Despite advances in Cardiovascular Disease (CVD) diagnosis and treatment, the total CVD prevalence for individuals ≥20 years of age in the United States is 48.6%.¹ Elevated low-density lipoprotein cholesterol (LDL-C) levels contribute to the pathogenesis of atherosclerotic cardiovascular disease (ASCVD) and cholesterol-lowering therapies predominantly target LDL-C, including statins and proprotein convertase subtilisin/kexin-type 9 (PCSK-9) inhibitors.^{2, 3} However, reports from several clinical trials suggest that a portion of the “residual risk” for adverse CVD outcomes in the setting of controlled LDL-C is associated with elevated lipoprotein(a) [Lp(a)].^{4, 5} Furthermore, studies have established that elevated Lp(a) is an independent, causal risk factor for the development of ASCVDs, including myocardial infarction.^{6, 7} Clinical guidelines currently support the measurement of Lp(a) in individuals with a personal or familial history of ASCVD⁸ or have advocated for the measurement of Lp(a) at least once for all adults.^9–12 The 2018 AHA/ACC Guidelines on the Management of Blood Cholesterol currently state that an Lp(a) ≥50 mg/dL constitutes a risk-enhancing factor, especially at higher Lp(a) concentrations,¹³ whereas the 2019 ESC/EAS Guidelines for the Management of Dyslipidemias state that Lp(a) concentrations ≥125 nmol/L constitute an increased CVD risk.¹¹ While guideline recommendations differ with respect to clinical decision cut-off concentrations, units and frequency of measurement, and patient populations for which Lp(a) measurements are recommended, the overarching consensus is that Lp(a) should be measured.^{8–12, 14} Despite guidelines and scientific statements recommending the use of Lp(a) for CVD risk stratification, Lp(a) is not routinely measured in clinical practice.^{4, 10, 14, 15} This is attributed, in part, to a lack of clinical assay standardization, which currently makes comparisons of patient measurements to clinical decision points and across clinical assays and studies challenging.^{16, 17}

Lp(a) is comprised of an apolipoprotein B (apoB)-containing LDL-like particle that is covalently linked to the apolipoprotein(a) [apo(a)] protein, which contains a plasminogen-like domain, 1 kringle five (KV) domain, and 10 kringle four (KIV) domains known as KIV₁ – KIV₁₀. The Lp(a) protein is synthesized in the liver and is highly heterogeneous due to apo(a) size polymorphisms that are genetically determined by the number of KIV₂ repeat sequences present, which can range in number from 1 to 40 copies. Depending on the number of KIV₂ repeats, apo(a) molecular weights can range between ~250 – 800 kDa, making the interpretation of Lp(a) concentrations difficult when results are reported in mass units (i.e., mg/dL) rather than molar units (nmol/L).⁵ Lp(a) measurement and data interpretation is further complicated by the presence of two different isoforms in circulation in most individuals (heterozygotes), which is typically not captured with current analytical assays. Also, differences in Lp(a) concentrations across ethnic groups have been described and their relationship to CVD risk has not been fully elucidated.^{6, 18, 19}

Lp(a) measurements performed on the same patient samples by different clinical assays were previously reported to exhibit high variability.^{17, 20} The consequences of Lp(a) inter-assay measurement variability include an inability to establish appropriate reference intervals and to universally apply clinical cut-off values. Additionally, a lack of assay agreement impedes Lp(a) patient sample measurement comparisons across clinical assays, interferes with measurement comparisons between clinical studies, and hinders assessments of pharmacologic efficacy, which is important given that several clinical trials are underway for Lp(a) lowering therapeutics. This inter-assay variability was previously attributed, in part, to assay antibody cross reactivity with the apo(a) KIV₂ domain.^{17, 21, 22} However, multiple factors likely contribute to current Lp(a) inter-assay measurement variability and must be addressed as Lp(a) assays are standardized. These factors include, but are not limited to, differences in how manufacturers perform assay calibration, how reference materials value assigned in molar units (nmol/L) are used for the calibration of assays reporting in mass units (mg/dL), and limited analytical measurement ranges, which may require the dilution of samples typically observed in the clinical setting.

