Abstract
Background
Atherogenic lipid-related risk may not be fully characterized by a single lipid measure. A dual-axis framework combining apolipoprotein B (ApoB), a measure of atherogenic particle number, with the LDL-C to total cholesterol ratio (LDL-C/TC), a derived measure of cholesterol partitioning, has been proposed to characterize lipid-related risk heterogeneity. The position of remnant cholesterol (RC) within this framework remains uncertain.
Methods
We analyzed 288,257 UK Biobank participants free of cardiovascular disease and lipid-lowering therapy. Participants were classified using dual-axis framework. Associations with incident atherosclerotic cardiovascular disease (ASCVD) were assessed using Cox models. RC was evaluated in combination with these axes across strata, including SCORE2-defined low-to-intermediate risk populations.
Results
Over a median follow-up of 13.9 years, the dual-axis framework differentiated ASCVD risk, with the dual-elevated phenotype (ApoB ≥ 90 mg/dL and LDL-C/TC ≥ 0.60) showing the highest risk (adjusted hazard ratio 1.23; 95% CI, 1.17–1.29). RC did not modify associations across dual-axis categories (P for interaction = 0.44), and higher RC levels were associated with attenuation of risk gradients across categories. In low-to-intermediate SCORE2 risk individuals, increasing numbers of elevated markers (ApoB, LDL-C/TC ratio, and RC) were associated with stepwise increases in risk, with absolute risk differences reaching 1.49% at 10 years for individuals with all three markers elevated versus those with none elevated. ApoB and the LDL-C/TC ratio produced small C-index increases (ΔC, 0.0013 and 0.0012), whereas RC produced no further improvement (ΔC, − 0.0001; P = 0.364).
Conclusions
RC did not materially modify associations across ApoB–LDL-C/TC phenotypes but identified absolute-risk heterogeneity within ApoB strata. Co-occurring marker elevations characterized higher-risk lipid phenotypes, although gains in discrimination were small and require external validation.
Graphical Abstract
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1186/s12944-026-03023-z.
Keywords: Apolipoprotein B, LDL-C/total cholesterol ratio, SCORE2, Atherosclerotic cardiovascular disease, Remnant cholesterol
Introduction
Atherosclerotic cardiovascular disease (ASCVD) develops as a consequence of prolonged exposure of the arterial wall to circulating atherogenic lipoproteins, rather than from a single cross-sectional lipid measurement [1, 2]. Despite this temporal nature, clinical risk assessment continues to rely predominantly on low-density lipoprotein cholesterol (LDL-C), which reflects the cholesterol mass within LDL particles at a given time point. Although LDL-C remains central to risk prediction and therapeutic decision-making [1, 2], substantial heterogeneity in ASCVD risk persists among individuals with similar LDL-C levels [3, 4]. This discordance reflects the fact that LDL-C provides an incomplete representation of atherogenic burden, without accounting for variation in lipoprotein particle number, lipid composition, or cholesterol distribution across lipoprotein fractions [3, 5]. Consequently, individuals with comparable LDL-C concentrations may differ markedly in their underlying lipoprotein profiles and cumulative atherogenic exposure, underscoring the need for approaches that more fully capture the multidimensional nature of lipid-related risk.
To better explain the observed heterogeneity in ASCVD risk, a structured framework [6] is needed to separate the different components of lipid-related risk that are embedded within LDL-C. A dual-axis lipid stratification approach has been developed to distinguish between atherogenic particle burden and lipid composition. Within this framework, apolipoprotein B (ApoB) provides a direct measure of the total number of atherogenic lipoprotein particles [7, 8], whereas the ratio of LDL-C to total cholesterol (LDL-C/TC) captures how cholesterol is distributed across lipoprotein fractions [6]. By jointly evaluating ApoB-defined particle burden and a derived measure of cholesterol partitioning, this approach may help characterize heterogeneity in lipid phenotypes that is not apparent from LDL-C alone.
Remnant cholesterol (RC), defined as the cholesterol carried within triglyceride-rich lipoproteins, has emerged as a clinically relevant marker of atherogenic risk beyond LDL-C alone [9, 10]. In current literature, RC generally refers to the cholesterol content of very-low-density lipoproteins, intermediate-density lipoproteins, and chylomicron remnants, although its definition and measurement are not entirely uniform across studies [11]. Higher RC levels have been associated with increased risks of cardiovascular events, stroke, and mortality, while genetic studies support an atherogenic role of triglyceride-rich lipoprotein pathways [11–15]. At the same time, RC is not a distinct lipoprotein class but a cholesterol measure derived from particles that are closely linked to ApoB-containing lipoprotein metabolism, and calculated and directly measured RC do not necessarily identify the same individuals [11]. Together, these features suggest that RC may capture risk information that is clinically relevant yet not fully distinct from broader measures of atherogenic particle burden and lipid composition, underscoring the importance of clarifying how RC fits within a multidimensional lipid stratification framework rather than evaluating it in isolation.
Clarifying the role of RC within a structured framework of lipid-related risk remains an important challenge. Contemporary risk prediction models, including Systematic Coronary Risk Evaluation 2 (SCORE2) [16], rely largely on conventional lipid measures and do not differentiate between atherogenic particle burden, lipid composition, and triglyceride-rich lipoproteins, potentially obscuring meaningful differences in underlying lipoprotein profiles. However, the position of RC within the risk gradients defined by ApoB and cholesterol distribution, and whether it improves risk stratification, remain uncertain, particularly among individuals at low-to-intermediate risk as defined by SCORE2.
In this study, we evaluated RC as a potential modifier within a dual-axis lipid framework defined by ApoB and the LDL-C/TC ratio in the general population, and assessed whether its integration with these dimensions refines risk stratification among individuals at low-to-intermediate risk according to SCORE2.
Methods
Study design and population
This study was conducted using the UK Biobank Resource under approved Application No. 531547. Within the UK Biobank cohort [17], participants were included if they were free of clinically recognized cardiovascular disease and not receiving lipid-lowering therapy at baseline. These criteria were applied to focus on a primary prevention population and to reduce the influence of treatment on lipid measurements. Cardiovascular disease at baseline was defined using standardized coding algorithms encompassing coronary heart disease, cerebrovascular disease, transient ischaemic attack, peripheral arterial disease, and prior coronary revascularization (Supplementary Table 1). Individuals with missing data on lipid variables, covariates, or outcome ascertainment were excluded. Among 368,336 cardiovascular disease–free participants not receiving lipid-lowering therapy, 80,079 individuals (21.7%) were excluded because of incomplete data on lipid exposures, covariates, or outcome ascertainment, leaving a final analytic sample of 288,257 participants (Supplementary Fig. 1).
Outcome definition
The primary endpoint was incident ASCVD, defined as the first occurrence of fatal or nonfatal myocardial infarction or stroke during follow-up. Outcome events were identified through linkage to national hospitalization and mortality records. Diagnoses were based on codes from the International Classification of Diseases, Tenth Revision (ICD-10), including I21–I22 for myocardial infarction and I61–I63 for stroke.
Biomarker assessment
Baseline lipid biomarkers were measured from non-fasting venous blood samples using standardized biochemical assays. Direct LDL-C concentrations were determined using a homogeneous enzymatic method. In parallel, LDL-C levels were estimated using established equations, including the Friedewald [18], Martin/Hopkins [19], and Sampson formulas [20], to enable comparisons across analytical approaches. For individuals with triglyceride levels ≥ 400 mg/dL, directly measured LDL-C values were used; otherwise, LDL-C was derived from the corresponding estimation methods. RC was defined as total cholesterol minus high-density lipoprotein cholesterol (HDL-C) minus LDL-C [10]. To maintain internal consistency, the LDL-C/TC ratio and RC were calculated using the same LDL-C definition within each analytical framework.
