Abstract
This study aimed to evaluate the independent and joint effects of adherence to healthy dietary patterns and slower biological aging on the incidence of diabetic microvascular complications in individuals with type 2 diabetes mellitus (T2DM), and to assess the mediating role of biological aging. In a prospective cohort of 13,294 T2DM participants without baseline DMCs, dietary quality was assessed using a validated 10-point score, while biological aging was calculated from nine biomarkers and chronological age. Cox regression models were used to assess associations, and mediation analysis was performed to estimate the mediating effects of biological aging. Over a mean follow-up of 11.9 years, 3197 participants developed DMCs, including 1392 cases of diabetic retinopathy (DR), 1908 of diabetic nephropathy (DN), and 598 of diabetic neuropathy (DPN). Higher dietary scores (6–10) were associated with reduced risks of composite DMCs (HR 0.845; 95% CI 0.742–0.962), DR (0.804; 0.659–0.981), and DN (0.766; 0.643–0.911), but not DPN. Phenotypic age acceleration (PhenoAgeAccel) ≤ 0 was also linked to a reduced risk of DMCs. In addition, biologically younger with higher dietary score (6–10 points) had 39.4%, 30.8%, 53.6%, and 41.9% lower risk of composite DMCs, DR, DN, and DPN, respectively. Mediation analysis revealed that PhenoAgeAccel accounted for 43.0%, 29.8%, and 33.5% of the diet association with composite DMCs, DR, and DN, respectively. The results suggest that healthier dietary patterns and slower biological aging can reduce the risk of DMCs in T2DM patients, with a substantial portion of the dietary benefits mediated through slower aging. Integrating dietary and aging-targeted interventions may offer a promising method to reduce DMC risk in T2DM.
Graphical Abstract
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1007/s11357-025-02002-z.
Keywords: Biological aging, Diabetic microvascular complication, Healthy dietary pattern, Mediation analysis
Introduction
Diabetic microvascular complications (DMCs), including diabetic retinopathy (DR), diabetic nephropathy (DN), and diabetic neuropathy (DPN), remain major causes of vision loss, renal failure, lower limb amputations, and impaired quality of life among individuals with diabetes [1–4]. Despite advances in glycemic control and comprehensive diabetes management, the burden of microvascular complications remains substantial, particularly in elder population [5, 6].
Increasing evidence suggests that dietary habits play a crucial role in the development and progression of diabetic complications. Adherence to a healthy dietary pattern rich in fruits, vegetables, whole grains, and healthy fats has been associated with improved metabolic profiles and reduced systemic inflammation [7, 8]. However, most studies have primarily focused on macrovascular or cardiometabolic outcomes [9], with limited evidence regarding the impact of dietary quality on the onset of microvascular complications in individuals with diabetes.
In addition, aging is a well-established risk factor for microvascular injury, characterized by impaired vasodilatory capacity, barrier disruption, and capillary rarefaction across multiple organs [10–13]. However, chronological age fails to capture interindividual heterogeneity in cumulative physiological decline [14]. Phenotypic age, a composite measure derived from chronological age and nine clinical biochemical biomarkers based on a Gompertz mortality model, provides a scalable estimate of biological aging and predicts morbidity, mortality, and chronic disease risk across diverse populations [15–17]. Individuals with accelerated biological aging (i.e., PhenoAge higher than chronological age, termed PhenoAge acceleration, PhenoAgeAccel) exhibit an increased risk of mortality and cardiovascular diseases (CVD) [18–20]. Emerging evidence further indicates that environmental and lifestyle factors, including unhealthy dietary habits, may accelerate biological aging [21, 22]. However, it remains unclear whether healthy dietary patterns and slower biological aging independently or jointly reduce the risk of DMCs, and to what extent biological aging mediates the association between diet and microvascular complications in diabetes.
In this study, we conducted a large-scale prospective cohort study using data from the UK Biobank to investigate the independent and combined associations of dietary patterns and PhenoAge with the incidence of diabetic microvascular complications. Furthermore, we evaluated the potential mediating role of PhenoAge in the relationship between dietary quality and DMC risk.
Methods
Study population
The UK Biobank is a prospective cohort of 502,144 adults aged 37–73 from 22 UK centers (2006–2010) [23]. Participants provided lifestyle, medical, and physical data, along with blood and urine samples for multi-omic analyses. The study was approved by the North West Multicenter Research Ethics Committee. This study has been approved under the UK Biobank application. For the current analysis, 20,127 participants with T2DM at baseline were identified using a validated UK Biobank algorithm [7, 24]. After excluding participants with eGFR < 60 mL·min⁻1·1.73 m2 (n = 1950), a history of CVD (n = 2815; Field ID 20002, non-cancer illness code, self-reported; excluding “deep venous thrombosis (dvt),” “heart arrhythmia,” “pulmonary embolism,” “stroke,” “venous thromboembolic disease,” “arterial embolism,” “peripheral vascular disease,” “vasculitis,” “aortic aneurysm,” “atrial flutter,” “angina,” and “cardiomyopathy”), missing baseline data on smoking or alcohol status at baseline (n = 157), incomplete clinical biomarkers required for the calculation of phenotypic age (n = 1702), and prevalent DR, DN, and DPN at baseline (n = 209), a total of 13,294 participants with T2DM were included in the final analytic cohort.
Dietary score assessment
Dietary patterns were evaluated using a healthy dietary score based on evidence-based recommendations for cardiovascular health [8, 25]. The score included ten dietary components: high intakes of fruits (Field ID 1309, 1319), vegetables (1289, 1299), whole grains (1438, 1448, 1458, 1468), fish (1329, 1339), dairy products (1408, 1418), and vegetable oils (1428, 2654, 1438); low intakes of refined grains (1438, 1448, 1458, 1468), processed meat (1349, 3680), unprocessed red meat (1359, 1369, 1379, 1389, 3680), and sugar-sweetened beverages (6144). Participants received one point for each target achieved, yielding a total score of 0–10, with higher scores indicating better diet quality. Scores were grouped as low (0–3), moderate (4–5), and high (6–10).
