Cyanocobalamin

Context.-Comparison of different analytical methods in proficiency surveys may be affected by the artificial nature of the survey material.

Objective.-To compare intermethod differences in proficiency survey results between 2 types of survey material, conventional proficiency testing material (PTM) and fresh frozen human serum (FFS), for 3 markers of anemia: ferritin, folate, and vitamin B12.

Design.-Data were gathered from a 2003 survey event in the College of American Pathologists Ligand ("K") Series, in which the specimens to be tested by each participating laboratory included 1 vial of FFS and 2 vials of PTM with different analyte concentrations. The more than 1600 laboratories subscribing to the survey were not advised as to the nature of the specimens.

Main Outcome Measures.-The bias of each method relative to the median of method means for each analyte and each type of survey material, and the interlaboratory coefficient of variation for each method.

Results.-For each of the 3 analytes, moderate to large method biases were observed. For ferritin, method biases correlated strongly between comparable PTM and FFS specimens (Spearman r = 0.863, P

Conclusions.-With ferritin, proficiency survey performance of PTM is similar to that of FFS, implying that method biases relate mainly to calibration. With folate and to a lesser extent with B12, PTM and FFS exhibit different method biases, implying that the biases reflect analyte heterogeneity and/or matrix effects.

(Arch Pathol Lab Med. 2005;129:323-327)

External proficiency surveys, as currently practiced, primarily serve the purpose of testing comparability of results among laboratories performing the same test by the same method. For this purpose it is desirable, but not essential, that the distributed specimens mimic actual patient specimens, that is, that the proficiency testing material (PTM) be commutable with native human material for the surveyed methods. Owing to concerns of cost, infection hazard, stability, and range of analyte concentration, clinical chemistry surveys generally do not distribute native human material, but rather materials derived from human and/or animal plasma but modified in various ways. A more ambitious goal for proficiency surveys is to document the absolute accuracy of test results from different laboratories using varied methods. It is natural, in reviewing survey data, to draw conclusions regarding the bias or interlaboratory imprecision of particular methods. However, the differences between PTM and native human material make any such analysis suspect.

The degree of commutability between PTM for chemistry surveys and native human serum or plasma has not been well characterized for most analytes. To address this problem, the College of American Pathologists recently included samples of fresh frozen human serum (FFS) along with traditional PTM in 2 survey events in clinical chemistry in 2003. The surveys included many general chemistry and ligand analytes, several of which are discussed in other articles. The present study addresses 3 markers of anemia: ferritin, folate, and vitamin B12. Survey data were analyzed to determine the degree to which intramethod and intermethod result variability observed with PTM matched that of the more authentic but expensive FFS.

MATERIALS AND METHODS

Survey Materials

The preparation and distribution of the FFS survey material, and of the PTM material used in the chemistry surveys (C surveys), is described in detail elsewhere.1 Proficiency testing material used in the Ligand surveys (K surveys) is prepared in a generally similar fashion to the C-series PTM (in brief, pooled plasma is defibrinated, dialyzed to remove anticoagulants, and spiked with the various analytes measured in the surveys). The ferritin used to spike the K-series material is prepared from human spleen; the folate is N^sup 5^-methyltetrahydrofolic acid, disodium salt; and the vitamin B12 is cyanocobalamin. Although additions of these analytes are made to meet certain targets, actual "weighed-in" values are not available.

Survey Data

The proficiency survey data for the present study were gathered from the K-A testing event in 2003, in which 1 specimen of FFS (K-02) and 2 specimens of PTM (K-01 and K-03) were sent to approximately 1600 participating laboratories. For ferritin, results were also gathered from the C-C Survey event (approximately 1400 participants) in 2003, which included 1 specimen of FFS (C-02, from the same pool as the K Survey) and 1 specimen of PTM (C-01, distinct from those in the K Survey).

Statistical Analysis

The results submitted by survey participants were first screened for outliers using the 2-pass, 3-SD test.2 Testing results that were greater than 3 SD from their peer group mean on the first and second pass were removed. Participant results from peer groups with fewer than 10 laboratories or other methods not specified were also excluded from the analysis. Method bias was calculated as the percent difference of the peer group mean from the median of all peer group means. Method imprecision was determined by calculating the coefficient of variation (CV) for each method and all methods combined. Finally, a nonparametric Spearman rank correlation on peer group means between FFS and comparable PTM was performed to determine the likely matrix effects. Precision values were compared between FFS and PTM using a nonparametric, paired Wilcoxon rank-sum test. P

RESULTS

Survey data are summarized in Tables 1 through 3. For each of the 3 analytes and each of 3 survey specimens, there were substantial differences among peer group means and method imprecision (CV).

