Abstract
Background
Despite achieving low-density lipoprotein cholesterol (LDL-C) targets, patients with type 2 diabetes mellitus (T2DM) remain at significant risk for microvascular complications. We aimed to investigate whether the atherogenic index of plasma (AIP), a marker of atherogenic dyslipidaemia, is associated with these complications in a real-world cohort of patients with controlled LDL-C.
Methods
We conducted a retrospective cohort study using electronic health records from community health centres in Shenzhen, China (2018–2024). We included 16,469 participants with T2DM and LDL-C<2.6 mmol/L at baseline, categorising them into tertiles (T1–3) based on baseline AIP. The primary outcome was a composite of incident diabetic retinopathy, diabetic nephropathy, and diabetic peripheral neuropathy, identified via ICD-10 codes. We used multivariable Cox proportional hazards models to assess associations.
Results
During a median follow-up of 4.6 years, 1,056 participants developed microvascular complications. We observed a graded association between AIP tertiles and the risk of microvascular complications. Participants in T3 had a significantly higher risk of total microvascular complications than those in T1 (hazard ratio (HR) = 1.34; 95% confidence interval (CI) = 1.15–1.56), while T2 showed no statistically significant increase in risk (HR = 1.10; 95% CI = 0.94–1.28). Specifically, compared to T1, T3 participants had significantly higher risks of diabetic nephropathy (HR = 1.62; 95% CI = 1.30–2.01) and diabetic peripheral neuropathy (HR = 1.39; 95% CI = 1.13–1.70), while no statistically significant association was found for diabetic retinopathy risk (HR = 1.23; 95% CI = 0.93–1.63). In contrast, compared with T1 participants, those in T2 had higher risks of nephropathy (HR = 1.28; 95% CI = 1.02–1.59), but no statistically significant associations were found for diabetic peripheral neuropathy or diabetic retinopathy.
Conclusions
We found that AIP is independently associated with increased risk of microvascular complications, particularly diabetic nephropathy and diabetic peripheral neuropathy, even in T2DM patients achieving LDL-C targets. These results suggest that AIP may serve as a simple, lipid-based marker to help identify patients with residual microvascular risk beyond conventional LDL-C control.
Keywords : type 2 diabetes, low-density lipoprotein cholesterol, atherogenic index of plasma, microvascular complications
The global burden of type 2 diabetes mellitus (T2DM) continues to rise rapidly, with 589 million adults affected worldwide and 148 million in China alone [1]. Among patients with T2DM, microvascular complications such as diabetic retinopathy (DR), diabetic nephropathy (DN), and diabetic peripheral neuropathy (DPN) represent major causes of morbidity and reduced quality of life [2].
Low-density lipoprotein cholesterol (LDL-C) remains the primary lipid target in T2DM management, with guidelines recommending levels <2.6 mmol/L [3,4]. However, notable residual risk of microvascular complications may still exist even after LDL-C targets are achieved [5–8]. This suggests that LDL-C alone may not fully capture microvascular risk and highlights the need for additional lipid-related markers.
The atherogenic index of plasma (AIP), calculated as log(triglycerides (TG)/high-density lipoprotein cholesterol (HDL-C)), reflects both the quantity and quality of atherogenic lipoproteins and is closely linked to insulin resistance and inflammation [9–11]. Previous studies have shown that elevated AIP is associated with increased risk of cardiovascular outcomes in the general T2DM population, with some evidence also suggesting a link to microvascular complications [10–14]. However, evidence regarding the association of AIP with microvascular complications specifically among patients who have already achieved LDL-C targets remains limited. Evaluating AIP in this population is of public health relevance, as it may help identify individuals with residual microvascular risk beyond conventional LDL-C control in real-world settings.
Therefore, we aimed to investigate the association between AIP and the risk of incident microvascular complications among patients with T2DM who achieved LDL-C targets (<2.6 mmol/L), using a large real-world community-based cohort in Shenzhen, China.
METHODS
Study design and participants
China's National Basic Public Health Service Standards mandate the creation and maintenance of electronic health records for individuals with T2DM at the community level, along with annual comprehensive health examinations. We conducted a retrospective cohort study using de-identified electronic health records from community health service centres in Shenzhen, China (2018–2024).
Participants were patients with a physician-diagnosed T2DM, as recorded in the community diabetes health management database according to the Chinese Diabetes Society guidelines and ICD-10 codes (E11). We included individuals with a confirmed diagnosis of T2DM, available baseline lipid measurements (i.e. LDL-C, TG, and HDL-C), and an LDL-C treatment target (<2.6 mmol/L) at baseline. We applied no strict age restrictions, as a broad age spectrum better reflects the real-world community diabetes population. We calculated follow-up time from the baseline examination date to the date of hospitalisation for microvascular complications or the study end date. Because outcome ascertainment relied on inpatient records, we imposed no minimum follow-up period and included all eligible participants regardless of follow-up duration. We excluded participants with pre-existing microvascular complications at baseline [12], LDL-C≥2.6 mmol/L at baseline [4], or missing baseline HDL-C or TG data.
The baseline database initially included 66,926 patients with T2DM in Shenzhen from 2018 to 2024 (Figure 1). After applying the eligibility criteria, we included 16,469 participants in the final analysis.
