| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Biochemical Genetics Laboratory, Department of Laboratory Medicine and Pathology, Mayo Clinic and Foundation, Rochester, MN 55905.
2
Department of New Technology and Test Development, South
Bend Medical Foundation, South Bend, IN 46601-1098.
3
Department of Neurology, Mayo Clinic and Foundation,
Rochester, MN 55905.
a Address correspondence to this author at: Mayo Clinic, Department of Laboratory Medicine and Pathology, Hilton 210, 200 First St. SW, Rochester, MN 55905. Fax 507-284-1399; e-mail
mcconnell.joseph{at}mayo.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: LDL subfractions were separated using the Quantimetrix LipoprintTM LDL System. High-resolution 3% polyacrylamide gel tubes were scanned densitometrically (610 nm) with a Helena EDC system. A computerized method to identify and quantitatively score the resolved LDL subfractions was developed. Results from the Quantimetrix method were compared using 51 plasma samples with values obtained by nondenaturing gradient gel electrophoresis (NDGGE) and nuclear magnetic resonance (NMR) spectroscopy.
Results: LDL subfractionation scores correlated significantly (P <0.05) with triglyceride, HDL-cholesterol, apolipoprotein B100, and LDL-cholesterol/apolipoprotein B100 (r = 0.591, -0.392, 0.454, and -0.411, respectively). For 51 samples, the Quantimetrix method classified 21 with small, 14 with intermediate, and 16 with large LDL. Of the 21 samples classified as small by Quantimetrix, 20 (95%) were classified as small (n = 18) or intermediate (n = 2) by NDGGE. All of the 16 specimens classified as large by Quantimetrix were either large (n = 14) or intermediate (n = 2) by NDGGE. LDL score was inversely correlated (r = -0.674; P <0.0001) with LDL particle size determined by NMR spectroscopy.
Conclusions: A quantitative method for the assessment of LDL particle size phenotype was developed using the Quantimetrix Lipoprint LDL System. The method can be performed in less than 3 h in batch mode and is suitable for routine use in clinical laboratories.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Despite this established role of small, dense LDL particles in cardiovascular risk assessment, current methods available for LDL subfractionation are technically demanding and not applicable in a routine clinical laboratory. These include density gradient ultracentrifugation (11)(12)(13)(14), nondenaturing gradient gel electrophoresis (NDGGE) (15)(16), and nuclear magnetic resonance (NMR) spectroscopy (17)(18)(19), procedures that have one or more of the disadvantages of being labor-intensive, technically demanding, expensive, and having a lengthy turnaround time.
We evaluated a polyacrylamide gel tube electrophoresis method (LipoprintTM LDL System; Quantimetrix, Redondo Beach, CA), which requires 25 µL of plasma or serum and has a total analysis time of <3 h. Although the equipment and all necessary reagents are commercially available, only minimal recommendations regarding densitometry are provided without reliable instructions to evaluate the scan or account for lot-to-lot inconsistencies. For these reasons, we developed a densitometric scanning system to generate raw data and a program to analyze data and calculate a numerical score for the electrophoretic scan pattern of the size-fractionated LDL. The data analysis system is effective in minimizing intraindividual interpretation bias, recalibration after lot changes, and particularly in establishing subfractionation scores corresponding to LDL size phenotypes.
We applied this modified method to plasma samples from 51 patients and compared the results with a commercially available NDGGE method (Zaxis) and an NMR method (LipoMed). Correlation between LDL subfractionation (LDLSF) scores and other measured lipid markers were also assessed.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Procedures
Quantimetrix Lipoprint LDL System.
High-resolution 3%
polyacrylamide gel tubes were used for electrophoresis. LDL
subfractionation was performed as described in the Lipoprint LDL System
product insert. Briefly, 25 µL of sample was mixed with 200 µL of
liquid loading gel. The loading gel contained Sudan Black B dye to
stain the lipoproteins. The resulting mixture was added to the top of
precast 3% polyacrylamide gel tubes. After photopolymerization at room
temperature for 30 min, samples were electrophoresed for 1 h (3
mA/gel tube). The electrophoresis was followed by "resting" the
tubes in the dark for 1 h before performing the densitometry. This
ambient incubation was added to increase the uniformity of the bands
before scanning.
