Clinical Biochemistry 47 (2014) 216–219

Contents lists available at ScienceDirect

Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem

Multisite evaluation of a monoclonal IMMULITE erythropoietin immunoassay William E. Owen a, Geralyn Lambert-Messerlian b, Cabrini Delaney c, Robert Christenson d, Bertrand Plouffe e, Rocio Ludewig e, Anne Woods e, Jyh-Dar Lei e, Stephan Thompson e, William L. Roberts a,f, Joely A. Straseski a,f,⁎ a

ARUP Institute for Clinical and Experimental Pathology®, Salt Lake City, UT, USA Women and Infants Hospital of Rhode Island, Providence, RI, USA Quest Diagnostics, Valencia, CA, USA d University of Maryland School of Medicine, Baltimore, MD, USA e Siemens Healthcare Diagnostics Inc., Tarrytown, NY, USA f Department of Pathology, University of Utah Health Sciences Center, Salt Lake City, UT, USA b c

a r t i c l e

i n f o

Article history: Received 19 September 2013 Received in revised form 14 November 2013 Accepted 17 November 2013 Available online 26 November 2013 Keywords: Method comparison Imprecision Interference Chemiluminescent immunoassay IMMULITE Serum erythropoietin Reference interval

a b s t r a c t Background: Erythropoietin (EPO) measurements are useful in diagnosing anemias and polycythemias. We conducted a multisite evaluation of a monoclonal IMMULITE® EPO immunoassay.1 Design and methods: The IMMULITE EPO assay is a solid-phase enzyme-labeled chemiluminescent immunometric assay. Method comparison to the Beckman ACCESS 2 assay using clinically characterized samples and reproducibility studies were conducted at three external independent laboratories. Internal evaluation conducted at Siemens included comparison of IMMULITE® 2000 and IMMULITE® 1000 assays to the ACCESS 2 assay; imprecision; linearity; limit of blank (LoB), limit of detection (LoD), and functional sensitivity; potential interference and cross-reactants; and reference interval determination. Results: External method comparison gave Deming regression of (IMMULITE 2000) = 0.96(ACCESS 2) + 2.57 IU/L, r = 0.98 (n = 217). Reproducibility ranged from 6.1% to 16.2%. Internal method comparisons gave Deming regressions of (IMMULITE 2000) = 1.09(ACCESS 2)− 3.51 IU/L, r = 0.98 and (IMMULITE 1000) = 0.95(ACCESS 2) + 0.52 IU/L, r = 0.95. Total imprecision ranged from 6.4% to 10.3% and linearity was confirmed from 3.5 to 562 IU/L. LoB, LoD, and functional sensitivity were 0.5, 1.0, and 1.5 IU/L, respectively. The assay was highly specific for EPO. Nonparametric reference interval was 4.3 to 29.0 IU/L (n = 170). Conclusions: The monoclonal IMMULITE EPO assay showed acceptable performance for EPO measurement. © 2013 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

Introduction Erythropoietin (EPO) is a 30 kDa highly glycosylated and acidic glycoprotein hormone comprised of 165 amino acids. EPO is primarily produced by the hepatocytes during fetal development, and postnatally by the peritubular cells of the renal cortex. EPO's primary role is to maintain red blood cell mass and hemoglobin at steady levels. Therefore, anemia and hypoxia are the main stimuli for the production of EPO. It binds to EPO receptors on the cell surface of the erythroid colonyforming unit and erythroid burst-forming units in the bone marrow to initiate erythropoiesis [1]. Anemia may occur if kidneys are damaged

Abbreviations: EPO, erythropoietin; LoB, limit of blank; LoD, limit of detection; CV, coefficient of variation. ⁎ Corresponding author at: c/o ARUP Laboratories, 500 Chipeta Way, Mail Code 115, Salt Lake City, UT 84108, USA. Fax: +1 801 584 5207. E-mail address: [email protected] (J.A. Straseski). 1 Not available for sale in the U.S. Product availability varies by country.

or unable to effectively respond to the demand for EPO, or if the patient's bone marrow stem cells do not respond to EPO stimulation. Hypoxia-induced expression of EPO has recently been attributed to activation of its promoter region by hypoxia-inducible-factor-2 [2]. Increased concentrations of EPO may be due to conditions of irondeficiency, megaloblastic or hemolytic anemias, myelodysplasia, chemotherapy, acquired immunodeficiency syndrome (AIDS), renal cell or adrenal carcinoma pheochromocytoma; while decreased concentrations may be associated with polycythemia vera and renal diseases [1,3]. Siemens has developed a monoclonal IMMULITE EPO assay using a one-step monoclonal–monoclonal sandwich format with a single 30 min incubation. The goal of this multisite study was to evaluate the overall analytical performance of the new IMMULITE EPO assay. Method comparison to the Beckman ACCESS 2 assay, reproducibility studies at three external independent laboratories, linearity, limit of blank (LoB), limit of detection (LoD), functional sensitivity, potential interference and cross-reactants, and reference interval determination for this immunoassay were all performed.

