Advanced lipoprotein testing for cardiovascular diseases risk assessment: a review of the novel approaches in lipoprotein profiling

Tube gel electrophoresis (TGE)

Since the first publications reporting lipoprotein separation on gel matrices in the early 1960s [37], gel electrophoresis (GE) underwent numerous developments and optimizations, especially concerning gel preparation, electrophoretic separation and staining. GE separates lipoproteins according to both their size and charge on gels of different natures. Depending on charge, particle mobility is more or less enhanced by the electric field while, in parallel, size hinders the migration in the gel network so that large particles migrate less than small ones. The main disadvantages of GE are the poor resolution and reproducibility of lipoprotein separation, notably due to the large variability of gel quality, and it is often considered a semiquantitative method [27]. However, the commercialization of ready-to-use tubes for GE (TGE) with reproducible gel matrices of similar properties, furnished as kits, greatly improved the method robustness.

LIPOPRINT™ is an example of semiautomated TGE system commercialized by Quantimetrix (Redondo Beach, CA, USA) for lipoprotein separation. The method involves GE kits, either specific to LDL or HDL particles profiling, consisting of ready-to-use 3% precast polyacrylamide gels in individual tubes [38]. Using a cholesterol-specific dye added before electrophoresis, lipoproteins, separated as different bands in the gel, are detected by densitometry to obtain the lipoprotein profile. In addition, because the dye added is cholesterol specific, bands intensity, i.e. the peak area determined by densitometry, is proportional to the relative amount of cholesterol associated with each lipoprotein subclass. This relative amount, when multiplied by the TC concentration, then corresponds to the absolute cholesterol concentration associated with each lipoprotein subclass [38], [39]. The major advantage of this method is that it comes with automated data processing software that posttreats densitometric measurements to determine the relative amounts of the different lipoprotein fractions. However, TC concentrations have to be measured separately with an independent method prior to analyses. Lipoprotein profiles are obtained in less than 3 h and compare well with those obtained by polyacrylamide GGE [38], [40]. The system was actually cleared by the FDA for LDL-C and LDL subfraction cholesterol concentration measurements [41]. The system can additionally measure lipoprotein sizes; however, comparability of the diameters obtained by this technique and by classic GGE was found perfectible [40], [42].

Cholesterol subclass concentrations measured after TGE rely on densitometry measurements. Depending on the method chosen to reveal lipoproteins after separation, the quantity actually measured is different. If a lipid or lipoprotein dye is used, the measurand is the complex formed between the dye and its target. However, if an enzymatic assay is used, then the measurand is the colored product of the reaction between the enzyme and its substrate. Very few data are available on the calibration materials used for particle concentration measurements derived from electrophoretic techniques, and most assays seem to have been developed using materials value assigned by field methods such as IT [38], [43]. However, for LIPOPRINT™, TC concentrations must be measured independently. If they were determined by a primary reference method, then the derived concentrations would be traceable to the SI units. It is to keep in mind though that because measurands are different, establishing SI traceability of the TC concentration would not guarantee the accuracy of the derived lipoprotein fractions cholesterol concentrations.

Gel permeation (GP) chromatography

Gel permeation-high-performance liquid chromatography (GP-HPLC) was first reported in 1980 by Hara et al. [44], and its clinical relevance to distinguish different lipoprotein profiles in serum was demonstrated shortly after [45], [46]. Recently, an automated GP-HPLC system adapted for routine and high throughput measurements of lipoprotein profiles, notably involving automated data processing, was released as the LipoSEARCH^® [47].

GP-HPLC separates lipoproteins as a function of their hydrated diameter according to size exclusion chromatography (SEC) principles. The large column pores allow the permeation of the smallest entities inside the pores while the biggest entities are not able to penetrate. Largest entities are thus eluted first as their path through the column is the shortest while the smallest entities elute last as their path through the pores is longer [48]. Detection and quantitation are performed using UV-Visible absorbance, either at 280 nm for apolipoproteins or at 550 nm after lipid derivatization [49]. LipoSEARCH^® uses two tandem TSK-like columns packed with polymethacrylate beads and measures simultaneously cholesterol and TG concentrations of the separated lipoproteins. Detection is performed by absorbance at 550 nm after enzymatic reaction of TG and cholesterol [50]. Automated data-processing software then deconvolutes chromatographic peaks and calculates the lipid concentrations associated with each lipoprotein subclass. Corresponding particle numbers can additionally be derived by the dedicated software using an algorithm developed and patented by Okazaki in 2015 that calculates particle number concentrations from TG and cholesterol concentrations and particle size [50], [51].

