Determination of the biological variation and reference change value of lipoprotein (a)

Abstract

Background

Understanding lipoprotein (a) [Lp(a)] measurement variability is essential in establishing its coronary heart disease (CHD) association, and optimizing assessment and management of atherosclerotic cardiovascular disease (ASCVD) risk. We established the components of biological variation (BV) and reference change value (RCV) of Lp(a) in a UK cohort.

Method

22 healthy individuals were recruited to the study. Blood samples were collected for six consecutive weeks and analysed in duplicate using the Lp(a) assay by Sentinel Diagnostics on the Beckman Coulter AU5800. Outlier, heterogeneity, normality, and trend analysis were performed, followed by CV-ANOVA to determine estimates of BV, adhering to the 14 BIVAC quality items. RCV was calculated based on estimated CV_A and CV_I.

Results

Four participants were excluded from the analysis as their mean Lp(a) levels fell below the functional sensitivity of the assay. Mean Lp(a) concentration ranged from 14 to 241 nmol/L. The overall estimate of CV_I for all participants was 10.9% (95% CI of 9.1 – 13.0%). The RCV for Lp(a) was +31.6%/-24.0%.

Conclusion

Our study obtained a CV_I estimate for Lp(a) that aligned consistently with recent studies adhering to the quality specifications outlined in the BIVAC checklist. The CV_I estimate was significantly lower than Lp(a) estimates reported in studies up to 2003. The CV_I estimate highlights the limitations of relying solely on a single Lp(a) measurement for prognosticating ASCVD risk and identifying candidates for novel Lp(a) therapies, particularly when the measured value is near clinical decision thresholds.

Keywords

Lipids analytes immunoassay laboratory methods cardiology clinical studies

Introduction

Lipoprotein (a) [Lp(a)] is a distinctive plasma lipoprotein composed of a low-density lipoprotein (LDL) particle, with apolipoprotein B-100 (ApoB-100) covalently linked to apolipoprotein(a) [apo(a)] by a disulphide bond.¹ The LPA gene primarily determines Lp(a) concentration in plasma.² Apo(a) comprises loop-like structures called kringles, stabilized by three internal disulphide bonds, with variability in the size of the apo(a) isoforms mainly due to number of kringle IV type-2 (KIV₂) repeats, with molecular mass ranging from 250 to 800 kDa.³ There is an inverse correlation of isoform size with Lp(a) concentration.⁴ Larger isoforms are associated with lower Lp(a) concentrations, likely due to an increase in the transit time through secretory pathway from hepatocytes and increased degree of pre-secretory degradation.⁵

Epidemiological studies have shown a consistent association between elevated Lp(a) levels and increased risks of myocardial infarction (MI), coronary heart disease (CHD), stroke, peripheral arterial disease and aortic valve stenosis, due to its pro-atherogenic and pro-inflammatory properties.^6–8 In the Emerging Risk Factors Collaboration, each standard deviation increase in Lp(a) concentration was linked to a 16% higher risk of CHD events.⁹ In the Copenhagen City Heart Study, individuals with Lp(a) levels in the 67th to 90th percentile, 90th to 95th percentiles, and above the 95th percentile had a 1.6, 1.9, and 2.6 times higher risk of myocardial infarction (MI), respectively, compared to those below the 22nd percentile.¹⁰ Doubling the Lp(a) concentration was associated with a 22% increased risk of MI. The Bruneck study found that higher Lp(a) levels significantly improved CHD risk prediction, with a 2.4-fold increased risk for those in the top quintile compared to the lower quintiles, even after adjusting for common risk factors.¹¹

Specific therapies targeting Lp(a) concentration have highlighted the need for accurate Lp(a) measurements, complicated by the unique structure of apo(a). KIV₂ repeats can result in significant challenges in measuring Lp(a).¹² Antibodies directed against the KIV₂ motif can recognize the protein multiple times, resulting in measurement bias. This leads to the underestimation of serum concentrations for small isoform Lp(a), while overestimating those of large isoform Lp(a).¹³ Lp(a) concentration should ideally be measured using a method where the effect of isoform size has been minimized using appropriate antibodies with calibrators certified for traceability of Lp(a) values to the WHO/IFCC reference material.¹⁴ The results should also ideally be expressed in particle units, and conversion from mass units should be strongly discouraged.¹⁴

