Abstract
Background
A higher atherogenic index of plasma (AIP; calculated as log10[triglycerides/high-density lipoprotein cholesterol]) is an indicator of atherogenic dyslipidemia. Whether the cumulative AIP (cAIP) burden provides prognostic information beyond the cumulative low-density lipoprotein cholesterol (cLDL-C) burden remains unclear.
Methods and Results
This retrospective study analyzed data for 11,402 adults with at least 3 health checkups. cAIP and cLDL-C burdens were calculated from the first 3 lipid-complete visits, with the third visit defined as the index date. The primary outcome was first non-fatal myocardial infarction (MI) or ischemic stroke. Cox models were adjusted for conventional cardiovascular risk factors and cLDL-C burden. Over a mean follow-up of 4.96 years, 435 primary events occurred. Per 1-SD increase, cAIP burden was independently associated with the primary outcome after full adjustment including cLDL-C burden (hazard ratio 1.14; 95% confidence interval 1.02–1.27). Restricted cubic splines showed a progressive increase in risk with higher AIP burden. In discordance analysis, low cLDL-C/high cAIP burden, but not high cLDL-C/low cAIP burden, was associated with a higher risk than low cLDL-C/low cAIP burden (hazard ratio 1.40; 95% confidence interval 1.04–1.89). The addition of AIP burden did not improve global discrimination or reclassification.
Conclusions
cAIP burden was associated with incident non-fatal MI or ischemic stroke and identified excess risk despite low cLDL-C burden, supporting its complementary value for characterizing residual lipid-related risk.
Key Words: Atherogenic index of plasma, Ischemic stroke, Low-density lipoprotein cholesterol, Myocardial infarction, Residual cardiovascular risk
Central Figure.
Atherosclerotic cardiovascular disease, including myocardial infarction (MI) and ischemic stroke, remains a leading cause of morbidity and mortality worldwide,1–4 and its prevention continues to be a major clinical and public health priority. Low-density lipoprotein cholesterol (LDL-C) is a central causal factor in the development of atherosclerotic cardiovascular disease,5,6 and lowering LDL-C is a cornerstone of both primary and secondary prevention. However, cardiovascular events continue to occur even when LDL-C is adequately controlled, highlighting the presence of residual risk.7,8 Part of this residual risk may be related to atherogenic dyslipidemia, characterized by elevated triglyceride (TG)-rich lipoproteins and low high-density lipoprotein cholesterol (HDL-C).9
The atherogenic index of plasma (AIP) is a simple lipid marker, with higher values reflecting the phenotype of TG-rich, low-HDL-C dyslipidemia.9,10 AIP has also been linked to small dense LDL particles and insulin resistance, and has therefore been considered a surrogate marker of an atherogenic lipoprotein profile.11 Previous studies have shown that a higher AIP is associated with coronary artery disease and adverse cardiovascular outcomes.12
However, most previous studies have relied on a single measurement of AIP and have not adequately captured long-term exposure to adverse lipid profiles. Atherosclerosis is a chronic process driven by cumulative exposure to atherogenic factors, and the concept of cumulative lipid burden has gained increasing importance in cardiovascular risk assessment.13,14 Consistent with this concept, recent evidence from Matsumoto et al.15 has emphasized the clinical relevance of long-term exposure to elevated LDL-C levels after coronary artery disease, supporting the importance of cumulative lipid burden rather than measurement at a single time point. Although recent studies have reported associations between cumulative AIP and cardiovascular outcomes, these analyses have largely focused on broad definitions of cardiovascular disease or a composite of major adverse cardiovascular events. Accordingly, it remains unclear whether the cumulative AIP burden provides prognostic information independent of, and complementary to, cumulative LDL-C burden.16
In addition, few studies have jointly evaluated the cumulative LDL-C burden and cumulative AIP burden to examine whether discordance between these 2 dimensions of lipid-related risk is clinically relevant. This question is particularly important because the LDL-C burden primarily reflects cholesterol exposure, whereas the cumulative AIP burden may better capture the long-term burden of atherogenic dyslipidemia characterized by TG-rich lipoproteins and low HDL-C. Data are especially limited for more specific and clinically interpretable atherosclerotic cardiovascular disease outcomes such as non-fatal MI and non-fatal ischemic stroke.
Therefore, in the present longitudinal cohort study based on annual health checkup data, we investigated the association between cumulative AIP burden derived from repeated measurements and incident non-fatal MI or ischemic stroke. We further examined whether this association persisted after accounting for cumulative LDL-C burden and performed a discordance analysis between the cumulative LDL-C burden and cumulative AIP burden to assess the complementary value of the cumulative AIP burden in capturing residual atherogenic cardiovascular risk.
Methods
Study Design and Population
This was a single-center retrospective longitudinal cohort study based on annual health checkup data. Health checkups were conducted from 2013 to 2020, and standardized data were collected at each visit, including demographic characteristics, anthropometric measurements, blood pressure, blood and urine test results, electrocardiographic findings, medical history, and medication use. Health checkup records were linked to hospital electronic medical record and inpatient databases using unique individual identifiers to ascertain subsequent cardiovascular events.
