Metrological Traceability and Assay Standardization in Laboratory Medicine STRESA, ITALY May 24, 2013
Standardization of ctni: Is there a light at the end of the tunnel? Jill Tate Pathology Queensland Brisbane, AUSTRALIA
Talk Outline Talk Outline Background IFCC ctni Pilot Study Serum pool as surrogate SRM Harmonisation capability Commutability across all assays Stability Next steps Preparation of candidate SRM Value assignment and uncertainty budget Value transfer to manufacturer s calibrator Harmonisation / commutability testing phase
ctni reference measurement system ctni reference measurement system Materials SI unit Procedures Traceability Pure substance Purified reference material: certified reference material NIST SRM 2921 Serum-based (commutable) reference Matrix material: certified reference material ctni-positive pool Manufacturer s master calibrator set Manufacturer s assay calibrators Patient serum sample Reference measurement procedure RP-LC and amino acid analysis Candidate RMP: ELISA or MS or Method harmonisation consensus approach Reference using commercial measurement ctni procedure assays Value Transfer Procedure (Reference Laboratory) Value Transfer Protocol Routine clinical assay Measurement Uncertainty Patient
Reference materials for ctni Reference materials for ctni NIST SRM 2921 Purified ICT complex Value assignment: RP-LC & amino acid analysis Not commutable in ~50% commercial assays Serum-based certified SRM Commutable in all commercial assays Lack of interferences e.g. ctni autoantibodies, heterophile antibodies Stable over long-term Standardised procedures in place for value assignment and value transfer to manufacturers master calibrators
Requirements for equivalent ctni measurements Measurand is defined unique, invariant part of molecule common to all components of the mixture present in serum Antibody specificity is defined Abies preferably recognise epitopes located in the stable part of ctni molecule all plasma ctni forms have equal reactivity or the difference in reactivity is not clinically relevant Assays are capable of being harmonised SRM is commutable across majority of assays Manufacturers have calibration traceability to SRM
ctni isoforms and assay recognition ctni isoforms and assay recognition Ratio Accu:Utlra ctn 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Changes in ctni ratio Accu:Ultra over time post chest pain onset ctnic + T ctnic + ctnitc + T ctnic + ctnitc + T 0 5 10 15 20 25 30 35 40 45 50 55 Time (h) AMI - Patient A AMI - Patient B AMI - Patient M ctnic + T Bates KJ et al. Clin Chem 2010; 56: 952-8
PROOF of PRINCIPLE: Serum pools as SRM for ctni Pools to consist of a blend of clinically relevant ctni forms and act as surrogate SRM for ctni rather than reflecting the ctni composition of each individual clinical sample Pools are commutable with patient samples covering the clinical ctni concentration range Pools lead to equivalent ctni measurement values
ctni Pilot Study in 2010-2012: AIMS Validation of the immunoassay reference measurement procedure for ctni Current status of commercial ctni assays Assessment of the commutability of blended serum pooled ctni candidate reference materials Evaluation of the stability of serum reference materials for ctni
ctni Pilot Study Samples ctni Pilot Study Samples Collection of samples from >90 patients with suspected AMI ctni concentrations in range 0.05-20 µg/l Collected from patients up to 72 h post presentation 30 samples per low, medium and high level 20 ml serum ( 50 ml blood) collected per patient Aliquotted within 4 h of collection & stored at -70 C Preparation of pools & sample kits at NIST Testing by NIST and Diagnostic Industry (NPL) January to May 2012 (1 lab in December 2012)
Participating Laboratories Participating Laboratories Beckman Access (AccuTnI) Roche Elecsys cobas e601 Biomerieux VIDAS TnI Ultra Roche Elecsys cobas e411 Siemens ADVIA Centaur (Ultra) Siemens Immulite 1000 TPI Siemens Immulite 2000/Xpi Siemens Dimension Vista Siemens Dimension EXL w/lm Siemens Dimension RxL Siemens Stratus CS Abbott Architect STAT hstni PATHFAST ctni (PF 1011-K) Abbott Architect i2000sr PATHFAST ctni-ii (PF 1101-K) Abbott AxSYM ctni-adv OCD Vitros 5600 crmp (at NIST)
Preparation of Serum Pools Preparation of Serum Pools Patient pools prepared in three ways by: addition of individual ctni-positive native patient samples dilution of high ctni concentration pool with low and medium concentration pools dilution of high & medium pools with a normal pool final concentration range 0.2-10 µg/l Normal Pool 500 ml pool from ~5-10 female donors (<30 y, BMI <25, & no reported history of heart disease) pre-screened for ctnaas none detected all participating labs also screened an aliquot.
