Bringing doctors together

or

A free, private network for verified physicians

Leading voices

Dr. Nishant Shah

Preventive Cardiology · Duke Hospital

View

Dr. Amit Khera

Preventive Cardiology · UT Southwestern

View

Dr. Martha Gulati

Preventive Cardiology · Houston Methodist

View

Dr. Pradeep Natarajan

Preventive Cardiology · MGH

View

Dr. Christopher Cannon

Preventive Cardiology · BWH

View

Dr. Nicholas Marston

Preventive Cardiology · BWH

View

Dr. Dorairaj Prabhakaran

General Cardiology

View

Dr. Pugazhendhi Vijayaraman

EP · Geisinger

View

Dr. Robert Schaller

EP · HUP

View

Dr. Trey Woods

General Cardiology · Emory Healthcare

View
  • Discover your community

  • Stay current

  • A profile that represents you

  • Find your people

About RoonPrivacyTerms

    Overview

    Lipoprotein(a), Insulin Resistance, and Cardiovascular Disease in the UK Biobank.

    Feb 11, 2026DOI: 10.1161/jaha.125.042361Journal ArticleAHA

    Written by: Richard KazibweChristopher L SchaichParag A ChevliJeff A KingsleySaeid MirzaiJuliana H NamutebiMuhammad Imtiaz AhmadAnurag MehtaHarpreet S BhatiaMitchell PauknerRishi RikhiErin D MichosMichael D Shapiro