Eligible participants were adults who underwent health checkups at Chutoen General Medical Center between 2013 and 2020 and had at least 3 visits with complete measurements of TGs, HDL-C, and LDL-C. The cumulative AIP burden and cumulative LDL-C burden were calculated using the first 3 lipid-complete visits for each participant. Individuals with a history of acute MI or ischemic stroke before the index date, defined as the third lipid-complete visit, were excluded, as were participants without follow-up data available for after the index date.
Using a fixed landmark design, cumulative lipid exposures were assessed only from the first 3 lipid-complete visits up to the index date. Follow-up for cardiovascular outcomes started on the index date and continued until occurrence of the first outcome event, the last available clinical follow-up, or February 2026, whichever came first.
Assessment of Lipid Exposure
At each health checkup, AIP was calculated from serum TG and HDL-C using the following formula:17
AIP = log10(TG / HDL-C)
The same units of measurement were used for TG and HDL-C values.
To evaluate cumulative exposure to atherogenic dyslipidemia, the cumulative AIP burden was calculated from the first 3 lipid-complete visits for each participant. We used a trapezoidal time-weighted average to account for differences in the time intervals between visits. Briefly, the mean of 2 adjacent AIP values was multiplied by the time interval between those visits, and the sum of these interval-specific values was divided by the total time from the first to the third lipid-complete visit, as follows:
where AIP1, AIP2, and AIP3 represent the AIP values at the first, second, and third lipid-complete visits, respectively, and t1, t2, and t3 represent the corresponding visit dates. The cumulative LDL-C burden was calculated from LDL-C values obtained at the same 3 visits using the same trapezoidal time-weighted averaging method.
Outcome Definition
The primary outcome was the first occurrence of non-fatal acute MI or non-fatal ischemic stroke after the index date. Acute MI was identified using International Classification of Diseases, Tenth Revision (ICD-10) codes I21–I22, and ischemic stroke was identified using ICD-10 code I63, based on physician-diagnosed events recorded in the institutional clinical database.18 At Chutoen General Medical Center, diagnoses of acute MI and ischemic stroke are routinely established according to standard clinical practice, including cardiac biomarker elevation and/or coronary imaging for MI, and imaging-confirmed cerebral infarction on computed tomography or magnetic resonance imaging for ischemic stroke.
Secondary outcomes were non-fatal acute MI alone and non-fatal ischemic stroke alone. All outcomes were evaluated as first events occurring after the index date.
Covariates
Covariates were selected a priori based on established cardiovascular risk factors and were obtained from the health checkup data at the index date. These included age, sex, body mass index (BMI), systolic blood pressure (SBP), smoking status, diabetes, the use of statins, and estimated glomerular filtration rate (eGFR).
Statistical Analyses
Continuous variables are presented as the mean ± SD or median with interquartile range (IQR), as appropriate, and categorical variables are presented as numbers and percentages. Participants were categorized into quartiles according to the distribution of cumulative AIP burden, and baseline characteristics were compared across quartiles. Quartiles of cumulative AIP burden were defined as Q1 (≤−0.0445), Q2 (>−0.0445 to ≤0.1328), Q3 (>0.1328 to ≤0.3367), and Q4 (>0.3367).
To evaluate the association between cumulative AIP burden and the primary outcome after the index date, Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Three hierarchical models were constructed. Model 1 was adjusted for age and sex. Model 2 was additionally adjusted for BMI, SBP, smoking status, diabetes, the use of statins, and eGFR. Model 3 further included cumulative LDL-C burden to examine whether the cumulative AIP burden provided prognostic information independent of cumulative LDL-C exposure.
The proportional hazards assumption was assessed using Schoenfeld residuals. To evaluate the dose–response relationship between the cumulative AIP burden and event risk, restricted cubic spline analyses were performed.
To examine the clinical relevance of discordance between the cumulative LDL-C burden and the cumulative AIP burden, participants were classified into 4 groups according to the median values of the 2 measures (LDL-C burden, 118.98 mg/dL; AIP burden, 0.1328), as follows: low LDL-C/low AIP burden, low LDL-C/high AIP burden, high LDL-C/low AIP burden, and high LDL-C/high AIP burden. The low LDL-C/low AIP burden group served as the reference category.
To assess the incremental prognostic value of the cumulative AIP burden beyond conventional cardiovascular risk factors and the cumulative LDL-C burden, we compared a base Cox model including age, sex, BMI, SBP, smoking status, diabetes, the use of statins, eGFR, and the cumulative LDL-C burden with an extended model additionally including the cumulative AIP burden. Model discrimination was evaluated using Harrell’s C-index, and improvement in risk prediction was further assessed using the continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI) at 3 years.
Prespecified subgroup and interaction analyses were performed to assess whether the association between the cumulative AIP burden and outcomes differed according to clinical background, including diabetes and impaired kidney function. Sensitivity analyses were performed to assess the robustness of the primary findings, including analyses excluding statin users and analyses using an alternative burden definition, in which the cumulative AIP burden and cumulative LDL-C burden were calculated as the simple mean of the first 3 measurements instead of the trapezoidal time-weighted average used in the primary analysis. Exploratory analyses evaluated long-term variability in AIP using the within-person standard deviation and coefficient of variation, and examined their associations with cardiovascular outcomes.
