Association of apolipoprotein C-III with atherosclerotic cardiovascular disease and the effectiveness of simulated interventions

Abstract

BACKGROUND

Despite effective control of low-density lipoprotein cholesterol, the residual risk of atherosclerotic cardiovascular disease (ASCVD) remains substantial, underscoring the urgent need to identify novel targets to further reduce ASCVD risk. This study aimed to investigate the association between apolipoprotein C-III (APOC3) levels and ASCVD, and further explore the potential impact of APOC3 intervention for reducing ASCVD risk.

METHODS

We leveraged the 20-year longitudinal data from the Chinese Multi-provincial Cohort Study–Proteomics Project, obtained totaling 5308 APOC3 measurements from 2550 participants, with up to four repeated measurements per participant. The intensity model was employed to estimate the association between time-varying APOC3 and ASCVD, accounting for time-varying potential confounders. The target trial emulation framework was used to emulate the effectiveness of various APOC3-targeted hypothetical interventions on ASCVD risk, with a guideline-directed low-density lipoprotein cholesterol-lowering strategy serving as a positive control.

RESULTS

Of 2550 participants aged 57.5 ± 7.9 years at baseline, and 56% of participants were females. The median level of APOC3 was 5554.8 (interquartile range: 3936.5-7913.8) Relative Fluorescence Units. The level of time-varying APOC3 level were associated with an increased risk of incident ASCVD, and a stronger association was observed in populations with metabolic abnormalities. Furthermore, more intensive reductions in APOC3 produced progressively larger decreases in ASCVD risk, with a maximum estimated reduction of 9.3% observed at a 90% APOC3 reduction.

CONCLUSIONS

Our findings suggest that APOC3 may play an important role in ASCVD incidence and that targeting APOC3 may offer additional cardiovascular benefits beyond conventional lipid-lowering strategies.

Atherosclerotic cardiovascular disease (ASCVD) remains the predominant cause of mortality and disability across the globe.^[1] Statin-based pharmacotherapy targeting cholesterol, in combination with lifestyle intervention, constitutes the cornerstone of current strategies for the prevention and treatment of atherosclerosis.^[2,3] However, more than 50% of residual cardiovascular risk remains despite the implementation of these contemporary therapies,^[4] underscoring the urgent need to identify novel therapeutic targets for preventing the onset and progression of ASCVD. Accumulating evidence indicates that, beyond the low-density lipoprotein cholesterol (LDL-C), triglyceride-rich lipoproteins (TRLs) and their remnant particles are major pathogenic drivers of atherosclerosis initiation and progression.^[5] Apolipoprotein C-III (APOC3) is a prominent constituent of TRLs and their remnants and serves as a key regulator of their metabolic processing and clearance.^[6,7] By inhibiting lipoprotein lipase (LPL) mediated TRL lipolysis and delaying hepatic clearance, elevated APOC3 levels lead to the accumulation of cholesterol rich remnant particles in plasma,^[8] and the deposition of these lipoproteins within the arterial wall together with their pro-inflammatory effect directly promotes the development and progression of atherosclerosis.^[9]

Converging epidemiological and human genetic evidence have supported the important role of APOC3 in ASCVD.^[10] Although previous studies found that elevated APOC3 levels are associated with an increased risk of ASCVD,^[11–16] existing studies on the association between APOC3 and ASCVD are primarily limited to baseline measurements of APOC3 and its association with long-term ASCVD outcomes or cross-sectional associations.^[10] The static nature of baseline exposure measurements may lead to erroneous estimation of the true relationship between exposure and long-term outcome, potentially resulting in biased estimates.^[17] Nevertheless, evidence incorporating repeated measurements of APOC3 over time remains limited, and the potential heterogeneity of the association between APOC3 level and ASCVD risk across different metabolic subgroups, such as individuals with hypertriglyceridemia, diabetes mellitus (DM), or hypertension, has yet to be clearly defined. Moreover, although APOC3-targeted nucleic acid–based therapies have demonstrated significant efficacy in lowering triglycerides (TGs) and remnant-related lipoproteins, most available randomized controlled trials (RCTs) have primarily focused on changes in TG levels as surrogate endpoints,^[18,19] without directly assessing the impact of APOC3-targeted intervention on ASCVD outcomes. The target trial emulation approach provides a framework for deriving causal inferences from observational data to mimic a RCT,^[20–22] offering a valuable approach to evaluate the long-term effects of APOC3 modulation on cardiovascular outcomes.