Lp(a) measurement variability persists, despite some Lp(a) assays being traceable to a common WHO/IFCC reference material, known as standard reference material (SRM)-2B,^{23, 24} with values assigned in nmol/L by an enzyme-linked immunosorbent assay (ELISA)-based, KIV₂ independent reference method.²⁵ Importantly, this ELISA-based reference method is no longer available and the SRM-2B reference material was depleted, thus establishing the need for a new, more sustainable reference method and materials that can be used in standardization programs. The standardization of Lp(a) assays will help improve measurement agreement across assays and will ensure that measurement results are comparable to clinical decision points, across methods, location, and over time. To date, two candidate mass spectrometry-based reference measurement procedures (RMP) have been proposed to replace the original ELISA-based RMP, including one endorsed by the International Federation of Clinical Chemistry and Laboratory Medicine’s Working Group for Apolipoproteins by Mass Spectrometry (IFCC WG APO-MS).^{17, 26} The IFCC WG APO-MS is also working to establish new serum-based reference materials for Lp(a) clinical assay calibration with traceability to the IFCC WG APO-MS RMP.^{27, 28}

The principal goal of CDC Clinical Standardization Programs (CSP) is to improve the accuracy and reliability of chronic biomarker measurements used in the patient care, public health, and clinical research settings. The CDC CSP is collaborating with the IFCC WG APO-MS to establish a formal Lp(a) standardization program as part of the IFCC’s proposed reference measurement system (RMS). In preparation for this, the CSP conducted a Lp(a) Interlaboratory Comparison Study with end-user clinical laboratories as participants that focused on Lp(a) clinical assays, to obtain information about current inter-assay variability, and to guide specific approaches for the implementation of Lp(a) standardization programs. Specifically, this study aimed to obtain information about the analytical performance of clinical assays routinely used in the patient care setting for Lp(a) measurements, to determine the current extent of and potential sources contributing to inter-assay variability, to assess how Lp(a) clinical assay measurements correlate with mass spectrometry values from the IFCC RMP, and to elucidate if Lp(a) inter-assay variability is attributable to differences in Lp(a) isoform sizes.

This Lp(a) Interlaboratory Comparison Study was designed and conducted by the CDC CSP to evaluate the accuracy and variability of serum Lp(a) measurements across current clinical assays routinely used in patient care settings, to investigate potential sources of measurement variability, and to inform stakeholders about ongoing Lp(a) standardization efforts. The results of this study demonstrate that inter-assay variability persists, especially at concentrations used to guide clinical decision making, despite previous efforts to standardize Lp(a) assays. The previous ELISA-based RMS, to which many assays are currently standardized, is no longer available and reference materials traceable to the previous RMS are now depleted. To fill this critical gap and further improve Lp(a) inter-assay agreement, the IFCC WG APO-MS established a new LC-MS-based RMP and is generating new reference materials value assigned in SI units and anchored to the new RMP for manufacturers to use for calibration purposes. Importantly, assay measurements in mg/dL showed good correlation with LC-MS values in nmol/L, suggesting a change from the previous ELISA-based RMS to the new LC-MS-based RMS may have minimal impacts on Lp(a) measurements. Study results also provided important information about factors that do not contribute to inter-assay variability, such as Lp(a) isoform sizes, as well as factors that may contribute to inter-assay variability, which will be explored further after assays are calibrated to a new common calibrator. While further improvements in Lp(a) standardization are necessary, the current lack of assay agreement should not preclude the measurement Lp(a) in patients according to multi-society clinical guidance documents that suggest measuring Lp(a) to identify those with higher CVD risk. Assay standardization takes time; thus, standardization efforts are being conducted in parallel with clinical Lp(a) measurements for risk assessment with the goal to improve Lp(a) measurement agreement over time.