Two-sided P<0.05 was considered statistically significant. Missing TG, HDL-C, and LDL-C values were not imputed, and only visits with complete measurements of all 3 lipid variables were used for exposure assessment. Accordingly, the analytic cohort was restricted to participants with at least 3 visits with complete lipid measurements. In contrast, missing values in covariates including age, sex, BMI, SBP, smoking status, diabetes, statin use, and eGFR were handled using multiple imputation by chained equations under the assumption that the data were missing at random. Analyses were performed across the imputed datasets, and estimates were combined using Rubin’s rules. All statistical analyses were performed using R version 4.5.1 (R Foundation for Statistical Computing, Vienna, Austria).
Ethical Considerations
This study was conducted in accordance with the principles of the Declaration of Helsinki and current ethical guidelines, and was approved by the Ethics Committee of Chutoen General Medical Center (Approval no. 1351260402). The requirement for written informed consent was waived in favor of an opt-out approach, and study information was publicly disclosed on the hospital website in accordance with the Act on the Protection of Personal Information. If a patient or a family member requested withdrawal from the study, the corresponding data were excluded from the analysis.
Results
Study Population
Between 2013 and 2020, 35,983 adults who underwent at least one health checkup at Chutoen General Medical Center were screened. Of these adults, 14,276 had at least 3 lipid-complete visits. After excluding 34 participants with prior MI before the index date, 112 with prior ischemic stroke before the index date, and 2,728 without follow-up data after the index date, 11,402 participants were included in the final analysis (Figure 1).
Figure 1.
Flow diagram of study participant selection from among adults who underwent health checkups between 2013 and 2020. Participants were screened according to the availability of at least 3 lipid-complete visits, a history of myocardial infarction or ischemic stroke before the index date (defined as the third lipid-complete visit), and the availability of follow-up data after the index date.
The mean age was 52.5 years, and 52.3% were men. The cumulative AIP burden and cumulative LDL-C burden were calculated from the first 3 lipid-complete visits, with the third visit defined as the index date. The median interval from the first to the third lipid-complete visit was 735 days (IQR 720–817 days), corresponding to 2.01 years (IQR 1.97–2.24 years). The median intervals from the first to second and from the second to third lipid-complete visits were 365 (IQR 354–404) and 368 (IQR 357–399) days, respectively. During a mean follow-up of 4.96 years, 435 primary outcome events occurred, including 136 MIs and 299 ischemic strokes.
Baseline Characteristics According to Cumulative AIP Burden
Participants were classified into quartiles (Q1–Q4) according to the cumulative AIP burden (Table 1). Those in the higher quartiles were more likely to be men and had higher BMI, SBP, glucose, and HbA1c levels, whereas eGFR was lower across increasing quartiles (all P<0.001). The prevalence of diabetes, current smoking, and statin use also increased progressively with higher cumulative AIP burden.
Table 1.
Baseline Characteristics According to Quartiles of Cumulative Atherogenic Index of Plasma Burden
| Overall (n=11,402) |
Q1 (n=2,851) |
Q2 (n=2,850) |
Q3 (n=2,850) |
Q4 (n=2,851) |
P value | |
|---|---|---|---|---|---|---|
| Demographic and clinical characteristics | ||||||
| Age (years) | 52.5±12.9 | 49.0±13.3 | 52.7±13.4 | 54.7±12.5 | 54.4±11.8 | <0.001 |
| Male sex | 5,958 (52.3) | 744 (26.1) | 1,219 (42.8) | 1,730 (60.7) | 2,265 (79.4) | <0.001 |
| BMI (kg/m2) | 22.5±3.48 | 20.5±2.56 | 21.8±2.89 | 23.1±3.31 | 24.7±3.58 | <0.001 |
| SBP (mmHg) | 119.6±15.7 | 114.5±15.2 | 118.1±15.7 | 121.5±15.3 | 124.2±15.1 | <0.001 |
| Current smoker | 1949 (17.1) | 236 (8.3) | 398 (14.0) | 546 (19.2) | 769 (27.0) | <0.001 |
| Diabetes | 562 (4.9) | 65 (2.3) | 103 (3.6) | 150 (5.3) | 244 (8.6) | <0.001 |
| Statin use | 1,216 (10.7) | 145 (5.1) | 250 (8.8) | 358 (12.6) | 463 (16.2) | <0.001 |
| eGFR (mL/min/1.73 m2) | 77.2±14.8 | 81.0±14.4 | 77.9±15.1 | 75.7±14.3 | 74.6±14.5 | <0.001 |
| Lipid and metabolic variables | ||||||
| Triglycerides (mg/dL) | 84.0 [60.0–121.0] | 52.0 [42.0–63.0] | 72.5 [60.0–87.0] | 97.0 [80.0–118.0] | 152.0 [119.5–199.0] | <0.001 |
| HDL-C (mg/dL) | 62.6±15.7 | 77.5±14.2 | 66.6±11.6 | 58.1±10.3 | 48.3±9.13 | <0.001 |
| LDL-C (mg/dL) | 119.4±29.6 | 106.3±27.1 | 117.4±26.9 | 125.1±28.1 | 129.0±30.9 | <0.001 |
| Non-HDL-C (mg/dL) | 141.3±32.8 | 123.9±29.0 | 135.6±28.5 | 145.5±29.3 | 160.2±32.9 | <0.001 |
| Total cholesterol (mg/dL) |
203.9±33.2 | 201.4±33.8 | 202.2±32.2 | 203.6±31.8 | 208.5±34.3 | <0.001 |
| HbA1c (%) | 5.74±0.59 | 5.59±0.40 | 5.66±0.44 | 5.76±0.56 | 5.93±0.78 | <0.001 |
| Follow-up duration (years) |
4.96±3.98 | 4.74±3.92 | 5.00±3.97 | 5.08±4.02 | 5.00±3.99 | 0.007 |
Unless indicated otherwise, values are presented as the mean ± SD, median [interquartile range], or n (%). P values were calculated using analysis of variance or the Kruskal–Wallis test for continuous variables, and the Chi-squared test for categorical variables, as appropriate. Quartiles of cumulative AIP burden were defined as Q1 (≤−0.0445), Q2 (>−0.0445 to ≤0.1328), Q3 (>0.1328 to ≤0.3367), and Q4 (>0.3367). AIP, atherogenic index of plasma; BMI, body mass index; eGFR, estimated glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; SBP, systolic blood pressure.
