CLIN.CHEM.30/11, 1797-1800 (1984)
Five Methods for Determining Low-Density LipoproteinCholesterol
Compared
Pierre N. Demacker, Anneke G. Hljmans, Bernard J. Brenninkmeljer,
We evaluated three precipitationmethods for determination
of low-density lipoprotein cholesterol in serum and an indirect
method involving the Fnedewald formula (Clin Chem 18:
499-502, 1972)by comparison
withresultsby ultracentrifu-
gation. The results of all methods for 83 sera, including 59
hyperlipidemic
type hA, IIB, and IV sera agreed very well, at
leastfor concentrationsof serumtriglyceridesbelow8 mmoll
L. The accuracy of the Friedewald formula was confirmedin
285 other sera, including66 sera with triglyceridescontent
between 4.52 and 8.0 mmol/L.For type Ill sera, the precipitation methodsproducedsimilarvalues to those obtainedwith
the Friedewald formula, all being much higher than the
ultracentrifugationvalues. Density-gradientultracentnfugation showed that the very-low-density lipoprotein remnants in
type Ill sera almost completelycoprecipitated with the lowdensity lipoproteins.The precipitationmethodsare not only
accuratebut also very precise (CV <5%); they can therefore
be used
in clinical laboratories to measure atherogenic
low-
density lipoproteinsplus the remnants of very-low-density
lipoproteins.
However,
when serum triglycerides
and high-
density lipoprotein cholesterol also are determined, the
Fnedewald
formula is a reliable alternative.
AddItIonal Koyphrases: Frisdewald formula
lipoprotein precipitation
type Ill hyperlipoproteinemia
atherogenesis
glycerides
very-low-density lipoprotein remnants
hyperlipidemia
ultracentrifugation
compared
.
The contribution
of serum cholesterol to the risk for
atherosclerosis
is determined by its distribution
among the
various lipoprotein fractions. A relatively large amount of
cholesterol in the low-density lipoprotem
(LDL)’ fraction is
atherogemc, whereas HDL-cholesterol
appears to be protective (1). For distinguishing
survivors of coronary heart
disease from healthy persons, the LDL-chol/HDL-chol
ratio
is reportedly threefold more sensitive than the measurement of total serum cholesterol (2). Thus, for selection of
high-risk patients, simple but reliable methods
for determining lipoprotein-associated
cholesterol, especially LDLchol, should be available.
Ultracentrifugation
for LDL-chol is time consuming
and
expensive and therefore not suitable for large-scale use.
Recently, three different procedures
for specific precipitation
and determination
of LDL-chol have become available
(3Department of Medicine, Division of General Internal Medicine,
and Division of Clinical Chemistry, St. Radboud Hospital, University of Numegen, Geert Grooteplein Zuid 8,6500 HE Nijmegen, The
Netherlands.
‘Nonstandard
abbreviations: VLDL, very-low-density
lipoproteins (d <1.006; Sf >20); VLDL-remnants, products of VLDL
catabolism
accumulated in type ifi hyperlipoproteinemia
(d
<1.019; S >15, especially the Sf 15-100); LDL, low-density lipoproteins (1.006 <d <1.063; 0 <S1<20); HDL, high-density lipoproteins
(1.063 <d <1.21); chol, cholesterol; TG, triglycerides; HLP, hyperli-
poproteinemia.
Received May 14, 1984;
accepted
August
14, 1984.
Ad P. Jansen, and Albert van ‘t Laar
6). At least one of these methods is considered suitable for
analyzing serum from patients with hyperlipoproteinemia
(HLP) type ifi (3) as well as from patients with type N or V
HLP (4). Therefore, we compared these methods with a
combined ultracentrifugationlprecipitation
method (7,8) for
various sera from normo- and hyperlipemic
subjects. In
addition, we considered the usefulness of the indirect method of Friedewald et al. (9), which is based on the determination of serum cholesterol, triglycerides,
and HDL-chol.
