| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
Servei de Bioquímica, Hospital de la Santa Creu i Sant Pau, 08025 Barcelona, Spain
2
Servei d’Endocrinologia, Hospital de la Santa Creu i Sant Pau, 08025 Barcelona, Spain
3
Departament de Bioquímica i Biologia Molecular, Universitat Autònoma, 08193 Bellaterra, Barcelona, Spain
a address correspondence to this author at: Departament de Bioquímica, Hospital de la Santa Creu i Sant Pau, Avinguda Sant Antoni Ma Claret 167, 08025 Barcelona, Spain; fax 34-93-2919196,
2038{at}hsp.santpau.es
LDL-cholesterol (LDL-C) is the main marker in the
evaluation and treatment of dyslipidemia (1). LDL-C is
typically estimated after isolation by ultracentrifugation
(2) or by calculation with the Friedewald formula
(3):
 |
where HDL-C is HDL-cholesterol. Beta-quantification, the most frequently used ultracentrifugation-based method to estimate LDL-C, includes all cholesterol associated with LDL, as well as cholesterol associated with lipoprotein(a) [Lp(a)] and intermediate-density lipoprotein (IDL) particles (4). Because beta-quantification is cumbersome and requires sophisticated equipment, most clinical laboratories use the Friedewald formula, which also includes LDL-C, IDL-cholesterol, and Lp(a)-cholesterol in the estimation of LDL-C. Although chylomicronemia and increased concentrations of VLDL and IDL particles are known to interfere with the Friedewald formula, unacceptable bias may also be found with triglyceride (Tg) concentrations <4.6 mmol/L (4)(5). Therefore, direct methods that reliably measure LDL-C would be of great interest in clinical practice.
Since the first introduction of an immunochemically based method (6), other methods based on the selective solubilization of LDL particles have also been reported (7). When compared with beta-quantification, most of these methods show a certain bias (8)(9)(10), which could be caused by unequal reactivity of reagents with the broad range of plasma lipoproteins estimated as LDL by commonly used procedures. Thus, before a direct method can be introduced into clinical practice, the cholesterol fraction(s) measured and its equivalence with beta-quantification must be evaluated. In this study, we assessed LDL-C Plus, a direct, homogeneous method for the measurement of LDL-C, and analyzed which lipoproteins are recognized as LDL.
The LDL-C Plus assay (cat. no. 1985604; Roche Diagnostics) was used
according to manufacturer instructions and calibrated with the
calibrator provided by the manufacturer. The principle of the
method has been described elsewhere (8)(9)(10). In the first
step, a mixture of Mg2+, sulfated
-cyclodextrin, and dextran sulfate was used to reduce the reactivity
of chylomicrons and VLDL-cholesterol (VLDL-C) with the enzymes used in
the final reaction.
Total cholesterol (TC) and Tg concentrations were measured by standard enzymatic methods (CHOD-PAP and GPO-PAP; Roche Diagnostics). HDL-C was measured by a direct method, and Lp(a) concentrations were measured by immunoturbidimetry (both from Roche Diagnostics). All assays were performed in an Hitachi 911 analyzer. LDL-C was determined by the Friedewald formula (LDLF) and by modified beta-quantification (LDLBQ), VLDL was separated at d <1.006 kg/L by ultracentrifugation (18 h; 105 000g; 4 °C), and HDL-C was measured in the infranatant by the direct method (instead of precipitation). During the period of the study (October–December 1999), the inaccuracies of TC, Tg, direct HDL, and LDLBQ measurements were controlled by a multilevel control from the Pacific Biometrics Research Foundation. Maximum inaccuracies were 2.2%, -4.4%, and -4.5% for TC, Tg, and direct HDL-C, respectively, whereas the inaccuracy for LDLBQ was 1.6–7.5%. Lipoprotein fractions were isolated by sequential ultracentrifugation from pooled sera using the following densities: d <0.95 kg/L for chylomicrons, 0.95–1.006 kg/L for VLDL, 1.006–1.019 kg/L for IDL, 1.019–1.050 kg/L for LDL, 1.050–1.100 kg/L for Lp(a), and >1.100 kg/L for HDL and the lipoprotein-deficient serum fraction. The range 1.019–1.050 kg/L was selected for LDL to avoid the inclusion of Lp(a) particles in this fraction. TC content in these fractions was measured as indicated above.