Figure 1.
Study population selection. LDL-C – low-density lipoprotein cholesterol, HDL-C – high-density lipoprotein cholesterol, TG – triglycerides.
Assessment of AIP
The AIP is a logarithmically transformed ratio of TG to HDL-C in molar concentration (mmol/L), and it is mathematically derived from log(TG/HDL-C) [9]. Nurses collected venous blood samples for laboratory testing, including measurements of TG, HDL-C, LDL-C, and other relevant parameters. We then classified the study population into three groups by AIP tertile: T1 (≤−0.43), T2 (−0.43–0.24), and T3 (>0.24).
Ascertainment of outcome
Microvascular complications served as the primary outcome for the cohort. The secondary outcomes included DR, DN, and DPN [15]. We extracted complication data from inpatient medical records. We identified cases using ICD-10 diagnostic codes for DR (E11.3, E14.3, H28.0 and H36.0), DN (E11.2, E14.2, N18.0, N18.1, N18.2, N18.3, N18.4, N18.5, N18.8 and N18.9), and DPN (E11.4, E14.4, G59.0, G62.9, G63.2 and G99.0).
Covariates
For this study, we derived covariate data from three sources: the diabetes baseline database, the annual health examination database, and electronic prescription records. Trained physicians or nurses collected baseline and annual health examination data through laboratory tests and structured interviews, while medication information was ascertained by matching drug names in electronic prescription records. We defined covariates as a set of potential confounders encompassing demographic, physical examination, lifestyle, and medical history factors. These included sex (male, female), age (year), educational level (primary school or below, middle school, high school or above) [16], body mass index (BMI), smoking status (never, former, current), drinking status (none, mild, moderate, heavy) [17], regular exercise (yes, no), duration of diabetes (>5 and ≤5 years), diagnosis of hypertension (yes, no), family history of diabetes (yes, no), use of oral antidiabetic drugs (yes, no), use of insulin (yes, no), and use of statins (yes, no). We calculated BMI as weight (kg) divided by height squared (m2) [18]. Heavy drinkers were defined as those who consumed alcohol every day. In contrast, self-reported occasional drinkers and frequent drinkers were defined as mild and moderate drinkers, respectively [17]. Trained physicians or nurses collected smoking status, drinking status, and regular exercise through face-to-face structured interviews during health examinations, in accordance with the standardised protocols of China's National Basic Public Health Service Standards. We recorded all interview data directly into the electronic health record system. We ascertained medication use from electronic prescription records in the community health database. For each participant, we extracted all medication records during the observation period and classified them into three categories based on drug name matching: oral antidiabetic drugs (including but not limited to metformin, sulfonylureas, DPP-4 inhibitors, SGLT-2 inhibitors, and thiazolidinediones), insulin preparations, and statins (including atorvastatin, rosuvastatin, simvastatin, and other lipid-lowering agents of the statin class). We considered a participant a user of a given medication category if the corresponding drug was prescribed at least once during the observation period. We included these medication variables as binary covariates (yes/no) in the multivariable-adjusted models. Hypertension was defined as a self-reported diagnosis, or an average systolic blood pressure ≥140 mm Hg, or an average diastolic blood pressure ≥90 mm Hg [19], or the administration of antihypertensive medication. Family history of diabetes was defined as at least one parent, sibling, or child with diabetes [20].
Statistical analysis
Baseline characteristics of participants were described according to AIP tertiles [21,22]. We presented normally distributed continuous variables as mean (standard deviation) and compared them between groups using one-way analysis of variance. We presented non-normally distributed continuous variables as median (interquartile range) and compared them using the Kruskal-Wallis H test. Lastly, we presented categorical variables as frequencies (percentages) and compared them using the χ2 test [23].
Missing data were examined for all covariates (Table S1 in the Online Supplementary Document). In the primary analysis, we performed multiple imputation by chained equations using the ‘mice’ package in R. We generated five imputed datasets with ten iterations each, using a fixed random seed to ensure reproducibility. Pooled estimates were obtained using Rubin’s rules.
To evaluate the association between AIP levels and the incidence of microvascular complications, we constructed multivariable Cox proportional hazards regression models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for each group. The proportional hazards assumption was verified using Schoenfeld’s residual test, and no violation was detected. We defined three models: Model 1 (unadjusted), Model 2 (adjusted for age, sex, and education level), and Model 3 (further adjusted for smoking status, drinking status, regular exercise, BMI, duration of diabetes, family history of diabetes, hypertension, use of oral antidiabetic drugs, use of insulin, and use of statins). To assess the dose-response association, AIP was entered into the Cox models as a continuous variable. We presented the results as HRs per one-unit increase in AIP using the three adjustment models.
We then performed subgroup and interaction analyses. We analysed the association between AIP and microvascular complications within predefined subgroups, including age, sex, BMI, and duration of diabetes. We also introduced interaction terms to test for effect modification.
To verify the robustness of the findings, we conducted sensitivity analyses. First, to address potential bias from multiple imputation, we excluded individuals with missing covariate values and conducted a complete-case analysis. Second, we excluded participants using oral antidiabetic drugs, insulin, or statins; Third, we additionally adjusted for total cholesterol and LDL-C levels in Model 3. Finally, we excluded participants developing microvascular complications within one year of follow-up to minimise potential reverse causality.
