- Email: [email protected]
Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform
CLB-08701; No. of pages: 5; 4C: Clinical Biochemistry xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Short Communication
Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform Isolde Seiden-Long a,b,⁎, Kareena Schnabl c, Wendy Skoropadyk c, Nola Lennon a, Arlayne McKeage a a b c
Calgary Laboratory Services, Calgary, AB, Canada Department of Pathology and Laboratory Medicine, University of Calgary, Canada Department of Laboratory Medicine and Pathology, University of Alberta, Canada
a r t i c l e
i n f o
Article history: Received 4 February 2014 Received in revised form 21 April 2014 Accepted 22 April 2014 Available online xxxx Keywords: Method evaluation Automated analyzers Application development Ammonia Enzymatic assays
a b s t r a c t Objective: Adaptation of the Randox Enzymatic Manual UV Ammonia method to be used on the Roche Cobas 6000 (c501) automated analyzer platform. Design and methods: The Randox ammonia reagent was evaluated for precision, linearity, accuracy and interference from hemolysis, icterus and lipemia on the Roche c501 analyzer. Comparison studies were conducted for the Randox reagent between Roche c501, Siemens Vista, Ortho Vitros 250, and Beckman DxC methods. Results: The Randox reagent demonstrates acceptable within-run (L1 = 65 μmol/L, CV 3.4% L2 = 168 μmol/L, CV 1.9%) and between-run precision (L1 = 29 μmol/L, CV 7.3% L2 = 102 μmol/L, CV 3.0%), Analytical Measurement Range (7–940 μmol/L), and accuracy. The method interference profile is superior for the Randox method (hemolysis index up to 600, icteric index up to 60, lipemic index up to 100) as compared to the Roche method (hemolysis index up to 200, icteric index up to 10, lipemic index up to 50). Comparison was very good between the Randox reagent and two other wet chemistry platforms. Conclusions: The Randox Enzymatic Manual UV Ammonia reagent is an available alternative to the Roche Cobas c501 reagent. The method is more robust to endogenous interferences and less prone to instrument error flags, thus allowing the majority of clinical specimens to be reported without additional sample handling at our institution. © 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Introduction Ammonia testing is performed in the clinical lab for two distinct patient populations. In the adult population, ammonia levels can be elevated during liver failure or impairment, urinary tract infection, gastrointestinal bacterial overgrowth, due to various medications e.g. valproic acid, chemotherapy, and in-patients on total parenteral nutrition. The lack of clinical specificity for elevated ammonia levels makes it a poor marker for diagnosis of hepatic encephalopathy in the emergency department [1]. However, persistent levels in patients with cirrhosis/hepatic failure can provide useful prognostic information [2]. In the pediatric population, blood ammonia levels can be increased in inherited defects of the urea cycle, organic acidurias, disorders of fatty acid oxidation, other illness in babies e.g. sepsis, asphyxia, Reye's syndrome and transient hyperammonemia of the newborn (normally seen within the first 24 h of life). Due to the underlying clinical causes
⁎ Corresponding author at: Department of Pathology and Laboratory Medicine, University of Calgary, Foothills Medical Centre, 1403-29 St. NW (C618B), T2N 2T9, Canada. Fax: +1 403 944 1364. E-mail address: [email protected] (I. Seiden-Long).
