| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Department of Pathology, Massachusetts General Hospital, 55 Fruit St., Boston, MA 02114;
aauthor for correspondence: fax 617-726-9206, e-mail asdighe{at}partners.org
Paraproteins can interfere in chemical measurements when they form precipitates during the testing procedure (1)(2)(3)(4)(5)(6)(7)(8)(9). Here we describe ways to identify paraproteins that interfere in methods for bilirubin and HDL on an automated analyzer. The approaches appeared to be effective in identifying these rare cases of interference and did not hinder the autovalidation of appropriately high or low values for either assay. Software analysis and reporting programs currently in use that rely on rigid two-point calculations derived from reaction monitors may be susceptible to errors in analysis.
We encountered an artifactually increased total bilirubin concentration and an artifactually low HDL in a patient with a monoclonal IgM paraprotein. The tests were initially performed on plasma samples with the Hitachi 917 automated analyzer (Roche Diagnostics) using the Roche total bilirubin assay and the Roche HDL-C Plus assay. For this patient, the reported total bilirubin was 318 mg/L, direct bilirubin was 2 mg/L, total protein was 127 g/L, and HDL was undetectable. Serum protein electrophoresis revealed a monoclonal IgM
M component at a concentration of 66 g/L. Similar results were seen on subsequent samples (both in plasma and serum), with a high total bilirubin and undetectable HDL. On a previous admission 2 years before, her total bilirubin concentration had been within the reference interval at 4 mg/L, and her HDL was measured as 390 mg/L by the same assay systems.
The Roche total bilirubin assay is a liquid, end-point, chromogenic assay based on a modification of the diazo method (10). The assay uses detergent as the accelerator and to help avoid protein precipitation. After solubilization with this reagent (reagent 1), the diazo reagent (reagent 2) is added to the cuvette to produce a pink color change that is proportional to the total bilirubin concentration and is monitored spectrophotometrically. Visually the patient sample demonstrating a high total bilirubin was nonicteric and showed no evidence of hemolysis or lipemia. To assess the possibility of interference from high concentrations of serum proteins, we diluted the patient sample two- and fourfold with normal saline. Dilution lowered the calculated concentration of total bilirubin but did not completely correct the error (fourfold dilution of the sample gave a calculated total bilirubin of 68 mg/L). On a different platform and methodology (Vitros 950; Ortho-Clinical Diagnostics), the total bilirubin was 2 mg/L and the HDL was 290 mg/L, in close agreement with the values for these analytes on testing 2 years earlier.
The reaction monitor for the total bilirubin assay as performed on the Hitachi 917 is shown in Fig. 1A
. A control sample with a high total bilirubin shows the typical reaction pattern with an early increase in absorbance readings after the addition of reagent 2. The serum from the patient with paraproteins showed a linear increase in absorbance late in the reaction, after the addition of the diazonium ion reagent. Visual inspection of the sample demonstrated that the increase in absorbance was attributable to the formation of a white insoluble precipitate and did not represent a color change. A similar reaction pattern has been described in two patients with myeloma (11). We attempted to determine how common this interference was in our population of patients with paraproteins by analyzing 15 patients with monoclonal immunoglobulin (5 IgM and 10 IgG) concentrations >40 g/L. None had an increased total bilirubin, indicating that even in this highly selected population this interference is uncommon.
|
To prospectively identify samples with the interference seen in the index case, we used open channels on the Hitachi 917 analyzer and manually programmed the analyzer to create a flag that was triggered when the difference between the absorbances at measurement points 12 and 14 (216 and 252 s, respectively) was >0.035. This flag was thus triggered when the absorbance increased late in the reaction, as occurs with patients with paraproteins. This flagging was appropriately triggered with numerous samples taken from our index patient. Examining the reaction plots presented in the report of Pantanowitz et al. (11), we believe that it is likely that this flag would have identified these patients as well. Importantly, the flag was not triggered when these parameters were used for >100 patients at our institution, including patients with bilirubin values as high as 350 mg/L. This suggests that the phenomenon is uncommon in the general hospital population and that "false-positive" flags are not being generated.
The assay for HDL used in this study was the Roche HDL-C Plus assay, a homogeneous colorimetric test. The reaction monitor for the HDL testing as run on the Hitachi 917 is shown in Fig. 1B
. As seen, the baseline absorbance reading for the sample from the index patient was markedly higher than the normal baseline reading, which is typically <0.100. The baseline reading occurs after the addition of reagent 1 of the Roche HDL-C assay to serum. Reagent 1 is a buffer containing magnesium sulfate and dextran sulfate that is designed to selectively form water-soluble complexes with LDL, VLDL, and chylomicrons, making them resistant to the subsequent enzymatic reaction. In the case of the index patient, the addition of the first reagent to the serum led to the rapid formation of an insoluble white precipitate that demonstrated a strong absorbance at 600 nm. On addition of reagent 2, the absorbance was reduced (presumably by dilution), and the enzymatic reaction proceeded. The analyzer calculated a HDL concentration of –110 mg/L, based on the baseline absorbance being higher than the final absorbance for the reaction.