0009-9120/$ – see front matter © 2013 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.clinbiochem.2013.11.012

W.E. Owen et al. / Clinical Biochemistry 47 (2014) 216–219

Materials and methods The IMMULITE EPO assay (Siemens Healthcare Diagnostics, Tarrytown, NY) is a solid-phase, two-site, one-cycle chemiluminescent enzyme immunometric assay. It utilizes a murine monoclonal capture antibody and an alkaline phosphatase-conjugated murine monoclonal signal antibody, requires one 30 min incubation and has a wide analytic measurement range (1.0 to 750 IU/L). Units reported are IU/L traceable to WHO 2nd IRP 67/343 [4]. This is a modification of the older reagent formulation, which uses an alkaline phosphatase-conjugated goat polyclonal signal antibody, requires two 30 min incubations, and has a narrower analytic measurement range (1.0 to 200 IU/L). External evaluations included method comparison and reproducibility studies conducted at three independent laboratories: ARUP Laboratories, Salt Lake City, UT; Women and Infants Hospital of Rhode Island, Providence, RI; and Quest Diagnostics, Valencia, CA. All clinical samples were collected with approval of Institutional Review Boards at the respective sites. Studies were performed using IMMULITE 2000 instruments and two reagent kit lots (201 and 202). For method comparison studies, the Beckman ACCESS 2 EPO assay (Beckman Coulter) was used as the comparator method and clinically characterized patient samples were divided among the three sites and tested in singleton (n = 217, 44% female, 56% male, ages 19–97 years, median = 58 years). A variety of medical conditions characterized by anemia or polycythemia were included (Table 1), as well as subjects currently undergoing recombinant EPO therapy (darbepoetin or epoetin; n = 64). Reproducibility studies included three replicates each of seven identical patient serum pools tested at all three sites and a minimum of four replicates of three control levels (Siemens) assayed with two reagent lots in two assays per day for 10 days. The total within-device imprecision was calculated according to Clinical Laboratory and Standards Institute Guideline EP5-A2 [5] and included variance components from day to day, assay to assay, and within assay. The total reproducibility imprecision included additional components from site to site and from lot to lot. Internal studies were conducted at Siemens and included additional method comparisons of IMMULITE 2000 and IMMULITE 1000 instruments to the Beckman ACCESS 2 assay (ACCESS 2 testing performed at University of Maryland School of Medicine, Baltimore, MD). Samples (n = 186) were from a unique clinically characterized collection and were not part of the external method comparison study. The study population included patients with anemia and polycythemia vera, with a subset receiving recombinant EPO therapy. Correlation across IMMULITE

Table 1 Diagnostic profile of study subjects. Primary diagnosis Anemia Tumor-induced anemia Chemotherapy-induced anemia Multiple myeloma Lymphoma Chronic renal failure anemia Nonrenal: acute blood loss Nonrenal: hereditary disorder Nonrenal: iron or vitamin B12 deficiency Nonrenal: chronic blood loss Nonrenal: anemia of chronic disease Nonrenal: microcytic anemia Aplastic anemia AZT-induced anemia in HIV infection Anemia subtotal Polycythemia Polycythemia vera Polycythemia, unknown etiology Secondary polycythemia Polycythemia subtotal Unknown etiology Total

