Validation of Low-Density Lipoprotein Cholesterol Equations in Pediatric Population

doi:10.21203/rs.3.rs-1612543/v2

Download PDF

Article

Validation of Low-Density Lipoprotein Cholesterol Equations in Pediatric Population

https://doi.org/10.21203/rs.3.rs-1612543/v2

This work is licensed under a CC BY 4.0 License

Version 2

posted

You are reading this latest preprint version

Purpose

This study aimed to validate different low-density lipoprotein cholesterol (LDL-C) estimating equations in the Turkish population, composed of children and adolescents.

Methods

In this study at Sivas Cumhuriyet University Hospital in Sivas (Turkey), a total of 3,908 children under the age of 18 were included. LDL-C was directly measured by direct homogeneous assays such as Roche, Beckman, and Siemens and was predicted by Friedewald’s, Martin-Hopkins’, extended Martin-Hopkins’ and Sampson’s formulas. Firstly, the concordances between the predictions obtained by the direct measurements and formulas were calculated, and then, these results were evaluated both overall and separately for the LDL-C, triglycerides (TG), and non-high-density lipoprotein cholesterol (non-HDL-C) sublevels. Linear regression analysis was performed, and residual error plots were generated between each prediction and direct measurement method. Coefficient of determination (R²) and mean absolute deviations were also evaluated.

Results

The overall concordance of the Friedewald, Sampson, Martin-Hopkins and the extended Martin-Hopkins formula were 64.6%, 69.9%, 69.4%, and 84.3% for the Roche direct assay, 69.8%, 71.6%, 73.6% and 80.4% for the Siemens direct assay, 66.5%, 68.8%, 68.9% and 82.1% for the Beckman direct assay, respectively. The extended Martin-Hopkins formula had the highest concordance coefficient in both overall and all sublevels of LDL-C, non-HDL-C, and TG. When estimating the LDL-C categories, the highest underestimation degrees were obtained with the Friedewald formula.

Conclusions

Our analysis, conducted in a large pediatric population, showed that the extended Martin-Hopkins equation gives more reliable results in predicting LDL-C compared to other equations.

cardiovascular disease

cholesterol

high-density lipoprotein

lipid metabolism

low-density lipoprotein

triglyceride

Many existing formulas have been proposed to estimate LDL-C levels.
Sampson formula is a novel LDL-C estimating equation and was found to provide accurate results in patients with low LDL-C level and/or hypertriglyceridemia.

Extended Martin-Hopkins equation gives more reliable results in predicting LDL-C compared to other formulas.

Cardiovascular disease (CVD) has been found to be the leading cause of morbidity and mortality in children and adolescents worldwide, as well as adults. Dyslipidemia is a condition related to abnormal lipoprotein metabolism and a common CVD risk factor. It causes atherosclerotic lesions in children aged 2-16, which can lead to an increased risk of cardiovascular disease in their adulthood [1, 2]. It is an important clinical problem in children and adolescents, especially living in western countries, due to the dietary causes and increasing incidence of obesity. According to the NHANES report, obesity affects 18.5 percent of all American youth and 20.6 percent of adolescents [3]. Although CVD does not usually manifest itself until the fourth decade of life, atherosclerosis, which is related to dyslipidemia, is known to begin at earlier stages of childhood. In fact, it has been shown that the early symptoms/signs of atherosclerosis, such as fatty lines, can appear as early as age 2 [2] Hence if the signs of the development of CVD can be diagnosed in childhood, cardiovascular disease can be possibly prevented or at least kept under control in adults by an earlier intervention such as the administration of lowering lipid levels or controlling lipid profiles which have been found to be helpful for primary and secondary prevention of the disease [4–6].

Low-density lipoprotein cholesterol, one of the most atherogenic lipoproteins and one of the critical parameters used for CVD risk assessment [7], has been identified as the key target for cholesterol-lowering therapy to minimize the risk of CVD [8]. Therefore, it is important to accurately measure the LDL-C serum levels to determine the further treatment of children and adolescents. Beta-quantification is regarded to be the gold standard method for LDL-C measurement; however, it has a number of drawbacks that limit its use in many laboratories due to its high cost and the test being time-consuming. For this reason, other direct methods, as well as the equations developed for LDL-C prediction, have been used to predict the LDL-C serum levels. The Friedewald is one of those equations, commonly used for estimating LDL-C serum concentrations [9]. However, it has some limitations in practice. For instance, it overestimates LDL-C levels when triglycerides are very low and underestimates LDL-C levels when TG is higher than 400 mg/dL. Also, it requires eight hours of overnight fasting which can be challenging for children and adolescents.

Due to the limitations of the Friedewald equation, other formulas have been developed such as the Martin-Hopkins, Sampson, etc., in order to predict LDL-C levels with higher accuracy [10–13]. Presently, there has been limited testing of these formulas resulting in a gap in the literature for estimating LDL-C levels in children and adolescents [8, 14–16].