We performed all statistical analyses using R, version 4.4.3 (R Core Team, Vienna, Austria). All P-values were two-tailed, and a P-value <0.05 was considered statistically significant.
RESULTS
Baseline characteristics
We included 16,469 participants. The median age was 57.0 years (interquartile range = 48.0–66.0), and 9067 (55.1%) were male. Missing data for key variables were generally low, including diabetes duration (2.59%), drinking status (0.56%), smoking status (0.37%), regular exercise (0.15%), BMI (0.12%), and lipid profiles (<0.3%). No missing data were present for the remaining covariates. Based on AIP tertiles, participants were categorised into three groups: T1 (n = 5,490), T2 (n = 5,491), and T3 (n = 5,488). Overall, individuals with higher AIP levels were more likely to be male, possess higher educational attainment, exhibit higher BMI, engage in heavy alcohol consumption and current smoking, while demonstrating lower rates of regular exercise (Table 1). Furthermore, higher AIP levels were associated with significantly higher rates of hypertension, family history of diabetes, use of oral antidiabetic medications, and statin therapy.
Table 1.
Characteristics of the study participants by AIP tertiles, n (%)
| Overall (n = 16,469) | T1 (n = 5,490) | T2 (n = 5,491) | T3 (n = 5,488) | P-value | |
|---|---|---|---|---|---|
|
Age in years*
|
57.0 (48.0, 66.0) |
58.0 (49.0, 67.0) |
58.0 (49.0, 67.0) |
55.0 (46.0, 65.0) |
<0.001 |
|
Sex
|
|
|
|
|
<0.001 |
| Male |
9,067 (55.1) |
2,874 (52.3) |
2,912 (53.0) |
3,281 (59.8) |
|
| Female |
7,402 (44.9) |
2,616 (47.7) |
2,579 (47.0) |
2,207 (40.2) |
|
|
Education
|
|
|
|
|
<0.001 |
| Primary school or below |
5,549 (33.7) |
1,884 (34.3) |
1,937 (35.3) |
1,728 (31.5) |
|
| Middle school |
5,747 (34.9) |
1,928 (35.1) |
1,830 (33.3) |
1,989 (36.2) |
|
| High school or above |
5,173 (31.4) |
1,678 (30.6) |
1,724 (31.4) |
1,771 (32.3) |
|
|
BMI (kg/m2)*
|
24.3 (22.4, 26.6) |
23.4 (21.5, 25.5) |
24.4 (22.5, 26.7) |
25.2 (23.3, 27.3) |
<0.001 |
|
LDL-C (mmol/L)*
|
2.1 (1.7, 2.4) |
2.0 (1.6, 2.3) |
2.1 (1.7, 2.4) |
2.1 (1.7, 2.4) |
<0.001 |
|
HDL-C (mmol/L)*
|
1.2 (1.0, 1.4) |
1.4 (1.2, 1.7) |
1.2 (1.0, 1.3) |
1.0 (0.9, 1.2) |
<0.001 |
|
TC (mmol/L)*
|
4.2 (3.6, 4.7) |
4.1 (3.6, 4.6) |
4.0 (3.5, 4.6) |
4.4 (3.8, 5.0) |
<0.001 |
|
TG (mmol/L)*
|
1.5 (1.0, 2.4) |
0.9 (0.7, 1.1) |
1.4 (1.2, 1.7) |
3.0 (2.2, 4.4) |
<0.001 |
|
Duration of diabetes in years
|
|
|
|
|
<0.001 |
| ≤5 |
11,148 (67.7) |
3,595 (65.5) |
3,667 (66.8) |
3,886 (70.8) |
|
| >5 |
5,321 (32.3) |
1,895 (34.5) |
1,824 (33.2) |
1,602 (29.2) |
|
|
Hypertension
|
|
|
|
|
<0.001 |
| No |
6,668 (40.5) |
2,528 (46.0) |
2,097 (38.2) |
2,043 (37.2) |
|
| Yes |
9,801 (59.5) |
2,962 (54.0) |
3,394 (61.8) |
3,445 (62.8) |
|
|
Family history of diabetes
|
|
|
|
|
<0.001 |
| No |
13,534 (82.2) |
4,631 (84.4) |
4,508 (82.1) |
4,395 (80.1) |
|
| Yes |
2,935 (17.8) |
859 (15.6) |
983 (17.9) |
1,093 (19.9) |
|
|
Drinking status
|
|
|
|
|
<0.001 |
| None |
13,866 (84.2) |
4,736 (86.3) |
4,671 (85.1) |
4,459 (81.2) |
|
| Mild |
1,517 (9.2) |
445 (8.1) |
471 (8.6) |
601 (11.0) |
|
| Moderate |
453 (2.8) |
115 (2.1) |
156 (2.8) |
182 (3.3) |
|
| Heavy |
633 (3.8) |
194 (3.5) |
193 (3.5) |
246 (4.5) |
|
|
Smoking status
|
|
|
|
|
<0.001 |
| Never |
12,854 (78.0) |
4,492 (81.8) |