Number of samples 11 23 17 16 46 1 5 7 7 1 1 7 1 143 39 7 2 48 26 217

217

platforms (1000 and 2000) and multiple lots was also assessed. Samples were eliminated if there was insufficient volume or if results were outside either analytic measurement range (total n = 173 for IMMULITE 2000, n = 148 for IMMULITE 1000). An imprecision study conducted at Siemens used seven patient pools with EPO values spanning the assay range tested in duplicate two assays per day using three lots of reagents over 20 days and on two instruments. Additional performance studies conducted at Siemens included dilution linearity; LoB, LoD, and functional sensitivity; interference and cross-reactant studies; and reference interval determination. Linearity was assessed by diluting six endogenous disease-state patient samples with EPO values spanning the assay range. LoB was determined using 60 replicates of the zero calibrator and LoD used 80 replicates of each of five pools with concentrations near the expected LoD (0.6 IU/L); both were determined using three IMMULITE 2000 instruments. Functional sensitivity was estimated using the combined data of the imprecision and LoD studies described above and data were fitted to a power curve (Variance Function Program, version 10.0, AACB, Alexandria, NSW, Australia). Potential cross-reactivity and interferences were assessed by spiking fourteen tested substances into calibrator material (recombinant EPO antigen in bovine calf serum). The reference interval study included apparently healthy in-house (Siemens) donors with normal hematocrit values (n = 170, ages 23–72 years). A nonparametric interval was estimated using Excel 2003 software according to the Clinical Laboratory and Standards Institute Guideline C28-A2 [6]. Deming regression statistics were calculated using Analyze-It software version 2.22 (Analyze-it Software, Ltd., Leeds, United Kingdom). Results The external EPO method comparison between the IMMULITE 2000 and the ACCESS 2 indicated good correlation with a slope of 0.96 and r = 0.98 (Fig. 1A) and a slight negative average bias of−5.4% (Fig. 1B). Reproducibility results averaged over three testing sites showed withinassay and total imprecision ranging from 3.4% to 8.8% and 6.1% to 16.2%, respectively, over concentrations of 3.9 to 613 IU/L (Table 2). The internal method comparison to ACCESS 2 gave the following Deming regression statistics including 95% confidence intervals: (IMMULITE 2000) = 1.09 ± 0.07(ACCESS 2) − 3.51 ± 2.77 IU/L, Sy/x = 30.01, r = 0.98 (n = 173) and (IMMULITE 1000) = 0.95 ± 0.55(ACCESS 2) + 0.52 ± 13.40 IU/L, S y / x = 17.24, r = 0.95 (n = 148). The within-assay and total imprecision using seven patient pools ranged from 3.6% to 6.8% and 6.4% to 10.3%, respectively, over concentrations of 3.9 to 615 IU/L. Dilution linearity (range 3.5 to 562 IU/L) studies demonstrated a mean recovery of 100% (range 90% to 130%, Table 3). LoB and LoD results were 0.5 and 1.0 IU/L, respectively. Based on these data, a functional sensitivity claim of 0.8 IU/L was determined (Fig. 2). However, taking into account the data points at higher concentrations associated with total CVs of 20% or higher, functional sensitivity was more conservatively estimated as 1.5 IU/L. The cross-reactant study showed that substances listed in Table 4 had no effect on results at the concentrations tested, within the precision of the assay. No interference was observed from bilirubin, hemoglobin, and triglycerides up to concentrations of 0.20 g/L, 5.97 g/L, and 30.0 g/L, respectively (data not shown). The nonparametric reference interval representing the 2.5 and 97.5 percentiles of a healthy adult population with a normal hematocrit (n = 170) was 4.3 to 29.0 IU/L (median = 10.6 IU/L). Discussion The current study expands upon a previously published evaluation of the monoclonal IMMULITE EPO immunoassay [7] by including results from three different laboratories, two reagent lots, and samples from patients with a variety of clinical conditions. Overall, both studies correlated well and indicated consistent performance of the assay, as well as good correlation with the ACCESS 2 immunoassay.

W.E. Owen et al. / Clinical Biochemistry 47 (2014) 216–219

875

IMMULITE 2000 EPO (IU/L)

A 700

525

350

175

0 0

175

350

525

700

875

% Difference IMMULITE 2000 - ACCESS 2

218

100

B

75 50 25 0 -25 -50 -75 -100 0

175

ACCESS 2 EPO (IU/L)

350

525

700

875

Mean EPO (IU/L)

Fig. 1. Multisite EPO comparison study. The monoclonal IMMULITE 2000 EPO immunoassay was compared to the ACCESS 2 immunoassay. Panel A; 217 samples over the range 1.3 to 818.7 IU/L were analyzed at a total of three testing sites. Deming regression analysis of all data points gave a slope of 0.96 ± 0.06, an intercept of 2.57 ± 4.23, Sy/x = 36.9, and r = 0.98. Solid line indicates the Deming regression line, dashed line indicates the identity line (x = y). Panel B; solid line indicates the mean percent difference and dashed lines indicate 2 standard deviations from the mean in Bland–Altman analyses. The mean difference was −5.4% and 2 standard deviations were 43.3%.