Therefore, in this study, we aimed to investigate the validity of the LDL-C levels predicted by the Friedewald, Sampson, Martin-Hopkins, and extended Martin-Hopkins formulas with the LDL-C levels measured by some direct assays (i.e., Roche, Siemens and Beckman) using a large cohort of Turkish children and adolescents.

Study population

The study was carried out in a retrospective design. The study includes 3,908 children and adolescents under the age of 18 who were referred to the Department of Biochemistry from different units of Sivas Cumhuriyet University Medical Hospital and whose lipid measurements (HDL-C, LDL-C, TG, and TC (total cholesterol)) were analyzed between March 3, 2011 and December 31, 2019. The study was conducted in accordance with the Declaration of Helsinki and adhered to Good Clinical Practice guidelines. The study was approved by the Ethics Committee of Sivas Cumhuriyet University (2022-03/17).

Lipid measurements

Roche Cobas 8000, c-702 and c-501, Siemens Advia 1800, and Beckman Coulter AU5800 systems were used for direct measurement of HDL-C, LDL-C, TG, and TC. In Siemens direct assay, HDL-C levels were measured with Trinder reaction. All other measurements in this and other direct assays are measured with colorimetric enzymatic reaction.

Lipid predictions

To predict LDL-C concentration, we used the Friedewald [9], Sampson [13], Martin-Hopkins and extended Martin-Hopkins formulas [12]. These formulas are given as follows.

Friedewald’s LDL-C prediction formula (LDL-C_F) consists of a linear combination of HDL-C, TG, and TC [9]. In this formula, a fixed TG/VLDL-C ratio is used.

The Martin-Hopkins LDL-C prediction formula (LDL-C_M) is a modification of Friedewald’s formula, where an adjustable factor () is used for the TG/VLDL-C ratio [12]. A 180-cell strata specific median TG/VLDL-C ratio table is used to predict . These median ratios are calculated according to the non-HDL-C and TG sublevels and changes between 3 and 12.

In the extended Martin-Hopkins LDL-C prediction formula (LDL-C_E), the same two-dimensional table format is used. In this case, strata specific median TG/VLDL-C ratios were calculated using Turkish population data.

Sampson’s LDL-C formula (LDL-C_S) is calculated using the following least-squares approach [13]:

Statistical analysis

The consistencies of each LDL-C estimating equation and direct assay were calculated using the overall concordance statistic. This statistic was determined as the ratio of direct LDL-C (LDL-CD) in the same category as predicted LDL-C depends on predicted LDL-C levels (< 110 mg/dL, 110 to 129 mg/dL, and ≥ 130 mg/dL). The concordance statistic was also calculated for non-HDL-C and TG sublevels (TG sublevels: < 75 mg/dL, 75 to 129 mg/dL, and ≥ 130 mg/dL; non-HDL-C sublevels: < 120 mg/dL, 120 to 144 mg/dL, and ≥ 145 mg/dL). To compare the predicted and measured LDL-C levels, ordinary least squares regression analysis models were fitted for each equation and each direct assay. For each combination, residual error plots were generated to display the change of the difference between measured and predicted LDL-C levels according to TG levels. All statistical analyses were performed using R 4.0.4 (www.r-project.org ) statistical software.

Patient characteristics

Table 1 shows the demographic characteristics of all participants. Females (54.7%) and males (45.3%), a total of 3,908 children, were included in the study. The mean age of all participants was 12.10±4.60. The median of TC, TG, HDL-C, and LDL-C measured by direct method levels were 153 mg/dL, 99 mg/dL, 44.3 mg/dL, and 92 mg/dL, respectively. The median TG/TC ratio and nonHDL-C level were calculated as 0.66 and 107 mg/dL, respectively. Lipid profiles of 2356 children (60.3%) were measured with Roche, 893 children (22.9%) with Beckman, and 659 children (16.9%) with Siemens direct assays. All demographic characteristics of the individuals were given separately for these assays as well.