Materials and Methods
Sera
Fresh serum samples from overnight-fasted
normal persons and from patients with various types of HLP were
stored at 4#{176}C
for no longer than four days, unless stated
otherwise. Patients were classified into the different phenotypes according to the criteria of Fredrickson et al. (10) with
cutoff limits for serum cholesterol, triglycerides,
and LDLchol of 7.3, 2.0, and 5.2 mmolJL, respectively (11). HLP was
classified as type ifi when high concentrations of serum
lipids were combined with a VLDL-chol/seruin
TG ratio
>0.69 and the absence of apoprotein E-3 and E-4 in the
isoelectric focusing pattern of the VLDL apoproteins
(11).
Because it was our intention to propose a clear criterion at
which the precipitation methods and the Friedewald
formula are equally accurate or inaccurate,
we arbitrarily
classified sera as type N or V HLP when LDL-chol was not
increased and serum TG were less or more than 8 mmol/L,
respectively.
We judged this adaptation to be necessary
because of the disagreement of the current electrophoresis
and refrigeration
tests for detecting chylomicrons
in serum
of different phenotypes (12). In our opinion, the interpretation of these tests is rather subjective, especially in the
analysis of type N and V sara.
Procedures
Comparison
method. The method we used for reference in
this study was a combined ultracentrifugationJprecipitation
procedure.
VLDL-chol
was determined
directly in the
VLDL-fraction isolated by ultracentrifugation
(7) and HDLchol was determined after precipitation
of the VLDL and
LDL in whole serum by polyethylene glycol 6000 (8). LDLchol was calculated by subtraction.
Precipitation
procedure
A (3, 4). The reagents
were obtained as a test-kit (LDL-cholesterin,
cat. no. 14992; Merck
A.G., D-6100, Darmstadt
1, F.R.G.). The procedure was
performed according to the manufacturer’s
directions: add
100 .tL of serum to 1000 zL of sodium citrate (64 mmol/L,
pH 5.12) containing
50000 U of heparin per litre. After
vortex-mixing,
incubate
the tubes for 10 mm at room
temperature,
centrifuge for 15 mm at 2800 x g at room
temperature
to remove LDL, and, using a Pasteur pipette,
aspirate the supernate, which contains the VLDL and HDL.
Measure the cholesterol in the supernate within 1 h.
Precipitation
procedure
B (5). This precipitation
reagent
was also obtained as a test-kit (LDL-cholesterol,
cat. no.
CLINICALCHEMISTRY,Vol. 30, No. 11, 1984
1797
726290;
Boehringer,
Mannheim,
F.R.G.). To 200 ML of
between the concentration of serum TG and VLDL-chol is
add 100 ML of the precipitation reagent containing,
dependent on age, sex, ethnic, and sociodemographic
variaper litre, 3.0 g of polyvinylsulfate and 170 g of the accelerations of the subjects studied (18).
tor polyethyleneglycol
methyl ether. After vortex-mixing
We determined
the accuracy of the Friedewald formula
the tubes’ contents and incubating for 15 mm at room
with 285 serum samples from normal individuals
and from
temperature, centrifuge for 10 mm at 2300 x g. Remove the
patients with various forms of HLP who were visiting our
supernate
and determine
its cholesterol content as described
out-patient clinic. Patients with type I, ifi, or V HLP were
above.
not included, so that this group we studied had serum TG
Precipitation
procedure
C (6). The precipitation reagent in
between 0.4 and 8.0 mmol/L. These subjects were divided
this procedure
(LDL-cholesterol,
cat. no. 61532; Bio-Merinto two groups, with serum TG concentrations greater or
less than 4.52 mmol/L (400 mg/100 mL). The ratios for
ieux, 69260 Charbonni#{232}resles Bains, France) contained
unspecified amphipathic polymers in imidazole buffer (25
VLDL-chollseruni
TG in these groups were 0.41 ± 0.10
mmol!L, pH 6.10). To 1000 MLof precipitation reagent (at 2(mean and SD; n = 219) and 0.47 ± 0.12 (n = 66),
8#{176}C)
add 50 MLof serum; vortex-mix and incubate at 2-8 #{176}Crespectively. The average ratio in the whole group was 0.42
± 0.11, which is in good agreement with Friedewald’s value
for 30 mm, then centrifuge for 5 mm at 2800 x g. In contrast
to the instructions
of the manufacturer,
we determined
of 0.45. In the following section we used the 0.42 value to
cholesterol in the supernate instead of redissolving
the LDLcalculate LDL-chol concentrations; these results (4.51 ±
precipitate in a small volume of reagent 2 of the test-kit, for
1.75 mmol/L) agreed excellently with those obtained by
consistency with the other methods.
ultracentritItgationlprecipitation
(4.49 ± 1.66 mmol/L). The
linear regression equation for this was LDL (calculated) =
Density-gradient
ultracentrifugation.
