Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP-OES, and ICP-MS in Diverse Laboratory Settings

Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP-OES, and ICP-MS in Diverse... Progress improving zinc nutrition globally is slowed by limited understanding of population zinc status. This challenge is compounded when small differences in measurement can bias the determination of zinc deficiency rates. Our objec - tive was to evaluate zinc analytical accuracy and precision among different instrument types and sample matrices using a standardized method. Participating laboratories analyzed zinc content of plasma, serum, liver samples, and controls, using a standardized method based on current practice. Instrument calibration and drift were evaluated using a zinc standard. Accuracy was evaluated by percent error vs. reference, and precision by coefficient of variation (CV). Seven laboratories in 4 countries running 9 instruments completed the exercise: 4 atomic absorbance spectrometers (AAS), 1 inductively coupled plasma optical emission spectrometer (ICP-OES), and 4 ICP mass spectrometers (ICP-MS). Calibration differed between individual instruments up to 18.9% (p < 0.001). Geometric mean (95% CI) percent error was 3.5% (2.3%, 5.2%) and CV was 2.1% (1.7%, 2.5%) overall. There were no significant differences in percent error or CV among instrument types ( p = 0.91, p = 0.15, respectively). Among sample matrices, serum and plasma zinc measures had the highest CV: 4.8% (3.0%, 7.7%) and 3.9% (2.9%, 5.4%), respectively (p < 0.05). When using standardized materials and methods, similar zinc concentration values, accuracy, and precision were achieved using AAS, ICP-OES, or ICP-MS. However, method development is needed for improvement in serum and plasma zinc measurement precision. Differences in calibration among instruments demonstrate a need for harmonization among laboratories. Keywords Plasma zinc · Zinc deficiency · Accur acy · Precision · Harmonization Introduction and growth [2–4]. In the absence of a severe zinc deficiency, circulating zinc is highly conserved [5], and there is no sen- Nearly 20% of the global population is at risk of inadequate sitive and specific indicator of zinc status in individuals [ 2, zinc intake [1]. Zinc is an essential micronutrient with 6]. The distribution of plasma or serum zinc concentration unique roles in protein structure and as a cofactor in sub- within a population does, however, allow the determination strate binding and enzymatic activity. Due to these molecu- of population zinc status [6]. Therefore, plasma or serum lar roles, zinc contributes to broad range of functions includ- zinc concentration is one of three recommended biomarkers ing DNA transcription and repair, cell signaling, energy for assessment of zinc status at the population level along metabolism, immune and central nervous system function, with dietary zinc intake and the prevalence of stunting among children under 5 [2]. Future progress towards reducing zinc deficiency glob - * Andrew G. Hall ally will depend on monitoring changes in plasma or serum [email protected]; [email protected] zinc status. Apart from dietary zinc intake, a number of fac- Benioff Children’s Hospitals; Children’s Hospital Oakland tors affect plasma or serum zinc concentration, including Research Institute, University of California San Francisco, systemic inflammation, time of specimen collection, fast - Oakland, CA, USA ing status, and variations in the processing and handling of Department of Nutritional Sciences and Toxicology, samples [2, 6–8]. To minimize the impact of some of these University of California Berkeley, Berkeley, CA, USA issues, recommendations have been made for the design of Department of Pediatrics, School of Medicine, University of California San Francisco, San Francisco, CA, USA Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… human studies assessing zinc status, and procedures for col- The methods document was sent to the three laboratories for lecting samples for zinc analysis [6, 9]. review and revised according to their feedback. The extent that analytical method and instrument may further contribute to variation in reported plasma or serum Laboratory Exercise A laboratory exercise was designed zinc values has seen limited study. The instruments typically for comparison of accuracy and precision of zinc concentra- used for zinc analysis vary in sensitivity and complexity tion measurements among different instruments and sample of operation. They include atomic absorption spectrome- matrices selected based on the literature review. The labo- ters (AAS, flame or graphite furnace), inductively coupled ratory exercise was designed to minimize variability due plasma optical (atomic) emission spectrometers (ICP-OES/ to sample preparation, reagent quality, and zinc reference ICP-AES), and ICP mass spectrometers (ICP-MS). Our pri- material. In addition to the Methods Reference Document, mary aim was an evaluation of accuracy and precision by specific instructions for instrument calibration, sample prep - instrument type and by sample type in diverse laboratory aration, and a standard set of materials for analysis were settings where zinc analysis of biological specimens from included. large surveys or clinical studies are typically conducted. Laboratories in low- and middle-income countries that had previously conducted zinc analysis studies for national- level surveys, and laboratories that had contributed proto- Materials and Methods cols for inclusion in the Methods Reference Document, were invited to participate in the laboratory exercise. Participat- We developed a standardized method based on a review of ing laboratories were instructed to use the identical sets of recent human studies, and instructed participating laborato- materials and supplies assembled at CHORI and shipped to ries to implement the method for the analysis of a standard- each laboratory. These included trace element analysis grade ized set of samples and controls. 68–70% nitric acid (Omnitrace, EMD Millipore, Burlington, MA; or BDH Aristar Plus, VWR International, Radnor, PA), Literature Review and Methods Reference Document To ultrapure water (Omnitrace Ultra, EMD Millipore, Burling- determine the typical sample matrices, preparation meth- ton, MA), and filter pipette tips (VWR International, Radnor, ods, and analytical methods for the measurement of zinc in PA) to minimize particulate contamination from the pipette human studies, a literature review was conducted. Human mechanism. studies published over the previous 5-years were identified Laboratories calibrated their instruments using the pro- in PubMed (National Center for Biotechnology Information, vided Standard Reference Material (SRM) 3168a zinc in Bethesda, MD) using the single search term “zinc” and fil - 10% nitric acid (National Institute of Standards and Technol- ters for human studies dated between February 2013 and ogy (NIST), Gaithersburg, MD). To detect potential differ - February 2018. Studies written in English with abstracts ences in calibration, and to assess instrument drift during the reporting the measurement of zinc in human derived sam- analysis, a dilute solution of SRM 3168a, containing 12.5 µg ples or foods were downloaded for further review. Articles zinc/dL in 5% nitric acid, was prepared in bulk at CHORI without zinc analytical data in the full text were excluded. A using the same ultrapure water and nitric acid provided to detailed description of the literature review is provided in the the participating laboratories. accompanying online material (Supplemental Appendix A). Sample matrices included human serum and plasma, and Three laboratories prominent in the literature review were powdered bovine liver (representative of prepared food com- contacted and their zinc analysis protocol(s) requested: Lab- posites). Unknowns (i.e., samples without a known value) oratory of Human Nutrition at the Institute of Food, Nutri- for each, as well as reference materials with certified values, tion and Health, Swiss Federal Institute of Technology were included for analysis. Unknowns or reference materials (Zurich, Switzerland); Section of Pediatric Nutrition, Uni- were dispensed into vials at CHORI prior to distribution to versity of Colorado School of Medicine (Aurora, CO); and the laboratories as follows: All vials were acid-washed prior Children’s Hospital Oakland Research Institute (CHORI), to dispensing. Human plasma and human serum unknowns UCSF Benioff Children’s Hospitals (Oakland, CA). Meth - (Zenbio, Research Triangle Park, NC) were shipped unfro- odological elements of zinc analysis protocols from these zen on cold pack from North Carolina to CHORI overnight laboratories, in addition to protocols provided by the Centers the day of collection, mixed and aliquoted into screw-cap for Disease Control and Prevention (CDC, Atlanta, GA) [10] polypropylene