The primary exposure was a prespecified dual-axis framework defined by ApoB (< 90 vs ≥ 90 mg/dL) and the LDL-C/TC ratio (< 0.60 vs ≥ 0.60) [6]. To further characterize how RC relates to these established lipid dimensions, exploratory combination analyses were conducted for ApoB with RC and for the LDL-C/TC ratio with RC. For these analyses, RC was evaluated using two prespecified cutoffs: 35 mg/dL (0.9 mmol/L), consistent with guideline-recommended thresholds for nonfasting lipid profiles [21], and 39 mg/dL (1.0 mmol/L), based on prior literature [9].
SCORE2-based risk stratification
To contextualize lipid-related risk within a clinically relevant framework, analyses were conducted using the SCORE2 cardiovascular risk model [16]. Participants were restricted to those eligible for SCORE2, including individuals aged 40–69 years without established ASCVD or major conditions that would preclude its use, such as diabetes mellitus, chronic kidney disease, or genetic lipid disorders [2]. Chronic kidney disease was identified using algorithmically defined end-stage renal disease outcomes in the UK Biobank. To further align with SCORE2 applicability, individuals with probable familial hypercholesterolaemia were excluded using a surrogate definition of directly measured LDL-C ≥ 4.9 mmol/L (≥ 190 mg/dL). Participants were categorized according to estimated 10-year SCORE2 risk. High risk was defined as ≥ 10% [2], and individuals classified as very high risk were combined with this group, resulting in two strata: low-to-intermediate risk and high risk. These strata were used as the clinical framework to evaluate whether integrating RC with the dual-axis lipid framework could further refine risk stratification among individuals at low-to-intermediate risk.
Statistical analysis
Continuous variables are summarized using medians and interquartile ranges, whereas categorical variables are reported as counts and proportions. Between-group differences in baseline characteristics were evaluated using the Wilcoxon rank-sum test for continuous variables and the Pearson χ2 test for categorical variables. Time-to-event associations between lipid-related exposures and incident ASCVD were analyzed using Cox proportional hazards models, with age as the time scale. Models were adjusted for key cardiovascular risk factors, including sex, ethnicity, smoking status, diabetes, body mass index (BMI), systolic blood pressure (SBP), and antihypertensive treatment. The proportional hazards assumption was verified based on Schoenfeld residuals.
Absolute risk was quantified under a competing-risk framework. Cumulative incidence functions for ASCVD were estimated with the Aalen–Johansen approach [22], accounting for non-ASCVD death as a competing event, and group differences were assessed using Gray’s test. Absolute risk differences at prespecified time horizons were derived from cumulative incidence estimates, and the number of individuals per excess ASCVD event was calculated as the reciprocal of these differences [23].
The primary exposure was defined using joint categories of ApoB and the LDL-C/TC ratio, modeled as categorical variables. Potential effect modification by RC was examined using prespecified binary thresholds (< 35 vs ≥ 35 mg/dL and < 39 vs ≥ 39 mg/dL). Robustness of the findings was evaluated through sensitivity analyses that repeated models across different LDL-C derivation methods, including directly measured and equation-based estimates. Additional adjustment for log-transformed triglyceride levels was performed to address potential effects related to non-fasting conditions.
Among participants classified as having low-to-intermediate SCORE2 risk, incremental discrimination was evaluated using nested Cox proportional hazards models. The base model included SCORE2 component variables, including sex, total cholesterol, HDL-C, SBP, and current smoking, with age used as the underlying time scale. Standardized ApoB, LDL-C/TC ratio, and RC were added sequentially. Harrell’s C-index was calculated for each model, and changes in C-index compared each model with the immediately preceding model. The corresponding 95% confidence intervals and P values were estimated from 1,000 bootstrap resamples. All statistical analyses were conducted in R (version 4.3.2). A two-sided P value < 0.05 was considered statistically significant.
Results
Baseline characteristics
Among 288,257 participants, those with higher remnant cholesterol (RC ≥ 35 mg/dL) exhibited a distinct cardiometabolic risk profile compared with those with lower RC levels (Table 1). Individuals in the higher RC group were slightly older and more frequently male, and had higher BMI and blood pressure, consistent with a metabolically adverse phenotype. Marked differences were observed across lipid-related measures. Participants with elevated RC had substantially higher ApoB (median, 126.9 vs 97.7 mg/dL) and triglyceride levels (241.3 vs 115.1 mg/dL), alongside lower HDL-C and higher non–high-density lipoprotein cholesterol (non–HDL-C) concentrations. Nearly all individuals with elevated RC had non–HDL-C ≥ 130 mg/dL, indicating that elevated RC occurs almost exclusively within states of increased atherogenic cholesterol burden.
Table 1.
Baseline characteristics of participants by remnant cholesterol levels (direct LDL-C–Based) in the UK Biobank
| Characteristics | RC < 35 mg/dL (n = 231,657)a | RC ≥ 35 mg/dL (n = 56,600)a | Total (n = 288,257) | P value |
|---|---|---|---|---|
| Age, y | 57.0 (49.0 to 63.0) | 58.0 (51.0 to 63.0) | 57.0 (50.0 to 63.0) | < 0.001 |
| Age | < 0.001 | |||
| < 50 | 58,309 (25.2) | 11,874 (21.0) | 70,183 (24.3) | |
| 50–54 | 35,839 (15.5) | 8971 (15.8) | 44,810 (15.5) | |
| ≥ 55 | 137,509 (59.4) | 35,755 (63.2) | 173,264 (60.1) | |
| Sex | < 0.001 | |||
| Female | 133,174 (57.5) | 25,690 (45.4) | 158,864 (55.1) | |
| Male | 98,483 (42.5) | 30,910 (54.6) | 129,393 (44.9) | |
| Ethnicity | < 0.001 | |||
| White | 218,418 (94.3) | 54,206 (95.8) | 272,624 (94.6) | |
| Black | 4434 (1.9) | 357 (0.6) | 4791 (1.7) | |
| Asian | 5136 (2.2) | 1292 (2.3) | 6428 (2.2) | |
| Other | 3669 (1.6) | 745 (1.3) | 4414 (1.5) | |
| Smoking | 0.569 | |||
| Never | 132,239 (57.1) | 32,289 (57.0) | 164,528 (57.1) | |
| Previous | 75,273 (32.5) | 18,485 (32.7) | 93,758 (32.5) | |
| Current | 24,145 (10.4) | 5826 (10.3) | 29,971 (10.4) | |
| Diabetes | 3206 (1.4) | 752 (1.3) | 3958 (1.4) | 0.320 |
| BMI, kg/m2 | 26.2 (23.7 to 29.3) | 28.2 (25.8 to 31.1) | 26.6 (24.1 to 29.8) | < 0.001 |
| BMI | < 0.001 | |||