Similar techniques using fast protein liquid chromatography (FPLC) systems for lipoprotein profiling were also reported in the late 1980s. These methods also separate lipoproteins by SEC, i.e. according to their size, but on a different type of resins: agarose gel Superose 6 [49], [52], [53]. Lipoprotein profiling and quantification protocols using these systems are similar to that of GP-HPLC. However, an important value added of FPLC is the possibility to use it either as an analytical or preparative method with improved reproducibility compared to UC [54]. Longer analyses procedures, lower throughput and higher pressures are the major downsides of FPLC. The method thus sparsely spread as a routine technique but does constitute a valuable tool for complementary analyses in research facilities or for clinical studies.

Concentration measurements by GP-HPLC or FPLC are performed mostly by continuous enzymatic reactions of cholesterol and TG or by absorbance measurements of the apolipoproteins at 280 nm. Overall, most methods described for lipoprotein profiling by GGE can be used for detection and quantification after GP lipoprotein separation. Assay calibration then depends on the method chosen to assign the TC and/or TG concentrations to the control materials. In parallel, GP columns pore size are generally verified using the same materials used for pore size calibration in GGE, i.e. either nanoparticle standards or protein molecular weight standard mixtures [55].

Apolipoprotein profiling by liquid chromatography isotopic-dilution mass spectrometry (LC-ID/MS)

Isotopic-dilution mass spectrometry (ID/MS) is the higher order reference method for many biomarkers in clinical chemistry and notably for TG and TC measurements [56], [57]. Apolipoprotein quantification by liquid chromatography ID/MS (LC-ID/MS) was first reported in the late 1990s by Barr et al. for apoA-I [58] and was further applied to other apolipoproteins (apoB, C and E) in the following years [59], [60], [61].

Apolipoprotein absolute quantification by LC-ID/MS relies on enzymatic trypsin digestion of serum apolipoproteins. After digestion, apolipoprotein-specific tryptic peptides were identified for each major class of apolipoproteins, and some of them were selected for quantification by ID/MS [60], [61], [62]. ID/MS quantification uses synthetic labeled entities with ¹³C, ¹⁵N or deuterium as internal standards (IS) to spike the samples. Depending on the method, this IS can be either the labeled recombinant protein or a synthetic labeled peptide characteristic of the protein. In the first case, both calibrators and samples undergo digestion, whereas in the latter case only the samples are digested. Using a recombinant protein as IS is considered the best practice for accurate and precise protein quantification by ID/MS as it allows taking into account the variability due to digestion. However, in some cases, using recombinant proteins is a challenge, and the alternative solution involving a synthetic peptide is more adapted. For quantification, calibrators consist of mixtures of unlabeled and labeled standards, i.e. protein or peptide, in different molar ratios. The peak area ratio of the unlabeled versus labeled entity is plotted as a function of the molar ratio to obtain the calibration curve. Knowing the concentration of the synthetic standards, the concentration of the endogenous peptide after digestion can be derived by calculating the peak area ratio of the endogenous peptide versus labeled peptide in the sample.

Although ID/MS results usually are reproducible and accurate, the initial tryptic digestion step may increase results variability, especially when using peptides as IS. The choice of fast releasing peptides and the verification that digestion is complete are then necessary to achieve accurate and precise quantification [63]. In addition, to establish the calibration curve, concentrations of the IS have to be accurately measured. This can be done with a primary reference method such as amino acid analysis by ID-LC/MS, using high purity amino acid certified reference materials (CRMs) as calibrators. Then, results of apolipoprotein quantification can be traceable to the SI units via an unbroken traceability chain. However, the only available reference materials (RMs) for apolipoprotein quantification were produced in the early 1990s to harmonize IN and IT assays [64]. These materials were endorsed by the World Health Organization (WHO) and widely used to recalibrate routine apolipoprotein immunoassays but were never intended for standardization purposes. It appears nevertheless that most ID-LC/MS methods for apolipoprotein quantification use these WHO RMs as external calibrators [60], [61], [62].

Given its high accuracy, good comparability with IN assays [59], [60], possible SI traceability and high throughput, ID-LC/MS is one of the candidate reference methods for apoB and apoA-I quantification in serum. However, this method uses expensive materials, and although “turn-key” approaches have been reported [62], ID-LC/MS requires trained technical staff and dedicated instrumentation. This method is thus mostly used for research purposes and has neither been involved in clinical studies nor transferred to clinical laboratories for routine.

