Resolving discordant samples in clinical laboratory practice.

The growing importance of the clinical laboratory in patient healthcare is evident from the fact that about 70% of medical decisions are now based on laboratory test results. (1) The role of the clinical laboratory, aided by advances in analytical science and increased automation, has become even more prominent, as cost pressures have reduced patient stays in the hospital. Laboratory results--normal or abnormal--help screening, diagnosis, indication of therapeutic pathways, and monitoring of therapy. But a variety of causes, pre-analytical or method-dependent, may lead to inaccurate "discordant" results; such results may lead to incorrect diagnosis and therapy, risking increased morbidity or mortality. (2-3) What causes lead to false-positive or false-negative results, especially in immunoassays and chemistry tests?

Errors in a test result may belong to three classes: random, systematic, and sporadic.

* Random error causes poor precision; however, automation and improvement in analytical science are advancing precision.

* Systematic errors are caused by incorrect calibration or differences in standardization of the test. Often, when a patient moves from one hospital to another, some lab results change because different methods may be used in those labs. Government regulatory authorities and international professional organizations have worked together to produce reference and survey materials, thus reducing the systematic error.

* The sporadic error, however, which occurs in some specimens but not others, is the largest cause of discordant results.

Causes of discordant results

What are the causes that produce discordant results in immunoassays? Using highly specific antibodies and binding proteins, these assays can measure very small concentrations (as little as femtomolar or [10.sup.-15] M level) of clinically important analytes in a specimen. Erroneous immunoassay results may be caused by two major sources: antibody specificity and assay interferents. While antibodies are very specific, they may recognize other molecules having a molecular "signature" similar to the analyte. Depending on the assay architecture, such molecules, called cross-reactants, may increase or decrease the assay signal, thereby falsely increasing or lowering the test results. (4-6) Such cross-reactants could be metabolites of the analyte with varying amounts of physiological activity, (7-8) a coadministered drug (9-10) or nutrition aid, or other endogenous component of the sample. Ideally, the cross-reactivity of the metabolites in an immunoassay should parallel their physiological activities. (11) An example of this concept is a chemiluminecsent immunoassay (CLIA) for the cardiac drug, digoxin, presented in Table 1. (12-13) The effect of cross-reactivity depends on the cross-reactant and analyte concentrations in the specimen, as shown in Figure 1. (14) Examples of endogenous cross-reactants include similar classes of compounds, like steroids and gonadotrophins, interfering in one another's assays. Thus, human chorionic gonadotrophin (hCG) often falsely increases results of other hormones with similar structures (e.g., leutrophin, LH, or thyroidstimulating hormone, TSH) in some assays. (15-16) Since hCG concentrations greatly increase during pregnancy, even a very small cross-reactivity of the assay antibodies to hCG results in large error.

Another example is the interference of endogenous digoxin-like immunoreactive factors (DLIF) in digoxin immunoassays. (17) DLIF concentration increases in certain groups of patients, like neonates, pregnant women, and patients with liver and renal diseases. (18) While it is easier to find out if a coadministered drug generates a discordant result, it is much more difficult to find out if the interferent is one or more components of a herbal medicine (e.g., Chinese medicines like ginseng, Chan Su, and Dan Shen). (19-20) Hemolysis, icterus, and turbidity from lipids are common sources of assay interference, and most kit manufacturers list them in the kit insert; however, conjugated bilirubin in serum specimens was reported to interfere in a fluorescence polarization immunoassay (FPIA) for vancomycin. (21) Imaging agents also have been reported to interfere in assays. Thus, radiological imaging agents, containing [.sup.99m]Tc or [.sup.131]I, have been reported to interfere in radioimmunoassays; and fluorescent dyes used in severe burn cases interfered in fluoroimmunoassays and FPIA. (22-23)