Measurement variability profoundly influences the robustness of disease associations, particularly in the context of lipoproteins. The variability in measurement can introduce uncertainty and conflicting estimates regarding the correlation of a specific biochemical risk factor with CHD. Biological variation (BV) encompasses the intra-individual coefficient of variation (CV_I), which is the natural fluctuation of the concentration of an analyte in the same individual, and the inter-individual coefficient of variation (CV_G), which is the variation of the concentration of an analyte among a group of individuals of sound health.¹⁵

Assessing BV helps determine the minimum precision needed for analyte measurement, which can be defined as either the desirable allowable imprecision or bias.¹⁶ Coupled with assay analytical imprecision, this determines the reference change value (RCV) of the assay. This is a statistical threshold used to determine whether the difference in analyte concentration between two consecutive measurements in the same individual is clinically significant. This can help determine the efficacy of an intervention or detriment of a change in health status. Understanding Lp(a) measurement variability is essential in optimizing assessment and management of atherosclerotic cardiovascular disease (ASCVD) risk, especially with the advent of therapies targeting Lp(a).

Past efforts have aimed to understand the temporal variability of Lp(a) measurements within individuals. Marcovina et al. (1994) showed significant differences in BV among subjects, ranging from 0.9 to 51%, inversely correlated with Lp(a).¹⁷ However, this study utilized a calibrator isolated from a donor with an intermediate-size isoform of apo(a), potentially introducing measurement bias as results were expressed in total Lp(a) mass.

Glueck et al. obtained serial measurements from 145 patients over 1.7 years, similarly finding a wide-ranging CV_I (1.7 to 58%) inversely correlated with Lp(a) mass concentration.¹⁸ Similar methodology in a diabetic population to the 1994 BV study by Marcovina et al., reported CV_I ranging from 23.7% to 42.3%, again inversely correlated with Lp(a) concentration.¹⁹ The implications were profound, suggesting that a single measurement could misclassify CHD risk, particularly near clinically significant cut-offs.

In a 2017 study, Marcovina et al. reassessed short-term temporal variability in Lp(a) using individuals from placebo groups in IONIS-APO(a)Rx and IONIS-APO(a)-LRx trials.²⁰ While mean changes in Lp(a) levels over time did not differ significantly among groups with varied Lp(a) concentrations, notable within-person variability was evident. Many individuals showed over a 25% change in Lp(a) level between two time points, highlighting Lp(a)’s dynamic nature over short-term intervals.

In 2018, The European Federation of Clinical Chemistry and Laboratory Medicine Working Group on Biological Variation and Finnish Group for the Biological Variation Database (EFLM WG-BV and TFG-BVD) aimed to tackle standardization issues affecting the application of BV results in clinical care.²¹ They introduced the Biological Variation Data Critical Appraisal Checklist (BIVAC) to assess the reliability of BV estimation studies, which was subsequently utilized to critically appraise existing studies on enzymes, lipids, lipoproteins, and diabetes-related analytes. In addition to most of the studies assessing the BV of Lp(a) being performed over 25 years ago and using outdated analytical methods,²² almost all the studies when appraised using the BIVAC resulted in a quality score of C on a scale of A to D.²¹

Clouet-Foraison et al. applied the BIVAC checklist in a recent study using samples from the European Biological Variation Study (EuBIVAS), estimating a CV_I of 8.9% (95% CI of 8.3 to 9.5%) for all individuals.²³ This study, conducted at the Bristol Royal Infirmary (BRI), sought to apply the BIVAC checklist to a UK cohort. The primary aim was to derive estimates of BV and RCV to inform local clinical decision-making regarding Lp(a)-lowering interventions and significant changes in health status.

Method

Analysis of imprecision and stability of Lp(a) assay

Assessment of stability and imprecision of the quantitative immunoturbidimetric Lp(a) assay by Sentinel Diagnostics on the Beckman Coulter AU5800 were performed according to best practice guideline,²⁴ and Clinical and Laboratory Standards Institute guidelines,²⁵ respectively.

Pooled samples [made up from spare serum from samples analysed for Lp(a)] at five clinically significant Lp(a) concentration levels (∼10 nmol/L, ∼50 nmol/L, ∼100 nmol/L, ∼200 nmol/L, and ∼400 nmol/L) were divided into five series with five replicate samples. Mean, standard deviation (SD), and coefficient of variation (CV) were calculated for each level. Imprecision was compared with the manufacturer’s specification and the total allowable error specified by the acceptable test performance criteria. The CV was calculated as: CV = (SD/mean) x 100. The functional sensitivity, defined as the lowest Lp(a) concentration with a CV of 20%, was estimated from data obtained in the experiment to define total precision.

