PUTRA-CV study protocol: a multicentre observational study of ethnic-specific genetic variants and dietary patterns in relation to lipoprotein(a) levels and their association with coronary artery disease severity in Malaysian adults

Abstract

Introduction

Although low-density lipoprotein cholesterol (LDL-C) is established as the primary cardiovascular disease (CVD) risk factor, some individuals with LDL-C within desirable limits still develop coronary artery disease (CAD). Lipoprotein(a) (Lp(a)) has emerged as a genetically determined independent risk factor for CVD. This study aims to investigate Lp(a) by determining its association with coronary artery stenosis severity, identifying its ethnic-specific genetic determinants and assessing its relationship with an energy-dense dietary pattern.

Methods and analysis

The PUTRA-CV study is a 3-year, multicentre, case-control observational study involving adult patients who have undergone coronary angiography. The primary outcome is the association between Lp(a) levels and the severity of angiographic CAD (assessed by Gensini or Syntax score). Secondary outcomes include the frequencies of Lp(a)-associated single nucleotide polymorphisms (SNPs) (rs10455872 and rs3798220) and the association between dietary patterns and Lp(a) levels. Lp(a) will be measured using a particle-enhanced immunoturbidimetric method, and SNPs will be genotyped using high-resolution melting. Dietary intake will be assessed using a validated semiquantitative food frequency questionnaire. Data will be analysed using SPSS. Descriptive statistics will be used to summarise population characteristics. Bivariate analyses will use chi-square (χ2), independent t-tests or Mann-Whitney U tests as appropriate. The independent association between Lp(a) and coronary artery stenosis severity will be determined using multivariable logistic regression, adjusting for confounders. Empirically driven dietary patterns will be derived using reduced rank regression, and their association with Lp(a) will be assessed. For genetic analysis, allele frequencies of the LPA SNPs rs10455872 and rs3798220 will be calculated and compared between cases and controls.

Ethics and dissemination

Ethical approval has been obtained from the ethics committees of the Ministry of Health Malaysia (NMRR ID-24-00877-2ID-IIR), Universiti Putra Malaysia (JKEUPM-2024–246), Universiti Teknologi MARA (REC/07/2024-OT/FB/2) and Universiti Malaya Medical Centre (MREC ID NO: 2 02 453–13692). The findings will be disseminated via peer-reviewed journals and conferences.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study employs matched case-control design using angiographically confirmed coronary artery disease (CAD) status, allowing for a well-defined and direct investigation of the association between Lipoprotein(a) (Lp(a)) and CAD severity.

• The study includes genetic analysis for ethnic-specific single nucleotide polymorphism frequencies related to Lp(a) and CAD within a multi-ethnic Malaysian cohort (Malay, Chinese, Indian), offering new opportunities for cardiovascular disease (CVD) risk stratification and supporting Lp(a) as an independent CVD risk marker.

• The study uses a validated semiquantitative food frequency questionnaire combined with reduced rank regression analysis to provide a methodologically rigorous approach to identify whole dietary patterns associated with Lp(a) levels, moving beyond single-nutrient analysis.

• Multicentre representative data collection that covers the regions of Greater Kuala Lumpur (Universiti Teknologi MARA Specialist Centre (UiTMSC), Northern region; Universiti Malaya Medical Centre (UMMC); Central region and Serdang Hospital Heart Centre (SHHC), Southern region).

• A potential limitation of this study is intercentre variability in patient management protocols, which may lead to inconsistencies in the data collected, while the self-reporting of certain predictor variables may be subject to recall bias.

