**4. Analytical accuracy of LDL-C in kidney disease and hemodialysis**

### **4.1 Metaanalysis of results achieved with Friedewald equation compared to reference ultracentrifugation method**

Our metaanalysis contains data from all available published papers dealing (entirely or as a part of broader study) with analytical accuracy of LDL-C assessment according to the Friedewald equation as compared with the reference UC method in CRI patients. The methodology of the metanalysis is described in our previous paper (Gaško et Sánchez-Meca, 2009). It includes studies from 1990 until the end of year 2010. Four studies were found but two of them were further divided according to the details of the study. The basic data of the studies are summarized in Table 10 and the results of metaanalysis in Fig 3.


Table 10. Studies included in metaanalysis, divided into subgroups according to inclusion criteria.

Fig. 3. Difference in means LDL-C with 95% CI in 4 studies included with 6 subgroups of patients with severe nephropathy on different forms of dialysis, with fixed and random effects model summary difference. Forest plot. Units: mmol/l. Patients included 586.

For patients in different dialysis regimes the summary differences of means of LDL-C is 0.234 mmol/l, what presents a bias of -4.9%. The information value of this result is much higher when compared with results on healthy probands od with results of patients with

Lipid and Lipoprotein Abnormalities in Chronic Renal Insufficiency: Review 359

Agrawal et al (2010) in USA in 2009 approximately 2200 laboratories used direct assays but 3300 calculated LDL-C according to Friedewald. In Czech and Slovak republics the ratio was different (159 laboratories on direct assays and only 83 on formula (Gaško et al, 2011). There is no official recommendation on international level about the methodology of LDL-C estimation. The choice depends solely on the decision of the laboratory and/or the health care provider. Another confounding factor is that the same laboratory can change the actual type of the direct method or use alternatively a direct method or the calculation mostly

We realized a multicentric prospective study to compare the results achieved by different direct method compared with the results of Friedewald calculation (Gaško et al, 2011) in 13 laboratories. All of them were controlled also in the frame of an external quality assesment system. All of them diclosed results of the control sera in the appropriate range. Results of the direct assay and the calculation were compared on a set of randomly selected 200 patient samples with a broad range of diseases including CRI. In Fig 5. the average difference between the results of direct (considred in this study as reference) method and the calculated results are shown. The differences are in the range from -8 % ap to + 30 %. Of course the evident limitation of this study is the lack of comparison with the reference ultacentrifugation method. Despite this limitation the results of this simple study combined with the results of metanalysis (4.1.) show an unacceptable degree of uncertainity of LDL-C

Lab Number

Fig. 5. Average differences (n = 200) in calculated LDL-C compared with results of the direct assay in 13 different laboratories. Value measured by the direct method in each laboratory is

A very interesting question is the situation after kidney transplatation. These patients at the first sight are "cured" but as a matter of fact they are far from being "healthy" beacause of

considered 100%. Data are shown as median and 2.5 to 97.5 percentiles.

1 2 3 4 5 6 7 8 9 10 11 12 13

estimation in individual patients not to mention those who are at special risk.

without any warning towards the clinician.

0

50

100

150

200

Difference LDL-F vs LDL-D in %

250

300

350

400

450

other diseases. We identified 23 publications including 17 213 patients analyzing accuracy of LDL-C assessment on healthy probands or in patients with hyperlipidemia from 1990 until 2009. The summary differences of means in this setting is 0.108 mmol/l and the bias is +2.3%. Furhter 3 papers followed analytical accuracy LDL-C in 350 patients with I and II diabetes mellitus or with hepatopathy. Total difference in means in this group was 0.234 mmol/l, (bias +6.8%; Fig 4).

From these findings it is possible to conclude that calculated concentrations of LDL-C according to the Friedewald equation can differ from the true values by almost 12 % systemic error. For clinical practice this analytical error can have a unwelcome situation because in approximately 7% of patients proper and necessary treatment according to guidelines is not prescribed and in another 5 % superfluous treatment is prescribed. This second group however is not in a danger of non-treated dyslipidemia because but there is problem of pharmacoeconomy.

Fig. 4. Difference in means LDL-C with 95% CI in 3 studies included with 6 subgroups of patients with diabetes mellitus or hepatopathy, with fixed and random effects model summary difference. Forest plot. Units: mmol/l. Patients included 350. Figure from Gaško & Sánchez-Meca, 2009, with permision.