With commercial controls containing two different LDL-C concentrations (Precinorm® L and Precipath® HDL/LDL; Roche Diagnostics), LDL-C Plus showed a run-to-run imprecision of <2.0%, which satisfied criteria established by the National Cholesterol Education Program (5). Recovery and serial dilutions of isolated LDL produced direct inaccuracy estimates of <5.1%. When LDL-C Plus was assessed in samples with chylomicrons and/or increased VLDL (n = 9), a positive difference was found (3.07 ± 1.73 mmol/L for LDL-C vs 2.71 ± 1.58 mmol/L for LDLBQ; P <0.0001, Wilcoxon t-test). However, when one sample with a serum VLDL-C/Tg ratio of 1.05 (in mmol/L) was excluded from analysis, the difference became negative (2.24 ± 0.93 mmol/L for LDL-C Plus vs 2.37 ± 1.00 mmol/L for LDLBQ; P <0.0001). These results suggest a strong dependence of LDL-C Plus values on serum VLDL-C/Tg (i.e., on serum IDL particle content).
Passing-Bablok regression analysis (11) comparing LDL-C Plus
and LDLBQ, assessed in 115 samples (TC,
3.38–10.56 mmol/L; Tg, 0.42–11.60 mmol/L; HDL-C, 0.52–2.32
mmol/L; LDL-CBQ, 0.69–8.28 mmol/L; 10 samples
with >4.6 mmol/L Tg, 5 of which contained chylomicrons), produced the
equation:
 |
The confidence interval of the slope included 1, but that of the y-intercept did not include 0. The regression analysis revealed a significant negative bias and lower concentrations of LDL-C Plus (3.15 ± 1.07 mmol/L) compared with LDLBQ (3.46 ± 0.99 mmol/L; P <0.001). This constant error has been described recently (9)(10) and persisted after the samples were classified according to whether Tg concentrations were lower (n = 64) or higher (n = 51) than 2.30 mmol/L. LDL-C Plus values were not correlated with Tg [Spearman correlation coefficient (rS) = 0.08], VLDL-C (rS = 0.004, obtained by the reference method), or HDL-C (Pearson correlation coefficient, r = 0.12), but a significant positive association was found with the serum VLDL-C/Tg ratio (rS = 0.28; P = 0.002). Again, these results suggest a relationship between LDL-C Plus and serum IDL-C content. In these samples, the Friedewald formula also produced significantly lower values (3.19 ± 1.04 mmol/L; P <0.0001) and a negative bias (LDLF = -0.225 + 1.017 x LDLBQ). In a subset of samples (n = 24) in which Lp(a) was measured, no significant association (rS = 0.02) was found between the Lp(a) logarithm and LDL-C Plus values.
To ascertain whether the negative bias of the LDL-C Plus method was
related to a differential reactivity of the reagents with various serum
lipoproteins, we investigated the amount of cholesterol that was
measured by the LDL-C Plus assay in two pooled sera. The reactivities
of these sera with LDL-C Plus are depicted in Fig. 1
. VLDL-C did not react significantly with LDL-C Plus, IDL-C
showed >60% reactivity in both pools, LDL-C showed the highest
reaction (70–80%), and the Lp(a) fraction with a density of
1.050–1.100 kg/L had a reactivity of 
35%. The reactivity of the
latter fraction could be attributable to the presence of LDL particles
in this density range, but not to the presence of Lp(a) particles
because reactivities in this fraction did not vary between pooled sera
with undetectable (<80 mg/L) and high (729 mg/L) Lp(a)
concentrations. No detectable reactivity was found in the HDL +
lipoprotein-deficient serum fraction. Our results are in accordance
with previous observations (12)(13)(14). A method that uses the
same mixture of Mg2+ ions, -cyclodextrin, and
dextran sulfate as LDL-C Plus to decrease the reactivity of
chylomicron-cholesterol and VLDL-C measures almost 80% of the
cholesterol associated with IDL particles (12). Indeed,
Lackner et al. (13), in dysbetalipoproteinemic patients, and
we, in hyperlipidemic genetically modified mice with very high IDL
content (14), have shown that these particles are recognized
by the HDL-C direct assay from Roche, which produces a positive bias.
Both the Roche HDL and LDL direct assays essentially share the same
mixture as first reagent (12)(15). Thus, both
the previous and the present results demonstrate that this mixture does
not eliminate the reactivity of IDL-C. This failure to eliminate the
reactivity of IDL-C, which interferes in HDL-C measurement, allows the
recognition of most IDL-C as LDL-C with the LDL-C Plus method, as
occurs with the Friedewald equation and beta-quantification.
|
According to these data, LDL-C Plus mainly recognized cholesterol associated with LDL and IDL, but not with Lp(a); the latter being strikingly different from LDLBQ and LDLF. However, the contribution of Lp(a) to TC is <0.2–0.3 mmol/L in subjects with Lp(a) concentrations <300 mg/L (16)(17). Thus, the substantial differences found between LDL-C Plus and LDLBQ are independent of which lipoprotein-cholesterol is recognized by the assay.