Densitometry was performed at 610 nm using a Helena EDC system (Helena Laboratories), which was improved by addition of a 12-tube holding device engineered and built by the Mayo Clinic Engineering Department. The holding device was constructed from 1/4-inch black Lucite
® CP acrylic (ICI acrylics, Inc.) engineered to fit securely in the Helena EDC densitometer. Twelve slots (7.6 x 0.6 cm) were cut through the acrylic sheet to hold the gel tubes (7.7 x 0.7 cm) in place. The holding device markedly improved the consistency of gel tube scanning by securing the tube during the scanning process. Raw data generated by the densitometer and the electrophoretic scan pattern of the size-fractionated LDL were analyzed using a computer program of our design to calculate a LDL subfractionation score and assign a phenotype (A, B, or intermediate).
Computer-assisted data analysis of densitometric scans.
Raw
data from the densitometer were imported into a Microsoft Excel
spreadsheet, and a scan was generated. Using the public domain NIH
Image program (developed at the US National Institutes of Health and
available on the Internet at
http://rsb.info.nih.gov/nih-image), we wrote a macro that
defined electrophoretic mobility (Rf)
values across the scan, divided the scan at designated
Rf values, and calculated the area
under the curve (AUC) for each fraction. A complete description of the
Rf designation may be found in the
Results under "Development of a LDLSF Scoring System".
The computer then summed the AUCs in a weighted manner
[(LDL1AUC · 1) +
(LDL2AUC · 2) + (LDL3AUC
· 4) + (LDL4AUC · 8) +
(LDL5AUC · 16) +... ] and divided the sum
by the total unweighted LDLAUC (i.e.,
LDL1AUC + LDL2AUC +
LDL3AUC + LDL4AUC + ...
) to generate a LDL score. This scoring system produced scores ranging
from as low as 1 to as high as 100 in patient samples with a great
preponderance of small, dense (fast migrating) LDL in the samples we
have tested to date. Division by the total unweighted
LDLAUC was performed to normalize the effect of
various sample LDL concentrations on the LDLSF score.
LDL scores were determined in plasma samples from 89 patients. On the basis of LDL migration rates on the scan, LDL phenotypes A (predominantly large, buoyant LDL), B (predominantly small, dense LDL) (1), or intermediate were assigned to score ranges as follows: LDLSF score <5.5 = phenotype A, 5.5–8.5 = intermediate, >8.5 = phenotype B. A complete description of the rationale for this phenotyping assignment is given in the Results under "Development of a LDLSF Scoring System".
Method comparison.
A method comparison study was performed
using 51 patient plasma samples to confirm proper assignment of LDLSF
score to phenotype. Samples were chosen to give a wide range of LDL
scores based on the Quantimetrix method and were sent to the South Bend
Medical Foundation (South Bend, IN) for LDL subfractionation using
NDGGE (LFS Lipogel System®; Zaxis) and to
LipoMed (Raleigh, NC) for LDL subfractionation using the LipoMed NMR
Lipoprofile®. Personnel performing subfraction
analysis at South Bend and LipoMed were blinded to the Quantimetrix
results until analysis was complete.
Zaxis LFS Lipogel System.
This system uses a bifocal gradient
polyacrylamide gel. The bottom half of the gel consists of an 8–27%
gradient, whereas the top half is a 0.5–5% gradient. Briefly, sample
was mixed with loading dye, and 4 µL was added to the
preelectrophoresed gel. Electrophoresis was performed for 15–20 h, and
gels were stained overnight with Sudan Black B at room temperature.
After destaining, the gels were fixed in 0.35 mol/L acetic acid.
The gels were scanned by densitometry, and LDL phenotype was assigned
by comparison of LDL migration distance in the sample to that of plasma
calibrators of known LDL phenotype. Plasma samples that were used as
calibrators for the Zaxis method were sent to Berkeley HeartLab
(Alameda, CA) for determination of LDL size and phenotyping by NDGGE.