Biological aging assessment
PhenoAge was computed via the published Gompertz mortality algorithm using albumin, creatinine, glucose, C-reactive protein, lymphocyte percentage, mean corpuscular volume, red cell distribution width, alkaline phosphatase, white blood cell count, and chronological age [17]. PhenoAgeAccel was defined as the residual of PhenoAge regressed on chronological age at baseline. PhenoAgeAccel > 0 indicates accelerated aging (biologically older), whereas PhenoAgeAccel ≤ 0 indicates decelerated aging (biologically younger).
Study outcomes
Outcomes were identified using ICD-10 codes from hospital and mortality records: diabetic retinopathy (DR: E11.3, E14.3, H28.0, H36.0), diabetic nephropathy (DN: E11.2, E14.2, N18.0–N18.5, N188, N18.9), and diabetic peripheral neuropathy (DPN: E11.4, E14.4, G59.0, G62.9, G63.2, G99.0) [7]. Only incident cases after baseline were included. Follow-up spanned from baseline to the earliest of outcome occurrence, death, loss to follow-up, or October 31, 2022.
Covariate assessment
Covariates included age, sex, ethnicity, Townsend deprivation index, education, smoking, alcohol use, sleep duration, BMI, HbA1c, family history of cardiovascular disease or hypertension, and use of aspirin, antihypertensive, lipid-lowering, or diabetes medications [7].
Statistical analysis
Baseline characteristics across dietary score categories were compared using ANOVA for continuous variables and chi-square tests for categorical variables. Missing values (< 10%) were imputed using sample means for normally distributed continuous variables, medians for skewed continuous variables, and modes for categorical variables. Cox proportional hazards models were used to estimate the associations between dietary score or PhenoAgeAccel with the risk of microvascular complications in participants with T2DM.
Three hierarchical models were constructed: model 1 included chronological age, sex, ethnicity, and Townsend deprivation index; model 2 included model 1 covariates plus education, family history of CVD or hypertension, and sleep duration; model 3 included model 2 covariates plus HbA1c, medication use, and BMI. Restricted cubic spline analysis was applied to assess the dose–response relationship between dietary score and the outcome, with nonlinearity tested using the Wald χ2 test. Survfit function in the survival package was used to estimate the cumulative incidences of DMCs across categories of dietary scores among individuals with T2DM.
The geom_density function in the ggplot2 package was used to estimate the distribution of PhenoAgeAccel across different disease state. Cox proportional hazards models were used to estimate the association between phenotypic ageing and risk of microvascular complications among individuals with T2DM, as well as the joint effects of dietary patterns and PhenoAgeAccel on DMCs risk. In addition, the mediation analysis was used to assess whether PhenoAgeAccel mediated the association between dietary score and microvascular complications in T2DM. Cox proportional hazards models were used to estimate HRs (95% CIs) of biologically younger according to healthy diet score among individuals with T2DM.
To assess effect heterogeneity, stratified analyses were conducted, and multiplicative interaction terms were tested between dietary score and prespecified subgroup factors, including age group (≤ 60 vs. > 60 years), sex, ethnicity, smoking status, alcohol drink status, medication use, and education. Cox proportional hazards models were further used to assess the robustness of results: (1) excluding cases occurring within the first 2 years to address reverse causality; (2) restricting to participants with complete covariate data; (3) including participants with prevalent CVD at baseline; (4) additionally adjusting for baseline eGFR in DN analyses; and (5) applying Fine and Gray subdistribution hazard models to account for competing risk of death.
All statistical analysis was performed using R software (version 4.2.1), and two-sided P-value < 0.05 was considered statistically significant.
Results
Baseline characteristics
Baseline characteristics of the study population are shown in Table 1. Among 13,294 participants with T2DM (mean age 59.8 ± 6.9 years; 62.6% male; 87.2% white), 6445 participants (48.5%) had dietary scores between 0 and 3, 5598 participants (42.1%) had scores between 4 and 5, and 1251 participants (9.4%) had scores between 6 and 10. Participants with higher diet scores tend to have lower BMI, HbA1c and less use of antihypertensive, lipid-lowering and diabetes medication (P-value < 0.001). Notably, PhenoAgeAccel decreased progressively with increasing dietary score from 0.2 ± 8.3 in the 0–3 point group to − 1.8 ± 7.9 years in the 6–10 point group. These cross-sectional differences are descriptive. As expected, several baseline characteristics differed across dietary score categories (Table 1), reflecting differences in health behaviors at study entry.
Table 1.