Introduction

Background

Coronary artery disease (CAD) is the leading cause of death globally. While low-density lipoprotein cholesterol (LDL-C) is an established risk factor, emerging evidence challenges its conventional prominence in CAD. Notably, individuals with normal-range LDL-C levels can still develop CAD. This residual cardiovascular disease (CVD) risk refers to the likelihood of cardiovascular (CV) events occurring despite treatment or the attainment of target levels for risk factors such as glycaemia, LDL-C and blood pressure (BP).¹

Lipoprotein(a) (Lp(a)) has emerged as an independent risk factor for atherosclerosis, CAD, stroke and aortic valve stenosis, with levels primarily governed by genetics, as demonstrated by epidemiological and genetic studies in diverse populations.² Lp(a) resembles LDL-C but carries an additional apolipoprotein(a) (apo(a)) moiety, conferring pro-atherogenic, pro-thrombotic and pro-inflammatory properties. It promotes plaque formation through enhanced arterial retention and oxidised phospholipid delivery while impairing fibrinolysis via plasminogen mimicry.³ Elevated Lp(a) levels therefore represent a potent genetic risk factor and an important unresolved therapeutic challenge in CVD.⁴

The LPA gene encodes the apo(a) and controls Lp(a) concentrations. The heritability of apo(a) varies among ethnicities.^{2 5} Populations of African descent have been shown to have the highest Lp(a) levels,⁶ while considerable variability exists among Asian populations.⁷ Compared with East Asians, who generally have lower Lp(a) concentrations,⁸ South Asians often exhibit higher levels comparable to those of Europeans.^{9 10} The PUTRA-CV (PUTRA denotes Universiti Putra Malaysia, the institution of the principal investigator of this study, while CV refers to cardiovascular) study is particularly relevant given Malaysia’s unique ethnic composition of Malays, Chinese and Indians. Findings from a study in Singapore support this ethnic-specific distribution, with Indians having the highest median Lp(a) levels, followed by Malays and Chinese having the lowest levels.⁷

Lp(a) levels typically remain consistent throughout an individual’s life.⁴ Understanding elevated Lp(a) presence is crucial because heightened levels increase the risk of atherosclerotic CVD (ASCVD), providing valuable information for clinical decision-making in risk management independent of conventional CVD risk factors. Evidence suggests that elevated Lp(a) levels, particularly exceeding 75 nmol/L (30 mg/dL), contribute to atherosclerotic plaque progression.³ Furthermore, once an individual with elevated Lp(a) is identified, cascade screening among family members can reveal additional cases due to its autosomal codominant inheritance pattern. International guidelines recommend measuring Lp(a) once in a patient’s lifetime.^{2 3 5 11 12}

To date, studies in Malaysia have only focused on Lp(a) concentrations between native ethnic communities,¹³ familial hypercholesterolaemia patients¹⁴ and atherosclerotic peripheral vascular disease.¹⁵ About 90% of plasma Lp(a) concentration is genetically determined and, therefore, hereditary,416,18 offering opportunities for risk stratification. However, genetic studies on Lp(a) have not been done in Malaysia. Hence, the PUTRA-CV study is the first in Malaysia to determine the ethnic-specific single nucleotide polymorphism (SNP) frequencies associated with Lp(a) in relation to CAD and establish the association between Lp(a) levels and coronary artery stenosis severity (measured by Gensini score or Syntax score) in patients undergoing angiogram.

While current evidence regarding dietary influences on Lp(a) levels remains contradictory, existing studies have primarily focused on isolated nutrients rather than comprehensive dietary patterns.^{19 20} Notably, research has shown that certain LDL-C-lowering strategies, particularly the replacement of saturated fats with carbohydrates or monounsaturated fats, may paradoxically elevate Lp(a) concentrations. This study addresses critical methodological and evidence gaps by being the first Malaysian investigation to examine empirically derived dietary patterns in relation to Lp(a), with specific focus on an ‘energy-dense, high sugar and saturated fat’ dietary pattern. This pattern was selected based on its established pro-inflammatory properties and association with insulin resistance pathways, both of which may influence Lp(a) metabolism.²¹ Unlike previous nutrient-specific approaches, our study employs reduced rank regression (RRR) to capture the complex interactions of whole dietary patterns, thereby providing more clinically relevant insights into how overall eating behaviours may modulate this genetically determined CVD risk factor.