of ammonia elevation in these populations, the clinical lab often receives lipemic or icteric samples for ammonia measurements. In addition, neonate samples are often hemolyzed. In our own facility, our lab receives approximately 25% of our specimens for ammonia testing either icteric, lipemic or hemolyzed. Due to the pre-analytical instability of ammonia levels in plasma samples, it is not feasible to centrifuge samples to remove lipemia without heating the specimen and causing unacceptable pre-analytical increase in ammonia levels. The only intervention that can be used to reduce interferent levels in an acceptable time frame is sample dilution. Consequently, an ammonia method must be relatively robust to these endogenous interferents to reliably report test results in the clinical diagnostic laboratory. Most commercial chemistry analyzers currently on the market use an enzymatic ammonia method in which glutamate dehydrogenase catalyzes the reaction of ammonium with α‐ketoglutarate and reduced nicotinic adenine dinucleotide (phosphate) (NAD(P)H) to form glutamate and NAD(P)+ and water. Absorbance decrease in [NADH] or [NADPH] can then be measured at a wavelength of 340 nm [3,4]. Assay reaction:
GLDH
þ
þ
NH3 þ α‐ketoglutarate þ NADPH → glutamate þ NADP þ H :
http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022 0009-9120/© 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
2
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
Out of 3849 labs reporting ammonia on the College of American Pathologists 2012 Chemistry C proficiency survey, 3259 were reporting a glutamate dehydrogenase enzymatic method and 556 by a thin film dry chemistry method. Dry-slide analyzers use a micro-diffusion apparatus to liberate free ammonia from the sample by alkalization. The ammonia is then measured colorimetrically by reflectance change of color of an indicator dye such as bromophenol blue. Our laboratory had been experiencing high frequencies of sample error flags (approximately 25% of all samples) using the Roche ammonia reagents on the Cobas c501 since the instrument was implemented in 2011. The instrument flags using the c501 reagent are often encountered in specimens with elevated levels of icterus and lipemia. Interferents in patient specimens increase background absorbance in the reaction which is not alleviated by subtracting the secondary absorbance reading at 700 nm. For the Cobas c501, this results in increased reagent background blank to a level which exceeds the absorbance maximums specified for the instrument detector once reagent 2 is added to the reaction (Fig. 1A–D). This sensitivity of the Roche method to endogenous interferents very common in ammonia blood samples is largely due to increasing reagent absorbance background in aging lots of the Cobas reagent. This results in instrument over-range absorbance flags even in samples which otherwise would have not received other interference flags (Fig. 1A, B). Roche has recently issued a technical bulletin advising customers to allow the reagent to rest open on the bench for 24 h to release ammonia build up prior to loading on the instruments [5]. This process was implemented in our laboratory with little effect on reducing instrument error flag rates. We describe here the adaptation of a third party reagent produced by Randox Inc. which our laboratory has used on the Roche Cobas c501 platform. The method principle utilizes glutamate dehydrogenase, similar to the Roche reagent, but stabilizers are added to the substrate which keep the background absorbance of the reagent well within the mid-range on the c501 detector limits making the method much more robust to endogenous interfering substances in human plasma samples.
Cassette filling and application parameters The following application parameters have been adapted from those provided by Randox.
Multi cassette filling for 150 tests/kit Bottle a — R1, 150 tests, 15 mL use volume, 20 mL fill volume Bottle b — not used Bottle c — R3, 150 tests, 3.8 mL use volume, 6.4 mL fill volume Instrument application — Roche c501 Test/type Assay/Rxn time/meas point Wave (2nd/Pri) Sample volume (norm.) (Dec.) (Inc.) Diluent Reagent Vol. R1 Reagent Vol. R2 Reagent Vol. R3 Linearity limit Prozone limit Abs. limit Cell detergent Stirring level Calibrator Calibration method Point/Span/Weight SD Limit Duplicate Limit Sensitivity Limit S1 ABS Limit Change over (to cassette) Autocalibration timeout Cassette Calib./timeout/days Range Unit Report name Data mode Technical limit H I L
AMM–PLASMA 2 point end — 10–36–70 700–340 8–0–0 4–0–0 16–0–0 Water 100 0 25 0 — decrease 0–0–0–0 0–0–0–0–0–0 Detergent 1 3 Linear 2–2–0 999 2–30 –3.6 – –2.7 –32000–32000 2 point Yes 2 point/blank/3 μmol/L NH3R Active 7–940 600 60 100
Materials and methods Reagents and samples Randox Enzymatic Manual UV Ammonia reagent was purchased from ESBE Scientific (Markham,ON, Canada). Reagents were prepared as per manufacturer's instructions. Ammonia calibrators and Roche Cobas 6000 c501 ammonia reagents and Multipack cassettes were purchased from Roche Diagnostics (Laval, QC, Canada). All samples were analyzed on a Roche Cobas c501 automated analyzer platform. Liquicheck Ethanol/Ammonia control materials were purchased from Bio-Rad Laboratories Canada Limited. Intralipid (20% emulsion) and unconjugated bilirubin were purchased from SigmaAldrich (St. Louis, MO,USA). Conjugated di-taurobilirubin was purchased from Millipore Corporation (MA, USA). Patient plasma specimens used for the purposes of this validation were collected as part of routine patient care at our institution. Use of these samples for the purposes of method evaluation is classified as a quality assurance/program evaluation activity to be used for assessment, management and/or improvement and is granted quality assurance exemption from ethics review requirement by the Conjoint Health Research Ethics Board at the University of Calgary. Briefly, K + EDTA vacutainer tubes (Becton Dickinson) are pre-chilled prior to collection and samples delivered to the lab on ice. Samples are centrifuged without delay at 4 °C ± 2 °C, aliquoted, and placed on ice. Plasma was stored at 4 °C for up to 1 h and analyzed or frozen for long term storage.
