Clin Chem Lab Med 2015; 53(1): 125–132

Tracy Teodoro-Morrison, Marcel J.W. Janssen, Jasper Mols, Ben H.E. Hendrickx, Mathieu H. Velmans, Johannes Lotz, Karl Lackner, Lieselotte Lennartz, David Armbruster, Gregory Maine and Paul M. Yip*

Evaluation of a next generation direct whole blood enzymatic assay for hemoglobin A1c on the ARCHITECT c8000 chemistry system Abstract Background: The utility of HbA1c for the diagnosis of type 2 diabetes requires an accurate, precise and robust test measurement system. Currently, immunoassay and HPLC are the most popular methods for HbA1c quantification, noting however the limitations associated with some platforms, such as imprecision or interference from common hemoglobin variants. Abbott Diagnostics has introduced a fully automated direct enzymatic method for the quantification of HbA1c from whole blood on the ARCHITECT chemistry system. Methods: Here we completed a method evaluation of the ARCHITECT HbA1c enzymatic assay for imprecision, accuracy, method comparison, interference from hemoglobin variants and specimen stability. This was completed at three independent clinical laboratories in North America and Europe. Results: The total imprecision ranged from 0.5% to 2.2% CV with low and high level control materials. Around the diagnostic cut-off of 48 mmol/mol, the total imprecision was 0.6% CV. Mean bias using reference samples from IFCC and CAP ranged from –1.1 to 1.0 mmol/mol. The enzymatic assay also showed excellent agreement with HPLC *Corresponding author: Paul M. Yip, Department of Laboratory Medicine and Pathobiology, University Health Network and University of Toronto, Toronto, Ontario, Canada, E-mail: [email protected] Tracy Teodoro-Morrison: Department of Laboratory Medicine and Pathobiology, University Health Network and University of Toronto, Toronto, Ontario, Canada Marcel J.W. Janssen, Jasper Mols, Ben H.E. Hendrickx and Mathieu H. Velmans: Viecuri Medical Center, Laboratory of Clinical Chemistry and Haematology, Venlo, The Netherlands Johannes Lotz and Karl Lackner: Laboratory Medicine, University Medical Center, Johannes Gutenberg University Mainz, Mainz, Germany Lieselotte Lennartz: European Scientific Affairs, Abbott GmbH, Wiesbaden, Germany David Armbruster and Gregory Maine: Global Scientific Affairs, Abbott Laboratories, Abbott Park, Illinois, USA

methods, with slopes of 1.01 and correlation coefficients ranging from 0.984 to 0.996 compared to Menarini Adams HA-8160, Bio-Rad Variant II and Variant II Turbo instruments. Finally, no significant effect was observed for erythrocyte sedimentation or interference from common hemoglobin variants in patient samples containing heterozygous HbS, HbC, HbD, HbE, and up to 10% HbF. Conclusions: The ARCHITECT enzymatic assay for HbA1c is a robust and fully automated method that meets the performance requirements to support the diagnosis of type 2 diabetes. Keywords: biological variation; diabetes; enzymatic assay; HbA1c; method evaluation. DOI 10.1515/cclm-2014-0310 Received March 28, 2014; accepted June 16, 2014; previously published online August 7, 2014

Introduction Diabetes mellitus is a growing epidemic that affects millions of people worldwide. The role of the laboratory in the diagnosis and management of both type 1 and type 2 diabetes includes the measurement of two main biomarkers: glucose and HbA1c. Glucose is traditionally measured for the diagnosis of diabetes from a fasting, random, or glucose-challenged blood sample and provides a snapshot of the patient’s immediate glycemic status. In contrast, hemoglobin A1c (HbA1c) is the preferred marker for monitoring long-term glycemic control of diabetes patients. Treatment targets for diabetes management are based on HbA1c levels, where the risk of complications from poorly controlled or uncontrolled diabetes can be reduced by achieving   ≤  53 mmol/mol HbA1c for most patients [1, 2]. The most recent Canadian, American and German Diabetes Associations published clinical practice guidelines recommend the use of HbA1c as one of the

Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 5/17/15 11:24 AM

126      Teodoro-Morrison et al.: Enzymatic HbA1c assay on ARCHITECT c8000 diagnostic criteria of type 2 diabetes, in addition to the traditional diagnostic tests relying on glucose [1, 2]. A value of   ≥  48 mmol/mol in asymptomatic individuals supports the diagnosis of type 2 diabetes. HbA1c testing offers several advantages over traditional glucose-based tests since it better reflects long-term glycemic status, has less biological variation and is stable after collection. There are over 150 methods available on the market which quantify HbA1c based on either of two principles: charge differences and structural differences [3]. The former principle is used in ion-exchange high-performance liquid chromatography (HPLC) and electrophoretic assays. The latter principle is used in immunoassays, enzymatic assays and boronate affinity chromatography. Currently, immunoassay methods followed by HPLC are the most widely used by laboratories in the US from a 2013 survey from the College of American Pathologists (CAP) [4]. Similarly, immunoassays followed by HPLC are the most commonly used methods in Germany [5]. Enzymatic HbA1c assays used in the clinical setting are relatively new and recent reports suggest this methodology to be accurate and precise [6, 7]. The principle of analysis begins with lysis of the whole blood sample and proteolytic digestion of glycated and non-glycated HbA1c. The resulting N-terminal fructosyl dipeptides are measured spectrophotometrically by the action of fructosyl dipeptide oxidase (FPOX) enzyme [7]. In vivo, glucose covalently binds the N-terminal valine of the HbA1c β-chain, which is then rearranged to a fructosyl moiety (Figure 1A). Therefore, the FPOX enzyme directly recognizes glycated HbA1c as a fructosyl dipeptide. The Abbott ARCHITECT assay has a total run-time of 10 min and is fully-automated. Quantification consists of two separate measurements: glycated dipeptide and total hemoglobin (Figure 1B), and the absorbance values are used to determine the percent HbA1c (NGSP units) or the HbA1c fraction in mmol/mol (IFCC units). In this study, we evaluated the performance characteristics of the next generation HbA1c assay on the Abbott ARCHITECT chemistry system in three clinical laboratories. Method comparison studies were performed against three HPLC systems, as well as imprecision, accuracy, effect of erythrocyte sedimentation and freezing, as well as interference from common hemoglobin variants.

Materials and methods Study design and instrumentation Three clinical laboratories participated in this method evaluation study of the Abbott ARCHITECT HbA1c enzymatic assay (Abbott Park,

IL, List #4P52): Center A at the Laboratory of Clinical Chemistry and Haematology, Viecuri Medical Center (Venlo, The Netherlands); Center B at the Institute for Laboratory Medicine, University Medical Center Johannes Gutenberg University (Mainz, Germany); and Center C at the Department of Clinical Biochemistry, Toronto General Hospital, University Health Network (Toronto, Canada). The routine HbA1c method at each center was used for comparison to the ARCHITECT enzymatic assay, including the Adams HA-8160 (Menarini Diagnostics, Firenze, Italy) at Center A, the Variant II (Bio-Rad Laboratories, Hercules, CA, USA) at Center B and the Variant II Turbo 2.0 (Bio-Rad Laboratories) at Center C. The Abbott ARCHITECT enzymatic assay was performed on either the c8000 (Centers A and C) or ci8200 (Center B) chemistry analyzers at each center. Briefly, red blood cells are aspirated and lysed on board the analyzer. Next, total hemoglobin is oxidized to methemoglobin and quantified photometrically at 476 nm. Following protease digestion, FPOX generates hydrogen peroxide coupled to peroxidase conversion of a chromogen that is measured at 660 nm. ARCHITECT HbA1c calibrators are traceable to the IFCC reference method for HbA1c and are certified by the NGSP program to ensure alignment with the Diabetes Control and Complications Trial (DCCT). Results were originally expressed in mmol/mol from instrumentation at centers A and B. Results from Center C were produced in NGSP units (%HbA1c) and converted to IFCC units (mmol/mol) by the master equation: NGSP(%) = 0.0915 × IFCC(mmol/mol)+2.15 [8–10]. All routine methods are also NGSP certified and traceable to the IFCC reference method.