    Find these authors on Roon

    AM
    Dr. Anurag Mehta
    Preventive Cardiology · Emory University

    METHODS

    All data used in this analysis are available online at https://www.ukbiobank.ac.uk.The UK Biobank is a prospective cohort of over 500 000 individuals aged 40 to 69 years, recruited between 2006 and 2010. Participants visited one of 22 centers across England, Wales, and Scotland where physical examinations were performed, and biological samples were collected. Each participant completed a baseline questionnaire administered by trained study personnel. Details on the UK Biobank study design have been previously published.14The UK Biobank was established with ethical approval from the North West Multi‐Centre Research Ethics Committee (REC reference: 11/NW/03820), and all participants provided written informed consent. The current study was conducted under the Application Number: 97439. The data used in this study were extracted in November 2023 and analyzed from December 2023 to April 2024.IR was assessed using the TyG, calculated using the formula: ln [triglyceride (mg/dL)×plasma glucose (mg/dL)/2].15 At baseline, participants in the UK Biobank study provided peripheral venous blood samples, with validated collection procedures.16 As random blood samples were drawn at each assessment center to support a variety of disease studies, fasting samples were not available. Non‐fasting serum glucose and triglyceride were measured using a central laboratory. The coefficients of variation were less than 3% for triglyceride and less than 2% for glucose.Baseline serum Lp(a) concentrations (in nanomole per liter) were measured using an immunoturbidimetric assay (Randox Laboratories, Crumlin, County Antrim, UK) on the Beckman Coulter AU5800 platform, which used the Denka Seiken Method.17 This assay is traceable to the World Health organization/International Federation of Clinical Chemistry (WHO/IFCC) standard reference material (SRM‐2B) and has demonstrated good concordance with this standard. It also shows minimal bias, even in individuals with very large Lp(a) particles.18 We used Lp(a) levels ≥125 nmol/L as the threshold to define individuals at high CVD risk.3The primary end point of this analysis was incident CVD (primary outcome), defined as a composite of events identified by International Classification of Diseases, Ninth or Tenth Revision (‐9) or codes for peripheral arterial disease, coronary artery disease, myocardial infarction, ischemic stroke, or death from cardiovascular causes. Secondary outcomes included individual components of the primary outcome. The procedure for identifying outcome events was based on a combination of self‐reported data confirmed by trained health care professionals, hospitalization records, and national procedure and death registries and has been previously published.19 We excluded participants with CVD at baseline. The diagnosis date was defined as the earliest date on which the diagnosis was confirmed. Participants were observed from enrollment until the occurrence of an incident CVD event, loss to follow‐up, or the final follow‐up date, set as November 20, 2023, when data were extracted for analysis.
    ICD
    ICD‐10
    Baseline assessments were conducted at study centers with the use of touchscreen questionnaires to collect data on sociodemographic factors, family history, lifestyle, and medical history, which were validated by trained personnel. Physical assessments involved recording vital signs, anthropometric measurements, and collecting blood and urine samples for laboratory analysis. Ethnicity was self‐reported and categorized as Asian, Black, White, or Mixed/Other; participants who did not self‐identify with any listed category or did not report ethnicity were included in the Mixed/Other category. The Townsend deprivation index, based on residential postcode, reflected unemployment, home/car ownership, and overcrowding.20 Participants also reported tobacco use (never, former, or current) and medication use at enrollment. Trained staff measured blood pressure, height, and weight. Body mass index (BMI) was calculated as weight (kg) divided by height (m2).
    Physical activity was assessed using modified questions from the validated short International Physical Activity Questionnaire.21 Activity time was converted into metabolic equivalent minutes/week. Hypertension at baseline was defined as self‐reported hypertension or use of antihypertensive medication, a prior hospital diagnosis of hypertension, systolic blood pressure (BP)≥140 mm Hg or diastolic BP ≥90 mm Hg at recruitment. The methods for blood and urine sample analysis have been previously detailed.16 Biochemical tests, including serum total cholesterol, high‐density lipoprotein cholesterol, triglyceride, and serum creatinine, were performed at a central laboratory. Creatinine was measured using a Beckman Coulter AU5800 with IDMS‐traceable methods. The estimated glomerular filtration rate in milliliters per minute per 1.73 m2 of body surface area was calculated with the use of the 4 variable Modification of Diet in Renal Disease equation.22
