| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Hôpital d’instruction, des armées PERCY, Clamart, France
aAddress correspondence to this author at: Hôpital d’instruction des armées PERCY, Laboratoire de biochimie, 101 avenue Henri Barbusse, 92140 Clamart, France. Fax 33-141-466-401; e-mail carinegarcia92{at}wanadoo.fr.
To the Editor:
LDL cholesterol (LDL-C) is commonly measured for evaluation and management of hypercholesterolemia. The Friedewald formula [LDL-C = (total cholesterol)—(HDL cholesterol)—triglycerides/2.2 for mmol/L] is commonly used to determine LDL-C, but this method has some well-established shortcomings and may not meet the National Cholesterol Education Program criteria of total error <12% (1)(2)(3). Another approach is the use of homogeneous methods for direct quantification of LDL-C. In our laboratory, we use the Kyowa Medex® method 2nd-generation assay on a Roche Cobas Integra 800® analyzer (1)(3)(4).
We report a case of falsely low LDL-C results in a man 69 years of age with type IIa dyslipidemia according to the Frederickson classification. The patient was admitted to the hospital for coronary heart disease and weight loss of 8 kg over a 3-week period. On admission, we performed a lipid evaluation on the Cobas Integra 800. The results revealed a discrepancy between the total cholesterol (TC; 9.18 mmol/L), and the sum of direct LDL-C (3.55 mmol/L) and HDL-C (0.27 mmol/L). Therefore, direct LDL-C and LDL-C calculated according to the Friedewald formula (8.31 mmol/L) did not match (see Fig. 1 in the Data Supplement that accompanies the online version of this Letter at http://www.clinchem.org/content/vol52/issue11). This discrepancy occurred consistently for 3 days. The patient then died of a myocardial infarction 3 days after admission.
To investigate the LDL-C measurement discrepancy, we performed various tests with the approval of our hospital ethics board.
We diluted the patient’s plasma in bovine serum albumin solution and found that this did not decrease the difference. We checked whether the error was the result of the TC or the HDL-C methods by performing the same measurements with various devices (Vitros 750®, Ortho Clinical Diagnostics; AU 600®, Olympus; RXL®, Dade Behring), and we obtained results similar to our own (see Fig. 1 in the online Data Supplement). TC results were comparable: 8.7 mmol/L with Vitros 750, 9.4 mmol/L with AU 600, and 9.1 mmol/L with RXL. HDL-C measured with the AU 600 was relatively high (0.93 mmol/L) compared with measurements made with other devices (0.23 mmol/L with Cobas Integra 800, 0.19 mmol/L with Vitros 750, and 0.21 mmol/L with electrophoresis) that may have been linked directly to the Daichi method.
Then, we performed lipoprotein electrophoresis (Hydragel 15 LIPO®, Sebia), which revealed an LDL-C fraction of 93.3% (7.16 mmol/L), indicating that LDL-C represented 93.3% of the lipoproteins quantified (LDL-C + HDL-C + VLDL-C + chylomicrons). We found no abnormal lipoproteins such as lipoprotein X or lipoprotein-a, which are reported to influence some homogeneous assays (5). We concluded that the Cobas Integra 800 LDL-C result was falsely low.
We found no reports in the medical literature or in data provided by Roche of interference occurring with the use of treatments the case patient had received, such as Fenofibrate 300®, Aldalix®, Verapamil®, Aspegic 250®, or Trinitrine®.
Several publications have reported a lack of specificity leading to false-positive results for intermediate (I) DL-C, rich apolipoprotein E HDL-cholesterol measured with the homogeneous LDL-C assays (1)(3)(6)(7), but we did not detect any abnormal lipoprotein fraction. Furthermore, laboratory results for our case patient showed serum TG at 1.33 mmol/L and total bilirubin at 50 µmol/L, values below the manufacturer-reported interference limits of 13.7 mmol/L for TG and 684 µmol/L for total bilirubin. Bile acids <200 µmol/L and lipoprotein X do not affect the direct LDL-C assay, and cholesterol associated with lipoprotein-a particles is not measured with this assay (7).
The only physiologic abnormality noted in our patient was cholestasis, indicated by an alkaline phosphatase concentration of 191 units/L (reference interval, 40–129 units/L) and a gamma glutamyl transferase (GGT) concentration of 568 units/L (reference interval, 8–61 units/L). There was also moderate liver cytolysis, with transaminases activity 3 times the reference value.
Because we did not find an obvious explanation for this interference, we reintroduced the Friedewald formula to screen for similar interferences in other patients. Eight cases were reported with transient falsely low or undetectable LDL-C, with reductions in LDL-C of 30%–100%. In each case, a period of cholestasis (defined as GGT >50 units/L and alkaline phosphatase >120 units/L) was noted. We considered the possible influence of cholestasis, as suggested by Fei et al. (6). We did observe that correlation coefficients for results obtained with the Friedewald formula (X) and direct analysis (Y) were lower (r2 = 0.73, Y = 0.986.X + 0.143; n = 5197) among patients with than among those without cholestasis (r2 = 0.96; Y = 0.803.X + 0.286; n = 593).
Homogeneous assays for LDL-C are commonly used, but these assays may not be suitable for all patients. Because we have, currently, no explanation for the observed discrepancies, further investigation is needed of the various causes of cholestasis and their potential implications in these falsely low LDL cholesterol results. Other abnormalities that must be considered include paraproteins, which were detected in 1 of our 8 study patients and have been reported as associated with artifactual undetectable HDL-C (8).
Acknowledgments
We thank Mrs. Hennache, Sebia® Laboratory.
References
-cyclodextrin sulfate. Clin Chem 1998;44:522-531.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