vials, and then frozen. The liver unknown, and the United States Department of Agriculture (USDA, food grade powdered bovine liver (CurEase, McEwen, TN), Washington, DC) [11], were reviewed and assembled into a was mixed for 5 min using a food processor (Cuisinart, Stan- Methods Reference Document (Supplemental Appendix B) ford, CT) prior to dispensing into polypropylene screwcap of techniques for determining zinc concentration in samples vials. from human studies using AAS, ICP-OES, and ICP-MS. A. G. Hall et al. Reference materials included SRM 1950 human plasma published (reference) zinc concentration for each reference and SRM 1577c powdered bovine liver (NIST, Gaithersburg, material: MD), Seronorm Trace Elements in Human Serum Levels 1 [Zn] − [Zn] measured reference and 2 (SERO AS, Billingstad, Norway), and a custom UTAK %error = × 100 [Zn] reference human plasma containing no added zinc (UTAK Laborato- ries, Valencia, CA). The SRM 1950 serum was distributed Percent bias was defined as the relative difference to each laboratory frozen in the sealed 1-mL glass ampoules between the zinc concentration measured, and the published with rubber stoppers, as provided by NIST. Lyophilized (reference) zinc concentration for each reference material Seronorm serum and UTAK plasma were reconstituted at (maintaining the sign of the difference from reference): CHORI using ultrapure water (Omnitrace Ultra, EMD Mil- [Zn] − [Zn] lipore, Burlington, MA) according to the manufacturer’s measured reference %bias = × 100 instructions, and dispensed into screw cap polypropylene [Zn] reference vials. The SRM 1577c bovine liver was also dispensed into The overall CV, % error, and % bias for a given instru- screwcap polypropylene vials prior to distribution to the ment were determined as the mean (or geometric mean) of laboratories. the respective indicator for all samples analyzed by each All sera and plasma were shipped on dry ice from CHORI instrument. The overall CV, % error, and % bias for a given to each participating laboratory. Zinc solutions, powdered sample matrix was determined as the overall mean (or geo- liver, ultrapure water, and pipette tips were shipped sepa- metric mean) value for the respective sample matrix. rately from CHORI at ambient temperature. The trace ele- All data were tabulated; % error, % bias, and CV were ment grade nitric acid was shipped directly from the supplier calculated using Microsoft Excel 2010 (Microsoft, Red- to each participating laboratory. To minimize the potential mond, WA). Descriptive statistics were calculated and for bias, individual laboratories were not informed of the statistical comparisons performed using SPSS 26 (IBM, known zinc values of any of the materials sent, with the Armonk, NY). Data were tested for normal distribution and exception of the pre-diluted zinc standard, reported as falling descriptive statistics calculated prior to statistical compari- within a range between 10 and 15 µg zinc/dL. Laboratories son by unpaired t-test or ANOVA. Unless stated otherwise, measured the zinc concentration of the pre-diluted zinc solu- all data are formatted in the text as “mean ± standard devia- tion in triplicate at the beginning and end of the analytical tion” or, where not normally distributed, as “geometric mean run, that of each reference material in triplicate, and each (95% confidence interval).” ICP-OES was not included in unknown 9 times. To minimize variability in viscosity that comparisons between instrument types, since there was could affect instrument sampling flow rates, a simple proto - only one such instrument. Values for CV and % error were col for digestion of serum and plasma in concentrated nitric log-transformed to achieve normal distribution prior to acid was specified. statistical comparison. Statistically significant differences were defined as p < 0.05 for all comparisons. As the study Data Analysis Instrument calibration was based on the was not designed to evaluate the accuracy and precision of measurement of zinc concentration in the pre-diluted SRM individual laboratories, the individual laboratories and their 3168a zinc solution for each instrument. Percent drift was individual instruments are not identified with respect to ana - defined as the relative change in measurement of the zinc lytical results. solution from start to end of the analytical run: [Zn] − [Zn] end start %drift = × 100% [Zn] start Results Precision was evaluated for all reference materials and unknowns by determining the coefficient of variation (CV), Literature Review Detailed results of the literature review i.e., sample mean divided by the sample standard deviation are provided in the supplemental material (Supplemental for each sample analyzed, expressed as a percentage: Appendix A). Briefly, out of 470 PubMed search hits, 134 peer-reviewed journal articles met the criteria for review. CV = × 100% Zinc concentration was most frequently determined in serum (n = 69), plasma (n = 48), and food composites (n = 12). Accuracy was evaluated using % error and % bias. Percent Instruments for zinc quantification included AAS ( n = 78 error was defined as the absolute value of the relative dif - AAS), ICP-MS (n = 22), and ICP-OES (n = 16). While 38 of ference between the zinc concentration measured, and the the studies using AAS described the instrument as a flame AAS, only three described using a graphite furnace AAS. Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… Seven studies used a plate reader or auto-analyzer for indi- rect zinc determination utilizing a zinc sensitive chemical dye or probe. Laboratory Exercise Seven laboratories in 4 countries, running 9 individual instruments, participated. The labora- tories and their instruments are listed in Table 1. Instrument calibration was assessed by measuring the zinc concentra- tion of the pre-diluted SRM 3168a solution (Fig. 1). Zinc concentration was 12.2 ± 0.7 µg/dL overall, and did not differ significantly ( p = 0.8) between AAS (12.0 ± 0.5 µg/dL, n = 4) and ICP MS (12.4 ± 1.0 µg/dL, n = 4). None of the individual instruments varied more than 10% from the expected value of 12.5 µg/dL. However, differences in calibration between individual instruments were as high as 18.9% (p < 0.001). Fig. 1 Instrument calibration. Pre-diluted SRM 3168a values from The pre-diluted zinc standard was measured again at the each instrument, with expected concentration (12.5  µg/dL, dotted end of the analytical run and instrument drift calculated rela- horizontal line), and range of ± 10% from the expected value (hori- zontal dashed lines). Data are displayed as mean ± standard deviation. tive to the initial values. The mean drift was 0.4% ± 3.0% Lowercase letters a, b, and c denote homogenous subsets where dif- (n = 8), and did not differ significantly ( p = 0.3) between fering letters represent statistically significant differences between AAS (1.8% ± 3.5%, n = 4) and ICP-MS-1.7% ± 1.7%, n = 3). individual instruments (p < 0.05) One AAS instrument drifted by more than 5% from its initial values (6.9% drift, from 11.8 ± 0.53 µg/dL to 12.6 ± 0.28 µg/ dL, p < 0.05). Data on drift were missing from one ICP-MS Serum and plasma unknowns yielded higher measure- ment variability (lower precision) compared with other sam- instrument. Zinc concentration of plasma, serum, and zinc content of ple matrices (Fig. 3). Measures of zinc concentration of the unknown serum had a CV of 4.8% (3.0%, 7.7%), and were liver samples is reported, according to each instrument type, in Fig. 2. No significant differences were observed between less precise (higher CV) than the pre-diluted SRM 3168a zinc standard solution, CV of 1.4% (0.7%, 2.9%) (p = 0.042), AAS and ICP-MS for any material measured. Precision between instrument types and samples was evaluated using and compared with SRM 1577c liver, CV of 1.1% (0.6%, 2.2%) (p = 0.006). Zinc measures of unknown plasma had the CV. The overall CV for each instrument was determined for all samples analyzed by that instrument. The instrument a CV of 3.9% (2.9%, 5.4%), and were less precise (higher) than SRM 1577c liver (p = 0.031). CV was 1.7% (0.8%, 3.7%) for AAS, 0.9% for ICP-OES, and 2.0% (1.1%, 3.4%) for ICP-MS, 1.7% (1.2%, 2.4%) overall. Accuracy between instrument types and samples was evaluated by the measurement error and bias. Reference val- No differences in precision were detected between AAS and ICP-MS (p = 0.61, n = 4 each for AAS and ICP-MS). ues for zinc used in calculation of accuracy were Seronorm Table 1 Participating laboratories and instruments Institute Location Instrument(s) Aga Khan University, Nutrition Research Laboratory Karachi, Pakistan iCE 3000 Flame AAS (Thermo Fisher Scientific, Waltham, MA) Children’s Hospital Oakland Research Institute, UCSF Oakland, California, USA 5100 SVDV ICP-OES (Agilent Technologies, Santa Benioff Children’s Hospitals Clara, CA) International Centre for Diarrhoeal