Baseline characteristics of participants
| Healthy dietary score | |||||
|---|---|---|---|---|---|
| Total | 0–3 | 4–5 | 6–10 | P-value | |
| Participants | N = 13,294 | N = 6445 | N = 5598 | N = 1251 | |
| Chronological age | 59.8 ± 6.9 | 59.5 ± 7.0 | 60.1 ± 6.8 | 60.4 ± 6.7 | < 0.001*** |
| Phenotypic age | 59.4 ± 10.3 | 59.6 ± 10.4 | 59.2 ± 10.2 | 58.7 ± 9.7 | 0.003** |
| PhenoAgeAccel | − 0.4 ± 8.3 | 0.2 ± 8.3 | − 0.9 ± 8.3 | − 1.8 ± 7.9 | < 0.001*** |
| Biological age group: | < 0.001*** | ||||
| Biologically older (%) | 5291 (39.8) | 2780 (43.1) | 2106 (37.6) | 405 (32.4) | |
| Biologically younger | 8003 (60.2) | 3665 (56.9) | 3492 (62.4) | 846 (67.6) | |
| Sex: | < 0.001*** | ||||
| Female (%) | 4971 (37.4) | 2096 (32.5) | 2362 (42.2) | 513 (41.0) | |
| Male (%) | 8323 (62.6) | 4349 (67.5) | 3236 (57.8) | 738 (59.0) | |
| Ethnicity: | < 0.001*** | ||||
| Others (%) | 1633 (12.3) | 625 (9.70) | 793 (14.2) | 215 (17.2) | |
| White (%) | 11,587 (87.2) | 5791 (89.9) | 4764 (85.1) | 1032 (82.5) | |
| Missing (%) | 74 (0.6) | 29 (0.5) | 41 (0.7) | 4 (0.3) | |
| Townsend deprivation index | − 0.56 ± 3.4 | − 0.45 ± 3.4 | − 0.67 ± 3.3 | − 0.64 ± 3.3 | 0.001*** |
| Smoking status: | < 0.001*** | ||||
| Current or former (%) | 7166 (53.9) | 3600 (55.9) | 2929 (52.3) | 637 (50.9) | |
| Never (%) | 6128 (46.1) | 2845 (44.1) | 2669 (47.7) | 614 (49.1) | |
| Alcohol drinking status: | < 0.001*** | ||||
| Current or former (%) | 12,194 (91.7) | 5977 (92.7) | 5081 (90.8) | 1136 (90.8) | |
| Never (%) | 1100 (8.3) | 468 (7.3) | 517 (9.2) | 115 (9.2) | |
| Education: | 0.459 | ||||
| College or university degree (%) | 10,366 (78.0) | 4987 (77.4) | 4408 (78.7) | 971 (77.6) | |
| Other levels (%) | 2696 (20.3) | 1346 (20.9) | 1092 (19.5) | 258 (20.6) | |
| Missing (%) | 232 (1.8) | 112 (1.7) | 98 (1.8) | 22 (1.8) | |
| Sleep duration (h) | 7.19 ± 1.3 | 7.22 ± 1.4 | 7.17 ± 1.3 | 7.17 ± 1.3 | 0.165 |
| Family history of CVD (%) | 7571 (57.0) | 3641 (56.5) | 3205 (57.3) | 725 (58.0) | 0.530 |
| Family history of hypertension (%) | 5410 (40.7) | 2562 (39.8) | 2340 (41.8) | 508 (40.6) | 0.074 |
| BMI (kg/m2) | 31.7 ± 5.8 | 31.9 ± 5.8 | 31.5 ± 5.7 | 31.2 ± 5.7 | < 0.001*** |
| HbA1c (mmol/mol) | 51 ± 13 | 53 ± 13 | 51 ± 13 | 50 ± 12 | < 0.001*** |
| Use of aspirin medication (%) | 6171 (46.4) | 3046 (47.3) | 2541 (45.4) | 584 (46.7) | 0.119 |
| Use of antihypertensive medication (%) | 5185 (39.0) | 2762 (42.9) | 1991 (35.6) | 432 (34.5) | < 0.001*** |
| Use of lipid-lowering medication (%) | 6144 (46.2) | 3236 (50.2) | 2366 (42.3) | 542 (43.3) | < 0.001*** |
| Use of diabetes medication (%) | 8806 (66.2) | 4477 (69.5) | 3555 (63.5) | 774 (61.9) | < 0.001*** |
Abbreviation: BMI, body mass index; CVD, cardiovascular disease
Missing smoking or alcohol data at baseline were exclusion criteria; values shown are descriptive distributions in the non-missing analytic sample
Association between dietary patterns and risk of microvascular complications among individuals with T2DM
During 157,718 person–years (mean 11.9 years; range 0.005 to 15.5 years), there were 3197 incident cases of microvascular complications (24.0%), including 1392 cases of DR (10.5%), 1908 cases of DN (14.4%), and 598 cases of DPN (4.5%). In fully adjusted models (model 3), participants with a dietary score of 6–10 had a 15.5% lower risk for composite DMCs (HR 0.845, 95% CI 0.742–0.962), a 19.6% lower risk for DR (HR 0.804, 95% CI 0.659–0.981), and a 23.4% lower risk of DN (HR 0.766, 95% CI 0.643–0.911). No significant association was observed between dietary score and DPN risk in any model (all P-value > 0.05) (Table 2). Additionally, sensitivity analyses excluding DMC cases occurring within the first 2 years yielded consistent results, reducing the likelihood of reverse causation (Table S1).
Table 2.