Previous studies showed minimal decrease in Lp(a) values over 25 months between fresh and frozen plasma.^17,26 Following Young et al.'s protocol for BV studies,²⁷ samples were collected at fixed intervals and frozen to facilitate simultaneous analysis to eliminate between-run imprecision. A stability study was conducted to confirm the reliability of the results. Blood samples from 20 individuals undergoing evaluation in the lipid clinic were collected into an SST tube, centrifuged and serum separated into seven aliquots. Baseline Lp(a) concentration was established from an initial aliquot. Two aliquots were chilled at 2–8°C and analysed after 72 h and 1 week for short-term stability. Two aliquots from each set were stored at −20°C and −80°C and analysed after two and 12 weeks for long-term stability.

Mean standardized values, 95% confidence intervals (CI), and P-values comparing baseline and subsequent time points were calculated. Results were plotted to demonstrate the rate of change in Lp(a) concentration over time. Figure 1 summarizes the processes used to confirm stability of estimates of Lp(a) concentration across varied temperature and temporal storage conditions.

Figure 1.

Schematic flowchart of process used to confirm stability of estimates of Lp(a) concentration across varied temperature and temporal storage conditions.

Study design of biological variation study

Healthy staff members at BRI participated in the study, following the protocol advocated by Cotlove and refined by Young et al.²⁷ Samples were collected at fixed intervals and frozen periodically for simultaneous batch analysis to eliminate between-run variability, compensating for within-run variability through replicate analyses. The study adhered to the 14 BIVAC quality items, including study design, recruitment, methodology of sample collection, storage and analysis and statistical handling.²¹ This enabled the estimates derived to be robust for safe application in clinical practice. The study was performed subject to favourable opinion and authorization from all necessary regulatory and other bodies, including REC, HRA, and University Hospitals Bristol and Weston NHS Trust.

Recruitment

A total of 22 healthy participants were enrolled in the study, determined based on Røraas et al.'s experiment estimating the number of individuals, repeat samples, and replicates needed to power within-person SD detection.²⁸ A power of 0.99 would be achieved with six repeat samples measured in duplicate from 10 healthy participants, with a ratio of one between the analytical coefficient of variation (CV_A) and CV_I.

The study included adults between the ages 18-65 years old. Participants were not included in the study if any of the following applied²⁹: acute hospital admission within last 3 months; diagnosis of diabetes mellitus; chronic kidney disease defined as an eGFR less than 60 mL/min; nephrotic syndrome defined as urine ACR greater than 3 mg/mmol; diagnosis of chronic liver disease due to any aetiology; untreated hypothyroidism or hyperthyroidism; on lipid lowering medications (including statins, ezetimibe, fibrates, bile acid sequestrants, inclisiran and PCSK9 inhibitors); and on exogenous sex steroids or growth hormone replacement. Written informed consent was obtained from participants prior to inclusion in the study.

Sample collection

Participants completed a health questionnaire covering general health, medical history, and medications. Data on age, gender, race, and alcohol consumption were recorded. Initial examination included measurements of height, weight, resting pulse, and blood pressure (BP).

Non-fasting blood samples were collected from participants in the morning over six weekly visits, using standard venepuncture and phlebotomy procedures. Blood samples were collected at each subsequent visit over a 6-week period by the chief investigator to minimize sampling bias. Two serum separator tubes were collected weekly for 6 weeks, centrifuged within 2 hours, and separated. Samples were then transferred into pre-labelled, pseudonymized aliquots and stored at −80°C until analysis, with records maintained by the chief investigator to link participant samples.

Laboratory analysis

All analyses were performed at the biochemistry laboratory of BRI using the Beckman Coulter AU5800. Lp(a) determinations were performed using the Sentinel Diagnostic Lp(a) Ultra (REF 11,504D), a quantitative immunoturbidimetric assay using latex particles coated with anti-Lp(a) antibodies. Agglutination resulting from the interaction between Lp(a) and the anti-Lp(a) antibody was detected through absorbance changes proportional to the Lp(a) concentration, reported in molar concentration units, in the sample. Calibration employed the Sentinel Diagnostic Lp(a) Cal Set (REF 11,522D), comprising lyophilized calibration serum at five levels (11.6 to 217 nmol/L). The Sentinel Diagnostic Lp(a) Control Set (REF 1,152,201), featuring lyophilized control serum at two levels, served as internal quality controls in each run. The Sentinel Diagnostic Lp(a) method is traceable to the reference method and uses calibrators standardized to the WHO-IFCC Reference Material SRM2B.