Table 1: Study population characteristics

Characteristic	Overall (N = 3,908)	Roche (N = 2,356)	Beckman (N = 893)	Siemens (N = 659)
Age (years)	13(9-16)	13(9-16)	14(9-16)	15(9-16)
Gender
Female	2138 (54.7)	1363 (57.9)	451 (50.5)	324 (49.2)
Male	1770 (45.3)	993 (42.1)	442 (49.5)	335 (50.8)
Lipid values
TC (mg/dL)	153(131-176)	150(129-171.8)	164(141-187.5)	149(127-174)
TG (mg/dL)	99(72-140)	103(74-143)	93(68-129)	99(70-145)
HDL-C (mg/dL)	44.3(37.5-52)	45(38-52)	46(39-53)	42.1(35.3-50.0)
Non-HDL-C (mg/dL)	107(87-129)	103(85-126)	117(98-139)	106(86-128.7)
TG - TC ratio	.66(.48-.90)	.69(.51-.95)	.57(.43-.78)	.67(.49-.94)
LDL-C_D (mg/dL)	92(74-113)	91(73-112)	104(86-121)	81(64-101)
LDL-C_F (mg/dL)	84.4(66.4-104.8)	80.0(63.4-100.0)	96.4(77.8-116.5)	84.2(64.1-103.1)
LDL-C_S (mg/dL)	86.4(67.8-107.0)	82.1(65.1-102.5)	97.5(79.7-118.2)	86.0(66.1-106.3)
LDL-C_M (mg/dL)	86.4(68.9-106.6)	82.6(65.9-102.7)	96.9(79.7-116.7)	85.7(67.9-106.2)
LDL-C_E (mg/dL)	93.3(76.9-114.1)	91.7(75.1-112.4)	104.6(89.1-121.0)	83.2(67.2-101.2)

Values are expressed as N(%), mean±SD or median(1^st – 3^rd quartiles). TC: total cholesterol; TG: triglycerides; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; non-HDL-C: non-high-density lipoprotein cholesterol; LDL-C_D: LDL-C measured by direct assay; LDL-C_F: LDL-C calculated by Friedewald formula; LDL-C_S: LDL-C calculated by Sampson formula; LDL-C_M: LDL-C calculated by Martin-Hopkins formula; LDL-C_E: LDL-C calculated by the extended Martin-Hopkins formula

Comparison of LDL-C concentrations calculated by various formulas versus direct assays

Overall concordances of the different equations for LDL-C prediction

Strata-specific median ratios of TG/VLDL-C were used to predict LDL-C levels. These predictions were made using the extended Martin-Hopkins’ formula. In order to see how the predicted change at different non-HDL-C and TG levels, the results of the calculations made according to the cut-off values determined for these variables are shown in Supplementary Table 1. This two-dimensional cross-table contained the median ratio of TG/VLDL-C with 180 cells were generated, with TG sublevels in the rows and non-HDL-C sublevels in the columns. In this 180-cell table, the cells display the median statistics for TG/VLDL-C ratio.

In Fig. 1, overall concordances of LDL-C predicted for each assay are given. It can be seen that the extended Martin-Hopkins formula produced the highest, and the Friedewald formula produced the lowest concordances within each direct assay. In Siemens direct assay, the concordance of Martin-Hopkins’ formula was higher than Sampson’s formula. In Beckman and Roche direct assays, the performances of these two formulas were found to be very similar.

The distribution density of LDL-C according to the results measured and estimated by different equations

In Fig. 2, the density plots show the distributions of the LDL-C levels measured by direct assays and predicted by the Friedewald, Sampson, Martin-Hopkins, and extended Martin-Hopkins equations. The difference between the median of predicted LDL-C levels by each equation and the median of LDL-C levels measured by direct methods were displayed with the red line in this figure. When Siemens direct assay was evaluated by the Friedewald and Sampson equations in Beckman and Roche direct assays, LDL-C levels were underestimated, whereas they were overestimated. When the distributions of the Martin-Hopkins and extended Martin-Hopkins were examined in Siemens direct assay, it was seen that the shape of Martin-Hopkins’ distributions was close to the measured LDL-C.

Concordances between the direct measured and predicted LDL-C by LDL-C strata

In Fig. 3, the concordances of the different equations for LDL-C prediction by different LDL-C sublevels are given for each assay. In all cases, the extended Martin-Hopkins equation gave the most concordant results. In the first category of LDL-C levels (<110 mg/dL), the Friedewald equation gave the least concordant results for each assay. The Sampson formula performed slightly better than the Martin-Hopkins equation for the Roche direct assay, but slightly worse than the Martin-Hopkins equation for the Siemens direct assay. Very similar results were observed for these two equations for the Beckman assay. In the second category of LDL-C levels (110 to 129 mg/dL), again, the lowest concordances were obtained with the Friedewald formula. The poor performance in the Friedewald formula for the Roche assay was striking. The performance of the Martin-Hopkins equation was higher than the Sampson equation for Beckman and Roche direct assays. These two approaches have approximately equal performance for Siemens direct assay. In the last category of LDL-C levels (>=130 mg/dL), the performance of the Sampson formula was relatively higher than the Martin-Hopkins equation for all direct assays. The performance of the Friedewald equation was higher than the Sampson and the Martin-Hopkins equation for Siemens direct assay.