The samples were classified by the Berkley HeartLab method as having
predominantly LDL particle sizes corresponding to the Berkley HeartLab
established cutpoints for phenotypes A and B, respectively. These
plasma samples were run as calibrators along with patient samples on
each gel and served as the basis by which patient samples were
phenotyped using the Zaxis method.
LipoMed NMR Lipoprofile.
This system measures lipoproteins
directly, without fractionation, in <5 min (20). An
automated liquid sample handler transfers plasma to the NMR instrument.
Each lipoprotein particle with a given diameter range broadcasts a
unique NMR signal, producing a NMR spectrum for the sample. Computer
algorithms are used to deconvolute the spectrum, providing a direct
measure of the concentrations of 15 different subclasses of VLDL, LDL,
and HDL. An overall LDL size measurement is determined and used to
categorize the patients phenotype as phenotype A (20.6–22.0 nm),
intermediate (20.4–20.5 nm), or phenotype B (19.0–20.3 nm). This
instrumentation is not currently available for purchase, and samples
must be sent to LipoMed for analysis.
Lipid measurements.
Total cholesterol and triglycerides were
assayed on a Hitachi 717 chemistry analyzer (Roche Diagnostics) using
Bayer/Technicon Omnipack and Bayer/Technicon RA/opeRA enzymatic
reagents (Bayer/Technicon), respectively. HDL-cholesterol was
determined using selective precipitation followed by the
enzymatic/colorimetric method for measuring cholesterol described
above. A solution of dextran sulfate and Ca2+
(1:10 solution of 40 g/L dextran sulfate in 111 g/L
CaCl2) was used to precipitate non-HDL
lipoproteins (chylomicrons, VLDL, and LDL). HDL-cholesterol in the
supernatant was measured using Technicon reagents on a Hitachi 717.
Calculated values for LDL-cholesterol were derived using the Friedewald
equation (21) as follows:
 |
Apolipoprotein AI and B100 were measured by automated immunoturbidimetric assays on a Hitachi 912 chemistry analyzer (Roche Diagnostics) using reagents from Diasorin Inc. Lipoprotein(a) was also measured by immunoturbidimetry using Diasorin reagents on a Hitachi 912.
analytical performance
Precision.
Intraassay precision was determined using plasma
samples from a phenotype A patient (mean LDL score = 3.2) and a
phenotype B patient (mean LDL score = 13.0) analyzed in 10 gel
tubes. Interassay precision was determined using plasma from 19
patients assayed on 3 separate days over a 1-week period.
Linearity.
Linearity of the method was determined by
performing mixing studies of a patient serum sample containing
predominantly large, buoyant LDL particles (phenotype A) with a sample
containing a high proportion of small, dense LDL (phenotype B).
Stability.
Fresh serum samples from 20 patients were
analyzed immediately after collection and after storage at 4, -20, and
-70 °C for 14 days. LDLSF scores were 2.9–15.8. Sample stability
at -20 and -70 °C was also investigated using six EDTA plasma
samples from phenotype B patients (LDLSF scores = 9.6–54).
Statistics.
Spearman rank correlation between LDLSF
score and other lipid markers as well as between LDLSF score and LDL
size (nm; LipoMed) was determined using StatView (Abacus Concepts,
Inc.).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
shows 12 gel tubes after completion of the electrophoresis.
Electrophoretic migration was from the top of the tube (cathode) to the
bottom (anode). Visible bands correspond to lipoprotein particles
stained with Sudan Black B dye before electrophoresis. Visual
inspection of the LDL region showed variable patterns, e.g., tube 7 has
predominantly fast-migrating LDL (small LDL), whereas the LDL in tube 8
is predominantly slow-migrating LDL (large LDL). A densitometric scan
of tube 8 is shown in Fig. 1B
.