| < 30 | 182,532 (78.8) | 37,762 (66.7) | 220,294 (76.4) | |
| ≥ 30 | 49,125 (21.2) | 18,838 (33.3) | 67,963 (23.6) | |
| SBP, mm Hg | 137.0 (125.0 to 151.0) | 142.0 (130.0 to 155.0) | 138.0 (126.0 to 152.0) | < 0.001 |
| DBP, mm Hg | 81.0 (74.0 to 89.0) | 85.0 (78.0 to 92.0) | 82.0 (75.0 to 89.0) | < 0.001 |
| Antihypertensive medication | 23,355 (10.1) | 5661 (10.0) | 29,016 (10.1) | 0.576 |
| ApoB, mg/dL | 97.7 (84.3 to 111.5) | 126.9 (113.3 to 142.0) | 102.5 (87.3 to 118.8) | < 0.001 |
| Total cholesterol, mg/dL | 211.8 (186.7 to 237.2) | 261.8 (237.0 to 288.2) | 220.5 (192.8 to 249.7) | < 0.001 |
| Triglycerides, mg/dL | 115.1 (85.3 to 155.9) | 241.3 (186.3 to 313.8) | 130.6 (92.1 to 188.8) | < 0.001 |
| HDL-C, mg/dL | 56.3 (47.2 to 67.0) | 48.3 (41.5 to 56.6) | 54.6 (45.8 to 65.2) | < 0.001 |
| Low HDL-Cb | 41,194 (17.8) | 18,539 (32.8) | 59,733 (20.7) | < 0.001 |
| Non–HDL-C, mg/dL | 154.0 (130.9 to 176.8) | 211.8 (191.1 to 234.0) | 163.3 (136.9 to 191.8) | < 0.001 |
| Non–HDL-C | < 0.001 | |||
| < 130 | 55,866 (24.1) | 186 (0.3) | 56,052 (19.4) | |
| ≥ 130 | 175,791 (75.9) | 56,414 (99.7) | 232,205 (80.6) | |
| Direct LDL-C, mg/dL | 131.1 (111.7 to 150.8) | 169.2 (149.6 to 189.2) | 137.6 (116.1 to 160.3) | < 0.001 |
| Friedewald LDL-C, mg/dL | 128.6 (106.5 to 151.0) | 165.0 (140.0 to 189.9) | 134.7 (110.8 to 159.6) | < 0.001 |
| Martin/Hopkins LDL-C, mg/dL | 130.0 (108.7 to 151.5) | 170.3 (149.0 to 192.2) | 136.8 (113.5 to 161.3) | < 0.001 |
| Sampson LDL-C, mg/dL | 131.4 (109.4 to 153.7) | 167.7 (144.1 to 191.7) | 137.7 (113.9 to 162.3) | < 0.001 |
| Remnant cholesterol (Direct LDL-C method), mg/dLc | 22.7 (17.6 to 27.8) | 40.9 (37.4 to 46.8) | 25.1 (18.9 to 32.6) | < 0.001 |
| Remnant cholesterol (Friedewald equation), mg/dLc | 23.0 (17.1 to 31.2) | 47.1 (37.1 to 58.6) | 26.1 (18.4 to 37.7) | < 0.001 |
| Remnant cholesterol (Martin/Hopkins equation), mg/dLc | 22.8 (18.5 to 28.0) | 40.8 (34.5 to 48.0) | 25.0 (19.4 to 32.6) | < 0.001 |
| Remnant cholesterol (Sampson equation), mg/dLc | 20.5 (15.2 to 27.7) | 44.2 (34.7 to 54.7) | 23.2 (16.4 to 33.9) | < 0.001 |
| Direct LDL-C/TC Ratio | 0.6 (0.6 to 0.6) | 0.6 (0.6 to 0.7) | 0.6 (0.6 to 0.7) | < 0.001 |
| Friedewald LDL-C/TC Ratio | 0.6 (0.6 to 0.6) | 0.6 (0.6 to 0.7) | 0.6 (0.6 to 0.7) | < 0.001 |
| Martin/Hopkins LDL-C/TC Ratio | 0.6 (0.6 to 0.6) | 0.6 (0.6 to 0.7) | 0.6 (0.6 to 0.7) | < 0.001 |
| Sampson LDL-C/TC Ratio | 0.6 (0.6 to 0.7) | 0.6 (0.6 to 0.7) | 0.6 (0.6 to 0.7) | < 0.001 |
| ApoB and Direct LDL-C/TC Ratio | < 0.001 | |||
| ApoB < 90 and LDL-C/TC ratio < 0.60 | 50,767 (21.9) | 1821 (3.2) | 52,588 (18.2) | |
| ApoB < 90 and LDL-C/TC ratio ≥ 0.60 | 30,681 (13.2) | 514 (0.9) | 31,195 (10.8) | |
| ApoB ≥ 90 and LDL-C/TC ratio < 0.60 | 20,119 (8.7) | 4425 (7.8) | 24,544 (8.5) | |
| ApoB ≥ 90 and LDL-C/TC ratio ≥ 0.60 | 130,090 (56.2) | 49,840 (88.1) | 179,930 (62.4) | |
| Incident ASCVDd | ||||
| No. of events | 10,061 | 3664 | 13,725 | — |
| Total follow-up (person-years) | 3,108,059 | 753,117 | 3,861,176 | — |
| Follow-up, y | 13.9 (13.1 to 14.5) | 14.0 (13.1 to 14.5) | 13.9 (13.1 to 14.5) | 0.400 |
| Rate/1000 person-y | 3.2 | 4.9 | 3.6 | — |
| All-cause deaths | 18,747 (8.1) | 4777 (8.4) | 23,524 (8.2) | 0.007 |
Values are median (Q1 to Q3) or n (%). P values were calculated using the Wilcoxon rank-sum test for continuous variables and the Pearson χ2 test for categorical variables. SI unit conversions: To convert cholesterol values from mg/dL to mmol/L, multiply by 0.0259; to convert triglyceride values from mg/dL to mmol/L, multiply by 0.0113
Abbreviations: ApoB Apolipoprotein B, ASCVD Atherosclerotic cardiovascular disease, BMI Body mass index, DBP Diastolic blood pressure, HDL-C High-density lipoprotein cholesterol, LDL-C Low-density lipoprotein cholesterol, LDL-C/TC ratio ratio of low-density lipoprotein cholesterol to total cholesterol, Non–HDL-C Non–high-density lipoprotein cholesterol, RC Remnant cholesterol, SBP Systolic blood pressure, TC Total cholesterol, UK Biobank United Kingdom Biobank
aRemnant cholesterol (RC) was calculated as total cholesterol minus HDL-C minus LDL-C using directly measured LDL-C; the cutoff of 35 mg/dL was used to define elevated RC, consistent with guideline-recommended thresholds for nonfasting lipid profiles [21]
bLow HDL-C was defined by sex-specific thresholds (< 40 mg/dL for men and < 50 mg/dL for women), with higher values considered normal [24]
cRemnant cholesterol was calculated as total cholesterol minus HDL-C minus LDL-C using different LDL-C estimation methods, including direct measurement, Friedewald equation, Martin/Hopkins equation, and Sampson equation
dIncident ASCVD was defined as a composite of fatal or nonfatal myocardial infarction and fatal or nonfatal stroke
Within the dual-axis framework, elevated RC was predominantly concentrated within the dual-elevated group (ApoB ≥ 90 mg/dL and LDL-C/TC ratio ≥ 0.60). The majority of individuals with RC ≥ 35 mg/dL were classified as dual-elevated (88.1%), compared with 56.2% among those with lower RC levels. In contrast, the dual-normal group (ApoB < 90 mg/dL and LDL-C/TC ratio < 0.60) was rare in the high-RC group, whereas discordant patterns were less frequent, suggesting a compression of individuals with elevated RC into high-risk lipid profiles. Correspondingly, incidence rates of ASCVD were higher among participants with elevated RC compared with those with lower RC (4.9 vs 3.2 events per 1000 person-years) over a median follow-up of 13.9 years.
Association of dual-axis lipid categories with ASCVD
In the overall population, the dual-axis framework differentiated ASCVD risk across lipid phenotypes (Table 2). Compared with the dual-normal group, risk was highest in the dual-elevated group (adjusted HR, 1.23; 95% CI, 1.17–1.29), whereas the high-ApoB/low-ratio phenotype was not associated with increased risk. Among individuals with RC < 35 mg/dL, the risk gradient across dual-axis categories remained apparent, with increased risk in the low-ApoB/high-ratio and dual-elevated groups. In contrast, associations were attenuated and less precisely estimated among individuals with RC ≥ 35 mg/dL. There was no evidence that RC materially modified the association between dual-axis categories and ASCVD (P for interaction = 0.44). Results were similar when RC was defined using the alternative threshold of 39 mg/dL (P for interaction = 0.64). Sensitivity analyses using different LDL-C estimation methods produced broadly consistent findings after additional adjustment for log-transformed triglyceride levels (Supplementary Table 2).