The participant samples were thawed at room temperature on the day of the analysis and mixed by inversion. Analysis was conducted by a lone operator using a single batch of reagents, quality control materials, and calibrators to minimize analytical variability. Samples were randomly added to the analyser, with quality control at start, middle and end of run, to minimize positional or time related biases.

Statistical analysis

Data processing and statistical analyses utilized various software tools. Initial data entry and cleaning occurred in Microsoft Excel 2016. Python (v3.11) was used for further analysis and modelling, employing libraries such as scipy. stats for tests, statsmodels for advanced modelling, pandas for manipulation, seaborn for visualization, and numpy for numerical operations. Specific statistical analyses were conducted in R (v4.3.2), employing the LMER library for linear mixed-effects models.

CV_I estimation was performed using CV-ANOVA within a multi-level nested model, as outlined by Røraas et al..³⁰ To facilitate this, each individual’s data point was transformed by dividing by the individual’s mean value, with ANOVA performed on the transformed data to estimate CV_I. Data homogeneity for CV_I estimation was assessed using the Cochran’s C test after excluding outlying participants whose variance of repeated measurements fell outside 1.5 times the interquartile range (IQR) of the group variance.^31,32 The homogeneity of the data used to estimate the CV_A was assessed using the Bartlett test after excluding outliers in the SD between duplicates of samples which fell outside 1.5 times the IQR of the group SD.^31,32 Verification of the normality of residuals was performed using the Shapiro-Wilk test. The CV_A was estimated from the difference between duplicate determinations for each sample taken per participant over the 6 weeks of repeat analysis, using the formula SD_A² = ∑d²/2N (d = difference between duplicate determinations, N = number of paired results). The formula, CV_A = SD_A/mean * 100, was then used to calculate the CV_A from the SD_A obtained. To assess steady-state conditions, linear regressions on pooled mean group sample concentrations were conducted. Steady-state was determined if the 95% CIs of the regression slope included zero. CV_I was estimated using a nested linear mixed-effects model on R.

Overall BV estimates were used to calculate the RCV for Lp(a) using the formula below, where Zα is set at 1.65 for the probability level of significant unidirectional change at 95% CI, using the log normal approach.³³ Figure 2 summarizes the processes used to estimate the components of BV and RCV of Lp(a).

{R C V}_{p o s / n e g} = 100 % X [e^{(\pm Z α x \sqrt 2 x \sqrt (S D A 2 + S D I 2))} - 1]

S D A 2 = \ln ({C V A}^{2} + 1)

S D I 2 = \ln ({C V I}^{2} + 1)

Figure 2.

Schematic flowchart of processes used to estimate biological variation and reference changes value of Lp(a).

Results

Analysis of pooled samples revealed CV within the desirable limits of imprecision per the analytical performance specification (APS) value on the EFLM BV database. The APS represents the minimum analytical quality standard required for analyte quantitation to ensure optimal health outcomes for patients. At approximately 12 nmol/L, the CV surpasses the desirable limits of imprecision. Acknowledging a functional sensitivity threshold at CV of 20%, a prudent approach was taken not to report numeric values and to exclude participants with Lp(a) concentration below this value. Aliquots analysed after 2 weeks and after 12 weeks storage at −80°C did not show any significant changes, with mean decreases of 0% (95% CI -4 to 4%, P = .941), and 2% (95% CI -6 to 2%, P = .2354), respectively. The remaining mean standardized values with 95% confidence intervals across various temperature storage conditions are summarized in Table 1.

Table 1.

Mean Standardized Values With 95% CI Across Varied Temperature Storage Conditions.

Storage conditions	Mean standardized value	95% CI	P-value
Baseline	1.00
72 h at 2–8°C	1.00	0.97–1.03	1.000
1 week at 2 °C–8°C	1.01	0.98–1.04	0.647
2 weeks at −20°C	0.97	0.93–1.02	0.238
12 weeks at −20°C	0.94	0.90–0.97	0.001
2 weeks at −80°C	1.00	0.96–1.04	0.941
12 weeks at −80°C	0.98	0.94–1.02	0.354

Baseline characteristics of Lp(a) biological variation study cohort

Twenty-two participants were initially recruited for the project to estimate Lp(a) BV. Table 2 summarizes participants’ baseline characteristics. All participants met study inclusion and exclusion criteria.

Table 2.

Summary of the Baseline Characteristics of Participants Recruited to Study.