Ferritin

Results for ferritin in the 2003 K-A Survey event were submitted from a total of 1603 laboratories, with 13 methods being represented by at least 10 laboratories (range, 11-390 laboratories) (Table 1). The all-method means for the FFS specimen and the 2 PTM specimens K-01 and K-03 were, respectively, 32.2, 24.6, and 546 ng/mL (32.2, 24.6, and 546 µg/L), making the K-01 PTM specimen more suitable for comparison to FFS. A scatterplot of method biases with respect to the median of method means for FFS and K-01 is shown in Figure 1. The observed correlation was strong (Spearman r = 0.863, P

The peculiar behavior of the Olympus method is difficult to explain. The raw data were reviewed to confirm that it was not a problem of data processing. There could possibly have been some change by the manufacturer in the several-month interval between the K- and C-survey events, there could be differences between the laboratories subscribing to the different surveys, or it may simply be a statistical fluke related to the relatively small number of laboratories using the Olympus method.

Folate

Results were submitted from a total of 1233 laboratories, with 12 methods being represented by at least 10 laboratories (range, 10-317 laboratories) (Table 2). The all-method means for FFS, K-01, and K-03 were 11.8,1.01, and 10.2 ng/mL (26.7, 2.3, and 23.1 nmol/L), respectively, making the K-03 PTM specimen more suitable for comparison with FFS. Method bias for FFS showed virtually no correlation with that for PTM (Figure 2; r = -0.224, P = .48), and the dispersion of results was lower with FFS (all-method CV = 17.9% for FFS, 37.2% for K-03). The Vitros ECI method had the highest bias (106%) for K-03, compared to a bias of -12% for FFS; even if this method were excluded, the correlation between FFS and PTM biases was still statistically insignificant, and dispersion remained greater with PTM. Within-method SDs and CVs, however, were both lower overall for PTM compared to FFS (P = .002 and P = .008, respectively).

Vitamin B12

Results were submitted from a total of 1394 laboratories, with 10 methods being represented by at least 10 laboratories (range, 10-411 laboratories). The all-method means for FFS, K-01, and K-03 were 471, 201, and 518 pg/ mL (348, 148, and 382 pmol/L), respectively, making the K-03 PTM specimen more suitable for comparison with FFS. The correlation between method biases for FFS and PTM was marginally significant (Figure 3; r = 0.55, P = .049). The all-method CV was similar between FFS (12%) and PTM (13%), and within-method CVs were generally similar but were lower overall with PTM (P = .02).

COMMENT

Ferritin, folate, and vitamin B12 are all important analytes in the diagnosis of anemia, but are of course very different types of chemical analytes. Ferritin is a protein known to have diverse forms (isoferritins) that may react differently in different immunoassays.1 The introduction of standard ferritin preparations appears to have reduced but not eliminated variability in commercial assays.4-6 Folate and vitamin B12, in contrast, are small molecules but nevertheless are also heterogeneous.7-10 Folate is a generic term for a group of compounds related to pteroic acid, including dihydrofolic acid, tetrahydrofolic acid, N5-methyltetrahydrofolic acid (the principal form in human serum), and forms containing up to 7 residues of glutamic acid.10 Vitamin B12 is similarly a generic term, comprising several molecules in the family of cobalamins, such as cyanocobalamin, methylcobalamin, deoxyadenosyl cobalamin, and aquocobalamin.10 These forms may react differently in different assays. Other complications that may relate to interassay variability for folate and B12 are their binding to plasma proteins and, especially in the case of vitamin B12, their low concentration.

For all 3 of these analytes, variability in commercial assays has been documented over the years, and no reference method is generally accepted. Proficiency surveys have a potential role in comparing the performance of different commercial assays, but questions arise concerning biases introduced by the artificial nature of the samples. For example, so that accuracy over a range of analyte concentrations can be tested, PTM is spiked with purified preparations of the analytes. Given analyte heterogeneity, these may not perfectly mimic the biological analyte (this is probably particularly true in the case of folate). Also, handling and dialysis of pooled serum may alter concentrations of ligands that affect protein binding of substances such as folate and B12, and may alter the binding proteins themselves, possibly affecting analytical methods to different degrees. When the analyte is itself a protein, such as ferritin, it is subject to degradation or possibly subtle changes that could alter binding to the antibody reagents used in commercial assays.

We found that commercial assays for ferritin gave substantially different peer group mean results for both PTM (range of means for 13 methods, 19.9-31.0 ng/mL [19.9-31.0 µg/L]; median of peer group means, 24.1 ng/mL [24.1 µg/L]) and FFS (range, 24.7-47.3 ng/mL [24.7-47.3 µg/L]; median of peer group means, 33.2 ng/mL [33.2 µg/L]). If these differences simply reflect method calibration, one would expect the method biases for FFS to correlate well with those for PTM. If, on the other hand, the nature of the PTM introduces method biases that may not apply to native patient material (as explained in the previous paragraph), then the method biases for the 2 materials could be very different. In fact, the method biases correlated closely, and intramethod CVs were similar for the 2 materials. It therefore appears that for ferritin, neither bias nor intramethod dispersion of results observed in proficiency surveys can be attributed to the artificial nature of PTM.