There is now a substantial body of epidemiological and genetic evidence that strongly supports a causal link between high Lp(a) levels and ASCVD, leading to the development of novel small interfering RNA (siRNA)-based therapies that can reduce Lp(a) by over 90%. Ongoing randomised clinical trials are also evaluating whether reducing Lp(a) can lower CV events.^{4 22 23} Hence, this novel data could be used to identify the residual CVD risk beyond LDL-C and justify the need for Lp(a) measurement as an independent CVD risk marker in Malaysia’s multi-ethnic population, considering the rapid development in Lp(a) targeted novel therapies.

Aims

The aims of this study are to determine the: (1) baseline characteristics (sociodemographics, clinical and laboratory parameters) of the study population; (2) association between Lp(a) and coronary artery stenosis severity in post-coronary angiography patients; (3) ethnic-specific SNP frequencies associated with Lp(a) in post-coronary angiography patients and in a control group; and (4) association between adherence to an empirically driven dietary pattern, high ‘energy dense, percentage energy from sugar and saturated fat’ and Lp(a) in study subjects.

Hypotheses

The following hypotheses will be tested:

• There is significant association between Lp(a) levels and severity of coronary artery stenosis.

• There are ethnic-specific SNP frequencies associated with Lp(a) in the Malaysian population.

• There is significant association between adherence to an empirically driven dietary pattern, high ‘energy dense, percentage energy from sugar and saturated fat’ and Lp(a) levels.

Method and analysis

Study settings, location and target population

Malaysia is a multi-ethnic country, and this observational study will recruit participants from the three major ethnic groups: Malays, Chinese and Indians. It is a multicentre, hospital-based case-control study conducted at three sites: Serdang Hospital Heart Centre (SHHC), Universiti Malaya Medical Centre (UMMC) and Universiti Teknologi MARA Specialist Centre (UiTMSC), which together serve the Greater Kuala Lumpur region. The target population consists of Malaysian adults aged 18 to 80 years. They are patients who have undergone coronary angiography and fit the eligibility criteria. The inclusion criteria for cases are: (1) post-coronary angiography patients with CAD. The inclusion criteria for controls are: (1) patients with no CAD confirmed by normal coronary angiography. Relevant exclusion criteria for both cases and controls include (1) non-fasting subjects, (2) pregnant women, (3) non-Malaysian citizens (to ensure genetic homogeneity and reduce confounding from unrelated ancestries) and (4) clinical evidence or history of familial hypercholesterolaemia, nephrotic syndrome, stage 4 or 5 chronic kidney disease (estimated glomerular filtration rate <30 mL/min/1.73 m²), liver dysfunction or thyroid disorders.

Sample size determination

The sample size is determined using a case-control matched pair calculation to estimate the number of subjects required to detect an odds ratio (OR) of 2 with an alpha (α) value of 0.05. The detailed sample size calculation is provided in online supplemental table S1. Table 1 summarises the sample size calculation using the two different SNPs.

Table 1. Sample size calculation³⁹.

SNP	P1	P2	Pe	m	M
rs3798220	0.0365	0.072	0.1033	91	881
rs10455872	0.1304	0.1795	0.2631	91	346

SNP, single nucleotide polymorphism.

Formula 1: M = m/p_e

where,

1. M is matched pairs

2. m is sample size

3. p_e is the probability of a discordant pair

4. m = [Z1-α/2/2 + Z1-β(P*(1-P*))^1/2]² / (P*−1/2)²

where,

• α is 0.05 (2-tailed test)

• Z1-α/2 is 1.96

• β is 10% (type II error)

• Z1-β is 1.282

• P* is OR/(1+OR), where OR is 2

• 5 1p_e, the probability of a discordant pair = (p₂(1-p₁) + p₁(1-p₂))

where,

1. p₁, proportion of exposed controls = frequency of disease allele in controls

2. p₂, proportion of exposed cases = frequency of disease allele in cases

Using the larger estimate, a total of 1762 subjects is required. To account for eligible individuals who might decline participation or provide incomplete data during the single study visit, a 5% non-response rate was incorporated into the sample size calculation:

Adjusted sample size = 1762/(1–0.05) = 1762/0.95 ≈ 1855 subjects

Accordingly, we plan to approach approximately 1855 eligible individuals, anticipating that about 5% will decline or provide unusable data, thereby achieving the target sample of 1762 participants. As the study involves two groups (cases and controls) across three centres, we aim to recruit approximately 309 cases and 309 age- (± 5 years) and sex-matched controls from each site: UMMC, SHHC and UiTMSC, resulting in a total of 927 cases and 927 controls. This approach ensures that the final analysable sample will meet or exceed the required size, thereby providing adequate statistical power.