|
To document that visual interpretation of scans is highly individual-dependent, 34 densitometric scans from randomly selected patient samples were obtained and reviewed by five separate individuals who independently assigned a phenotype to each scan. Individuals were asked to assign a phenotype (A, B, or intermediate) to each scan. No restrictions were placed as to how each individual ascribed phenotypes. All reviewing individuals had been involved with initial evaluation of the Quantimetrix method, understood the electrophoretic process, and were familiar with LDL size phenotyping. However, complete agreement was achieved for only 11 (32%) of the 34 samples. Of the 11 samples with complete agreement, 10 were phenotype A and 1 was phenotype B. For nine (26%) of the samples, opposite phenotypes (A vs B) were assigned for the same sample. This survey supported our hypothesis that a standardized method for phenotype assignment would be helpful in avoiding subjective differences in result interpretation.
development of a ldlsf scoring system
Using an image processing program (NIH Image), we wrote a
calibration macro to set separation lines at defined points across the
densitometric scan. Vertical separation lines were defined using 30
different samples with type A phenotypes, based on qualitative scan
appraisal. Five vertical lines were then placed at defined points for
each of the 30 samples, as depicted in Fig. 2
. The lines were placed as follows: Line 1 was placed at the
beginning of the VLDL peak; lines 2 and 5 were identified as reference
points at the apex of the VLDL (Rf = 0)
and HDL (Rf = 1) peaks, respectively.
Line 3 was placed at the valley between VLDL and intermediate-density
lipoprotein (IDL). Line 4 was placed at a point representing 10% of
the height from the valley between IDL and LDL to the top of the LDL
peak to reduce the likelihood of including small IDL particles in the
LDL scoring region. Relative to the two reference points, all 30
Rf values for each of the three lines (1, 3,
and 4) were averaged. The Rf means were then
used to identify lipoprotein windows: the window between lines 1 and 3
was used for VLDL, the window between lines 3 and 4 was used for IDL,
and the window from line 4 to line 5 was used for LDL.
|
The average Rf values were then
entered into a second NIH Image sample analysis macro, which was
designed to calculate a LDLSF score for patient samples and controls.
The program dropped vertical lines at appropriate
Rf locations after the technologist
identified the apex of the VLDL and HDL peaks. A total of 18 lines were
dropped, creating 17 fractions. The first fraction contained VLDL. The
IDL window was divided into 3 equidistant fractions, and the LDL window
into 12. HDL was contained in the final fraction. The program then
calculated the AUC for each fraction in the LDL window and performed
logarithmic transformation of the AUC data, as described in
Materials and Methods, to produce a LDLSF score. Fig. 3
depicts the scoring process for phenotype A, intermediate, and
B samples (LDLSF score = 2.5, 7.1, and 14.2, respectively). The
computer system allowed a high degree of consistency and a means to
alleviate interpretative bias.
|
Scored densitometric scans from 89 patient plasma samples were used to establish assignment of LDL particle size phenotype to LDLSF score ranges. This was accomplished as follows: If the combined area of the first 3 LDL fractions was greater than two-thirds of the total area, phenotype A was assigned. If the sum of the AUCs of fractions 1–3 was equal to or greater than one-half but less than two-thirds of the total area, an intermediate classification was made. If the sum of the AUCs of fractions 1–3 was less than one-half of the total area, phenotype B was assigned. This allowed classification of 55 patients with pattern A, 19 as intermediate, and 15 with pattern B. The LDLSF scores (mean ± SD) for the phenotype A, intermediate, and phenotype B groups were 3.7 ± 1.1, 6.6 ± 0.8, and 15.0 ± 7.2, respectively. On the basis of these results, the following phenotyping criteria were established: LDLSF score <5.5, phenotype A; score 5.5–8.5, intermediate; score >8.5, phenotype B. The SD for phenotype B overlaps with the intermediate range because of very high LDL scores associated with some phenotype B specimens.