To determine
the
flotation/density
characteristics
of the lipoproteins
precip1.03 (±0.03) LDL (reference) - 0.12 (±0.13); SE was 0.38, r
was 0.98, and the mean difference between LDL obtained
itated in procedure A, we used the density-gradient ultracentrifugation
procedure described elsewhere (13). Dissolve
with both methods was -0.01 ± 0.39 mmolIL (p = 0.80, n =
285).
the precipitates obtained from 3 mL of either a fresh
normolipidemic serum or a type III serum (chol 9.3 mmol/L;
Analytical
recoveries of cholesterol
in the ultracentrifugation and precipitation
method. The sum of cholesterol in the
TG 6.5 mmoLfL) in 3 mL of sodium citrate buffer (64
VLDL and the LDL + HDL fractions was 103.7 ± 2.6%
mmol/L, pH 7.7). Increase the density of the clear solution to
(mean ± SD, n = 28) of the value in total serum. The sum of
1.10 kgfL by dissolving
0.14 g of KBr per 3 mL, then transfer
2.7 mL of this to a 10-mL ECI B60 polycarbonate
ultracencholesterol recovered in the supernates containing the
VLDL + HDL and in the precipitated
LDL dissolved with
trifuge tube (14.5 x 96 mm, the same length as the tubes
0.1 moltL NaOH was 97.9 ± 2.0%, 99.0 ± 4.6%, and 101.6 ±
used in ref. 13). Overlayer this sample with, in order, 2 mL
5.1%, in precipitation
methods A, B, and C, respectively (n
of 1.065 kg/L, 2 mL of 1.020 kg/L, and 2.3 mL of 1.006 kg/L
=
28). These results indicate that the ultracentrifugation!
solutions
and ultracentrifuge
for 83 nun at 40 000 rpm
precipitation
method is accurate and that none of the
(170 000 x gay) and 17.5 h at 37 000 rpm in the SW41
precipitation
reagents interferes with the cholesterol deterswinging-bucket rotor at 20 #{176}C
to float VLDL particles with
mination.
respective S. values of >100 and 15-100 to the top of the
Comparison
of the five LDL methods
in normal
and
tube. After the first run, aspirate the lipoproteins in the top
hyperlipemic
sera. We applied the various methods for
0.7 mL, then refill the tube with 0.7 mL of d 1.006 solution.
After the second run, collect the fraction with Sf 15-100 by
determining
LDL-chol in normal sara (n = 32) and in sera of
patients with type HA (n = 21), type II B (n = 9), type ifi (n
aspirating the top 2 mL. If the fractions have been precol=
9), type N (n = 21), and type V (n = 5). The results
ored with Coomassie Brilliant Blue R (14), a clear LDL band
obtained for the sara of patients
with type ifi and V HLP
is visible about halfway up the tube; aspirate this band also.
were analyzed separately. In the nine sara of patients with
Analytical
methods.
Cholesterol was determined
manualtype ifi HLP, the LDL-chol concentrations
obtained by the
ly with an enzymatic test-kit (CHOD-PAP, cat. no 290319;
Boehringer test kit no. 8 in ref. 15). Incubate 2 mL of
four methods under study were significantly
greater than
and did not correlate significantly with those obtained by
reagent with 20 ML of serum or standard, 100 ML of
ultracentrifugationlprecipitation
(Table 1).
supernate A, 50 ML of supernate
B, or 200 tL of supernate C
Surprisingly, the LDL-chol values obtained by the precipaccording
to the instructions.