In this study, we leveraged the 20-year longitudinal data from the Chinese Multi-provincial Cohort Study–Proteomics Project (CMCS-PP), with up to four repeated APOC3 measurements per participant, to first estimate the time-varying association of the serum APOC3 protein levels with ASCVD risk, then investigate the potential heterogeneity of the association across different metabolic subgroups. Further, we emulate the effectiveness of various APOC3-targeted hypothetical interventions on ASCVD prevention by using the target trial emulation framework.

METHODS

Study Population

The present study was conducted as an extension of the Chinese Multi-provincial Cohort Study (CMCS), a community-dwelling, population-based prospective study initiated in 1992, the design of which has been described elsewhere.^[23,24] Briefly, a total of 21,953 healthy participants aged > 35 years were recruited during the baseline survey. Among these participants, up to 5966 individuals were invited to attend follow-up visits in 2002, 2007, 2012, and 2020–2023. A multistage sampling approach was applied to select 2621 individuals from the Beijing region, with the 2002 examination serving as the baseline for the establishment of the CMCS-PP. In the present study, a total of 2550 participants were included in the final analysis after excluding 71 individuals who experienced cardiovascular disease events before the 2002 visit. Among these participants, 1321 participants were from the CMCS-Beida subcohort and had up to four repeated proteomic measurements, whereas the remaining 1229 participants were from the CMCS-Shougang subcohort and had a single proteomic measurement at the 2002 baseline visit. All the participants were prospectively followed for the incidence of ASCVD through the end of 2023. The study protocol was approved by the Ethics Committee of Beijing Anzhen Hospital, Capital Medical University, Beijing, China (No.ks2019029), and was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants.

APOC3 Measurements

Serum APOC3 levels were measured using the SomaScan 11K^® assay. Briefly, 50 μL aliquots of serum were shipped on dry ice to Accuramed Inc. Laboratory (Shanghai, China), a facility certified by SomaLogic Operating Co., Inc. (Colorado, USA), for protein measurements. The SomaScan^® platform employs modified single-stranded DNA aptamers (SOMAmers) that selectively bind target proteins, which are subsequently quantified using DNA microarray technology. Protein abundances are reported as relative fluorescent units, as specified by the manufacturer. Quality control procedures were conducted in accordance with the technical specifications of the SomaScan^® assay. Based on internal quality control samples provided by the platform, the median intra-assay coefficient of variation was 4.56% for CMCS-Beida and 4.62% for CMCS-Shougang, indicating high intra-batch reproducibility. The inter-assay coefficient of variation between CMCS-Beida and CMCS-Shougang was 6.12%, reflecting acceptable inter-batch variability.

Risk Factor Measurements

In each survey, demographic information, lifestyles, and personal medical history was collected using a standardized questionnaire complied with the protocol of the World Health Organization-MONICA (Monitoring of Trends and Determinants in Cardiovascular Disease).^[25] Current smoking was defined as at least one cigarette per day. Alcohol consumption was measured by the frequency of drinking liquor, beer, and wine, and current drinking was defined as drinking at least once a week. Anthropometric measurements and blood pressure were obtained during standardized physical examinations, and the body mass index was calculated as weight in kilograms divided by the square of height in meters. Blood pressure was measured in the right arm in a sitting position using a mercury sphygmomanometer after at least 5 min of rest, and the mean value of two consecutive blood pressure measurements was used. Hypertension was defined by a systolic blood pressure ≥ 140 mmHg or a diastolic blood pressure ≥ 90 mmHg, or by the reported use of antihypertensive medication during the preceding two weeks. DM was defined as a fasting blood glucose level ≥ 7.0 mmol/L, or self-reported physician diagnosis. Hypercholesterolemia was defined as total cholesterol ≥ 6.2 mmol/L, or LDL-C ≥ 4.1 mmol/L, or self-reported use of cholesterol-lowering medications during the preceding two weeks. Hypertriglyceridemia was defined as TG ≥ 1.7 mmol/L, or self-reported use of cholesterol-lowering medications during the preceding two weeks. Remnant cholesterol (RC) was calculated as total cholesterol minus LDL-C minus high-density lipoprotein cholesterol (HDL-C).^[26] Obesity was defined as body mass index ≥ 28 kg/m². Venous blood samples were collected from the antecubital vein into lavender-top tubes containing ethylenediaminetetraacetic acid anticoagulant in the morning after at least 8 h of fasting, and the fasting blood glucose, total cholesterol, TG, HDL-C, and LDL-C levels were measured using standard enzymatic methods according to previous reports. The 10-year predicted ASCVD risk was estimated using the risk prediction model recommended by the Chinese guideline for the primary prevention of cardiovascular disease, which incorporates age, sex, smoking status, LDL-C, HDL-C, systolic blood pressure, and use of antihypertensive medication as predictor variables.^[2] This risk framework was further applied in a simulated positive control analysis to evaluate risk reduction across different strata of LDL-C lowering.