Parameters for onboard reagent stability and calibration were determined during an extended between run precision study using a failure of our QC rule set for ammonia (13S/22S/R4S/41S) to trigger a calibration event. Onboard reagent stability was verified up to 30 days. Blank calibration required every 3 days. Two point calibration was required every 21 days.
Method evaluation Comparison samples run within-site were run on Roche and Randox reagents simultaneously to minimize pre-analytical increase in ammonia. The cross-site multiple instrument comparison samples were frozen at −80 °C, distributed to 4 locations and analyzed by standardized sample handling protocol on the same day at all 4 locations with timed thaw-to-analyze intervals to allow comparability of results. For Beckman DxC, Siemens Vista and Ortho Vitros 250 methods, the reference locations utilized the manufacturer's reagent and calibrators for the comparison study. Within-run (N = 10 replicates) and betweenrun precision (N = 30 replicates) experiments were conducted using quality control materials at clinically relevant ammonia levels. Linearity experiments were conducted using high and low ammonia pools in human plasma matrix spanning the analytical measurement range (AMR) of the assay measured N = 5 times at each level. Lower limit of quantitation (LLQ) was conducted by serial dilution and defined as
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
3
Fig. 1. Reaction trace for typical patient specimens analyzed for ammonia level on Cobas 6000. The patient specimen is added at the same time as Reagent 1 in both reactions. A and B) Sample which is mildly icteric and lipemic and the instrument can provide a numeric result for Randox reagent (B), but NAbs flag for Roche reagent (A). C and D) sample exceeds maximum Icteric index of 10 for the Roche method but is still reportable by the Randox method (Icteric max 60). E) Reaction curve for Randox reagent for high specimen. Reaction rate is slower with the Randox reagent than the Roche reagent, but complete for the last 4 reading points of the 10 minute application.
the lowest measured analyte level achieving a 20% CV. Accuracy was verified using proficiency survey material from the College of American Pathologists (Survey C-A 2013). Interference studies for lipemia and icterus were conducted by spiking an EDTA plasma pool with intralipid or conjugated or unconjugated bilirubin stock solution. The hemolysis interference experiment was conducted as described by Dimeski [6], with hemolysate made fresh prior to testing to distinguish true hemolysis interference from spurious elevations due to long term storage of hemolysate. There is no biological variation data available for ammonia measurement, thus criteria for significant interference were based on assay precision (b 50 μmol/L ± 20%, N = 50 μmol/L ± 10%). Data was analyzed using GraphPad Prism Software ver 6.01 (GraphPad Software Inc., La Jolla, CA, USA).
Results and discussion In order to achieve the performance reported here for the Randox reagent on the Cobas 6000 platform, several modifications from the vendor's stated application were incorporated into the final application presented in the Materials and methods section of this paper to achieve calibration stability, comparability amongst instruments and better long term assay performance. The reaction reading window in the application had to be adjusted to commence reading after the addition of the GLDH enzyme (Reagent 2) as this adjustment alleviated differences in background absorbance of substrate in the cassette after sitting onboard the analyzer. The adjustment of the reading window also minimized the impact of any background absorbance increases due to
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
4
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
endogenous interfering substances on the analyte measurement results. The initial reaction rates for the Randox reagent were not rapid enough to ensure full reaction completion necessary for an end-point calculation method in the 10 minute standard application length on the Roche c501 analyzer. This leads to variability in calibration results in high samples and poor correlation between instruments of the same type. This issue was alleviated by increasing the mixing settings on the application, and increasing enzyme concentration above that recommended in the manufacturer's package insert which allows the reaction to proceed at a faster rate and reach completion in the allotted 10 minute time frame. Within-run precision for this method was L1 = 65 μmol/L, CV 3.4% and L2 = 168 μmol/L, CV 1.9%, and between-run precision was L1 = 29 μmol/L, CV 7.3% L2 = 102 μmol/L, CV 3.0%. The Randox ammonia reagent was verified linear 8–940 μmol/L. The LLQ was proven to be 7 μmol/L, with a final AMR of 7–940 μmol/L. Accuracy was verified using proficiency survey material from the College of American Pathologists Survey C-A 2013 CHM-1 and the N = 3 samples analyzed with the Randox reagent recovered within 5% of the All Method Mean on the Survey. Survey samples analyzed with the Roche reagent also recovered with 5% of the peer group mean for this survey. A total of 123 samples were included in the method comparison study between the Roche and Randox reagents (Fig. 2A). However, 29/123 (24%) of clinical specimens received instrument flags with the Roche reagent for exceeding the absorbance maximum limits on the detector on the Cobas c501. The instrument flags using the c501 reagent were mostly seen in specimens with elevated levels of icterus and lipemia common in clinical specimens in our lab. The 29 samples receiving instrument flags were not included in calculations for correlation between Roche and Randox methods. The agreement between the Roche c501 and Randox ammonia reagents was very good for the remaining samples (Fig. 2A). The use of the Roche calibrator material with the Randox reagent for the purposes of this evaluation contributed to improving the correlation between the methods.