Primary outcome events were more frequent in the higher quartiles, increasing from 63 in Q1 to 147 in Q4. The relationship between the cumulative AIP burden and cumulative LDL-C burden was examined further in the adjusted and discordance analyses.
Association Between Cumulative AIP Burden and the Primary Outcome
The cumulative AIP burden was significantly associated with the primary outcome in Cox proportional hazards models (Table 2). When analyzed as a continuous variable, each 1-SD increase in cumulative AIP burden was associated with a higher risk of the primary outcome in Model 1 (HR 1.23; 95% CI 1.11–1.36; P<0.001), Model 2 (HR 1.15; 95% CI 1.03–1.28; P=0.011), and Model 3, which was additionally adjusted for cumulative LDL-C burden (HR 1.14; 95% CI 1.02–1.27; P=0.022).
Table 2.
Association of Cumulative Atherogenic Index of Plasma Burden With the Primary Outcome (First Non-Fatal Myocardial Infarction or Ischemic Stroke)
| Exposure | Model 1 | Model 2 | Model 3 | |||
|---|---|---|---|---|---|---|
| HR (95% CI) | P value | HR (95% CI) | P value | HR (95% CI) | P value | |
| Continuous analysis | ||||||
| Per 1-SD increase in cumulative AIP burden |
1.23 (1.11–1.36) | <0.001 | 1.15 (1.03–1.28) | 0.011 | 1.14 (1.02–1.27) | 0.022 |
| Quartile analysis | ||||||
| Quartile 1 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Quartile 2 | 1.01 (0.73–1.41) | 0.934 | 0.95 (0.68–1.32) | 0.742 | 0.92 (0.66–1.28) | 0.621 |
| Quartile 3 | 1.47 (1.08–1.99) | 0.013 | 1.31 (0.95–1.79) | 0.094 | 1.25 (0.91–1.72) | 0.172 |
| Quartile 4 | 1.51 (1.11–2.05) | 0.009 | 1.23 (0.89–1.72) | 0.208 | 1.17 (0.84–1.64) | 0.352 |
Model 1 was adjusted for age and sex. Model 2 was additionally adjusted for BMI, SBP, smoking status, diabetes, statin use, and eGFR. Model 3 was further adjusted for cumulative LDL-C burden. Quartiles of cumulative AIP burden were defined as Q1 (≤−0.0445), Q2 (>−0.0445 to ≤0.1328), Q3 (>0.1328 to ≤0.3367), and Q4 (>0.3367). CI, confidence interval; HR, hazard ratio. Other abbreviations as in Table 1.
In quartile-based analyses, a higher cumulative AIP burden was associated with increased risk in Model 1, with significant associations for Q3 and Q4 compared with Q1. This association was attenuated after further adjustment, and the quartile-based estimates were no longer statistically significant in Model 3. Overall, the continuous analysis supported an independent association between the cumulative AIP burden and the primary outcome beyond cumulative LDL-C burden.
The addition of the cumulative AIP burden to the model including conventional cardiovascular risk factors and the cumulative LDL-C burden resulted in minimal improvement in model discrimination, with essentially no change in the C-index (0.698 for both). Risk reclassification metrics were also modest and not statistically significant (continuous NRI 0.043 [95% CI −0.038, 0.137]; IDI 0.00001 [95% CI −0.00042, 0.00076]; Supplementary Table 1).
Dose–Response Relationship Between Cumulative AIP Burden and the Primary Outcome
Restricted cubic spline analysis showed a progressive increase in the risk of the primary outcome with higher cumulative AIP burden (Figure 2). The increase in risk appeared gradual across the observed range, without a clear threshold.
Figure 2.
Dose–response relationship between cumulative atherogenic index of plasma (AIP) burden and the primary outcome (first non-fatal myocardial infarction or ischemic stroke). Restricted cubic spline curve showing the association between cumulative AIP burden and the adjusted hazard ratio for the primary outcome. The solid line indicates the estimated hazard ratio, and the shaded area represents the 95% confidence interval.