ASCVD Ascertainment

Participants were actively followed at intervals of 1–2 years for the occurrence of fatal or non-fatal acute coronary events, stroke events, and deaths from other causes, with additional case ascertainment through local disease surveillance systems. The primary outcome was ASCVD, defined as incident fatal or non-fatal acute coronary events and ischemic stroke, whereas hemorrhagic stroke and non-cardiovascular death were treated as competing events. Through December 31, 2023, 463 participants (18.2%) experienced ASCVD during a median follow-up of 20.3 years (interquartile range: 18.4-21.6 years), and 264 participants (10.4%) experienced competing events.

In the target trial emulation based on the CMCS-Beida cohort, 227 participants (17.2%) experienced ASCVD events, and 148 participants (11.2%) experienced competing events during follow-up.

Statistical Analysis

Data preparation

Continuous variables are presented as mean ± SD or medians (interquartile range), as appropriate according to their distribution, while categorical variables are summarized as counts (percentages). Serum APOC3 and apolipoprotein B (apoB) concentrations were normalized using a rank-based inverse normal transformation, and the differences in APOC3 levels between groups were evaluated using the Wilcoxon rank-sum test. To account for the potential time-varying effects of repeated APOC3 measurements, follow-up time was partitioned into 5-year intervals in a person-time data format, with APOC3 levels aligned to the corresponding re-examination year. The missing values were imputed using the last observation carried forward method.

Association of time-varying APOC3 level and ASCVD

We employed an intensity model to capture the effects of multiple repeated measurements,^[27] adjusting for age, age², time, time², and sex, to estimate the association between APOC3 levels and the risk of incident ASCVD. The APOC3-ASCVD association was quantified by the ratio of cumulative ASCVD risk (i.e., risk ratio) at the end of follow-up between a hypothetical intervention group with a per SD increase in time-dependent measured APOC3 over the study period and a control group with its natural levels. Sequential models were then constructed with stepwise adjustment for lipid-lowering medication use and established cardiovascular risk factors recommended by the 2020 Chinese guideline on the primary prevention of cardiovascular diseases, including smoking status, LDL-C, HDL-C, systolic blood pressure, use of antihypertensive medication, and DM status. Subgroup analyses by age (< 60 years and ≥ 60 years), sex (males and females), obesity status (yes and no), hypertriglyceridemia status (yes and no), hypercholesterolemia status (yes and no), DM status (yes and no), and hypertension status (yes and no) at baseline are also conducted separately. To mitigate selection bias, we applied multiplied inverse probability weights accounting for survival from the 1992-1993 baseline to the 2002 re-examination and for remaining uncensored and alive throughout follow-up. Specifically, we first calculated survival probabilities at each follow-up time point and constructed time-varying inverse probability weights. Then, we estimated the probability of being selected for serum proteomic measurement in 2002 based on baseline covariates (age, sex, blood pressure, fasting glucose, lipid levels, smoking status, and medication use), and truncated weights at the 1^st and 99^th percentiles to reduce the impact of extreme values. Multiplying survival and selection weights generated a combined weight, which was then applied to adjust the study sample.