Folate also exhibited substantial method biases for both PTM and FFS. However, unlike ferritin, the method biases for folate correlated very poorly between PTM and FFS. Hence, it is likely that the pure folate used to spike PTM differs substantially from the distribution of species found in native human serum, although differences in the protein matrix could also account for some of the variability. It is noteworthy that with folate the intramethod CVs were lower with PTM than with FFS. This observation might again reflect different chemical forms of folate in the different materials, better stability in the case of PTM, or some other matrix effect.

Vitamin B12 was an intermediate case, with moderate method biases that exhibited a borderline significant correlation between PTM and FFS, similar all-method CVs, and modestly lower intramethod CVs with PTM.

This study demonstrates that the effect of using artificial material instead of native human material in proficiency testing of serum components is very much analyte dependent. Although use of native serum generally allows more valid analysis of method biases, in many specific cases its regular use may offer no advantage or only a minimal advantage not worth the increased cost.

References

1. Miller WC, Myers GL, Ashwood ER, et al. Creatinine measurement: state of the art in accuracy and interlaboratory harmonization. Arch Pathol Lab Med. 2005;129:297-304.

2. Barneft V, LevvisT. Outliers in Statistical Data. Chichester, NY: Wiley & Sons; 1994.

3. van Suijlen JD, van Noord PC, Leijnse B. Accuracy of serum territin determinations in tissue preparations and human serum. J Clin Chem Clin Biochem. 1990;28:43-48.

4. Thorpe SJ, Walker D, Arosio P, Heath A, Cook JD, Worwood M. International collaborative study to evaluate a recombinant L territin preparation as an International Standard. Clin Chem. 1997;43:1 582-1587.

5. Lotz J, Hafner C, Prellwitz W. Reference values for a homogeneous ferritin assay and traceability to the 3rd International Recombinant Standard for Ferritin (NIBSC code 94/572). Clin Chem Lab Med. 1999;37:821-825.

6. Stacy DL, Han P. Serum ferritin measurement and the degree of agreement using 4 techniques. Am I Clin Pathoi 1992;98:511-515.

7. Gunter EW, Bowman BA, Caudill SP, Twite DB, Adams MJ, Sampson EJ. Results of an international round robin for serum and whole-blood folate. Clin Chem. 1996:42:1689-1694.

8. Lee DS, Griffiths BW. Human serum vitamin B12 assay methods: a review. CUn Biochem. 1985:18:261-266.

9. Pfeiffer CM, Fazili Z, McCoy L, Zhang M, Gunter EW. Determination of folate vitamers in human serum by stable-isotope-dilution tandem mass spectrometry and comparison with radioassay and microbiologie assay. Clin Chem. 2004:50:423-432.

10. Fairbanks VF, Klee GG. Biochemical aspects of hematology. In: Burtis CA, Ashwood ER, eds. Tietz Textbook of Clinical Chemistry. 3rd ed. Philadelphia, Pa: WB Saunders; 1999:1642-1710.

Jay L. Bock, MD, PhD; David B. Endres, PhD; Ronald J. Elin, MD, PhD; Edward Wang, PhD; Bruce Rosenzweig, PhD; George C. Klee, MD, PhD

Accepted for publication September 20, 2004.

From the Department of Pathology, Stony Brook University, Stony Brook, NY (Dr Bock); Department of Pathology and Laboratory Medicine, Keck School of Medicine, University of Southern California, Los Angeles (Dr Endres); Department of Pathology and Laboratory Medicine, University of Louisville, Louisville, Ky (Dr Elin); College of American Pathologists, Northfield, III (Dr Wang); Diagnostic and Molecular Medicine Health Care Croup, Long Beach VA Medical Center, Long Beach, Calif (Dr Rosenzweig); and the Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, Minn (Dr Klee).

Presented in part at the 2004 Annual Meeting of the American Association for Clinical Chemistry, Los Angeles, Calif, ]uly 29, 2004.

Dr Klee declares that he has received research grants from Becton Coulter and Biosite for work unrelated to the preparation of this manuscript. All other authors also have no relevant financial interest in the products or companies described in this article.

Reprints: lay L. Bock, MD, PhD, Department of Pathology, Stony Brook University, Stony Brook, NY 11794-7300 (e-mail: jbock@notes. cc.sunysb.edu).

Copyright College of American Pathologists Mar 2005
Provided by ProQuest Information and Learning Company. All rights Reserved