Concordances between the direct measured and predicted LDL-C by LDL-C strata by triglycerides levels

In Fig. 4, overall concordances for LDL-C predicts by TG sublevels (< 75 mg/dL, 75 to 129 mg/dL and ≥ 130 mg/dL) were given for each assay. The highest calculated concordance values were seen in the extended Martin-Hopkins when the TG sublevel was analyzed for each assay. Nevertheless, the concordances of the equations against increased TG levels decreased with any direct assay. The calculated concordance of the equations given by TG sublevels for LDL-C sublevels is shown in three figures (Supplementary Fig. 1-Fig. 3).

In patients with LDL-C < 110 mg/dL, the extended Martin-Hopkins equation gave the highest performance by comparison with other equations, i.e., the Friedewald, Sampson, and Martin-Hopkins for all TG levels. Remarkably, the extended Martin-Hopkins equation decreased slightly for Beckman and Roche direct assays, whereas this formula had an almost equal performance for Siemens direct assay as the TG level increased from < 75 mg/dL to ≥ 130 mg/dL. When the TG was < 75 or between 75 mg/dL and 129 mg/dL, the Friedewald and Sampson equations had similar performance for each assay. However, the performance of the Friedewald method was lower than the Sampson method when TG level was ≥ 130 mg/dL for each assay. Even though the performance of the Martin-Hopkins equation decreased as the TG level increased for Beckman and Roche direct assays, this formula had an almost equal performance for Siemens direct assay. The performance of the Friedewald formula significantly reduced as the TG level increased for Beckman and Roche direct assay. For LDL-C between 110 to 129 mg/dL, the performance of the extended Martin-Hopkins equation was again the highest for each assay when TG level was < 75 mg/dL. The concordance of the extended Martin-Hopkins approach decreased as TG level increased for Roche direct assay, whereas firstly, it decreased then, it increased for Beckman and Siemens direct assays. In TG < 75 mg/dL, as the Friedewald and Sampson had similar performance for each assay, the distinction between these equations increased while the TG level increased for each assay. Even though the performance of the Friedewald, Sampson, and Martin-Hopkins equations generally decreased while the TG level increased for Beckman and Roche direct assays, these performances first decreased and then increased for Siemens direct assay. In patients with LDL-C ≥ 130 mg/dL, the performance of the Martin-Hopkins and the extended Martin-Hopkins equations constantly decreased while the TG levels were increasing for Roche direct assay; however, it first increased and then decreased for Siemens and Beckman direct assay. The Friedewald and Sampson formulas showed the same performance in the TG level < 75 mg/dL for each assay. However, while this situation changed in the TG levels between 75 and 129 mg/dL for Siemens and Roche direct assays, it remained the same for Beckman direct assay. While the performance of the Friedewald equation was lower than the Sampson equation in the TG levels between 75 and 129 mg/dL for Roche direct assay, it had a reverse situation for Siemens direct assay.

While the performance of the Friedewald equation was lower than Sampson equation for Beckman and Roche direct assays, these performances were the same for Siemens direct assay at the TG level ≥ 130 mg/dL.

Concordances between the direct measured and predicted LDL-C by LDL-C strata by non-HDL-C levels

In Fig. 5, the concordances of LDL-C calculated according to the three sublevels (<120 mg/dL, 120 to 144 mg/dL, and ≥145 mg/dL) created for non-HDL-C are given for all assays. The results indicated that the highest concordance values belonged to the extended Martin-Hopkins for all non-HDL sublevels and each assay. Furthermore, for Siemens and Roche direct assays, as the non-HDL-C level changed from <120 mg/dL to ≥145 mg/dL, there was a decrease in the concordance values calculated using the Friedewald equation. The calculated concordance of the equations given by non-HDL-C sublevels for LDL-C strata is shown in three figures (Supplementary Fig. 4-Fig. 6).

For the patients in the first category of LDL-C (<110 mg/dL) in all assays, the concordance calculated by the extended Martin-Hopkins equation as the non-HDL-C level changed from <120 mg/dL to >=145 mg/dL, values decreased. The performance of the Friedewald, Sampson, and the Martin-Hopkins equations were almost similar in the non-HDL sublevel <144 mg/Dl; however, this situation changed when non-HDL-C exceeded 144 mg/dL. The patients in the second category of LDL-C (between 110 to 129 mg/dL), as the non-HDL-C levels changed from <120 mg/dL to ≥145 mg/dL, the concordance of the Martin-Hopkins increased, but the concordance of extended Martin-Hopkins decreased for Roche direct assay. The patients in the third category of LDL-C (≥130 mg/dL), extended Martin-Hopkins gave the highest concordance for each assay, and all sublevels of the non-HDL-C except non-HDL-C was between 120 to 144 mg/dL for Siemens direct assay.