HRs (95% CIs) of microvascular complications according to healthy dietary score among individuals with T2DM
| Healthy dietary score | ||||||
|---|---|---|---|---|---|---|
| 0–3 | 4–5 | 6–10 | P-value | HR continuous | P-value | |
| Microvascular complications | ||||||
| Cases/person-years | 1619/75,868 | 1307/66,677 | 271/15,173 | |||
| Model 1 | 1.00 |
0.894 (0.830, 0.962) |
0.784 (0.689, 0.892) |
< 0.001*** |
0.889 (0.842, 0.938) |
< 0.001*** |
| Model 2 | 1.00 |
0.893 (0.829, 0.961) |
0.784 (0.689, 0.892) |
< 0.001*** |
0.888 (0.841, 0.938) |
< 0.001*** |
| Model 3 | 1.00 |
0.938 (0.871, 1.010) |
0.845 (0.742, 0.962) |
0.011** |
0.927 (0.878, 0.979) |
0.006** |
| Diabetic retinopathy | ||||||
| Cases/person-years | 713/6242 | 565/4950 | 114/1010 | |||
| Model 1 | 1.00 |
0.870 (0.778, 0.972) |
0.741 (0.607, 0.903) |
0.003** |
0.864 (0.796, 0.939) |
< 0.001*** |
| Model 2 | 1.00 |
0.868 (0.777, 0.970) |
0.741 (0.607, 0.903) |
0.003** |
0.864 (0.795, 0.938) |
< 0.001*** |
| Model 3 | 1.00 |
0.915 (0.818, 1.023) |
0.804 (0.659, 0.981) |
0.032* |
0.904 (0.832, 0.983) |
0.018** |
| Diabetic nephropathy | ||||||
| Cases/person-years | 979/8615 | 781/6870 | 148/1334 | |||
| Model 1 | 1.00 |
0.882 (0.802, 0.970) |
0.706 (0.593, 0.840) |
< 0.001*** |
0.858 (0.799, 0.921) |
< 0.001*** |
| Model 2 | 1.00 |
0.881 (0.801, 0.968) |
0.706 (0.593, 0.839) |
< 0.001*** |
0.857 (0.799, 0.920) |
< 0.001*** |
| Model 3 | 1.00 |
0.930 (0.846, 1.023) |
0.766 (0.643, 0.911) |
0.003** |
0.898 (0.837, 0.964) |
0.003** |
| Diabetic neuropathy | ||||||
| Cases/person-years | 294/2224 | 254/1984 | 50/419 | |||
| Model 1 | 1.00 |
1.018 (0.860, 1.206) |
0.864 (0.640, 1.167) |
0.341 |
0.965 (0.852, 1.092) |
0.568 |
| Model 2 | 1.00 |
1.016 (0.857, 1.203) |
0.864 (0.640, 1.168) |
0.342 |
0.964 (0.851, 1.091) |
0.558 |
| Model 3 | 1.00 |
1.085 (0.916, 1.286) |
0.956 (0.707, 1.292) |
0.768 |
1.019 (0.900, 1.154) |
0.761 |
Model 1: Chronological age, sex, ethnicity, Townsend deprivation index;
Model 2: Model 1 covariates plus education, family history of CVD, family history of hypertension, and sleep duration;
Model 3: Model 2 covariates plus HbA1c, use of aspirin, use of antihypertensive medication, lipid-lowing medication or diabetes medication, and BMI
Abbreviation: CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; T2DM, type 2 diabetes mellitus
Restricted cubic spline analyses revealed an approximately linear dose–response relationship between higher dietary scores and lower risk of composite DMCs, DR, and DN (all P_overall < 0.01; P_nonlinear > 0.05), whereas no significant association was observed for DPN (P_overall = 0.29; P_nonlinear = 0.16) (Fig. 1). Standardized cumulative incidence plots showed a progressively lower risk of DMCs with increasing dietary score categories (Fig. S1).
Fig. 1.
Dose–response relationship of healthy diet with the risk of microvascular complications among individuals with T2DM. Restricted cubic spline analysis was used to estimate the association between dietary score and the risk of the composite microvascular complications (a), diabetic retinopathy (b), diabetic nephropathy (c), and diabetic neuropathy (d) among individuals with T2DM. Blue zone were 95% CIs. Multivariable-adjusted models were adjusted for chronological age, sex, ethnicity, Townsend deprivation index, education, family history of CVD or hypertension, sleep duration, HbA1c, use of medicine (aspirin, antihypertensive medication, lipid-lowing medication, or diabetes medication), and BMI. All P-values for overall association were < 0.01 (except for diabetic neuropathy: P-value = 0.2946). Abbreviation: CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; T2DM, type 2 diabetes mellitus
Association between phenotypic ageing and risk of microvascular complications among individuals with T2DM
PhenoAgeAccel distributions were significantly higher among participants who developed microvascular complications compared to those without events (Fig. S2). In the fully adjusted model (model 3), biologically younger participants (PhenoAgeAccel ≤ 0) had 31.1%, 17.6%, 45.1%, and 29.9% lower risk of composite DMCs, DR, DN, and DPN, respectively. Table S2 showed that biologically younger individuals (PhenoAgeAccel ≤ 0) had significantly lower risks of microvascular complications, including composite DMCs, DR, DN, and DPN, compared to biologically older individuals.
Joint effects of dietary patterns and PhenoAgeAccel on risk of microvascular complications among individuals with T2DM
The risk of DMCs was lowest among participants who were biologically younger and adhered to a healthy diet. Compared with the high-risk reference group (biologically older with 0–3 points dietary score), those biologically younger with higher dietary score (6–10 points) had 39.4%, 30.8%, 53.6%, and 41.9% lower risk of composite DMCs, DR, DN, and DPN, respectively (Fig. 2).
Fig. 2.
Joint effects of dietary patterns and PhenoAgeAccel on the risk of microvascular complications among individuals with T2DM. Cox proportional hazard models were used to estimate joint effects of dietary patterns and PhenoAgeAccel on the risk of composite microvascular complications (a), diabetic retinopathy (b), diabetic nephropathy (c), and diabetic neuropathy (d). Dietary patterns were categorized into 0–3, 4–5, and 6–10 points dietary score. Multivariable-adjusted models were adjusted for chronological age, sex, ethnicity, Townsend deprivation index, education, family history of CVD or hypertension, sleep duration, HbA1c, use of medicine (aspirin, antihypertensive medication, lipid-lowing medication, or diabetes medication), and BMI. Abbreviation: CVD, cardiovascular disease; PhenoAgeAccel, phenotypic age acceleration; T2DM, type 2 diabetes mellitus
Mediation effect of PhenoAgeAccel between healthy dietary patterns and the risk of microvascular complications among individuals with T2DM
PhenoAgeAccel partially mediated the association between healthy dietary patterns and the risk of microvascular complications. The proportion of the effect mediated was estimated at 43.0% for composite microvascular complications, 29.8% for DR, and 33.5% for DN. No significant mediation effect was observed for DPN (Table 3; Fig. 3). Additionally, individuals with healthier dietary patterns (dietary score ≥ 5 points) were significantly more likely to be biologically younger (Table S3).
Table 3.
