Evaluation of a Particle-Enhanced Turbidimetric

Evaluation of a Particle-Enhanced Turbidimetric Immunoassay for the Measurement of Ferritin:Application to Patients Participating in an AutologousBlood Transfusion ProgramFREDERIC GO´MEZ,JOSEP M a SIMO ´,JORDI CAMPS,XAVIER CLIVILLE ´,NU ´RIA BERTRAN,NATA `LIA FERRE ´,CLAUDI BOFILL,and JORGE JOVEN Centre de Recerca Biome `dica and Unitat de Medicina Transfusional,Hospital Universitari de Sant Joan,Carrer Sant Joan s/n,43201-Reus,Catalunya,SpainObjectives:To evaluate a turbidimetric immunoassay for the mea-surement of ferritin,and to assay this method in a group of patients undergoing an autologous blood transfusion program.Design and methods:We used an ILab 900analyzer.This instru-ment automates a particle-enhanced immunoturbidimetric assay with an analysis time of 9min.This technique was compared with a microparticle immunoassay.The turbidimetric assay was used to measure ferritin in a group of 30patients undergoing an autologous blood transfusion program.Results:The assay was linear in the range 3–1400␮g/L (r ϭ0.9999).The intra-and inter-assay imprecision (CV)at 20,97and 469␮g/L were Ͻ3.0and Ͻ5.0%,respectively.Recovery was 88.7to 97.4%.The detection limit was 3␮g/L.Hemoglobin (Յ4g/L),mild hyperbilirubi-linemia (bilirubin Յ50␮mol/L),triglycerides (Յ10mmol/L)and my-eloma paraproteins did not interfere with the assay.The assay showed good correlation with a microparticle enzymoimmunoassay (r ϭ0.994)with a mean difference between methods of Ϫ6Ϯ16␮g/L.This method was sensitive,accurate,and fast enough for an efficient follow-up of autologous blood transfusion patients.Conclusions:The new automated serum assay for ferritin is an attractive alternative that avoids the need for dedicated instrumen-tation.Copyright ©2000The Canadian Society of Clinical ChemistsKEY WORDS:anemia;autologous blood transfusion;automation;ferritin;turbidimetric immunoassay.IntroductionFerritin measurement in serum is considered to be the most reliable method for the evaluation of iron stores (1,2).The specificity and sensitivity of this measurement are greater to those of total iron or transferrin determi-nations (3–5).The clinical consequences of ferropenic anemia,the increasing practice of autologous blood transfusion (ABT,6–8)and the therapeutic use of re-combinant human erythropoietin (9–12)have generated considerable demand for a sensitive,convenient,and inexpensive assay of ferritin.Several methods for measuring serum ferritin have been reported.These include immunoradiometric as-say (13),radioimmunoassay (14),enzyme-linked im-munosorbent assay (ELISA)(15),microparticle enzy-moimmunoassay (MEIA),and chemiluminiscence assay (16,17).These methods accomplish the rigorous requirements for sensitivity and specificity that allow the detection of a few ␮g/L of ferritin,which are often found in patients with low iron stores.However,they may be time-consuming and need special,often dedi-cated,instruments.Conversely,turbidimetric immunoassay is a well-established technique for the rapid quantification of analytes at concentrations of mg/L (18).The concen-tration range for turbidimetric assays may be ex-tended by using the light-scattering properties of the immunoaggregates obtained by attaching the anti-body to latex particles;termed particle-enhanced turbidimetric immunoassay (PETIA,19–22).These techniques do not require specific instruments,merely a general automated analyzer,and have the advantages of decreased operating time,reduced costs together with an overall integration with other clinical chemistry analyses.The present study has been aimed to (a)evaluate a PETIA for the measurement of ferritin adapted for use in an ILab 900automatic analyzer (Instrumentation Lab-oratories,Milan,Italy);and (b)assay this method in a group of patients participating in an ABT program.Methods A PPARATUSFerritin measurements were performed in an ILab 900(Instrumentation Laboratories),an auto-Correspondence:Dr.J.Camps,Centre de Recerca Bio-me `dica,Hospital Universitari de Sant Joan,C/.Sant Joan s/n,43201-Reus,Catalunya,Spain.Fax:ϩ34-977-31-25-69;E-mail:jcamps@Manuscript received January 4,2000;revised March 2,2000;accepted March 2,2000.Clinical Biochemistry,Vol.33,No.3,191–196,2000Copyright ©2000The