Disease Research Dhaka, Bangladesh AA-7000 Flame AAS (Shimadzu, Kyoto, Japan) Oklahoma State University, Department of Nutritional Stillwater, Oklahoma, USA ELAN 9000 ICP-MS (PerkinElmer, Waltham, MA) Sciences Laboratory of Human Nutrition; Institute of Food, Zurich, Switzerland 240FS Flame AAS (Agilent Technologies, Santa Clara, Nutrition and Health; Swiss Federal Institute of CA), Technology iCAP RQ ICP-MS (Thermo Fisher Scientific, Waltham, MA) Interdisciplinary Center for Plasma Mass Spectrometry; Davis, California, USA 8900 ICP-MS (Agilent Technologies, Santa Clara, CA) University of California, Davis Section of Pediatric Nutrition, University of Colorado Aurora, Colorado, USA AAnalyst 700 Flame AAS (PerkinElmer, Waltham, MA), School of Medicine 7700 ICP-MS (Agilent Technologies, Santa Clara, CA) A. G. Hall et al. Fig. 2 Zinc concentration of reference and unknown serum, plasma, and liver. Zinc concentration of each serum, plasma, and liver sample was determined by AAS (n = 4), ICP-OES (n = 1), and ICP-MS (n = 4). Data are displayed as mean ± standard deviation. No differences were detected between AAS and ICP-MS for any material, or overall Fig. 3 Zinc measurement precision by material analyzed. Data are expressed as geomet- ric mean and 95% confidence interval. Lowercase letters a, b, and c denote homogenous subsets where differing letters represent statistically significant differences ( p < 0.05) L1 serum, 109.7 µg/dL; Seronorm L2 serum, 161.7 µg/dL; measures of SRM 1950 plasma (10.8% ± 20.6%) were biased UTAK plasma, 59.4 µg/dL; SRM 1950 plasma, 71.3 µg/ high compared with Seronorm L1 serum (− 1.3% ± 5.8%, dL; and SRM 1577c liver, 181.1 µg/g. The reference value p < 0.05), UTAK plasma (− 4.9% ± 7.6%, p < 0.01), and for SRM 1950 plasma was converted to µg/dL using the SRM 1577c liver (− 3.0% ± 4.6%, p < 0.01) (Fig. 4(B)). reported density of 1.02086 g/mL from the certificate of analysis. Error for all instruments was 3.5% (2.2%, 5.6%) overall: 3.3% (0.8%, 14.1%) for AAS, 4.4% for ICP-OES, and 3.4% Discussion (1.8%, 6.5%) for ICP-MS. No significant difference in error was observed between AAS and ICP-MS (p = 0.95, n = 4 This is the first multi-site evaluation of accuracy and preci - each). Higher error was observed with SRM 1950 plasma sion of elemental zinc analysis in samples typical of human compared with Seronorm L1, 10.2% (4.6%, 22.5%) vs. 1.7% studies that the authors are aware of. Comparable results (0.5%, 5.8%), respectively (p < 0.05) (Fig. 4(A)). for the measurement of zinc were achievable with a variety The bias for all instruments was 1.0% ± 6.4% overall: of instrumentation in geographically diverse settings, using 2.0% ± 9.7% for AAS, 0.3% for ICP-OES, and 0.1% ± 3.8% methods and samples typical of human research. However, for ICP-MS. No significant difference in bias was observed the results for calibration, accuracy, and precision under- between AAS and ICP-MS (p = 0.74, n = 4 each). Overall, score potential pitfalls and areas for improvement in the Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… Fig. 4 Zinc measurement accuracy by reference material analyzed. Lowercase letters a and b denote homogenous subsets where differing letters represent statistically significant differences between materials (p < 0.05). (A) Percent error is expressed as geometric mean and 95% CI. (B) Percent bias is expressed as mean ± SEM implementation of zinc analytical methods in the clinical are substantially higher than the recommended cutoffs for research setting. zinc deficiency (i.e., from 57 to 74 µg/dL, depending on the Although differences in calibration were as high as 18% population, fasting status, and time of collection), as well among the individual instruments, each instrument was as the serum zinc concentrations of the populations from within 10% of the expected value, an accepted cutoff for which the cutoffs were derived (ranging from 76 ± 16% to calibration verification [ 10]. Nonetheless, variation in the 98 ± 14% µg/dL (geometric mean ± CV)) [6, 15]. number estimated to be zinc deficient, resulting from dif - Lyophilized reference materials such as those produced ferences in calibration, would be concerning for multi-site by Sero and UTAK require reconstitution in the laboratory, studies estimating population differences in zinc status. For leaving room for variability in zinc content due to varia- example, consider a hypothetical population of children with tion in diluent volume or the completeness of reconstitution a serum zinc concentration of 86 ± 13 µg/dL, and a cutoff for into solution. To avoid this inherent limitation, the authors zinc deficiency of < 65 µg/dL. A 15% difference in calibra - sought a reference material that did not require reconstitu- tion, where one laboratory was reporting zinc values 10% tion before use. NIST SRM 1950, Metabolites in Frozen below the true value, and the other 5% above, would cause Human Plasma, has a certified zinc content of 0.698 mg zinc the prevalence of zinc deficiency to vary from 2 to 17%. per kilogram (71.3 µg/dL based on the certified density). Synchronization of instrument calibration, and verification This zinc concentration is also more comparable to typical that instruments maintain calibration over time, is essential cutoffs for zinc deficiency. in order to avoid such bias. Surprisingly, measures of the zinc concentration in SRM Serum or plasma zinc is the only recommended bio- 1950 had the greatest error, and were biased above the ref- marker of zinc status in human populations [2]. However, erence value. Reasons behind this finding were not deter - in the current study, zinc measures in the donor serum and mined. Although high zinc values may be due to a number of plasma yielded the lowest precision of any of the sample factors, they are often due to contamination. If zinc contami- matrices analyzed. Although we lack data on the specific nation were introduced in the laboratories where the vial was cause, one possibility is the formation of clots or precipitates initially opened, it is expected that other samples handled in in vitro. Fibrin and fibrinogen have affinity for zinc [ 12]. the same way would show a similar bias. This was not the Zinc containing proteins, notably alpha-2-macroglobulin, case. The rubber stopper used in the SRM 1950 ampoule, are also present in clots [13]. Although clot formation and however, may be a source of contamination. Sealing rub- precipitates are causes for sample rejection for coagulation ber typically contains zinc oxide and other zinc compounds biomarkers [14], we are not aware of zinc methods specify- [16], and rubber stoppers are documented contaminators of ing this rejection. We propose that our observation of higher zinc and other metals in sterile pharmaceutical preparations variability in these samples warrants further study. using sealed vials similar to those used for SRM 1950 [17]. Care should also be taken to ensure adequate choice in Upon their review of these data, NIST initiated the process reference material for verification of accuracy. For exam - of removing the zinc value for SRM 1950 [18]. ple, Seronorm Trace Elements, the most commonly men- This study has several strengths. Multiple zinc analyti- tioned reference material for serum and plasma zinc analysis cal laboratories were included in low- and high-income based on our literature review, was available in two zinc countries around the world, reflecting a global collabora - concentration levels, i.e., 110 and 162 µg/dL. These values tive effort. A thorough literature review and input from A. G. Hall et al. already been assigned to the Author Accepted Manuscript version that multiple experienced zinc analytical laboratories informed might arise from this submission. the design of the laboratory activity to ensure its appropri- ateness. The analysis considered multiple issues, including Data Availability Supporting material are available as a supplement. calibration and differences by sample type. And, impor - Data will be made available upon request to the corresponding author. tantly, tight control was exerted over materials and pro- cess, so that the cause of significant differences could be Declarations effectively evaluated. There are also several limitations to this study and the Ethical Approval Not applicable (no human subjects). interpretation of its results. Budget limited the study to Consent to Participate. Not applicable (no human subjects). a small sample size, with only one ICP-OES instrument, limiting statistical comparisons. However, ICP-OES is also Consent to Publish. Not applicable (no human subjects). the least commonly used instrument, accounting for 15% of studies identified in our literature review, nearly half of Conflict of Interest The authors declare no competing interests. which used the same ICP-OES instrument included in the present study. Another limitation is that calibration was only Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- assessed at one concentration. Error in serially diluted cali- tion, distribution and reproduction in any medium or format, as long brators depends on concentration, and additional concentra- as you give appropriate credit to the original author(s) and the source, tions would allow more precise quantification of calibration provide a link to the Creative Commons licence, and indicate if changes error. Future synchronization activities between laboratories were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated should therefore include pre-diluted standards over the range otherwise in a credit line to the material. If material is not included in of the standard curve. the article's Creative Commons licence and your intended use is not Clinicians and public health scientists rely on data com- permitted by statutory regulation or exceeds the permitted use, you will paring the nutritional status of population groups to inform need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . the design, implementation, and evaluation of nutrition interventions. Previous research has established the impor- tance of minimizing analytical variability for biomarkers of nutritional status [19]. Studies revealing analytical variabil- References ity in specific nutritional biomarkers between laboratories have led to programs for the harmonization of laboratory 1. Wessells KR, Brown KH (2012) Estimating the global preva- methods for folate [20, 21], vitamin A [22], and vitamin D lence of zinc deficiency: results based on zinc availability in [23–25]. Future studies evaluating the health consequences national food supplies and the prevalence of stunting. PLoS ONE of zinc deficiency, as well as the impact of interventions to 7(11):e50568. https:// doi. org/ 10. 1371/ journ al. pone. 00505 68 2. King JC, Brown KH, Gibson RS, Krebs NF, Lowe NM, Siek- correct zinc deficiency, will require the comparison of zinc mann JH, Raiten DJ (2016) Biomarkers of Nutrition for Develop- concentrations in serum, plasma, and food sources in diverse ment (BOND)-zinc review. J Nutr. https://d oi.o rg/1 0.3 945/j n.1 15. settings. Our results add to the justification for the further harmonization of laboratories analyzing zinc. 3. Maret W (2017) Zinc in cellular regulation: the nature and signifi - cance of “zinc signals.” Int J Mol Sci. 18(11):2285. https:// doi. org/ 10. 3390/ ijms1 81122 85 Supplementary Information The online version contains supplemen- 4. Prasad AS (2013) Discovery of human zinc deficiency: its impact tary material available at https://doi. or g/10. 1007/ s12011- 021- 02883-z . on human health and disease. Adv Nutr 4(2):176–190. https://doi. org/ 10. 3945/ an. 112. 003210 Acknowledgements The authors acknowledge Junaid Iqbal, Kehkashan 5. King JC, Shames DM, Lowe NM, Woodhouse LR, Sutherland B, Begum, David Killilea, Md. Munirul Islam, Rubhana Raqib, Barbara Abrams SA, Turnlund JR, Jackson MJ (2001) Effect of acute zinc Stoecker, Workhneh Agidew, Christophe Zeder, Austin Cole, and Jamie depletion on zinc homeostasis and plasma zinc kinetics in men. Westcott, for input on methods and efforts analyzing samples. We also Am J Clin Nutr 74(1):116–124. https://doi. or g/10. 1093/ a jcn/74.1. thank CDC and USDA for providing zinc analytical methods developed by their respective institutions. 6. Brown KH, Rivera JA, Bhutta Z, Gibson RS, King JC, Lonnerdal B, Ruel MT, Sandstrom B, Wasantwisut E, Hotz C (2004) Inter- Author Contribution A. G. H., C. M. M., and J. C. K. designed the national Zinc Nutrition Consultative Group (IZiNCG) technical study; A. G. H. conducted the research; A. G. H. and C. M. M. analyzed document #1 Assessment of the risk of zinc deficiency in popula - the data; A. G. H., C. M. M., and J. C. K. wrote the manuscript. tions and options for its control. Food nutrition bullet 25(1):Suppl 2-S99 203 Funding This study was funded by grants OPP1150161 and INV- 7. Brown KH (1998) Effect of infections on plasma zinc concentra - 007767 from the Bill and Melinda Gates Foundation. This work was tion and implications for zinc status assessment in low-income supported, in whole or in part, by the Bill and Melinda Gates Founda- countries. Am J Clin Nutr 68(2 Suppl):425S-429S. https://do i. tion (Grant Number: INV-007767). Under the grant conditions of the org/ 10. 1093/ ajcn/ 68.2. 425S Foundation, a Creative Commons Attribution 4.0 Generic License has Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… 8. Wessells KR, Peerson JM, Brown KH (2019) Within-individual 18. Yu LL (2020) NIST e-mail communication regarding certified differences in plasma ferritin, retinol-binding protein, and zinc zinc value for SRM 1950 concentrations in relation to inflammation observed during a 19. Blanck HM, Bowman BA, Cooper GR, Myers GL, Miller DT short-term longitudinal study are similar to between-individ- (2003) Laboratory issues: use of nutritional biomarkers. J Nutr ual differences observed cross-sectionally. Am J Clin Nutr 133(Suppl 3):888S-894S. https:// doi. org/ 10. 1093/ jn/ 133.3. 888S 109(5):1484–1492. https:// doi. org/ 10. 1093/ ajcn/ nqz014 20. Gunter EW, Bowman BA, Caudill SP, Twite DB, Adams MJ, 9. IZiNCG Technical Brief 2, 2nd ed.: Assessing population zinc sta- Sampson EJ (1996) Results of an international round robin for tus with serum zinc concentration (2012). Davis, CA: University serum and whole-blood folate. Clin Chem 42(10):1689–1694 of California. www. izincg. org. Accessed 30 Dec 2019 21. Pfeiffer CM, Zhang M, Jabbar S (2018) Framework for laboratory 10. Multi-Elements in Serum by ICP-DRC-MS. DLS 3006.8–03 harmonization of folate measurements in low- and middle-income (2018). Inorganic radiation analytical toxicology, Division of countries and regions. Ann N Y Acad Sci 1414(1):96–108. https:// Laboratory Sciences, National Center for Environmental Health, doi. org/ 10. 1111/ nyas. 13532 Centers for Disease Control and Prevention, Atlanta, Georgia, 22. Haynes BM, Schleicher RL, Jain RB, Pfeiffer CM (2008) The USA CDC VITAL-EQA program, external quality assurance for serum 11. Determination of Metals by ICP-MS and ICP-OES. CLG-TM3.05 retinol, 2003–2006. Clin chimica acta; Int J Clin Chem 390(1– (2016). Office of Public Health and Science, Food Safety and 2):90–96. https:// doi. org/ 10. 1016/j. cca. 2008. 01. 009 Inspection Service, United States Department of Agriculture, 23. Binkley N, Krueger D, Cowgill CS, Plum L, Lake E, Hansen KE, Washington, D.C., USA DeLuca HF, Drezner MK (2004) Assay variation confounds the 12. Marx G (1988) Zinc binding to fibrinogen and fibrin. Arch Bio - diagnosis of hypovitaminosis D: a call for standardization. J Clin chem Biophys 266(1):285–288. https:// doi. org/ 10. 1016/ 0003- Endocrinol Metab 89(7):3152–3157. https:// doi. org/ 10. 1210/ jc. 9861(88) 90260-32003- 031979 13. Stachowicz A, Siudut J, Suski M, Olszanecki R, Korbut R, Undas 24. Binkley N, Krueger DC, Morgan S, Wiebe D (2010) Current sta- A, Wisniewski JR (2017) Optimization of quantitative proteomic tus of clinical 25 hydroxyvitamin D measurement an assessment analysis of clots generated from plasma of patients with venous of between-laboratory agreement. Clin chimica acta Int J Clin thromboembolism. Clin Proteomics 14:38. https:// doi. or g/ 10. Chem 411(23–24):1976–1982. https://doi. or g/10. 1016/j. cca. 2010. 1186/ s12014- 017- 9173-x08. 018 14. Magnette A, Chatelain M, Chatelain B, Ten Cate H, Mullier F 25. Sempos CT, Heijboer AC, Bikle DD, Bollerslev J, Bouillon R, (2016) Pre-analytical issues in the haemostasis laboratory: guid- Brannon PM, DeLuca HF, Jones G, Munns CF, Bilezikian JP, ance for the clinical laboratories. Thromb J 14:49. https://doi. or g/ Giustina A, Binkley N (2018) Vitamin D assays and the defini - 10. 1186/ s12959- 016- 0123-z tion of hypovitaminosis D: results from the First International 15. Pilch SM, Senti FR (1985) Analysis of zinc data from the second Conference on Controversies in Vitamin D. Br J Clin Pharmacol National Health and Nutrition Examination Survey (NHANES 84(10):2194–2207. https:// doi. org/ 10. 1111/ bcp. 13652 II). J Nutr 115(11):1393–1397. https://doi. or g/10. 1093/ jn/ 115. 11. 1393 Publisher's note Springer Nature remains neutral with regard to 16. Chandrasekaran VC (2010) Rubbers, Chemicals and Compound- jurisdictional claims in published maps and institutional affiliations. ing for ‘O’ Rings and Seals, in Rubber Seals for Fluid and Hydrau- lic Systems. William Andrew, Oxford 17. Li X, Qian P (2017) Identification of an e xposure risk to heavy metals from pharmaceutical-grade rubber stoppers. J Food Drug Anal 25(3):723–730. https:// doi. org/ 10. 1016/j. jfda. 2016. 07. 008 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Biological Trace Element Research Springer Journals

Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP-OES, and ICP-MS in Diverse Laboratory Settings

Download PDF
8 pages

Loading next page...