### **4.2 Unreliability of LDL-C assessment in everyday laboratory practice**

There is only one method providing a true picture about the composition of the particles ultracentrifugation with subsequent chemical analysis of their composition. This method is the golden standard and the anchor of all methods used in routine clinical practice.

For a couple of years there was only one indirect way to estimate the cholesterol content of LDL particles. Despite modern methods estimating total cholesterol, HDL-cholesterol and triglycerides with a high level of precision and accuracy they have a certain degree of uncertainity and the overall uncertainity is too high to give a reliable picture on LDLcholesterol even in probands without any confounding factors. In CRI patients the situation is even worse.

Recently clinical chemistry laboratories can use one commercial kits based on 7 different method for direct LDL-C assays. Some laboratories however calculate LDL-C according to the Friedewald formula or use both procedures (e.g. direct assay only in the case of elevated total cholesterol or in patients with otherwise elevated cardiovascular risk). According to

other diseases. We identified 23 publications including 17 213 patients analyzing accuracy of LDL-C assessment on healthy probands or in patients with hyperlipidemia from 1990 until 2009. The summary differences of means in this setting is 0.108 mmol/l and the bias is +2.3%. Furhter 3 papers followed analytical accuracy LDL-C in 350 patients with I and II diabetes mellitus or with hepatopathy. Total difference in means in this group was 0.234

From these findings it is possible to conclude that calculated concentrations of LDL-C according to the Friedewald equation can differ from the true values by almost 12 % systemic error. For clinical practice this analytical error can have a unwelcome situation because in approximately 7% of patients proper and necessary treatment according to guidelines is not prescribed and in another 5 % superfluous treatment is prescribed. This second group however is not in a danger of non-treated dyslipidemia because but there is

**Model Study name Statistics for each study Difference in means and 95% CI Difference Lower Upper in means limit limit**

Fig. 4. Difference in means LDL-C with 95% CI in 3 studies included with 6 subgroups of patients with diabetes mellitus or hepatopathy, with fixed and random effects model

**4.2 Unreliability of LDL-C assessment in everyday laboratory practice** 

the golden standard and the anchor of all methods used in routine clinical practice.

summary difference. Forest plot. Units: mmol/l. Patients included 350. Figure from Gaško &

There is only one method providing a true picture about the composition of the particles ultracentrifugation with subsequent chemical analysis of their composition. This method is

For a couple of years there was only one indirect way to estimate the cholesterol content of LDL particles. Despite modern methods estimating total cholesterol, HDL-cholesterol and triglycerides with a high level of precision and accuracy they have a certain degree of uncertainity and the overall uncertainity is too high to give a reliable picture on LDLcholesterol even in probands without any confounding factors. In CRI patients the situation

Recently clinical chemistry laboratories can use one commercial kits based on 7 different method for direct LDL-C assays. Some laboratories however calculate LDL-C according to the Friedewald formula or use both procedures (e.g. direct assay only in the case of elevated total cholesterol or in patients with otherwise elevated cardiovascular risk). According to

**-1,00 -0,50 0,00 0,50 1,00 Favours FF Favours BQ**

Matas et al 2, 1994 0,370 0,214 0,526 Matas et al 3, 1994 0,310 0,215 0,405 Hirany et al 1, 1997 0,180 0,099 0,261 Hirany et al 2, 1997 0,280 0,234 0,326 Wägner et al, 2000 0,240 0,197 0,283 Wägner et al, 2003 0,090 0,038 0,142 Fixed 0,220 0,196 0,245 Random 0,234 0,160 0,309

mmol/l, (bias +6.8%; Fig 4).

problem of pharmacoeconomy.

Sánchez-Meca, 2009, with permision.

is even worse.

Agrawal et al (2010) in USA in 2009 approximately 2200 laboratories used direct assays but 3300 calculated LDL-C according to Friedewald. In Czech and Slovak republics the ratio was different (159 laboratories on direct assays and only 83 on formula (Gaško et al, 2011).

There is no official recommendation on international level about the methodology of LDL-C estimation. The choice depends solely on the decision of the laboratory and/or the health care provider. Another confounding factor is that the same laboratory can change the actual type of the direct method or use alternatively a direct method or the calculation mostly without any warning towards the clinician.