Of the samples excluded from comparison between Roche and Randox methods, 9 could be included in the comparison with the Beckman DxC method during the inter-site comparison study and correlated well (Fig. 2D). The cross-platform comparison study demonstrates good agreement between the Randox ammonia reagent and the wet chemistry ammonia methods on the Beckman and Siemens platforms (Fig. 2C and D) and a significant low proportional bias by Ortho Vitros 250 dry chemistry ammonia (Fig. 2B). The manufacturer has been advised of the bias and is currently investigating the issue. The Siemens Vista and Roche c501 chemistries were the most susceptible to endogenous interfering substances in the cross-platform comparison study samples. For those comparisons 31/40 samples run for crosssite comparison could be reported on Siemens and 64/80 on Roche platforms respectively without intervention (i.e. dilution) to remove interference. The Randox reagent was tested for interference from hemolysis, icterus and lipemia and demonstrated no significant interferences up to levels of 6 g/L hemoglobin for hemolysis (H index = 600), 1026 μmol/L conjugated or unconjugated bilirubin for Icterus (I index = 60) and 1 g/L intralipid for lipemia (L index = 100). It should be noted that lipemia index is a measure of sample turbidity on the c501 platform and does not correlate well with triglyceride levels in patient specimens. Hemolysis index can be affected by the under filling of EDTA tubes. Unger et al. have reported that at EDTA concentrations 3 times the usual, hemolysis index measurements can be 50% higher than the values in EDTA plasma collected under standard conditions [7]. However, in our patient population, hemolyzed specimens are the minority, while icteric and lipemic specimens are more abundant, and if there were under filling issues for the EDTA tubes in this study, they were insufficient to create any spurious hemolysis flags. The Randox Enzymatic Manual UV Ammonia reagent presents an alternative to the Roche c501 ammonia reagent on the Cobas 6000 automated analyzer platform as it is less prone to interference from endogenous interfering substances such as lipemia and icterus which
Fig. 2. Comparisons between chemistry ammonia methods for 4 automated chemistry platforms. A) Roche vs. Randox on Cobas 6000 (c501) platform; B) Ortho Vitros 250 vs Randox; C) Siemens Vista vs Randox and D) Beckman DxC vs Randox.
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
are present at rates as high as 25% in specimens sent to the lab for the measurement of ammonia. This method allows the clinical lab to report the majority of specimens received from the clinical units in the hospital setting without additional interventions.
Acknowledgments ESBE Scientific provided 1 sample kit for evaluation of the Randox reagent. All subsequent reagent kits used to complete this study were purchased by Calgary Laboratory Services. The authors would like to thank Sandra Fortier, Allison Venner and Kathy Carter for assisting with the inter-site comparison samples and Dr. Lawrence de Koning for his review of the inter-site correlation protocol.
5
References [1] Gundling F, Zelihic E, Seidl H, Haller B, Umgelter A, Schepp W, Dodt C. How to diagnose hepatic encephalopathy in the emergency department. Ann Hepatol 2013;12(1):108–14. [2] Kumar R, Shalimar, Sharma H, Prakash S, Panda SK, Khanal S, Acharya SK. Persistent hyperammonemia is associated with complications and poor outcomes in patients with acute liver failure. Clin Gastroenterol Hepatol 2012;10(8):925–31. [3] Neeley WE, Phillipson J. Automated enzymatic method for determining ammonia in plasma, with 14-day reagent stability. Clin Chem 1988;34(9):1868–9. [4] Pesh-Imam M, Kumar S, Willis CE. Enzymatic determination of plasma ammonia: evaluation of Sigma and BMC Kits. Clin Chem 1978;24(11):2044–6. [5] Roche. NH3L — unstable QC results on cobas c 311, c 501 and c 502. Urgent notification RR2013-05/13-009. Roche Customer Care; 2013. [6] Dimeski G. Effects of hemolysis on the Roche ammonia method for Hitachi analyzers. Clin Chem 2004;50(5):976–7. [7] Unger J, Filippi G, Patsch W. Measurements of free hemoglobin and hemolysis index: EDTA- or lithium-heparinate plasma? Clin Chem 2007;53(9):1717–8.