4,345 (79.1) |
4,017 (73.2) |
|
| Ever |
1167 (7.1) |
390 (7.1) |
416 (7.6) |
361 (6.6) |
|
| Current |
2,448 (14.9) |
608 (11.1) |
730 (13.3) |
1,110 (20.2) |
|
|
Weekly exercise
|
|
|
|
|
<0.001 |
| No |
5,089 (30.9) |
1,561 (28.4) |
1,622 (29.5) |
1,906 (34.7) |
|
| Yes |
11,380 (69.1) |
3,929 (71.6) |
3,869 (70.5) |
3,582 (65.3) |
|
|
Use of oral antidiabetic drugs
|
|
|
|
|
<0.001 |
| No |
1,999 (12.1) |
800 (14.6) |
625 (11.4) |
574 (10.5) |
|
| Yes |
14,470 (87.9) |
4,690 (85.4) |
4,866 (88.6) |
4,914 (89.5) |
|
|
Use of insulin
|
|
|
|
|
<0.001 |
| No |
14,129 (85.8) |
4,602 (83.8) |
4,736 (86.3) |
4,791 (87.3) |
|
| Yes |
2,340 (14.2) |
888 (16.2) |
755 (13.7) |
697 (12.7) |
|
|
Use of statin
|
|
|
|
|
<0.001 |
| No |
11,613 (70.5) |
4,057 (73.9) |
3,793 (69.1) |
3,763 (68.6) |
|
| Yes | 4,856 (29.5) | 1,433 (26.1) | 1,698 (30.9) | 1,725 (31.4) |
AIP – atherogenic index of plasma, BMI – body mass index, HDL-C – high-density lipoprotein cholesterol, LDL-C – low-density lipoprotein cholesterol, T – tertile, TC – total cholesterol, TG – triglycerides
*Presented as median (interquartile range) unless specified otherwise.
Association between AIP and the risk of diabetic microvascular complications
During a median follow-up of 4.6 years, 1,056 participants developed microvascular complications, including 595 (3.6%) DPN cases, 535 (3.2%) DN cases, and 330 (2.0%) DR cases (Figure 2, Panel A–D). The incidence density (per 1000 person-years) was 7.9 for DPN, 7.1 for DN, and 4.4 for DR (Table 2). In addition, the higher AIP groups showed a significantly increased cumulative incidence of total microvascular complications over the follow-up period after the baseline survey (log-rank test P < 0.05).
Figure 2.
Kaplan-Meier curves for microvascular complications by AIP tertile. Panel A. Total microvascular complication. Panel B. Nephropathy. Panel C. Neuropathy. Panel D. Retinopathy. AIP – atherogenic index of plasma.
Table 2.
Incidence density of diabetic complications (per 1000 person-years)
| Cases/n | Person-years | DPN | DN | DR | |
|---|---|---|---|---|---|
|
Overall
|
1,056/16,469 |
75,554.4 |
595 (7.9) |
535 (7.1) |
330 (4.4) |
|
T1
|
432/5,490 |
25,344.6 |
179 (7.1) |
145 (5.7) |
108 (4.3) |
|
T2
|
481/5,491 |
25,249.7 |
188 (7.4) |
182 (7.2) |
111 (4.4) |
| T3 | 547/5,488 | 24,960.1 | 228 (9.1) | 208 (8.3) | 111 (4.4) |
DN – diabetic nephropathy, DPN – diabetic peripheral neuropathy, DR – diabetic retinopathy, T – tertile
We performed Cox proportional hazards regression analyses to evaluate the association between AIP levels and the risk of diabetic microvascular complications. The proportional hazards assumption was satisfied (P = 0.58) (Table S2 in the Online Supplementary Document). The results showed that participants in T3 had a significantly higher risk of total microvascular complications than those in T1 in Model 1 (HR = 1.24; 95% CI = 1.07–1.44), Model 2 (HR = 1.26; 95% CI = 1.08–1.46), and Model 3 (HR = 1.34; 95% CI = 1.15–1.56). On the contrary, T2 showed no statistically significant increase in risk across all models (P > 0.05) (Table 3). However, the P for trend was significant across all models (P for trend <0.05), supporting a dose-response relationship between increasing AIP levels and the hazard of total microvascular complications.
Table 3.
