A large real-world study found an association in the younger-onset group alone, although formal interaction testing did not establish age at diagnosis as an effect modifier.
In a cross-sectional study of 6,996 patients with type 2 diabetes published in Endocrine, elevated serum uric acid was associated with higher odds of prevalent diabetic retinopathy (DR) among those diagnosed before 65 years of age, but not among those diagnosed at 65 years or older.1
Don’t miss out on hearing about our latest peer-reviewed articles, expert opinions, conference news, podcasts and more.
Why does this matter?
DR remains a leading cause of preventable vision loss in adults with diabetes, and identifying additional, low-cost markers that could support risk stratification is of ongoing interest. Serum uric acid is inexpensive to measure and is commonly included in metabolic or renal assessments.
Proposed mechanisms linking elevated uric acid with microvascular injury include inflammation, oxidative stress, endothelial dysfunction and vascular endothelial growth factor-related pathways. However, the study did not investigate these mechanisms directly, and the authors provide only biological context for an observational, cross-sectional association.
What did the study investigate?
Researchers analyzed electronic health records from a single tertiary center, Lishui People’s Hospital in Zhejiang Province, China, covering January 2018 to June 2023. Of 7,891 patients with type 2 diabetes, 6,996 met the eligibility criteria after the exclusion of patients with comorbid retinal disease, previous ocular surgery, significant renal or hepatic impairment, autoimmune conditions or missing data.
Patients were grouped by age at diabetes onset: younger onset, defined as diagnosis before 65 years of age (n=6,004), and older onset, defined as diagnosis at 65 years or older (n=992). The authors selected this threshold a priori based on World Health Organization and American Diabetes Association definitions of an older diabetes population. It differs from the lower age thresholds, commonly under 40 or 50 years, conventionally used to define young-onset type 2 diabetes in other literature.
Elevated uric acid was defined using sex-specific thresholds of ≥420 µmol/L for men and ≥380 µmol/L for women. DR was assessed from fundus photographs graded independently by two masked ophthalmologists using the Early Treatment Diabetic Retinopathy Study protocol, with disagreements resolved by a third specialist.
The primary outcome was the presence of any DR. Multivariable logistic regression adjusted for age, sex, body mass index, blood pressure, low-density lipoprotein cholesterol, triglycerides, glycated hemoglobin, diabetes duration, estimated glomerular filtration rate (eGFR) and medication use.
What do the data show?
Overall, 1,359 of 6,996 patients (19.4%) had prevalent DR.
In the younger-onset group, DR prevalence was 25.5% among patients with elevated uric acid and 20.0% among those with normal levels. Elevated uric acid was associated with higher adjusted odds of DR in this group (adjusted odds ratio [aOR], 1.42; 95% confidence interval [CI], 1.17–1.73).
In the older-onset group, DR prevalence was 12.6% among patients with elevated uric acid and 11.5% among those with normal levels. The association was not statistically significant (aOR, 1.05; 95% CI, 0.61–1.79). This smaller stratum included 992 patients and 116 DR events overall, resulting in a relatively imprecise estimate.
A penalized-likelihood model in the older-onset group produced a near-identical estimate (aOR, 1.06), suggesting that the null result was not simply an artifact of the conventional regression method.
Critically, a formal statistical test for interaction between uric acid and age-at-onset group was not significant (P for interaction=0.295). This remained the case in sensitivity analyses using alternative age cutoffs of 40, 50 and 60 years, with all interaction P values ≥0.30.
The authors therefore state that the findings do not establish age at onset as a true effect modifier. Although the association reached statistical significance in the younger-onset group alone, the contrast between the two strata should be interpreted as descriptive rather than as confirmed evidence that the association differs according to age at diagnosis.
When DR was analyzed by severity, the estimated association was larger for proliferative DR (aOR, 2.39; 95% CI, 1.24–4.61) than for nonproliferative DR (aOR, 1.31; 95% CI, 1.09–1.58). However, only 64 proliferative cases were available, resulting in a relatively imprecise estimate.
