Letter to the Editor

Ann Lab Med 2024; 44(4): 363-366

Published online July 1, 2024

Copyright © Korean Society for Laboratory Medicine.

Direct LDL Cholesterol Assay vs. Estimated Equations in Patients With Hypertriglyceridemia or Low LDL Cholesterol Levels

Jennifer Rodríguez-Domínguez, RD.J.* , Álvaro Piedra-Aguilera, PA.A.* , María Martínez-Bujidos, MB.M. , Susana Malumbres-Serrano, MS.S. , Cristian Morales-Indiano, MI.C. , and Carla Fernández-Prendes , FP.C.

Laboratory Medicine Department, Hospital Universitari Germans Trias i Pujol, Badalona, Spain

Correspondence to: Carla Fernández-Prendes, FP.C.
Laboratory Medicine Department, Hospital Universitari Germans Trias i Pujol, Carretera de Canyet S/N, 08916 Barcelona, Badalona, Spain

* These authors contributed equally to this study as co-first authors.

Received: September 29, 2023; Revised: December 12, 2023; Accepted: January 4, 2024

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Dear Editor,

Low-density lipoprotein–cholesterol (LDL-C) levels strongly correlate with cardiovascular disease risk. Guidelines recommend targeting low LDL-C levels to reduce this risk. Therefore, precise measurement is essential [1].

The LDL-C reference method requires ultracentrifugation by trained personnel and large serum samples [2, 3]. Equations, such as Friedewald, Martin–Hopkins, or Sampson, are often used instead to calculate LDL-C levels [4, 5]. Equations have limitations, particularly for the calculation LDL-C≤70 mg/dL or triglycerides levels (TG) ≥400 mg/dL. Furthermore, bias and imprecision from the different measurements used in the calculation can adversely affect LDL-C calculation accuracy [6].

LDL-C clinical decision limits are based on the Friedewald equation. Despite the additional cost, the direct LDL-C method (dLDL-C) is faster than ultracentrifugation and more accurate than equations. dLDL-C results may vary either among manufacturers or depending on the reagent-calibrator-instrument combination. Therefore, it is advisable to compare the dLDL-C with the reference method [7].

We assessed and compared the dLDL-C and equations for calculating LDL-C against the reference method to identify quick and accurate alternatives. This was a prospective study developed in Spain, from March 2022 to March 2023.

In total, 212 serum samples stored at 4ºC were analyzed and categorized into two groups (IRB; PI-21-036): group 1 (TG≥400 mg/dL, positive lipemic index, N=113) and group 2 (TG≤200 mg/dL, LDL-C≤70 mg/dL, N=99). Group 1 was subdivided into 1a (TG=400–500 mg/dL, N=62) and 1b (TG=500–900 mg/dL, N=51). Group 2 was subdivided into 2a (LDL-C=10–40 mg/dL, N=49) and 2b (LDL-C=40–70 mg/dL, N=50).

dLDL-C level was measured using the enzymatic selective protection method [8] with an AU5800 analyzer (Beckman Coulter, Brea, CA, USA). The dLDL-C calibrator value is traceable to the reference method. LDL-C was calculated using the Friedewald, Martin–Hopkins (180-cell strata), and Sampson equations. Very-low-density lipoprotein–cholesterol (VLDL-C) was isolated from 1.5 mL of serum utilizing sequential density gradient ultracentrifugation at 100,000×g for 18 hrs with an F50L fixed-angle rotor (Thermo Scientific, Basingstoke, UK) using KBr for density adjustment (1,006 g/mL). TC, HDL-C, TG, and VLDL-C contents were measured using the AU5800 analyzer, and LDL-C level was calculated as follows: LDL-C=TC−HDL-C−VLDL-C [2, 3, 4].

Statistical analysis was conducted using MedCalc v19.6 (MedCalc Software, Ostend, Belgium). Means were compared using t-test for paired samples to compare the different methods against the reference method.

Bias between methods was calculated as follows:


where Cn and Cx represent the LDL-C determined using the reference and alternative methods, respectively. Bias was compared with the reference change value (RCV), which was calculated for LDL-C considering a unilateral Z statistic with 95% confidence (Z=1.65), as follows:


where CVa and Cvi are the analytical CV and within-subject biological variation according to the European Federation of Clinical Chemistry and Laboratory Medicine Biological Variation Database, respectively [9]. The RCV was calculated to be 20.0% based on CVa=2.18 and CVi=8.3. Bland–Altman difference plots were used to compare the reference and alternative methods in groups 1 and 2.

Table 1 reveals significant differences in LDL-C between the reference and alternative methods, except for dLDL-C in group 1. However, the bias did not exceed the RCV for dLDL-C and Martin–Hopkins. In group 2, LDL-C differed significantly from those determined using the reference method for all alternative methods; however, the bias did not exceed the RCV for dLDL-C.

Table 1 . Comparison of methods for direct LDL-C assay and estimated equations for LDL-C with a reference method

Group 1 (TG≥400 mg/dL)Median (IQR)148 (116–177)149 (130–170)85 (59–112)121 (101–142)101 (81–122)
Bias (%)0.5−42.4−18.5−31.7
Group 1a (TG=400–500 mg/dL)Median (IQR)151 (115–180)149 (122–174)93 (61–124)126 (101–151)106 (78–132)
Bias (%)−1.5−38.4−16.6−30.1
Group 1b (TG=500–900 mg/dL)Median (IQR)148 (124–174)148 (132–165)77 (55–98)114 (102–134)98 (82–112)
Bias (%)−0.2−48.3−22.7−34.0
Group 2 (LDL-C≤70 mg/dL)Median (IQR)58 (49–62)61 (54–68)41 (34–46)45 (37–50)43 (36–48)
Bias (%)5.3−29.6−22.9−25.4
Group 2a (LDL-C=10–40 mg/dL)Median (IQR)49 (44–54)55 (51–63)34 (29–38)36 (30–41)36 (29–40)
Bias (%)10.9−31.8−25.8−26.7
Group 2b (LDL-C=40–70 mg/dL)Median (IQR)62 (58–66)65 (61–69)46 (43–49)49 (46–53)48 (45–51)
Bias (%)5.8−25.1−20.2−21.9

*Reference method.

