Quantification of HDL particle number (HDL-P) by proton NMR: Don’t believe the numbers

Abstract

Purpose of review:

Proton nuclear magnetic resonance (NMR) can rapidly assess lipoprotein concentrations and sizes in biological samples. It may be especially useful for quantifying high density lipoprotein (HDL), which exhibits diverse particle sizes and concentrations. We provide a critical review of the strengths and limitations of NMR for quantifying HDL subclasses.

Recent findings:

Recent studies using NMR have shed light on HDL’s role in various disorders, ranging from residual cardiovascular risk to host susceptibility to infection. However, accurately quantifying HDL particle number, size, and concentration (HDL-P) remains a challenge. Discrepancies exist between NMR and other methods such as gel electrophoresis, ion mobility analysis and size-exclusion chromatography in estimating the abundance of HDL species and the ratio of apolipoprotein A-I (APOA1) to HDL particles.

Summary:

NMR is a low-cost method for quantifying HDL-P that is readily applicable to clinical and translational studies. However, inconsistencies between the results of NMR quantification of HDL-P and other independent methods hinder the interpretation of NMR results. Because proton NMR apparently fails to accurately quantify the sizes and concentrations of HDL, the relevance of such studies to HDL biology poses challenges. This limits our understanding of pathophysiological implications of HDL-P as determined by NMR, particularly in determining cardiovascular disease (CVD) risk.

Introduction.

Many lines of evidence indicate that low levels of HDL, quantified by its cholesterol content (HDL-C), are a powerful risk factor for premature atherosclerotic cardiovascular disease (CVD) [1]. However, recent evidence has cast doubt on the idea that increasing HDL-C levels necessarily provides therapeutic benefits [2]. For instance, genetic variations associated with changes in HDL-C levels do not consistently correlate with alterations in the risk of cardiovascular disease. Additionally, trials of two drugs aimed at raising HDL-C levels were terminated prematurely due to an apparent lack of clinical benefit in individuals already suffering from established CVD. These observations have led to the suggestion that low HDL-C is a marker for other conditions, such as insulin resistance, that promote CVD rather than a direct contributor to the disease.

A significant challenge in clinical practice is establishing a clear connection between the molecular mechanisms responsible for HDL’s protective effects on the heart and HDL metrics applicable to translational and clinical research [3]. This in part reflects the heterogeneity of HDL: HDLs exhibit a wide range of sizes and concentrations, reflecting their structural and functional diversity within the body [4]. These particles can vary significantly in diameter, spanning from approximately 6 to 14 nanometers. Additionally, their concentrations in the bloodstream can vary across individuals and physiological conditions. This heterogeneity underscores the complex nature of HDL’s metabolism and its role in lipid transport and cardiovascular health [2]. Understanding the diverse sizes and concentrations of HDL particles is crucial for unraveling HDL’s physiological functions and their roles in disorders such as CVD [3].

These observations suggest that measuring HDL particle concentration (HDL-P)—the concentration of total HDL particles and the sizes and concentrations of HDL subclasses—might offer a more accurate picture of HDL’s heart-protective effects. Thus, methods such as nuclear magnetic resonance (NMR) and ion mobility analysis (IMA) have been used to quantify HDL-P, providing potentially more informative data than traditional cholesterol measurements [5].

Quantifying HDL-P by proton NMR.

Otvos and colleagues first described the quantification of lipoprotein particles by proton NMR in a seminal paper published in 1991 [6]. They suggested that the sizes and concentrations of specific subclasses of lipoproteins might serve as a new metric for assessing the risk of CVD in humans that was distinct from the cholesterol content of lipoproteins.

Proton NMR is a powerful technique for quantifying the concentrations and sizes of lipoproteins, including HDL particles, in biological samples. Lipoproteins are characterized by the signals from the methyl group hydrogen atoms (protons) within their molecular structures. Each type of lipoprotein, such as HDL, LDL, and VLDL, produces distinctive NMR signals due to differences in lipid and protein composition affecting chemical shift in the proton resonance [7].