Forms and formats contribute to discordance 

Another source of assay discordance is the heterogeneity of the analyte, especially for the peptide analytes. Additionally, the analyte may be present in the specimen as free, as well as bound to other proteins. Thus, the prostate-cancer marker, prostate-specific antigen (PSA), stays in serum as both a free form and bound to serum proteins like [a.sub.1]-anti chymotrypsin and [a.sub.2]-macroglobulin, (24) or the cardiac marker troponin-I (cTnI) may exist as free or complexed with troponins C and T. (25) If two assays recognize these forms differently, they will report dissimilar results for the same sample. (26) Table 2 shows that five commercial cTnI assays have different responses to the free vs. complexed (with troponin C) forms of the analyte. (27) Prolactin exists in serum as a monomer, dimer, trimer, and polymer. Since these forms react differently to different antibodies, discordance has been observed among different prolactin immunoassays. (28)

Some discordant results may arise from the particular immunoassay format used. For example, in a two-site immunometric assay, which uses two separate antibodies to the analyte to form a "sandwich" complex, the concentration of the complex is directly proportional to that of the analyte. In such an assay, if the antigen concentration is very high, separate analyte molecules may bind the two separate antibodies, thus reducing the complex formation, and thereby falsely lowering the test results. Such examples of "hook" or "prozone" effects are found in assays measuring a wide range of the analytes, (e.g., hCG and PSA); based on reagent formulations, the hook analyte concentrations are different in different assays. (29-30)

Other culprits in discordance

Some pre-analytical factors also cause discordant results. Most important among them is the type of specimen-collection tube used, and how the specimen is stored. Thus, if the analyte's or assay's reagents are affected by the various additives found in sample-collection tubes, different results will be obtained for the same patient depending on the type of collection tube. An example is cTnI assays, where [Ca.sup.2+] ions are needed to maintain the complexed forms of the analyte. Since anticoagulants like EDTA or heparin bind [Ca.sup.2+] ions, some cTnI assays show different results for serum and different types of plasma. (31) If an assay uses an enzyme to generate the signal, (e.g., alkaline phosphatase or horse-radish peroxidase), the presence of the similar enzymes or substances that affect the enzyme activities in the sample or reagents may affect the reported result. (32)

Some analytes are very sensitive to storage conditions. Thus, plasma samples for ammonia measurement must be analyzed as soon as possible after collection and should be kept refrigerated until analysis, minimizing atmospheric exposure. (33) Free PSA and parathyroid hormone are also unstable and must be stored refrigerated. (34-35) While automated analyzers have greatly improved lab productivity and increased the quality of results, they also bring some system-specific issues interfering with assay results (e.g., sample or reagent probe carryover, goodness of fit with the calibration curve algorithm, and reaction cuvette carryover). Assay manufacturers, however, take utmost steps to minimize such interferences.

One of the most common assay interferences found in the clinical lab is from antibodies in the specimen. While most such interference is from endogenous antibody (the most common being the 'heterophilic' antibody and autoantibody), therapeutic antibody may also cause assay interference. An example of the latter is Digibind, which consists of Fabfragments of digoxin antibody, administered to neutralize digitalis toxicity. Digibind, especially in high concentration in a sample, causes false-positive interference in many digoxin immunoassays. Ultrafiltration is a simple experiment to resolve such discordance; the specimen is ultrafiltered through a membrane with a molecular weight cutoff range of 10 kDa. Most protein interferents, as well as those which are highly bound to proteins (e.g., DLIF), are retained while the relatively less-protein-bound analyte, digoxin, collects in the filtrate and is analyzed. (36)

Autoantibodies are antibodies that exist in a patient, binding preferentially to the analyte of choice (or components of the reagents used in the assay). Such interference is often found in thyroid assays. (37) Autoantibodies to some assay reagent components (e.g., antiavidin or antienzyme) may yield incorrect results for reagents using avidin or those enzymes. (38-39) Rheumatoid factors, another type of autoantibody, also can interfere in immunoassays. (40)

