Methods for LDL-C Measurement

Several methods have been used to measure LDL-C. The first, a reference laboratory procedure,
involves ultracentrifugation to separate LDL from other lipoproteins, followed by analysis as
described above to measure cholesterol. This method is not extensively discussed here. A much
more common second method uses the Friedewald formula to calculate LDL-C. Finally, more
recently developed homogeneous methods for measuring LDL-C are now available.

Friedewald Calculation.

LDL-C can be determined by using the Friedewald formula, originally described by Friedewald,
Levy, and Fredrickson ( Friedewald, 1972 ). Generally, in fasting plasma samples LDL contains
the cholesterol that is not present in HDL or VLDL. Thus LDL-C can be determined by the
following equation in which concentrations are expressed in mmol/L, and the term [Plasma
TG]/2.175 is used to represent VLDL-C.

(17-1) 

The term [Plasma TG]/5 is used when concentrations are expressed in mg/dL. In this method, the
plasma TC, TG, and HDL-C concentrations are determined as described. Because most plasma
triglyceride is carried by VLDL, VLDL-C concentration is estimated from the ratio of
triglyceride to cholesterol in VLDL:

(17-2) 

It has been reported that the factor [Plasma TG]/2.825 gives a more nearly accurate estimate of
VLDL-C ( DeLong, 1986 ). This is equivalent to Plasma TG/6.5, when concentrations are
expressed in mg/dL. However, the factor that gives the best estimate of VLDL-C, and therefore
the best estimate of LDL-C, varies among populations and depending upon the triglyceride
method used; on balance, the NCEP Working Group on Lipoprotein Measurement preferred the
unmodified Friedewald equation ( NCEP Working Group on Lipoprotein Measurement, 1995 ).

The Friedewald formula has significant limitations ( Sniderman, 2003). These limitations arise
largely from two assumptions upon which the method is based. First, the calculation assumes
that essentially all plasma triglycerides are carried in VLDL. Second, the method assumes the
triglyceride/cholesterol ratio of VLDL is invariant. Neither assumption is entirely true; as a
result, this method is unsuitable for nonfasting samples that contain chylomicrons or samples that
contain β-VLDL. As compared to VLDL, the ratio of triglycerides to cholesterol in chylomicrons
is much higher. Thus, when CM is present the use of the factor TG/2.175 to account for non-
HDL, non-LDL cholesterol can overestimate the amount of cholesterol in VLDL, leading to
underestimation of LDL-C. Similarly the ratio of triglyceride to cholesterol in β-VLDL is much
lower than in VLDL, and the use of the factor TG/2.175 in the presence of β-VLDL can
underestimate VLDL-C and thus overestimate LDL-C. A patient with type 3
hyperlipoproteinemia can be misclassified as having an elevated LDL-C. It is important to
distinguish the two conditions because their treatments differ.
Even in chylomicron-free samples the ratio of VLDL-C to triglyceride changes as triglyceride
levels increase, and can lead to errors in estimates of VLDL-C. Because VLDL generally only
carries about 25% of the TC in plasma, the resulting errors in LDL-C are usually less than 5-10
mg/dl (0.130-0.260) mmol/L). However, the calculation is not suitable for samples with high
triglyceride concentrations. Errors in LDL-C become noticeable at triglyceride levels > 2.26
mmol/L (200 mg/dL), and are felt to become unacceptably large at triglyceride levels > 4.52
mmol/L (400 mg/dL). The accuracy of LDL-C calculation also suffers at low LDL-C levels,
indicating that calculated LDL-C values may not provide the best assessment of cardiac risk in
patients who are already receiving cholesterol-lowering therapy ( Sniderman, 2003 ). Provided
that its limitations are appreciated, the Friedewald equation has broad utility, both as a screening
tool and for following patients.

Another issue to consider is non-LDL lipoproteins. Generally, LDL contributes most of the
cholesterol to the measurement, and IDL and Lp(a) contribute only a few mg/dL each. However,
these lipoproteins can contribute significantly to cholesterol measurements in some
hyperlipidemia patients. Since Lp(a) levels are not lowered by a number of treatments that
effectively lower LDL levels, Lp(a) measurements can, in some circumstances, reveal why a
patient does not respond well to LDL-lowering therapy.

Direct LDL-C Measurement.

To a large extent the ability to calculate LDL-C has eliminated the need for direct measurements.
However, homogeneous direct LDL-C methods are useful when triglycerides are elevated
because they are not subject to interference by triglycerides even at relatively high
concentrations (=600 mg/dL) ( Bachorik, 2000 ). Direct assays have been adapted for use on a
variety of analyzers and have been extensively reviewed elsewhere ( Nauck, 2002 ; Miller,
2002). While these methods differ significantly, in general they use a combination of two
‘reagents.’ The first reagent usually selectively removes non-LDL lipoproteins (and/or stabilizes
or inhibits LDL from reacting with enzymes) and the second reagent releases cholesterol from
LDL so that it can be measured enzymatically.

In one method (Equal Diagnostics, Exton, PA; Genzyme Diagnostics, Cambridge MA), the first
reagent uses a detergent polymer mixture to disrupt non-LDL lipoproteins, releasing their
cholesterol. The cholesterol is then de-esterified, and acted upon by cholesterol oxidase to
generate hydrogen peroxide, which further reacts to form a colorless compound. The second
reagent contains a detergent that releases cholesterol from LDL. After de-esterification, the LDL-
C proceeds through a similar set of reactions, except that the final step generates a colored
compound. The intensity of the color is proportional to the concentration of LDL-C. Another
method (Roche Diagnostics, Indianapolis, IN) is based on selective micellary solubilization of
LDL by a nonionic detergent, as well as the interaction of a sugar compound with HDL, VLDL
and chylomicrons to inhibit the participation in the measurement assay ( Sugiuchi, 1998). A third
method exploits the fact that reactivity of cholesterol in the different lipoproteins is affected by
the hydrophile:lipophile balance (HLB) of solubilizing detergents. In this method non-LDL-C is
reacted with cholesterol esterase and cholesterol oxidase at conditions under which LDL-C is
inhibited and the resulting peroxide eliminated by catalase. A second reagent then alters the HLB
of the detergent, creating conditions under which LDL-C reacts. This reagent also contains azide
that inhibits catalase and allows the colorimetric detection of peroxide (see Reaction 17-3)
(Polymedco Inc., Cortlandt Manor, NY; Reference Diagnostics, Bedford, MA). In a fourth
method, the first reagent contains amphoteric surfactants that protect LDL, and allow the
elimination of non-LDL-C and resulting peroxide, as above. The second reagent contains non-
ionic surfactants that displace the protecting surfactants and allow measurement of LDL-C
(Sigma Diagnostics, St Louis, MO).

While these methods exhibit good precision, they produce discrepancies as compared to the
reference ultracentrifugation procedures in a number of circumstances, including presence of
abnormal lipoproteins. When the triglyceride level is < 400 mg/dL, homogeneous methods
perform no better than the Friedewald calculation for classifying patients into treatment groups (
Miller, 2002). However, unlike the calculation, homogeneous assays can provide clinically
useful results when triglycerides > 400 mg/dL. Another potential benefit is the convenience of
measuring LDL-C in nonfasting individuals though some have recommended against this
practice ( Miller, 2002). ATP III recommendations do not favor replacing calculated LDL-C with
direct LDL-C since the constituents of the calculation have to be measured in any case. Running
direct LDL-C would only add to the expense.