To quantify lipoprotein concentrations, proton NMR spectroscopy measures the intensity of specific NMR signals corresponding to different lipoprotein subclasses in a blood sample based on subtle differences in chemical shift.[7] By comparing these signal intensities to known standards or calibration curves, the concentrations of different lipoprotein subclasses can be determined. Moreover, proton NMR can provide information about the sizes of lipoprotein particles from the chemical shift differences arising from increasing particle size and hence chemical environment in the particle. By analyzing the linewidths of these signals, researchers can estimate the average sizes of lipoprotein particles within a sample. This allows them to characterize lipoprotein size distributions and identify alterations in lipoprotein metabolism associated with various physiological conditions such as dyslipidemia and cardiovascular disease[7].

NMR quantification of HDL-P is thus a potentially powerful metric because HDL is composed of heterogenous mixtures of sizes of particles that vary greatly in lipid composition and relative concentration, both between the different subclasses in an individual and between different individuals and has been commercially implemented in two major approaches – NMR LipoProfile (by Labcorp) and Nightingale platform (by Nightingale Health).

Advantages of NMR for clinical studies.

Because NMR is inexpensive and high-throughput method, NMR quantification of HDL-P has been widely used in clinical studies, the associations between specific classes of lipoprotein particles and a wide range of factors ranging from risk of cardiovascular disease to menopause [8–15]. Studies support the proposal that the association of HDL-P with CVD risk is independent of HDL’s cholesterol content (HDL-C). For example, in a randomized primary prevention trial comparing rosuvastatin treatment to placebo (a nested case-control study within the JUPITER trial—Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin), researchers examined subjects with normal low-density lipoprotein cholesterol levels but elevated C-reactive protein levels to see whether HDL-P (quantified by NMR), HDL-cholesterol (HDL-C), apolipoprotein A-I (APOA1, HDL’s major protein), or cholesterol efflux capacity best predicted residual CVD risk. Among those four HDL-related biomarkers, HDL-P emerged as the most robust inverse predictor of incident events and as a biomarker indicating residual risk [16, 17].

A recent study found that higher HDL particle number (HDL-P) was linked to reduced risk of heart failure [18]. Conversely, larger HDL particle size associated with heightened risk. No significant association was found between HDL-C levels and heart failure risk after adjusting for other cardiovascular risk factors [18].

NMR links HDL-P to infection.

Recent clinical studies found an association between HDL and the body’s defense mechanisms. For example, in a study of more than 400,000 people who were monitored for 6 years, HDL-C, LDL cholesterol, and triglyceride levels associated with risk of hospitalization due to infectious disease [19] Adjusting for HDL-C levels eliminated the association between LDL cholesterol, triglycerides and risk, pointing to HDL as the primary risk factor. To validate these findings, the researchers developed a polygenic risk score based on genetic variations related to lipid levels. Only the HDL cholesterol polygenic score significantly and inversely associated with risk of infection-related hospitalizations. These results strongly imply that low HDL-C levels increase the risk of infection.

Harsløf et al. recently extended these observations to HDL-P by determining whether low levels of certain HDL sizes were also linked to risk of infectious disease [20]. In a study of more than 30,000 participants from Copenhagen’s general population, they found that low levels of small and medium-sized HDL, but not large and extra-large HDL, associated with higher risk of morbidity and mortality from infectious disease. These findings are intriguing as they might indicate underlying mechanisms; it is known that cholesterol levels in cell membranes help regulate Toll-like receptor signaling in immune cells. However, the association of HDL size with infectious disease risk was not adjusted for low HDL-C levels.

Another recent NMR study of HDL-P linked low concentration of small HDL particles to a higher risk of sepsis, sepsis-related death, and admission to critical care due to sepsis. Genetic analyses suggested the relationship was not causal, however. Instead, the findings suggested that increased IL-6 signaling could potentially confound the associations observed with reduced HDL particle count. The authors proposed that this confounding effect might partly explain the observed association between (small) HDL particle count and sepsis [21].

Limitations of quantifying HDL-P by NMR.