Correct the results for diluitation method A, B, and C did not differ significantly from
tion. For example, analysis of 20 MLof standard and 100 ML
the values calculated by the F’rmedewald formula: respective
of supernate A, combined with an 11-fold sample dilution
correlation coefficients were 0.92,0.98, and 0.94 (S = 0.57,
during the precipitation step, requires a correction factor of
(2100/2020) x (1/5) x 11. To calculate the LDL-chol values,
0.37, and 0.51, respectively).
we subtract the cholesterol in the supernate from the total
serum cholesterol.
Table 1. LDL-Chol Values in Sera of Nine Patients
Triglycerides
were determined
by semi-automated
colorwith Type Ill Hyperlipoprotelnemla as Determined
imetry (16). All concentrations
are expressed
in mmolJL.
by the Five Different Methods
Statistical
analyses. We analyzed results by Student’s
Ca
Mean (SD)
-y,x
paired or unpaired t-test. Results obtained by different
Serum
cholesterol,
mmol/L
8.6
(1.9)
methods were correlated
by using Pearson’s correlation test.
Serum triglycerides, mmol/L
4.0 (2.1)
The standard
errors in the intercept, slope, and estimate
VLDL-chol/serumTG
1.04 (0.31)
(S,) were also calculated (17).
LDL, mmol/L
3.74 (1.13)
Ultracentrifugation
Results
Friedewald formula
5.89 (1.60)
Method A
5.69 (1.37)
Evaluation of the accuracy of the Friedewald formula. The
0.57
Method B
6.08 (1.66)
0.37
accuracy of the Friedewald
formula
(LDL-chol = total chol
Method C
5.73 (1.39)
0.51
- HDL-chol
0.45 x serum TG) is of course influenced by
method
A,
B,
or
C;
x
Fnedewald
formula.
the accuracy of the methods used for the determination of
bSignificantly different (p <0.001) from results by the other four methods.
cholesterol, TG, and HDL-chol. Furthermore, the relation
serum
-
=
1798
CLINICAL CHEMISTRY, Vol.30, No.11,1984
=
LDL-chol values obtained by the four methods
under
study in type V sara occasionally disagreed by more than
±1 mmol/L. Use of these methods in type V sara is therefore
not recommended.
In the 83 sara with TG concentrations
less than 8.0 mmol/
L (mean and SD: 2.47 and 1.68 mmol/L), the correlation
between the results obtained by the four methods under
study and our reference method was excellent (Table 2). The
accuracy of all three precipitation
methods was very satisfactory. Surprisingly,
the values found with the Friedewald
formula correlated best and gave the lowest S
(0.38 vs
0.52-0.56).
Precision of the precipitation
methods and stability of the
sera stored at -20 #{176}C
or 4#{176}C.
We divided into 1-mL aliquots
fresh serum from a patient with type Ill HLP (chol 10.3
mmol/L, TG 4.9 mmol/L), from a patient with type N HLP
(chol 6.5 mmol/L, TG 3.6 mmol/L), and from four normolipidemic persons. Portions were analyzed when fresh and after
storage at 4#{176}C
or at -20 #{176}C
for 1, 6, 10, 13, and 21 days.
The apparent LDL-chol values in HLP type ifi serum
stored at -20 #{176}C,
as measured with reagents A and C,
increased by about 14%, but when these samples were
stored at 4#{176}C,
the values gradually decreased by as much as
50%. With reagent B we obtained widely variable results
because of incomplete sedimentation
of the precipitated
lipoproteins.
For the other five sara, the LDL-chol values obtained with
reagents B and C at days 1, 6, 10, 13, and 21 at either
temperature did not differ significantly from the values
obtained in the fresh samples. CVs less than 5.1% were
obtained with both precipitation procedures (Table 3). Procedure A tended (0.05 <p <0.1) to give 4-5% lower values
after storage at -20 #{176}C
for 10, 13, or 21 days. The LDL-chol
values for the samples stored at 4#{176}C
and measured with
reagent A decreased
as much as 18%, which explains the
higher CV obtained under these conditions (Table 3).
Analysis
of LDL precipitates by density-gradient
ultracentrifugation.