The Lp(a) literature has long attributed inter-assay Lp(a) measurement variability to the use of polyclonal antibodies and potential antibody cross-reactivity with the KIV₂ sequence in the apo(a) repeat region,^{17, 37} which was believed to result in Lp(a) concentrations that do not accurately reflect the true Lp(a) concentration. To determine if or how Lp(a) isoform size currently contributes to Lp(a) inter-assay measurement variability, individual samples were phenotyped and selected based on the total KIV# of the dominant Lp(a) isoform. Samples with the same or similar apo(a) isoform sizes and different Lp(a) concentrations were specifically selected for study inclusion to determine how much variability stems from isoform size. The data from individual samples in this this study suggest that Lp(a) measurement variability is not significantly correlated with isoform size, suggesting additional sources of variability should be considered. These same serum units, pre-selected based on apo(a) isoform content, were used to generate 3 CDC QC serum pools with 6 KIV each. Interestingly, the 3 QCs, 2 with the same isoform sizes, did not exhibit higher inter-assay variability, again suggesting isoform size independence. The variability observed for the QCs pools did not reflect that observed in clinical samples, where QC variability was lower than individual donor samples with comparable concentrations, suggesting that pooled materials may not provide comprehensive information about measurement variability that occurs in individual donor samples used in patient care and clinical research.

Often, inter-assay variability is a product of several different factors that must be addressed as assays are standardized. All assays in this study utilized a 5-point calibration system and demonstrated good linearity when compared to the all lab mean concentration (mg/dL). Assays included in this study also exhibited higher sample scatter and biases at higher Lp(a) concentrations, which occurred just below or above the Lp(a) clinical decision point of 50 mg/dL (125 nmol/L). Interestingly, assays included in this study with stated traceability to SRM-2B, a single level RM value assigned in nmol/L by the previous RMS, still exhibit both positive and negative biases when compared to the all lab mean. Additional factors that may potentially contribute to Lp(a) inter-assay variability include differences in how manufacturers perform assay calibration and how reference materials, such as SRM-2B, value assigned in molar units (nmol/L) were used for the calibration of assays reporting in mass units (mg/dL). Furthermore, the composition of calibration solutions used by assay manufacturers are unknown and could be a potential contributing factor. One would expect calibration differences to impact sample measurements in a consistent manner and may explain some of the differences in slopes across assays but do not likely explain differences in scatter. The increased scatter may stem from how manufacturers use secondary reference materials value assigned in molar units (nmol/L) to calibrate assays measuring in mass units (mg/dL). This is often done through the application of proprietary mathematical formulas for unit conversions, which could be a potential source of continued inter-assay variability. Indeed, studies have demonstrated that no clear molar to mass ratio can be universally applied.³⁷ However, the need for unit conversions would be ameliorated by a transition of all Lp(a) assays to molar unit measurements. Worth noting, a recent scientific statement from the American Heart Association recommended that labs use an “assay that is reported in nanomoles per liter”;¹⁴ however, the clinical decision limits currently used in the US are in mg/dL and a switch to molar units requires stakeholder support and time for implementation.

Lp(a) measurement variability may also be due to the potential impacts of sample dilutions, or the diluent used, that are performed when the analytical measurement range does not cover the concentration range observed in patient samples.²² In this study, some individual samples had concentrations outside the analytical measurement range for 7 out of 8 assays. Preanalytical differences, such as freeze-thawing or sample handling, may also lend to increased Lp(a) measurement variability.³⁸ Notably, the biases for the same assays run by different laboratories did not always align, where Lp1 and Lp6 exhibited positive and negative biases, respectively. Disparate bias trends were also observed between Lp8 and Lp9, which also exhibited significantly different individual sample average CVs of 4.5% and 11.4%, respectively. These differences occurred despite laboratory participants reporting use of the same assay on the same instrument platform and using calibrators from the same manufacturer. The primary discrepancy noted was a wider degree of sample scatter that is not likely attributable to lot-to-lot reagent variations, which would cause a systematic shift across all measurement results. These results suggest that potential differences in preanalytical sample handling between Lp1 and Lp6, for example, seem to be relevant and requires further investigation.