In addition, to assess the relationships between APOC3 and other lipid parameters, partial correlation analyses were performed, adjusting for age, sex, and the use of lipid-lowering medications. Correlation coefficients (r) were computed using the Spearman’s rank method.

Emulated the effectiveness of hypothetical APOC3 interventions on ASCVD

The effectiveness of APOC3-targeted interventions for ASCVD prevention was evaluated using a series of emulated RCTs under the target trial emulation framework using 20-year follow-up data from CMCS-Beida subcohort. The protocols are provided in Table 1.

Table 1. Specification and emulation of a target trial of APOC3 and treat-to-target LDL-C interventions to prevent atherosclerotic cardiovascular disease.

Protocol component	Target trial specification	Target trial emulation
APOC3: apolipoprotein C-III; LDL-C: low-density lipoprotein cholesterol.
Eligibility criteria	No history of cardiovascular disease at baseline In-person examination at the baseline	Same as for the target trial
Treatment strategies	Natural course of no intervention APOC3 interventions were defined as sustained percentage reductions from the observed level, ranging from 10% to 90% Treat-to-target LDL-C lowering intervention based on cholesterol-lowering targets recommended by the 2020 Chinese guideline on the primary prevention of cardiovascular diseases	Same as for the target trial
Treatment assignment	Eligible participants are randomly assigned to the corresponding intervention strategies at the baseline visit in 2002 or during the follow-up visit	We implemented the hypothetical interventions by discretizing the follow-up times into a 5-year interval and initiating the intervention for those eligible participants
Outcomes	Atherosclerotic cardiovascular disease	Same as for the target trial
Follow-up	Follow-up period starts at the baseline and ends at the year of recording atherosclerotic cardiovascular disease, deaths, non-cardiovascular deaths, loss to follow-up, 20 years after baseline, or administrative end of follow-up on 31 December 2023, whichever comes first	Same as for the target trial We defined the start of follow-up period (i.e., time zero) as the initiating time of the intervention
Causal contrast	Per-protocol effect	Observational analog of per-protocol effect
Statistical analysis	Per-protocol analysis of protein-lowering effect under various strategies	Same per-protocol analysis with sequential emulation and adjustment for covariates

Eligibility criteria　We included 1321 eligible participants who were free of ASCVD at the baseline in CMCS-Beida subcohort, as previously stated. Characteristics of participants are summarized in Table 2.

Table 2. Characteristics of study participants.