Discordance Analysis Between Cumulative LDL-C Burden and Cumulative AIP Burden
To examine the clinical relevance of discordance between the cumulative LDL-C burden and cumulative AIP burden, participants were classified into 4 groups according to the median values of the 2 measures (LDL-C burden, 118.98 mg/dL; AIP burden, 0.1328; Figure 3; Supplementary Table 2). The absolute event rates were 2.23% in the low LDL-C/low AIP burden group, 5.50% in the low LDL-C/high AIP burden group, 3.22% in the high LDL-C/low AIP burden group, and 4.76% in the high LDL-C/high AIP burden group.
Figure 3.
Discordance analysis of cumulative low-density lipoprotein cholesterol (LDL-C) burden and cumulative atherogenic index of plasma (AIP) burden. (A) Forest plot showing adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for the primary outcome (first non-fatal myocardial infarction or ischemic stroke) according to 4 groups defined by the median values of cumulative LDL-C burden and cumulative AIP burden. The low LDL-C/low AIP burden group was used as the reference group. (B) Kaplan–Meier curves showing the cumulative incidence of the primary outcome across the same 4 discordance groups. Clear separation of event curves was observed, particularly for the low LDL-C/high AIP burden group, supporting the excess risk identified in the adjusted analyses.
Using the low LDL-C/low AIP burden group as the reference, the low LDL-C/high AIP burden group had a significantly higher risk of the primary outcome (HR 1.40; 95% CI 1.04–1.89; P=0.028). In contrast, the high LDL-C/low AIP burden group was not significantly associated with a higher risk of the primary outcome (HR 1.12; 95% CI 0.81–1.55; P=0.503), whereas the high LDL-C/high AIP burden group had a significantly elevated risk (HR 1.38; 95% CI 1.04–1.83;P=0.024).
Notably, the low LDL-C/high AIP burden group had more than twice the absolute event rate of the low LDL-C/low AIP burden group, supporting the finding that cumulative AIP burden may identify excess risk not fully explained by the cumulative LDL-C burden alone. Cumulative event curves for the 4 discordance groups are shown in Figure 3.
Association Between Cumulative AIP Burden and Secondary Outcomes
In analyses of secondary outcomes, the cumulative AIP burden was significantly associated with MI alone (HR 1.36 per 1-SD increase; 95% CI 1.13–1.64; P=0.001) in the fully adjusted model (Table 3). In contrast, the association with ischemic stroke alone was not statistically significant (HR 1.04; 95% CI 0.90–1.19; P=0.599).
Table 3.
Association of Cumulative Atherogenic Index of Plasma Burden With the Secondary Outcomes in the Fully Adjusted Model
| Outcome | HR (95% CI) | P value |
|---|---|---|
| Non-fatal MI alone | 1.36 (1.13–1.64) | 0.001 |
| Non-fatal ischemic stroke alone | 1.04 (0.90–1.19) | 0.599 |
HRs and 95% CIs for the associations of cumulative atherogenic index of plasma burden with non-fatal MI alone and non-fatal ischemic stroke alone were estimated using Cox proportional hazards models adjusted for age, sex, BMI, SBP, smoking status, diabetes, statin use, eGFR, and cumulative LDL-C burden. MI, myocardial infarction. Other abbreviations as in Tables 1,2.
Subgroup, Sensitivity, and Exploratory Analyses
In prespecified subgroup analyses, the association between cumulative AIP burden and the primary outcome was broadly directionally consistent across most subgroups, although statistical significance was more evident in men, older participants, and those without chronic kidney disease (Supplementary Figure). The association did not reach statistical significance in women, younger participants, participants with diabetes, or those with chronic kidney disease.
Formal interaction testing did not show statistically significant interactions by sex (Pinteraction=0.051), age group (Pinteraction=0.559), diabetes (Pinteraction=0.865), or chronic kidney disease status (Pinteraction=0.159). Therefore, these subgroup findings should be interpreted as exploratory.
Sensitivity analyses yielded largely similar results after the exclusion of statin users, the application of an alternative burden definition, and comparison of complete-case and imputed datasets (Table 4), supporting the robustness of the main findings. In exploratory analyses, AIP variability was not significantly associated with the primary outcome either without adjustment for cumulative AIP burden (HR 1.00; 95% CI 0.90–1.10; P=0.932) or after additional adjustment for cumulative AIP burden (HR 0.97, 95% CI 0.88–1.07; P=0.590).
Table 4.
Sensitivity Analyses for the Primary Outcome
| Sensitivity analysis | Model/dataset | Exposure term | HR (95% CI) | P value |
|---|---|---|---|---|
| Excluding lipid-lowering medication users |
Participants not receiving lipid-lowering medication; adjusted for age, sex, BMI, SBP, diabetes, current smoking, eGFR, and cumulative LDL-C burden |
Cumulative AIP burden (per 1-SD increase) |
1.06 (0.94–1.20) | 0.364 |
| Alternative burden definition | Fully adjusted model using alternative burden definitions for both AIP and LDL-C |
Alternative cumulative AIP burden (per 1-SD increase) |
1.14 (1.02–1.28) | 0.021 |
| Comparison by missing-data strategy |
Complete-case analysis | Cumulative AIP burden (per 1-SD increase) |
1.18 (1.04–1.33) | 0.008 |
| Imputed dataset analysis | Cumulative AIP burden (per 1-SD increase) |
1.14 (1.02–1.27) | 0.023 |
HRs were estimated for cumulative AIP burden per 1-standard deviation increase using Cox proportional hazards models across predefined sensitivity analyses. In the alternative burden definition, the cumulative AIP burden and cumulative LDL-C burden were calculated as the simple mean of the first 3 measurements instead of using the trapezoidal time-weighted average (as in the primary analysis). The imputed dataset analysis used multiple imputation for missing covariates. Abbreviations as in Tables 1,2.