The previous Lp(a) ELISA RMS is no longer available and the SRM-2B reference material is depleted, resulting in the need for a new RMS and new reference materials. The IFCC WG APO-MS is establishing a new LC-MS RMS and reference materials that will be value assigned in nmol/L. To evaluate how moving to the new IFCC-endorsed LC-MS RMS might impact Lp(a) assays currently performing measurements in mg/dL, sample measurements in mg/dL (except Lp3 in nmol/L) were compared to LC-MS values in nmol/L. While assay-specific Lp(a) measurements in mg/dL demonstrated good linear correlation with values from the LC-MS RMP, a comparable degree of scatter was still observed at concentrations ≥50 mg/dL (or 125 nmol/L). This suggests that the scatter is not likely attributable to non-specificity, which would result in higher scatter in measurement comparisons to the LC-MS values than to the all lab mean, which was not the case. The use of Lp(a) assays reporting in two different units of measure have hindered the establishment of universal clinical thresholds. Interestingly, a recent publication from Szarek, et al. utilized data from the ODYSSEY OUTCOMES trial to investigate if Lp(a) concentration measurements in 11,970 samples by 3 different assays were associated with risk of major adverse cardiovascular events (MACE).³⁹ The study found that Lp(a) measurements from two immunoassays, 1 reporting in mg/dL and 1 in nmol/L, and by the IFCC LC-MS assay were all similarly prognostic for MACE risk.³⁹ In the study from Szarek, both immunoassays were well-correlated with the LC-MS method (r>0.96), consistent with a previous report.^{28, 39} In our study, we also found good correlation between immunoassay measurements and the LC-MS RMP values. Research to assess whether a switch to the IFCC LC-MS RMS would affect clinical interpretation is ongoing.

As the data presented in this study suggest, there is clearly a remaining need to standardize Lp(a) assays to 1) allow for clinical trial measurement comparisons and 2) to help ensure that patient measurement results are comparable across methods, location, and over time. This is important given that there are several small interfering RNA-based pharmacologic therapies in various clinical trial phases (NCT04606602 - SLN360, NCT04023552 - TQJ230 or Pelacarsen, and NCT04270760 - AMG 890 or Olpasiran) that reduce hepatic Lp(a) expression. The standardization of Lp(a) assays will not only allow researchers to appropriately compare clinical trial outcomes, but it will also allow physicians to effectively monitor patient treatment efficacy irrespective of the assay used or laboratory that performs the Lp(a) measurements. While Lp(a) assays are currently used in patient care, these assays exhibit inter-measurement variability despite previous standardization efforts. Therefore, further Lp(a) assay standardization is necessary to improve inter-assay Lp(a) measurement consistency and, thus, clinical utility, as indicated in a previous study.²⁷ However, the lack of assay standardization should not preclude the measurement Lp(a) in patients in line with society clinical guidance documents that suggest measuring Lp(a) to identify those with higher CVD risk. Assay standardization takes time; thus, standardization efforts will be conducted in parallel with continued clinical risk assessments with the goal to improve inter-assay agreement over time by providing manufacturers with new reference materials anchored to the new IFCC mass spectrometry-based RMS.

The data from this study suggest that Lp(a) measurement variability is not concentration dependent (similar CVs across the concentration range), nor does variability increase with isoform size, necessitating further investigations into potential sources of assay variability. This study provides new insights into current Lp(a) inter-assay variability and assay performance that will guide future CDC CSP standardization efforts. Ultimately, Lp(a) standardization, which includes reporting results in SI units, will enable measurement comparisons across clinical assays and between clinical studies, which is currently challenging, thus helping improve patient care and public health by increasing evidenced-based clinical decision making.