Fuchs’ Endothelial Corneal Dystrophy and Cardiometabolic Comorbidities

Unstructured Abstract

In this retrospective study utilizing data from the National Institutes of Health All of Us Research Program, individuals with Fuchs’ endothelial corneal dystrophy (n=548) exhibited a higher burden of cardiometabolic disease than matched controls (n=5,480).

Report

Fuchs’ endothelial corneal dystrophy (FECD) is the leading indication for keratoplasty in the United States (U.S.).¹ Preclinical studies have identified multiple pathophysiological mechanisms in FECD, including oxidative stress-induced apoptosis, aberrant extracellular matrix deposition, and mitochondrial dysfunction.² Similar pathways have been implicated in cardiometabolic conditions, including type 2 diabetes mellitus (T2DM), hypertension (HTN), hypercholesterolemia, coronary artery disease (CAD), myocardial infarction (MI), congestive heart failure (CHF), cerebrovascular disease, peripheral artery disease (PAD), non-alcoholic fatty liver disease (NAFLD), and obesity.³ Given the rising prevalence of cardiometabolic disorders, we investigated whether such conditions are associated with FECD.

We analyzed de-identified data from the Registered Tier Dataset (V8) of the National Institutes of Health All of Us (AoU) Research Program, approved by the AoU Institutional Review Board. The AoU database links electronic health records, demographics, and survey responses from participants who provided informed consent, with purposeful recruitment from groups historically underrepresented in biomedical research.⁴ Participants enroll through a participating health care provider organization or online as direct volunteers.⁴ Additional research ethics board review was not required. We adhered to the Declaration of Helsinki and followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Cases of FECD were identified using the following Observational Medical Outcomes Partnership (OMOP) concepts: 443740 (“Corneal endothelial dystrophy”), 4109696 (“Fuchs’ corneal dystrophy”), or 4162284 (“Corneal guttata”). The earliest diagnosis constituted the index date. Controls had no history of FECD-related OMOP concepts and were observable on the matched case’s index date.

We performed 1:10 greedy nearest-neighbour propensity score matching using year of birth, body mass index (BMI), sex, race/ethnicity, smoking history, alcohol use, educational attainment, household income, health insurance status, and eye care utilization frequency. Eligibility was limited to participants who had an eye examination by an eye care professional (i.e., ophthalmologist or optometrist) within the past year. An absolute standardized mean difference (SMD) <0.1 indicated negligible imbalance.

Statistical analyses were conducted in Python. The primary analysis examined associations between FECD and pre-index cardiometabolic comorbidities (T2DM, HTN, hypercholesterolemia, CAD, MI, CHF, cerebrovascular disease, PAD, NAFLD, and obesity) via conditional logistic regressions accounting for matching. The secondary analysis examined associations between FECD and cardiometabolic comorbidities, irrespective of diagnosis order. Exploratory analyses examined cardiometabolic diagnosis timing relative to FECD and age at first cardiometabolic diagnosis. P-values were adjusted using the Benjamini-Hochberg false discovery rate procedure.

548 participants with FECD were matched to 5,480 unique controls. The mean age and BMI were 63.6 ± 9.7 years and 29.1 ± 6.6 kg/m², respectively. Most participants were female (59.7%) and self-identified as White (84.8%). Participant characteristics are summarized in Table S1 (available at www.aaojournal.org).

Percentages denote the prevalence of each condition among FECD cases and matched controls, respectively. Compared to matched controls, participants with FECD had higher odds of prior T2DM (19.3% vs. 15.5%), HTN (52.7% vs. 41.1%), hypercholesterolemia (25.9% vs. 20.9%), cerebrovascular disease (9.7% vs. 6.0%), and obesity (26.5% vs. 19.7%) (all p_adjusted≤0.03). Odds of prior CAD, CHF, MI, NALFD, or PAD did not differ. These associations are summarized in Figure 1.

Figure 1.