Variable	2002 (n = 2550)	2007 (n = 1021)	2012 (n = 1015)	2020-2023 (n = 722)
Data are presented as means ± SD or n (%). *Presented as median (interquartile range). APOC3: apolipoprotein C-III; CMCS: Chinese Multi-provincial Cohort Study; CMCS-PP: Chinese Multi-provincial Cohort Study–Proteomics Project; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; RC: remnant cholesterol.
Age at measurement, yrs	57.5 ± 7.9	65.2 ± 7.8	69.6 ± 7.8	76.0 ± 7.5
Males	1123 (44.0%)	466 (45.6%)	453 (44.6%)	316 (43.8%)
Smoking	324 (12.7%)	80 (7.8%)	58 (5.7%)	28 (4.2%)
Drinking	772 (30.3%)	104 (10.2%)	59 (5.8%)	43 (6.0%)
Body mass index, kg/m2	25.4 ± 3.3	24.7 ± 3.2	24.4 ± 3.3	24.8 ± 3.5
Systolic blood pressure, mmHg	131.4 ± 19.1	137.5 ± 17.4	138.4 ± 15.7	145.8 ± 17.8
LDL-C, mmol/L	3.27 ± 0.82	3.31 ± 0.88	2.97 ± 0.90	2.21 ± 0.67
HDL-C, mmol/L	1.37 ± 0.31	1.33 ± 0.26	1.34 ± 0.30	1.46 ± 0.33
Total cholesterol, mmol/L	5.38 ± 1.00	5.36 ± 1.00	5.09 ± 1.07	4.79 ± 1.08
Triglyceride, mmol/L	1.34 (0.95–1.94)*	1.55 (1.15–2.14)*	1.35 (0.95–1.86)*	1.28 (0.96–1.76)*
RC, mmol/L	0.67 (0.41–0.96)*	0.64 (0.44–0.87)*	0.71 (0.58–0.89)*	1.10 (0.85–1.36)*
Apolipoprotein B, Relative Fluorescence Units	3232.3 (1855.4–8366.7)*	3080.5 (1913.0–6056.8)*	1448.5 (1143.9–1872.0)*	1948.2 (1456.3–2688.8)*
Fasting blood glucose, mmol/L	5.1 ± 1.5	5.91 ± 1.04	5.71 ± 1.25	6.25 ± 1.50
Hypertension	1199 (47.0%)	660 (64.7%)	697 (68.8%)	567 (79.6%)
Diabetes mellitus	296 (11.6%)	170 (16.7%)	205 (20.2%)	188 (26.4%)
Hypercholesterolemia	563 (22.1%)	298 (29.2%)	386 (38.0%)	391 (54.8%)
Hypertriglyceridemia	937 (36.7%)	487 (47.7%)	522 (51.4%)	450 (62.3%)
Antihypertension drugs	621 (24.4%)	468 (45.8%)	528 (52.0%)	389 (54.0%)
Lipid-lowering drugs	180 (7.1%)	148 (14.5%)	312 (30.7%)	347 (48.2%)
Antidiabetic drugs	118 (4.6%)	110 (10.8%)	158 (15.6%)	147 (20.4%)
Any use of drug	858 (33.7%)	554 (54.3%)	659 (64.9%)	514 (71.4%)
CMCS-PP subcohort
CMCS-Beida	1321 (51.8%)	1021 (100.0%)	1015 (100.0%)	722 (100.0%)
CMCS-Shougang	1229 (48.2%)	0	0	0
APOC3, Relative Fluorescence Units	5554.8 (3936.5–7913.8)*	5516.9 (4133.2–7654.7)*	4365.2 (3273.0–5854.5)*	4478.7 (3332.7–5967.3)*

Intervention strategies　Eligible participants were assigned to each of 10 hypothetical intervention strategies targeting APOC3 reduction, with relative reductions ranging from 10% to 90% in 10% increments, or to a natural course without intervention. In addition, treat-to-target LDL-C lowering strategies were evaluated as a positive control, in which interventions were initiated and titrated according to risk to achieve guideline-recommended LDL-C targets specified by the 2020 Chinese guideline on the primary prevention of cardiovascular diseases, as described previously.^[2] All interventions were initiated at the time risk-based criteria were met (time zero) and were maintained throughout follow-up until LDL-C targets were achieved. In contrast, APOC3 interventions were defined as sustained relative reductions applied uniformly to APOC3 levels at each follow-up time point throughout the observation period.

Intervention assignments　To emulate randomization, eligible CMCS- Beida participants were cloned and assigned at baseline (time zero) to hypothetical APOC3 intervention strategies (relative reductions ranging from 10% to 90%) or to the natural course without intervention. This approach approximates randomized treatment allocation while retaining the structure of the observed data.

Outcome　ASCVD, as stated in “ASCVD ascertainment”.

Follow-up　Eligible participants were followed from the 2002 visit until the first occurrence of ASCVD, a competing event (as mentioned above), loss to follow-up, 20 years of follow-up after the 2002 visit, or administrative censoring on December 31, 2023, whichever occurred first.

Statistical analysis　Targeted maximum likelihood estimation (TMLE) was used to quantify the effects of the hypothetical intervention strategies targeting APOC3 on the risk of ASCVD.^[28,29] To emulate randomization, eligible CMCS-Beida participants were cloned into 10 copies and assigned to predefined intervention strategies at time zero. Both the treatment and outcome processes were modeled by incorporating time-fixed and time-varying confounders using a Super Learner ensemble, which included a generalized linear model, an intercept-only model, and a random survival forest. Time-fixed confounders included baseline age and sex. Time-varying covariates comprised smoking status, systolic blood pressure, LDL-C, HDL-C, DM status, use of antihypertensive and lipid-lowering medications, and the contemporaneous 10-year ASCVD risk category within the sampling framework. Missing data were handled using last observation carried forward to impute. Stabilized inverse probability weights, as described above, were additionally applied to account for selection bias. The participants who experienced the competing events were treated as censored in the TMLE analyses. Five-fold cross-validation was implemented within the TMLE framework to reduce the risk of overfitting.