Table 2.
Association of combined Apolipoprotein B and LDL-C/TC ratio categories with incident ASCVD overall and by remnant cholesterol levels in the UK Biobank
| Variable | Events/Sample | Incidence Rate (Per 1,000 Person-Years) | Unadjusted HR (95% CI) | Adjusted HR (95% CI) | P value |
|---|---|---|---|---|---|
| Overall population | |||||
| ApoB < 90 mg/dL and LDL-C/TC Ratio < 0.6 | 2,010/52,588 | 2.85 | 1 [Reference] | 1 [Reference] | |
| ApoB < 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 1,435/31,195 | 3.43 | 1.30 (1.21–1.39) | 1.08 (1.01–1.16) | 0.02 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio < 0.6 | 890/24,544 | 2.70 | 0.87 (0.80–0.94) | 0.98 (0.91–1.06) | 0.68 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 9,390/179,930 | 3.90 | 1.33 (1.27–1.39) | 1.23 (1.17–1.29) | < 0.001 |
| Remnant cholesterol (< 35 vs ≥ 35 mg/dL); P for effect modification = 0.44 | |||||
| < 35 mg/dL (n = 231,657) | |||||
| ApoB < 90 mg/dL and LDL-C/TC Ratio < 0.6 | 1,898/50,767 | 2.79 | 1 [Reference] | 1 [Reference] | |
| ApoB < 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 1,399/30,681 | 3.40 | 1.32 (1.23–1.42) | 1.10 (1.02–1.18) | 0.009 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio < 0.6 | 632/20,119 | 2.33 | 0.75 (0.69–0.82) | 0.90 (0.82–0.99) | 0.03 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 6,132/130,090 | 3.51 | 1.23 (1.17–1.29) | 1.16 (1.10–1.22) | < 0.001 |
| ≥ 35 mg/dL (n = 56,600) | |||||
| ApoB < 90 mg/dL and LDL-C/TC Ratio < 0.6 | 112/1,821 | 4.66 | 1 [Reference] | 1 [Reference] | |
| ApoB < 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 36/514 | 5.26 | 1.22 (0.84–1.78) | 1.17 (0.81–1.71) | 0.40 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio < 0.6 | 258/4,425 | 4.38 | 0.91 (0.73–1.14) | 1.04 (0.83–1.29) | 0.76 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 3,258/49,840 | 4.91 | 1.03 (0.85–1.24) | 1.17 (0.97–1.41) | 0.10 |
| Remnant cholesterol (< 39 vs ≥ 39 mg/dL); P for effect modification = 0.64 | |||||
| < 39 mg/dL (n = 251,382) | |||||
| ApoB < 90 mg/dL and LDL-C/TC Ratio < 0.6 | 1,934/51,448 | 2.81 | 1 [Reference] | 1 [Reference] | |
| ApoB < 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 1,417/30,946 | 3.42 | 1.32 (1.23–1.41) | 1.10 (1.02–1.18) | 0.008 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio < 0.6 | 696/21,321 | 2.43 | 0.78 (0.71–0.85) | 0.93 (0.85–1.01) | 0.08 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 7,200/147,667 | 3.64 | 1.26 (1.20–1.33) | 1.18 (1.12–1.24) | < 0.001 |
| ≥ 39 mg/dL (n = 36,875) | |||||
| ApoB < 90 mg/dL and LDL-C/TC Ratio < 0.6 | 76/1,140 | 5.10 | 1 [Reference] | 1 [Reference] | |
| ApoB < 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 18/249 | 5.44 | 1.14 (0.68–1.91) | 1.14 (0.68–1.91) | 0.62 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio < 0.6 | 194/3,223 | 4.54 | 0.86 (0.66–1.12) | 0.96 (0.73–1.25) | 0.75 |
| ApoB ≥ 90 mg/dL and LDL-C/TC Ratio ≥ 0.6 | 2,190/32,263 | 5.11 | 0.96 (0.76–1.20) | 1.11 (0.88–1.40) | 0.38 |
The definition of incident ASCVD is provided in the Table 1 footnote. Two prespecified cutoffs for remnant cholesterol were examined: 35 mg/dL (0.9 mmol/L), consistent with guideline-recommended thresholds for nonfasting lipid profiles [21], and 39 mg/dL (1.0 mmol/L), based on prior literature [9]. Associations were evaluated using multivariable Cox models with age as the time scale, adjusting for sex, ethnicity, smoking status, diabetes, body mass index, systolic blood pressure, and antihypertensive treatment
Abbreviations: ApoB apolipoprotein B, ASCVD atherosclerotic cardiovascular disease, BMI body mass index, HR hazard ratio, LDL-C low-density lipoprotein cholesterol, LDL-C/TC ratio ratio of low-density lipoprotein cholesterol to total cholesterol, SBP systolic blood pressure, TC total cholesterol
Cumulative incidence of ASCVD
Cumulative incidence analyses demonstrated clear separation of ASCVD risk across dual-axis categories among individuals with lower RC (< 35 mg/dL) (Fig. 1A). A stepwise gradient was observed across categories, with the highest incidence in the dual-elevated group and the lowest in the high-ApoB/low-ratio group relative to the dual-normal group (Gray’s test, P < 0.001). In contrast, among individuals with elevated RC (≥ 35 mg/dL), cumulative incidence curves were largely overlapping across dual-axis categories, with no significant differences observed (Gray’s test, P = 0.26) (Fig. 1B). These findings were consistent using an alternative RC threshold of 39 mg/dL (Supplementary Fig. 2).
Fig. 1.
Cumulative incidence of ASCVD across combined ApoB and LDL-C/TC ratio categories by remnant cholesterol threshold of 35 mg/dL (0.9 mmol/L). Panel A shows participants with remnant cholesterol <35 mg/dL (<0.9 mmol/L), and panel B shows participants with remnant cholesterol ≥35 mg/dL (≥0.9 mmol/L). ApoB, apolipoprotein B; ASCVD, atherosclerotic cardiovascular disease; LDL-C, low-density lipoprotein cholesterol; LDL-C/TC ratio, ratio of low-density lipoprotein cholesterol to total cholesterol; TC, total cholesterol. Both the LDL-C/TC ratio and remnant cholesterol were derived using directly measured LDL-C. Cumulative incidence of ASCVD was estimated using the Aalen–Johansen method with non-ASCVD death as a competing risk. Group differences were assessed using Gray’s test, with corresponding P values reported
Risk stratification in low–intermediate SCORE2 population
Among individuals with low–intermediate SCORE2 risk, cumulative incidence analyses demonstrated distinct patterns across lipid marker combinations (Fig. 2). Using a prespecified threshold of 35 mg/dL for RC, both the LDL-C/TC ratio and RC identified higher-risk groups, with RC showing more pronounced separation in cumulative incidence across ApoB strata. Consistent findings were observed at a threshold of 39 mg/dL.
Fig. 2.