Approval and recruitment

Ethical approvals to conduct this study were granted by the Universiti Putra Malaysia (UPM)’s Ethics Committee for Research Involving Human Subjects (JKEUPM) followed by Medical Research and Ethics Committee (MREC) of the Ministry of Health Malaysia as well as the UiTM Research Ethics Committee and UMMC Medical Research Ethics Committee. These approvals were obtained before contacting the cardiologists, who will also serve as co-investigators, to coordinate subject recruitment, schedule dates and times and plan the recruitment process with co-investigators from all three centres.

Adult patients aged 18 to 80 years who are scheduled to undergo coronary angiography will be invited to join the study. They will undergo screening for the exclusion criteria outlined above, based on retrospective laboratory data, medical history and medication regimens obtained from the Hospital Information System (HIS). Participants who either declare or are found to have one or more of the above conditions will be excluded from the study. Written informed consent will be obtained from eligible participants who agree to join the study.

Participants will be informed that their involvement in the study is entirely voluntary and that they may withdraw consent at any time without penalty or loss of benefits to which they are otherwise entitled. Their decision to participate or withdraw will not affect their medical care. No compensation or token of appreciation will be provided for participation; however, participants may request to be informed of their Lp(a) blood test results.

Data collection

Data collection is ongoing from September 2024 and will continue through August 2027. Data collection is currently underway, starting at one centre and progressing to the next once the desired sample size for both controls and cases is achieved. The estimated duration at each centre depends on the response and workload, leading to a variation of the duration at each centre. The first batch of participants is from UiTMSC, followed by SHHC and UMMC.

After a thorough verbal explanation of the study, the participants who agree and sign the consent form will be requested to fill out a self-administered semiquantitative food frequency questionnaire (FFQ). The FFQ is adapted from the Malaysian Adults Nutrition Survey (MANS) 2014, used in other studies.²⁴

Ten millilitres of blood will be collected from the participant in serum separator tube (SST) and ethylenediaminetetraacetic acid (EDTA) tubes by a trained phlebotomist for Lp(a) and genetic analysis, respectively. Participants will be informed about the possible risks of the study, which are similar to those associated with venepuncture. These risks include excessive bleeding (especially for those taking anticoagulants), fainting or feeling light-headed, haematoma, infection and the need for multiple punctures to locate the veins. Measures aimed at mitigating risks and ensuring participant safety and comfort during and after venepuncture include having skilled personnel to conduct the procedure, thoroughly cleaning the puncture site with an alcohol swab before the procedure, using sterile equipment to prevent infection, applying immediate pressure to avoid excessive bleeding and applying ice and pressure to the affected area to reduce swelling and prevent haematoma. If the participant experiences faintness or light-headedness, they will be asked to lie down, and fluids will be provided if necessary. Participants with elevated Lp(a) levels will be referred to their cardiologist, who is also a co-researcher in this study, for timely expert management. Figure 1 presents a visual summary of the study flow.

Figure 1. The data collection process in the PUTRA-CV Study. FFQ, food frequency questionnaire; Lp(a), lipoprotein(a); SHHC, Serdang Hospital Heart Centre; UiTMSC, Universiti Teknologi MARA Specialist Centre; UMMC, Universiti Malaya Medical Centre.