    We used Spearman’s rank correlation coefficient to assess cross sectional relationship between TyG and Lp(a). We conducted a Cox multivariable analysis to estimate the risk of CVD associated with log‐transformed Lp(a), because of its skewed distribution, and TyG, reporting the adjusted hazard ratios (aHRs) for each variable while adjusting for the other. Next, the study population was stratified by Lp(a) levels using a cutoff of 125 nmol/L, which corresponds to a clinically significant threshold associated with high CVD risk. As there are no established normal values for TyG to define IR,23 we classified IR into high and low categories using the 75th percentile of TyG as the cutoff, which corresponded to a value of 9.07 in this cohort.
    We then created 4 groups based on the combination of Lp(a) and TyG status as follows: Group 1: Lp(a) <125 nmol/L with low IRGroup 2: Lp(a) ≥125 nmol/L with low IRGroup 3: Lp(a) <125 nmol/L with high IRGroup 4: Lp(a) ≥125 nmol/L with high IR
    Group 1: Lp(a) <125 nmol/L with low IR
    Group 2: Lp(a) ≥125 nmol/L with low IR
    Group 3: Lp(a) <125 nmol/L with high IR
    Group 4: Lp(a) ≥125 nmol/L with high IR
    We summarized the demographic and clinical characteristics of participants by the 4 groups (Group 1–4). We used Chi‐square tests for categorical variables, while a generalized linear model was applied to continuous variables, computing least squares means for each group. Pairwise comparisons between groups were adjusted using the Bonferroni method to control for multiple comparisons in normally distributed data,24 or the Kruskal‐Wallis test followed by Dunn’s test with adjustments for multiple comparisons in skewed data.25
    In all instances, we used 2 Cox models: a minimally adjusted model (Model 1) that included age, sex, ethnicity, and Townsend deprivation index; and a second model (Model 2) additionally adjusting for body mass index, smoking status, systolic blood pressure, high‐density lipoprotein cholesterol, statin therapy, and estimated glomerular filtration rate.
    To ensure the validity of our Cox proportional hazards models, we conducted diagnostic checks in the primary analyses where log‐Lp(a) and TyG were modeled as continuous variables. The proportional hazards assumption was evaluated using Schoenfeld residuals, which showed no evidence of violation (Table S1). We further assessed the potential influence of individual observations using deviance residuals (Figure S1) and examined the linearity of continuous covariate effects on the log‐hazard scale using martingale residuals (Figure S2). These diagnostic evaluations did not indicate any significant violations, supporting the robustness of the models.
    We conducted exploratory analyses, including testing for interaction, to assess the association between Lp(a), IR categories, and primary outcome across subgroups defined by sex, age (<50, 50–60, >65 years), baseline hypertension status (self‐report or use of antihypertensive medication, a prior ICD diagnosis, or BP ≥140/90 mm Hg), obesity (BMI ≥30 kg/m2), current smoking status, family history of heart disease, and chronic kidney disease (CKD) (defined as estimated glomerular filtration rate <60 mL/min per 1.73 m2).
    We conducted sensitivity analyses to evaluate the robustness of our findings. First, we re‐fitted Cox Model 2 with an additional adjustment for baseline glycated hemoglobin level. Second, we repeated the analysis stratifying the study population by diabetes status. Diabetes was defined as a self‐reported diagnosis or documented history of diabetes, use of antidiabetic medications, a random blood glucose level of 200 mg/dL or higher, or a glycated hemoglobin level of 6.5% or above. These sensitivity analyses were necessary to address potential confounding by diabetes status, as IR is often a precursor or feature of diabetes.26 Furthermore, some studies suggest that very low Lp(a) may be associated with diabetes risk.6 Third we also conducted additional sensitivity analyses excluding participants with triglyceride levels above 300 mg/dL, as triglyceride, used in calculating TyG, may be inversely associated with Lp(a) levels, particularly in individuals with very high triglyceride levels.27, 28 Finally we conducted an additional sensitivity analysis including participants taking lipid‐lowering therapies, as these may affect Lp(a), triglyceride, glucose metabolism, and diabetes status.29, 30
    In all instances, separate analyses were performed for the primary outcome, and the secondary outcomes. The CIs were calculated at a 95% confidence level. However, their widths have not been adjusted to account for multiple comparisons and should not be considered a substitute for formal hypothesis testing.
    A 2‐sided P‐value of <0.05 was considered statistically significant the main effects. However, for testing interaction terms, we pre‐specified a P‐value threshold of 0.1 to identify potential interactions, acknowledging that tests for interaction are often underpowered and require a less stringent threshold than main effect testing.31, 32
    All analyses were performed using SAS version 9.4 (SAS Institute Inc.) and the R statistical computing environment (version 4.1.3; http://www.r‐project.org).