Mediation effect of PhenoAgeAccel on the associations between healthy diet and microvascular complications among individuals with T2DM
| Direct effect β (95% CI) |
Mediated effect β (95% CI) |
Proportion mediated effect % (95% CI) |
P-value | |
|---|---|---|---|---|
| Model 1 | ||||
| Microvascular complications | − 0.005 (− 0.012, 0.00) | − 0.004 (− 0.005, 0.00) | 0.430 (0.256, 0.960) | 0.002** |
| Diabetic retinopathy | − 0.005 (− 0.010, 0.00) | − 0.002 (− 0.003, 0.00) | 0.298 (0.168, 0.820) | < 0.001*** |
| Diabetic nephropathy | − 0.006 (− 0.011, 0.00) | − 0.003 (− 0.004, 0.00) | 0.335 (0.212, 0.690) | < 0.001*** |
| Diabetic neuropathy | 0.00 (− 0.003, 0.00) | − 0.001 (− 0.001, 0.00) | 0.863 (− 7.20, 8.180) | 0.410 |
| Model 2 | ||||
| Microvascular complications | − 0.005 (− 0.011, 0.00) | − 0.004 (− 0.004, 0.00) | 0.349 (0.196, 0.986) | 0.004** |
| Diabetic retinopathy | − 0.005 (− 0.011, 0.00) | 0.00 (− 0.001, 0.00) | 0.133 (0.053, 0.475) | 0.004** |
| Diabetic nephropathy | − 0.006 (− 0.012, 0.00) | − 0.003 (− 0.004, 0.00) | 0.323 (0.200, 0.640) | < 0.001*** |
| Diabetic neuropathy | 0.00 (− 0.002, 0.00) | 0.00 (− 0.001, 0.00) | − 1.085 (− 6.01, 5.69) | 0.744 |
Model 1: Univariate model; model 2: chronological age, sex, ethnicity, Townsend deprivation index, education, family history of CVD or hypertension, sleep duration, HbA1c, use of aspirin, use of antihypertensive medication, lipid-lowing medication or diabetes medication, and BMI. Abbreviation: CVD, cardiovascular disease; PhenoAgeAccel, phenotypic age acceleration; T2DM, type 2 diabetes mellitus
Fig. 3.
Mediated effect of PhenoAgeAccel in the association between healthy diet and diabetic microvascular complications. Mediation analysis models were used to estimate mediated effect of PhenoAgeAccel in the association between healthy dietary scores and composite microvascular complications, diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy with unadjusted. Bars show risk difference on the probability scale (Mediated effect/Direct effect). Red represents mediated effect, blue represents directed effect. ***P-value < 0.001 and **P-value < 0.01. Two-sided P-values and 95% CIs were obtained from the empirical bootstrap distribution
Stratified analysis and sensitivity analysis
To examine whether the association between dietary patterns and DMCs varies across subgroups, stratified analyses were performed. No significant interactions were observed between healthy dietary score and any stratified factors after accounting for multiple comparisons (Fig. S3), indicating that the relationship between dietary patterns and DMCs was consistent across these subgroups.
Sensitivity analyses further confirmed the robustness of the main findings. Results remained consistent after excluding events occurring within the first 2 years, including participants with baseline cardiovascular disease, restricting analyses to individuals with complete covariate data, adjusting for baseline eGFR, or applying Fine-Gray subdistribution hazard models to account for competing risks (Tables S4–S7).
Discussion
In this prospective cohort study of 13,294 participants with T2DM from the UK Biobank, we found that adherence to a healthy dietary pattern and slower biological aging were associated with a reduced risk of microvascular complications that impact vision and quality of life. Notably, up to 43.0% of the protective effect of diet was mediated through slower biological aging. These results highlight the intertwined roles of dietary patterns and aging biology in shaping microvascular health in diabetes and support the potential of dietary interventions to mitigate microvascular risk.
Diabetic retinopathy, nephropathy, and neuropathy remain leading causes of visual loss, renal failure, and disability worldwide despite advances in glycemic management. Increasing evidence has suggested diet is a powerful lifestyle lever for chronic disease risk [26, 27]. Consistent with growing global dietary recommendations that advocate for whole-diet patterns [28, 29], our results showed that a higher intake of fruits, vegetables, whole grains, fish, dairy, and vegetable oils, together with a lower intake of red/processed meat and sugar-sweetened beverages, was associated with a reduced risk of microvascular complications in T2DM. These findings were supported by large prospective samples, objective endpoints, and comprehensive adjustment for confounders. However, the weaker association observed for DPN suggests that its pathogenesis may be less sensitive to dietary modification or may require a longer latency period for effects to emerge.
Our study provides novel insights into biological mechanism linking diet to microvascular complications by highlighting the mediating role of biological aging. PhenoAgeAccel, which integrates chronological age with multiple inflammatory and metabolic markers, has been shown to capture aging across diverse biological systems [30, 31]. In our cohort, participants with lower dietary scores exhibited accelerated biological aging, suggesting that poor diet may contribute to microvascular damage via accelerated aging pathways. Notably, for DR and DN—where the association with dietary score was strongest—approximately one-third of the dietary benefit was mediated through PhenoAgeAccel. These prospective findings reinforce the relevance of biological aging as a clinically meaningful risk factor in diabetes.
Previous studies have shown that poor dietary habits promote systemic inflammation, oxidative stress, and metabolic dysregulation, all of which accelerate aging and increase susceptibility to adverse outcomes [32, 33]. The biomarkers composing phenotypic age (e.g., CRP, glucose, albumin) reflect systemic inflammation, metabolic stress, and hematopoietic dysfunction, which are the core processes of microvascular dysfunction. Pro-inflammatory markers (CRP, WBC, lower lymphocyte%) contribute to endothelial activation, leukostasis, and barrier disruption [34, 35]. Hyperglycemia activates polyol/PKC and AGE-RAGE pathways, inducing oxidative stress that thickens the basement membrane and causes neuro-ischemic injury [36–38]. Lower albumin reflects chronic inflammation/nutritional compromise and associates with capillary leak [39, 40]. Higher alkaline phosphatase reflects disordered mineral metabolism and vascular calcification, which may stiffen the microvasculature [41, 42]. Hematologic indices (higher RDW, abnormal MCV) imply impaired erythrocyte deformability and tissue hypoxia, a driver of VEGF and inflammatory pathways [43, 44]. Higher creatinine indicates greater pre-existing microvascular burden and uremic endothelial toxicity [45, 46]. These mechanisms correspond to the observed risks for DR, DKD, and DPN.