Cumulative incidence of ASCVD across combined lipid categories based on ApoB, LDL-C/TC ratio, and remnant cholesterol in participants with low–intermediate SCORE2 risk in the UK Biobank. Panel A shows categories based on ApoB and the direct LDL-C/TC ratio. Panel B shows categories based on ApoB and remnant cholesterol using a cutoff of 35 mg/dL (0.9 mmol/L), and panel C shows categories based on ApoB and remnant cholesterol using a cutoff of 39 mg/dL (1.0 mmol/L). The study population comprised UK Biobank participants with low–intermediate SCORE2 risk, with derivation of this subgroup shown in Supplementary Fig. 1. SCORE2 risk categories were defined according to the 2025 ESC/EAS guidelines [2], with high risk defined as a predicted SCORE2 risk ≥ 10%, encompassing both high and very high risk levels. Eligibility was restricted to individuals aged 40–69 years without prior ASCVD, diabetes, chronic kidney disease, or genetic or rare lipid or blood pressure disorders. Both the LDL-C/TC ratio and remnant cholesterol were derived using directly measured LDL-C. Two prespecified cutoffs for remnant cholesterol were examined: 35 mg/dL (0.9 mmol/L), consistent with guideline-recommended thresholds for nonfasting lipid profiles [21], and 39 mg/dL (1.0 mmol/L), based on prior literature [9]
Consistent with these observations, RC conferred greater absolute risk differences than the LDL-C/TC ratio when comparing elevated versus lower levels within ApoB strata (Supplementary Table 3), with slightly larger absolute risk differences observed among individuals with lower ApoB levels. At 10 years, the absolute risk difference for RC (≥ 35 vs < 35 mg/dL) was 1.22% among individuals with ApoB < 90 mg/dL and 1.09% among those with ApoB ≥ 90 mg/dL, corresponding to 82 and 92 individuals per excess ASCVD event, respectively. In contrast, the corresponding differences for the LDL-C/TC ratio (≥ 0.60 vs < 0.60) were smaller (0.42% and 0.85%, respectively).
Elevated RC (≥ 35 mg/dL), defined as the combined proportion of individuals with high RC across both ApoB strata (ApoB < 90 and ≥ 90 mg/dL), was present in a relatively small proportion of the population (approximately 15%). In contrast, elevations in the LDL-C/TC ratio (≥ 0.60), similarly defined across ApoB strata, were substantially more prevalent (approximately 71%) (Fig. 2). Accordingly, RC identified a more concentrated high-risk subgroup, whereas the LDL-C/TC ratio captured a substantially larger proportion of the population at modestly increased risk.
When considered jointly, the number of elevated lipid markers (ApoB, LDL-C/TC ratio, and RC) was associated with a graded increase in cumulative incidence (Fig. 3). Absolute risk differences increased stepwise with the number of elevated markers, reaching 1.49% at 10 years when comparing individuals with all three markers elevated versus those with none elevated, corresponding to 67 individuals per excess ASCVD event, with progressively lower numbers of individuals per excess event as marker burden increased (Supplementary Table 4).
Fig. 3.
Cumulative incidence of ASCVD according to the number of elevated lipid markers (ApoB, LDL-C/TC Ratio, and Remnant Cholesterol) in participants with low–intermediate SCORE2 Risk in the UK Biobank. Panel A defines elevated markers as ApoB ≥90 mg/dL, LDL-C/TC ratio ≥0.60, and remnant cholesterol ≥35 mg/dL (0.9 mmol/L). Panel B uses the same ApoB and LDL-C/TC ratio thresholds but defines elevated remnant cholesterol as ≥39 mg/dL (1.0 mmol/L). The study population, LDL-C calculation method, and SCORE2 definition were identical to those described for Fig. 2
Exploratory analyses of RC in relation to lipid markers
In exploratory analyses, combinations of RC with ApoB and the LDL-C/TC ratio were associated with graded differences in ASCVD risk, with the highest risk observed when RC and the corresponding lipid marker were both elevated (Supplementary Table 5). Groups with elevated RC had higher adjusted risk estimates than the common reference group across both ApoB- and LDL-C/TC ratio–based combinations, with the highest risk observed in the dual-elevated groups. Risk was highest when elevated RC co-occurred with either elevated ApoB or a higher LDL-C/TC ratio. These patterns were consistent across both RC thresholds (35 and 39 mg/dL).
Incremental discrimination analysis
Among participants with low-to-intermediate SCORE2 risk, the C-index of the model incorporating SCORE2 component variables was 0.6449 (95% CI, 0.6395–0.6504). Sequential addition of ApoB and the LDL-C/TC ratio produced small improvements in discrimination (ΔC, 0.0013; 95% CI, 0.0007–0.0021; P < 0.001, and ΔC, 0.0012; 95% CI, 0.0005–0.0021; P < 0.001, respectively). Further addition of RC did not improve discrimination (ΔC, − 0.0001; 95% CI, − 0.0002 to 0.0001; P = 0.364) (Supplementary Table 6).
Discussion
Principal findings
In this large population-based primary prevention cohort free of lipid-lowering therapy, RC did not materially modify the associations between dual-axis categories and ASCVD risk, although risk separation appeared attenuated within the higher-RC stratum. Within the SCORE2-defined low-to-intermediate risk population, when considered in conjunction with ApoB, RC identified a more concentrated, higher-risk subgroup, whereas the LDL-C/TC ratio identified a substantially larger proportion of individuals at modestly increased risk. Joint consideration of ApoB, the LDL-C/TC ratio, and RC demonstrated a graded increase in ASCVD risk with increasing number of elevated markers. Together, these findings indicate that RC provides complementary risk information within the proposed dual-axis framework, although the present interaction analysis does not determine whether RC is statistically or biologically independent of other lipid measures. Incremental discrimination analyses provided a complementary perspective. Adding ApoB and the LDL-C/TC ratio resulted in statistically significant but small increases in the C-index, whereas RC provided no further improvement, indicating that the observed phenotype-level differences and absolute-risk gradients should not be interpreted as substantial gains in population-level discrimination.
Comparison with previous studies
The present study extends previous dual-axis analysis across six US cohorts, in which the joint assessment of ApoB and the LDL-C/TC ratio identified distinct ASCVD risk phenotypes, including increased risk in the low-ApoB/high-ratio phenotype [6]. Whereas that study demonstrated risk stratification by integrating particle burden with LDL cholesterol distribution, it did not evaluate how RC fits within this structure. By incorporating RC into the dual-axis framework in the UK Biobank, the current study examined whether RC modified associations across dual-axis phenotypes and whether it further stratified absolute risk within ApoB-defined categories.
Previous primary-prevention studies have generally supported the prognostic relevance of RC, although they addressed somewhat different questions. In a pooled analysis of 17,532 individuals, Quispe et al. found that RC remained associated with incident ASCVD after adjustment for conventional risk factors, LDL-C, and ApoB [25]. Doi et al. subsequently reported that adding markedly elevated RC to conventional risk models appropriately reclassified a proportion of individuals who subsequently developed myocardial infarction or ischemic heart disease into higher 10-year risk categories, with similar findings in a sensitivity analysis restricted to participants with a SCORE2-estimated 10-year cardiovascular risk of 1% to 10% [26]. More recently, an EPIC-Norfolk analysis found that RC remained associated with major cardiovascular events after adjustment for ApoB and LDL-C, while hsCRP mediated only a small proportion of this association [15]. Our observation that elevated RC identified higher absolute risk across ApoB strata and contributed to graded risk across increasing numbers of elevated lipid markers is broadly consistent with these findings.
However, evidence regarding whether remnant-related cholesterol provides risk information fully separable from ApoB is less uniform. In a residual-discordance analysis of the UK Biobank and Framingham Heart Study, Bilgic et al. used VLDL-C as a remnant-related cholesterol measure and found that its association with ASCVD became small after HDL-C was additionally considered, whereas the association of ApoB remained comparatively robust [27]. Because VLDL-C is not identical to calculated RC, these results are relevant but not directly equivalent to the present analysis. Conversely, a recent Copenhagen General Population Study found that individuals with high RC, high ApoB, and low LDL-C had an ASCVD HR of 1.45, compared with an HR of 1.20 among individuals with low RC, high ApoB, and high LDL-C, with both groups compared against participants with concordantly low lipid values [28]. These findings suggest that ApoB captures the total number of circulating atherogenic particles, while cholesterol distribution across ApoB-containing lipoproteins may still contribute to heterogeneity in risk. In this context, the absence of interaction between RC and the dual-axis categories in our study should not be interpreted as evidence that RC lacks independent prognostic information. Rather, RC did not behave as a clearly separable effect-modifying axis within the present categorical framework, while still providing complementary absolute-risk information and helping define graded risk across increasing lipid-marker burden.