Participants, no.	22
Median age, years (IQR)	39 (32–50)
Male gender, no. (%)	9 (40.9%)
Ethnicity
Caucasian, no. (%)	16 (72.7%)
South East Asian, no. (%)	5 (22.7%)
Japanese, no. (%)	1 (4.5%)
Habitual alcohol consumption
None, (%)	55.0%
<14 units, (%)	30.0%
≥14 units, (%)	15.0%
Median BMI, kg/m² (IQR)	24.5 (22.7–26.8)

Estimation of Lp(a) biological variation and reference change value

Two participants were excluded after screening as their Lp(a) concentrations were below the limit of quantitation of the assay. A further two participants were excluded from the analysis as their mean Lp(a) levels fell below the functional sensitivity of the assay. Additionally, four participants were excluded as outliers as the variance of their repeated measurements fell outside 1.5 times the IQR of the group variance.^31,32 Cochran’s C statistic result was 2.10 (P-value = .02). Significant outliers in the SD between duplicates of samples were excluded if they fell outside 1.5 times the IQR of the group SD. All remaining participants had at least four paired samples included in the final model. Shapiro-Wilk test indicated normal distribution for residuals of individual variation. No significant overall trends in Lp(a) concentration were observed over the 6-week study.

Figure 3 displays Lp(a) concentration distribution for all participants in the final analysis, with 95% CI around mean values, after outlier exclusion. Mean Lp(a) concentration ranged from 14 to 241 nmol/L.

Figure 3.

Mean values (dots) and 95% CI of Lp(a) concentrations for each individual included in the study, after exclusion of outliers.

After exclusion of outliers, the overall estimation for CV_I for all participants was 10.9% (95% CI of 9.1 – 13.0%). The CV_A for all participants based on duplicates of each sample was 4.7% (95% CI of 4.5 – 4.8%). The RCV for Lp(a) was calculated using the estimated CV_A and CV_I, with the results summarized in Table 3.

Table 3.

BV Components for Lp(a) Concentration Derived From Study Participants, After Exclusion of Outliers.

Variable	Value	95% CI
Participants, no.	14
Total duplicates of samples, no.	73
CV_A, %	4.7	4.5–4.8
CV_I, %	10.9	9.1–13.0
RCV increase, %	31.6	26.8–37.6
RCV decrease, %	24.0	21.1–27.3

Discussion

The study assessed the imprecision and stability of the Sentinel Diagnostic Lp(a) assay, crucial for BV studies,^26,27 and focused on assessing BV in a healthy UK cohort, adhering to BIVAC standards.

The CV_I of 10.9% in our study closely mirrored the CV_I of 10% (4.3–26.7%) reported in a meta-analysis of Lp(a) BV studies documented in the EFLM BV database.²² This similarity was also seen in comparison to the CV_I estimate of 8.9% (8.3–9.5%) derived from the BV study conducted by Clouet-Foraison et al..²³ Our study could not reliably estimate CV_I separately for males and females. Clouet-Foraison et al. identified a slight gender-based difference in CV_I, reporting values of 6.7% (5.9–7.5%) for males and 10.6% (ranging from 9.8–11.6%) for females.²³ Given the slightly higher proportion of females in our study, it is worth considering whether this demographic distribution influenced the slightly elevated estimate of CV_I.

It is important to highlight the differences in the BV estimates of Lp(a) reported in studies up to 2003.^17–19 This may be explained by outdated assays not meeting current standards of isoform insensitivity, variations in study designs, and less restrictive statistical analyses that employed heterogeneous cohorts, resulting in unreliable CV_I estimations with wide CIs. However, there is a potential risk associated with stringent exclusion criteria applied in BV studies, specifically tailored for APS purposes. Such rigorous criteria might yield estimates that lack external validity in the context of routine clinical practice. In our study, four out of 18 participants were excluded due to non-homogeneity of CV_I, emphasizing the delicate balance required in data exclusion to ensure reliability without sacrificing generalizability.

The minimum number of samples required from each participant to achieve an estimate within 10% of their true homeostatic set point with a 95% probability can be determined using the formula: n = {1.96 * ([CV_A² + CV_I²]^1/2/10)}².³⁴ Based on the CV_A and CV_I estimates obtained in the study, a minimum of five samples is necessary to obtain an estimate for Lp(a) concentration within 10% of a healthy participant’s true homeostatic set point, within our laboratory setting.