Association between AIP tertiles and incidence of diabetic microvascular complications*
| Model 1 |
Model 2 |
Model 3 |
||||
|---|---|---|---|---|---|---|
| HR (95%CI) | P-value | HR (95%CI) | P-value | HR (95%CI) | P-value | |
|
Total microvascular complications
|
|
|
|
|
|
|
| T1 |
ref |
|
ref |
|
ref |
|
| T2 |
1.07 (0.92–1.25) |
0.371 |
1.07 (0.92–1.25) |
0.378 |
1.10 (0.94–1.28) |
0.242 |
| T3 |
1.24 (1.07–1.44) |
0.004 |
1.26 (1.08–1.46) |
0.002 |
1.34 (1.15–1.56) |
<0.001 |
|
P-value for trend |
|
0.004 |
|
0.002 |
|
<0.001 |
|
DN
|
|
|
|
|
|
|
| T1 |
ref |
|
ref |
|
ref |
|
| T2 |
1.26 (1.01–1.57) |
0.038 |
1.27 (1.02–1.58) |
0.030 |
1.28 (1.02–1.59) |
0.031 |
| T3 |
1.46 (1.18–1.81) |
<0.001 |
1.55 (1.25–1.92) |
<0.001 |
1.62 (1.30–2.01) |
<0.001 |
|
P-value for trend |
|
<0.001 |
|
<0.001 |
|
<0.001 |
|
DPN
|
|
|
|
|
|
|
| T1 |
ref |
|
ref |
|
ref |
|
| T2 |
1.05 (0.86–1.29) |
0.622 |
1.05 (0.86–1.29) |
0.639 |
1.09 (0.89–1.34) |
0.418 |
| T3 |
1.30 (1.07–1.58) |
0.01 |
1.28 (1.06–1.56) |
0.013 |
1.39 (1.13–1.70) |
0.002 |
|
P-value for trend |
|
0.009 |
|
0.012 |
|
0.002 |
|
DR
|
|
|
|
|
|
|
| T1 |
ref |
|
ref |
|
ref |
|
| T2 |
1.03 (0.79–1.34) |
0.821 |
1.03 (0.79–1.34) |
0.835 |
1.15 (0.88–1.50) |
0.311 |
| T3 |
1.05 (0.80–1.36) |
0.742 |
0.99 (0.76–1.30) |
0.967 |
1.23 (0.93–1.63) |
0.142 |
| P-value for trend | 0.743 | 0.966 | 0.141 | |||
CI – confidence interval, DN – diabetic nephropathy, DPN – diabetic peripheral neuropathy, DR – diabetic retinopathy, HR – hazard ratio, ref – reference, T – tertile
*Model 1 was unadjusted; Model 2 was adjusted for age, sex, and education level; Model 3 was further adjusted for smoking status, drinking status, regular exercise, BMI, diabetes duration, family history of diabetes, hypertension, use of oral antidiabetic drugs, use of insulin, and use of statins.
Analyses for the secondary outcomes revealed a specific association between AIP levels and the risk of diabetic complications. The association was most pronounced for DN. Using T1 as the reference, T3 showed an increased DN risk in Model 1 (HR = 1.46; 95% CI = 1.18–1.81), Model 2 (HR = 1.55; 95% CI = 1.25–1.92), and Model 3 (HR = 1.62; 95% CI = 1.30–2.01), indicating that elevated exposure levels were consistently associated with increased DN risk. Trend tests further supported significant relationships across tertiles in all models (P for trend <0.001), suggesting that higher AIP is closely associated with elevated risk of microvascular complications. For DPN, significant associations were only evident in T3 across all models. In Model 3, T3 had an HR of 1.39 (95% CI = 1.13–1.70) compared to T1, with a significant trend (P for trend = 0.013). However, no statistically significant associations were detected between AIP levels and incident DR in any model, with the 95% CIs for all HRs including 1.0.
When AIP was modelled as a continuous variable, each one-unit increase was significantly associated with higher risk of DN (HR = 1.81; 95% CI = 1.42–2.31, P < 0.001), DPN (HR = 1.34; 95% CI = 1.07–1.69, P = 0.013), and the composite microvascular complications (HR = 1.37; 95% CI = 1.15–1.64, P < 0.001) after full adjustment (Table S3 in the Online Supplementary Document). For DR, the association was in the same direction but was not statistically significant (HR = 1.33; 95% CI = 0.97–1.81, P = 0.073).
Subgroup and sensitivity analysis
To explore the association between AIP and diabetic microvascular complications in more detail, we categorised the study population based on age, sex, BMI, and duration of diabetes. We then performed stratified analyses to test for interactions and adjust for confounders within subgroups (Table S4 in the Online Supplementary Document). Notably, we observed a significant interaction between age and AIP in diabetic DPN risk (Pinteraction = 0.009). Specifically, a high AIP index was more strongly associated with an elevated risk of DPN among people aged ≥65 years than those aged <65 years. However, we did not detect any interaction among other subgroups.
We conducted several sensitivity analyses to assess the robustness of the main findings (Table S5–8 in the Online Supplementary Document). The associations of AIP and microvascular complications remained consistent after excluding individuals with missing covariate values. This association also remained unchanged after removing glucose/lipid-lowering drug users at baseline. Additionally, adjusting for total cholesterol and LDL-C levels in Model 3 did not compromise the stability of the results. The results also persisted after excluding participants with follow-up durations of less than one year.
DISCUSSION
Analysing 16,469 patients with T2DM in Shenzhen who achieved target LDL-C levels (<2.6 mmol/L), we found that elevated AIP levels remain significantly and positively associated with the risk of diabetic microvascular complications, particularly DPN and DN. However, elevated AIP levels were not associated with DR.