Pro forma documentation 

The pro forma questionnaire comprises sociodemographic factors (age, gender, race, education level, smoking history, alcohol intake and physical activity). Trained medical personnel will be asked to adhere to standardised protocols before data collection to reduce variability in anthropometric measurements across the three study sites. These measurements will include weight, height, waist circumference (WC) and BP assessment. Subjects will be weighed to the nearest 0.1 kg on a calibrated weighing scale. Standing height will be measured to the nearest 0.1 cm with a wall-mounted stadiometer. Body mass index will be calculated using the formula: weight/height² (kg/m²). WC will be measured using a flexible plastic tape in between ribs and hip bone. Resting BP will be measured using an automated BP monitor (Omron, Omron Healthcare, Ic., Illinois, USA). Two consecutive readings, taken 30 s apart, will be recorded while the subject is seated, with measurements taken on the left arm. The average of the two systolic and diastolic readings will be recorded. The medical history, medications, post-angiography findings and laboratory data will be recorded. All the data in the pro forma will be obtained from the patient records in the HIS of the respective centres.

Biochemical data

Retrospective laboratory data during the angiogram will be electronically retrieved from the subjects’ medical records. These data include full blood count, liver function test, renal profile, fasting plasma glucose, haemoglobin A1c, fasting serum lipid, thyroid stimulating hormone, free thyroxine and high-sensitivity C-reactive protein. The data will be reviewed to ensure that subjects meet the inclusion criteria. LDL-C will be derived using the Friedewald equation, non-high-density lipoprotein cholesterol (non-HDL-C) will be obtained by subtracting HDL-C from the total cholesterol value and the Atherogenic Index of Plasma will be calculated with the formula log (triglycerides divided by HDL-C).

Lipoprotein(a) (Lp(a)) analysis

After an overnight fast, a blood sample for Lp(a) will be obtained from the subject’s cubital vein into a SST tube. The collection tube will be transported immediately in an icebox to the Chemical Pathology Laboratory, Faculty of Medicine and Health Sciences, UPM. Samples for Lp(a) will be centrifuged and aliquoted on the same day and sera will be stored at −80°C for no longer than 6 months prior to batch analysis. Frozen samples at −80°C will be thawed at room temperature for analysis. The assay uses a particle-enhanced immunoturbidimetric method on the Roche Cobas c system (Roche Diagnostics, Mannheim, Germany), where human Lp(a) binds to latex particles coated with anti-Lp(a) antibodies, leading to agglutination. The resulting turbidity is then quantified at 659 nm. The measurement range extends from 7 to 240 nmol/L, with a detection limit of 7 nmol/L. Samples exceeding 240 nmol/L will be diluted for further analysis. Calibration and quality control (QC) will be performed in conjunction with normal laboratory operations. A threshold value of 75 nmol/L (30 mg/dL) will be considered indicative of elevated CVD risk.³ The intra-assay and inter-assay coefficients of variation (CV) for Lp(a) were 2.4% and 2.4% at level 1 QC (32 nmol/L) and 1.0% and 1.1% at level 2 QC (101 nmol/L), respectively.

Genetic analysis

Genetic analysis will be performed to identify specific genetic variants or polymorphisms associated with elevated Lp(a) levels. Deoxyribonucleic acid (DNA) will be extracted from buffy coat leukocytes of a venous blood sample collected in an EDTA tube using the commercially available QIAmp Blood Mini Kit (QIAGEN) and transported to the Genetics Laboratory, Faculty of Medicine and Health Sciences, UPM. A spectrophotometer (PCR MAX Lambda 63500, Bibby Scientific Ltd, UK) will be used to determine the concentration and purity (1.7–1.9) of DNA samples, and 1% agarose gel electrophoresis will be used to analyse the integrity of the DNA. The LPA SNPs rs10455872 and rs3798220 will be analysed using high-resolution melting (HRM). Real-time polymerase chain reaction (PCR) will be conducted using the LightCycler 480 system (Roche Diagnostics, Mannheim, Germany). After amplification, the HRM analysis will be carried out using the LightCycler High Resolution Melting Master (Roche Diagnostics, Mannheim, Germany). The different melting points will be observed using normalised graphs, and HRM analysis software will be used to determine the genotypes. To date, the PCR amplification rate has been 100% on all genomic DNA samples that passed the quality check. As a crucial data quality assurance step, after HRM analysis, a total of 10% of the samples from each predicted genotype will be confirmed with Sanger sequencing. The observed genotype frequency in both controls and cases will then be tested for any significant deviation from Hardy-­Weinberg equilibrium.