Changes in Lp(a) concentrations with novel medications such as pelacarsen, olpasiran, or lepodisiran are usually evident, given the substantial reductions, typically more than 90%, conferred by these new injectable therapies. However, selecting patients for these treatments based on Lp(a) concentration cut-offs, such as those used in clinical trial inclusion criteria, requires careful consideration. While it is impractical to repeat five samples in routine clinical care to obtain a true Lp(a) value for a patient, it is crucial to recognize the limitations of a single measurement when formulating long-term management plans for patients with elevated Lp(a).

The study-derived RCV provides critical guidance for both clinicians and laboratories in interpreting Lp(a) results. For clinicians, the RCV enhances the utility of a single Lp(a) measurement by establishing a range within which the true patient value likely resides. This aids in refining risk prognostication and determining potential eligibility for Lp(a)-lowering therapies, particularly when results are near decision thresholds outlined in current guidelines. For laboratories, the RCV serves as a benchmark to assess significant changes in Lp(a) levels between consecutive measurements. Exceeding this limit warrants further investigation into potential pre-analytical errors such as sample mislabelling or collection issues or underlying secondary causes of elevated Lp(a), including hypothyroidism, nephrotic syndrome, or renal disease.

Harb et al. highlighted the value of repeat Lp(a) measurements in reclassifying cardiovascular risk using European Atherosclerosis Society (EAS) categories.³⁵ Among 98 patients with intermediate-risk Lp(a) levels (30–50 mg/dL), 20% shifted to low-risk and 33% to high-risk after retesting. Similarly, Shamsudeen and Hegele reported intra-individual variability, with 10 of 11 patients reclassified around a 50 mg/dL cut-off over 10 years.³⁶ While Trinder et al. observed strong correlations between baseline and follow-up Lp(a) levels, considerable scatter around the correlation line suggests that reliance on a single result can be misleading, particularly when significant change was taken as exceeding 25% or 120 nmol/L.³⁷

In our cohort of 22 participants, applying EAS-based categories (<90, 90–199, ≥200 nmol/L), three shifted from low- to moderate-risk and two from moderate- to high-risk. While a single lifetime measurement of Lp(a) may be practical given resource constraints, this study highlights the need for caution when making long-term management decisions based on a single Lp(a) value, particularly when the measurement is close to clinical decision thresholds.

Three participants showed Lp(a) concentrations above 125 nmol/L. Acknowledging the limited participant pool, this distribution aligns with the observed Lp(a) concentration pattern in the Copenhagen General Population Study.³⁸ Although this falls outside the primary scope of our study, it underscores the necessity for a more systematic approach to Lp(a) screening in the general population in identifying individuals who might be at an elevated risk of ASCVD due to underlying genetic factors. Elevated Lp(a) levels correlate with increased ASCVD risk, exhibiting a concentration-dependent relationship.^9–11 Ongoing clinical trials are investigating novel Lp(a) lowering therapies’ impact on ASCVD, with the hope of addressing residual ASCVD risk not directly linked to LDL cholesterol.^39,40

Study limitations

A limitation of the study is the exclusion of results in the BV estimation due to non-homogeneity of variance and instances where Lp(a) concentration fell below the quantification limit or functional sensitivity threshold. While necessary for the reproducibility of estimates using the CV-ANOVA method, its potential impact on the clinical applicability and generalizability to the broader population should be acknowledged.

It is important to highlight that the study refrained from calculating a CV_G for Lp(a) due to the wide range of observed concentrations stemming from underlying genetic factors within the population. Furthermore, the utility of a CV_G for Lp(a) concentration is doubtful, as interpretation of a Lp(a) result is inherently linked to an individual’s own ASCVD risk.

Separate estimations of CV_I for male and female would have been ideal, considering slight differences in prior studies. However, limited participants precluded statistically valid estimates. Stratifying data based on concentration ranges for consistent CV_I estimates was unfeasible due to small sample size, although prior studies found no significant differences across groups.

Conclusions

Our findings yielded a CV_I estimate for Lp(a) that aligned consistently with recent studies adhering to the quality specifications outlined in the BIVAC checklist. Our findings mirrored those reported by Clouet-Foraison et al., who conducted their study using the EuBIVAS population, and were in line with estimates documented in the EFLM BV database, derived from a meta-analysis of prior studies. Importantly, our CV_I estimate was significantly lower than Lp(a) estimates reported in studies up to 2003, a discrepancy likely attributed to methodological and assay advancements over the years. The CV_I estimate highlights the limitations of relying solely on a single Lp(a) measurement for prognosticating ASCVD risk and identifying candidates for novel Lp(a) therapies, particularly when the measured value is near clinical decision thresholds..