Several studies have explored the role of AIP in diabetic complications. A Chinese cross-sectional study of 3,278 patients undergoing coronary angiography showed that AIP is positively correlated with the risk of coronary artery disease in patients with T2DM [12]. In addition, the China Health and Retirement Longitudinal Study, which included 8,727 participants aged ≥45 years, demonstrated that high AIP levels are associated with an increased risk of stroke in individuals with prediabetes and diabetes [13]. Another secondary analysis based on data from the Action to Control Cardiovascular Risk in Diabetes trial, involving 10,251 patients with long-standing T2DM, indicated that AIP was independently associated with the risk of major adverse cardiovascular events in patients with T2DM [14]. However, previous research has paid insufficient attention to diabetic microvascular complications [10]. Among the few studies that have examined this association, findings have been inconsistent. Li et al. reported a positive correlation between AIP and microvascular complications in a general T2DM population [10], while others have found weaker or non-significant associations, particularly for DR [24,25]. We extend the existing evidence by focusing specifically on patients who have already achieved LDL-C targets, a population in which residual risk remains poorly characterised, and the association with AIP has been less well studied.
We found that the association between AIP and microvascular complications is highly organ-specific. After adjusting for confounders, higher AIP was associated with significantly increased risks of both DPN and DN, but not with DR. The continuous variable further supported the association between AIP and microvascular risk, consistent with the primary tertile-based findings. Several potential mechanisms may explain the observed association between AIP and microvascular complications. As one of the most severe complications of T2DM, DPN significantly increases the risks of ulcers, non-traumatic amputations, and foot infections [26]. Abnormal lipid levels can increase blood viscosity and promote microthrombus formation [27], while an inadequate blood supply to the nervous system can disrupt energy metabolism and impair signal transmission [28], particularly in patients with high AIP levels. Consistent with previous research, the risk of diabetic kidney disease remains high even when LDL-C levels are well-controlled [29]. Levels of TG and/or HDL-C are considered residual risk factors for DN [30]. The renal microvasculature appears particularly sensitive to lipid abnormalities, as triglyceride-rich lipoproteins can promote glomerular injury through lipotoxicity, inflammation, and oxidative stress [31,32]. In contrast, the link between lipid profiles and DR appears less robust [24]. The blood-retinal barrier may offer some protection against damage from plasma lipoproteins, and intraretinal lipid transport may play a more critical role in DR development than plasma lipid levels [25]. However, some studies have found a significant association between AIP levels and DR in patients with T2DM [33,34]. These studies focused on the general T2DM population, whereas our investigation specifically targeted the subgroup of patients who had achieved LDL-C targets. This distinction may account for the discrepant findings regarding DR. Therefore, further multicentre clinical investigations are warranted to clarify the association between AIP and DR across different risk strata, particularly in populations with controlled conventional risk factors.
This study also has certain limitations. First, as an observational study, it cannot establish causality. Although we performed extensive multivariable adjustment and multiple sensitivity analyses, we cannot entirely rule out residual confounding. Important potential sources include glycaemic control (e.g. HbA1c), which had high missing rates and was not systematically recorded in the community health database. In addition, this real-world data set largely lacked longitudinal lipid profiles. Nevertheless, we adjusted for a wide range of demographic, lifestyle, clinical, and medication-related confounders, and the main findings remained robust in multiple sensitivity analyses. Future studies with more comprehensive data collection are warranted to further minimise residual confounding. Second, we identified microvascular complications exclusively from inpatient medical records using ICD-10 codes. Since DR, DN, and DPN are frequently diagnosed and managed in outpatient settings, some milder or early-stage cases may have been missed. This could have resulted in under-ascertainment of outcomes, potentially underestimating the incidence rates observed in this cohort. Thus, the significant associations we observed are likely to be conservative estimates of the true effect. Furthermore, the study population consists solely of Chinese individuals from the Shenzhen region, and the generalisability of the findings requires validation in other ethnic groups. Finally, AIP was calculated from single-time point measurements, which may not capture the potential impact of longitudinal changes on risk.
CONCLUSIONS
We found that AIP is independently associated with diabetic microvascular complications, particularly DN and DPN, even when LDL-C targets are met. Our results suggest that AIP may serve as a useful complementary marker for identifying residual risk in this population. Future studies are warranted to confirm its clinical utility and value in routine diabetes management.
Additional material
Acknowledgements
We sincerely thank all the medical staff who contributed to this project.
Ethics statement: The ethics committee of Huazhong University of Science and Technology approved this study (Number: 2022S004).
Footnotes
Funding: The Shenzhen Science and Technology Innovation Commission supported this study (Grant No. JCYJ20230807115703006).
Authorship contributions: SZ, YS, and YZ conceptualised the study, wrote the original draft, and were responsible for the methodology. YS analysed the data. LM, YW, HY, RC, ZC, and HR reviewed and edited the manuscript and were responsible for validation and visualisation. XY reviewed and edited the manuscript, provided the administrative support, and supervised the study. YS had full access to all study data and is responsible for data integrity and the accuracy of the data analysis.
Disclosure of interest: The authors completed the ICMJE Disclosure of Interest Form (available upon request from the corresponding author) and disclose no relevant interests
Data availability: The data sets and codes analysed in this study are available from the corresponding author upon reasonable request.