 
/lp/springer-journals/comparison-of-serum-plasma-and-liver-zinc-measurements-by-aas-icp-oes-XTN5AHTerv

References (24)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
ISSN
0163-4984
eISSN
1559-0720
DOI
10.1007/s12011-021-02883-z
Publisher site
See Article on Publisher Site

Abstract

Progress improving zinc nutrition globally is slowed by limited understanding of population zinc status. This challenge is compounded when small differences in measurement can bias the determination of zinc deficiency rates. Our objec - tive was to evaluate zinc analytical accuracy and precision among different instrument types and sample matrices using a standardized method. Participating laboratories analyzed zinc content of plasma, serum, liver samples, and controls, using a standardized method based on current practice. Instrument calibration and drift were evaluated using a zinc standard. Accuracy was evaluated by percent error vs. reference, and precision by coefficient of variation (CV). Seven laboratories in 4 countries running 9 instruments completed the exercise: 4 atomic absorbance spectrometers (AAS), 1 inductively coupled plasma optical emission spectrometer (ICP-OES), and 4 ICP mass spectrometers (ICP-MS). Calibration differed between individual instruments up to 18.9% (p < 0.001). Geometric mean (95% CI) percent error was 3.5% (2.3%, 5.2%) and CV was 2.1% (1.7%, 2.5%) overall. There were no significant differences in percent error or CV among instrument types ( p = 0.91, p = 0.15, respectively). Among sample matrices, serum and plasma zinc measures had the highest CV: 4.8% (3.0%, 7.7%) and 3.9% (2.9%, 5.4%), respectively (p < 0.05). When using standardized materials and methods, similar zinc concentration values, accuracy, and precision were achieved using AAS, ICP-OES, or ICP-MS. However, method development is needed for improvement in serum and plasma zinc measurement precision. Differences in calibration among instruments demonstrate a need for harmonization among laboratories. Keywords Plasma zinc · Zinc deficiency · Accur acy · Precision · Harmonization Introduction and growth [2–4]. In the absence of a severe zinc deficiency, circulating zinc is highly conserved [5], and there is no sen- Nearly 20% of the global population is at risk of inadequate sitive and specific indicator of zinc status in individuals [ 2, zinc intake [1]. Zinc is an essential micronutrient with 6]. The distribution of plasma or serum zinc concentration unique roles in protein structure and as a cofactor in sub- within a population does, however, allow the determination strate binding and enzymatic activity. Due to these molecu- of population zinc status [6]. Therefore, plasma or serum lar roles, zinc contributes to broad range of functions includ- zinc concentration is one of three recommended biomarkers ing DNA transcription and repair, cell signaling, energy for assessment of zinc status at the population level along metabolism, immune and central nervous system function, with dietary zinc intake and the prevalence of stunting among children under 5 [2]. Future progress towards reducing zinc deficiency glob - * Andrew G. Hall ally will depend on monitoring changes in plasma or serum [email protected]; [email protected] zinc status. Apart from dietary zinc intake, a number of fac- Benioff Children’s Hospitals; Children’s Hospital Oakland tors affect plasma or serum zinc concentration, including Research Institute, University of California San Francisco, systemic inflammation, time of specimen collection, fast - Oakland, CA, USA ing status, and variations in the processing and handling of Department of Nutritional Sciences and Toxicology, samples [2, 6–8]. To minimize the impact of some of these University of California Berkeley, Berkeley, CA, USA issues, recommendations have been made for the design of Department of Pediatrics, School of Medicine, University of California San Francisco, San Francisco, CA, USA Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… human studies assessing zinc status, and procedures for col- The methods document was sent to the three laboratories for lecting samples for zinc analysis [6, 9]. review and revised according to their feedback. The extent that analytical method and instrument may further contribute to variation in reported plasma or serum Laboratory Exercise A laboratory exercise was designed zinc values has seen limited study. The instruments typically for comparison of accuracy and precision of zinc concentra- used for zinc analysis vary in sensitivity and complexity tion measurements among different instruments and sample of operation. They include atomic absorption spectrome- matrices selected based on the literature review. The labo- ters (AAS, flame or graphite furnace), inductively coupled ratory exercise was designed to minimize variability due plasma optical (atomic) emission spectrometers (ICP-OES/ to sample preparation, reagent quality, and zinc reference ICP-AES), and ICP mass spectrometers (ICP-MS). Our pri- material. In addition to the Methods Reference Document, mary aim was an evaluation of accuracy and precision by specific instructions for instrument calibration, sample prep - instrument type and by sample type in diverse laboratory aration, and a standard set of materials for analysis were settings where zinc analysis of biological specimens from included. large surveys or clinical studies are typically conducted. Laboratories in low- and middle-income countries that had previously conducted zinc analysis studies for national- level surveys, and laboratories that had contributed proto- Materials and Methods cols for inclusion in the Methods Reference Document, were invited to participate in the laboratory exercise. Participat- We developed a standardized method based on a review of ing laboratories were instructed to use the identical sets of recent human studies, and instructed participating laborato- materials and supplies assembled at CHORI and shipped to ries to implement the method for the analysis of a standard- each laboratory. These included trace element analysis grade ized set of samples and controls. 68–70% nitric acid (Omnitrace, EMD Millipore, Burlington, MA; or BDH Aristar Plus, VWR International, Radnor, PA), Literature Review and Methods Reference Document To ultrapure water (Omnitrace Ultra, EMD Millipore, Burling- determine the typical sample matrices, preparation meth- ton, MA), and filter pipette tips (VWR International, Radnor, ods, and analytical methods for the measurement of zinc in PA) to minimize particulate contamination from the pipette human studies, a literature review was conducted. Human mechanism. studies published over the previous 5-years were identified Laboratories calibrated their instruments using the pro- in PubMed (National Center for Biotechnology Information, vided Standard Reference Material (SRM) 3168a zinc in Bethesda, MD) using the single search term “zinc” and fil - 10% nitric acid (National Institute of Standards and Technol- ters for human studies dated between February 2013 and ogy (NIST), Gaithersburg, MD). To detect potential differ - February 2018. Studies written in English with abstracts ences in calibration, and to assess instrument drift during the reporting the measurement of zinc in human derived sam- analysis, a dilute solution of SRM 3168a, containing 12.5 µg ples or foods were downloaded for further review. Articles zinc/dL in 5% nitric acid, was prepared in bulk at CHORI without zinc analytical data in the full text were excluded. A using the same ultrapure water and nitric acid provided to detailed description of the literature review is provided in the the participating laboratories. accompanying online material (Supplemental Appendix A). Sample matrices included human serum and plasma, and Three laboratories prominent in the literature review were powdered bovine liver (representative of prepared food com- contacted and their zinc analysis protocol(s) requested: Lab- posites). Unknowns (i.e., samples without a known value) oratory of Human Nutrition at the Institute of Food, Nutri- for each, as well as reference materials with certified values, tion and Health, Swiss Federal Institute of Technology were included for analysis. Unknowns or reference materials (Zurich, Switzerland); Section of Pediatric Nutrition, Uni- were dispensed into vials at CHORI prior to distribution to versity of Colorado School of Medicine (Aurora, CO); and the laboratories as follows: All vials were acid-washed prior Children’s Hospital Oakland Research Institute (CHORI), to dispensing. Human plasma and human serum unknowns UCSF Benioff Children’s Hospitals (Oakland, CA). Meth - (Zenbio, Research Triangle Park, NC) were shipped unfro- odological elements of zinc analysis protocols from these zen on cold pack from North Carolina to CHORI overnight laboratories, in addition to protocols provided by the Centers the day of collection, mixed and aliquoted