We realized a multicentric prospective study to compare the results achieved by different direct method compared with the results of Friedewald calculation (Gaško et al, 2011) in 13 laboratories. All of them were controlled also in the frame of an external quality assesment system. All of them diclosed results of the control sera in the appropriate range. Results of the direct assay and the calculation were compared on a set of randomly selected 200 patient samples with a broad range of diseases including CRI. In Fig 5. the average difference between the results of direct (considred in this study as reference) method and the calculated results are shown. The differences are in the range from -8 % ap to + 30 %. Of course the evident limitation of this study is the lack of comparison with the reference ultacentrifugation method. Despite this limitation the results of this simple study combined with the results of metanalysis (4.1.) show an unacceptable degree of uncertainity of LDL-C estimation in individual patients not to mention those who are at special risk.

Fig. 5. Average differences (n = 200) in calculated LDL-C compared with results of the direct assay in 13 different laboratories. Value measured by the direct method in each laboratory is considered 100%. Data are shown as median and 2.5 to 97.5 percentiles.

A very interesting question is the situation after kidney transplatation. These patients at the first sight are "cured" but as a matter of fact they are far from being "healthy" beacause of

Lipid and Lipoprotein Abnormalities in Chronic Renal Insufficiency: Review 361

Proper clinical and biochemical evaluation of the underlying kidney disease and continuous monitoring of its progression. Special attenttion should be devoted to proteinuria and hypalbuminemia because they directly influence the metabolism of LDL-type particles and

Evaluation of the risk factors not directly associated with the kidney disease (smoking,

Evaluation of lipid parameters – total cholesterol, LDL-cholesterol measured directly, HDLcholesterol, triglycerides, apoprotein B100, apoporotein AI and Lp(a). The evaluation should not be a mechanistical process but should be based on pathophysiology of atherosclerosis in renal disease and should consider the strengths and weaknesses of the employed methods. Measurement of other parameters not fully validated yet from analytical point of view and from their clinical usefulness (small dense LDLs, parameters of the oxidative stress and

Agrawal, M., Spencer, H.J., Fass, F.H. (2010) Method of LDL cholesterol measurement

*Journal of Investigative Medicine*, Vol.58, No.8, pp. 945-949, ISSN 1081-5589 Attman, P.O., Samuelson, O. & Alaupovic P. (2011) The effect of decreasing renal function

Bairaktari, E., Tzallas, C., Kalientzidou, M. & al. (2004) Evaluation of alternative calculation

patiens. *Clinical Biochemistry*, Vol.37, No.10, pp. 937-940, ISSN 0009-9120 Bairaktari, E., Elisaf, M., Tzallas, Ch. & al. (2001) Evaluation of five methods for determining

*Biochemistry*, Vol.34, No.8, pp. 593-602, ISSN 0009-9120

influences classification of LDL cholesterol treatment goals: Clinical research study.

on lipoprotein profiles. *Nephrol Dialysis Transplant* doi:10.1093/ndt/gfq762, ISSN

methods for determining low-density lipoprotein cholesterol in hemodialysis

low-density lipoprotein cholesterol (LDL-C) in hemodialysis patients. *Clinical* 

their concentration in the blood.

**6. List of abbreviations** 

**7. References** 

1460 - 2385

CHY Chylomicron

HD Hemodialysis

TC Total cholesterol TG Triglyceride

obesity, diabetes, hypertension, etc.)

antioxidant systems) is possible for research purposes.

CKD Chronic kidney disease CRI Chronic renal insufficiency ESRD End stage renal disease

HDL High-density lipoprotein

LDL Low-density lipoprotein

TRL Triglyceride rich particle VLDL Very low-density lipoprotein

IDL Intermediate-density lipoprotein LCAT Lecithin-cholesterol acyltransferase

C Cholesterol (as LDL-C, HDL-C, nonHDL-C)

the long-life immunosuppression therapy and of the slightly decreased kidney function in most of them. Balal et al (2010) compared results from 103 renal transplant recipients and the sum of differences between the calculation according to Friedewaldovho and the direct method was – 6,5 % ± 6,6 % (mean ± SD). According to Tsimihodimos et al (2008) after transplantation one very important atherogenic factor disappears – the increased level of Lp(a).