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
Contents lists available at ScienceDirect
Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem
Short Communication
Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform Isolde Seiden-Long a,b,⁎, Kareena Schnabl c, Wendy Skoropadyk c, Nola Lennon a, Arlayne McKeage a a b c
Calgary Laboratory Services, Calgary, AB, Canada Department of Pathology and Laboratory Medicine, University of Calgary, Canada Department of Laboratory Medicine and Pathology, University of Alberta, Canada
a r t i c l e
i n f o
Article history: Received 4 February 2014 Received in revised form 21 April 2014 Accepted 22 April 2014 Available online xxxx Keywords: Method evaluation Automated analyzers Application development Ammonia Enzymatic assays
a b s t r a c t Objective: Adaptation of the Randox Enzymatic Manual UV Ammonia method to be used on the Roche Cobas 6000 (c501) automated analyzer platform. Design and methods: The Randox ammonia reagent was evaluated for precision, linearity, accuracy and interference from hemolysis, icterus and lipemia on the Roche c501 analyzer. Comparison studies were conducted for the Randox reagent between Roche c501, Siemens Vista, Ortho Vitros 250, and Beckman DxC methods. Results: The Randox reagent demonstrates acceptable within-run (L1 = 65 μmol/L, CV 3.4% L2 = 168 μmol/L, CV 1.9%) and between-run precision (L1 = 29 μmol/L, CV 7.3% L2 = 102 μmol/L, CV 3.0%), Analytical Measurement Range (7–940 μmol/L), and accuracy. The method interference profile is superior for the Randox method (hemolysis index up to 600, icteric index up to 60, lipemic index up to 100) as compared to the Roche method (hemolysis index up to 200, icteric index up to 10, lipemic index up to 50). Comparison was very good between the Randox reagent and two other wet chemistry platforms. Conclusions: The Randox Enzymatic Manual UV Ammonia reagent is an available alternative to the Roche Cobas c501 reagent. The method is more robust to endogenous interferences and less prone to instrument error flags, thus allowing the majority of clinical specimens to be reported without additional sample handling at our institution. © 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Introduction Ammonia testing is performed in the clinical lab for two distinct patient populations. In the adult population, ammonia levels can be elevated during liver failure or impairment, urinary tract infection, gastrointestinal bacterial overgrowth, due to various medications e.g. valproic acid, chemotherapy, and in-patients on total parenteral nutrition. The lack of clinical specificity for elevated ammonia levels makes it a poor marker for diagnosis of hepatic encephalopathy in the emergency department [1]. However, persistent levels in patients with cirrhosis/hepatic failure can provide useful prognostic information [2]. In the pediatric population, blood ammonia levels can be increased in inherited defects of the urea cycle, organic acidurias, disorders of fatty acid oxidation, other illness in babies e.g. sepsis, asphyxia, Reye's syndrome and transient hyperammonemia of the newborn (normally seen within the first 24 h of life). Due to the underlying clinical causes
⁎ Corresponding author at: Department of Pathology and Laboratory Medicine, University of Calgary, Foothills Medical Centre, 1403-29 St. NW (C618B), T2N 2T9, Canada. Fax: +1 403 944 1364. E-mail address: [email protected] (I. Seiden-Long).
of ammonia elevation in these populations, the clinical lab often receives lipemic or icteric samples for ammonia measurements. In addition, neonate samples are often hemolyzed. In our own facility, our lab receives approximately 25% of our specimens for ammonia testing either icteric, lipemic or hemolyzed. Due to the pre-analytical instability of ammonia levels in plasma samples, it is not feasible to centrifuge samples to remove lipemia without heating the specimen and causing unacceptable pre-analytical increase in ammonia levels. The only intervention that can be used to reduce interferent levels in an acceptable time frame is sample dilution. Consequently, an ammonia method must be relatively robust to these endogenous interferents to reliably report test results in the clinical diagnostic laboratory. Most commercial chemistry analyzers currently on the market use an enzymatic ammonia method in which glutamate dehydrogenase catalyzes the reaction of ammonium with α‐ketoglutarate and reduced nicotinic adenine dinucleotide (phosphate) (NAD(P)H) to form glutamate and NAD(P)+ and water. Absorbance decrease in [NADH] or [NADPH] can then be measured at a wavelength of 340 nm [3,4]. Assay reaction:
GLDH
þ
þ
NH3 þ α‐ketoglutarate þ NADPH → glutamate þ NADP þ H :
http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022 0009-9120/© 2014 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
2
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