Positioning remnant cholesterol within the ApoB-defined lipid risk framework
Our findings suggest that, within the proposed framework, RC is most appropriately interpreted in relation to the ApoB-defined atherogenic lipoprotein system rather than as a clearly separable effect-modifying axis. Although RC has been consistently associated with ASCVD in observational and genetic studies [14, 29], whether it represents a distinct dimension of risk remains debated [30]. Because RC reflects cholesterol carried within triglyceride-rich lipoproteins and their remnants—particles that are themselves ApoB-containing—its apparent statistical independence from ApoB should be interpreted cautiously, as biological coupling within the same atherogenic particles limits clear separation between particle number and cholesterol content [30]. In this context, RC did not materially modify the associations defined by the dual-axis model and, at higher levels, was associated with attenuation of risk separation across lipid-defined categories, suggesting convergence of risk across subgroups. Together, these findings suggest that RC-associated risk substantially overlaps with broader atherogenic particle burden, while potentially retaining information related to the composition and metabolic context of ApoB-containing particles.
Within this ApoB-centered framework, RC provided additional phenotypic and absolute-risk characterization, although it did not further improve model discrimination. Whereas ApoB captures the total number of atherogenic particles and defines the primary structural axis, RC appears to reflect variation in their remnant-related composition and metabolic behavior. Accordingly, RC identified a more concentrated subgroup with disproportionately higher absolute risk, while the LDL-C/TC ratio captured a broader population at more modestly increased risk. Notably, elevated RC identified higher absolute risk even within the lower ApoB and lower LDL-C/TC ratio strata, suggesting that it may capture remnant-related disturbances not fully reflected by particle number or cholesterol partitioning alone. At the same time, elevated RC was rarely observed as an isolated abnormality, instead clustering within profiles characterized by higher ApoB and adverse lipid composition, indicating that it more often reflects globally increased atherogenic burden rather than a distinct pathway. These observations support a hierarchical model in which ApoB defines the backbone of risk, while RC provides context-dependent refinement within that structure.
These findings are consistent with evidence from lipid-modifying intervention studies and subsequent analyses [31, 32], including the PROMINENT trial, in which substantial reductions in triglycerides and remnant-related cholesterol were not accompanied by cardiovascular benefit in the absence of a reduction in ApoB-containing particle burden. This pattern suggests that lowering RC alone may be insufficient to reduce risk if the total number of atherogenic particles remains unchanged or increases, reinforcing the central role of ApoB as an integrative constraint on atherogenic exposure. More broadly, these findings support a framework in which cardiovascular risk reflects the combined effects of atherogenic particle number and cholesterol content, with ApoB serving as the primary integrative measure.
The distributional patterns observed in this study further support a mechanistic interpretation of RC as a marker of an adverse remnant-rich metabolic phenotype. Elevated RC was accompanied by higher triglycerides, higher ApoB, lower HDL-C, and higher non–HDL-C, consistent with enrichment of triglyceride-rich lipoproteins, altered cholesterol partitioning, and impaired remnant metabolism. Remnant particles are sufficiently small to enter the arterial wall and relatively enriched in cholesterol, providing a plausible basis for their atherogenic potential, and may also contribute to vascular injury through lipolysis-related inflammatory pathways [29, 33–35]. The clustering of these abnormalities, together with the marked concentration of participants in the dual-elevated category, may have contributed to the attenuated and less precisely estimated risk separation observed at higher RC levels. Conversely, the persistence of RC-associated risk in the absence of overt abnormalities in ApoB or cholesterol distribution suggests that RC may also capture earlier or subtler disturbances in triglyceride-rich lipoprotein metabolism. Taken together, these findings indicate that RC reflects both the severity and the metabolic context of atherogenic lipoprotein dysfunction, rather than emerging as a clearly separable risk dimension within the present categorical framework.
Positioning non–HDL-C and Lipoprotein(a) within the framework
The proposed framework should be interpreted alongside, rather than as a replacement for, established lipid measures. Non–HDL-C reflects the cholesterol mass carried by all ApoB-containing lipoproteins and remains a guideline-recognized measure of atherogenic cholesterol burden. ApoB, in contrast, quantifies the circulating number of atherogenic particles; the two measures therefore provide related but noninterchangeable information [2]. In the present analysis, non–HDL-C was not incorporated as an additional classification axis because it is mathematically coupled to the measures under investigation: calculated RC equals non–HDL-C minus LDL-C, while the LDL-C/TC ratio shares LDL-C and total cholesterol with these measures. Accordingly, the present analyses characterize co-occurring, correlated lipid phenotypes and do not establish that each marker represents a fully independent biological pathway. The findings should also not be interpreted as demonstrating superiority of the proposed framework over non–HDL-C.
Lipoprotein(a) [Lp(a)] represents a distinct and predominantly genetically determined cardiovascular risk modifier, and contemporary guidelines recommend measuring it at least once during adulthood [2]. Although each Lp(a) particle contains ApoB-100 and therefore contributes to measured ApoB, total ApoB does not distinguish Lp(a)-related particle burden or its associated risk. Lp(a) is therefore best considered an additional inherited risk-modifying layer alongside, rather than subsumed within, the proposed ApoB-centered framework.
Potential role of polygenic risk scores
Polygenic risk scores (PRS) may provide an additional inherited risk layer beyond circulating lipid phenotypes. In individuals with hypercholesterolemia but without an identifiable pathogenic familial hypercholesterolemia variant, an LDL-C PRS may help characterize a potential polygenic contribution to elevated cholesterol, whereas a coronary artery disease (CAD) PRS may capture broader inherited susceptibility to ASCVD among individuals with otherwise similar lipid levels [36]. This distinction is relevant because cardiovascular risk may vary according to genetic background even at comparable LDL-C concentrations. Conceptually, PRS would complement rather than replace the present framework: ApoB, the LDL-C/TC ratio, and RC characterize the circulating lipid phenotype at baseline, whereas PRS reflects lifelong inherited susceptibility. However, PRS has not yet been recommended for routine cardiovascular risk assessment, and its incremental predictive value, ancestry-specific calibration, transportability, and clinical utility require further evaluation before widespread implementation [36].
Clinical interpretation and future evaluation
The findings suggest that joint assessment of ApoB, the LDL-C/TC ratio, and RC may help characterize lipid-related risk heterogeneity among participants with low-to-intermediate SCORE2 risk [16]. Co-occurring elevations in these markers were associated with progressively higher absolute ASCVD risk, with more apparent risk separation when at least two markers were elevated. Because the LDL-C/TC ratio and calculated RC are derived from overlapping components of the standard lipid profile, however, the marker count should not be interpreted as representing three independent biological pathways or as establishing a clinical action threshold.
ApoB serves as the measure of atherogenic particle number within the proposed framework. The LDL-C/TC ratio and RC may provide complementary phenotypic information regarding cholesterol partitioning and remnant-related cholesterol burden when interpreted alongside ApoB and established clinical risk assessment. Neither derived measure should be used to determine whether ApoB measurement is necessary, and the proposed framework should not be interpreted as replacing non–HDL-C or the separate assessment of Lp(a). Before clinical implementation, its incremental predictive performance, external validity, and consequences for clinical decision-making require formal evaluation.
Limitations
Several limitations should be considered. First, this was an observational analysis, and residual or unmeasured confounding limits causal inference. Second, RC was calculated rather than directly measured and may not correspond exactly to directly measured remnant cholesterol. ApoB is biologically related to the derived lipid measures, while the LDL-C/TC ratio and calculated RC are mathematically coupled through overlapping components of the standard lipid profile. Calculated RC equals non–HDL-C minus LDL-C, and the LDL-C/TC ratio is derived from LDL-C and total cholesterol. Consequently, analyses based on the number of elevated markers may partly reflect clustering of correlated features within the same lipid profile and should not be interpreted as demonstrating additive effects of independent biological pathways. Similarly, the absence of statistical interaction does not demonstrate homogeneity of risk across dual-axis phenotypes or exclude independent prognostic information from RC.