    RESULTS

    At enrollment, the 328 031 participants included in this analysis had a mean age of 56.4 years (SD 8.1), with 54.7% being women, 94.0% identified as White, 30.4% had hypertension, 8.3% had diabetes, 10.3% were current smokers, and 13.0% reported taking lipid‐lowering medications.
    The participants with Lp(a) ≥125 nmol/L and high IR (Group 4) showed significant differences compared with those with Lp(a) <125 nmol/L and low IR (reference group) in several baseline characteristics and CVD risk factors. They had a lower percentage of female participants (43.2% versus 71.8%) and were older (57.3 versus 56.0 years). They also reported higher rates of current smoking (12.7% vs. 9.5%), family history of heart disease (7.0% versus 4.2%), diabetes (16.7% versus 5.2%), hypertension (41.4% versus 26.8%), CKD (6.8% versus 3.7%), and higher total cholesterol levels (239.8 mg/dL versus 216.7 mg/dL). Additionally, statin use was more common in the group with Lp(a) ≥125 nmol/L and high IR compared with the reference group (20.5% versus 10.6%) (Table 1).
    Baseline Characteristics of Study Population by Lipoprotein(a) and Insulin Resistance Categories (N=328 031)
    Low IR was defined as triglyceride‐index (TyG)<75 percentile; High IR was defined as Tyg ≥75 percentile. BMI indicates body mass index; BP, blood pressure; eGFR, estimated glomerular filtration rate; HDL‐C, high‐density lipoprotein cholesterol; IQR, interquartile range; IR, insulin resistance; LDL‐C, low‐density lipoprotein cholesterol; Lp(a), lipoprotein(a); MET, metabolic equivalent.
    P<0.05 compared with the Lp(a) <125 nmol/L with Low IR (reference group). P‐value is for the comparison between categories of LP(a) and TyG, tested with the Chi‐square tests for categorical variables. Pairwise comparisons between groups were adjusted using the Bonferroni method to control for multiple comparisons in normally distributed data, or the Kruskal–Wallis test followed by Dunn’s test with adjustments for multiple comparisons in skewed data.
    There was a minimal correlation between log‐Lp(a) and TyG with the Spearman correlation coefficient of −0.047 (P<0.0001).
    Over a median follow‐up period of 14.6 years (interquartile range: 13.7 to 15.4 years), 26 865 primary outcome events were recorded. In the fully adjusted model, the aHRs for the primary outcome were 1.08 (95% CI, 1.06–1.09) per SD increase in log‐Lp(a) and 1.06 (95% CI, 1.04–1.07) per SD increase in TyG, with each adjusted for the other. Using an Lp(a) cutoff value of 125 nmol/L or higher, the risk of the primary outcome was elevated, with an aHR of 1.17 (95% CI: 1.12–1.21) compared with those with Lp(a) below 125 nmol/L, after adjusting for TyG. Similarly, high IR was associated with the risk of the primary outcome, with an aHR of 1.10 (95% CI: 1.06–1.12) compared with low IR, after adjusting for Lp(a). We observed a suggestive interaction between log‐Lp(a) and TyG (P=0.07), which met our pre‐specified threshold of P<0.1 for interaction testing (Table 2).
    Risk of Cardiovascular Disease Associated With Log‐Lp(a) and TyG, Adjusted for Each Other
    Primary outcome defined as a composite of myocardial infarction, coronary artery disease, peripheral artery disease, ischemic stroke, or cardiovascular mortality. Model 1 adjusted for age, sex, ethnicity, Townsend Deprivation Index. Model 2 additionally adjusted for body mass index, smoking status, systolic blood pressure (BP), high‐density lipoprotein cholesterol, statin therapy, and estimated glomerular filtration rate. aHR indicates adjusted hazard ratio; CAD, coronary artery disease; Lp(a), lipoprotein(a); MI, myocardial infarction; PAD, peripheral artery disease; and TyG, triglyceride‐glucose index.
    Multiplicative interaction between log‐Lp(a)*TyG calculated with Model 2 covariates.
    The results from joint‐effect analyses, which assessed CVD risk by grouping participants with baseline Lp(a) and IR, are shown in Table 3. Relative to participants in the reference group with Lp(a) <125 nmol/L and low IR(Group 1), the aHRs for the primary outcome were 1.15 (95% CI: 1.10–1.20) for participants with Lp(a) ≥125 nmol/L and low IR(Group 2), 1.09 (95% CI: 1.06–1.12) for participants with Lp(a) <125 nmol/L and high IR (Group 3), and 1.32 (95% CI: 1.24–1.41) for participants with Lp(a) ≥125 nmol/L and high IR (Group 4).
    Risk of Cardiovascular Events Stratified by Lipoprotein(a) and Insulin Resistance Categories