Thus, the mediating role of PhenoAgeAccel for the strong association between healthy dietary patterns and lower risk of DMCs is biologically plausible. However, it is necessary to emphasize that the causality cannot be inferred from these associations identified in observational data, although mediation analysis provides support to plausible biological pathways. PhenoAgeAccel is currently well-validated and widely used aging measures in terms of their mimicking of the aging process, whereas the complexity of aging and the unclear definition of normal aging make it difficult to explicitly develop a perfect aging measure and relate these aging measures to mechanisms of “normal aging” [47, 48]. Regardless, our findings highlight the need for future study to elucidate molecular pathways linking diet and biological aging, and further studies are warranted to replicate and clarify these associations.
Our findings indicate that higher diet quality and slower biological aging are associated with lower risk of microvascular complications in individuals with T2DM, consistent with growing evidence that healthy dietary patterns promote healthy aging and slow disease accumulation. In two large prospective cohorts with up to 30 years of follow-up, long-term adherence to Alternative Healthy Eating Index (AHEI), Dietary Approaches to Stop Hypertension (DASH), Alternative Mediterranean Index (aMED), Mediterranean-DASH (MIND), and related patterns (emphasizing higher intakes of fruits, vegetables, whole grains, unsaturated fats, nuts, legumes, and low-fat dairy products, with lower intake of trans fats, sodium, sugary beverages, and red or processed meats (or both)) was associated with greater odds of healthy aging across cognitive, physical, mental, and disease-free domains (with the AHEI having the strongest effect) [49]. A 15-year cohort of 2473 community-dwelling older adults showed that higher adherence to MIND, AHEI, or aMED diets was linked to a slower annual accumulation of multimorbidity, whereas a pro-inflammatory dietary pattern was associated with faster accumulation [50]. Together, these studies suggest that an “anti-aging” dietary pattern, similar to our high-score diet for T2DM (high in fruits, vegetables, whole grains, fish, dairy, and vegetable oils, and low in refined grains, processed and unprocessed red meats, and sugary drinks), may reduce DMC risk. Such a diet may exert protective effects by reducing inflammation, oxidative stress, glycation/AGE-RAGE signaling, and endothelial dysfunction—mechanisms partly captured by PhenoAgeAccel. Accordingly, measures of biological aging could serve as modifiable intermediate targets in future dietary intervention trials for DMC prevention in T2DM.
However, several limitations should be acknowledged. First, dietary patterns were assessed only at baseline, which may not reflect long-term changes in dietary habits over the follow-up period. Second, PhenoAgeAccel, derived from multi-system blood biomarkers, primarily captures systemic biological aging and may not fully represent organ- or tissue-specific aging, particularly those related to microvascular function. Third, although extensive covariate adjustments were performed, the possibility of residual confounding cannot be completely excluded. Finally, as the UK Biobank cohort predominantly comprises White participants, the generalizability of our findings to more diverse populations may be limited. Further studies in ethnically and geographically varied cohorts are warranted to confirm these associations and elucidate the underlying molecular mechanisms.
In conclusion, adherence to a healthy dietary pattern and slower biological aging were independently and jointly associated with lower risks of diabetic microvascular complications, particularly diabetic retinopathy and nephropathy. Moreover, biological aging partially mediated the protective effect of a healthy diet. These findings underscore the potential of integrated prevention strategies that simultaneously target modifiable dietary behaviors and biological aging processes to alleviate microvascular complication burden in diabetes.
Supplementary Information
Below is the link to the electronic supplementary material.
(PDF 687 KB)
Acknowledgements
We are grateful to all the participants of UK Biobank and all the people involved in building the UK Biobank study.
Author contribution
Study conception and design: H.Z, K.L, N.L, and B.Y; data curation: H.Z, S.B, Q.M, RK.Y, XY.Y, and J.L; formal analysis: H.Z, B.S, Q.M, and RK.Y; interpretation of data: Z.H, RK. Y, J.X, XY. Y, J.L, Q.L; writing: Z.H, S.B, Q.L, JY.Z, and Z.Y; reviewing: H.Z, Q.L, Z.Y, K.L, N.L, and B.Y. All authors read and approved the final version of the manuscript.
Funding
This study was supported by the Shanghai Municipal Health Commission Collaborative Innovation Cluster Project (2024CXJQ02 to BY) and the Fundamental Research Funds for the Central Universities (YG2025ZD04 to BY and YG2025QNA43 to HZ).
Data availability
This study has been conducted using the UK Biobank Resource under Application Number 107451. The code, models, algorithms, protocols, methods, and other useful materials related to the project are available upon reasonable request to the corresponding authors.
Declarations
Ethics approval and consent to participate
Not applicable. UK Biobank has full ethical approval from the NHS National Research Ethics Service (16/NW/0274). All participants gave written informed consent.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Han Zhou, Shuai Ben and Qian Ma contributed equally to this work.
Contributor Information
Na Li, Email: nx_linalab@163.com.
Kun Liu, Email: drliukun@sjtu.edu.cn.
Biao Yan, Email: yanbiao@sjtu.edu.cn.