All statistical analyses were performed using R statistical software 4.4.1 (R Foundation for Statistical Computing, Boston, MA, USA). All tests were two-sided, and the P-value < 0.05 was considered statistically significant.

RESULTS

Characteristics of Study Participants

The study included 2550 participants, contributing a total of 5308 APOC3 measurements, with up to four repeated measurements per participant: 516 participants completed four measurements, 468 participants completed three measurements, 274 participants completed two measurements, and 1292 participants completed one measurement. Table 2 shows the characteristics of eligible CMCS-PP participants across four surveys. Of 2550 participants aged 57.5 ± 7.9 years at baseline, and 56% of participants were females. The median level of APOC3 was 5554.8 (interquartile range: 3936.5-7913.8) Relative Fluorescence Units, ranging from 910.0 to 60,403.0 Relative Fluorescence Units at baseline. Across the study period, upward trends were noted in the prevalence of hypertension, hypercholesterolemia, hypertriglyceridemia, and DM, as well as in the proportion of participants receiving antihypertensive, lipid-lowering, and antidiabetic therapies. As shown in Figure 1, the levels of APOC3 distributions were generally comparable across age (< 60 years vs. ≥ 60 years) and smoking status, drinking status, and obesity status, whereas higher APOC3 levels were observed in males than in females. In addition, APOC3 distributions were shifted toward higher values among participants with hypertriglyceridemia, hypercholesterolemia, hypertension, or DM (all of P < 0.05). In addition, as shown in supplemental material, Figure 1S, the APOC3 levels are moderately to highly correlated with TG (r = 0.63) and RC (r = 0.50), while the correlation with apoB is weaker (r = 0.31).

Figure 1.

Distribution of circulating APOC3 levels across subgroups by age, sex, lifestyles, and metabolic conditions.

The median and interquartile range of APOC3 are presented on the log₂-transformed scale. Violin plots display the distribution of values, with embedded boxplots indicating the median and interquartile range. P-values were calculated using the Wilcoxon rank-sum test. APOC3: apolipoprotein C-III.

Association Between Time-varying APOC3 Levels and ASCVD Risk

Using an intensity model incorporating repeated measurements of APOC3, per-standard deviation higher APOC3 was positively associated with a 37% increased ASCVD risk [cumulative risk ratio (RR) = 1.37, 95% CI: 1.25–1.50] (Figure 2). This association remained consistent after further adjustment for lipid-lowering medication use (RR = 1.38, 95% CI: 1.26–1.52). In the fully adjusted model, elevated APOC3 levels remained positively associated with incident ASCVD (RR = 1.31, 95% CI: 1.19–1.44).

Figure 2.

Association of apolipoprotein C-III with atherosclerotic cardiovascular disease.

Model 1: adjusted for age, sex, and cohort effects. Model 2: adjusted for variables in Model 1 plus the lipid-lowering medication. Model 3: adjusted for variables in Model 2 plus the smoking status, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, systolic blood pressure, and use of antihypertensive medication.

Subgroup analyses demonstrated statistically significant effect modification in the association between time-varying APOC3 levels and ASCVD risk by age, sex, DM status, hypertension status, and hypertriglyceridemia status (Figure 2). Specifically, stronger associations were observed among participants aged ≥ 60 years compared with those aged < 60 years (RR = 1.49, 95% CI: 1.30–1.72 vs. RR = 1.12, 95% CI: 0.97–1.28, P_interaction = 0.01), and among females compared with males (RR = 1.39, 95% CI: 1.23–1.57 vs. RR = 1.06, 95% CI: 0.89–1.25, P_interaction = 0.027). The association between APOC3 levels and ASCVD risk was stronger in participants with DM, hypertension, or hypertriglyceridemia than in those without these conditions, with statistically significant interactions observed (Figure 2). In contrast, no statistically significant interaction was observed between APOC3 levels and hypercholesterolemia status in relation to ASCVD risk (P_interaction = 0.221).

