| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Lipids, Lipoproteins, and Cardiovascular Risk Factors |
1 Quest Diagnostics Nichols Institute, San Juan Capistrano, CA; 2 Children’s Hospital Oakland Research Institute (CHORI), Oakland, CA; 3 Lawrence Livermore National Laboratory, Livermore, CA.
aAddress correspondence to this author at: Quest Diagnostics Nichols Institute, 33608 Ortega Highway, San Juan Capistrano, CA 92675. E-mail michael.p.caulfield{at}questdiagnostics.com.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: After an ultracentrifugation step to remove albumin, we determined the size and concentrations of lipoprotein particles in serum samples using IM. Scan time is 2 min and covers a particle range of 17.2–540.0 Å. After scanning, data are pooled by totaling the particle number across a predetermined size range that corresponds to particular lipoprotein subclasses. IM results were correlated with those of standard methods for cholesterol and apolipoprotein analysis.
Results: Intra- and interassay coefficients of variation for LDL particle size were <1.0%. The intra- and interassay variation for LDL and HDL particle subfraction measurements was <20%. IM-measured non-HDL correlated well with apolipoprotein B (r = 0.92).
Conclusions: The IM method provides accurate, reproducible, direct determination of size and concentration for a broad range of lipoprotein particles. Use of this methodology in studies of patients with cardiovascular disease and other pathologic states will permit testing of its clinical utility for risk assessment and management of these conditions.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Currently, there is no standardized method for measuring concentrations and distributions of lipoprotein particle subpopulations; each of the clinically available methods (vertical ultracentrifugation, nuclear magnetic resonance, and gradient gel electrophoresis) requires some form of algorithm to generate the lipoprotein profiles and/or particle concentrations.
Gas-phase differential electrical mobility, also known as ion mobility (IM), is well established in the field of aerosol science for particle analysis and measurement (10). The development of an electrospray interface (11) permitted the system to be used to analyze particles in solution, and its use was soon extended to analysis of DNA (12), proteins(13), and a variety of other biological materials (14)(15)(16)(17). A patent describes its use for measuring the size distribution of lipoprotein particles (18). In this study, we describe and validate the use of IM to measure the distributions and concentrations of plasma lipoprotein particles, covering the spectrum of HDL (approximately 75 Å diameter) to larger, very-low-density lipoprotein (VLDL) (approximately 520 Å diameter), and establish summary statistics characterizing the result distributions commonly seen in healthy adults.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. In the IM method for lipoprotein fractionation, the lipoproteins in a volatile solution are introduced into a flow of air (1.6 L/min) containing approximately 5% CO2 by means of an electrospray. In the electrospray chamber, the desolvated and highly charged lipoprotein particles are nearly neutralized by ionized air, introduced by a polonium 
-particle emitter present in the chamber. The proportion of singly charged particles emerging from the electrospray chamber can be calculated using Fuch charge distribution (19). Fuch charge distribution describes the probability that a submicron-diameter particle of any given diameter will have exactly N charges. The particles are then carried in the airflow to the differential mobility analyzer (DMA). As the particles enter the top of the DMA, they are confined in a thin flow stream by a laminar concurrent flow of air called a sheath flow. The particle-free sheath flow recirculates through the DMA at a 20 L/min flow rate. As the particles are carried through the DMA, an electric potential across the sheath flow causes the particles to drift toward a collection slit. Ramping the applied potential causes particles of different diameters to pass through the slit, thus allowing lipoprotein particles between 30 and 542 Å to be sampled; the size range can be extended to include larger particles if desired. At any given electrical potential, particles of predictable size pass through the collection slit and enter a separate air stream (1.6 L/min) that carries them to a particle counter. The particle counter first enlarges the particles by condensing a vapor onto each particle and then detects the droplets, now several microns in diameter, via light scatter. Knowledge of the electrical potential applied to the DMA, the dimensions of the DMA, and the flow rate of air passing through the DMA permits accurate calculation of particle diameter and the number of particles in a discrete size range. Particle diameter is determined from the following equation:
 |
|
The number of particles detected in 0.1-s bins is recorded and used to determine distributions. The data in a time bin is converted into particle number concentration (number/mL) and also into particle mass concentration (mass/mL). When these values are plotted against particle diameter, particle number or particle mass distributions are generated. An example of the raw particle count data is shown in Fig. 2A
(particle number corrected for dilution) for a control sample, prepared by an overnight ultracentrifugation. Fig. 2B
shows the data converted to arbitrary mass units using the formula: mass = density x volume (a density of 1.0 was used for all lipoprotein particles).