Vertical auto profile (VAP^®)

Vertical auto profile (VAP) was developed in the 1980s by Chung et al. [65] and Cone et al. [66] and is a semiautomated system which protocol was derived from lipoprotein separation by sequential UC. The system underwent various optimizations and was commercialized by Atherotech (Birmingham, AL, USA) until 2016 in different upgrades: VAP-II-fingerstick^® (VAP-II-fs), which provides the lipoprotein profile of a patient from a minimum amount of plasma (18 μL) [67], and VAP-II, a similar system with better resolution and performances that however necessitates larger amounts of plasma [68].

VAP-II^® analysis is a two-step procedure. First, lipoproteins are separated by Single Vertical Spin density-gradient UC [65], [69]. Plasma density is adjusted with KBr to a density equal to, or greater than, the densest lipoprotein to be separated. This density-adjusted plasma is then introduced into a centrifuge tube and layered under a KBr or NaCl solution of density equal to, or lower than, the density of the least dense lipoprotein to be separated. The tube is then ultracentrifuged in a vertical rotor at approximately 720,000 g for an hour. The discontinuous gradient ensures the sufficient separation of lipoproteins according to their respective flotation rates: a function of their density, size and conformation. The densest ones thus end up in the bottom of the tube, whereas the least dense ones are in the upper zone of the tube [69]. These separated lipoproteins are then quantified using their cholesterol content with an automated continuous enzymatic assay. For that purpose, the centrifuge tubes are placed in a gradient fractionator, and the gradient is continuously drawn from the tube and mixed in a chamber with an enzymatic cholesterol reagent. Absorbance is then measured at 505 nm [65]. The continuous measure of the absorbance is reported as a function of the relative gradient position, and deconvolution of the signal by software-assisted data processing finally allows the determination of the cholesterol concentration associated with each lipoprotein class and subclass, thus providing the lipoprotein profile [66]. Additionally, cholesterol concentrations can be further converted into apoB-equivalent concentrations via an algorithm included in the software [70].

VAP-II^® targets the cholesterol contained in lipoproteins using a specific enzymatic assay. The quantity measured is thus the colored product of the reaction between cholesterol and enzymatic reagent. Method linearity and reproducibility proved to be satisfactory with coefficients of variation lower than 5% for TC and for cholesterol subclasses concentrations [71]. Concerning accuracy, VAP-II LDL-C concentrations were compared to those measured with the Abell Kendall method at the Northwest Lipid Research Laboratory (NWLRL, Seattle, WA, USA), a reference laboratory for cholesterol measurements. Results were found in good agreement [71]. Concerning apoB concentration measurements by VAP-II, the equations involved to convert cholesterol concentrations into apoB equivalents were determined by correlation with an IT assay calibrated with the WHO RM for apolipoproteins [70] rendering apoB concentrations by VAP traceable to this material.

Nuclear magnetic resonance (NMR)

In 1991, Otvos reported a new method for lipoprotein quantification using a spectroscopic method: proton nuclear magnetic resonance (H-NMR) [72]. This method was automated and commercialized as the NMR LipoProfile^® (LabCorp, Burlington, NC, USA) assay and was recently cleared by the FDA for cholesterol concentration measurements [73]. Its clinical relevance for CVD risk assessment was demonstrated several times in the context of clinical trials [15], [21], [32]. Very recently, a new two-dimensional H-NMR assay, the DOSY LipoScale^® (Biosfer Teslab, Tarragona, Spain), was also developed for lipoprotein quantification [74], [75].

When submitted to a high-frequency magnetic field, protons contained in molecules or proteins produce resonance spectra that are specific to their chemical environment. It was evidenced by Otvos et al. that lipoproteins in plasma had specific resonance signatures [76], and Lounila et al. demonstrated a relationship between H-NMR resonance frequency and lipoprotein diameters [77]. Lipoprotein H-NMR spectroscopy measures the specific resonance signature of the particles’ lipid terminal methyl groups [78]. Otvos suggested an interesting analogy with bells to further explain NMR principles [72]. Similarly to bells of different size having different sound signals, different lipoproteins broadcast different lipid signals depending on their size. In addition, as the sound loudness is expected to be proportional to the number of bells, the amplitude of the lipid resonance signal is expected to feature the amount of lipids in the particle. H-NMR analysis thus deconvolutes a composite sound signal of lipoproteins in plasma to extract their specific signal and amplitude, i.e. respectively their diameter and concentration.

LipoProfile^® was the first available assay for lipoprotein quantification by H-NMR. It uses a linear least square regression model to deconvolute the H-NMR spectra measured [79]. To reconstruct the different peaks corresponding to each lipoprotein class and subclass, this software relies on a library of lipid H-NMR spectra obtained from lipoprotein fractions prepared by UC and further characterized in size and lipid composition using GGE or electron microscopy and chemical analyses [78]. Particle concentrations are then expressed either as TG or TC concentrations [72], and results can be extracted afterward as a proprietary test report adapted for physicians and detailing lipid concentrations and risk status.

