Higher remnant cholesterol was associated with a greater likelihood of developing diabetic retinopathy in adults with type 2 diabetes, a prospective community-based cohort study found. Among 1,580 participants who had no retinopathy at the start, 304 developed it during a mean follow-up of three years. After adjustment, participants with elevated remnant cholesterol had a 50% higher hazard of retinopathy than those with normal levels. A one-unit increase in log-transformed remnant cholesterol was linked to a 26% higher hazard.
Remnant cholesterol, or RC, was calculated by subtracting high-density lipoprotein cholesterol, known as HDL-C, and low-density lipoprotein cholesterol, or LDL-C, from total cholesterol. The study defined RC of at least 0.8 mmol/L as elevated and compared it with levels below that cutoff. RC was examined both as a continuous measure and as an elevated-versus-normal category.
What the cohort found
The researchers enrolled 2,963 people. Of those, 2,165 were free of retinopathy; 140 were excluded because baseline data were incomplete and 445 because they had no subsequent follow-up. That left 1,580 participants in the closed-cohort analysis. Elevated RC was present in 325 people, or 20.57%, at baseline.
The reported incidence rate was 7.82 per 100 person-years among people with elevated baseline RC, compared with 6.22 among those without elevated RC. The overall rate was 6.53 per 100 person-years. These are incidence rates, not cumulative percentages of participants who developed retinopathy. A hazard ratio compares how quickly events occur over follow-up. In the fully adjusted analysis, the hazard ratio for elevated versus normal RC was 1.50, with a 95% confidence interval of 1.11 to 2.03. For each one-unit increase in the natural log of RC, the adjusted hazard ratio was 1.26, with a 95% confidence interval of 1.08 to 1.47.
The sex-stratified analysis found a significant association in women but not men. In women, a one-unit increase in log-transformed RC was associated with a 37% higher hazard of retinopathy, while elevated RC was associated with a 48% higher hazard compared with normal RC. The corresponding adjusted hazard ratios were 1.37, with a 95% confidence interval of 1.12 to 1.69, and 1.48, with a confidence interval of 1.02 to 2.16. The association was also stronger among participants who had had diabetes for at least 10 years, with an adjusted hazard ratio of 1.54 and a 95% confidence interval of 1.16 to 2.03.
New cases were defined as the first retinopathy diagnosis by an ophthalmologist. Seven-field photographs were taken at baseline and follow-up and graded independently by two ophthalmologists. Disagreements, reported at less than 8%, were adjudicated by a senior ophthalmologist, and the more severely affected eye determined the incident grade. The main analysis used time-dependent Cox models, allowing RC and most other covariates to vary over time while treating age, sex, education and diabetes duration as fixed.
A signal, not a treatment result
The association remained positive in several conventional lipid-target groups. For continuous RC, the model estimated hazard ratios of 1.39 among participants with triglycerides below 1.7 mmol/L, 1.37 among those with LDL-C below 2.6 mmol/L, and 1.77 among those meeting the HDL-C target. But the elevated-versus-normal estimate in the LDL-C target group was 1.47, with a 95% confidence interval from 0.87 to 2.47. Because that interval crossed 1, the categorical comparison in that group was less certain.
Adding RC to a base model modestly improved its ability to distinguish participants who did and did not develop retinopathy. The area under the curve, or AUC, rose from 0.602 to 0.609 overall and from 0.616 to 0.629 among women. The dose-response pattern was linear, with no evidence of nonlinearity, and the analysis found a sex interaction.
Because this was an observational cohort, the results show an association rather than proof that RC causes retinopathy. They also do not show that lowering RC or adding it to routine monitoring prevents the condition. RC was calculated from total, HDL-C and LDL-C, and unmeasured lifestyle, medication and healthcare factors could still have influenced the association. The cohort consisted of Chinese adults with type 2 diabetes, so the findings may not apply to other populations or other forms of diabetes.
Results remained consistent after researchers excluded cases diagnosed during the first year of follow-up and participants taking lipid-lowering drugs. That consistency supports the stability of the association in this dataset, but it does not turn the study into a treatment test. Whether lowering RC changes retinopathy outcomes remains unanswered.
Paper data and sources
Original title: Association between remnant cholesterol and incident diabetic retinopathy in patients with type 2 diabetes: a prospective cohort study
Authors: Cheng H, Chen J, Li Y et al.
Journal/Repository: Annals of medicine
Status: Peer-reviewed
First online: 2026-10-09
DOI: Not available
Original paper · Full text