Footnotes

Acknowledgements

I would like to thank the staff in the Biochemistry Department at BRI for all their help with the setup of the assay. I would like to acknowledge the help given to me by the R&D department at Bristol Royal infirmary. I would like to express my heartfelt thanks to the staff at BRI, who volunteered their time and samples to the project.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) did not receive any financial support for the research, authorship, and/or publication of this article. Dr Paul Downie has received speaker/consulting fees from Besins Healthcare, Daiichi-Sankyo, Sanofi, Amgen, Sobi, Novartis and Amarin. Dr Paul Downie has received financial support for travel and accommodation to attend national/international conferences from Sanofi, Amgen and Daichii-Sankyo. Dr Kofi Antwi has received financial support for travel and accommodation to attend an international conference from Daichii-Sankyo.

Ethical approval

IRAS 319104 Biological variation of Lp(a) and its application in a lipid clinic - HRA & HCRW Approval issued 31/08/2023.

Guarantor

Dr Kofi Antwi.

Contributorship

Dr Kofi Antwi wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

ORCID iD

Kofi Antwi

References

Schmidt

Noureen

Kronenberg

, et al. Structure, function, and genetics of lipoprotein (a). J Lipid Res 2016; 57(8): 1339–1359.

Kraft

Köchl

Menzel

, et al. The apolipoprotein (a) gene: a transcribed hypervariable locus controlling plasma lipoprotein (a) concentration. Hum Genet 1992; 90(3): 220–230.

Utermann

Menzel

Kraft

, et al. Lp(a) glycoprotein phenotypes. Inheritance and relation to Lp(a)-lipoprotein concentrations in plasma. J Clin Investig 1987; 80(2): 458–465.

Kronenberg

Utermann

. Lipoprotein(a): resurrected by genetics. J Intern Med 2013; 273(1): 6–30.

Brunner

Lobentanz

Pethö-Schramm

, et al. The number of identical kringle IV repeats in apolipoprotein(a) affects its processing and secretion by HepG2 cells. J Biol Chem 1996; 271(50): 32403–32410.

Leibundgut

Scipione

Yin

, et al. Determinants of binding of oxidized phospholipids on apolipoprotein (a) and lipoprotein (a). J Lipid Res 2013; 54(10): 2815–2830.

van der Valk

Bekkering

Kroon

, et al. Oxidized phospholipids on lipoprotein(a) elicit arterial wall inflammation and an inflammatory monocyte response in humans. Circulation 2016; 134(8): 611–624.

Garg

Guan

Karger

, et al. Lipoprotein (a) and risk for calcification of the coronary arteries, mitral valve, and thoracic aorta: the Multi-Ethnic Study of Atherosclerosis. J Cardiovasc Comput Tomogr 2021; 15(2): 154–160.

Emerging Risk Factors Collaboration Erqou

Kaptoge

Perry

, et al. Lipoprotein(a) concentration and the risk of coronary heart disease, stroke, and nonvascular mortality. JAMA 2009; 302(4): 412–423.

10.

Kamstrup

Tybjaerg-Hansen

Steffensen

, et al. Genetically elevated lipoprotein(a) and increased risk of myocardial infarction. JAMA 2009; 301(22): 2331–2339.

11.

Willeit

Kiechl

Kronenberg

, et al. Discrimination and net reclassification of cardiovascular risk with lipoprotein(a): prospective 15-year outcomes in the Bruneck Study. J Am Coll Cardiol 2014; 64(9): 851–860.

12.

Kronenberg

Tsimikas

. The challenges of measuring Lp(a): a fight against Hydra? Atherosclerosis 2019; 289: 181–183.

13.

Marcovina

Albers

Gabel

, et al. Effect of the number of apolipoprotein(a) kringle 4 domains on immunochemical measurements of lipoprotein(a). Clin Chem 1995; 41(2): 246–255.

14.

Cegla

Neely

RDG

France

, et al. HEART UK consensus statement on Lipoprotein(a): a call to action. Atherosclerosis 2019; 291: 62–70.

15.

Harris

. Distinguishing physiologic variation from analytic variation. J Chron Dis 1970; 23(7): 469–480.

16.

Fraser

. Biological variation data for setting quality specifications. Interneto prieiga: https://wwwwestgardcom.2010

17.

Marcovina

Gaur

Albers

. Biological variability of cholesterol, triglyceride, low- and high- density lipoprotein cholesterol, lipoprotein(a), and apolipoproteins A-I and B. Clin Chem 1994; 40(4): 574–578.

18.