|
sample preparation
We briefly mix serum samples or controls by vortex-mixing, then mix 5 µL of sample or control with 20 µL of an albumin removal reagent [7.5 g/L Reactive green 19 dextran (RGD), Sigma-Aldrich; 2.5 g/L dextran sulfate, Sigma-Aldrich; and 0.5 g/L EDTA, Spectrum Chemicals] and incubate them on ice for 15 min. After incubation, we overlay the sample mixture on 200 µL deuterium oxide (Medical Isotopes) in a 42.2 ultracentrifuge tube (Beckman Coulter). The samples are ultracentrifuged at 10 °C for 135 min at 223 000g (42 000 rpm), and then we remove the top 85 µL (i.e., the lipid fraction) of the sample. We dilute the samples 1:800 for HDL analysis using 25 mmol/L ammonium acetate, 0.5 mmol/L ammonium hydroxide, pH 7.4. For LDL analysis, samples are diluted 1:200 with the same diluent containing 5 µg/mL dextran sulfate to help prevent LDL particles from sticking to the capillary surfaces. Final dilutions are made in deep-well 96-well plates and placed in a Leap HTC PAL autosampler (Eksigent) with the cooled stack maintained at 6 °C.
lipoprotein analysis
The autosampler is connected to the electrospray generator (Model 3480; TSI) via methyl-deactivated silica capillary (50 µm i.d.; SGE). Flow is introduced by nano-LC pumps (Eksigent) running a mobile phase of 25 mmol/L ammonium acetate, 0.5 mmol/L ammonium hydroxide, pH 7.4. By means of a capillary metal union (Upchurch Scientific), an autosampler injects 10 µL sample at 6 µL/min into a transfer capillary (methyl-deactivated, SGE, 50 µm, 33 cm long). High voltage (2.1 kV) is applied to the metal capillary union located 33 cm upstream of the electrospray unit. The electrospray Taylor cone is monitored visually and amperometrically to ensure stability. After the sample has filled the capillary and reached the electrospray chamber, the flow is decreased to 200 nL/min and the data recording process is started. The air (containing approximately 5% CO2) flowing into the electrospray chamber is regulated at 1.6 L/min.
The electrosprayed particles pass through a particle-charge neutralizing chamber and then enter the DMA. Scan time is 2 min and covers a particle range of 17.2 to 542.0 Å. After a scan is completed, data for specific ranges of particles corresponding to lipoprotein subclasses are pooled by totaling the particles across a predetermined set of 0.1-s bins that corresponds to particular subclasses, and the predominant LDL particle size (modal diameter) is determined.
volunteers for determination of reference distributions for lipoprotein particle size and subfraction concentrations
Nonmedicated (including hormone replacement therapy and birth control), apparently healthy Quest Diagnostics employee volunteers were recruited to establish reference distributions for the lipoprotein fractions detected by IM. The study was approved by the institutional review board, and all volunteers gave informed consent. We analyzed samples from 650 individuals (469 women, 181 men) by IM before applying specific exclusion criteria. All subjects had normal complete blood counts and chemistry panels. For inclusion in the reference population, individuals had to meet NCEP ATP guidelines (2) for optimal plasma lipid and lipoprotein concentrations: total cholesterol <200 mg/dL, triglycerides <150 mg/dL, LDL cholesterol <130 mg/dL, and HDL cholesterol >40 mg/dL (men) or >50 mg/dL (women). After applying the exclusion criteria, 68 men and 191 women remained. Total cholesterol, triglycerides, and HDL cholesterol to determine eligibility for inclusion in the reference distributions and for method comparison were run on a Hitachi 917 (Roche Diagnostics) using Roche Diagnostics reagents.