Angiographic evaluation of coronary artery disease (CAD) severity

This retrospective data will be obtained electronically from patient records. The angiographic severity of CAD will be assessed using the score determined by the interventional cardiologist who performed the procedure. Two primary scoring systems for evaluating CAD severity are Syntax and Gensini. Syntax score can be calculated using the online tool available at (Syntax Score Calculator) (http://syntaxscore.org/calculator/start.htm). It is an angiographic grading system designed to assess the complexity of CAD. The Syntax Score is the total of points allocated to each coronary lesion with a diameter narrowing exceeding 50% in vessels larger than 1.5 mm.²⁵ In contrast, the Gensini score evaluates the extent of CAD by assigning severity points to each lesion based on the degree of stenosis and the importance of the lesion’s location.^{21 26 27}

In-person interviews of dietary intake using a semiquantitative food frequency questionnaire (FFQ)

The 165-item FFQ, adapted from the MANS 2014, will be used to evaluate participants' habitual food intake during the preceding year and takes approximately 30 to 45 min to complete. For analysis, these items will be collapsed into 13 food groups, classified according to their nutrient profiles or culinary usage. The NutriPro software will be used to assess the hypothesised contribution to coronary artery stenosis severity.

Individual interviews will be conducted to evaluate the frequency of consumption for each food item, with responses recorded as the number of times per day, per week, per month or never. Only one option is allowed for each food item selected. Participants will additionally be required to indicate the number of servings consumed each time they intake the food items listed in the FFQ. Each food item will be allocated a standard serving size (medium) based on the Album Saiz Sajian Makanan Malaysia (Food Portion Sizes of Malaysian Foods Album, 2002/2003). Serving sizes for food items such as fruits, meats and beverages will be estimated using household measurement aids, including the number of pieces or whole fruits, spoons, cups, matchbox-sized portions and other familiar references. However, for elderly participants or those who need assistance, the researcher will help by asking questions, obtaining answers and filling out the FFQ on their behalf. The conversion of food frequency to the quantity of food consumed (grams or calories) will be carried out using the following formula:

Amount of food (g) per day = frequency of intake (conversion factor) × serving size × total number of servings × weight of food in one serving (Wessex Institute of Public Health 1995)

Dietary misreporting is prevalent in studies examining diet and health outcomes.²⁸ To mitigate this, a standardised equation will be employed, based on the ratio of energy intake to total energy expenditure, with 95% confidence limits as cut-offs to assess the extent of dietary misreporting.²⁹ In the prospective modelling, the variable of dietary misreporting, which includes plausible, over- and under-reporting, will be linked to possible covariates.

Dietary pattern

This study will use RRR to identify dietary patterns characterised by high energy density, sugar and saturated fat, key factors implicated in Lp(a) modulation through inflammatory and metabolic pathways. Unlike exploratory methods like principal component analysis, RRR offers distinct advantages by incorporating a priori biological knowledge through carefully selected response variables (dietary energy density, percentage energy from sugar and saturated fat) that serve as plausible mediators between diet and CVD risk. The RRR approach mathematically derives dietary patterns that maximally explain variance in these specific intermediate biomarkers, creating a more direct link to the biological pathways of interest than purely data-driven methods.³⁰ By focusing on 13 predefined food groups (online supplemental table S2) as predictors and these three evidence-based dietary response variables, our analysis bridges the gap between broad dietary patterns and their specific metabolic effects. The first derived RRR pattern is particularly meaningful as it captures the greatest proportion of variance in our targeted dietary mediators. This methodology aligns perfectly with our study aims, as it allows us to move beyond simple associations to identify dietary exposures that directly influence the metabolic pathways potentially affecting Lp(a) concentrations, while avoiding the limitations of single-nutrient approaches that have produced inconsistent findings in previous research. The selected response variables reflect current scientific understanding of how energy-dense, high-sugar, high-saturated fat diets may promote oxidative stress, hepatic lipogenesis and inflammation, all mechanisms potentially relevant to Lp(a) metabolism and CVD risk.1931,33