An important issue with any analytical method is the accuracy of its results. A number of observations using alternative methods suggest that using NMR to measure HDL-P fails to accurately assess the sizes and concentrations of subclasses of HDL particles (Figure). In a study that linked HDL-P to risk of infection, a calculation based on median values of the sizes of APOA1 and HDL gave a stoichiometry of APOA1 per HDL particle of ~7.7 mol/mol ([20]). Another commercial NMR method yielded a stoichiometry of ~1.5 APOA1 per particle [22]. Both values are inconsistent with a wide range of biochemical and biophysical studies demonstrating a ratio of APOA1 per HDL particle that ranges from 2 to 5 [4, 23, 24].

Figure. Comparison of the distribution of HDL particles measured by different methods.

Distribution of HDL particles in plasma of a single subject was measured by 1H-NMR (7 classes measured by LP4-algorithm normalized to total HDL-P) (A), calibrated differential ion mobility analysis (cIMA) (full particle distribution profile) (B), and native non-denaturing gradient gel electrophoresis with anti-APOA1 antibody staining (NDGGE) (C), and of the same subject’s HDL isolated by sequential density ultracentrifugation measured by negative stain transmission electron microscopy (TEM) (D).

The Nightingale NMR method can simultaneously quantify a wide range of metabolic biomarkers, including all classes of lipoproteins. It was extensively characterized in a large population-based study [25]. That study produced a value a ratio of 6.7 for APOA1 per HDL particle [25]. When the authors examined the scatter plot of HDL-P and APOA1, they found a linear correlation of r=0.86 (Supplemental Figure S7 in ref. [25]). In contrast, calculations based on the Shen model of HDL yield a curvilinear relationship between size and the APOA1/HDL ratio [25]. A similar linear relationship (r=0.93) was found for LDL-P and APOB [25]. It is well established that LDL carries one molecule of APOB per particle. In contrast, as noted above, the stoichiometry of APOA1 per particle varies from 2 to 5 for HDLs that range in size from small to large. These observations suggest that the method quantifies APOA1 rather than HDL-P.

When the Nightingale method determines the distribution of HDL particle sizes, it produces a log-linear pattern (Supplemental Figure S4 in ref [25]) with the smallest HDL particles being the most abundant (55%). However, gel electrophoresis detects 3 major sizes of HDL, with medium-sized HDL particles as the most abundant subpopulation in plasma [4][26]. The electrophoretic method cannot assess particle concentration quantitatively, however, because it quantifies APOA1, the major HDL protein, without considering the APOA1/HDL particle stoichiometry of different sizes of HDL. Nevertheless, it clearly shows that small HDLs are only a minor component of the total HDL population. A careful quantitative study of HDL isolated by sequential floatation and size-exclusion chromatography also demonstrated that medium-sized HDL particles represent the most abundant subclass [27].

Other potential limitations of proton NMR.

Proton NMR detects and quantifies intensity and chemical shift of the line shapes of lipids’ methyl and methylene resonances [6, 7]. An unresolved issue is how the lipid composition of HDL particles affects the proton resonances quantified by NMR. It is well established that HDL particles vary widely in lipid composition in different clinical conditions and that diet has a major impact on HDL’s lipid composition. These observations raise the possibility that changes in HDL size and concentration as determined by NMR actually reflect changes in the particles’ lipid composition.

There are other issues with using NMR to quantify HDL-P. Commercial algorithms are proprietary, making it difficult to independently assess the basis for particle quantification. Also, the method is not standardized among different laboratories. As noted above, two commercially available methods give very different values for the stoichiometry of APOA1 per particle. Also, the assays have not been standardized against other validated methods for quantifying HDL particles. It is not known if the algorithms have become modified over time, making it difficult to compare data between different studies. Collectively, these issues raise major concerns over using NMR to quantify physiologically relevant HDL-P. Indeed, in our opinion the values reported by NMR should not be called HDL-P. Instead, they might be better termed HDL-NMR.

Quantifying lipoprotein size by NMR.