Discussion
In this longitudinal health checkup cohort study, a higher cumulative AIP burden was independently associated with incident MI or ischemic stroke, even after adjustment for cumulative LDL-C burden. The most clinically relevant finding was the discordance analysis: participants with low cumulative LDL-C burden but high cumulative AIP burden remained at increased risk, whereas those with high cumulative LDL-C burden but low cumulative AIP burden did not show a statistically significant excess risk. In addition, the association appeared stronger for MI than for ischemic stroke. Together, these findings suggest that the cumulative AIP burden may provide complementary information on residual atherogenic risk beyond cumulative LDL-C exposure. Although cumulative AIP burden showed minimal improvement in global discrimination metrics, this finding is consistent with the modest effect size observed in the primary analysis and suggests that cumulative AIP burden is not a strong global predictor of cardiovascular events.
These findings are biologically plausible and broadly consistent with prior literature. LDL-C remains a central causal factor in atherosclerosis, and cumulative LDL-C exposure has emerged as an important determinant of future cardiovascular risk.5,13 At the same time, substantial residual risk persists despite LDL-C lowering, and TG-rich lipoproteins and their remnants are increasingly recognized as important contributors to that risk.9
AIP, calculated from TGs and HDL-C, is thought to reflect a broader atherogenic dyslipidemic phenotype linked to TG-rich lipoproteins, low HDL-C, insulin resistance, and small dense LDL particles.11 Recent studies have also highlighted the prognostic relevance of lipid-related phenotypes beyond LDL-C concentration alone, including small dense LDL-C and lipid ratios such as the LDL-C/HDL-C ratio.19,20 These findings support the concept that composite or phenotype-based lipid assessment may provide complementary information for cardiovascular risk characterization. Previous studies and recent meta-analyses have reported associations between higher AIP and coronary artery disease, stroke, and adverse cardiovascular outcomes.12 A recent community-based cohort study further showed that cumulative AIP was associated with incident major adverse cardiovascular events, stroke, and MI.16
Our results extend these previous findings in 2 ways. First, rather than examining cumulative AIP alone, we evaluated the cumulative AIP burden together with the cumulative LDL-C burden. This is important because the cumulative LDL-C burden mainly captures cholesterol exposure, whereas the cumulative AIP burden may better reflect long-term TG-rich, low-HDL-C dyslipidemia. Second, the discordance analysis suggests that an elevated cumulative AIP burden may identify individuals with excess risk despite a relatively low cumulative LDL-C burden. This finding supports the concept that the cumulative AIP burden captures a lipid-related risk dimension not fully reflected by LDL-C alone, which is consistent with current evidence that TG-rich lipoprotein remnants contribute to atherosclerosis independently of LDL-C.9 Thus, the clinical value of the cumulative AIP burden may lie less in improving overall risk prediction performance and more in identifying specific TG-rich, low-HDL-C cardiometabolic phenotypes that are not adequately captured by conventional lipid measures.
The stronger association observed for MI than for ischemic stroke may also be meaningful. Coronary events may be more directly linked to atherogenic dyslipidemia characterized by TG-rich lipoproteins and remnant particles, whereas ischemic stroke is biologically more heterogeneous and may be influenced by competing mechanisms beyond lipid-related atherosclerosis.21 Importantly, our findings should be interpreted as hypothesis-generating rather than practice-changing. The limited improvements in the C-index, NRI, and IDI indicate that the cumulative AIP burden is unlikely to serve as a strong global risk prediction marker when added to conventional risk factors and the cumulative LDL-C burden. Instead, its potential value may lie in identifying a specific cardiometabolic phenotype characterized by TG-rich and low-HDL-C dyslipidemia among individuals with an apparently low cumulative LDL-C burden. Therefore, the AIP should be considered a complementary marker for risk characterization rather than a stand-alone target for treatment decision-making. Recent studies are consistent with this cautious interpretation. High cumulative non-HDL-C exposure has been associated with arterial stiffness, small dense LDL-C has predicted ischemic heart disease regardless of LDL-C level, and combined control of LDL-C and TGs has been linked to better long-term outcomes after PCI in patients with diabetes.22–24 These findings support the concept that LDL-C alone may not fully characterize lipid-related cardiovascular risk.