into screw-cap for Disease Control and Prevention (CDC, Atlanta, GA) [10] polypropylene vials, and then frozen. The liver unknown, and the United States Department of Agriculture (USDA, food grade powdered bovine liver (CurEase, McEwen, TN), Washington, DC) [11], were reviewed and assembled into a was mixed for 5 min using a food processor (Cuisinart, Stan- Methods Reference Document (Supplemental Appendix B) ford, CT) prior to dispensing into polypropylene screwcap of techniques for determining zinc concentration in samples vials. from human studies using AAS, ICP-OES, and ICP-MS. A. G. Hall et al. Reference materials included SRM 1950 human plasma published (reference) zinc concentration for each reference and SRM 1577c powdered bovine liver (NIST, Gaithersburg, material: MD), Seronorm Trace Elements in Human Serum Levels 1 [Zn] − [Zn] measured reference and 2 (SERO AS, Billingstad, Norway), and a custom UTAK %error = × 100 [Zn] reference human plasma containing no added zinc (UTAK Laborato- ries, Valencia, CA). The SRM 1950 serum was distributed Percent bias was defined as the relative difference to each laboratory frozen in the sealed 1-mL glass ampoules between the zinc concentration measured, and the published with rubber stoppers, as provided by NIST. Lyophilized (reference) zinc concentration for each reference material Seronorm serum and UTAK plasma were reconstituted at (maintaining the sign of the difference from reference): CHORI using ultrapure water (Omnitrace Ultra, EMD Mil- [Zn] − [Zn] lipore, Burlington, MA) according to the manufacturer’s measured reference %bias = × 100 instructions, and dispensed into screw cap polypropylene [Zn] reference vials. The SRM 1577c bovine liver was also dispensed into The overall CV, % error, and % bias for a given instru- screwcap polypropylene vials prior to distribution to the ment were determined as the mean (or geometric mean) of laboratories. the respective indicator for all samples analyzed by each All sera and plasma were shipped on dry ice from CHORI instrument. The overall CV, % error, and % bias for a given to each participating laboratory. Zinc solutions, powdered sample matrix was determined as the overall mean (or geo- liver, ultrapure water, and pipette tips were shipped sepa- metric mean) value for the respective sample matrix. rately from CHORI at ambient temperature. The trace ele- All data were tabulated; % error, % bias, and CV were ment grade nitric acid was shipped directly from the supplier calculated using Microsoft Excel 2010 (Microsoft, Red- to each participating laboratory. To minimize the potential mond, WA). Descriptive statistics were calculated and for bias, individual laboratories were not informed of the statistical comparisons performed using SPSS 26 (IBM, known zinc values of any of the materials sent, with the Armonk, NY). Data were tested for normal distribution and exception of the pre-diluted zinc standard, reported as falling descriptive statistics calculated prior to statistical compari- within a range between 10 and 15 µg zinc/dL. Laboratories son by unpaired t-test or ANOVA. Unless stated otherwise, measured the zinc concentration of the pre-diluted zinc solu- all data are formatted in the text as “mean ± standard devia- tion in triplicate at the beginning and end of the analytical tion” or, where not normally distributed, as “geometric mean run, that of each reference material in triplicate, and each (95% confidence interval).” ICP-OES was not included in unknown 9 times. To minimize variability in viscosity that comparisons between instrument types, since there was could affect instrument sampling flow rates, a simple proto - only one such instrument. Values for CV and % error were col for digestion of serum and plasma in concentrated nitric log-transformed to achieve normal distribution prior to acid was specified. statistical comparison. Statistically significant differences were defined as p < 0.05 for all comparisons. As the study Data Analysis Instrument calibration was based on the was not designed to evaluate the accuracy and precision of measurement of zinc concentration in the pre-diluted SRM individual laboratories, the individual laboratories and their 3168a zinc solution for each instrument. Percent drift was individual instruments are not identified with respect to ana - defined as the relative change in measurement of the zinc lytical results. solution from start to end of the analytical run: [Zn] − [Zn] end start %drift = × 100% [Zn] start Results Precision was evaluated for all reference materials and unknowns by determining the coefficient of variation (CV), Literature Review Detailed results of the literature review i.e., sample mean divided by the sample standard deviation are provided in the supplemental material (Supplemental for each sample analyzed, expressed as a percentage: Appendix A). Briefly, out of 470 PubMed search hits, 134 peer-reviewed journal articles met the criteria for review. CV = × 100% Zinc concentration was most frequently determined in serum (n = 69), plasma (n = 48), and food composites (n = 12). Accuracy was evaluated using % error and % bias. Percent Instruments for zinc quantification included AAS ( n = 78 error was defined as the absolute value of the relative dif - AAS), ICP-MS (n = 22), and ICP-OES (n = 16). While 38 of ference between the zinc concentration measured, and the the studies using AAS described the instrument as a flame AAS, only three described using a graphite furnace AAS. Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… Seven studies used a plate reader or auto-analyzer for indi- rect zinc determination utilizing a zinc sensitive chemical dye or probe. Laboratory Exercise Seven laboratories in 4 countries, running 9 individual instruments, participated. The labora- tories and their instruments are listed in Table 1. Instrument calibration was assessed by measuring the zinc concentra- tion of the pre-diluted SRM 3168a solution (Fig. 1). Zinc concentration was 12.2 ± 0.7 µg/dL overall, and did not differ significantly ( p = 0.8) between AAS (12.0 ± 0.5 µg/dL, n = 4) and ICP MS (12.4 ± 1.0 µg/dL, n = 4). None of the individual instruments varied more than 10% from the expected value of 12.5 µg/dL. However, differences in calibration between individual instruments were as high as 18.9% (p < 0.001). Fig. 1 Instrument calibration. Pre-diluted SRM 3168a values from The pre-diluted zinc standard was measured again at the each instrument, with expected concentration (12.5  µg/dL, dotted end of the analytical run and instrument drift calculated rela- horizontal line), and range of ± 10% from the expected value (hori- zontal dashed lines). Data are displayed as mean ± standard deviation. tive to the initial values. The mean drift was 0.4% ± 3.0% Lowercase letters a, b, and c denote homogenous subsets where dif- (n = 8), and did not differ significantly ( p = 0.3) between fering letters represent statistically significant differences between AAS (1.8% ± 3.5%, n = 4) and ICP-MS-1.7% ± 1.7%, n = 3). individual instruments (p < 0.05) One AAS instrument drifted by more than 5% from its initial values (6.9% drift, from 11.8 ± 0.53 µg/dL to 12.6 ± 0.28 µg/ dL, p < 0.05). Data on drift were missing from one ICP-MS Serum and plasma unknowns yielded higher measure- ment variability (lower precision) compared with other sam- instrument. Zinc concentration of plasma, serum, and zinc content of ple matrices (Fig. 3). Measures of zinc concentration of the unknown serum had a CV of 4.8% (3.0%, 7.7%), and were liver samples is reported, according to each instrument type, in Fig. 2. No significant differences were observed between less precise (higher CV) than the pre-diluted SRM 3168a zinc standard solution, CV of 1.4% (0.7%, 2.9%) (p = 0.042), AAS and ICP-MS for any material measured. Precision between instrument types and samples was evaluated using and compared with SRM 1577c liver, CV of 1.1% (0.6%, 2.2%) (p = 0.006). Zinc measures of unknown plasma had the CV. The overall CV for each instrument was determined for all samples analyzed by that instrument. The instrument a CV of 3.9% (2.9%, 5.4%), and were less precise (higher) than SRM 1577c liver (p = 0.031). CV was 1.7% (0.8%, 3.7%) for AAS, 0.9% for ICP-OES, and 2.0% (1.1%, 3.4%) for ICP-MS, 1.7% (1.2%, 2.4%) overall. Accuracy between instrument types and samples was evaluated by the measurement error and bias. Reference val- No differences in precision were detected between AAS and ICP-MS (p = 0.61, n = 4 each for AAS and ICP-MS). ues for zinc used in calculation of accuracy were Seronorm Table 1 Participating laboratories and instruments Institute Location Instrument(s) Aga Khan University, Nutrition Research Laboratory Karachi, Pakistan iCE 3000 Flame AAS (Thermo Fisher Scientific, Waltham, MA) Children’s Hospital Oakland Research Institute, UCSF Oakland, California, USA 5100 SVDV ICP-OES (Agilent Technologies, Santa Benioff Children’s Hospitals