determination of particle diameter boundary intervals for lipoprotein fractions quantified by the im procedure
Fig. 3
displays quantitative profiles of lipoprotein particles as a function of particle diameter for 3 representative subjects. Shown in this figure are the boundaries used to define major lipoprotein subfraction groupings. We ascertained 2 HDL subfraction categories, HDL large (equivalent to HDL2b, 105–145 Å) and HDL small (equivalent to HDL3 + 2a, 76.5–105 Å), using visual inspection of individual profiles and based on conformity of these size intervals with those previously determined by other methods (20). HDL large was discerned as a distinct peak in most samples (Fig. 3
). Particle size boundaries within the IDL to LDL range were determined by IM analysis of subfractions isolated by density gradient ultracentrifugation from 8 subjects using published procedures (21)(22). The major subfractions were defined according to previous nomenclature (6) as IDL large (IDL1, 268.2–296 Å), IDL small (IDL2, 238–268.2 Å), LDL large (LDL I, 219.9–238 Å), LDL medium (LDL II, 211–219.9 Å), LDL small (LDL III, 201.7–211 Å), and LDL very small (LDL IV, 180–201.7 Å). In the case of VLDL-sized particles, IM particle distributions were inspected visually to select boundaries that were used to define 3 regions: small (296–335 Å), medium (335–424 Å), and large (424–520Å). VLDL particles >520 Å were not measured in this scan but will be the focus of future studies.
|
interference
Matrix specificity was minimized in this procedure through a) initial enrichment of the lipoproteins from the serum/plasma proteins and b) separation of the lipoprotein particles by size in the IM. Compounds such as hemoglobin are removed during the ultracentrifugation; small compounds such as bilirubin, while not completely sedimenting in the ultracentrifuge, appear before the lipoproteins in the IM analysis because of their small size.
statistics
We used regression analysis to assess the relationships between biochemical and IM lipoprotein measurements.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
20%, could not be determined because of the lack of unique standards for lipoproteins of known size and concentration; therefore the mean + 10 SD was used as a substitute (23). We determined the concentration of particles in each lipoprotein fraction mathematically using Avogadro’s number, the volume of sample measured, and the sample collection time. (This was possible as the method quantifies particles directly.) The formula to convert from particle number to concentration is:
 |
Based on these data, the LOD and LOQ for particle concentrations ranged from 0.01 nmol/L and 0.02 nmol/L, respectively, for the VLDL large category to 10.6 nmol/L and 38.6 nmol/L for the HDL small category (see Supplemental Table 1 that accompanies the online version of this article at http://www.clinchem.org/content/vol54/issue8).
reproducibility and precision
The intraassay precision for all lipoprotein fractions was <20% (see Supplemental Table 2, A–C, in the online Data Supplement). The highest CVs were associated with IDL and VLDL, fewer particles of which were present. The variation seen in these fractions is a combination of the sample preparation as well as the IM analysis. The intraassay variation of a control sample run 22 times was <9.0% for all HDL and LDL fractions and <13.1% for all lipoprotein fractions (see Supplemental Table 2D in the online Data Supplement).
The reproducibility (interassay mean, SD, and CV) for each sample is shown in Supplemental Table 3 in the online Data Supplement. As was the case for intraassay variation, higher CVs were observed for fractions with lower particle concentrations. CVs were <20% for LDL and HDL and <26% for IDL and VLDL. Both intra- and interassay CVs for LDL particle size were <1.0%.
specimen stability
We used EDTA plasma samples from 5 individuals to evaluate the stability of the lipoproteins at room temperature (20–26 °C), under refrigeration (2–8 °C), after freezing (<–20 °C), and after multiple freeze/thaw cycles at <–70 °C. Samples were divided into sufficient numbers of aliquots for the appropriate time points at each temperature. Values were expressed as a percentage of the day 0 time-point value. Samples were stable for 5 days at 20–26 °C, 7 days at 2–8 °C, and 28 days at –20 °C. Five freeze-thaw cycles had no significant effects on the results (see Supplemental Table 4 in the online Data Supplement).