    Low IR was defined as triglyceride‐index (TyG)<75 percentile; High IR was defined as Tyg ≥75 percentile. Primary outcome was defined as a composite of myocardial infarction, coronary artery disease, peripheral arterial disease, ischemic stroke, or cardiovascular mortality. Primary outcome was defined as a composite of myocardial infarction, coronary artery disease, peripheral arterial disease, ischemic stroke, or cardiovascular mortality. Model 1 adjusted for age, sex, ethnicity, Townsend Deprivation Index. Model 2 additionally adjusted for body mass index, smoking status, systolic blood pressure, high‐density lipoprotein cholesterol, statin therapy, and estimated glomerular filtration rate. aHR indicates adjusted hazard ratio; CVD, cardiovascular disease; IR, insulin resistance; and Lp(a), lipoprotein(a).
    The cumulative incidence curves for primary outcome, myocardial infarction, and cardiovascular death events among participants in the UK Biobank, according to the combination groups of Lp(a) and TyG, are shown in the Figure. The highest cumulative CVD incidence was observed in participants with Lp(a) ≥125 nmol/L and high IR, while the lowest was in those with Lp(a) <125 nmol/L and low IR.
    The primary outcome displayed in (A) represents the first major adverse cardiovascular event, defined as a composite of coronary artery disease, myocardial infarction, peripheral arterial disease, ischemic stroke, or cardiovascular death. (B) Shows the cumulative incidence of first myocardial infarction events. (C) Displays cardiovascular death. Participants were classified into 4 groups: Group 1, Lp(a) <125 nmol/L with low insulin resistance; Group 2, Lp(a) ≥125 nmol/L with low insulin resistance; Group 3, Lp(a) <125 nmol/L with high insulin resistance; Group 4, Lp(a) ≥125 nmol/L with high insulin resistance. Low insulin resistance was defined as triglyceride‐index <75th cohort percentile; high insulin resistance was defined as triglyceride‐index ≥75th cohort percentile.
    In Scenario 1, which examined the effect of IR status on the primary outcome within different Lp(a) groups, the adjusted hazard ratio (aHR) was 1.08 (95% CI: 1.05–1.12) among individuals with Lp(a) <125 nmol/L and 1.10 (95% CI: 1.01–1.20) among those with Lp(a) ≥125 nmol/L. In Scenario 2, which assessed the effect of Lp(a) status within different IR groups, the aHR for the primary outcome was 1.15 (95% CI: 1.10–1.20) in individuals with low IR and 1.21 (95% CI: 1.13–1.29) in those with high IR (Table 4).
    Risk Reclassification of Cardiovascular Risk Based on Lipoprotein and Insulin Resistance.
    Low insulin resistance was defined as triglyceride‐glucose index <75 percentile; High insulin resistance was defined as triglyceride‐glucose index ≥75 percentile. Primary outcome was defined as a composite of peripheral artery disease, coronary artery disease, myocardial infarction, ischemic stroke, or cardiovascular mortality. Scenario 1: Effect of triglyceride‐glucose index on incident cardiovascular disease between Lp(a) groups; Scenario 2: effect of Lp(a) on incident cardiovascular disease between triglyceride‐glucose index groups. aHR indicates adjusted hazard ratio; CAD, coronary artery disease; IR, insulin resistance; Lp(a), lipoprotein(a); MI, myocardial infarction; and PAD, peripheral artery disease.
    In the subgroups we included, compared with Lp(a) <125 nmol/L with low IR (Group 1, reference), co‐exposure to Lp(a) ≥125 nmol/L and high IR (Group 4) was associated with increased CVD risk. The aHRs for Group 4 were: men, 1.26 (95% CI: 1.16–1.36); women, 1.44 (95% CI: 1.28–1.60); individuals with hypertension, 1.29 (95% CI: 1.10–1.39); those without hypertension, 1.34 (95% CI: 1.20–1.51); individuals with CKD, 1.20 (95% CI: 1.03–1.41); and those without CKD, 1.31 (95% CI: 1.22–1.41) (Table 5).
    Association of Lipoprotein(a) and Insulin Resistance With Primary Outcome Onset in Subgroups.
    Low IR was defined as triglyceride‐index <75 percentile; high IR was defined as triglyceride‐index ≥75 percentile. Primary outcome was defined as a composite of myocardial infarction, coronary artery disease, peripheral arterial disease, ischemic stroke, or cardiovascular mortality. Model adjusted for age, sex, ethnicity, Townsend Deprivation Index, body mass index, smoking status, systolic blood pressure, high‐density lipoprotein cholesterol, statin therapy, and estimated glomerular filtration rate. aHR indicates adjusted hazard ratio; CVD, cardiovascular disease; IR, insulin resistance; and Lp(a), lipoprotein(a).
    Consistent with the main analysis, the sensitivity analysis that included further adjustment for glycated hemoglobin did not significantly change the results (Table S2). Co‐exposure to Lp(a) ≥125 nmol/L and high IR was associated with increased risk of CVD in individuals with and without diabetes (Table S3). Separate sensitivity analyses, excluding participants with triglyceride levels above 300 mg/dL or those on lipid‐lowering therapies showed no significant differences in the findings, as presented in (Tables S4 and S5).