Clara, CA) International Centre for Diarrhoeal Disease Research Dhaka, Bangladesh AA-7000 Flame AAS (Shimadzu, Kyoto, Japan) Oklahoma State University, Department of Nutritional Stillwater, Oklahoma, USA ELAN 9000 ICP-MS (PerkinElmer, Waltham, MA) Sciences Laboratory of Human Nutrition; Institute of Food, Zurich, Switzerland 240FS Flame AAS (Agilent Technologies, Santa Clara, Nutrition and Health; Swiss Federal Institute of CA), Technology iCAP RQ ICP-MS (Thermo Fisher Scientific, Waltham, MA) Interdisciplinary Center for Plasma Mass Spectrometry; Davis, California, USA 8900 ICP-MS (Agilent Technologies, Santa Clara, CA) University of California, Davis Section of Pediatric Nutrition, University of Colorado Aurora, Colorado, USA AAnalyst 700 Flame AAS (PerkinElmer, Waltham, MA), School of Medicine 7700 ICP-MS (Agilent Technologies, Santa Clara, CA) A. G. Hall et al. Fig. 2 Zinc concentration of reference and unknown serum, plasma, and liver. Zinc concentration of each serum, plasma, and liver sample was determined by AAS (n = 4), ICP-OES (n = 1), and ICP-MS (n = 4). Data are displayed as mean ± standard deviation. No differences were detected between AAS and ICP-MS for any material, or overall Fig. 3 Zinc measurement precision by material analyzed. Data are expressed as geomet- ric mean and 95% confidence interval. Lowercase letters a, b, and c denote homogenous subsets where differing letters represent statistically significant differences ( p < 0.05) L1 serum, 109.7 µg/dL; Seronorm L2 serum, 161.7 µg/dL; measures of SRM 1950 plasma (10.8% ± 20.6%) were biased UTAK plasma, 59.4 µg/dL; SRM 1950 plasma, 71.3 µg/ high compared with Seronorm L1 serum (− 1.3% ± 5.8%, dL; and SRM 1577c liver, 181.1 µg/g. The reference value p < 0.05), UTAK plasma (− 4.9% ± 7.6%, p < 0.01), and for SRM 1950 plasma was converted to µg/dL using the SRM 1577c liver (− 3.0% ± 4.6%, p < 0.01) (Fig. 4(B)). reported density of 1.02086 g/mL from the certificate of analysis. Error for all instruments was 3.5% (2.2%, 5.6%) overall: 3.3% (0.8%, 14.1%) for AAS, 4.4% for ICP-OES, and 3.4% Discussion (1.8%, 6.5%) for ICP-MS. No significant difference in error was observed between AAS and ICP-MS (p = 0.95, n = 4 This is the first multi-site evaluation of accuracy and preci - each). Higher error was observed with SRM 1950 plasma sion of elemental zinc analysis in samples typical of human compared with Seronorm L1, 10.2% (4.6%, 22.5%) vs. 1.7% studies that the authors are aware of. Comparable results (0.5%, 5.8%), respectively (p < 0.05) (Fig. 4(A)). for the measurement of zinc were achievable with a variety The bias for all instruments was 1.0% ± 6.4% overall: of instrumentation in geographically diverse settings, using 2.0% ± 9.7% for AAS, 0.3% for ICP-OES, and 0.1% ± 3.8% methods and samples typical of human research. However, for ICP-MS. No significant difference in bias was observed the results for calibration, accuracy, and precision under- between AAS and ICP-MS (p = 0.74, n = 4 each). Overall, score potential pitfalls and areas for improvement in the Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… Fig. 4 Zinc measurement accuracy by reference material analyzed. Lowercase letters a and b denote homogenous subsets where differing letters represent statistically significant differences between materials (p < 0.05). (A) Percent error is expressed as geometric mean and 95% CI. (B) Percent bias is expressed as mean ± SEM implementation of zinc analytical methods in the clinical are substantially higher than the recommended cutoffs for research setting. zinc deficiency (i.e., from 57 to 74 µg/dL, depending on the Although differences in calibration were as high as 18% population, fasting status, and time of collection), as well among the individual instruments, each instrument was as the serum zinc concentrations of the populations from within 10% of the expected value, an accepted cutoff for which the cutoffs were derived (ranging from 76 ± 16% to calibration verification [ 10]. Nonetheless, variation in the 98 ± 14% µg/dL (geometric mean ± CV)) [6, 15]. number estimated to be zinc deficient, resulting from dif - Lyophilized reference materials such as those produced ferences in calibration, would be concerning for multi-site by Sero and UTAK require reconstitution in the laboratory, studies estimating population differences in zinc status. For leaving room for variability in zinc content due to varia- example, consider a hypothetical population of children with tion in diluent volume or the completeness of reconstitution a serum zinc concentration of 86 ± 13 µg/dL, and a cutoff for into solution. To avoid this inherent limitation, the authors zinc deficiency of < 65 µg/dL. A 15% difference in calibra - sought a reference material that did not require reconstitu- tion, where one laboratory was reporting zinc values 10% tion before use. NIST SRM 1950, Metabolites in Frozen below the true value, and the other 5% above, would cause Human Plasma, has a certified zinc content of 0.698 mg zinc the prevalence of zinc deficiency to vary from 2 to 17%. per kilogram (71.3 µg/dL based on the certified density). Synchronization of instrument calibration, and verification This zinc concentration is also more comparable to typical that instruments maintain calibration over time, is essential cutoffs for zinc deficiency. in order to avoid such bias. Surprisingly, measures of the zinc concentration in SRM Serum or plasma zinc is the only recommended bio- 1950 had the greatest error, and were biased above the ref- marker of zinc status in human populations [2]. However, erence value. Reasons behind this finding were not deter - in the current study, zinc measures in the donor serum and mined. Although high zinc values may be due to a number of plasma yielded the lowest precision of any of the sample factors, they are often due to contamination. If zinc contami- matrices analyzed. Although we lack data on the specific nation were introduced in the laboratories where the vial was cause, one possibility is the formation of clots or precipitates initially opened, it is expected that other samples handled in in vitro. Fibrin and fibrinogen have affinity for zinc [ 12]. the same way would show a similar bias. This was not the Zinc containing proteins, notably alpha-2-macroglobulin, case. The rubber stopper used in the SRM 1950 ampoule, are also present in clots [13]. Although clot formation and however, may be a source of contamination. Sealing rub- precipitates are causes for sample rejection for coagulation ber typically contains zinc oxide and other zinc compounds biomarkers [14], we are not aware of zinc methods specify- [16], and rubber stoppers are documented contaminators of ing this rejection. We propose that our observation of higher zinc and other metals in sterile pharmaceutical preparations variability in these samples warrants further study. using sealed vials similar to those used for SRM 1950 [17]. Care should also be taken to ensure adequate choice in Upon their review of these data, NIST initiated the process reference material for verification of accuracy. For exam - of removing the zinc value for SRM 1950 [18]. ple, Seronorm Trace Elements, the most commonly men- This study has several strengths. Multiple zinc analyti- tioned reference material for serum and plasma zinc analysis cal laboratories were included in low- and high-income based on our literature review, was available in two zinc countries around the world, reflecting a global collabora - concentration levels, i.e., 110 and 162 µg/dL. These values tive effort. A thorough literature review and input from A. G. Hall et al. already been assigned to the Author Accepted Manuscript version that multiple experienced zinc analytical laboratories informed might arise from this submission. the design of the laboratory activity to ensure its appropri- ateness. The analysis considered multiple issues, including Data Availability Supporting material are available as a supplement. calibration and differences by sample type. And, impor - Data will be made available upon request to the corresponding author. tantly, tight control was exerted over materials and pro- cess, so that the cause of significant differences could be Declarations effectively evaluated. There are also several limitations to this study and the Ethical Approval Not applicable (no human subjects). interpretation of its results. Budget limited the study to Consent to Participate. Not applicable (no human subjects). a small sample size, with only one ICP-OES instrument, limiting statistical comparisons. However, ICP-OES is also Consent to Publish. Not applicable (no human subjects). the least commonly used instrument, accounting for 15% of studies identified in our literature review, nearly half of Conflict of Interest The authors declare no competing interests. which used the same ICP-OES instrument included in the present study. Another limitation is that