We also assessed the stability of prepared samples to determine the length of time they could be stored before analysis. Analysis of samples stored in the cooled autosampler tray for a period of 3 days showed no change in lipoprotein profile (see Supplemental Fig. 1 in the online Data Supplement).
sample carryover
The capillaries in the system are flushed between samples to minimize the potential for carryover. To confirm that this washing is sufficient, we analyzed sample diluent (ammonium acetate with dextran sulfate) used for LDL sample dilutions after a 1:200 dilution sample. There was a small amount of carryover in the LDL region (<1.3%). The overall lipoprotein carryover was <1.5% (see Supplemental Fig. 2 in the online Data Supplement).
effect of specimen type
Matched samples were drawn in red top (serum), serum separator (SST), EDTA, and heparin collection tubes. Each collection type from the same individual was prepared and analyzed in the same assay. We evaluated the data using the serum as the target concentration for each lipoprotein region and determining the percent recovery of the other sample types relative to it. Recoveries for the 3 collection tubes were as follows: SST, 82%–101%; EDTA plasma, 91%–114%; and heparin plasma, 88%–95%.
method comparison
We correlated IM-determined concentrations of total HDL, HDL large, total LDL, total IDL, total VLDL, and non-HDL cholesterol (sum of LDL, IDL, and VLDL fractions) with biochemically measured triglycerides, HDL cholesterol, LDL cholesterol (calculated), and plasma apoA1 and apoB. Fig. 4
shows the individual data and correlation (r) for the comparisons among the most closely related measurements. As expected, given that IM measures particle concentrations and not the lipid or protein content of lipoproteins, there were varying degrees of correlation among the measurements. Notably, total LDL particle concentrations as measured by IM correlated more strongly with apoB (r = 0.90) than with LDL cholesterol (r = 0.79), and the correlation with apoB increased slightly for the non-HDL (r = 0.92) (Fig. 4
, A–C). Although linear regression analysis fit these data well, the results using a polynomial fit were slightly improved and went through the origin (non-HDL, r = 0.94). Significant correlations were also observed between VLDL and triglycerides, and between HDL large and total HDL and HDL cholesterol (Fig. 4D
–F). Similar correlations were seen for apoA1 (see Supplemental Fig. 3 in the online Data Supplement).
|
Table 1
gives summary statistics characterizing the result distributions observed in our healthy adult population for the respective lipoprotein fractions as described in Materials and Methods and shown for representative subjects in Fig. 3
.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The method described here is accurate and reproducible, with intra- and interassay CVs of <1.0% for LDL particle size and <18% and 16% for HDL and LDL fractions, respectively. There was more variation in the IDL and VLDL regions (intra- and interassay CVs of <26%) because of the absolute lower number of particles. This variation could be reduced by incorporating a lower dilution and an additional scan on the IM. We did not investigate this in the present study.
The IM method offers advantages over other procedures for lipoprotein particle analysis in that it not only measures particle size accurately on the basis of physical principles, but also directly counts the particles present at each size. This approach thereby provides the only direct measurement of lipoprotein particle size and concentration for each lipoprotein subclass, from small HDL to large VLDL.