References
- 1.Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan BB, et al. IDF diabetes atlas: global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022;183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Teo ZL, Tham YC, Yu M, Chee ML, Rim TH, Cheung N, et al. Global prevalence of diabetic retinopathy and projection of burden through 2045: systematic review and meta-analysis. Ophthalmology. 2021;128(11):1580–91. [DOI] [PubMed] [Google Scholar]
- 3.Hu Q, Jiang L, Yan Q, Zeng J, Ma X, Zhao Y. A natural products solution to diabetic nephropathy therapy. Pharmacol Ther. 2023;241. [DOI] [PubMed] [Google Scholar]
- 4.Feldman EL, Callaghan BC, Pop-Busui R, Zochodne DW, Wright DE, Bennett DL, et al. Diabetic neuropathy. Nat Rev Dis Primers. 2019;5(1). [DOI] [PubMed] [Google Scholar]
- 5.Lyssenko V, Vaag A. Genetics of diabetes-associated microvascular complications. Diabetologia. 2023;66(9):1601–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bellary S, Kyrou I, Brown JE, Bailey CJ. Type 2 diabetes mellitus in older adults: clinical considerations and management. Nat Rev Endocrinol. 2021;17(9):534–48. [DOI] [PubMed] [Google Scholar]
- 7.Geng T, Zhu K, Lu Q, Wan Z, Chen X, Liu L, et al. Healthy lifestyle behaviors, mediating biomarkers, and risk of microvascular complications among individuals with type 2 diabetes: a cohort study. PLoS Med. 2023;20(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mozaffarian D. Dietary and policy priorities for cardiovascular disease, diabetes, and obesity: a comprehensive review. Circulation. 2016;133(2):187–225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.He P, Li H, Ye Z, Liu M, Zhou C, Wu Q, et al. Association of a healthy lifestyle, life’s essential 8 scores with incident macrovascular and microvascular disease among individuals with type 2 diabetes. J Am Heart Assoc. 2023;12(17). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Bloom SI, Islam MT, Lesniewski LA, Donato AJ. Mechanisms and consequences of endothelial cell senescence. Nat Rev Cardiol. 2023;20(1):38–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Augustin HG, Koh GY. A systems view of the vascular endothelium in health and disease. Cell. 2024;187(18):4833–58. [DOI] [PubMed] [Google Scholar]
- 12.Seals DR, Jablonski KL, Donato AJ. Aging and vascular endothelial function in humans. Clin Sci Lond. 2011;120(9):357–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rossman MJ, Kaplon RE, Hill SD, McNamara MN, Santos-Parker JR, Pierce GL, et al. Endothelial cell senescence with aging in healthy humans: prevention by habitual exercise and relation to vascular endothelial function. Am J Physiol Heart Circ Physiol. 2017;313(5):H890–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Elliott ML, Caspi A, Houts RM, Ambler A, Broadbent JM, Hancox RJ, et al. Disparities in the pace of biological aging among midlife adults of the same chronological age have implications for future frailty risk and policy. Nat Aging. 2021;1(3):295–308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Pugashetti JV, Kim JS, Bose S, Adegunsoye A, Linderholm AL, Chen CH, et al. Biological age, chronological age, and survival in pulmonary fibrosis: a causal mediation analysis. Am J Respir Crit Care Med. 2024;210(5):639–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Liu Z, Kuo PL, Horvath S, Crimmins E, Ferrucci L, Levine M. A new aging measure captures morbidity and mortality risk across diverse subpopulations from NHANES IV: a cohort study. PLoS Med. 2018;15(12). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Wang T, Duan W, Jia X, Huang X, Liu Y, Meng F, et al. Associations of combined phenotypic ageing and genetic risk with incidence of chronic respiratory diseases in the UK Biobank: a prospective cohort study. Eur Respir J. 2024;63(2). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Li X, Cao X, Zhang J, Fu J, Mohedaner M, Danzengzhuoga, et al. Accelerated aging mediates the associations of unhealthy lifestyles with cardiovascular disease, cancer, and mortality. J Am Geriatr Soc. 2024;72(1):181–93. [DOI] [PMC free article] [PubMed]
- 19.Duan S, Wu Y, Zhu J, Wang X, Fang Y. Associations of polycyclic aromatic hydrocarbons mixtures with cardiovascular diseases mortality and all-cause mortality and the mediation role of phenotypic ageing: a time-to-event analysis. Environ Int. 2024;186. [DOI] [PubMed] [Google Scholar]
- 20.Chen L, Yin X, Zhao Y, Chen H, Tan T, Yao P, et al. Biological ageing and the risks of all-cause and cause-specific mortality among people with diabetes: a prospective cohort study. J Epidemiol Community Health. 2022;76(9):771–8. [DOI] [PubMed] [Google Scholar]
- 21.Yang G, Cao X, Li X, Zhang J, Ma C, Zhang N, et al. Association of unhealthy lifestyle and childhood adversity with acceleration of aging among UK Biobank participants. JAMA Netw Open. 2022;5(9). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Dominguez LJ, Veronese N, Barbagallo M. Dietary patterns and healthy or unhealthy aging. Gerontology. 2024;70(1):15–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Caleyachetty R, Littlejohns T, Lacey B, Bešević J, Conroy M, Collins R, et al. United Kingdom Biobank (UK Biobank): JACC Focus Seminar 6/8. J Am Coll Cardiol. 2021;78(1):56–65. [DOI] [PubMed] [Google Scholar]
- 24.Said MA, Verweij N, van der Harst P. Associations of combined genetic and lifestyle risks with incident cardiovascular disease and diabetes in the UK Biobank study. JAMA Cardiol. 2018;3(8):693–702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nat Rev Endocrinol. 2018;14(2):88–98. [DOI] [PubMed] [Google Scholar]