Imprecision The assay imprecision was determined at Centers A and B following the Clinical Laboratory Standards Institute EP05-A2, based on 20 replicate of the Bio-Rad Lyphochek whole blood Diabetes control material as well as human whole blood samples [11]. The samples were tested in duplicate in two runs, separated by a minimum of 2 h and run over 5 days. Between runs, samples were stored at 2–8 °C. Assay imprecision was also estimated from patient samples used for method comparison and assayed in duplicate on the enzymatic assay at all centers. The standard error of the estimate (SEE) was determined from replicates of method comparison patient samples (see below) and used as a measure of reproducibility.

Method comparison Fresh surplus EDTA whole blood samples retrieved after routine measurement were used during the study at each center: Center A, n = 129, range 24–126 mmol/mol; Center B, n = 123, range 15–121 mmol/mol; Center C, n = 123, range 23–116 mmol/mol. At all centers, patient samples were assayed in duplicate on the ARCHITECT enzymatic assay.

Accuracy and linearity Accuracy and linearity were assessed by measurement of samples with values assigned by reference laboratories. At Centers A and B, accuracy testing included eight frozen whole blood samples from the European Reference Laboratory for Glycohemoglobin (ERL) with

Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 5/17/15 11:24 AM

Teodoro-Morrison et al.: Enzymatic HbA1c assay on ARCHITECT c8000      127

A OH H OH OH

+

-His-Val(NH2)

CHO H OH H

-His-Val-N

N

H C

N H

H C H

H

OH H OH OH CH2OH + H2O H OH H

-His-Val-N

(Glucose)

CH2OH

(Schiff base)

H

O H OH OH

(Fructosyl-HbA1c)

CH2OH OH H

H

B Protease

FPOX H2O2

-His-Val-Fruc

Peroxidase

Azido-Met-Hb (476 nm)

Chromogen (660 nm) Figure 1 Schematic representation of the generation of HbA1c in vivo and the enzymatic reaction for measurement of HbA1c by the Abbott ARCHITECT direct method. (A) Addition of glucose to hemoglobin occurs non-enzymatically in vivo, and rearranges to form a fructosyl moiety covalently attached to the N-terminal valine of the hemoglobin β-chain. (B) The ARCHITECT direct enzymatic reaction in vitro quantifies fructosyl-Hb by the selective action of fructosyl dipeptide oxidase (FPOX) on the glycated dipeptide His-Val-fructosyl, and is expressed relative to total hemoglobin. HbA1c values determined by the IFCC reference method. At Center C, fresh whole blood from the CAP 2013 LN15-A survey was used for calibration verification and linearity with assigned values determined from NGSP laboratory results and known admixture ratios. Samples were tested in duplicate at all sites.

as control where at each time point two aliquots of the sample were either re-suspended or allowed to settle prior to HbA1c measurement (performed at Center B).

Interference from hemoglobin variants Fresh and frozen patient sample comparison A subgroup of 20 whole blood samples from the method comparison study pool were collected and quantified for HbA1c at Center B in Mainz on the enzymatic system. Specimens were collected and aliquoted. One aliquot was assayed immediately and the second aliquot was immediately frozen at –20 °C and thawed 7 days later for reassessment on identical instrumentation. Statistical analysis included t-test, Wilcoxon test, and Passing-Bablok analysis for the comparison of fresh and frozen specimens.