Third, although directly measured LDL-C was used in the primary analysis, classification based on fixed ApoB, LDL-C/TC ratio, and RC cutoffs remains sensitive to values near the thresholds. These prespecified cutoffs were used for phenotypic classification and should not be interpreted as guideline-endorsed treatment thresholds. Sex-specific calibration or thresholds for the LDL-C/TC ratio were not evaluated. In addition, participants with elevated RC were predominantly concentrated in the dual-elevated category, resulting in relatively small discordant subgroups and less precise estimates within the high-RC stratum. Fourth, lipid measures were assessed only at baseline, and subsequent changes in lipid levels or initiation of lipid-lowering therapy during follow-up were not incorporated.
Fifth, the proposed framework was not directly compared with non–HDL-C and did not incorporate Lp(a), an established and predominantly genetically determined cardiovascular risk modifier. PRS were also not evaluated because the study was prespecified to examine circulating lipid phenotypes rather than inherited risk prediction. Evaluation of PRS would require separate selection and validation of LDL-C– and CAD-specific scores, consideration of genetic ancestry and population structure, and assessment of transportability and incremental predictive performance. Sixth, although incremental discrimination was evaluated within the SCORE2-eligible UK Biobank population, these analyses were conducted in the same cohort and therefore require external validation. No decision-curve, treatment-benefit, or cost-effectiveness analyses were performed. Accordingly, the findings do not establish an optimal clinical testing sequence or support using the LDL-C/TC ratio or RC to determine whether ApoB should be measured. Finally, the healthy volunteer profile, predominantly White population, exclusion of participants with missing data, and restriction to untreated primary-prevention participants may limit generalizability to other populations, treated individuals, and secondary-prevention settings.
Conclusion
In this large untreated primary-prevention cohort, RC did not materially modify the associations across ApoB–LDL-C/TC phenotypes, although elevated RC marked higher absolute ASCVD risk within ApoB strata. Co-occurring elevations in ApoB, the LDL-C/TC ratio, and RC were associated with progressively higher ASCVD risk, with greater risk separation at higher marker counts. However, the addition of ApoB and the LDL-C/TC ratio produced only small improvements in discrimination, and RC provided no further improvement. Because these measures are correlated and partly mathematically coupled, the findings should be interpreted as joint lipid phenotypes rather than as independent biological dimensions or as establishing a clinical action threshold. The proposed ApoB-centered framework may provide complementary phenotypic characterization, but its predictive performance requires external validation, and its clinical utility requires further decision-analytic and prospective evaluation.
Supplementary Information
Acknowledgements
This research has been conducted using the UK Biobank Resource under Application Number 531547. This work uses data provided by patients and collected by the NHS as part of their care and support. We gratefully acknowledge the UK Biobank participants and the research team for their invaluable contributions that made this study possible.
Abbreviations
- ApoB
Apolipoprotein B
- ASCVD
Atherosclerotic cardiovascular disease
- BMI
Body mass index
- CI
Confidence interval
- CAD
Coronary artery disease
- CVD
Cardiovascular disease
- DBP
Diastolic blood pressure
- HDL-C
High-density lipoprotein cholesterol
- HR
Hazard ratio
- IQR
Interquartile range
- LDL-C/TC ratio
Ratio of low-density lipoprotein cholesterol to total cholesterol
- LDL-C
Low-density lipoprotein cholesterol
- Lp(a)
Lipoprotein(a)
- Non–HDL-C
Non–high-density lipoprotein cholesterol
- PRS
Polygenic risk scores
- RC
Remnant cholesterol
- SBP
Systolic blood pressure
- SCORE2
Systematic Coronary Risk Evaluation 2
- TC
Total cholesterol
Authors’ contributions
B.W. conceived the study, conducted the analyses, interpreted the data, and drafted the manuscript. D.L. supervised the study, contributed to data interpretation, and critically revised the manuscript for important intellectual content. Both authors reviewed the results, provided critical feedback, and approved the final version of the manuscript.
Funding
This work was supported by the Beijing Natural Science Foundation (IS25067) and the National Institute of Hospital Administration (LY2025JY008).
Data availability
Access to UK Biobank data is available to qualified researchers through a standard application via the official UK Biobank platform (https://www.ukbiobank.ac.uk/).
Declarations
Ethics approval and consent to participate
No additional ethical approval was required for this study. UK Biobank received ethical approval from the North West Multi-Centre Research Ethics Committee, and all participants provided written informed consent at enrollment.
Consent for publication
Not applicable.
Competing interests
The authors declare no conflicts of interest. The views expressed in this manuscript are those of the authors and do not necessarily reflect those of the UK Biobank.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Bin Wang, Email: wangb.2023@tsinghua.org.cn.
Dong Li, Email: lidong15@mail.tsinghua.edu.cn.
References
- 1.Writing Committee M, Blumenthal RS, Morris PB, Gaudino M, Johnson HM, Anderson TS, et al. 2026 ACC/AHA/AACVPR/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Dyslipidemia: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2026;153:e1154–276. 10.1161/CIR.0000000000001423. [DOI] [PubMed] [Google Scholar]
- 2.Mach F, Koskinas KC, Roeters van Lennep JE, Tokgozoglu L, Badimon L, Baigent C, et al. Focused Update of the 2019 ESC/EAS Guidelines for the management of dyslipidaemias. Eur Heart J. 2025;2025(46):4359–78. 10.1093/eurheartj/ehaf190. [DOI] [PubMed] [Google Scholar]
- 3.Marston NA, Giugliano RP, Melloni GEM, Park JG, Morrill V, Blazing MA, et al. Association of apolipoprotein B-containing lipoproteins and risk of myocardial infarction in individuals with and without atherosclerosis: distinguishing between particle concentration, type, and content. JAMA Cardiol. 2022;7:250–6. 10.1001/jamacardio.2021.5083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Sniderman AD, Dufresne L, Pencina KM, Bilgic S, Thanassoulis G, Pencina MJ. Discordance among apoB, non-high-density lipoprotein cholesterol, and triglycerides: implications for cardiovascular prevention. Eur Heart J. 2024;45:2410–8. 10.1093/eurheartj/ehae258. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sayed A, Peterson ED, Virani SS, Sniderman AD, Navar AM. Individual variation in the distribution of apolipoprotein B levels across the spectrum of LDL-C or non-HDL-C levels. JAMA Cardiol. 2024;9:741–7. 10.1001/jamacardio.2024.1310. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wang B, Nurmohamed NS, Kraaijenhof JM, Li D. ASCVD Risk Stratification Using Apolipoprotein B and the LDL-C to Total Cholesterol Ratio. Eur J Prev Cardiol. 2026. 10.1093/eurjpc/zwag260. [DOI] [PubMed] [Google Scholar]