Emulated the Effectiveness of Hypothetical APOC3 Interventions on ASCVD

We estimated the effectiveness of APOC3-targeted intervention via target trail emulation framework. Figure 3 shows the observational analog of the per-protocol effects of hypothetically lowering APOC3 level on the cumulative risk of ASCVD over the study period. Compared with the natural course of no intervention (cumulative risk = 16.3%, 95% CI: 11.0%–21.6%), lowering APOC3 level was showed a substantial reduction in ASCVD risk. When the level APOC3 was reduced by 60%, the absolute ASCVD risk was 14.1% (95% CI: 4.7%–23.4%, absolute risk reduction = −2.3%). Furthermore, a more pronounced reduction in APOC3 levels (90%) corresponded to up to a 9.3% reduction in ASCVD risk.

Figure 3.

Estimated atherosclerotic cardiovascular disease absolute risk reduction under feasible interventions: guideline-directed LDL-C lowering and APOC3 reduction.

Absolute risk reduction (%) across intervention strategies with a reduction in APOC3 protein ranging from 10% to 90% (increments of 10%). APOC3: apolipoprotein C-III; LDL-C: low-density lipoprotein cholesterol.

DISCUSSION

In this 20-year longitudinal cohort with up to four repeated serum APOC3 measurements, we found that the level of APOC3 was consistently associated with an increased risk of incident ASCVD. Subgroup analyses further identified stronger association between APOC3 and ASCVD in participants with metabolic abnormalities, such as DM, hypertension, and hypertriglyceridemia. Furthermore, more intensive reductions in APOC3 produced progressively larger decreases in ASCVD risk, with a maximum estimated reduction of 9.3% observed at a 90% APOC3 reduction.

To the best of our knowledge, this is the first study to evaluate the association between time-varying APOC3 levels and ASCVD risk, and emulate the potential effectiveness of APOC3-targeted interventions on ASCVD risk using a target trial emulation framework. By leveraging multiple APOC3 measurements, we demonstrated that higher APOC3 levels were associated with an increased risk of ASCVD, after accounting for the time-varying effects of APOC3 and potential covariates during long-term follow-up. Our results are consistent with previous epidemiological and human genetic studies implicating APOC3 positively associated with ASCVD risk.^[10] Compared with these previous studies, our analysis advances the field by incorporating repeated proteomic measurements and accounting for time-varying exposure, thereby reducing regression dilution bias and better providing a more accurate representation of long-term APOC3 exposure. Furthermore, we also found that APOC3 levels are moderately to highly correlated with TG and RC, further supporting the feasibility of APOC3 inhibition to lower TG and RC levels and thereby reduce ASCVD risk. In contrast, the correlation between APOC3 and apoB was relatively weak, consistent with prior studies showing that common APOC3 variants associated with modest increases in circulating APOC3 were strongly linked to elevated TRLs but not with apoB.^[30]

Notably, we observed a significant modification effect by metabolic status, including DM, hypertension, and hypertriglyceridemia, indicating heterogeneity in the strength of the association between APOC3 levels and ASCVD risk across these conditions. Stronger associations among individuals with DM, hypertension, or hypertriglyceridemia might be attributed to increased accumulation of remnant lipoproteins and vascular injury caused by APOC3 in these metabolic disorders. In hypertriglyceridemia, APOC3 inhibits LPL activity and hepatic clearance of TRL remnants, increasing RC exposure and further exacerbating atherosclerosis risk,^[31,32] thereby strengthening the association between APOC3 and ASCVD in this population. As for DM, Kanter, et al.^[8] found that baseline APOC3 levels associate with cardiovascular risk more strongly in individuals with type 2 DM than in those without DM, which is consistent with us. Moreover, in the hypertensive population, endothelial dysfunction and vascular inflammation may increase arterial susceptibility to APOC3–associated TRL remnants,^[31] thereby amplifying the translation of RC exposure into the process of atherosclerosis. Therefore, in populations with metabolic abnormalities, particular attention should be given to the ASCVD risk mediated by APOC3, highlighting the need for targeted interventions in these high-risk populations. Moreover, our findings indicate that pharmaceutical interventions lowering APOC3 might be more effective in patients with these metabolic disorders.