Erythrocyte sedimentation At Center A whole blood patient samples (n = 6) were mixed thoroughly by inversion, immediately followed by pipetting 1 mL aliquots into seven EDTA-containing collection tubes to mimic routine patient sampling. One aliquot was immediately tested on the ARCHITECT analyzer while the other aliquots remained undisturbed on a rack at room temperature. The remaining aliquots were loaded gently without mixing one at a time and spaced by 1 h intervals over a course of 6 h. Each aliquot was tested in duplicate. One patient sample served

Patient samples containing heterozygous HbS, HbC, HbD or HbE variants were acquired from the NGSP reference laboratory (Diabetes Diagnostic Laboratory, University of Missouri School of Medicine, Columbia, MO, USA) (n = 15–25) and analyzed at Center C. Additionally, samples containing varying concentrations of HbF were similarly obtained (n = 25).

Statistical analysis All calculations were determined using either EP Evaluator software version 7.0 (David G. Rhodes Associates, Kennett Square, PA, USA), Analyse-It (Analyse-it Software, Leeds, UK) or Microsoft Excel version 14.0 (Microsoft, Redmond, WA, USA).

Results This study evaluated the analytical performance of the Abbott ARCHITECT enzymatic HbA1c assay independently

Brought to you by | New York University Bobst Library Technical Services Authenticated Download Date | 5/17/15 11:24 AM

128      Teodoro-Morrison et al.: Enzymatic HbA1c assay on ARCHITECT c8000 at three clinical laboratories. The participating laboratories and comparator instruments at each center are described in the Materials and methods. Several performance characteristics were examined including imprecision, linearity, accuracy, method comparison and interference with common hemoglobin variants. Imprecision was assessed using commercial quality control (QC) material and patient samples. The detailed within-run and between-run imprecision results are summarized in Table 1. Imprecision of the Bio-Rad Lyphochek whole blood diabetes control tended to be higher in the low QC range of 27–29 mmol/mol than at the high QC levels of 75–80 mmol/mol with the total imprecision was 0.5%–0.6% CV at two centers (A and B). Patient samples showed similar imprecision at middle and high HbA1c levels. Around the diagnostic cut-off for type 2 diabetes (48 mmol/mol) imprecision was   0.98, slopes of 1.01, and y-intercepts of –4.3–3.3 mmol/mol. We also assessed the effect of sample stability by freezing the specimen for 1 week at –20 °C. There was no significant difference in HbA1c quantification with freezing. From Passing-Bablok analysis, the comparison of fresh (x-axis) and frozen (y-axis) samples on the enzymatic assay was y = 1.00x+0.02, with a relative mean bias of –0.2%. Also, a comparison of results from fresh and frozen specimens showed no significant difference using t-test and Wilcoxon test analysis (p = 0.571). The ARCHITECT HbA1c method is fully-automated, involving the direct sampling of whole blood with an on-board probe. To test the effect of erythrocyte sedimentation prior to loading, specimens were aliquoted and initially measured followed by measurements over time (total 6  h) without prior mixing, allowing erythrocytes to settle. Six patient samples ranging from 32 to 53 mmol/mol were used and imprecision of all time points for each sample ranged from 0.5% to 1.0% CV. Comparing all results to baseline at time point 0 for a given sample, mean biases of 0.4–0.7 mmol/mol and relative biases of 0.9%–1.9% were obtained. Finally, comparison of one sample handled without mixing to the same sample with mixing at each time point produced a mean bias of –0.3 mmol/mol. Assessment of potential interference from common hemoglobin variants is shown in Table 3. From 20 to 25 heterozygous patient samples, results showed similar HbA1c measurement from the ARCHITECT enzymatic assay compared to NGSP values, with slopes of 0.90–0.98, correlation coefficients ranging from 0.98 to 1.00, and mean biases of –1.3 to –2.6 mmol/mol. HbF samples initially tested here ranged from 5% to 28% content. Only when HbF content was restricted to