To elucidate the role of calibration in the negative bias of LDL-C
Plus, we assessed 68 additional serum samples (TC, 3.42–8.90 mmol/L;
Tg, 0.43–12.5 mmol/L; HDL-C, 0.55–2.51 mmol/L;
LDL-CBQ = 1.77–6.19 mmol/L; 11 samples with
>4.6 mmol/L Tg, of which 6 contained chylomicrons) after adding 0.31
mmol/L (the difference between means observed in the first set of 115
samples) to the initial value of the calibrator and obtained the
equation:
 |
The confidence intervals for the slope and
y-intercept included 1 and 0, respectively. LDL-C Plus and
LDLBQ concentrations were not significantly
different (3.63 ± 1.34 mmol/L for LDL-C Plus and 3.71 ±
1.30 mmol/L for LDLBQ), but in the same samples,
LDLF was significantly lower (3.27 ± 1.47
mmol/L; P <0.001). Of samples with <4.6 mmol/L and 
4.6
mmol/L Tg, 81% and 45%, respectively, had a bias lower than 10%
according to LDL-C Plus, whereas 79% and 18% showed this bias
according to LDLF. Thus, it is likely that the
differences found between LDL-C Plus and LDLBQ
could be attributed to the calibration of the assay. This has also been
suggested by two recent studies (9)(10), but our
data show, for the first time, the reactivity of LDL-C Plus reagents
with the broad range of lipoproteins estimated as LDL by currently used
methods.
In conclusion, we demonstrate that the LDL-C Plus assay measures cholesterol associated with IDL and LDL but not with Lp(a) particles. The significant negative bias observed may be attributed to suboptimal assay calibration. After readjusting the calibration, we found a close relationship between LDL-C Plus and LDLBQ values, and LDL-C Plus became a reliable alternative to beta-quantification and a better approach for LDL-C measurement than the Friedewald formula.
Acknowledgments
We thank Artur Palet and Santiago Juvé from Roche Diagnostics, S.L. (Barcelona, Spain) for kindly providing the reagents.
References
-cyclodextrin sulfate. Clin Chem 1998;44:522-531.-cyclodextrin. Clin Chem 1995;41:717-723.The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  C. Garcia-Hejl, P. Vest, C. Renard, A. Merens-Gonthier, A. Boukhira, and H. Thefenne-Astier Falsely Low LDL Cholesterol Results and Cholestasis. Clin. Chem., November 1, 2006; 52(11): 2125 - 2127. [Full Text] [PDF]
|
|
|
|
 |
|
 E. T. Bairaktari, K. I. Seferiadis, and M. S. Elisaf Evaluation of Methods for the Measurement of Low-Density Lipoprotein Cholesterol Journal of Cardiovascular Pharmacology and Therapeutics, January 1, 2005; 10(1): 45 - 54. [Abstract] [PDF]
|
|
|
|
|
|
S. Usui, H. Kakuuchi, M. Okamoto, Y. Mizukami, and M. Okazaki Differential Reactivity of Two Homogeneous LDL-Cholesterol Methods to LDL and VLDL Subfractions, as Demonstrated by Ultracentrifugation and HPLC Clin. Chem., November 1, 2002; 48(11): 1946 - 1954. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. G. Miller, P. P. Waymack, F. P. Anderson, S. F. Ethridge, and E. C. Jayne Performance of Four Homogeneous Direct Methods for LDL-Cholesterol Clin. Chem., March 1, 2002; 48(3): 489 - 498. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Nauck, G. R. Warnick, and N. Rifai Methods for Measurement of LDL-Cholesterol: A Critical Assessment of Direct Measurement by Homogeneous Assays versus Calculation Clin. Chem., February 1, 2002; 48(2): 236 - 254. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, pool A, which contains high cholesterol and Tg, low HDL-C,
and undetectable Lp(a) concentrations. 
, pool B, which
contains high cholesterol and Lp(a) (>700 mg/L) concentrations. TC,
VLDL-C, TG, and Lp(a) concentrations were higher, whereas HDL-C
was lower in pool A. For details, see text.
ND, not detectable; LPDS,
lipoprotein-deficient serum.