- 7.Soffer DE, Marston NA, Maki KC, Jacobson TA, Bittner VA, Pena JM, et al. Role of apolipoprotein B in the clinical management of cardiovascular risk in adults: an expert clinical consensus from the National Lipid Association. J Clin Lipidol. 2024;18:e647–63. 10.1016/j.jacl.2024.08.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.De Oliveira-Gomes D, Joshi PH, Peterson ED, Rohatgi A, Khera A, Navar AM. Apolipoprotein B: bridging the gap between evidence and clinical practice. Circulation. 2024;150:62–79. 10.1161/CIRCULATIONAHA.124.068885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wadstrom BN, Pedersen KM, Wulff AB, Nordestgaard BG. Elevated remnant cholesterol, plasma triglycerides, and cardiovascular and non-cardiovascular mortality. Eur Heart J. 2023;44:1432–45. 10.1093/eurheartj/ehac822. [DOI] [PubMed] [Google Scholar]
- 10.Sturzebecher PE, Katzmann JL, Laufs U. What is “remnant cholesterol”? Eur Heart J. 2023;44:1446–8. 10.1093/eurheartj/ehac783. [DOI] [PubMed] [Google Scholar]
- 11.Wulff AB, Varbo A, Nordestgaard AT, Nordestgaard BG. Feasibility of remnant cholesterol as a therapeutic target for atherosclerotic cardio-vascular disease. Expert Opin Ther Targets. 2026;30:229–44. 10.1080/14728222.2026.2639678. [DOI] [PubMed] [Google Scholar]
- 12.Yang XH, Zhang BL, Cheng Y, Fu SK, Jin HM. Association of remnant cholesterol with risk of cardiovascular disease events, stroke, and mortality: a systemic review and meta-analysis. Atherosclerosis. 2023;371:21–31. 10.1016/j.atherosclerosis.2023.03.012. [DOI] [PubMed] [Google Scholar]
- 13.Jung HN, Heo JH, Kim MJ, Han KD, Kang JG, Lee SJ. Remnant Cholesterol as an Independent Cardiovascular Risk Factor in Young Adults: An Age-Stratified Cohort Study. J Am Coll Cardiol. 2025;86:2047–58. 10.1016/j.jacc.2025.09.015. [DOI] [PubMed] [Google Scholar]
- 14.Bjornson E, Adiels M, Taskinen MR, Burgess S, Rawshani A, Boren J, et al. Triglyceride-rich lipoprotein remnants, low-density lipoproteins, and risk of coronary heart disease: a UK Biobank study. Eur Heart J. 2023;44:4186–95. 10.1093/eurheartj/ehad337. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kraaijenhof JM, Kerkvliet MJ, Nurmohamed NS, Grefhorst A, Kroon J, Wareham NJ, et al. The role of systemic inflammation in remnant cholesterol-associated cardiovascular risk: insights from the EPIC-Norfolk study. Eur J Prev Cardiol. 2026;33:729–39. 10.1093/eurjpc/zwaf037. [DOI] [PubMed] [Google Scholar]
- 16.group Sw, collaboration ESCCr. SCORE2 risk prediction algorithms: new models to estimate 10-year risk of cardiovascular disease in Europe. Eur Heart J. 2021;42:2439–54. 10.1093/eurheartj/ehab309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779. 10.1371/journal.pmed.1001779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499–502. [PubMed] [Google Scholar]
- 19.Martin SS, Blaha MJ, Elshazly MB, Toth PP, Kwiterovich PO, Blumenthal RS, et al. Comparison of a novel method vs the Friedewald equation for estimating low-density lipoprotein cholesterol levels from the standard lipid profile. JAMA. 2013;310:2061–8. 10.1001/jama.2013.280532. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Sampson M, Ling C, Sun Q, Harb R, Ashmaig M, Warnick R, et al. A new equation for calculation of low-density lipoprotein cholesterol in patients with normolipidemia and/or hypertriglyceridemia. JAMA Cardiol. 2020;5:540–8. 10.1001/jamacardio.2020.0013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Nordestgaard BG, Langsted A, Mora S, Kolovou G, Baum H, Bruckert E, et al. Fasting is not routinely required for determination of a lipid profile: clinical and laboratory implications including flagging at desirable concentration cut-points-a joint consensus statement from the European Atherosclerosis Society and European Federation of Clinical Chemistry and Laboratory Medicine. Eur Heart J. 2016;37:1944–58. 10.1093/eurheartj/ehw152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: competing risks and multi-state models. Stat Med. 2007;26:2389–430. 10.1002/sim.2712. [DOI] [PubMed] [Google Scholar]
- 23.Suissa S. Calculation of number needed to treat. N Engl J Med. 2009;361:424–5. 10.1056/NEJMc0903274. [DOI] [PubMed] [Google Scholar]
- 24.Grundy SM, Stone NJ, Bailey AL, Beam C, Birtcher KK, Blumenthal RS, et al. 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA guideline on the management of blood cholesterol: a report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines. Circulation. 2018;139:e1082–143. 10.1161/CIR.0000000000000625. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Quispe R, Martin SS, Michos ED, Lamba I, Blumenthal RS, Saeed A, et al. Remnant cholesterol predicts cardiovascular disease beyond LDL and ApoB: a primary prevention study. Eur Heart J. 2021;42:4324–32. 10.1093/eurheartj/ehab432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Doi T, Langsted A, Nordestgaard BG. Elevated remnant cholesterol reclassifies risk of ischemic heart disease and myocardial infarction. J Am Coll Cardiol. 2022;79:2383–97. 10.1016/j.jacc.2022.03.384. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bilgic S, Pencina KM, Pencina MJ, Cole J, Dufresne L, Thanassoulis G, et al. Discordance analysis of VLDL-C and ApoB in UK Biobank and Framingham Study: a prospective observational study. Arterioscler Thromb Vasc Biol. 2024;44:2244–51. 10.1161/ATVBAHA.124.321165. [DOI] [PubMed] [Google Scholar]
- 28.Hvid K, Balling M, Nordestgaard BG, Afzal S: Elevated apoB due to high remnant cholesterol confers high risk of ASCVD without prompting sufficient lipid-lowering therapy: a guideline-based limitation and an unmet medical need. Eur J Prev Cardiol. 2025. 10.1093/eurjpc/zwaf735. [DOI] [PubMed]
- 29.Ginsberg HN, Packard CJ, Chapman MJ, Boren 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;42:4791–806. 10.1093/eurheartj/ehab551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Bilgic S, Remaley AT, Sniderman AD. Triglyceride-rich lipoprotein cholesterol and cardiovascular risk. Curr Opin Lipidol. 2023;34:259–66. 10.1097/MOL.0000000000000905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Das Pradhan A, Glynn RJ, Fruchart JC, MacFadyen JG, Zaharris ES, Everett BM, et al. Triglyceride lowering with pemafibrate to reduce cardiovascular risk. N Engl J Med. 2022;387:1923–34. 10.1056/NEJMoa2210645. [DOI] [PubMed] [Google Scholar]
- 32.Doi T, Langsted A, Nordestgaard BG. Remnant cholesterol, LDL cholesterol, and apoB absolute mass changes explain results of the PROMINENT trial. Atherosclerosis. 2024;393:117556. 10.1016/j.atherosclerosis.2024.117556. [DOI] [PubMed] [Google Scholar]
- 33.Heo JH, Jo SH. Triglyceride-rich lipoproteins and remnant cholesterol in cardiovascular disease. J Korean Med Sci. 2023;38:e295. 10.3346/jkms.2023.38.e295. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Balling M, Afzal S, Varbo A, Nordestgaard BG, Langsted A. Remnant cholesterol: quantification, concentrations by sex and age, and risk of ischemic heart disease. Clin Chem. 2025;71:451–62. 10.1093/clinchem/hvae217. [DOI] [PubMed] [Google Scholar]
- 35.Nordestgaard AT: Triglyceride-Rich Lipoproteins and Inflammation: Independent and Synergistic Roles in Cardiovascular Disease. Curr Atheroscler Rep. 2026;28. 10.1007/s11883-026-01393-4. [DOI] [PubMed]
- 36.Bosco G, Mszar R, Piro S, Sabouret P, Gallo A. Cardiovascular risk estimation and stratification among individuals with hypercholesterolemia. Curr Atheroscler Rep. 2024;26:537–48. 10.1007/s11883-024-01225-3. [DOI] [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
Access to UK Biobank data is available to qualified researchers through a standard application via the official UK Biobank platform (https://www.ukbiobank.ac.uk/).