Recently, a new alternative to H-NMR spectroscopy was suggested using 2D diffusion-ordered H-NMR spectroscopy (DOSY) and was patented as a new assay for lipoprotein quantification: the LipoScale^® [74], [80]. This method measures H-NMR spectra under a strength gradient resulting in changes in the H-NMR resonance intensities. These changes depend notably on the diffusion coefficient of the species that generate the resonance. Lipoprotein diffusion coefficients are estimated by DOSY and derived as hydrated radii using the Stokes-Einstein equation [81]. DOSY cholesterol concentrations of lipoprotein subclasses are calculated using the same principle as 1D-H-NMR, i.e. using data-processing software similar to that of LipoProfile^®. This software was however further developed and modified to improve peak deconvolution using statistical approaches [82].

H-NMR thus measures proton resonance of lipids contained in lipoproteins. As this resonance is produced by all lipids in the particle, it is often considered that H-NMR directly measures lipoprotein particles. H-NMR requires trained technical staff and dedicated instrumentation, but the development of automatic and affordable assays for lipoprotein profiling has enabled its widespread use in clinical trials [15], [21], [32], [83]. Absence of lipoprotein separation steps and short run times have made H-NMR one of the most used ALT methods for clinical trials and research over these past decades. However, H-NMR measures lipid concentrations by resonance and derives this concentration further into lipoprotein particle concentrations, relying on the hypothesis that a fixed mean quantity of lipids is contained in each lipoprotein. In addition, lipoprotein quantification accuracy largely relies on the processing software used for signal deconvolution which processes highly complex spectra using an experimental library. No data were found on the way processing algorithms were established nor on system’s calibration and therefore, traceability of the results remains unclear.

Electrospray differential mobility analysis (ES-DMA)

Electrospray differential mobility analysis (ES-DMA), also known as ion mobility analysis, is the most recent ALT method. First reported in 1998 by Kaufman, it was initially applied to nanoparticle and macro-ion size measurements [84]. Its first application to lipoprotein profiling was reported in 2008 in a publication by Caulfield et al. [85], [86]. This method is now available as a routine diagnostic test exclusively run by Quest Diagnostics (Madison, NJ, USA) and has already been involved in clinical trials [33], [87].

ES-DMA is a system that selects and counts intact lipoprotein particles in the aerosol phase. Lipoproteins in serum are aerosolized with an electrospray interface including a neutralization source used to apply a known charge distribution to the generated aerosol. Downstream, aerosolized lipoproteins are selected using a differential mobility analyzer (DMA) composed of a drift tube in which lipoproteins, submitted to an electric field ramp at atmospheric pressure, are selected gradually depending on their electrical mobility diameter. The selected lipoproteins are then counted by laser detection in a condensation particle counter [88]. Finally, results are reported as a number size distribution that represents the number of particles counted per cubic centimeter of air at each mobility diameter. Integrating the peaks of interest, i.e. summing all counts on a diameter range, thus provides the particle concentration measured by the system in the aerosol phase. However, the key step for particle number quantification by ES-DMA is the postanalytical processing to further derive this aerosol phase particle concentration into a liquid phase particle concentration, i.e. the concentration in the initial sample. Various approaches have been reported, but to date, debates remain concerning the most adapted method and harmonization of the process have not yet been achieved [86], [89], [90].

Although ES-DMA has proven its relevance for lipoprotein testing, it has been very little implemented in clinical and research laboratories. Indeed, although automation and high throughput proved to be achievable for ES-DMA, these remain costly and require expert technical staff. In addition, ES-DMA is sensitive to interferences, especially generated by serum proteins, and specific sample preparation steps are often necessary to obtain an accurate lipoprotein profile. Nevertheless, one advantage of ES-DMA is that it measures in a short time the full lipoprotein profile of the sample without data deconvolution. Lipoprotein classes may not be fully resolved, but their concentrations can be calculated over specified diameter ranges [86]. Contrary to most methods reported for lipoprotein profiling and quantification, ES-DMA is the only one for which the measurand is the full intact lipoprotein [91]. However, it was demonstrated that calibration with appropriate standards is necessary to achieve accurate quantification and a study of results comparability with IN revealed important variability depending on the calibration material chosen [90]. As for diameter measurement accuracy, ES-DMA systems are generally calibrated with CRMs of inorganic nanoparticles and provide highly precise measurements [90], [92]. It is however important to note that the diameters measured with ES-DMA are electrical mobility diameters, i.e. dry diameters, which can therefore not be compared to hydrated diameters such as ones measured by GGE.