calibration was only Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- assessed at one concentration. Error in serially diluted cali- tion, distribution and reproduction in any medium or format, as long brators depends on concentration, and additional concentra- as you give appropriate credit to the original author(s) and the source, tions would allow more precise quantification of calibration provide a link to the Creative Commons licence, and indicate if changes error. Future synchronization activities between laboratories were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated should therefore include pre-diluted standards over the range otherwise in a credit line to the material. If material is not included in of the standard curve. the article's Creative Commons licence and your intended use is not Clinicians and public health scientists rely on data com- permitted by statutory regulation or exceeds the permitted use, you will paring the nutritional status of population groups to inform need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . the design, implementation, and evaluation of nutrition interventions. Previous research has established the impor- tance of minimizing analytical variability for biomarkers of nutritional status [19]. Studies revealing analytical variabil- References ity in specific nutritional biomarkers between laboratories have led to programs for the harmonization of laboratory 1. Wessells KR, Brown KH (2012) Estimating the global preva- methods for folate [20, 21], vitamin A [22], and vitamin D lence of zinc deficiency: results based on zinc availability in [23–25]. Future studies evaluating the health consequences national food supplies and the prevalence of stunting. PLoS ONE of zinc deficiency, as well as the impact of interventions to 7(11):e50568. https:// doi. org/ 10. 1371/ journ al. pone. 00505 68 2. King JC, Brown KH, Gibson RS, Krebs NF, Lowe NM, Siek- correct zinc deficiency, will require the comparison of zinc mann JH, Raiten DJ (2016) Biomarkers of Nutrition for Develop- concentrations in serum, plasma, and food sources in diverse ment (BOND)-zinc review. J Nutr. https://d oi.o rg/1 0.3 945/j n.1 15. settings. Our results add to the justification for the further harmonization of laboratories analyzing zinc. 3. Maret W (2017) Zinc in cellular regulation: the nature and signifi - cance of “zinc signals.” Int J Mol Sci. 18(11):2285. https:// doi. org/ 10. 3390/ ijms1 81122 85 Supplementary Information The online version contains supplemen- 4. Prasad AS (2013) Discovery of human zinc deficiency: its impact tary material available at https://doi. or g/10. 1007/ s12011- 021- 02883-z . on human health and disease. Adv Nutr 4(2):176–190. https://doi. org/ 10. 3945/ an. 112. 003210 Acknowledgements The authors acknowledge Junaid Iqbal, Kehkashan 5. King JC, Shames DM, Lowe NM, Woodhouse LR, Sutherland B, Begum, David Killilea, Md. Munirul Islam, Rubhana Raqib, Barbara Abrams SA, Turnlund JR, Jackson MJ (2001) Effect of acute zinc Stoecker, Workhneh Agidew, Christophe Zeder, Austin Cole, and Jamie depletion on zinc homeostasis and plasma zinc kinetics in men. Westcott, for input on methods and efforts analyzing samples. We also Am J Clin Nutr 74(1):116–124. https://doi. or g/10. 1093/ a jcn/74.1. thank CDC and USDA for providing zinc analytical methods developed by their respective institutions. 6. Brown KH, Rivera JA, Bhutta Z, Gibson RS, King JC, Lonnerdal B, Ruel MT, Sandstrom B, Wasantwisut E, Hotz C (2004) Inter- Author Contribution A. G. H., C. M. M., and J. C. K. designed the national Zinc Nutrition Consultative Group (IZiNCG) technical study; A. G. H. conducted the research; A. G. H. and C. M. M. analyzed document #1 Assessment of the risk of zinc deficiency in popula - the data; A. G. H., C. M. M., and J. C. K. wrote the manuscript. tions and options for its control. Food nutrition bullet 25(1):Suppl 2-S99 203 Funding This study was funded by grants OPP1150161 and INV- 7. Brown KH (1998) Effect of infections on plasma zinc concentra - 007767 from the Bill and Melinda Gates Foundation. This work was tion and implications for zinc status assessment in low-income supported, in whole or in part, by the Bill and Melinda Gates Founda- countries. Am J Clin Nutr 68(2 Suppl):425S-429S. https://do i. tion (Grant Number: INV-007767). Under the grant conditions of the org/ 10. 1093/ ajcn/ 68.2. 425S Foundation, a Creative Commons Attribution 4.0 Generic License has Comparison of Serum, Plasma, and Liver Zinc Measurements by AAS, ICP‑OES, and ICP‑MS in Diverse… 8. Wessells KR, Peerson JM, Brown KH (2019) Within-individual 18. Yu LL (2020) NIST e-mail communication regarding certified differences in plasma ferritin, retinol-binding protein, and zinc zinc value for SRM 1950 concentrations in relation to inflammation observed during a 19. Blanck HM, Bowman BA, Cooper GR, Myers GL, Miller DT short-term longitudinal study are similar to between-individ- (2003) Laboratory issues: use of nutritional biomarkers. J Nutr ual differences observed cross-sectionally. Am J Clin Nutr 133(Suppl 3):888S-894S. https:// doi. org/ 10. 1093/ jn/ 133.3. 888S 109(5):1484–1492. https:// doi. org/ 10. 1093/ ajcn/ nqz014 20. Gunter EW, Bowman BA, Caudill SP, Twite DB, Adams MJ, 9. IZiNCG Technical Brief 2, 2nd ed.: Assessing population zinc sta- Sampson EJ (1996) Results of an international round robin for tus with serum zinc concentration (2012). Davis, CA: University serum and whole-blood folate. Clin Chem 42(10):1689–1694 of California. www. izincg. org. Accessed 30 Dec 2019 21. Pfeiffer CM, Zhang M, Jabbar S (2018) Framework for laboratory 10. Multi-Elements in Serum by ICP-DRC-MS. DLS 3006.8–03 harmonization of folate measurements in low- and middle-income (2018). Inorganic radiation analytical toxicology, Division of countries and regions. Ann N Y Acad Sci 1414(1):96–108. https:// Laboratory Sciences, National Center for Environmental Health, doi. org/ 10. 1111/ nyas. 13532 Centers for Disease Control and Prevention, Atlanta, Georgia, 22. Haynes BM, Schleicher RL, Jain RB, Pfeiffer CM (2008) The USA CDC VITAL-EQA program, external quality assurance for serum 11. Determination of Metals by ICP-MS and ICP-OES. CLG-TM3.05 retinol, 2003–2006. Clin chimica acta; Int J Clin Chem 390(1– (2016). Office of Public Health and Science, Food Safety and 2):90–96. https:// doi. org/ 10. 1016/j. cca. 2008. 01. 009 Inspection Service, United States Department of Agriculture, 23. Binkley N, Krueger D, Cowgill CS, Plum L, Lake E, Hansen KE, Washington, D.C., USA DeLuca HF, Drezner MK (2004) Assay variation confounds the 12. Marx G (1988) Zinc binding to fibrinogen and fibrin. Arch Bio - diagnosis of hypovitaminosis D: a call for standardization. J Clin chem Biophys 266(1):285–288. https:// doi. org/ 10. 1016/ 0003- Endocrinol Metab 89(7):3152–3157. https:// doi. org/ 10. 1210/ jc. 9861(88) 90260-32003- 031979 13. Stachowicz A, Siudut J, Suski M, Olszanecki R, Korbut R, Undas 24. Binkley N, Krueger DC, Morgan S, Wiebe D (2010) Current sta- A, Wisniewski JR (2017) Optimization of quantitative proteomic tus of clinical 25 hydroxyvitamin D measurement an assessment analysis of clots generated from plasma of patients with venous of between-laboratory agreement. Clin chimica acta Int J Clin thromboembolism. Clin Proteomics 14:38. https:// doi. or g/ 10. Chem 411(23–24):1976–1982. https://doi. or g/10. 1016/j. cca. 2010. 1186/ s12014- 017- 9173-x08. 018 14. Magnette A, Chatelain M, Chatelain B, Ten Cate H, Mullier F 25. Sempos CT, Heijboer AC, Bikle DD, Bollerslev J, Bouillon R, (2016) Pre-analytical issues in the haemostasis laboratory: guid- Brannon PM, DeLuca HF, Jones G, Munns CF, Bilezikian JP, ance for the clinical laboratories. Thromb J 14:49. https://doi. or g/ Giustina A, Binkley N (2018) Vitamin D assays and the defini - 10. 1186/ s12959- 016- 0123-z tion of hypovitaminosis D: results from the First International 15. Pilch SM, Senti FR (1985) Analysis of zinc data from the second Conference on Controversies in Vitamin D. Br J Clin Pharmacol National Health and Nutrition Examination Survey (NHANES 84(10):2194–2207. https:// doi. org/ 10. 1111/ bcp. 13652 II). J Nutr 115(11):1393–1397. https://doi. or g/10. 1093/ jn/ 115. 11. 1393 Publisher's note Springer Nature remains neutral with regard to 16. Chandrasekaran VC (2010) Rubbers, Chemicals and Compound- jurisdictional claims in published maps and institutional affiliations. ing for ‘O’ Rings and Seals, in Rubber Seals for Fluid and Hydrau- lic Systems. William Andrew, Oxford 17. Li X, Qian P (2017) Identification of an e xposure risk to heavy metals from pharmaceutical-grade rubber stoppers. J Food Drug Anal 25(3):723–730. https:// doi. org/ 10. 1016/j. jfda. 2016. 07. 008

Journal

Biological Trace Element Research – Springer Journals

Published: Jun 1, 2022

Keywords: Plasma zinc; Zinc deficiency; Accuracy; Precision; Harmonization

There are no references for this article.