Heterophilic antibody interference is a major source of discordant results reported in the literature. These antibodies interact with the assay antibodies (raised mostly in mice, goat, sheep, or rabbits) used in the reagents. The heterophilic antibodies may form spontaneously or in response to close contact with the animals or antibodies used earlier in their therapy or diagnosis. In most of the affected immunoassays, heterophilic antibodies interact with the two analyte-specific reagent antibodies, forming a "pseudo-complex" that falsely raises the test result. (41) Human antimouse antibody (HAMA) interference is a special case of heterophilic antibody interference, where the HAMA interacts with murine antibodies. Assay manufacturers often add agents to block heterophilic antibodies; derived primarily from nonspecific animal antibodies, the blockers bind to the sample's interfering antibodies, neutralizing them before they can interact with the analyte-specific antibodies. (42) This, however, is not a foolproof method, because the blockers can neutralize only up to certain concentrations of interfering antibodies. If the specimen has a very high concentration of the heterophilic antibodies, these can saturate out the blocking reagent and then affect the assay.

[FIGURE 1 OMITTED]

Resolving discordant results

How can discordant samples be identified among the large number of samples that a typical laboratory processes? While it is easier to detect such discordance with abnormal results (because they are flagged by the auto analyzers), the discordant result that falls in the normal range is the most difficult to detect. This is where the skill and experience of the laboratorian helps. Patient history also helps: Does an analyte result match with other results from the lab? Does the patient chart alert of any discordant results found earlier? Is the patient taking any medication that is known to interfere with some assays? Of course, spotting a discordant result is easier if two different assays with the same specimen report very different results (e.g., when method comparison is done in evaluating a new assay). Once identified, a discordant sample should be confirmed by repeating the assay(s) to rule out instrument error. If possible, a third "referee" assay may be run to learn which result is correct. Dilution studies often confirm and resolve discordance. If heterophilic interference is suspected as the cause of discordance, the sample may be incubated with blocking reagents and repeated in the assay.

While laboratory tests are increasingly used to aid earlier and faster diagnosis--in turn, improving healthcare and reducing costs--laboratorians must constantly be on watch to detect and resolve discordant results.

                                                  Cross-reactivity, %
Metabolite                     Cardioactivity, %  RRA         CLIA

Digoxigenin bis-digitoxoside   77                 51.8        90
Digoxigenin mono-digitoxoside  66                 47.3        68
Digoxigenin                     4-21              11          <1
Dihydrodigoxin                  2-6               <1          <1

Table 1. Cross-reactivity of the major digoxin metabolites in a
commercial chemiluminescent immunoassay (CLIA) and an experimental
radio-receptor assay (RRA) for digoxin. The RRA used the actual digoxin
receptor from human heart muscles (instead of an antibody). The CLIA
used an antibody for digoxin, screened to match the metabolite
cross-reactivity. The CLIA does have metabolite cross-reactivity
comparable to the metabolities' cardioactivity and RRA.

cTnI,       Ratio of observed cTnI, complex/free
[micro]g/L  CLIA (1)  FIA (1)  FIA (2)  FIA (3)  CLIA (2)

 0          NA        NA       NA       NA       NA
 5          1.00       4.43    1.25     2.07     1.23
10          0.89       5.36    1.80     2.02     1.07
20          1.08       6.24    1.36     3.14     1.22
40          1.44      11.58    1.20     2.75     1.35

Table 2. The isoforms of cardiac troponin-I (free and complexed) have
different responses in several commercial cTnI assays (CLIA,
chemiluminescent immunoassay; FIA, fluoroimmunoassay). Thus, if a
specimen contained both forms of cTnI, it would produce discordant
results among the assays studied. The complexed form of cTnI used in
this study was that with troponin-C.

References

1. Forsman R. Communicating the Laboratory's Value to Health Care [interview]. Clin Lab News. 2004;30(7):12.

2. Kroll, E. Interference with clinical laboratory analyses. Clin Chem. 1994;40:1996.

3. Cole LA, Shahabi S, Butler S, et al. Utility of commonly used commercial human chorionic gonadotropin immunoassays in the diagnosis and management of trophoblastic diseases. Clin Chem. 2001;47:308-315.