Remarkably few studies have directly compared NMR results with those obtained by other methods for quantifying lipoproteins. One study compared gradient gel electrophoresis with NMR quantification of low density lipoprotein particle size in 324 individuals, including 152 with type 1 diabetes and 172 controls[28]. The two methods correlated only modestly, with an average difference of 5.38 nm (NMR showing smaller sizes). Agreement varied, as fewer than half of those classified as pattern B on gradient gel electrophoresis were classified as such on NMR. Pattern B, which associates predominantly with small and dense LDL cholesterol particles, is often seen when HDL cholesterol levels are low, triglyceride levels are elevated, and there is a tendency toward high blood sugar levels and type II diabetes mellitus. Agreement between the two methods was lower in diabetic subjects, women, and those with lower triglyceride levels. Moreover, another detailed analysis found very poor agreement between NMR and other methods [5, 29]. These findings suggest that NMR and gradient gel electrophoresis measurements of LDL and HDL size are not consistent with each other, particularly in women and diabetic people.

Clinical utility of NMR.

The potential clinical utility of proton NMR in lipoprotein analysis is notable for several reasons, making it a valuable tool in clinical studies [25]. First, it is relatively inexpensive, high-throughput, and requires only small volumes of plasma or serum; therefore, it is practical for large-scale clinical research and routine medical practice. This accessibility facilitates widespread adoption and utilization across various healthcare settings. Second, it has the potential to comprehensively quantify diverse classes of lipoprotein particles.

Numerous studies have highlighted the potential of HDL-NMR to provide clinically relevant information. For instance, findings from the JUPITER trial suggest that low levels of HDL-NMR could serve as a marker for identifying statin-treated individuals who still have residual increased risk of CVD and therefore would benefit from more aggressive therapy [30]. This underscores the potential importance of HDL-NMR in risk stratification and personalized treatment approaches.

While HDL-NMR holds promise, the associations between HDL-NMR and CVD risk found in multiple studies has been inconsistent. Moreover, a notable study in women found that while NMR lipoprotein profiles predicted CVD as well as standard lipid measurements and immunoassays of apolipoproteins, they did not demonstrate superiority [31]. This emphasizes the need for further research to elucidate the comparative strengths and limitations of NMR-based lipoprotein particle analysis in clinical risk assessment.

An alternative to NMR.

Ion mobility analysis (IMA, also known as differential ion mobility analysis, DMA) is a method for quantifying lipoprotein particles and especially HDL-P [22, 32]. One of its main strengths is its ability to precisely determine particle size, as it relies on the physics of charged particles moving in an electric field. However, IMA is only semi-quantitative because various factors affect the ionization and detection of particles during analysis. To address this issue, we developed calibrated IMA, where known particle concentrations are used to calibrate the concentrations of HDL particles [22]. Calibration allows the conversion of IMA signal intensity, originally a relative measurement, into a metric for absolute particle concentration (μmol/mol). Importantly, this method was validated using known concentrations of monodisperse gold nanoparticles and reconstituted HDL particles [22]. Quantifying HDL particle concentration with calibrated IMA produces a distribution of HDL subclasses that aligns well with results from gel electrophoresis and transmission electron microscopy (Figure) or size-exclusion chromatography. Furthermore, the APOA1/particle stoichiometry of 3 to 4 obtained with calibrated IMA is consistent with our current understanding of HDL structure.

Conclusions.

In summary, the interpretation of NMR studies is challenged by the unclear relationships between HDL-NMR profiles and the actual sizes and concentrations of HDL subpopulations. This is potentially a major problem because we and others have proposed that some sizes of HDL reduce CVD risk more strongly than others [33, 34]. Without a clear understanding of these relationships, it becomes difficult to discern the pathophysiological relevance of NMR findings and their implications for HDL biology. Thus, ongoing efforts to elucidate the associations between NMR-derived lipoprotein profiles and the underlying physiology of HDL metabolism are essential for optimizing the clinical utility of this technique and informing advances in cardiovascular risk assessment and management.

Key Points:

• Proton NMR is a high-throughput method for quantifying HDL-P that is readily applied to clinical and translational research.

• NMR has been proposed to rapidly quantify the sizes and concentrations of HDL subclasses as well as total HDL particles.

• NMR assessment of key HDL subclasses differ in relative abundance and size from those determined by orthogonal, independent methods.

• The relevance of HDL-P to the actual concentration of HDL subclasses is unclear, making it difficult to link those metrics to our current understanding of HDL biology.