Out of 3849 labs reporting ammonia on the College of American Pathologists 2012 Chemistry C proficiency survey, 3259 were reporting a glutamate dehydrogenase enzymatic method and 556 by a thin film dry chemistry method. Dry-slide analyzers use a micro-diffusion apparatus to liberate free ammonia from the sample by alkalization. The ammonia is then measured colorimetrically by reflectance change of color of an indicator dye such as bromophenol blue. Our laboratory had been experiencing high frequencies of sample error flags (approximately 25% of all samples) using the Roche ammonia reagents on the Cobas c501 since the instrument was implemented in 2011. The instrument flags using the c501 reagent are often encountered in specimens with elevated levels of icterus and lipemia. Interferents in patient specimens increase background absorbance in the reaction which is not alleviated by subtracting the secondary absorbance reading at 700 nm. For the Cobas c501, this results in increased reagent background blank to a level which exceeds the absorbance maximums specified for the instrument detector once reagent 2 is added to the reaction (Fig. 1A–D). This sensitivity of the Roche method to endogenous interferents very common in ammonia blood samples is largely due to increasing reagent absorbance background in aging lots of the Cobas reagent. This results in instrument over-range absorbance flags even in samples which otherwise would have not received other interference flags (Fig. 1A, B). Roche has recently issued a technical bulletin advising customers to allow the reagent to rest open on the bench for 24 h to release ammonia build up prior to loading on the instruments [5]. This process was implemented in our laboratory with little effect on reducing instrument error flag rates. We describe here the adaptation of a third party reagent produced by Randox Inc. which our laboratory has used on the Roche Cobas c501 platform. The method principle utilizes glutamate dehydrogenase, similar to the Roche reagent, but stabilizers are added to the substrate which keep the background absorbance of the reagent well within the mid-range on the c501 detector limits making the method much more robust to endogenous interfering substances in human plasma samples.
Cassette filling and application parameters The following application parameters have been adapted from those provided by Randox.
Multi cassette filling for 150 tests/kit Bottle a — R1, 150 tests, 15 mL use volume, 20 mL fill volume Bottle b — not used Bottle c — R3, 150 tests, 3.8 mL use volume, 6.4 mL fill volume Instrument application — Roche c501 Test/type Assay/Rxn time/meas point Wave (2nd/Pri) Sample volume (norm.) (Dec.) (Inc.) Diluent Reagent Vol. R1 Reagent Vol. R2 Reagent Vol. R3 Linearity limit Prozone limit Abs. limit Cell detergent Stirring level Calibrator Calibration method Point/Span/Weight SD Limit Duplicate Limit Sensitivity Limit S1 ABS Limit Change over (to cassette) Autocalibration timeout Cassette Calib./timeout/days Range Unit Report name Data mode Technical limit H I L
AMM–PLASMA 2 point end — 10–36–70 700–340 8–0–0 4–0–0 16–0–0 Water 100 0 25 0 — decrease 0–0–0–0 0–0–0–0–0–0 Detergent 1 3 Linear 2–2–0 999 2–30 –3.6 – –2.7 –32000–32000 2 point Yes 2 point/blank/3 μmol/L NH3R Active 7–940 600 60 100
Materials and methods Reagents and samples Randox Enzymatic Manual UV Ammonia reagent was purchased from ESBE Scientific (Markham,ON, Canada). Reagents were prepared as per manufacturer's instructions. Ammonia calibrators and Roche Cobas 6000 c501 ammonia reagents and Multipack cassettes were purchased from Roche Diagnostics (Laval, QC, Canada). All samples were analyzed on a Roche Cobas c501 automated analyzer platform. Liquicheck Ethanol/Ammonia control materials were purchased from Bio-Rad Laboratories Canada Limited. Intralipid (20% emulsion) and unconjugated bilirubin were purchased from SigmaAldrich (St. Louis, MO,USA). Conjugated di-taurobilirubin was purchased from Millipore Corporation (MA, USA). Patient plasma specimens used for the purposes of this validation were collected as part of routine patient care at our institution. Use of these samples for the purposes of method evaluation is classified as a quality assurance/program evaluation activity to be used for assessment, management and/or improvement and is granted quality assurance exemption from ethics review requirement by the Conjoint Health Research Ethics Board at the University of Calgary. Briefly, K + EDTA vacutainer tubes (Becton Dickinson) are pre-chilled prior to collection and samples delivered to the lab on ice. Samples are centrifuged without delay at 4 °C ± 2 °C, aliquoted, and placed on ice. Plasma was stored at 4 °C for up to 1 h and analyzed or frozen for long term storage.
Parameters for onboard reagent stability and calibration were determined during an extended between run precision study using a failure of our QC rule set for ammonia (13S/22S/R4S/41S) to trigger a calibration event. Onboard reagent stability was verified up to 30 days. Blank calibration required every 3 days. Two point calibration was required every 21 days.