The presumed LDL precipitates
obtained
with
reagent A from a serum with normal lipid values and a type
Ill serum were analyzed by density-gradient
ultracentrifugation as described above. In the normal serum 83% of the
lipoproteiii-associated
cholesterol was found in the 0 <Sf
<15 fraction and had the flotation characteristics
of LDL;
the remainder
was isolated in the fractions with S1 >100
(2%) and 15 <Sf <100 (15%). In the type ifi serum these
percentages were 30,23, and 47%, respectively. Thus, in the
type ifi serum not less than 70% of the precipitated lipoprotein-associated
cholesterol was non-LDL.
Discussion
The three recently introduced precipitation methods have
been claimed to specifically precipitate
LDL without coprecipitating
the VLDL and HDL. Our results confirm this.
LDL-chol values obtained by the precipitation methods and
by ultracentrifugation
correlated veiy well. However, till
Tabie 3. Precision (CV, %) of the LDLPrecipitation Methods and of Total Cholesterol
Measurement in Sera at Different Storage
Conditions
4G(
Mean
Range
Mean
Rang.
Method A
3.7
1.9-6.6
7.0
3.8-10.8
Method B
3.2
MethodC
2.9
1.7-5.0
1.9-4.4
2.8
3.8
1.5- 4.8
2.6- 5.1
Total cholesterol
1.1
0.8-1.3
1.2
0.8-
1.6
Five sera divided into I -mL aliquots were analyzed freshly or after storage at
-20#{176}C
or4 ‘C for1,6, 10,13,and21daysbythe three precipitation methods.
comparison, the CV for the
See also text.
For
total cholesterol
measurement
is also given.
now the precipitation behavior of VLDL remnants accumulated in serum of patients with type ifi HLP, but also
present in fasting sara of normolipemics
and hyperlipemic
non-type ifi subjects, remained
in question.
The VLDL
remnants have physicochemical
characteristics
intermediate between VLDL and LDL, being primarily
1.006 <d
<1.019. In type III HLP, large quantities of these remnants
are present in the VLDL fraction. Earlier (19) we studied
with the same ultracentrifugation
method 15 patients with
type ifi HLP, whose serum lipids and LDL-chol values were
similar to those of the patients in the present study. We
found that the mean concentration
of LDL1-chol (d 1.0061.019) in non-type ifi patients was 0.38 to 0.81 mmol of
cholesterol per litre, and was related to the amounts of
intermediate
-pre-$
lipoproteins
(VLDL remnants)
on
agarose gel electrophoresis
(19). The mean LDLj-chol concentration in type ifi patients was 1.47 mmol/L. Thus LDLchol corrected for remnant-associated
cholesterol in the
present study should be approximately
3.74 - 1.47 = 2.27
mmolJL, considerably less than the LDL-chol concentration
found by precipitation (5.80 mmoIIL). This indicates that
VLDL remnants are precipitated even when their density is
<1.006 kgfL. This is confirmed by density-gradient
ultracentrifugation: in a normal serum and a type ifi serum, 17
and 70% of the precipitated lipoprotein-associated
cholesterol had flotation properties characteristic for VLDL remnants. Binding of heparin or other polyanions to the LDL
particles does not explain these results, because we found
that binding of heparin does not influence the density
distribution of LDL.
Taken together, the precipitation methods evaluated here
measure the same apparent LDL-chol values as ultracentrifugation but also precipitate the VLDL remnants. All
precipitation methods were equally effective, at least in nontype ifi sera with TG concentrations less than 8 mmol/L,
and all were equally inaccurate in the analysis of type Ill
sara.