An analysis treating uric acid as a continuous variable found an approximately linear positive association across the observed uric acid range, with no significant departure from linearity (P=0.31).
The younger-onset association persisted after additional adjustment for albuminuria (aOR, 1.37) and in a fully expanded model that included smoking, alcohol use, urate-lowering therapy and hypertension severity (aOR, 1.38).
The E-value was 1.67 for the point estimate and 1.38 for the confidence limit closest to the null, suggesting that residual confounding of moderate magnitude could potentially account for the association.
Were the findings consistent across subgroups?
No significant interaction was observed by sex.
A nominally significant interaction by eGFR category (P=0.040) suggested a stronger association among patients with an eGFR below 60 mL/min/1.73 m² (aOR, 1.61) than among those with an eGFR at or above this threshold (aOR, 1.12).
A significant interaction by lipid-lowering medication use (P<0.001) suggested no association among medication users. However, the medication-user subgroup was small, and the estimate was wide and imprecise (aOR, 0.67; 95% CI, 0.27–1.69).
The authors explicitly caution that these subgroup analyses were exploratory and were not corrected for multiple comparisons. They should therefore not be interpreted as evidence that statins, fenofibrate or other lipid-lowering therapies protect against DR.
Why is this relevant for clinical practice?
The findings provide a hypothesis-generating signal in patients diagnosed with type 2 diabetes before 65 years of age.
An exploratory analysis suggested that the association may be more pronounced in patients with reduced eGFR, but this requires confirmation. The findings do not support using uric acid as a treatment target or as a basis for changing DR screening intervals.
The authors also note that renal dysfunction accompanying more advanced DR could itself increase serum uric acid levels. The association may therefore be bidirectional rather than reflecting a causal effect of uric acid on retinal damage.
Limitations
The cross-sectional design means that exposure and outcome were measured at the same time, so temporality cannot be established and reverse causation cannot be excluded.
Uric acid was measured only once. Given its known short-term biological variability, measurement error could plausibly account for some of the apparent contrast between the age-at-onset groups.
Age at onset and diabetes duration are conceptually overlapping variables, although the reported collinearity statistics were modest. The subgroup and interaction analyses were exploratory, were not corrected for multiple comparisons and included several strata with few events.
Medication exposure was recorded only as current binary use, without information about treatment duration, dose or adherence. Differences in healthcare use and retinal-screening intensity may also have introduced detection bias, because DR could only be identified in patients who underwent eye examination.
Finally, the study was conducted at a single center in one Chinese province, which limits the generalizability of the results to other populations and healthcare settings.
Clinical takeaway
In this large, cross-sectional real-world study, elevated uric acid was associated with higher odds of prevalent DR among patients diagnosed with type 2 diabetes before 65 years of age, but not in the smaller group diagnosed at 65 years or older.
However, the formal interaction test was not statistically significant, meaning that the apparent age-related contrast should be regarded as descriptive rather than as proven effect modification.
The findings are hypothesis-generating. Prospective, longitudinal studies incorporating repeated uric acid measurements are needed before this marker can inform risk stratification or clinical decision-making.
Reference
- Fan M, Li J, Zhou Q, Shao H, Ni L. Age at diabetes onset modifies the association between hyperuricemia and diabetic retinopathy: a real-world cross-sectional study. Endocrine. 2026;91:237. doi:10.1007/s12020-026-04703-7
Cite: Elevated uric acid and diabetic retinopathy by age at diabetes onset. touchOPHTHALMOLOGY. July 17, 2026.
Disclosure: Also featured on touchIMMUNOLOGY.com and available here. This content was created by the touchOPHTHALMOLOGY team utilizing AI as an editorial tool (ChatGPT [large language model]. OpenAI). The content was developed, fact-checked and edited by human editors. The authors declared no competing interests.
Editor: Nicola Cartridge, Director of Content