Canadian Society of Clinical ChemistsPrinted in the USA.All rights reserved0009-9120/00/$–see front matterPII S0009-9120(00)00064-3mated,random-access,discrete clinical chemistryanalyzer.This instrument facilitates the perfor-mance of spectrophotometric and turbidimetric re-actions at37°C,using one or two reagents,inbar-code primary tubes.Up to50different tech-niques can be processed simultaneously on-line inthe same sample.For comparison studies we usedthe MEIA technique from Abbott Diagnostics(Ab-bott Park,IL,USA)in an AxSYM௡immunoassayautomatic analyzer(Abbott).The analyzer is a ran-dom-access immunoassay system capable of per-forming20techniques simultaneously,at an ap-proximate speed of40results per hour.The MEIAtechnique is a type of enzymoimmunoassay in whichthe sample reacts with antibody-coated latex micro-particles and after incubation,the bound and un-bound components are separated into a glass-fibermatrix(23).R EAGENTS AND PROCEDURESThe ferritin PETIA assay was performed withcommercial reagents obtained from Biokit(QuantexFerritin kit,Biokit,S.A.,Barcelona,Spain).Theassay required two reagents.In the first step,3␮Lof serum were diluted with190␮L of100mMHEPES buffer(pHϭ7.5)with NaCl(0.5M)andsodium azide(0.9g/L)and incubated for3.4min.Inthe second step,the diluted sample reacted with asuspension of60␮L of polystyrene latex particles ofuniform size(200nm diameter)coated with antihu-man ferritin monoclonal antibody.The resultantagglutination was read at570nm twice,at36s andat5.4min after the addition of the latex reagent.Both reagents are liquid and do not require anypreparation prior to the analysis.MEIA technique was performed according to themanufacturer’s instructions.S TANDARD AND CONTROL MATERIALSThe calibrators of the PETIA assay were alsosupplied by Biokit(Quantex Ferritin Standard mul-tipoint)and consisted of five vials of liquid stan-dards from human source,with0,25,100,200,and500␮g/L of ferritin.Ferritin values higher than500␮g/L were automatically flagged and diluted one-fourth by the analyzer.A sixth standard with aferritin concentration of1000␮g/L was used as aprozone control.The standards were calibratedagainst the World Health Organization3rd Interna-tional Reference Preparation of Human Serum Fer-ritin(94/572).The calibration of the MEIA assay has been pre-viously described(24).Control levels2(ferritinϭ115Ϯ29␮g/L)and3(ferritinϭ458Ϯ92␮g/L)were from Biokit.Control1was the Immunoassay Control,level1(ferritinϭ34Ϯ1␮g/L)from Dade Behring(Newark,DE,USA).Three pools of sera were also used for performancestudies and were designated as“low”(20Ϯ2␮g/L),“medium”(97Ϯ1␮g/L)and“high”(469Ϯ3␮g/L). The sera were obtained from routine clinical sam-ples,mixed and frozen.When required they were thawed and mixed gently for20min on a Coulter mixer(Coulter Electronics Ltd.,Hialeah,FL,USA). S AMPLESFor comparison purposes,blood samples were obtained from100patients referred to the Clinical Laboratory of Hospital Universitari de Sant Joan de Reus.We chose samples from patients with a range of clinical conditions to include low,normal,and high concentrations of serum ferritin.Blood was allowed to clot at37°C and serum was obtained by centrifugation at1000g for10min and processed the same day.The measurements of ferritin in subjects benefit-ing from ABT were performed in30patients(12 men,18women),aged between24and78years. They were scheduled for major orthopedic surgery (nϭ15),hysterectomy(nϭ12),and prostatectomy (nϭ3).There was one visit in our outpatients clinic before each of the donations that were scheduled at 4-day intervals(up to4visits),one visit24h after the last donation(24h before surgery),one visit immediately before surgery,and the last one was performed20days after discharge(7).The fre-quency of blood donation was decided according the clinical evaluation,the predicted needs,and if a hemoglobin level of at least11g/dL was maintained. The amount of blood donated in each visit was450Ϯ30mL,which was expected to significantly deplete iron stores(6).Patients were fully aware of the study protocol as approved by the Bioethics Commit-tee of our Institution,and all patients gave informed consent.S TATISTICAL METHODSData are presented as meansϮSD.Mean values for ferritin by the two methods were compared by