Given the current understanding of the relationship between diet and CV health, we anticipate that RRR would identify patterns linked to our primary outcome of interest: Lp(a) concentration. In RRR, the number of dietary patterns identified always equals the number of response variables used in the model. The first pattern derived by RRR is optimal because it always explains the most variation in response variables compared with other derived patterns. Individual scores or z-standardised intakes for each dietary pattern are calculated by combining standardised food intakes with scoring weights (scoring coefficients) as shown in the conceptual equation below:

Dietary pattern score = β1FG1 + β2FG2 + β3FG3 …… (n number of food groups) where

β = weight or scoring coefficient for each food group, derived from RRR (eigenvectors of the covariance matrix)

FG = individual’s standardised intake of food groups in grams (FG1 indicates first food group and FG2 as the second food group and so on).

In particular, the coefficients obtained from the RRR analysis for each food group will be multiplied by the individual’s intake (in grams per serving) of those food groups. For instance, if fried potatoes have a high positive coefficient in the dietary pattern, an individual who consumes a large quantity of fried potatoes will have a higher score for that pattern. Every individual in the study will have a score for each pattern, which discriminates how strongly the dietary intake corresponds to each pattern. These scores are usually computed as z-scores, with a mean of zero and a variance of one, ensuring that differences in food group intake do not affect the findings. Individuals with negative z-scores imply lower adherence to the derived dietary pattern, while those with positive z-scores are more likely to adhere.

Data analysis

Statistical calculations will be performed using the standard software package, IBM Statistical Package for the Social Sciences (SPSS) for Windows, V.29.0. Armonk, NY: IBM Corp. In descriptive analysis, mean and standard deviation (SD) will be used to summarise normally distributed continuous variables, and median and interquartile range (IQR) will be presented for skewed distribution, whereas count (n) and percentage (%) will be calculated for categorical variables. The association between two categorical variables will be analysed using the chi-square (χ2) tests and independent t-test (parametric)/Mann-Whitney U test (non-parametric) will be used for the comparison of continuous variables between two groups. Logistic regression will be used to evaluate the association between angiographic CAD and Lp(a) levels, followed by stepwise multivariate logistic regression to assess the impact of additional CVD risk factors on Lp(a). Adjustments for potential confounders (eg, sociodemographic variables) will be made. Standardised regression coefficients and ORs with 95% confidence intervals (CIs) will be reported.

The RRR analysis will be conducted with the partial least squares procedure in the Statistical Analysis System (SAS) software (SAS Institute). The extracted dietary patterns are uncorrelated due to the orthogonality of eigenvector scores. Therefore, different dietary patterns can be used in multivariate analyses as independent variables to evaluate the relationship between dietary patterns, diseases of interest and other population characteristics without confounding each other.

The allele frequencies of the SNPs rs10455872 and rs3798220 will be determined.³⁴ In 2009, two LPA variants, rs10455872 (AG/GG) and rs3798220 (CT/CC), were linked to elevated levels of Lp(a), a lower copy number in LPA (which influences the frequency of kringle IV–type 2 repeats), a smaller Lp(a) size and an increased risk of CAD.³⁵ These two SNPs also ‘tag’ about half of the small apo(a) isoforms.⁴

Data management and confidentiality

Data will be managed in accordance with the Data Protection Act. All information collected via the pro forma and FFQ will remain confidential as no personal identifiers will be recorded. Data will be entered into a password-protected computer and upon study completion, transferred to encrypted external drives (eg, USB flash drives or external hard drives) before being deleted from the computer. The drives will be stored in a locked office for 7 years, after which the data will be permanently deleted.