- 26.Tada H, Takamura M, Kawashiri MA. The effect of diet on cardiovascular disease, heart disease, and blood vessels. Nutrients. 2022;14(2). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Santos-Pujol E, Noguera-Castells A, Casado-Pelaez M, García-Prieto CA, Vasallo C, Campillo-Marcos I, et al. The multiomics blueprint of the individual with the most extreme lifespan. Cell Rep Med. 2025;6(10). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Tingö L, Bergh C, Rode J, Rubio MFR, Persson J, Johnson LB, et al. The effect of whole-diet interventions on memory and cognitive function in healthy older adults-a systematic review. Adv Nutr. 2024;15(9). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Granic A, Sayer AA, Cooper R, Robinson SM. Nutrition in the prevention and treatment of skeletal muscle ageing and sarcopenia: a single nutrient, a whole food and a whole diet approach. Proc Nutr Soc. 2024:1–16. [DOI] [PMC free article] [PubMed]
- 30.Liu W, Wang J, Wang M, Hou H, Ding X, Ma L, et al. Oxidative stress factors mediate the association between life’s essential 8 and accelerated phenotypic aging: NHANES 2005–2018. J Gerontol A Biol Sci Med Sci. 2024;79(1). [DOI] [PubMed] [Google Scholar]
- 31.Hu Z, Xu J, Shen R, Lin L, Su Y, Xie C, et al. Combination of biological aging and genetic susceptibility helps identifying at-risk population of venous thromboembolism: a prospective cohort study of 394 041 participants. Am J Hematol. 2025;100(4):575–83. [DOI] [PubMed] [Google Scholar]
- 32.Aleksandrova K, Koelman L, Rodrigues CE. Dietary patterns and biomarkers of oxidative stress and inflammation: a systematic review of observational and intervention studies. Redox Biol. 2021;42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Frazier K, Kambal A, Zale EA, Pierre JF, Hubert N, Miyoshi S, et al. High-fat diet disrupts REG3γ and gut microbial rhythms promoting metabolic dysfunction. Cell Host Microbe. 2022;30(6):809-23.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Sinha SK, Nicholas SB, Sung JH, Correa A, Rajavashisth TB, Norris KC, et al. Hs-CRP is associated with incident diabetic nephropathy: findings from the jackson heart study. Diabetes Care. 2019;42(11):2083–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Vestweber D. How leukocytes cross the vascular endothelium. Nat Rev Immunol. 2015;15(11):692–704. [DOI] [PubMed] [Google Scholar]
- 36.González P, Lozano P, Ros G, Solano F. Hyperglycemia and oxidative stress: an integral, updated and critical overview of their metabolic interconnections. Int J Mol Sci. 2023;24(11). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Wautier JL, Wautier MP, Schmidt AM, Anderson GM, Hori O, Zoukourian C, et al. Advanced glycation end products (AGEs) on the surface of diabetic erythrocytes bind to the vessel wall via a specific receptor inducing oxidant stress in the vasculature: a link between surface-associated AGEs and diabetic complications. Proc Natl Acad Sci USA. 1994;91(16):7742–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Kang Q, Yang C. Oxidative stress and diabetic retinopathy: molecular mechanisms, pathogenetic role and therapeutic implications. Redox Biol. 2020;37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Cai YW, Zhang HF, Gao JW, Cai ZX, Cai JW, Gao QY, et al. Serum albumin and risk of incident diabetes and diabetic microvascular complications in the UK Biobank cohort. Diabetes Metab. 2023;49(5). [DOI] [PubMed] [Google Scholar]
- 40.Soeters PB, Wolfe RR, Shenkin A. Hypoalbuminemia: pathogenesis and clinical significance. JPEN J Parenter Enteral Nutr. 2019;43(2):181–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Kim DW, Hwang SY, Nam YJ, Kim D, Shin SJ, et al. The combined prognostic significance of alkaline phosphatase and vascular calcification in patients with end-stage kidney disease. Nutr Metab Cardiovasc Dis. 2020;30(9):1476–83. [DOI] [PubMed] [Google Scholar]
- 42.Romanul FC, Bannister RG. Localized areas of high alkaline phosphatase activity in endothelium of arteries. Nature. 1962;195:611–2. [DOI] [PubMed] [Google Scholar]
- 43.Hausladen A, Qian Z, Zhang R, Premont RT, Stamler JS. Optimized S-nitrosohemoglobin synthesis in red blood cells to preserve hypoxic vasodilation via βCys93. J Pharmacol Exp Ther. 2022;382(1):1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Ebrahimi S, Bagchi P. A computational study of red blood cell deformability effect on hemodynamic alteration in capillary vessel networks. Sci Rep. 2022;12(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Cunha RSD, Santos AF, Barreto FC, Stinghen AEM. How do uremic toxins affect the endothelium? Toxins (Basel). 2020;12(6). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.He D, Gao B, Wang J, Yang C, Zhao MH, Zhang L. The difference between cystatin c- and creatinine-based estimated glomerular filtration rate and risk of diabetic microvascular complications among adults with diabetes: a population-pased cohort study. Diabetes Care. 2024;47(5):873–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Rutledge J, Oh H, Wyss-Coray T. Measuring biological age using omics data. Nat Rev Genet. 2022;23(12):715–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Cohen AA, Kennedy BK, Anglas U, Bronikowski AM, Deelen J, Dufour F, et al. Lack of consensus on an aging biology paradigm? A global survey reveals an agreement to disagree, and the need for an interdisciplinary framework. Mech Ageing Dev. 2020;191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Tessier AJ, Wang F, Korat AA, Eliassen AH, Chavarro J, Grodstein F, et al. Optimal dietary patterns for healthy aging. Nat Med. 2025;31(5):1644–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Abbad-Gomez D, Carballo-Casla A, Beridze G, Lopez-Garcia E, Rodríguez-Artalejo F, Sala M, et al. Dietary patterns and accelerated multimorbidity in older adults. Nat Aging. 2025;5(8):1481–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
(PDF 687 KB)
Data Availability Statement
This study has been conducted using the UK Biobank Resource under Application Number 107451. The code, models, algorithms, protocols, methods, and other useful materials related to the project are available upon reasonable request to the corresponding authors.