Method evaluation Comparison samples run within-site were run on Roche and Randox reagents simultaneously to minimize pre-analytical increase in ammonia. The cross-site multiple instrument comparison samples were frozen at −80 °C, distributed to 4 locations and analyzed by standardized sample handling protocol on the same day at all 4 locations with timed thaw-to-analyze intervals to allow comparability of results. For Beckman DxC, Siemens Vista and Ortho Vitros 250 methods, the reference locations utilized the manufacturer's reagent and calibrators for the comparison study. Within-run (N = 10 replicates) and betweenrun precision (N = 30 replicates) experiments were conducted using quality control materials at clinically relevant ammonia levels. Linearity experiments were conducted using high and low ammonia pools in human plasma matrix spanning the analytical measurement range (AMR) of the assay measured N = 5 times at each level. Lower limit of quantitation (LLQ) was conducted by serial dilution and defined as
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
3
Fig. 1. Reaction trace for typical patient specimens analyzed for ammonia level on Cobas 6000. The patient specimen is added at the same time as Reagent 1 in both reactions. A and B) Sample which is mildly icteric and lipemic and the instrument can provide a numeric result for Randox reagent (B), but NAbs flag for Roche reagent (A). C and D) sample exceeds maximum Icteric index of 10 for the Roche method but is still reportable by the Randox method (Icteric max 60). E) Reaction curve for Randox reagent for high specimen. Reaction rate is slower with the Randox reagent than the Roche reagent, but complete for the last 4 reading points of the 10 minute application.
the lowest measured analyte level achieving a 20% CV. Accuracy was verified using proficiency survey material from the College of American Pathologists (Survey C-A 2013). Interference studies for lipemia and icterus were conducted by spiking an EDTA plasma pool with intralipid or conjugated or unconjugated bilirubin stock solution. The hemolysis interference experiment was conducted as described by Dimeski [6], with hemolysate made fresh prior to testing to distinguish true hemolysis interference from spurious elevations due to long term storage of hemolysate. There is no biological variation data available for ammonia measurement, thus criteria for significant interference were based on assay precision (b 50 μmol/L ± 20%, N = 50 μmol/L ± 10%). Data was analyzed using GraphPad Prism Software ver 6.01 (GraphPad Software Inc., La Jolla, CA, USA).
Results and discussion In order to achieve the performance reported here for the Randox reagent on the Cobas 6000 platform, several modifications from the vendor's stated application were incorporated into the final application presented in the Materials and methods section of this paper to achieve calibration stability, comparability amongst instruments and better long term assay performance. The reaction reading window in the application had to be adjusted to commence reading after the addition of the GLDH enzyme (Reagent 2) as this adjustment alleviated differences in background absorbance of substrate in the cassette after sitting onboard the analyzer. The adjustment of the reading window also minimized the impact of any background absorbance increases due to
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
4
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
endogenous interfering substances on the analyte measurement results. The initial reaction rates for the Randox reagent were not rapid enough to ensure full reaction completion necessary for an end-point calculation method in the 10 minute standard application length on the Roche c501 analyzer. This leads to variability in calibration results in high samples and poor correlation between instruments of the same type. This issue was alleviated by increasing the mixing settings on the application, and increasing enzyme concentration above that recommended in the manufacturer's package insert which allows the reaction to proceed at a faster rate and reach completion in the allotted 10 minute time frame. Within-run precision for this method was L1 = 65 μmol/L, CV 3.4% and L2 = 168 μmol/L, CV 1.9%, and between-run precision was L1 = 29 μmol/L, CV 7.3% L2 = 102 μmol/L, CV 3.0%. The Randox ammonia reagent was verified linear 8–940 μmol/L. The LLQ was proven to be 7 μmol/L, with a final AMR of 7–940 μmol/L. Accuracy was verified using proficiency survey material from the College of American Pathologists Survey C-A 2013 CHM-1 and the N = 3 samples analyzed with the Randox reagent recovered within 5% of the All Method Mean on the Survey. Survey samples analyzed with the Roche reagent also recovered with 5% of the peer group mean for this survey. A total of 123 samples were included in the method comparison study between the Roche and Randox reagents (Fig. 2A). However, 29/123 (24%) of clinical specimens received instrument flags with the Roche reagent for exceeding the absorbance maximum limits on the detector on the Cobas c501. The instrument flags using the c501 reagent were mostly seen in specimens with elevated levels of icterus and lipemia common in clinical specimens in our lab. The 29 samples receiving instrument flags were not included in calculations for correlation between Roche and Randox methods. The agreement between the Roche c501 and Randox ammonia reagents was very good for the remaining samples (Fig. 2A). The use of the Roche calibrator material with the Randox reagent for the purposes of this evaluation contributed to improving the correlation between the methods.