assay imprecision
Patient plasma samples were used to assess the imprecision of the
assay. Intraassay imprecision (n = 10) for two plasma samples gave
CVs of 8.1% and 8.4% at LDL subfractionation scores of 3.2 and 13.0,
respectively. For 19 patients with LDL scores ranging from 2.9 to 16.5
assayed on 3 days over a 1-week period, the mean interassay CV was
13%. Improved precision was demonstrated by analysis of each
patient sample in duplicate. The intraassay CVs for patient samples
analyzed 10 times in duplicate were 4.6% and 4.3% at LDLSF scores of
3.4 and 13.3, respectively.
lot-to-lot variability
To determine whether variability between lots of reagents (lots
included gel tubes, loading gel with stain, and electrophoresis buffer)
existed, 12 patient samples with a range of LDLSF scores were assayed
in duplicate by each of two lots, and the results were compared. With
the first lot of reagents, LDLSF scores of 3.6–9.9 (mean = 5.8)
were observed. With the second lot, scores were 2.3–5.0 (mean =
3.1). Thus, from one lot to the next, the absolute mean difference in
LDL score was 46.8%. Significant lot-to-lot variability was noted when
other lots were acquired and tested. This problem was addressed by
acquiring 30 phenotype A samples, performing analysis, and applying the
NIH Image calibration macro described above to drop lines at defined
points across the scans. This established mean
Rf values for the new lot. The new
Rf values were then incorporated into
the sample analysis macro, producing good correlation between reagent
lots. When the updated macro was used to score the same 12 samples, the
LDLSF scores were 3.8–11.1 (mean = 5.7) and the mean lot-to-lot
difference was 1.04% (Table 1
).
|
linearity
Serum samples containing predominantly large, buoyant LDL
particles (phenotype A) were mixed with samples containing a high
proportion of small, dense LDL (phenotype B). The LDL scores progressed
in a uniform manner, becoming larger with increasing proportions of
phenotype B sample. Plotting the LDLSF score (x-axis) vs the
percentage of phenotype B (y-axis) yielded the following
linear equation: y = 16.8x - 42.1,
with a correlation coefficient (r) of 0.9663 (Fig. 4
).
|
sample type and stability
EDTA plasma and serum samples were simultaneously collected from
33 volunteers. LDLSF scores ranged from 2.2 to 53.9 (EDTA) and from 2.4
to 50.5 (serum). Mean ± SD values for EDTA plasma and serum
samples were 7.8 ± 9.9 and 8.4 ± 10.7, respectively. The
regression equation was: serum = 1.039(plasma) + 0.252;
r = 0.9559.
There was no discernible difference in LDL score between fresh controls (<1 week old) and those stored at 4 °C for up to 2 months. Specimen stability was assessed in 20 patient serum samples analyzed fresh and after storage for 14 days. The mean recoveries (± SD) of LDLSF scores at 2 weeks were 105% ± 16%, 69% ± 10%, and 74% ± 8.9% for samples stored at 4, -20, and -70 °C, respectively. Larger decreases (as a percentage) in the LDL score after freezing were noted for phenotype B samples. The mean recoveries of the LDLSF score in six EDTA plasma samples (phenotype B) frozen for 2 months at -20 and -70 °C were 95% ± 29% and 102% ± 15%, respectively.
interferences
Two plasma samples were supplemented with ascorbic acid to
a final concentration of 50 mg/L and with hemoglobin to a final
concentration of 2000 mg/L. Neither substance caused significant
interference with LDLSF score determination (data not shown).
correlation with other lipid measures
The correlations between the LDL score and other lipid markers are
shown in Table 2
. The correlation was significant (P 
0.05) for
triglycerides, HDL-cholesterol, apolipoprotein B100, and
LDL-cholesterol/apolipoprotein B100. Statistically significant
correlation was not observed between LDLSF score and total cholesterol,
LDL-cholesterol, or apolipoprotein AI.
|
method comparisons
To verify appropriate assignment of LDL size phenotype to LDLSF
score range, the Quantimetrix method was compared with a NDGGE method
(Zaxis) and a NMR method (LipoMed). EDTA plasma samples from 51
patients, selected to give a mixture of phenotypes (A, B, and
intermediate, based on LDL score) were sent for comparative analysis.