Surprisingly,
the LDL-chol values calculated
by the
Friedewald formula were also accurate, except for sara from
Table 2. Results for LDL-Chol In 83 Sera (TG <8.0 mmol/L)5 by the Four Methods (y) Compared with
Those by the Ultracentrifugation Method (x)
LDL-chol, mmol/L5
Method
Friedewald
MethodA
MethodB
MethodC
a
Unear regression equation
y
1.02(0.02)x-O.1O(0.13)#{176}
y = 0.97(0.03)x+0.25(0.17)
y
1.03(0.04)x-0.14(0.19)
y 0.98(0.03)x+0.08(0.16)
=
=
=
r
S)
0.98
0.96
0.96
0.96
0.38
0.52
0.56
0.54
Mean
5.00 (1.83)#{176}
5.10 (1.78)
5.04 (1.90)
5.00 (1.81)
x-y
0.01
-0.09
-0.03
0.00
(0.38)
(0.52)
(0.56)
(0.53)
p
0.87
0.13
0.62
0.95
Sera of patients with type Ill HLP excluded. Mean LDL-chol by ultracentrifugationwas 5.00 (SD 1.76) mmolJL. No. in parentheses indicate SD.
CLINICAL CHEMISTRY, Vol. 30, No. 11, 1984
1799
with type Ill and V HLP. As mentioned earlier, the
mean value we obtained for the VLDL-chol/serum
TG ratio
in 285 sera was close to that found by Friedewald
et al. (9).
In contrast to their original report, we also found the
Friedewald
formula to be reliable in sara with TG concentrations between 4.5 and 8 mmoIJL. This disagreement
strengthens the assumption
by Friedewald
et al. (9) that the
larger percentage
of errors seen in this group of type IV
patients might be due in part to laboratory
errors in the
ultracentrifugation
calculation
of LDL-chol rather than
greater inaccuracy of the estimation procedure. Our findings enhance
considerably
the number of sara in which
ultracentrifugation
can be omitted for the determination
of
LDL-chol, provided that the rare type ifi HLP has been
patients
excluded.
Both the precipitation
methods and the Friedewald formula produced similar values for type ifi sara, but these
were considerably higher than the values obtained by
ultracentrifiigationiprecipitation.
This is reasonable because the Friedewald
formula
is based on the assumption
that the choLPl’G ratio of the VLDL fraction is normal;
contamination
of this fraction with VLDL remnants invalidates the formulas. The good agreement between the apparent LDL-chol values produced by the precipitation methods
or by the Friedewald calculation again confirms that in type
ifi sara most, if not all, of the VLDL remnants
also
precipitate.
However, the co-precipitation
of VLDL remnants is in disagreement
with the data of Wieland and
Seidel (3), who reported preliminary findings that type ifi
VLDL is not precipitated.
The precipitation
methods appeared to be precise (except
for HLP type ifi serum), and the stability of serum stored at
or at -20 #{176}C
was good. Using the same storage conditions as for HDL-chol
should permit reliable results for
LDL-chol by precipitation.
We recommend storing the sara
at -20 #{176}C
(maximum one month), preferably, or at 4#{176}C
(one
week). Precipitation
method A is the most sensitive to
changes in the storage conditions, perhaps because during
storage the pH of serum increases (8), which may influence
the precipitation of LDL by reagent A (3).
From our results, we conclude that the introduction
of
methods for specific precipitation
of LDL seems superfluous,
because we obtained the same or even better results by
using the Friedewald formula. However, this is only a good
alternative
for those laboratories that have accurate
and
precise methods for the determination
of both serum TG and
HDL-chol.
Determination
of HDL-chol in sara with abovenormal TG concentrations
may be an especial source of error
(20).
If LDL-chol is precipitated
to differentiate
between LDLchol and HDL-chol, we recommend using the Friedewald
formula. The interference of VLDL remnants
in the precipitation method for LDL is, in our opinion, rather an advantage than a disadvantage
because
both classes of lipoproteins are atherogemc in contrast
to ordinary
VLDL (21).
Thus, by using the precipitation methods or the Friedewald
formula, one can more nearly accurately quantify the concentrations of atherogenic lipoproteins present, even in type
ifi HLP. In this respect these methods are superior to
ultracentrifugation.
We thank Dea van Sommeren-Zondag
and Pieternel van Heijst
for excellent technical assistance, Erny Meij-van Ke8teren and
1800 CLINICAL CHEMISTRY, Vol.30, No. 11, 1984
Ineke ten Have for the preparation of the manuscript, and the
NetherlandsHeart Foundation for partial financial support.
References
1. Kannel WB, Castelli WP, Gordon T. Cholesterol in the prediction of atherosclerotic disease. Ann Intern Med 90, 85-91 (1979).