Forest plot illustrating the odds of having a history of cardiometabolic disorders among participants with compared to those without Fuchs’ endothelial corneal dystrophy. Filled circles denote statistically significant associations (p_adjusted<0.05), while hollow circles denote non-significant associations (p_adjusted>0.05). Error bars represent 95% confidence intervals. Conditional comparisons were performed within strata matched for age, sex, body mass index, self-reported race or ethnicity, health insurance status, smoking, alcohol use, education, household income, and eye examination frequency.

Relative to individuals without cardiometabolic disease, the odds of FECD increased progressively among those with a history of one cardiometabolic condition (OR=1.54, 95% CI=[1.20, 1.96], p<0.01), two cardiometabolic conditions (OR=2.05, 95% CI=[1.58, 2.66], p<0.01), and three or more cardiometabolic conditions (OR=2.12, 95% CI=[1.65, 2.72], p<0.01).

Secondary analysis results are presented in Table S2 (available at www.aaojournal.org). All findings were consistent with the primary analysis, except that the odds of ever having NAFLD were higher in the FECD group (6.0% vs. 3.8%; p_adjusted=0.04) and the odds of ever having T2DM were similar between groups (27.9% vs. 24.1%; p_adjusted=0.07).

Figure S1 (available at www.aaojournal.org) depicts median intervals between cardiometabolic diagnoses and FECD diagnosis, contextualizing the time by which each comorbidity was documented relative to the index date. Individuals with FECD and controls were diagnosed with each cardiometabolic condition at a similar age (all p_adjusted>0.05).

In this large, diverse U.S. cohort, individuals with FECD had a greater prevalence of common cardiometabolic comorbidities than matched controls. Further research is warranted to clarify underlying mechanisms and clinical implications.

Individuals with FECD were more likely to have a history of T2DM than matched controls, complementing recent U.S. Department of Veterans Affairs data reporting a 25% greater likelihood of T2DM among FECD cases.⁵ Hyperglycemia may be linked to corneal endothelial dysfunction through increased oxidative stress and biomechanical alterations.⁵ Shared genetic mechanisms might also be relevant, as the FECD-associated TCF4 gene has been implicated in insulin gene regulation.⁵

The stepwise increase in FECD odds with greater cardiometabolic multimorbidity suggests an association between systemic cardiometabolic burden and corneal endothelial vulnerability, potentially mediated by shared underlying mechanisms. These insights extend upon single-center U.S. data showing higher hyperlipidemia prevalence among FECD cases than controls (74% vs. 50%; p=0.02),⁶ with nonsignificant differences for HTN, CHF, CAD, peripheral vascular disease, T2DM, and stroke, likely reflecting limited statistical power (all p≥0.10).⁶ Another German cohort reported earlier FECD diagnosis in obese than non-obese individuals, with adipose-derived estrogen metabolites hypothesized to promote corneal endothelial oxidative stress.⁷

The temporal distribution of cardiometabolic diagnoses relative to FECD is subject to wide range of influences, including variability in clinical detection and documentation. Greater healthcare or eye care utilization in the setting of cardiometabolic disease may increase the likelihood that FECD is detected, if present. To mitigate this, we matched FECD cases and controls on a proxy for eye care utilization frequency, strengthening comparability between groups. However, residual detection bias remains possible, as lifetime eye examination history was unavailable for matching.

Limitations include potential diagnostic misclassification in the AoU dataset, and OMOP concept codes did not permit stratification by disease severity. Temporal analyses relied on first diagnosis dates and should be interpreted cautiously given retrospective documentation limitations. Participants with missing covariates were excluded from the matching process, potentially introducing selection bias. In a post-hoc sensitivity analysis mutually adjusting for all cardiometabolic comorbidities, only HTN remained associated with FECD; however, interpretation is limited by collinearity among interrelated conditions and potential overadjustment. Given the observational design, all results should be considered hypothesis-generating rather than causal.

Data Availability Statement:

This study used data from the All of Us Research Program’s Registered Tier Dataset Version 8, available to authorized users on the Researcher Workbench.