REFERENCES
- 1.Genitsaridi I, Salpea P, Salim A, Sajjadi SF, Tomic D, James S, et al. 11th edition of the IDF Diabetes Atlas: global, regional, and national diabetes prevalence estimates for 2024 and projections for 2050. Lancet Diabetes Endocrinol. 2026;14:149–156. 10.1016/S2213-8587(25)00299-2 [DOI] [PubMed] [Google Scholar]
- 2.Arnold SV, Khunti K, Tang F, Chen H, Cid-Ruzafa J, Cooper A, et al. Incidence rates and predictors of microvascular and macrovascular complications in patients with type 2 diabetes: results from the longitudinal Global DISCOVER study. Am Heart J. 2022;243:232–9. 10.1016/j.ahj.2021.10.181 [DOI] [PubMed] [Google Scholar]
- 3.Yang YS, Kim HL, Kim SH, Moon MK.Lipid management in Korean people with type 2 diabetes mellitus: Korean Diabetes Association and Korean Society of Lipid and Atherosclerosis consensus statement. J Lipid Atheroscler. 2023;12:12–22. 10.12997/jla.2023.12.1.12 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Chinese Diabetes Society Guideline for the prevention and treatment of type 2 diabetes mellitus in China (2020 edition) (Part 1). Zhongguo Shiyong Neike Zazhi. 2021;41:668–95. [Google Scholar]
- 5.Judge EP, Phelan D, O’Shea D.Beyond statin therapy: a review of the management of residual risk in diabetes mellitus. J R Soc Med. 2010;103:357–62. 10.1258/jrsm.2010.100033 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Shaya GE, Leucker TM, Jones SR, Martin SS, Toth PP.Coronary heart disease risk: low-density lipoprotein and beyond. Trends Cardiovasc Med. 2022;32:181–94. 10.1016/j.tcm.2021.04.002 [DOI] [PubMed] [Google Scholar]
- 7.Nichols GA, Philip S, Reynolds K, Granowitz CB, Fazio S.Increased residual cardiovascular risk in patients with diabetes and high versus normal triglycerides despite statin-controlled LDL cholesterol. Diabetes Obes Metab. 2019;21:366–71. 10.1111/dom.13537 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Luo Y, Peng D.Residual atherosclerotic cardiovascular disease risk: focus on non-high-density lipoprotein cholesterol. J Cardiovasc Pharmacol Ther. 2023;28:10742484231189597. 10.1177/10742484231189597 [DOI] [PubMed] [Google Scholar]
- 9.Qin Z, Zhou K, Li Y, Cheng W, Wang Z, Wang J, et al. The atherogenic index of plasma plays an important role in predicting the prognosis of type 2 diabetic subjects undergoing percutaneous coronary intervention: results from an observational cohort study in China. Cardiovasc Diabetol. 2020;19:23. 10.1186/s12933-020-0989-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Li Z, Huang Q, Sun L, Bao T, Dai Z.Atherogenic index in type 2 diabetes and its relationship with chronic microvascular complications. Int J Endocrinol. 2018;2018:1765835. 10.1155/2018/1765835 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Xu J, Zhou H, Xiang G.Association of atherogenic index of plasma with retinopathy and nephropathy in adult patients with type 2 diabetes mellitus aged >18 years. Can J Diabetes. 2022;46:708–14. 10.1016/j.jcjd.2022.04.008 [DOI] [PubMed] [Google Scholar]
- 12.Zhou K, Qin Z, Tian J, Cui K, Yan Y, Lyu S.The atherogenic index of plasma: a powerful and reliable predictor for coronary artery disease in patients with type 2 diabetes. Angiology. 2021;72:934–41. 10.1177/00033197211012129 [DOI] [PubMed] [Google Scholar]
- 13.Qu L, Fang S, Lan Z, Xu S, Jiang J, Pan Y, et al. Association between atherogenic index of plasma and new-onset stroke in individuals with different glucose metabolism status: insights from CHARLS. Cardiovasc Diabetol. 2024;23:215. 10.1186/s12933-024-02314-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fu L, Zhou Y, Sun J, Zhu Z, Xing Z, Zhou S, et al. Atherogenic index of plasma is associated with major adverse cardiovascular events in patients with type 2 diabetes mellitus. Cardiovasc Diabetol. 2021;20:201. 10.1186/s12933-021-01393-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.von Ferber L, Koster I, Hauner H.Medical costs of diabetic complications total costs and excess costs by age and type of treatment results of the German CoDiM Study. Exp Clin Endocrinol Diabetes. 2007;115:97–104. 10.1055/s-2007-949152 [DOI] [PubMed] [Google Scholar]
- 16.Wang T, Zhang M, Shi W, Li Y, Zhang T, Shi W.Atherogenic index of plasma, high sensitivity C-reactive protein and incident diabetes among middle-aged and elderly adults in China: a national cohort study. Cardiovasc Diabetol. 2025;24:103. 10.1186/s12933-025-02653-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kim SH, Cho YK, Kim YJ, Jung CH, Lee WJ, Park JY, et al. Association of the atherogenic index of plasma with cardiovascular risk beyond the traditional risk factors: a nationwide population-based cohort study. Cardiovasc Diabetol. 2022;21:81. 10.1186/s12933-022-01522-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.He W, Li Q, Yang M, Jiao J, Ma X, Zhou Y, et al. Lower BMI cutoffs to define overweight and obesity in China. Obesity (Silver Spring). 2015;23:684–91. 10.1002/oby.20995 [DOI] [PubMed] [Google Scholar]