Discussion

This study aims to better understand the importance of Lp(a) in CVD within the context of Malaysia’s multi-ethnic population. The National Health and Morbidity Survey (NHMS) is a nationwide, population-based survey that provides critical data on Malaysia’s disease burden, health problems, healthcare needs and expenditures. The 2023 cycle focused on non-communicable diseases (NCDs) and revealed concerning trends in conventional CVD risk factors, such as the increase in diabetes prevalence from 11.2% in 2011 to 15.6% in 2023. Hypertension prevalence stood at 29.2%, while overweight and obesity rates grew from 44.5% in 2011 to 54.4% in 2023.³⁶ In this context, the study on Lp(a) in Malaysia will play a critical role in identifying residual CVD risk that persists beyond these conventional factors, highlighting Lp(a) as an important CVD risk marker. This is particularly relevant in diverse populations where LDL-C alone may not fully capture an individual’s CVD risk profile.

The study also aims to explore dietary patterns linked to Lp(a) levels, focusing on high energy-dense foods, sugar and saturated fat intake. The cross-sectional design of our dietary assessment, while valuable for identifying associations between dietary patterns and Lp(a) levels, inherently limits our ability to establish causal relationships. By capturing dietary intake and Lp(a) concentrations at a single time point, we cannot determine the temporal sequence of exposure and outcome, specifically whether observed dietary patterns preceded or resulted from Lp(a) variations. This design also precludes assessment of potential longitudinal effects of dietary modifications on Lp(a) metabolism. Future prospective cohort studies with repeated dietary and Lp(a) measurements would be necessary to elucidate the potential causal nature of these relationships and evaluate the long-term impact of dietary interventions on Lp(a) concentrations. Understanding these associations can inform dietary recommendations to reduce global CVD risk and support public health guidelines and targeted nutritional interventions. This study is also aligned with the global third sustainable development goal, which aims to ensure healthy lives and promote well-being for all, particularly by reducing early mortality from NCDs.

This study, conducted in Malaysia’s multi-ethnic population of Malays, Chinese and Indians, provides insights that may be relevant to other Southeast Asian countries with similar ancestral backgrounds and rising burdens of NCDs, particularly CVD. However, the applicability of these findings is limited by regional differences in genetic admixture. Further validation in other well-characterised Southeast Asian cohorts is needed to guide the development of tailored, context-specific strategies for CVD risk assessment and management.³⁷

International guidelines recommend measuring Lp(a) once in a patient’s lifetime, especially for those at elevated risk of CVD, particularly when other risk factors such as a strong family history of early heart disease or established CV conditions are present.^{2 3 5 6} In Malaysia, the Clinical Practice Guidelines for the Management of Dyslipidemia suggest considering Lp(a) testing for patients with recurrent CV events, if the assay is available.³⁸ Investigating ethnic-specific SNP frequencies associated with Lp(a) and CAD offers a promising avenue for enhancing CVD risk stratification.² Genotyping these SNPs can lead to more precise risk assessments and treatment strategies tailored to the multi-ethnic Malaysian population. This approach emphasises the importance of incorporating Lp(a) measurements into routine risk evaluations, particularly in light of emerging Lp(a)-targeted therapies, which could provide new opportunities for managing CVD risk more effectively.

The siRNA agent olpasiran and the antisense oligonucleotide pelacarsen have demonstrated reductions in Lp(a) levels of over 90% and approximately 80%, respectively. Large-scale phase 3 outcome trials are underway, including the fully enrolled Lp(a) HORIZON trial (pelacarsen; >8000 participants)²² and the OCEAN(a) trial (olpasiran),²³ which are expected to report results around 2025. These studies will provide definitive evidence on whether Lp(a) lowering translates into reduced CV events. In this context, our local findings could help identify residual CVD risk beyond LDL-C and support the inclusion of Lp(a) measurement as an independent CVD risk marker in Malaysia’s multi-ethnic population.

Ethics and dissemination

Ethical approvals to conduct this study were first granted by the UPM’s Ethics Committee for Research Involving Human Subjects (JKEUPM-2024–246), followed by MREC of the Ministry of Health Malaysia (NMRR ID-24-00877-2ID- IIR), UiTM Research Ethics Committee (REC/07/2024-OT/FB/2) and UMMC Medical Research Ethics Committee (MREC ID NO:2 02 453–13692). Verbal and written consent will be obtained voluntarily from all participants before the data collection. The findings will be disseminated via peer-reviewed journals and conferences.