4. Valdes R, Miller JJ. Increasing the specificity of immunoassays. J Clin Immunoassay. 1992;15:87-96.

5. Jortani SA, Miller JJ, Helm RA, Johnson NA, Valdes R. Suppression of immunoassay results by crossreactivity. J. Clin Ligand Assay. 1997;20:177-179.

6. Dasgupta A, Saffer H, Wells A, Datta P. Bidirectional (Positive/Negative) Interference of Spironolactone, Canrenone and Potassium Canreonate on Serum Digoxin Measurement: Elimination of Interference by Measuring Free Digoxin or Using a Chemiluminescent Assay for Digoxin. J Clin Lab Analysis. 2002; 16:172-177.

7. Sutfin TA, Jusco WJ. Compendium of active drug metabolites. In: Wilkinson GR, Rawlins MD, eds. Drug metabolism and disposition: considerations in clinical pharmacology. Lancaster, UK: MTP Press;1985:91-159.

8. Datta P, Graves S. Patients with Suspicious Digoxin Levels: Evaluation of the Potential Contribution of Cross-reactive Digoxin Metabolites. Clin Biochem. 1998;31:51-54.

9. Datta P. Oxaprozin and 5-(p-hydroxyphenyl)-5-phenylhydantoin Interference in Phenytoin Immunoassays. Clin Chem. 1997;43:1468-1469.10.11.12.

10. Datta P, Dasgupta A. A New Turbidometric Digoxin Immunoassay on the ADVIA 1650 Analyzer Is Free from Interference by Spironolactone, Potassium Canrenoate, and Their Common Metabolite Canrenone Ther Drug Monit. 2003;25:47811.

11. Gault MH, ed. Drug metabolites and immunoassays. Clin Chem. 1994;40:1863-1864.12.13.

12. Datta P, Larsen F. Specificity of Digoxin Immunoassays toward Digoxin Metabolites. Clin Chem. 1994;40:1348-1349.13.

13. Miller J, Straub RW, Valdes R. Digoxin immunoassays with cross-reactivity of digoxin metabolites proportional to their biological activity. Clin Chem. 1994;40:1898-1903.

14. Datta P, Dasgupta A, Bi-directional (Positive/negative) interference in a digoxin immunoassay: importance of antibody specificity. Ther Drug Monit. 1998;20:352-357.

15. Taylor AE, Khoury RH. Interferences in immunometric assays for gonadotropins. J Clin Ligand Assay 1997;20:190-199.16.

16. Gow SM, Kellett HA, Seth J, et al. Limitations of new thyroid function tests in pregnancy. Clin Chim Acta. 1985;152:325-333.17.

17. Datta P, Xu L, Malik S, et al. Effect of Antibody Specificity on Results of Selected Digoxin Immunoassays among Various Clinical Groups. Clin Chem. 1996;42:373-379.

18. Datta P, Hinz V, Klee G. Comparison of Four Digoxin Immunoassays with Respect to Interference from Digoxin-like Immunoreactive Factors, Clin Biochem. 1996;29:541-547.

19. Dasgupta A, Wu S, Actor J, Olsen M, Wells A, Datta P. Effect of Asian and Siberian ginseng on serum digoxin measurement by five digoxin immunoassays. Significant variation in digoxin-like immunoreactivity among commercial ginsengs. Amer J Clin Pathol. 2003;119:298-303.

20. Datta P, Dasgupta A. Effect of Chinese Medicines Chan Su and Danshen in EMIT 2000 and Randox Digoxin immunoassays: Wide variation in Digoxin-Like immunoreactivity and magnitude of interference in Digoxin measurement by different brands of the same product. Ther Drug Monit. 2002;24:637-644.

21. Wood FL, Earl JW, Nath C, Coakley JC. Falsely low vancomycin results using the Abbott TDx. Ann Clin Biochem. 2000;37:411-413.

22. Bloom JN, Herman DC, Elin RJ, et al. Intravenous fluorescein interference with clinical laboratory tests. Am J Ophthalmol. 1989;108:375-379.

23. Leal T, Dupret P, Hassoun A, Wallemacq PE. Topical application of eosin to burns produces interference in measurement of serum vancomycin by fluorescence polarization immunoassay. Clin Chem. 1997;43:1288-1289.