    DISCUSSION

    In this prospective cohort of 320 803 participants from the UK Biobank, all free of CVD at baseline, we found compelling evidence of increased cardiovascular risk over 15 years of follow‐up, based on a combined baseline measure of Lp(a) and IR. First, Lp(a) and IR each independently contributed to cardiovascular risk, with additive value when both were considered together. Second, the combination of Lp(a) and the IR provided the greatest magnitude of risk for CVD. Finally, our results suggest that the CVD risk associated with Lp(a) was modified by IR.
    These findings have several significant implications for CVD prevention. First, blood biomarkers beyond low‐density lipoprotein cholesterol are gaining recognition within the clinical community for their role in understanding cardiovascular disease processes and guiding preventive interventions.33 Several ongoing trials are investigating the effectiveness of new agents for Lp(a) reduction on clinical outcomes.29 Therefore, it is important to examine the usefulness of including cardiometabolic health measures, such as IR, in the enhancement of CVD risk assessment beyond Lp(a) alone. It has long been known that IR promotes atherosclerosis through mechanisms such as hyperinsulinemia, chronic hyperglycemia, impaired adipose function, inflammation, and dyslipidemia.7 These processes may interact with the atherogenic potential of Lp(a), further increasing the risk of clinical CVD. TyG not only serves as a cost‐effective and convenient proxy for IR but has also been shown to predict CVD morbidity and mortality across diverse populations.34, 35 Notably, the majority of prospective cohort studies on the TyG’s association CVD have focused on Asian populations,36, 37, 38, 39, 40 with limited research conducted in other populations.41, 42, 43 But a recent UK Biobank cohort study found a similar association between TyG and CVD risk, supporting its predictive relevance beyond Asian cohorts.43
    Unlike insulin sensitivity, which can fluctuate over a person’s lifetime,7 in most adults, Lp(a) remains generally stable.3 Despite this, our study shows that even a single assessment of IR could also provide valuable additional information on CVD risk beyond what is obtained from measuring Lp(a) levels alone. Our findings are consistent with previous studies that demonstrated the ability of IR in forecasting CVD events years into the future.8, 44 This study builds on existing knowledge by demonstrating that in assessing CVD risk and selecting preventive interventions, clinicians can confidently evaluate both Lp(a) levels and IR simultaneously.
    We observed a suggestive interaction between log‐Lp(a) and TyG (P = 0.07), which met our pre‐specified threshold of P<0.1 for interaction testing. Further studies are needed to validate this finding and clarify its clinical relevance in CVD risk stratification. Regardless, comprehensive protection against Lp(a)‐associated CVD risk may require adjunctive strategies to improve insulin sensitivity. Several antidiabetic agents, including metformin, sulfonylureas, sodium‐glucose cotransporter 2 (SGLT2) inhibitors, dipeptidyl peptidase 4 (DPP‐4) inhibitors, thiazolidinediones, and glucagon‐like peptide 1 receptor agonists, have demonstrated effectiveness in improving insulin sensitivity.7 The use of glucagon‐like peptide‐1 receptor agonists, in particular, is expected to increase rapidly over the coming years because of recent clinical trials demonstrating their benefits for weight loss and cardiovascular risk reduction, even among individuals without diabetes.45 Multiple clinical trials have explored targeted interventions for IR,7 but currently there are no medications approved specifically for targeting IR. Lifestyle modifications, such as diet, physical activity, and weight loss for individuals who are overweight or obese, remain the cornerstone for improving insulin sensitivity, but have no effect on Lp(a) levels.1, 7 This study suggests the need for future research to explore whether medications that improve insulin sensitivity could benefit individuals with high Lp(a) for primary CVD prevention.
    In the subgroup analyses, the co‐exposure to elevated Lp(a) and IR was associated with higher CVD risk among women, younger individuals, and those without hypertension, obesity, or CKD at baseline. The relative impact of Lp(a) and IR appeared more pronounced in these traditionally lower‐risk groups. This pattern may reflect differences in baseline risk, medication use, or unmeasured factors, and warrants further investigation.
    While not the primary focus of our study, we found a minimal inverse correlation between Lp(a) level and IR. Data on the relationship between Lp(a) and IR are currently limited; however, a prior study has reported an inverse correlation between IR and Lp(a).46 Future studies are needed to confirm whether a similar paradoxical inverse relationship exists between Lp(a) and IR, akin to the suggested association between low Lp(a) and high diabetes risk.6 Nonetheless, we found that elevated Lp(a) levels in individuals with high IR, regardless of diabetes status, are associated with a higher CVD risk, perhaps mirroring the Lp(a)‐associated residual CVD risk observed in individuals with diabetes.6
    Previous studies have reported an inverse relationship between Lp(a) and triglyceride levels, particularly when triglyceride levels exceed 300 mg/dL.27, 28 In our cohort, only a small proportion of individuals (6.7%, n=21 941) had triglyceride levels above this threshold, and their exclusion did not alter our primary findings. This suggests that TyG, a triglyceride‐based surrogate for IR, can be reliably used to further stratify Lp(a)‐associated CVD risk. Additionally, our results were consistent in a cohort of patients who were taking lipid‐lowering medications which can affect glucose metabolism, triglyceride levels, and Lp(a).29, 30
    The limitations of our study should be considered. First, this study emphasizes that IR and Lp(a) are independent contributors to CVD risk, yet the observational design precludes conclusions about causality. Second, a random blood sample was used to measure TyG in UK Biobank participants. Although non‐fasting glucose was a limitation, we addressed it through sensitivity analysis adjusting for glycated hemoglobin and by excluding participants with diabetes in a repeated analysis. Third, the TyG was measured from a single blood sample at baseline, limiting our ability to evaluate how changes in the IR over time may impact CVD risk. Some studies suggest that variability in the IR may also be a strong predictor of CVD.47 Fourth, because of the observational design of this study, residual confounding effects cannot be completely ruled out, despite adjusting for several major confounding factors in our analysis. Specifically, given the close relationship between IR and diabetes, we conducted additional sensitivity analyses to confirm that this relationship did not confound our findings. Finally, the UK Biobank cohort is not nationally representative and may be subject to a “healthy volunteer” selection bias. The UK Biobank also consists predominantly of individuals of European ancestry, resulting in limited racial diversity within the study population. This lack of racial diversity is significant given that Lp(a) levels vary across racial groups, with Black individuals generally exhibiting higher levels and Hispanic individuals lower levels compared with White individuals.48 However, valid assessments of associations between exposures and cardiovascular outcomes can often be widely generalizable, even if participants are not fully representative of the general population.