- 19.Lenfant C, Chobanian AV, Jones DW, Roccella EJ.Seventh report of the Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7): resetting the hypertension sails. Hypertension. 2003;41:1178–9. 10.1161/01.HYP.0000075790.33892.AE [DOI] [PubMed] [Google Scholar]
- 20.Hu X, Yu W, Yang L, Pan W, Xu K, Chen X, et al. First-degree family history of diabetes is associated with nonalcoholic fatty liver disease independent of glucose metabolic status. J Diabetes Complications. 2022;36:108083. 10.1016/j.jdiacomp.2021.108083 [DOI] [PubMed] [Google Scholar]
- 21.He Y, Li Z, Yu L, Liu Y, Li L, Yang R, et al. Association between the atherogenic index of plasma and carotid artery plaques in patients with coronary heart disease in different glucose metabolism states: an RCSCD-TCM study in Tianjin, China. Endocrine. 2023;81:252–61. 10.1007/s12020-023-03389-5 [DOI] [PubMed] [Google Scholar]
- 22.Liu T, Li W, Zhang S, Wang Y, Zou X, Zhao Y, et al. Association of the atherogenic index of plasma and other lipid indices with abnormal glucose metabolism in patients with ischemic stroke. Front Endocrinol (Lausanne). 2025;16:1677653. 10.3389/fendo.2025.1677653 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Gan M, Chen B, Qin M, Zhang L, Wang X.Association between atherogenic index of plasma (AIP) and all-cause and cardiovascular mortality in patients with metabolic dysfunction-associated steatotic liver disease (MASLD): a cohort study based on NHANES 1999-2018. Cardiovasc Diabetol. 2025;24:391. 10.1186/s12933-025-02941-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Eid S, Sas KM, Abcouwer SF, Feldman EL, Gardner TW, Pennathur S, et al. New insights into the mechanisms of diabetic complications: role of lipids and lipid metabolism. Diabetologia. 2019;62:1539–49. 10.1007/s00125-019-4959-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zhang J, Liu C, Peng Y, Fang Q, Wei X, Zhang C, et al. Impact of baseline and trajectory of the atherogenic index of plasma on incident diabetic kidney disease and retinopathy in participants with type 2 diabetes: a longitudinal cohort study. Lipids Health Dis. 2024;23:11. 10.1186/s12944-024-02003-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Wu B, Niu Z, Hu F.Study on risk factors of peripheral neuropathy in type 2 diabetes mellitus and establishment of prediction model. Diabetes Metab J. 2021;45:526–38. 10.4093/dmj.2020.0100 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Zhang HH, Han X, Wang M, Hu Q, Li S, Wang M, et al. The association between genomic DNA methylation and diabetic peripheral neuropathy in patients with type 2 diabetes mellitus. J Diabetes Res. 2019;2019:2494057. 10.1155/2019/2494057 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Tracey TJ, Steyn FJ, Wolvetang EJ, Ngo ST.Neuronal lipid metabolism: multiple pathways driving functional outcomes in health and disease. Front Mol Neurosci. 2018;11:10. 10.3389/fnmol.2018.00010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Russo GT, De Cosmo S, Viazzi F, Pacilli A, Ceriello A, Genovese S, et al. Plasma triglycerides and HDL-C levels predict the development of diabetic kidney disease in subjects with type 2 diabetes: the AMD Annals Initiative. Diabetes Care. 2016;39:2278–87. 10.2337/dc16-1246 [DOI] [PubMed] [Google Scholar]
- 30.Zhu L, Lv T, Song S, Tan Y, She Y, Zhou X, et al. Association of atherogenic index of plasma with kidney dysfunction in diabetic individuals: findings from two national population-based studies. BMC Endocr Disord. 2025;25:105. 10.1186/s12902-025-01925-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Tramontano D, D’Erasmo L, Larouche M, Brisson D, Lauziere A, Di Costanzo A, et al. The vicious circle of chronic kidney disease and hypertriglyceridemia: what is first, the hen or the egg? Atherosclerosis. 2025;403:119146. 10.1016/j.atherosclerosis.2025.119146 [DOI] [PubMed] [Google Scholar]
- 32.Opazo-Ríos L, Mas S, Marín-Royo G, Mezzano S, Gómez-Guerrero C, Moreno JA, et al. Lipotoxicity and diabetic nephropathy: novel mechanistic insights and therapeutic opportunities. Int J Mol Sci. 2020;21:2630. 10.3390/ijms21072632 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Zhang Y, Chen G, Jing Y.Correlation between the atherogenic index of plasma and risk of diabetic retinopathy. Front Med (Lausanne). 2025;12:1565414. 10.3389/fmed.2025.1565414 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Wei B, Zhou L, Shu BL, Huang QY, Chai H, Yuan HY, et al. Association of diabetic retinopathy with plasma atherosclerotic index, visceral obesity index, and lipid accumulation products: NHANES 2005-2008. Lipids Health Dis. 2024;23:331. 10.1186/s12944-024-02325-4 [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
Data Availability Statement
Data availability: The data sets and codes analysed in this study are available from the corresponding author upon reasonable request.