Table 3
compares phenotypes assigned by each method. Quantitative
analysis of LDL size was not performed with the Zaxis NDGGE method;
therefore, a correlation plot could not be generated. A significant
inverse correlation was observed (r = -0.674;
P <0.0001) between the LDLSF score and LDL size (nm)
determined by the LipoMed method. The inverse relationship was expected
because of the nature of the LDLSF scoring system, which produces
higher LDL scores as particle size decreases.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The assay could be performed reliably on EDTA plasma or refrigerated serum, but LDL subfraction scores were not stable in frozen serum samples. Freezing serum at -20 or -70 °C decreased LDLSF scores, particularly in samples with high scores (phenotype B). For long-term storage, serum samples should not be used. The LDLSF score was demonstrated to be stable in EDTA plasma for at least 2 months when stored at -70 °C. LDL subfractionation can be performed reliably in serum samples stored refrigerated for up to 2 weeks.
The precision of the Quantimetrix method (between-run CV = 13%) was improved (mean CV = 4.5%) if samples were assayed in duplicate and reported as the average LDL subfraction score. For this reason, we run all of our routine LDL subclass analyses in duplicate and report the average score. This may be particularly important if the method is used in therapeutic studies to monitor treatment effects because duplicate measures decrease analytical variability, allowing more discrete determination of changes resulting from dietary or pharmacologic therapy.
In 51 samples used for method comparison, the LDLSF score was found to
correlate significantly with triglyceride (r = 0.59;
P <0.0001) and HDL-cholesterol (r =
-0.392; P = 0.006) concentrations as well as with
other lipid markers (Table 2
). This observation is consistent with
previous reports including that of the Quebec Cardiovascular Study
(7), which reported correlation coefficients of -0.46
(P <0.001) and 0.39 (P <0.001) between LDL size
(nm) and triglyceride and HDL-cholesterol concentrations, respectively.
In the Quebec Cardiovascular Study (7), the presence of
small, dense LDL was independently associated with increased risk of
developing ischemic heart disease even after adjustment for plasma
triglyceride and HDL-cholesterol concentrations.
Phenotypes (A, B, or intermediate) were ascribed to LDL score ranges, and the method was compared with NDGGE and NMR methods for LDL subfractionation. The correlation of results between methods was satisfactory, indicating the viability of the Lipoprint LDL System for assessing LDL particle size phenotype. However, 5 of 21 samples classified as phenotype B by the Quantimetrix method were classified as type A by NMR. All five of these samples were classified as type B or intermediate by NDGGE. Additionally, one sample classified as type B by the Quantimetrix method was classified as type A by NDGGE. NMR classified this sample as intermediate. Adjustment of LDLSF score range did not improve the method comparison results. This comparison confirmed that appropriate assignment of LDLSF score ranges to LDL particle size phenotypes was achieved. These data indicate that although there is general agreement among the three methods, standardization of LDLSF methods is needed, and that the methods should not be used interchangeably.