2. Avogaro F, Cazzolato G, Bon GB, Belussi G. High density
lipoproteins. Ric Clin Lab 12, 87-95 (1982).
3. Wieland H, Seidel D. A simple specific method for precipitation
of low density lipoproteins. J Lipid Res 24, 904-909 (1983).
4. Kirch PT, Wurzberg U. The direct determination of low density
lipoprotein cholesterol by precipitation. Clin Chem 29, 1227 (1983).
Abstract
5. Marev J, Draeger
B, Wehxneyer G, et a!. Method for the
quantitation of serum low density lipoprotein cholesterol. Ibid. p
1173. Abstract.
6. Troyez G, Alcindor LG. Patent to be published.
7. Demacker PNM, Vos-Janssen HE, Jansen AP, van ‘t Laar A.
Evaluation of the dual-precipitation method by comparison with the
ultracentrifugation method for measurement of lipoproteins in
serum. Clin Chem 23, 1238-1244 (1977).
8. Demacker PNM, Humans AG, Vos-Janssen HE, et al. A study of
the use of polyethylene glycol in estimating cholesterol in highdensity lipoprotein. Ibid. 26, 1775-1779 (1980).
9. Friedewald MiT, Levy RI, Fredrickson
DS. Estimation of the
concentration
of low density lipoprotein cholesterol in plasma,
without use of the preparative ultracentrifuge. Ibid. 18, 499-502
(1972).
10. Fredrickson DS, Levy RI, Lees RS. Fat transport in lipoproteins. N Engi J Med 276, 34-44, 94-103, 148-156, 2 15-226, 273-
281 (1967).
11. Stuyt PMJ, Demacker PNM, van ‘t Laar A. Serum lipids,
lipoproteins and apolipoprotein E phenotypes in relatives of patienta with type ifi hyperlipoproteinemia. Eur J Clin Invest 14,
219-226 (1984).
12. McNeely S, Seatter K, Yubaniak J, Kashyap ML. The 16-hour
standing test and lipoprotein electrophoresis for detection of chylomicrons in plasma. Clin Chem 27, 731-732 (1981).
13. Redgrave TR, Carlson LA. Changes in plasma very low density
and low density lipoprotein content, composition, and size after a
fatty meal in normo- and hypertriglyceridemic man. JLipidRes 20,
217-229 (1979).
14. Demacker PN, van Sommeren-Zondag DF, Stalenhoef AF, et
al. Ultracentrifugation
in swinging
bucket and fixed-angle rotors
evaluated for isolation and determination of high-density lipoprotein subfractions HDL2 and HDL3. Clin Chem 29, 656-663 (1983).
15. Demacker PNM, Boerma GJM, Baadenhuijsen H, et al. Evaluation of accuracy of 20 different test kits for the enzymic determination of cholesterol. Ibid. pp 1916-1922.
16. Demacker PNM, van Oppenraay JB, Baadenhuijsen H, Jansen
AP. An unproved
semi-automated
method for the colorimetric
determination of triglycerides in serum. Clin Chim Acta 64, 45-50
(1975).
17. Davis RB, Thompson JE, Psi-due HL. Characteristics
cal parameters
used
to interpret
least-squares
results.
of statistiClin Chem
24, 611-620 (1978).
18. Wilson PW, Abbott RD. Garrison RJ, Castelli WP. Estimation
of very-low density lipoprotein cholesterol from data on triglyceride
concentration in plasma. Ibid. 27, 2008-2010 (1981).
19. Demacker PNM, Vos-Janssen HE, Stuyt PMJ, et a!. Application and comparison of different criteria used for the diagnosis of
type ifi hyperlipoproteinemia.
Ibid. 24, 1445-1451 (1978).
20. Demacker PNM, Vos-Janssen HE, Hijmans AGM, et a). Measurement of high-density lipoprotein cholesterol in serum: Comparison of six isolation methods combined with enzymic cholesterol
analysis. Ibid. 26, 1775-1786 (1980).
21. Havel RJ. Approach to the patient with hyperlipidemia. Med
Clin North Am 66, 319-333 (1982).