Student’s t test.Intragroup comparisons of ferritin means were performed by one-way ANOVA.The association between variables was measured by lin-ear regression analysis.The degree of agreement between both methods of measurement was esti-mated by the Bland-Altman graphical procedure (25).Statistical calculations were performed with the SPSS statistical program(26).ResultsP ERFORMANCE EVALUATION OF FERRITIN MEASUREMENTKinetic curves.Kinetic studies for a200␮g/L calibrator,for Control level3and for a reagent blank are shown in Figure1.The best-fit model(rՆ0.999) for these curves was quadratic(curves with the general model:yϭb0ϩb1xϩb2x2).The upper limit of the assay was set at1000␮g/L.Samples withGO´MEZ ET AL.higher ferritin values were automatically flagged and diluted (1/4)by the analyzer.Imprecision.Intra-assay imprecision was deter-mined with 15replicate analyses of the three com-mercial controls and the three pools.To assess inter-assay imprecision,aliquots of these controls stored at Ϫ20°C were analyzed over 15consecutive days.The CV values are shown in Table 1.Recovery.Recovery was estimated by spiking the three pools with the World Health Organizationstandard at a ratio of nine parts of pool and 1part of ferritin solution (500␮g/L).The percentage recover-ies (calculated as measured ferritin/theoretical fer-ritin ϫ100)from triplicate measurements were 90.2Ϯ3.0%for the low pool,88.7Ϯ2.3%for the medium pool and 97.4Ϯ3.1%for the high pool.Linearity and detection limit.Linearity was as-sessed by measuring serial dilutions of three high samples (valued at 1417,1413,and 1320␮g/L)in physiological saline solution from neat up to a dilu-tion of 1/128.The assessment of linearity took into account the automatic dilution one-fourth of the high samples by the analyzer.The regression line of observed vs expected values was y ϭ(0.99Ϯ0.01)x ϩ(11.7Ϯ3.1)(r ϭ0.9999Ϯ0.0001).To determine the detection limit,the absorbance of the reagent blank was measured 15times,the mean ϮSD was calculated and the detection limit was de-fined as the ferritin concentration corresponding to an absorbance equal to the mean of the reagent blank value ϩ2SD.The detection limit thus calcu-lated was 3␮g/L.Interference.Assessment of interference from trig-lycerides,hemoglobin and bilirubin was performed as previously described (2).The medium sera pool was supplemented with chylomicrons,hemoglobin or bilirubin at varying concentrations.Results are shown in Figure 2.There was no substantial inter-ference from hemoglobin up to 4g/L nor from mild hyperbilirubinemia (bilirubin Յ50␮mol/L)and mild hypertrigliceridemia (triglycerides Յ10mmol/L).Effect of paraproteinemia.The effect of nonspecific aggregation and/or cross-reactivity from parapro-teinemia on the ferritin determination was assessed as described (2).Serial dilutions (300␮L)in physi-ological saline of sera from a patient with IgA myeloma (IgA ϭ27.0g/L)were added to identical volumes of the medium pool,gently mixed on a Coulter mixer for 20min and subsequently ana-lyzed.The regression line of the observed (y )versus the expected (x )ferritin values was y ϭ0.99x ϩ0.01(r ϭ0.999).These results indicate that ferritin measurements were not influenced by the parapro-tein concentration.Comparison of methods.For 100samples,the mean values for ferritin were 67Ϯ119␮g/L for the PETIA assay and 73Ϯ128␮g/L for the MEIA assay.The equation of the regression line was y ϭ0.92x ϩ0.99;r ϭ0.994(Figure 3A).The degree of agreement between methods was assessed using the Bland-Altman graphical technique (Figure 3B).The mean difference was Ϫ6Ϯ16␮g/L.Results showed a tendency of the PETIA assay to undervalue ferritin concentrations with respect to MEIA.Nevertheless,only 3samples had ferritin values Ͼ2SD from the mean difference,and all of them were Ͼ350␮g/L,indicating a proportionally low percentage of error.Ferritin measurements in patients receiving an ABT.Results are presented in Table 2.As expected,there was a significant decrease in serum ferritin concentrations throughout the donation period,re-flecting the depletion of iron stores secondary totheFigure 1—Kinetics of the agglutination reaction.The increase in absorbance at 3.4min corresponds to the latex addition.