24. Lilja H, Christensson A, Dahlen U, et al. Prostate-specific antigen in serum occurs predominantly in complex with [a.sub.1]-antichymotrypsin. Clin Chem. 1991;37:1618-1625.

25. Wu AHB, Feng Y-J, Morre R, et al. Characterization of cardiac troponin subunit release into serum after acute myocardial infarction and comparison of assays for troponin T and I. Clin Chem. 1998;44:1198-1208.

26. Zhou AM, Tewari PC, Bluestein B, et al. Multiple forms of prostate-specific antigen in serum: differences in immunorecognition by monoclonal and polyclonal assays. Clin Chem. 1993;39:2483-2491.

27. Datta P, Foster K, Dasgupta A. Response to free and complex human cardiac troponin-I compared in five commercial troponin-I assays. Clin Chem. 1999;45:2266-2269.

28. Ismail AAA, Walker PL, Fahis-Wilson MN, et al. Prolactin and macroprolactin: a case report of hyperprolactinaemia highlighting the interpretation of discrepant results. Ann Clin Biochem. 2003;40:298.

29. Khosravi MS. Shifting the hook effect in one-step immunometric assays. Clin Chem. 1990;36:169-170.

30. Datta P, Dasgupta A. The evaluation of an automated chemiluminescent immunoassay for complexed PSA on the Bayer ACS:180O System. J Clin Lab Analysis. 2003;17:174-178.

31. Gerhardt W, Nordin G, Herbert A-K, et al. Troponin T and I assays show decreased concentrations in heparin plasma compared with serum: lower recoveries in early than in late phases of myocardial injury. Clin Chem. 2000;46:817-821.

32. Dasgupta A, Chow L, Wells A, Datta P. Effect of elevated concentration of alkaline phosphatase on cardiac troponin I assays. J Clin Lab Analysis. 2001;15:175-177.

33. Burtis CA, Ashwood ER, eds. Tietz Testbook of Clinical Chemistry, 2d edition. Philadelphia, PA: WB Saunders Company; 1994:1487-1488.

34. Sokoll LJ, Bruzek DJ, Dua R, et al. Short term stability of the molecular forms of PSA and effect on percent free PSA (%fPSA) and percent complexed PSA (%cPSA). (Abstract). Clin Chem. 2001;47:A139.

35. Teal TK, Wood JL, Stevens PE, Lamb EJ. Stability of bio-intact (1-84) parathyroid hormone ex vivo in serum and EDTA plasma from hemodialysis patients. (Letter) Clin Chem. 2004;50:1711.

36. Datta P. A blocker reagent to reduce Digibind interference in a non-pretreatment digoxin assay [abstract]. Ther Drug Monit 1995;17:407.

37. Despres N, Grant AM. Antibody interference in thyroid assays: a potential for clinical misinformation. Clin Chem. 1998;44:440-454.

38. Banfi G, Pontillo M, Sidoli A, et al. Interference from antiavidin antibodies in thyroid testing in a woman with multiendocrine neoplasia syndrome type 2B. J Clin Ligand Assay. 1995;18:248.

39. Levinson SS. Test interferences from endogenous antibodies. J Clin Ligand Assay. 1997;20:180.

40. Dasgupta A, Banerjee SK, Datta P. False positive troponin I in the MEIA assay due to presence of rheumatoid factors in serum: elimination of this interference by using a polyclonal antisera against rheumatoid factors. Am J Clin Pathol. 1999;112:753-756.

41. Kricka LJ. Human anti-animal antibody interferences in immunological assays [review]. Clin Chem. 1999;45:942-956.

42. Reinsberg J. Different efficacy of various blocking reagents to eliminate interferences by human antimouse antibodies with a two-site immunoassay. Clin Biochem. 1996;29:145-148.

By Pradip Datta, PhD, DABCC 

Pradip Datta, PhD, DABCC, is a senior staff scientist in the diagnostics division of Bayer HealthCare in Tarrytown, NY.