Regression analysis for differences between the direct and calculated LDL-C with different assay methods

In Fig. 6, the linear regression analyses were performed to examine the correlation between the LDL-C levels predicted by formulas and measured by direct assays. It has been seen that the LDL-C values estimated by the extended Martin-Hopkins equation form a better regression model. The R square value calculated with extended Martin-Hopkins equation was found to be higher than those calculated with the other equations, 0.95 for Roche direct measurement method, 0.90 for Beckman direct measurement method, and 0.92 for Siemens direct measurement method. As a result, the extended Martin-Hopkins equation is highly associated with the LDL-C measured by direct assay equation.

Residual error plots for differences between the direct and calculated LDL-C with different assay methods

Fig. 7 shows the variation of the difference between direct LDL-C measurement and calculated LDL-C over the change in TG level. It is clear that in the case of increased TG levels, the LDL-C levels were underestimated with the Friedewald and Sampson formula. There was also some underestimated condition in the Martin-Hopkins equation, although not as large as in the Friedewald and Sampson for Beckman and Roche direct assays. It was observed that compared with the other assays Siemens direct assay in terms of TG levels was less for the Martin-Hopkins formula. The differentiation of TG levels did not affect the extended Martin-Hopkins formula in each assay and remained close to zero at all TG levels. The extended Martin-Hopkins formula was calculated lowest mean absolute deviation statistics for each assay.

The proportion of misclassified samples per direction by predicted low-density lipoprotein cholesterol category

In Fig. 8, diverging bar charts show the total percentage of underclassified and overclassified samples within each LDL-C category. In each bar, how well different equations concordant the LDL-C categories or how many categories they predict up/down are expressed in different colors. The concordance of the extended Martin-Hopkins equations was the highest according to other equations in all sublevels of the LDL-C levels and each assay. In LDL-C levels between 110 to 129 mg/dL, the underestimation of LDL-C occurred in 0.7% with the extended Martin-Hopkins equation compared with 34.1% with the Friedewald equation, 24.1% with the Sampson equation, and 20.0% with the Martin-Hopkins equation for Roche direct assay.

In LDL-C < 110 mg/dL, the degree of underestimation was least pronounced with the extended Martin-Hopkins equation, with only 7.6% of patients underclassified for Roche assay and with only 9.0% of patients underclassified for Beckman assay. In Beckman and Roche direct assays, this statistic was 29.3% and 31.6% for the Friedewald equation, 29.3% and 27.3% for the Sampson equation, 27.4% and 27.7% for the Martin-Hopkins equation, respectively. In general, the concordant prediction of LDL-C for the Friedewald equation was the lowest according to other equations in all sublevels of the LDL-C levels and each assay. In most cases, LDL-C levels were overestimated in the Sampson and the Martin-Hopkins equations. For instance, in LDL-C levels between 110 and 129 mg/dL, the degree of overestimation with the Sampson and the Martin-Hopkins formulas was 20.9%, while this degree was 19.4% for the Friedewald and 10.5% for the extended Martin-Hopkins in Siemens direct assay. The highest degree of underestimation was observed in the Friedewald equation for all subclasses of LDL-C levels and all types of assays.

The disadvantages of the Friedewald formula, such as requiring fasting serum and underestimating LDL-C levels below 70 and above 150 mg/dL, led to the recommendation of new formulas. The Martin-Hopkins and Sampson are among these formulas, and validation of these formulas was examined for adult data in several populations [17–25]. Studies investigating the validation of these recent formulations in the pediatric population are very limited. Roper et al. compared the Friedewald and Martin-Hopkins formulas in a study of 127 pediatric data [16]. The authors recommended using the Martin-Hopkins formula in cases where LDL-C is below 100 mg/dL, and TG is between 150 and 399 mg/dL. Garoufi et al. [8] investigated the Friedewald and Anandaraja formulas in the data of 1005 children, 317 of which were dyslipidemic, and as a result, suggested that the Friedewald formula may be a good screening tool, and the Anandaraja formula may be more appropriate in the follow-up of patients with dyslipidemia. Cicero et al. [14] conducted a study with 2605 pediatric data, compared the Friedewald and Sampson formulas, and as a result, stated that the Sampson formula was more reliable in estimating LDL-C. Our study, up till now, is the study with the largest sample size investigating the validation of LDL-C formulas in the pediatric population. Evaluation of the different direct assays in the performed comprehensive analyzes, as well as the median statistics of our own population were calculated for the extended Martin-Hopkins formula, and the important contributions have been made to the study by using these. The results of our study revealed that there is a significant difference in the performance of the equations used in the LDL-C prediction for different assays. The method with the highest overall concordance coefficient was the extended Martin-Hopkins formula. The extended Martin-Hopkins formula showed the best performance in all LDL-C, non-HDL-C, and TG subgroups. The lowest-performing equation was the Friedewald formula.