Study Limitations
Several limitations should be acknowledged. First, this was a retrospective single-center study of participants undergoing health checkups, and the relatively young cohort and intermediate follow-up duration may limit generalizability and statistical power for some subgroup and secondary outcome analyses. Second, cardiovascular events were identified using institutional clinical records and ICD-10 codes; therefore, events treated at other hospitals or out-of-hospital deaths may not have been captured, leading to possible underestimation or misclassification of outcomes. Third, residual confounding cannot be excluded because information on diet, alcohol intake, physical activity, socioeconomic status, and other lifestyle factors was unavailable. In addition, lipid-lowering therapy was assessed only at the index date, and treatment changes during follow-up were not incorporated. Fourth, the cumulative lipid burden was calculated from the first 3 lipid-complete health check-ups and may not fully reflect longer-term lipid exposure after the index date. Finally, AIP is a composite surrogate marker derived from TGs and HDL-C and should not be interpreted as a specific biological pathway independent of insulin resistance, obesity, or metabolic syndrome. Direct measures of apolipoprotein B, remnant cholesterol, lipoprotein(a), small dense LDL particles, and inflammatory biomarkers were not available. Moreover, adding the cumulative AIP burden did not materially improve discrimination or reclassification metrics; therefore, these findings support its role as a complementary marker for risk characterization rather than a stand-alone tool for treatment decision-making.
Conclusions
In conclusion, the cumulative AIP burden was independently associated with incident MI or ischemic stroke beyond the cumulative LDL-C burden. The excess risk observed in individuals with a low cumulative LDL-C burden but high cumulative AIP burden supports the complementary role of the cumulative AIP burden in identifying residual atherogenic risk.
Disclosures
The authors declare that they have no competing interests.
IRB Information
This study was approved by the Ethics Committee on Medical Research of Chutoen General Medical Center (Reference no. 1351260402).
Supplementary Files
Supplementary Table 1. Supplementary Table 2. Supplementary Figure.
Acknowledgments
The authors thank Takahiro Imaizumi, MD, PhD, and Yuji Itho, MD (Department of Cardiology, Chutoen General Medical Center, Kakegawa, Japan), for their contributions to this study. The authors also thank all the physicians who made this study possible. During manuscript preparation, the authors used an AI-based language model (ChatGPT, OpenAI) solely to assist with English translation and linguistic refinement. All content was critically reviewed, verified, and edited by the authors to ensure scientific accuracy and integrity. The authors take full responsibility for the content and conclusions of the manuscript.
Funding Statement
Sources of Funding: This research did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.
Data Availability
The data underlying this article will be shared upon reasonable request to the corresponding author.
References
- 1. Chong B, Jayabaskaran J, Jauhari SM, Chan SP, Goh R, Kueh MTW, et al.. Global burden of cardiovascular diseases: Projections from 2025 to 2050. Eur J Prev Cardiol 2025; 32: 1001–1015, doi:10.1093/eurjpc/zwae281. [DOI] [PubMed] [Google Scholar]
- 2. Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al.. Global burden of cardiovascular diseases and risk factors, 1990–2019: Update from the GBD 2019 study. J Am Coll Cardiol 2020; 76: 2982–3021, doi:10.1016/j.jacc.2020.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Tokgozoglu L, Kayıkcioglu M, Roeters van Lennep J.. Sex-specific differences in cardiovascular risk factors and their management. Atherosclerosis 2026; 414: 120641, doi:10.1016/j.atherosclerosis.2026.120641. [DOI] [PubMed] [Google Scholar]
- 4. Martin SS, Aday AW, Allen NB, Almarzooq ZI, Anderson CAM, Arora P, et al.. 2025 heart disease and stroke statistics: A report of US and global data from the American Heart Association. Circulation 2025; 151: e41–e660, doi:10.1161/CIR.0000000000001303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ference BA, Ginsberg HN, Graham I, Ray KK, Packard CJ, Bruckert E, et al.. Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies. A consensus statement from the European Atherosclerosis Society Consensus Panel. Eur Heart J 2017; 38: 2459–2472, doi:10.1093/eurheartj/ehx144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Mach F, Baigent C, Catapano AL, Koskinas KC, Casula M, Badimon L, et al.. 2019 ESC/EAS Guidelines for the management of dyslipidaemias: Lipid modification to reduce cardiovascular risk. Eur Heart J 2020; 41: 111–188, doi:10.1093/eurheartj/ehz455. [DOI] [PubMed] [Google Scholar]