    CONCLUSION

    We found that a single measurement of lipoprotein(a) and insulin resistance, assessed using the triglyceride‐glucose index, independently predict incident cardiovascular events in individuals free of CVD. These data support accounting for insulin resistance when assessing Lp(a)‐related CVD risk.

    Sources of Funding

    Dr Bhatia is supported by the National Institutes of Health, Grant 1K08HL166962. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

    Disclosures

    Dr Bhatia reports consultant/advisor relationships with Kaneka, Novartis, Arrowhead and Abbott. Dr Shapiro is supported by institutional grants from Amgen, Arrowhead, Boehringer Ingelheim, 89Bio, Esperion, Novartis, Ionis, Merck, and New Amsterdam; and he has participated in Scientific Advisory Boards with Amgen, Agepha, Ionis, Novartis, New Amsterdam, and Merck. He has also served as a consultant for Ionis, Novartis, Regeneron, Aidoc, Shanghai Pharma Biotherapeutics, Kaneka, Novo Nordisk, Arrowhead, and Tourmaline. Dr Mehta is supported by institutional grants from Amgen and Novartis. Dr Mirzai is supported by the NHLBI of the NIH (T32‐HL‐076132). The remaining authors have no disclosures to report.

    Supporting information

    Tables S1–S5.
    Figures S1–S2.
    Strobe Checklist.
    AMDr. Anurag Mehta

    Dr. Anurag Mehta

    Cardiology · Preventive Cardiology
    Emory University
    anuragm