LDL-C cut-off points of <110 mg/dL and ≥130 mg/dL have been used to define acceptable and abnormal results in children and adolescents, respectively [26]. In this study, we found higher concordance between the extended Martin-Hopkins formula and direct measurements compared to other formulas for both <110 mg/dL and ≥130 mg/dL categories. The Friedewald is the most commonly used formula in the clinical laboratory to predict LDL-C values [27]. Considering the category of ≥130 mg/dL in Roche direct assay, it was determined that the Friedewald formula underestimates 55% of the individuals compared to the direct measurement method. This ratio was found to be only 13.2% in the extended Martin-Hopkins formula. Similar results were obtained for Siemens and Beckman assays. The Friedewald formula underestimates 26% and 38.7% of individuals, while the extended Martin-Hopkins formula underestimates 16% and 20.9% of individuals for Siemens and Beckman assays, respectively. Accordingly, we think that the use of the Friedewald formula in children and adolescents may increase the possibility of a false-negative diagnosis of dyslipidemia. Coronary artery disease (CAD) is the leading cause of global mortality and has a significant health-economic impact [28, 29] The number of CVD deaths gradually accelerated from 12.1 million to 18.6 million by 2019 [28]. Increased LDL-C levels have been linked to an increased risk of CVD mortality [30]. It has been shown that the average LDL-C reduction of 27 mg/dL (12.6%) led to a 19% risk reduction in the primary CVD death and nonfatal myocardial infarction. Moreover, regardless of age, sex, baseline LDL-C, or previous CVD status, a 39 mg/dL LDL-C reduction resulted in around 20% fewer CVD occurrences.[31] Thus, we think that underestimation of the LDL-C levels may cause disruptions in the follow-up and treatment of patients and increase the risk of future CVD. Although the extended Martin-Hopkins has the highest concordance at both ≥110 mg/dL and 110-129 mg/dL categories with direct measurement, it was determined that the extended Martin-Hopkins formula overestimates 10.4% and 5.7% of individuals for Roche direct assay, 12% and 10.5% of individuals for Siemens direct assay, and 9.8% and 10.1% of individuals for Beckman direct assay respectively. In most cases, this percentage was found to be lower in the other formula for Beckman and Roche assays. Therefore, we think that using the extended Martin-Hopkins formula may increase the false positive rate, albeit at a low rate, thereby increasing the risk of unnecessary medical procedures. However, as a generalized result, it can be said that the extended Martin-Hopkins formula gave the lowest proportion of misclassified samples for all assays.

Increased TG level causes chylomicrons to accumulate at high levels. This may change the relationship between TG and cholesterol, and LDL-C prediction has larger errors in the Friedewald formula [12]. Although many previous studies evaluated the effect of high TG levels on the accuracy of the equations developed for LDL-C prediction in adults, little is known regarding the effect of higher TG levels in children and adolescents. As we indicated in the previous paragraph, the cut-off value of ≥130 mg/dL is being used to define abnormal high TG levels in children and adolescents. Thus, we evaluated the concordance between direct measurement and equations developed for LDL-C prediction at the TG levels ≥130 mg/dL. We determined that the extended Martin-Hopkins formula has higher concordance with direct measurement compared to other formulas. Accordingly, we think that the extended Martin-Hopkins formula may be used to obtain the best performance in predicting the LDL-C levels in the case of higher TG levels.

The performance of the equations used in LDL-C prediction varies in different direct assays. The extended Martin-Hopkins formula may be a suitable equation when evaluating LDL-C levels in the pediatric population. Researchers can calculate the extended Martin-Hopkins formula using the median statistics of their own population. Further validation is needed in different populations.

CAD

Coronary artery disease

CVD

Cardiovascular disease

LDL-C

low-density lipoprotein cholesterol

LDL-CD

Direct LDL-C

LDL-C_F

Friedewald’s LDL-C prediction formula

LDL-C_M

Martin-Hopkins LDL-C prediction formula

LDL-C_S

Sampson’s LDL-C formula

Non-HDL-C

Non-high-density lipoprotein cholesterol

Total cholestrol

Triglycerides

Funding

This study was supported by the Research Fund of Erciyes University [TSG-2021-10912]. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Availability of Data

Data cannot be shared publicly because of the restrictions of the Sivas Cumhuriyet University Ethics committee. Data are available upon a reasonable request for researchers who meet the criteria for access to confidential data. A non-author point of contact is: Demet Kablan, E-mail: [email protected] .

Competing Interests

The authors have no relevant financial or non-financial interests to disclose.