The differences between IM and standard measurements of plasma lipid and lipoprotein cholesterol concentrations are due in part to interindividual variations in the compositions of lipoprotein particles. However, there were strong correlations between plasma concentrations of apoB and IM measurements of both LDL and non-HDL particle concentrations (r = 0.90 and 0.92, respectively). The correlations reflect the fact that 1 molecule of apoB is present in each of these particles, and hence that apoB provides a measure of total particle concentration. The data for the large, apoB-containing particles were fitted slightly better by a polynomial regression passing through zero than by a linear regression. This indicates that directly measured particle concentrations are overestimated by apoB, particularly at lower plasma concentrations. This may be due to a limitation of the immunoassay to accurately measure apoB in all lipoprotein particles, and/or the immunoassay measures both apoB100 and apoB48. Further investigation will be required to determine the basis for this finding, but it may indicate a limitation in the use of plasma apoB concentration as a surrogate measure of lipoprotein particle number. The poorer correlation (r = 0.79) between total LDL and LDL cholesterol is likely due to the varying concentrations of cholesterol in these particles. VLDL correlated with triglycerides (r = 0.84) and demonstrated the expected increased content of triglycerides with larger VLDL particles. HDL cholesterol was strongly correlated with HDL large (r = 0.86), but the correlation was weaker for total HDL (r = 0.59). This may reflect a greater variation in either the recovery or composition of HDL small, the densest lipoprotein particle. Similar results were obtained for correlations of HDL large and total HDL with plasma apoA1.
We used the IM procedure to determine reference distributions for particle concentrations for a series of fractions throughout the lipoprotein particle spectrum, as well as for peak LDL diameter, based on a population of individuals who met the NCEP ATP criteria for optimal plasma lipid and lipoprotein cholesterol measurements (2). Use of this methodology in studies of patients with cardiovascular disease and other pathologic states will permit testing of its clinical utility in the assessment of risk and management of these conditions.
|
Acknowledgments |
|---|
Financial Disclosures: M. P. Caulfield, S. Li, G. Lee, W. A. Salameh, and R. E. Reitz are employees of Quest Diagnostics Inc. W. H. Benner was a consultant for Quest Diagnostics during this study. R. M. Krauss received financial support from Quest Diagnostics for his laboratory during this study.
Acknowledgments: We thank J. C. Geaney and Katie Wojnoonski for technical assistance, Colette Scheele and Esther Carlton for their help in collecting and analyzing the reference distribution data, Earl Cornell for help with the programming of the ion mobility instrument interface, and Jeff Radcliff for help with the manuscript preparation.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  K. Musunuru, M. Orho-Melander, M. P. Caulfield, S. Li, W. A. Salameh, R. E. Reitz, G. Berglund, B. Hedblad, G. Engstrom, P. T. Williams, et al. Ion Mobility Analysis of Lipoprotein Subfractions Identifies Three Independent Axes of Cardiovascular Risk Arterioscler Thromb Vasc Biol, November 1, 2009; 29(11): 1975 - 1980. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. R. Superko Advanced Lipoprotein Testing and Subfractionation Are Clinically Useful Circulation, May 5, 2009; 119(17): 2383 - 2395. [Full Text] [PDF]
|
|
|
|
|
|
S. Mora Advanced Lipoprotein Testing and Subfractionation Are Not (Yet) Ready for Routine Clinical Use Circulation, May 5, 2009; 119(17): 2396 - 2404. [Full Text] [PDF]
|
|
|
|
 |
|
 J. H. Contois, J. P. McConnell, A. A. Sethi, G. Csako, S. Devaraj, D. M. Hoefner, and G. R. Warnick Apolipoprotein B and Cardiovascular Disease Risk: Position Statement from the AACC Lipoproteins and Vascular Diseases Division Working Group on Best Practices Clin. Chem., March 1, 2009; 55(3): 407 - 419. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. Otvos, L. L. Rudel, and J. P. McConnell Concerns Regarding Lipoprotein Particle Measurement by Ion Mobility Analysis Clin. Chem., December 1, 2008; 54(12): 2086 - 2087. [Full Text] [PDF]
|
|
|
|
|
|
M. P. Caulfield, S. Li, G. Lee, P. A. Blanche, W. A. Salameh, W. H. Benner, R. E. Reitz, and R. M. Krauss In Reply Clin. Chem., December 1, 2008; 54(12): 2088 - 2089. [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Ruby, D. K. Nomura, C. S. S. Hudak, L. M. Mangravite, S. Chiu, J. E. Casida, and R. M. Krauss Overactive endocannabinoid signaling impairs apolipoprotein E-mediated clearance of triglyceride-rich lipoproteins PNAS, September 23, 2008; 105(38): 14561 - 14566. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