In conclusion, the Quantimetrix Lipoprint LDL System combined with the automated scoring methodology developed in our laboratory enables quantitative LDL subfractionation in <3 h in batch mode; thus, it is suitable for routine use in clinical laboratories. The Quantimetrix Corporation has recently incorporated an adaptation of our scoring system into their method, is working on production of a completely automated LDL subfraction analysis system, and is seeking Food and Drug Administration approval for their product. Quantimetrix is additionally taking steps to minimize the observed lot-to-lot variability of their product. Studies are currently underway to assess the clinical utility of this method as a CAD risk marker, in predicting overt CAD, and for monitoring therapy.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: CAD, coronary artery disease; NDGGE, nondenaturing gradient gel electrophoresis; NMR, nuclear magnetic resonance; LDLSF, LDL subfractionation; AUC, area under the curve; and IDL, intermediate-density lipoprotein.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  H. Y. AYDOGAN, S. ISBIR, O. KURNAZ, U. GORMUS, and T. ISBIR Associations of Lipoprotein Lipase S447X and Apolipoprotein E Genotypes with Low-density Lipoprotein Subfractions in Turkish Patients with Coronary Artery Disease In Vivo, January 1, 2009; 23(1): 155 - 161. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Satoh, A. Shimatsu, K. Kotani, N. Sakane, K. Yamada, T. Suganami, H. Kuzuya, and Y. Ogawa Purified Eicosapentaenoic Acid Reduces Small Dense LDL, Remnant Lipoprotein Particles, and C-Reactive Protein in Metabolic Syndrome Diabetes Care, January 1, 2007; 30(1): 144 - 146. [Full Text] [PDF]
|
|
|
|
 |
|
 M. Okazaki, S. Usui, A. Fukui, I. Kubota, and H. Tomoike Component Analysis of HPLC Profiles of Unique Lipoprotein Subclass Cholesterols for Detection of Coronary Artery Disease Clin. Chem., November 1, 2006; 52(11): 2049 - 2053. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Ensign, N. Hill, and C. B. Heward Disparate LDL Phenotypic Classification among 4 Different Methods Assessing LDL Particle Characteristics Clin. Chem., September 1, 2006; 52(9): 1722 - 1727. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. J. Kullo, K. Ding, E. Boerwinkle, S. T. Turner, and M. de Andrade Quantitative trait loci influencing low density lipoprotein particle size in African Americans J. Lipid Res., July 1, 2006; 47(7): 1457 - 1462. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Gazi, E. S. Lourida, T. Filippatos, V. Tsimihodimos, M. Elisaf, and A. D. Tselepis Lipoprotein-Associated Phospholipase A2 Activity Is a Marker of Small, Dense LDL Particles in Human Plasma Clin. Chem., December 1, 2005; 51(12): 2264 - 2273. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Okazaki, S. Usui, M. Ishigami, N. Sakai, T. Nakamura, Y. Matsuzawa, and S. Yamashita Identification of Unique Lipoprotein Subclasses for Visceral Obesity by Component Analysis of Cholesterol Profile in High-Performance Liquid Chromatography Arterioscler. Thromb. Vasc. Biol., March 1, 2005; 25(3): 578 - 584. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. J. Kullo and C. M. Ballantyne Conditional Risk Factors for Atherosclerosis Mayo Clin. Proc., February 1, 2005; 80(2): 219 - 230. [Abstract] [PDF]
|
|
|
|
 |
|
 J. S. Volek, M. J. Sharman, A. L. Gomez, C. DiPasquale, M. Roti, A. Pumerantz, and W. J. Kraemer Comparison of a Very Low-Carbohydrate and Low-Fat Diet on Fasting Lipids, LDL Subclasses, Insulin Resistance, and Postprandial Lipemic Responses in Overweight Women J. Am. Coll. Nutr., April 1, 2004; 23(2): 177 - 184. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Sharman, A. L. Gomez, W. J. Kraemer, and J. S. Volek Very Low-Carbohydrate and Low-Fat Diets Affect Fasting Lipids and Postprandial Lipemia Differently in Overweight Men J. Nutr., April 1, 2004; 134(4): 880 - 885. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. G. Davies, J. M. Graham, and B. A. Griffin Rapid Separation of LDL Subclasses by Iodixanol Gradient Ultracentrifugation Clin. Chem., November 1, 2003; 49(11): 1865 - 1872. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. S. Volek, M. J. Sharman, A. L. Gomez, T. P. Scheett, and W. J. Kraemer An Isoenergetic Very Low Carbohydrate Diet Improves Serum HDL Cholesterol and Triacylglycerol Concentrations, the Total Cholesterol to HDL Cholesterol Ratio and Postprandial Lipemic Responses Compared with a Low Fat Diet in Normal Weight, Normolipidemic Women J. Nutr., September 1, 2003; 133(9): 2756 - 2761. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Sharman, W. J. Kraemer, D. M. Love, N. G. Avery, A. L. Gomez, T. P. Scheett, and J. S. Volek A Ketogenic Diet Favorably Affects Serum Biomarkers for Cardiovascular Disease in Normal-Weight Men J. Nutr., July 1, 2002; 132(7): 1879 - 1885. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