Table 2 Reproducibility study results across three testing sites. Sample

Total number of replicates

Mean value (IU/L)

Within-assay CV (%)

Within-device CV (%)

Total CV (%)

Control 1 Control 2 Control 3 Pool 1 Pool 2 Pool 3 Pool 4 Pool 5 Pool 6 Pool 7

489 481 482 359 360 360 362 360 360 360

11.8 45.2 123.2 3.9 10.9 28 68 112 192 613

5.8 4.6 4.7 8.8 5.0 4.6 3.7 3.8 4.4 3.4

7.1 5.7 6.4 11.6 6.8 6.2 5.4 5.6 5.8 5.4

7.2 6.1 6.9 16.2 8.6 7.9 6.6 6.8 6.6 6.6

Table 3 Dilution linearity and recovery.

1

2

3

4

5

6

Dilution ratio

Observed concentrations (IU/L)

Expected concentrations (IU/L)

Recovery (%)

8 in 8 4 in 8 2 in 8 1 in 8 8 in 8 4 in 8 2 in 8 1 in 8 8 in 8 4 in 8 2 in 8 1 in 8 8 in 8 4 in 8 2 in 8 1 in 8 8 in 8 4 in 8 2 in 8 1 in 8 8 in 8 4 in 8 2 in 8 1 in 8

29.8 14.1 6.74 3.46 62.1 28.9 14.5 7.90 121 56.3 30.0 15.0 209 103 67.7 28.9 295 160 81.3 41.4 562 276 133 64.9

– 14.9 7.45 3.73 – 31.1 15.5 7.76 – 60.5 30.3 15.1 – 105 52.3 26.1 – 148 73.8 36.9 – 281 141 70.3

– 95% 90% 93% – 93% 94% 102% – 93% 99% 99% – 99% 130% 111% – 108% 110% 112% – 98% 95% 92%

In examining the Bland Altman plot, one high outlier (N 4 standard deviations) was observed. The results for both assays were quite low (ACCESS 2 = 2.36 IU/L and IMMULITE 2000 = 6.78 IU/L). This patient had a primary diagnosis of polycythemia vera with a normal hematocrit, thus the EPO results would be expected to be at or below the low end of the reference interval as was observed for this sample. Therefore, it is presumed that either EPO result would not have significantly changed the clinical evaluation of this patient. Additionally, higher imprecision was observed for both assays at these low concentrations. Measurements were also performed in singleton, therefore a random error could not be ruled out. It is noteworthy that those samples with negative biases exceeding −50%, with the exception of two, were all patients being treated with darbepoetin. These results were consistent with previously published recovery studies of darbepoetin-supplemented samples [7]. One of the two exceptions was a patient with a diagnosis of polycythemia vera, and results for both ACCESS 2 (2.75 IU/L) and IMMULITE 2000 (1.55 IU/L) were consistent with this diagnosis. Medication information was not available for the other patient, therefore the presence of darbepoetin could not be ruled out. Measuring concentrations of synthetic EPO preparations is not considered clinically necessary to monitor therapy. Their detection by commercially available EPO immunoassays varies significantly due to differences in structure and overall immunoreactivity. Similarly, using commercial EPO immunoassays as a screening test for suspected athletic doping is problematic [8]. Challenges to using immunoassays

Table 4 Cross-reactive compounds. Compound

mg/L added

Percent cross-reactivity

Human serum albumin α-1Antitrypsin α-Acid glycoprotein Human α-globulin Human transferrin (iron-saturated) Human transferrin (nonsaturated) Acetaminophen Acetylsalicylic acid Ibuprofen α-2 Macroglobulin Gamma globulin EPO receptor Recombinant human thrombopoietin Rheumatoid factor

35,000 5000 1400 50,000 4000 4000 1000 1000 2000 3750 500 0.05 0.05 1.22

ND ND ND ND ND ND ND ND ND ND ND ND ND ND

ND = nondetectable.

W.E. Owen et al. / Clinical Biochemistry 47 (2014) 216–219

35%

results for bilirubin, hemoglobin, and triglycerides also indicated good resistance to the effects of common interferents. The reference interval was slightly higher than that determined in a previous study (3.3–23.4 IU/L, n = 129, ages 21–67 years) [7]. The reason for this difference is unknown. However, unlike the present study, subjects in this earlier study were not screened for anemia which would be expected to cause higher values if present. Our results do confirm that each laboratory should establish its own reference interval.

