring biomechanical protection.⁴⁰ However, the adolescent predominance of KC and its sex-related differences cannot be adequately explained by allergy or immature physiologic crosslinking alone, suggesting the involvement of additional systemic modulators. Increasing evidence has implicated endocrine influences in corneal homeostasis.^41,42 In the present study, we provide convergent clinical, cellular, and in vivo evidence demonstrating that sex hormone imbalance disrupts stromal collagen metabolism and weakens corneal biomechanics, supporting a contributory role of hormonal regulation in KC progression.

Clinically, both systemic and local hormonal levels were observed in KC patients. E2 levels were elevated in plasma and tears, and tear E2 concentrations correlated significantly with multiple indicators of disease severity, including mean keratometry, minimum corneal thickness, diopter, and BCVA. However, previous reports have described different systemic hormone profiles, such as unchanged salivary E2 but increased dehydroepiandrosterone levels.³⁰ These discrepancies likely reflect heterogeneity in age distribution, sampling approaches, and biological matrices. Notably, tear measurements may more directly reflect the local corneal microenvironment than systemic biofluids,⁴³ thereby strengthening the link between local hormonal exposure and stromal changes. Elevated E2 levels were observed predominantly in male patients. Because males typically exhibit lower baseline estrogen exposure,⁴⁴ relative increases may produce a greater perturbation of the corneal hormonal microenvironment and thereby exert stronger effects on stromal homeostasis, although this hypothesis warrants further validation.

Notably, the current clinical findings appear to be driven predominantly by the male subgroup, whereas female-specific effects remain uncertain. The lack of significant hormonal or MMP2 differences in the female subgroup may be related to the limited sample size and the absence of menstrual-cycle staging, factors that introduce substantial physiological variability and reduce statistical power. Because estrogen fluctuates markedly across the menstrual cycle, pregnancy, and menopause, timing effects may obscure biologically relevant associations. Clinical observations of KC progression during pregnancy,²⁶ with systemic estrogen therapy,²⁸ and during in vitro fertilization⁴⁵ further suggest that estrogen-related effects in females may be context dependent rather than absent. Larger, well-controlled cohorts with hormonal staging will be required to further clarify sex-specific mechanisms in KC.

Consistent with these findings, tear MMP2 concentrations were also increased and positively correlated with E2 levels, suggesting MMP2 as a potential molecular link between hormonal signaling and stromal remodeling. Among the MMPs implicated in KC, MMP2 is constitutively expressed in the corneal stroma and primarily mediates basal turnover of COL1 and basement membrane components, which may better reflect chronic ECM dysregulation.⁴⁶ KC is characterized by progressive thinning, protrusion, and biomechanical weakening, changes that reflect disruption of stromal ECM homeostasis, particularly dysregulated collagen metabolism. Persistent MMP2 activation combined with insufficient TIMP-mediated inhibition would favor chronic proteolysis and collagen loss.⁴⁷ These observations led us to hypothesize that hormonal imbalance promotes KC progression through modulation of collagen metabolism.

Our in vitro data further support this hypothesis. High estrogen exposure shifted HCSCs toward a degradative phenotype, characterized by disruption of the MMP/TIMP balance, reduced expression of collagen crosslinking and ECM-stabilizing genes (including LOX and LUM), decreased COL1 protein levels, and increased hydroxyproline release. Together, these changes indicate that elevated estrogen impaired collagen stabilization and enhanced matrix degradation. In contrast, androgen deficiency induced only modest increases in MMP2 transcription without parallel changes in collagen degradation products, suggesting a comparatively limited role in active stromal remodeling. Prior studies similarly describe androgen signaling as cell type–specific and primarily homeostatic rather than directly degradative,^48,49 supporting a potential supportive or protective role in maintaining corneal stability.

Sex hormones regulate cellular function primarily through binding to specific nuclear receptors that modulate downstream gene transcription.³⁹ To further define the mechanism underlying these estrogen-associated matrix changes, sex hormone receptor subtypes were examined in HCSCs. ERβ, but not ERα or ARs, was selectively upregulated under high-estrogen conditions. Pharmacologic inhibition of ERβ partially reversed estrogen-induced increases in the MMP2/TIMP2 ratio, restored COL1 expression, and reduced collagen degradation products. These findings support a predominant role for an E2-ERβ-MMP2 signaling axis linking hormonal exposure to stromal matrix remodeling.

Although cell-based models provide valuable mechanistic insights, they do not fully recapitulate the structural and biomechanical complexity of intact corneal tissue. Previous ex vivo studies using supraphysiologic β-estradiol exposure reported reduced corneal Young's modulus, suggesting estrogen-associated biomechanical weakening.⁵⁰ However, direct in vivo evidence remains limited. To extend these findings in vivo, we developed a high-estrogen rat model to mimic the hormonal milieu of KC. Male rats were selected based on the more pronounced estradiol elevation observed in male KC patients and to avoid the large estrous-cycle fluctuations in females that confound exogenous estrogen studies. Sustained estradiol exposure resulted in reduced CCT, disorganized collagen fibril architecture, and decreased stromal density. Such disruption of collagen architecture is consistent with diminished resistance to deformation.⁵¹ These structural alterations translated into measurable biomechanical impairment, as evidenced by a significant reduction in Young's modulus on tensile testing. Notably, these changes mirror hallmark features of KC,⁵² thereby extending our mechanistic findings to the tissue level and providing direct evidence that estrogen imbalance can compromise corneal stroma structure and biomechanical stability in vivo.

Research limitations are inevitable. The relatively modest clinical sample size may limit the robustness of sex-specific inferences. In addition, detailed information regarding menstrual cycle phase was not available in female participants, which may introduce variability in circulating sex hormone levels. These factors should be considered when interpreting the observed sex-stratified associations. Further studies with larger, well-characterized cohorts and systematic assessment of menstrual cycle–related hormonal variability are warranted. Moreover, in vitro HCSC models cannot fully recapitulate the complex corneal microenvironment, including interactions with epithelial and endothelial cells, immune components, and tear film factors. Finally, the downstream molecular mechanisms regulated by ERβ remain incompletely defined. Future studies integrating multiomics approaches may systematically clarify ERβ-dependent signaling networks, providing deeper insight into the molecular basis of sex hormone–induced corneal remodeling and identifying potential therapeutic targets.

In summary, sex hormone imbalance, particularly elevated estrogen levels, plays a pivotal role in corneal collagen metabolism disruption and biomechanical destabilization. Dysregulation of the MMP/TIMP balance and activation of the E2–ERβ–MMP2 pathway shifts stromal remodeling toward proteolysis, resulting in reduced CCT, disorganized collagen fiber architecture, and significantly impaired corneal biomechanical properties. When considered together with the clinical features of KC, our mechanistic results suggest a clear biological route by which estrogen-related signaling could drive abnormal stromal remodeling. Collectively, the study extends the current conceptual framework of KC pathophysiology by highlighting sex hormone–responsive matrix dynamics as a potentially relevant component of disease progression.