Author Contributions

Gozde Erturk Zararsiz conceived and designed the experiments, performed the analyses, contributed reagents/materials/analysis tools, figures and/or tables, authored or reviewed drafts of the paper, approved the final draft. Ahu Cephe and Necla Kochan prepared the original draft, contributed to analyses, helped perform the analysis with constructive discussions, discussed the results and authored or reviewed drafts of the paper, approved the final draft. Serra İlayda Yerlitas validated the results and helped in data curation. Serkan Bolat and Halef Okan Doğan collected the data, contributed the methodology, authored or reviewed drafts of the paper. Gökmen Zararsız supervised the study, helped in writing, discussed the results, authored or reviewed drafts of the paper, approved the final draft.

Ethics approval

The study was performed in accordance with the Declaration of Helsinki and adhered to Good Clinical Practice guidelines. The study was approved by the Ethics Committee of Sivas Cumhuriyet University (2022-03/17).

Consent to participate

No informed consent was obtained.

Consent for publication

Written informed consent was obtained from all participants.

Daniels SR, Greer FR (2008) Lipid screening and cardiovascular health in childhood. Pediatrics. https://doi.org/10.1542/peds.2008-1349
Berenson GS, Srinivasan SR, Bao W, Newman W, Tracy RE, Wattigney WA, Newman 3rd WP, Tracy RE, Wattigney WA (1998) Association between multiple cardiovascular risk factors and atherosclerosis in children and young adults. The Bogalusa Heart Study. N Engl J Med 338:
Sanyaolu A, Okorie C, Qi X, Locke J, Rehman S (2019) Childhood and Adolescent Obesity in the United States: A Public Health Concern. Glob Pediatr Heal. https://doi.org/10.1177/2333794X19891305
McGill HC, McMahan CA, Zieske AW, Malcom GT, Tracy RE, Strong JP (2001) Effects of nonlipid risk factors on atherosclerosis in youth with a favorable lipoprotein profile. Circulation. https://doi.org/10.1161/01.CIR.103.11.1546
Gidding SS (2001) Cholesterol Guidelines Debate. Pediatrics. https://doi.org/10.1542/peds.107.5.1229
Fox KM, Wang L, Gandra SR, Quek RGW, Li L, Baser O (2016) Clinical and economic burden associated with cardiovascular events among patients with hyperlipidemia: A retrospective cohort study. BMC Cardiovasc Disord. https://doi.org/10.1186/s12872-016-0190-x
National Heart, Lung and BI (2011) Expert Panel on Integrated Guidelines for Cardiovascular Health and Risk Reduction in Children and Adolescents: Summary Report. Pediatrics 128:S213–S256
Garoufi A, Drakatos A, Tsentidis C, Klinaki E, Paraskakis I, Marmarinos A, Gourgiotis D (2017) Comparing calculated LDL-C with directly measured LDL-C in healthy and in dyslipidemic children. Clin Biochem. https://doi.org/10.1016/j.clinbiochem.2016.05.026
Friedewald Wt Fau - Levy RI, Levy Ri Fau - Fredrickson DS, Fredrickson DS, Clin C Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.
Yu HH, Markowitz R, De Ferranti SD, Neufeld EJ, Farrow G, Bernstein HH, Rifai N (2000) Direct measurement of LDL-C in children: Performance of two surfactant- based methods in a general pediatric population. Clin Biochem. https://doi.org/10.1016/S0009-9120(00)00055-2
Palmer MK, Barter PJ, Lundman P, Nicholls SJ, Toth PP, Karlson BW (2019) Comparing a novel equation for calculating low-density lipoprotein cholesterol with the Friedewald equation: A VOYAGER analysis. Clin Biochem. https://doi.org/10.1016/j.clinbiochem.2018.10.011
Martin SS, Blaha MJ, Elshazly MB, Toth PP, Kwiterovich PO, Blumenthal RS, Jones SR (2013) Comparison of a novel method vs the Friedewald equation for estimating low-density lipoprotein cholesterol levels from the standard lipid profile. JAMA - J Am Med Assoc. https://doi.org/10.1001/jama.2013.280532
Sampson M, Ling C, Sun Q, et al (2020) A New Equation for Calculation of Low-Density Lipoprotein Cholesterol in Patients with Normolipidemia and/or Hypertriglyceridemia. JAMA Cardiol. https://doi.org/10.1001/jamacardio.2020.0013
Cicero AFG, Fogacci F, Patrono D, et al (2021) Application of the Sampson equation to estimate LDL-C in children: Comparison with LDL direct measurement and Friedewald equation in the BLIP study. Nutr Metab Cardiovasc Dis. https://doi.org/10.1016/j.numecd.2021.02.034