Although RCTs have clearly demonstrated the efficacy of APOC3-targeted interventions in lowering TG levels, their effects on ASCVD outcomes remain to be established. No ASCVD prevention trials targeting APOC3 are ongoing to date, but a phase 3 clinical trial has been announced.^[33] In this study, our emulated target trial demonstrated that targeting APOC3 with an intervention could effectively reduce the risk of ASCVD. Furthermore, a more pronounced reduction in APOC3 levels corresponded to up to a 9.3% reduction in ASCVD risk. A recent study based on a large-scale European population lowered baseline RC to mimic the effect of APOC3 inhibitors on reducing APOC3 levels and assessed the impact on ASCVD risk. The study also found that an 80% reduction in RC could lower 10-year ASCVD risk by up to 4%.^[34] Additionally, several drugs targeting APOC3, such as olezarsen and volanesorsen, have been developed and shown in published RCTs to reduce APOC3 levels by 60%–80%.^{[18,19,35,36]} These reductions may lead to a 2%–6% decrease in ASCVD risk. Moreover, no major adverse events were observed, further supporting the feasibility of targeting APOC3 as an intervention for ASCVD.

Mechanistic insights into how TG-lowering interventions translate into cardiovascular benefit can provide context for these findings. APOC3 inhibitors lower TGs primarily by enhancing the clearance of TRLs through LPL-dependent and independent pathways, which confers a more comprehensive TG-lowering effect. Angiopoietin-like protein 3 inhibitors are also a well-characterized class of TG-lowering agents that reduce TG levels and modulate cardiovascular risk primarily by inhibiting LPL and endothelial lipase, thereby enhancing the catabolism of TRLs and affecting multiple lipid fractions including LDL-C and HDL-C. Fibrates are a class of PPAR-α agonists that lower TG and influence ASCVD risk by promoting LPL-mediated TG hydrolysis, increasing fatty acid oxidation, and reducing hepatic TG synthesis, while broadly regulating lipid metabolism. The mechanisms of these interventions differ, with APOC3 inhibition being highly TRL-specific, angiopoietin-like protein 3 inhibition broader across lipoprotein classes, and fibrates acting pleiotropically.^[37–40] The TG-lowering effects of these drugs through distinct mechanisms may help guide the selection of targeted therapies and support individualized cardiovascular risk management.

STRENGTHS AND LIMITATIONS

Several strengths of the current study are worth noting. To our knowledge, this is the first study to examine the association between time-varying APOC3 levels and ASCVD risk by leveraging a longitudinal cohort with up to four repeated APOC3 measurements over a 20-year follow-up. This analysis captures the longitudinal association between APOC3 levels and ASCVD risk beyond baseline measurements. In addition, we leveraged ensemble machine learning–based causal inference methods to support the causal role of APOC3 in ASCVD and to emulate the potential effects of APOC3-lowering interventions on ASCVD risk. Despite these advantages, several possible study limitations in this study should be pointed out. Although advanced causal inference approaches and target trial emulation were applied, residual confounding inherent to observational studies cannot be fully excluded. In addition, the estimates from the target trial emulation did not reach statistical significance, likely due to the limited sample size. However, the risk reduction associated with LDL-C lowering, used as a positive control, was consistent with prior studies, showing a reduction of approximately 2%–3%,^[41,42] supporting the validity of the emulated framework. Furthermore, it should be acknowledged that the findings were based on simulation models, and their validity needs to be further confirmed in RCTs. Last but not least, as the APOC3 measurements are not absolute quantitative values, future studies with absolute quantification are warranted to enable clinically interpretable findings.

CONCLUSIONS

In summary, our findings suggest that time-varying APOC3 levels are associated with ASCVD risk, with a stronger association observed in populations with metabolic abnormalities. The simulation of APOC3 interventions indicates a potential trend towards risk reduction, further supporting that targeting APOC3 could be effective in reducing ASCVD risk.

SUPPLEMENTARY DATA

Supplementary data to this article can be found online.