Glueck

Tracy

Sieve-Smith

, et al. Whether, to what degree, and why lipoprotein(a) levels change over time. Clin Chim Acta 1995; 238(1): 11.

19.

Hernández

Francisco

Chacón

, et al. Biological variation of lipoprotein(a) in a diabetic population. Analysis of the causes and clinical implications. Clin Chem Lab Med 2003; 41(8): 1075–1080.

20.

Marcovina

Viney

Hughes

, et al. Temporal variability in lipoprotein(a) levels in patients enrolled in the placebo arms of IONIS-APO(a)(Rx) and IONIS-APO(a)-L(Rx) antisense oligonucleotide clinical trials. J Clin Lipidol 2018; 12(1): 122–129.

21.

Aarsand

Røraas

Fernandez-Calle

, et al. The biological variation data critical appraisal checklist: a standard for evaluating studies on biological variation. Clin Chem 2018; 64(3): 501–514.

22.

Aarsand

Fernandez-Calle

Webster

, et al. The EFLM biological variation database. Avaiable at: https://biologicalvariationeu/ (Accessed December 20th. 2020).

23.

Clouet-Foraison

Marcovina

Guerra

, et al. Analytical performance specifications for lipoprotein(a), apolipoprotein B-100, and apolipoprotein A-I using the biological variation model in the EuBIVAS population. Clin Chem 2020; 66(5): 727–736.

24.

Gomez-Rioja

Von Meyer

Cornes

, et al. Recommendation for the design of stability studies on clinical specimens. Clin Chem Lab Med 2023; 61(10): 1708–1718.

25.

Carey

. User verification of precision and estimation of bias: approved guideline. Clinical and Laboratory Standards Institute, 2014.

26.

Kronenberg

Lobentanz

König

, et al. Effect of sample storage on the measurement of lipoprotein[a], apolipoproteins B and A-IV, total and high density lipoprotein cholesterol and triglycerides. J Lipid Res 1994; 35(7): 1318–1328.

27.

Young

Harris

Cotlove

. Biological and analytic components of variation in long-term studies of serum constituents in normal subjects IV. Results of a Study Designed to Eliminate Long-Term Analytic Deviation 1971; 17: 403.

28.

Røraas

Petersen

Sandberg

. Confidence intervals and power calculations for within-person biological variation: effect of analytical imprecision, number of replicates, number of samples, and number of individuals. Clin Chem 2012; 58(9): 1306–1313.

29.

Enkhmaa

Berglund

. Non-genetic influences on lipoprotein(a) concentrations. Atherosclerosis 2022; 349: 53–62.

30.

Røraas

Støve

Petersen

, et al. Biological variation: the effect of different distributions on estimated within-person variation and reference change values. Clin Chem 2016; 62(5): 725–736.

31.

Braga

Panteghini

. Generation of data on within-subject biological variation in laboratory medicine: an update. Crit Rev Clin Lab Sci 2016; 53(5): 313–325.

32.

Tukey

. Exploratory data analysis. Vol. 2. Reading, MA: Addison-Wesley Publishing Company, 1977.

33.

Fokkema

Herrmann

Muskiet

FAJ

, et al. Reference change values for brain natriuretic peptides revisited. Clin Chem 2006; 52(8): 1602–1603.

34.

Fraser

Harris

. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989; 27(5): 409–437.

35.

Harb

Ziogos

Blumenthal

, et al. Intra-individual variability in lipoprotein(a): the value of a repeat measure for reclassifying individuals at intermediate risk. Eur Heart J Open 2024; 4(5): oeae064.

36.

Shamsudeen

Hegele

. Lipoprotein(a) variability over time in adult lipid clinic patients. Atherosclerosis 2022; 359: 52–53.

37.

Trinder

Paruchuri

Haidermota

, et al. Repeat measures of lipoprotein(a) molar concentration and cardiovascular risk. J Am Coll Cardiol 2022; 79(7): 617–628.

38.

Nordestgaard

Chapman

Ray

, et al. Lipoprotein(a) as a cardiovascular risk factor: current status. Eur Heart J 2010; 31(23): 2844–2853.

39.

O’Donoghue

Rosenson

Gencer

, et al. Small interfering RNA to reduce lipoprotein(a) in cardiovascular disease. N Engl J Med 2022; 387(20): 1855–1864.

40.

Tsimikas

Karwatowska-Prokopczuk

Gouni-Berthold

, et al. Lipoprotein(a) reduction in persons with cardiovascular disease. N Engl J Med 2020; 382(3): 244–255.