ŒϭControl 3;ᮀϭ200␮g/L calibrator;●ϭreagent blank.T ABLE 1Analytical imprecision of the PETIA for ferritinContral material a Mean (␮g/L)IntraassayInterassay SD (␮g/L)CV,%SD (␮g/L)CV,%Control 134 1.1 3.3 1.2 3.6Control 2115 2.1 1.8 4.5 3.8Control 345810.4 2.38.8 1.8Low pool 200.6 3.0 1.0 5.0Medium pool 97 1.3 1.3 3.8 3.9High pool4698.71.812.62.7aAliquots of the three controls and pools were kept at Ϫ20°C until analyzed (n ϭ15).MEASUREMENT OF FERRITIN BY PETIArepeated bleeding.In all the subjects studied,serum ferritin concentrations remained always between the range 3–346␮g/L.DiscussionResults of all the performance characteristics of the PETIA were acceptable when compared with recommended values (1,27).The determination of ferritin concentrations in serum by this assay is an efficient and convenient alternative to MEIA.As demonstrated,this assay may be easily adapted to general analyzers,reducing the needs for dedicated instruments and allowing a much faster laboratoryresponse (9vs.20min to process a single sample).This possibility is of a great interest for Transfu-sional Medicine Units,since it facilitates the achievement of ferritin results within the time frame of patients clinic visits,thereby improving clinical decisions for blood donors and forautologousFigure 2—Interference of hemoglobin (●),bilirubin (Œ),and triglycerides (s )in the ferritin assay.The figure shows the effects on the medium pool (ferritin ϭ97Ϯ1␮g/L)of serial dilutions of free hemoglobin (up to 32g/L),bilirubin (up to 100␮mol/L)and chylomicrons (triglycer-ides up to 10mmol/L).Hemoglobin presented with a progressive negative interference from dilution one-fourth (8g/L)to undiluted (32g/L).Bilirubin presented with a positive interference when undiluted (100␮mol/L).Figure 3—Regression (A)and Bland-Altman plot (B)forferritin in the subjects studied.In (B),the dashed lines represent Ϯ2SD of the mean difference.T ABLE 2Ferritin concentrations (␮g/L)in ABT patients Visit No.A B C 148Ϯ4878Ϯ64132Ϯ96237Ϯ3466Ϯ56114Ϯ843—63Ϯ5295Ϯ964——75Ϯ62a 34Ϯ3149Ϯ3969Ϯ50b 33Ϯ2848Ϯ3669Ϯ52c49Ϯ5154Ϯ35101Ϯ88A:Patients with 2blood donations.B:Patients with 3blood donations.C:Patients with 4blood donations.a:24h before surgery.b:immediately before surgery.c:20days after discharge.In all the groups ferritin concentration significantly decreased from visit 1to b (p Ͻ0.001by ANOVA).GO´MEZ ET AL .blood handling.Readily available serum ferritin determinations may be used to regulate oral iron supplementation(28)in these patients,or to contrib-ute together with hemoglobin to determine the fre-quency and number of blood donations. Comparison with MEIA showed an acceptable degree of agreement,specially at low and normal ferritin concentrations.The agreement was not so good in samples with high ferritin levels and we do not know the reason for this discrepancy.This fact has little relevance in the study of patients with low iron stores,but it implies that high ferritin levels, which are commonly found in patients with hemo-siderosis,chronic diseases,or several types of can-cer,should be interpreted with caution when mea-sured by this method.The sensitivity of the PETIA assay is sufficient to accurately measure ferritin concentrations in ABT patients.Ferritin can be measured together with other general chemistry tests decreasing the total volume of blood that must be drawn from a patient;an aspect that has a special relevance in anemic pa-tients or in subjects submitted to repeated phlebot-omies.In addition,in our laboratory,this method has decreased by35%the reagent cost(compared to MEIA)of performing any one,single ferritin mea-surement.In conclusion,the present study demonstrates that this PETIA is an effective,reliable and a readily automated method for measuring ferritin concentra-tions in serum.Together these add a significant contribution to reducing laboratory costs and ratio-nalizing sample processing.AcknowledgementsThe authors thank Izasa S.A and Biokit S.A.for gifts of reagents and financial support.Nata`lia Ferre´was in receipt of a grant from Fundacio´Privada Reddis(1998). Technical assistance was by Peter R.Turner of t-SciMed (Reus,Spain).References1.Worwood M.The laboratory assessment of iron sta-tus—an update.Clin Chim Acta1997;259:3–23. 