We next attempted to determine how common this interference was by analyzing 15 patients with high concentrations of monoclonal immunoglobulins (5 IgM and 10 IgG; all with immunoglobulin values >40 g/L). Of these 15 patients, 2 had HDL concentrations <200 mg/L (HDL of 80 mg/L in a patient with 45 g/L of a monoclonal IgG and HDL of –80 mg/L in a patient with 54 g/L of a monoclonal IgM). These patients displayed reaction monitors similar to those for the index patient, demonstrating a highly increased baseline absorbance after the addition of reagent 1 (patient 2 in Fig. 1B
). In addition, when these samples were analyzed on an alternative platform and methodology, the Vitros 950 Clinical Chemistry System (Ortho-Clinical Diagnostics), this testing revealed total HDL concentrations of 350 and 460 mg/L, in close agreement with the values for HDL obtained in previous testing for these patients, at a time when the patients lacked paraproteins.
The lowering of HDL concentrations in the presence of paraproteins has been noted previously (12). However, there has been no solution for how to screen for these abnormalities before release of results. To prospectively identify these patients and prevent erroneous results from being reported to our clinicians, we have used open channels on the Hitachi 917 analyzer and manual programming to create a flag that is triggered when the net absorbance reading from measurement point 16 (288 s; immediately before the addition of reagent 2) is >1. Thus, this flag is triggered when an increase in absorbance occurs as a result of precipitate formation. The flag was not triggered when >100 patients lacking paraproteins were analyzed.
A high-volume laboratory has few options to prospectively screen for interference in common analytes such as HDL or bilirubin. For total bilirubin, a visual check for a lack of icterus in the sample or use of the icteric index may be helpful, but many laboratories are simply too busy to use these labor- and time-intensive techniques, especially for high-volume analytes. The existing Hitachi 917 analyzer software allows these potentially inaccurate results to be reported, and it is then left to the clinician to bring up the question of interference. This is unlikely to be an issue restricted to this platform because most automated analyzers use similar routines to calculate analyte concentrations. This may lead to an unnecessary workup for hemolysis or liver disease, the missing of a diagnosis of a paraprotein, or inaccurate assessment of cardiac risk in the case of HDL. We find that prospective use of reaction monitor flagging can be used to flag these cases before release to clinicians.
Although individual laboratories can create these solutions, the optimum solution rests with instrument manufacturers. Instrument manufacturers do not generally support efforts by end users to manually program custom flags into their instruments because there is concern that it may change the performance characteristics of the analyzer. It is hoped that reaction monitor analysis will become a feature of automated analyzers to allow these results to be prospectively flagged as abnormal without requiring individual laboratories to create their own interventions.
Acknowledgments
We thank Dr. Laura Olson of the Revere Health Center for alerting us to this potential interference, and the staff of the Massachusetts General Hospital clinical immunology and clinical chemistry laboratories for assistance.
References
from a myeloma patient. Clin Chem 1997;43:2435-2437.The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. R. Murali, A. Kratz, and K. E. Finberg Case 40-2006 -- A 64-Year-Old Man with Anemia and a Low Level of HDL Cholesterol N. Engl. J. Med., December 28, 2006; 355(26): 2772 - 2779. [Full Text] [PDF]
|
|
|
|
 |
|
 G. Dimeski and A. Carter Rare IgM Interference with Roche/Hitachi Modular Glucose and {gamma}-Glutamyltransferase Methods in Heparin Samples Clin. Chem., November 1, 2005; 51(11): 2202 - 2204. [Full Text] [PDF]
|
|
|
|
|
|
A. Nauti, A. Barassi, G. Merlini, and G. V. M. d'Eril Paraprotein Interference in an Assay of Conjugated Bilirubin Clin. Chem., June 1, 2005; 51(6): 1076 - 1077. [Full Text] [PDF]
|
|
|
|
|
|
C. A. Sheppard, R. C. Allen, G. E. Austin, A. N. Young, M. A. Ribeiro, and C. R. Fantz Paraprotein Interference in Automated Chemistry Analyzers Clin. Chem., June 1, 2005; 51(6): 1077 - 1078. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) had bilirubin measured as 333 mg/L, control patients [low total bilirubin (
) and high total bilirubin (
)] had bilirubin measured as 2 and 235 mg/L, respectively. The open rectangles indicate the points at which measurements are taken by the instrument to calculate results. (B), the index patient (