Of the samples excluded from comparison between Roche and Randox methods, 9 could be included in the comparison with the Beckman DxC method during the inter-site comparison study and correlated well (Fig. 2D). The cross-platform comparison study demonstrates good agreement between the Randox ammonia reagent and the wet chemistry ammonia methods on the Beckman and Siemens platforms (Fig. 2C and D) and a significant low proportional bias by Ortho Vitros 250 dry chemistry ammonia (Fig. 2B). The manufacturer has been advised of the bias and is currently investigating the issue. The Siemens Vista and Roche c501 chemistries were the most susceptible to endogenous interfering substances in the cross-platform comparison study samples. For those comparisons 31/40 samples run for crosssite comparison could be reported on Siemens and 64/80 on Roche platforms respectively without intervention (i.e. dilution) to remove interference. The Randox reagent was tested for interference from hemolysis, icterus and lipemia and demonstrated no significant interferences up to levels of 6 g/L hemoglobin for hemolysis (H index = 600), 1026 μmol/L conjugated or unconjugated bilirubin for Icterus (I index = 60) and 1 g/L intralipid for lipemia (L index = 100). It should be noted that lipemia index is a measure of sample turbidity on the c501 platform and does not correlate well with triglyceride levels in patient specimens. Hemolysis index can be affected by the under filling of EDTA tubes. Unger et al. have reported that at EDTA concentrations 3 times the usual, hemolysis index measurements can be 50% higher than the values in EDTA plasma collected under standard conditions [7]. However, in our patient population, hemolyzed specimens are the minority, while icteric and lipemic specimens are more abundant, and if there were under filling issues for the EDTA tubes in this study, they were insufficient to create any spurious hemolysis flags. The Randox Enzymatic Manual UV Ammonia reagent presents an alternative to the Roche c501 ammonia reagent on the Cobas 6000 automated analyzer platform as it is less prone to interference from endogenous interfering substances such as lipemia and icterus which
Fig. 2. Comparisons between chemistry ammonia methods for 4 automated chemistry platforms. A) Roche vs. Randox on Cobas 6000 (c501) platform; B) Ortho Vitros 250 vs Randox; C) Siemens Vista vs Randox and D) Beckman DxC vs Randox.
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022
I. Seiden-Long et al. / Clinical Biochemistry xxx (2014) xxx–xxx
are present at rates as high as 25% in specimens sent to the lab for the measurement of ammonia. This method allows the clinical lab to report the majority of specimens received from the clinical units in the hospital setting without additional interventions.
Acknowledgments ESBE Scientific provided 1 sample kit for evaluation of the Randox reagent. All subsequent reagent kits used to complete this study were purchased by Calgary Laboratory Services. The authors would like to thank Sandra Fortier, Allison Venner and Kathy Carter for assisting with the inter-site comparison samples and Dr. Lawrence de Koning for his review of the inter-site correlation protocol.
5
References [1] Gundling F, Zelihic E, Seidl H, Haller B, Umgelter A, Schepp W, Dodt C. How to diagnose hepatic encephalopathy in the emergency department. Ann Hepatol 2013;12(1):108–14. [2] Kumar R, Shalimar, Sharma H, Prakash S, Panda SK, Khanal S, Acharya SK. Persistent hyperammonemia is associated with complications and poor outcomes in patients with acute liver failure. Clin Gastroenterol Hepatol 2012;10(8):925–31. [3] Neeley WE, Phillipson J. Automated enzymatic method for determining ammonia in plasma, with 14-day reagent stability. Clin Chem 1988;34(9):1868–9. [4] Pesh-Imam M, Kumar S, Willis CE. Enzymatic determination of plasma ammonia: evaluation of Sigma and BMC Kits. Clin Chem 1978;24(11):2044–6. [5] Roche. NH3L — unstable QC results on cobas c 311, c 501 and c 502. Urgent notification RR2013-05/13-009. Roche Customer Care; 2013. [6] Dimeski G. Effects of hemolysis on the Roche ammonia method for Hitachi analyzers. Clin Chem 2004;50(5):976–7. [7] Unger J, Filippi G, Patsch W. Measurements of free hemoglobin and hemolysis index: EDTA- or lithium-heparinate plasma? Clin Chem 2007;53(9):1717–8.
Please cite this article as: Seiden-Long I, et al, Evaluation of a third party enzymatic ammonia method for use on the Roche Cobas 6000 (c501) automated platform, Clin Biochem (2014), http://dx.doi.org/10.1016/j.clinbiochem.2014.04.022