30% 25%

% CV

219

20% 15%

Conclusions 10% 5% 0% 0.1

0.8 1

10

100

1000

EPO IU/L Fig. 2. EPO Functional Sensitivity Study Results. Data from the imprecision and LoD studies performed at Siemens are displayed. The x-axis shows the EPO concentrations tested and the y-axis shows the imprecision (%CV). The plot is a power curve fitted to the data points. The concentration (0.8 IU/L) which coincides with an imprecision of 20% CV from the power curve is indicated.

designed for serum and plasma for screening in these cases include the small doses of synthetic EPO used in doping (often below the assay's limit of detection), the exorbitant number of EPO preparations that can be obtained worldwide, its rapid clearance from the body and overall low recovery by these assays. Further studies would be required to determine the utility of an assay such as IMMULITE EPO in detecting therapeutic or illicit EPO preparations. The internal method comparison study indicated consistent performance across the IMMULITE family of instruments, though there was a minor statistically significant difference in slope between the IMMULITE 2000 results and those of the IMMULITE 1000. This may have been due to differences in assay calibrations or the patient sample numbers used for each platform, both of which would have affected the range of the confidence intervals. Results for the external reproducibility and internal imprecision studies showed good performance and were consistent with earlier IMMULITE EPO immunoassay performance studies [7,9,10]. The internal dilution linearity study results also indicated the assay was linear over a wide dynamic range. The LoB, LoD, and functional sensitivity were all well below the reference interval determined for the assay (4.3–29.0 IU/L), which is important in the diagnosis of polycythemia vera [11–14]. The cross-reactant study indicated the assay was very specific for EPO. The interference

EPO measurement is useful for the diagnosis of anemias and polycythemias. The monoclonal IMMULITE EPO immunoassay was evaluated across multiple testing sites, reagent lots and instrument systems and showed good performance for all characteristics evaluated. It is suitable for routine EPO measurement on the IMMULITE 1000 and IMMULITE 2000 systems. References [1] Marsden JT. Erythropoietin—measurement and clinical applications. Ann Clin Biochem 2006;43:97–104. [2] Jelkmann W. Regulation of erythropoietin production. J Physiol 2011;589:1251–8. [3] Pagana KD, Pagana TJ. Mosby's Manual of Diagnostic and Laboratory Tests. 3rd ed. St. Louis: Mosby Elsevier; 2006. [4] Annable L, Cotes PM, Mussett MV. The second international reference preparation of erythropoietin, human, urinary, for bioassay. Bull World Health Organ 1972;47:99–112. [5] CLSI. Evaluation of Precision Performance of Quantitative Measurement Methods; Approved Guideline. CLSI Document EP5-A2. Wayne, PA: Clinical Laboratory and Standards Institute; 2004. [6] CLSI. Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline—Second Edition. CLSI Document C28-A2. Wayne, PA: Clinical Laboratory and Standards Institute; 2000. [7] Owen WE, Roberts WL. Performance characteristics of a new Immulite® 2000 system erythropoietin assay. Clin Chim Acta 2011;412:480–2. [8] Reichel C. Recent developments in doping testing for erythropoietin. Anal Bioanal Chem 2011;401:463–81. [9] Benson EW, Hardy R, Chaffin C, Robinson CA, Konrad RJ. New automated chemiluminescent assay for erythropoietin. J Clin Lab Anal 2000;14:271–3. [10] Owen WE, Roberts WL. Performance characteristics of the IMMULITE 2000 erythropoietin assay. Clin Chim Acta 2004;340:213–7. [11] Pearson TC, Messinezy M. The diagnostic criteria of polycythaemia rubra vera. Leuk Lymphoma 1996;22(Suppl. 1):87–93. [12] Johansson P, Safai-Kutti S, Lindstedt G, Suurkula M, Kutti J. Red cell mass, spleen size and plasma erythropoietin in polycythaemia vera and apparent polycythaemia. Acta Haematol 2002;108:1–7. [13] Mossuz P, Girodon F, Donnard M, et al. Diagnostic value of serum erythropoietin level in patients with absolute erythrocytosis. Haematologica 2004;89:1194–8. [14] Michiels JJ, Bernema Z, Van Bockstaele D, De Raeve H, Schroyens W. Current diagnostic criteria for the chronic myeloproliferative disorders (MPD) essential thrombocythemia (ET), polycythemia vera (PV) and chronic idiopathic myelofibrosis (CIMF). Pathol Biol (Paris) 2007;55:92–104.