2.Simo´JM,Joven J,Cliville´X,Sans T.Automated latexagglutination immunoassay of ferritin with a centrif-ugal analyzer.Clin Chem1994;40:625–9.3.Jacobs A,Miller F,Worwood M,Beamish MR,Ward-rop CA.Ferritin in the serum of normal subjects and patients with iron deficiency and iron overload.Br Med J1972;4:206–8.4.Lipschitz DA,Cook JD,Fimch CA.Clinical evaluationof serum ferritin as an index of iron stores.N Engl J Med1974;290:1213–16.5.Palencia-Domı´nguez T,Martı´nez-Cervera J,Fuentes-Arderiu X.Evaluation of a new measurement proce-dure for the concentration of ferritin in serum.Eur J Clin Chem Clin Biochem1997;35:117–20.6.Bofill C,Joven J,Bages J,et al.Response to repeatedphlebotomies in patients with non-insulin-dependent diabetes mellitus.Metabolism1994;43:614–20.7.Cliville´X,Bofill C,Joven J,et al.Hemorheological,coagulative and fibrinolytic changes during autolo-gous blood donation.Clin Hemorheol Micro1998;18: 265–72.8.Kasper SM,Lazansky H,Stark C,Klimek M,Laub-inger R,Bo¨rner U.Efficacy of oral iron supplementa-tion is not enhanced by additional intravenous iron during autologous blood donation.Transfusion1998;38:764–70.9.Sherman LA,Lippmann MB,Ahmed P,Buchholz DH.Effect on cardiovascular function and iron metabolism of the acute removal of2units of red cells.Transfu-sion1994;34:573–7.10.Sans T,Bofill C,Joven J,et al.Effectiveness of very lowdoses of subcutaneous recombinant human erythropoi-etin in facilitating autologous blood donation before orthopedic surgery.Transfusion1996;36:822–6.11.Goodenough LT,Marcus RE.The erythropoietic re-sponse to erythropoietin in patients with rheumatoid arthritis.J Lab Clin Med1997;130:381–6.12.Sowade O,Warnke H,Scigalla P,et al.Avoidance ofallogenic blood transfusions by treatment with epo-etin beta(recombinant human erythropoietin)in pa-tients undergoing open-heart surgery.Blood1997;89: 411–18.13.Addison GM,Beamish MR,Hales CN,Hodgkins M,Jacobs A,Llewellin P.An immunoradiometric assay for ferritin in the serum of normal subjects and patients with iron deficiency and iron overload.J Clin Pathol1972;25:326–9.14.Halliday JW,Gera KL,Powell LW.Solid phase radio-immunoassay for serum ferritin.Clin Chim Acta 1975;58:207–14.15.Fairbanks VF,Klee GG,Biochemical aspects of He-matology.In:Burtis CA,Ashwood ER,Eds.Tietz textbook of clinical chemistry.Pp.1974–2072.2nd ed.Philadelphia:WB Saunders Co.,1994.16.Weeks,J,Woodhead JS.Chemiluminiscence immuno-assays.J Clin Immunoassay1984;7:82–6.17.Steitz WR.Immunoassay labels in chemiluminiscenceand bioluminiscence.Clin Biochem1984;17:120–4.18.Price CP,Spencer K,Whicher J.Light scatteringimmunoassay of specific proteins:a review.Ann Clin Biochem1983;20:1–14.19.Price CP,Newmann DJ.Light scattering immunoas-say.In:Price CP,Newman DJ,Eds.Principles and practice of immunoassay.Pp.443–80.2nd ed.London: Macmillan,1997.20.Thakkar H,Newman DJ,Holownia P,et al.Develop-ment and validation of a particle-enhanced turbidi-metric inhibition assay for urine albumin on the Dade aca௡analyzer.Clin Chem1997;43:109–13.21.Finney H,Newman DJ,Gruber W,Merle P,Price CP.Initial evaluation of cystatin C measurment by parti-cle-enhanced immunonephelometry on the Behring nephelometer systems(BNA,BN II).Clin Chem1997;43:1016–22.22.Thompson JC,Craig AR,Davey CL,et al.Kinetics andproposed mechanism of the reaction of an immunoin-hibition,particle-enhanced immunoassay.Clin Chem 1997;43:2384–9.23.Maclin E,Young DS.Automation in the clinical labo-ratory.In:Burtis CA,Ashwood ER,Eds.Tietz text-book of clinical chemistry.Pp.313–82.2nd ed.Phila-delphia:WB Saunders Co.,1994.man N,Juul-Jorgensen B,Bentzon MW.Calibra-tion of Abbott AxSYM ferritin kit using the WHO human liver ferritin International Standard80/602.Eur J Clin Chem Clin Biochem1997;35:631–2. 25.Bland JM,Altman DG.Statistical methods for assess-MEASUREMENT OF FERRITIN BY PETIAing agreement between two methods of clinical ncet1986;i:397–10.26.Norusis MJ.SPSS PCϩ.Advanced statistics4.0Gor-niche:SPSS International BV,1990.27.Fraser CG,Hyltoft Petersen P,Rico´s C,Haeckel R.Proposed quality specifications for the imprecisionand inaccuracy of analytical systems for clinical chem-istry.Eur J Clin Chem Clin Biochem1992;30:311–7.28.Birgegard G,Hogman C,Johansson A,Killander A,Simonsson B,Wide L.Serum ferritin in the regulation of iron therapy in blood donors.Vox Sang1980;38: 29–35.GO´MEZ ET AL.。