Molavi F, Namazi N, Asadi M, Sanjari M, Motlagh ME, Shafiee G, Qorbani M, Heshmat R, Kelishadi R (2020) Comparison common equations for LDL-C calculation with direct assay and developing a novel formula in Iranian children and adolescents: The CASPIAN v study. Lipids Health Dis. https://doi.org/10.1186/s12944-020-01306-7
Roper SM, Cao J, Tam E, Devaraj S (2017) Performance of calculated and directly measured low-density lipoprotein cholesterol in a pediatric population. Am J Clin Pathol 148:42–48
Sathiyakumar V, Blumenthal R, Elshazly, M B, Blumenthal RS. (2020) New information on accuracy of LDL-C estimation. Am Coll Cardiol 1–11
Sathiyakumar V, Park J, Quispe R, et al (2018) Impact of novel low-density lipoprotein-cholesterol assessment on the utility of secondary non-high-density lipoprotein-c and apolipoprotein b targets in selected worldwide dyslipidemia guidelines. Circulation 138:244–254
Chaen H, Kinchiku S, Miyata M, Kajiya S, Uenomachi H, Yuasa T, Takasaki K, Ohishi M (2016) Validity of a novel method for estimation of low-density lipoprotein cholesterol levels in diabetic patients. J Atheroscler Thromb 23:1355–1364
Shin D, Bohra C, Kongpakpaisarn K (2018) Novel method versus the Friedewald method for estimating low-density lipoprotein cholesterol in determination of the eligibility for statin treatment for primary prevention in the United States. Med (United States). https://doi.org/10.1097/MD.0000000000010612
Song Y, Lee HS, Baik SJ, et al (2021) Comparison of the effectiveness of Martin’s equation, Friedewald’s equation, and a Novel equation in low-density lipoprotein cholesterol estimation. Sci Rep 11:1–11
Piani F, Cicero AFG, Ventura F, et al (2021) Evaluation of twelve formulas for LDL-C estimation in a large, blinded, random Italian population. Int J Cardiol 330:221–227
Vargas-Vázquez A, Bello-Chavolla OY, Antonio-Villa NE, Mehta R, Cruz-Bautista I, Aguilar-Salinas CA (2021) Comparative assessment of LDL-C and VLDL-C estimation in familial combined hyperlipidemia using Sampson’s, Martin’s and Friedewald’s equations. Lipids Health Dis 20:1–9
Zafrir B, Saliba W, Flugelman MY (2020) Comparison of novel equations for estimating low-density lipoprotein cholesterol in patients undergoing coronary angiography. J Atheroscler Thromb 27:1359–1373
Zararsız GE, Bolat S, Cephe A, Kochan N, Yerlitaş SI, Doğan HO, Zararsız G (2022) Validation of Friedewald, Martin / Hopkins and Sampson Equations in the Low-Density Lipoprotein Cholesterol Estimation with Different Assays. https://doi.org/https://doi.org/10.1101/2022.01.29.22270067
Lim JS, Kim EY, Kim JH, Yoo JH, Yi KH, Chae HW, Choi JH, Kim JY, Hwang IT (2020) 2017 clinical practice guidelines for dyslipidemia of korean children and adolescents. Ann Pediatr Endocrinol Metab. https://doi.org/10.6065/apem.2040198.099
Vujovic A, Kotur-Stevuljevic J, Spasic S, Bujisic N, Martinovic J, Vujovic M, Spasojevic-Kalimanovska V, Zeljkovic A, Pajic D (2010) Evaluation of different formulas for LDL-C calculation. Lipids Health Dis. https://doi.org/10.1186/1476-511X-9-27
Roth GA, Mensah GA, Johnson CO, et al (2020) Global Burden of Cardiovascular Diseases and Risk Factors, 1990–2019: Update From the GBD 2019 Study. J Am Coll Cardiol. https://doi.org/10.1016/j.jacc.2020.11.010
Gheorghe A, Griffiths U, Murphy A, Legido-Quigley H, Lamptey P, Perel P (2018) The economic burden of cardiovascular disease and hypertension in low- and middle-income countries: A systematic review. BMC Public Health. https://doi.org/10.1186/s12889-018-5806-x
Abdullah SM, Defina LF, Leonard D, et al (2018) Long-Term Association of Low-Density Lipoprotein Cholesterol with Cardiovascular Mortality in Individuals at Low 10-Year Risk of Atherosclerotic Cardiovascular Disease: Results from the Cooper Center Longitudinal Study. Circulation. https://doi.org/10.1161/CIRCULATIONAHA.118.034273
Stein EA, Raal FJ (2014) Targeting LDL: Is lower better and is it safe? Best Pract Res Clin Endocrinol Metab 28:309–324

No competing interests reported.

Download PDF

Version 2

posted

You are reading this latest preprint version

Validation of Low-Density Lipoprotein Cholesterol Equations in Pediatric Population

Status:

Version 2

Abstract

Purpose

Methods

Results

Conclusions

Figures

What Is Known

What Is New

Introduction

Materials And Methods

Results

The proportion of misclassified samples per direction by predicted low-density lipoprotein cholesterol category

Discussion

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 2