- 7. Ginsberg HN, Packard CJ, Chapman MJ, Borén J, Aguilar-Salinas CA, Averna M, et al.. Triglyceride-rich lipoproteins and their remnants: Metabolic insights, role in atherosclerotic cardiovascular disease, and emerging therapeutic strategies: A consensus statement from the European Atherosclerosis Society. Eur Heart J 2021; 42: 4791–4806, doi:10.1093/eurheartj/ehab551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Reijnders E, van der Laarse A, Jukema JW, Cobbaert CM.. High residual cardiovascular risk after lipid-lowering: Prime time for predictive, preventive, personalized, participatory, and psycho-cognitive medicine. Front Cardiovasc Med 2023; 10: 1264319, doi:10.3389/fcvm.2023.1264319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Chapman MJ, Packard CJ, Björnson E, Ginsberg HN, Borén J.. Triglyceride-rich lipoproteins, remnants and atherosclerotic cardiovascular disease: What we know and what we need to know. Atherosclerosis 2025; 410: 120529, doi:10.1016/j.atherosclerosis.2025.120529. [DOI] [PubMed] [Google Scholar]
- 10. Dobiásová M, Frohlich J.. The plasma parameter log(TG/HDL-C) as an atherogenic index: Correlation with lipoprotein particle size and esterification rate in apoB-lipoprotein-depleted plasma (FERHDL). Clin Biochem 2001; 34: 583–588, doi:10.1016/S0009-9120(01)00263-6. [DOI] [PubMed] [Google Scholar]
- 11. Onat A, Can G, Kaya H, Hergenç G.. Atherogenic index of plasma (log10 triglyceride/high-density lipoprotein-cholesterol) predicts high blood pressure, diabetes, and vascular events. J Clin Lipidol 2010; 4: 89–98, doi:10.1016/j.jacl.2010.02.005. [DOI] [PubMed] [Google Scholar]
- 12. Rabiee Rad M, Ghasempour Dabaghi G, Darouei B, Amani-Beni R.. The association of atherogenic index of plasma with cardiovascular outcomes in patients with coronary artery disease: A systematic review and meta-analysis. Cardiovasc Diabetol 2024; 23: 119, doi:10.1186/s12933-024-02198-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Ference BA, Braunwald E, Catapano AL.. The LDL cumulative exposure hypothesis: Evidence and practical applications. Nat Rev Cardiol 2024; 21: 701–716, doi:10.1038/s41569-024-01039-5. [DOI] [PubMed] [Google Scholar]
- 14. Zhang Y, Pletcher MJ, Vittinghoff E, Clemons AM, Jacobs DR Jr, Allen NB, et al.. Association between cumulative low-density lipoprotein cholesterol exposure during young adulthood and middle age and risk of cardiovascular events. JAMA Cardiol 2021; 6: 1406–1413, doi:10.1001/jamacardio.2021.3508. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Matsumoto I, Mukaida K, Wada K, Kurozumi M, Namba T, Takagi Y.. Long-term exposure to high low-density lipoprotein cholesterol levels and clinical course in secondary prevention after coronary artery disease. Circ Rep 2025; 7: 1116–1124, doi:10.1253/circrep.CR-25-0091. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Liu Z, Zhang L, Wang L, Li K, Fan F, Jia J, et al.. The predictive value of cumulative atherogenic index of plasma (AIP) for cardiovascular outcomes: A prospective community-based cohort study. Cardiovasc Diabetol 2024; 23: 264, doi:10.1186/s12933-024-02350-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Dobiásová M.. Atherogenic index of plasma [log(triglycerides/HDL-cholesterol)]: Theoretical and practical implications. Clin Chem 2004; 50: 1113–1115, doi:10.1373/clinchem.2004.033175. [DOI] [PubMed] [Google Scholar]
- 18. Ono Y, Taneda Y, Takeshima T, Iwasaki K, Yasui A.. Validity of claims diagnosis codes for cardiovascular diseases in diabetes patients in Japanese administrative database. Clin Epidemiol 2020; 12: 367–375, doi:10.2147/CLEP.S245555. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Koba S, Yokota Y, Satoh N, Ito Y, Tsunoda F, Sakai K, et al.. Small dense low-density lipoprotein cholesterol predicts long-term coronary artery disease recurrence in patients without metabolic dyslipidemia. Circ J 2026; 90: 999–1009, doi:10.1253/circj.CJ-25-0805. [DOI] [PubMed] [Google Scholar]
- 20. Nakashima H, Ikeda S, Shinohara K, Matsumoto S, Yoshida D, Nakashima R, et al.. Prognostic value of the low-density lipoprotein cholesterol/high-density lipoprotein cholesterol ratio for cardiovascular events in statin-treated type 2 diabetes with diabetic retinopathy without prior cardiovascular disease. Circ J 2026; 90: 492–501, doi:10.1253/circj.CJ-25-0841. [DOI] [PubMed] [Google Scholar]
- 21. Wadström BN, Wulff AB, Pedersen KM, Jensen GB, Nordestgaard BG.. Elevated remnant cholesterol increases the risk of peripheral artery disease, myocardial infarction, and ischaemic stroke: A cohort-based study. Eur Heart J 2022; 43: 3258–3269, doi:10.1093/eurheartj/ehab705. [DOI] [PubMed] [Google Scholar]
- 22. Chen G, Chen Y, Yao Y, Ding L, Wu S, Wu W.. High cumulative non-high-density lipoprotein-cholesterol concentration increases the risk of new-onset arterial stiffness: A prospective cohort study. Circ J 2025; 89: 629–637, doi:10.1253/circj.CJ-24-0921. [DOI] [PubMed] [Google Scholar]
- 23. Endo K, Tanaka M, Sato T, Inyaku M, Nakata K, Kawaharata W, et al.. High level of estimated small dense low-density lipoprotein cholesterol as an independent risk factor for the development of ischemic heart disease regardless of low-density lipoprotein cholesterol level: A 10-year cohort study. Circ J 2025; 89: 1182–1189, doi:10.1253/circj.CJ-24-0770. [DOI] [PubMed] [Google Scholar]
- 24. Maruo T, Ike A, Takamiya Y, Matsuoka Y, Shigemoto E, Kato Y, et al.. Impact of controlling serum low-density lipoprotein cholesterol and triglycerides on long-term clinical outcomes in diabetic patients who have undergone percutaneous coronary intervention. Circ Rep 2024; 6: 573–582, doi:10.1253/circrep.CR-24-0081. [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
Supplementary Table 1. Supplementary Table 2. Supplementary Figure.
Data Availability Statement
The data underlying this article will be shared upon reasonable request to the corresponding author.