Abbreviations: IQR, interquartile range; TG, triglycerides; LDL-C, low-density-lipoprotein cholesterol; dLDL-C, direct LDL-C assay.

In groups 1a and 1b, LDL-C measured using the reference method differed significantly from the values determined using the alternative methods, except for dLDL-C. The bias never exceeded the RCV for dLDL-C in groups 1a and 1b nor for Martin–Hopkins in group 1a. In groups 2a and 2b, all methods differ significantly from the reference method; however, the bias did not exceed the RCV for dLDL-C (Table 1).

Bland–Altman plots demonstrated minimal bias of dLDL-C vs. the reference method in groups 1 and 2 (Fig. 1).

Figure 1. Bland–Altman plots for the alternative methods and the ultracentrifugation (reference) method. (A) dLDL-C, (B) Friedewald, (C) Martin–Hopkins, and (D) Sampson methods applied in group 1. (E) dLDL-C, (F) Friedewald, (G) Martin–Hopkins, and (H) Sampson methods applied in group 2.
Abbreviations: TG, triglycerides; LDL-C, low-density lipoprotein–cholesterol; dLDL-C, direct low-density lipoprotein–cholesterol.

Notably, for TG≥400 mg/dL, dLDL-C was the superior method. For TG=500–900 mg/dL, results from all alternative methods differed from ultracentrifugation results, except for dLDL-C. For TG=400–500 mg/dL, both dLDL-C and Martin–Hopkins methods were suitable, with dLDL-C being the most accurate. dLDL-C is applicable for LDL-C<70 mg/dL.

For LDL-C<70 mg/dL or TG≥400 mg/dL, Friedewald, Martin–Hopkins, and Sampson tend to underestimate LDL-C. Martin–Hopkins and Sampson equations reportedly have improved accuracy over Friedewald with TG≥400 mg/dL or LDL-C<70 mg/dL [5, 10]. However, these studies did not compare these methods with ultracentrifugation, but with dLDL-C.

In conclusion, the study showed that both equations and the Friedewald method continue to underestimate LDL-C. The findings support that dLDL-C is an excellent choice for TG≥400 mg/dL or LDL-C<70 mg/dL. These results are specific for the Beckman Coulter dLDL-C; differences among manufacturers have been reported [7]. Similarly, TC, HDL-C, and TG assays used to calculate LDL-C are not perfectly standardized among manufacturers.

Martínez-Bujidos M, Morales-Indiano C and Fernández-Prendes C contributed to the conception and design of the study; Rodríguez-Domínguez J, Piedra-Aguilera A, and Fernández-Prendes C interpreted the results; Piedra-Aguilera A performed the statistical analysis; Rodríguez-Domínguez J, Piedra-Aguilera A, Malumbres-Serrano S and Fernández-Prendes C drafted the manuscript; and Fernández-Prendes C supervised the study. All authors read and approved the final manuscript.

  1. Wilson PWF, Polonsky TS, Miedema MD, Khera A, Kosinski AS, Kuvin JT. Systematic review for the 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Blood Cholesterol: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Circulation 2019;139:e1144-61.
  2. Havel RJ, Eder HA, Bragdon JH. The distribution and chemical composition of ultracentrifugally separated lipoproteins in human serum. J Clin Invest 1955;34:1345-53.
    Pubmed KoreaMed CrossRef
  3. Ordovas JM. Fast Ultracentrifugation Methods for the Separation of Plasma Lipoproteins. Methods Mold Biol 1998;110:93-103.
    Pubmed CrossRef
  4. Frost PH, Havel RJ. Rationale for use of non-high-density lipoprotein cholesterol rather than low-density lipoprotein cholesterol as a tool for lipoprotein cholesterol screening and assessment of risk and therapy. Am J Cardiol 1998;81:26B-31B.
    Pubmed CrossRef
  5. Ertürk Zararsız G, Bolat S, Cephe A, Kochan N, Yerlitaş SI, Doğan HO, Zararsız G. Validation of Friedewald, Martin-Hopkins and Sampson low-density lipoprotein cholesterol equations. PLoS One 2022;17:e0263860.
    Pubmed KoreaMed CrossRef
  6. Martin SS, Blaha MJ, Elshazly MB, Brinton EA, Toth PP, McEvoy JW, et al. Friedewald-estimated versus directly measured low-density lipoprotein cholesterol and treatment implications. J Am Coll Cardiol 2013;62:732-9.
    Pubmed CrossRef
  7. Nauck M, Warnick GR, Rifai N. Methods for measurement of LDL-cholesterol: a critical assessment of direct measurement by homogeneous assays versus calculation. Clin Chem 2002;48:236-54.
    Pubmed CrossRef
  8. Miki Y. A homogeneous assay for the selective measurement of LDL-cholesterol in serum. Enzymatic selective protection method. Clin Lab 1999;45:398-401.
  9. EFLM Biological Variation Database. (Updated on April 2023).
  10. Azimi V, Farnsworth CW, Roper SM. Comparison of the Friedewald equation with Martin and Sampson equations for estimating LDL cholesterol in hypertriglyceridemic adults. Clin Biochem 2022;108:1-4.
    Pubmed CrossRef