ctni Candidate Serum Pools ctni Candidate Serum Pools Pool A B C D E F G Description 18 low ctni patient samples pooled using volumes which ranged from 1.25 ml to 8.0 ml 21 medium ctni patient samples pooled using volumes which ranged from 1.5 ml to 6.0 ml 21 high ctni patient samples pooled using volumes which ranged from 0.75 ml to 10.75 ml 28.0 ml Pool A and 7.0 ml Pool C 14.0 ml Pool C and 21.0 ml Pool B 4.0 ml Pool C and 36.0 ml Pool NORM 4.0 ml Pool B and 36.0 ml Pool NORM
ctni Pilot Study: data analysis and results Imprecision Duplicate measurements for 90 patient samples and 7 duplicate vials of pools Current status of commercial ctni assays and crmp Between-method variation Commutability Pools vs 90 patient samples Harmonisation capability Between-method agreement
ctni Pilot Study - imprecision ctni Pilot Study - imprecision Mean imprecision for 17 assays (%CV) 12.0 10.0 8.0 6.0 4.0 2.0 L16 L21 L12 Low patient samples - 4.1% CV High patient samples - 2.1% CV Medium patient samples - 2.2% CV Serum pools - 2.2% CV 0.0 0.0 0.1 1.0 10.0 100.0 Mean ctni for 17 assays (µg/l)
Commutability Assessment of Pools Commutability Assessment of Pools Paired comparisons Assay 2 vs. Assay 4 Assay 4 [ctni], μg/l 20 15 10 5 0 Patient Samples serum pools 0 5 10 15 20 25 Assay 2 [ctni], μg/l Assay 2 and 4 from same manufacturer
Commutability Assessment of Pools Commutability Assessment of Pools Paired comparisons Assay 14 vs. Assay 15 14 Assay 15 [ctni], μg/l 12 10 8 6 4 2 0 Patient Samples serum pools 0 5 10 15 20 Assay 14 [ctni], μg/l Assay 14 and 15 from different manufacturer
Commutability Assessment of Pools Paired comparisons Assay 9 vs. Assay 11 Assay 5 vs. Assay 7 Assay 11 [ctni], μg/l 14 12 10 8 6 4 2 0 Patient Samples serum pools 0 1 2 3 4 Assay 9 [ctni], μg/l Assay 7 [ctni],?g/l 12 10 8 6 4 2 0 Patient Samples serum pools 0 10 20 30 40 Assay 5 [ctni], µg/l Assay 3 vs. Assay 9 Assay 6 vs. Assay 11 Assay 9 [ctni], µg/l 4 3.5 3 2.5 2 1.5 1 0.5 0 Patient Samples serum pools 0 5 10 15 20 25 Assay 3 [ctni], µg/l Most of the 136 paired comparisons looked like these Assay 11 [ctni], µg/l 14 12 10 8 6 4 2 0 Patient Samples serum pools 0 5 10 15 20 25 30 Assay 6 [ctni], µg/l
Commutability Assessment of Pools Commutability Assessment of Pools crmp (NIST) [ctni], μg/l 9 8 7 6 5 4 3 2 1 0 Assay 7 vs. crmp (NIST) Patient Samples serum pools Paired comparisons 0 2 4 6 8 10 12 Assay 7 [ctni], μg/l Nearly all paired comparisons of the crmp vs. commercial assays looked like this
Current status of ctni assays in 2012 Current status of ctni assays in 2012 For commercial assays ~10-fold difference in concentration between assays crmp shows poor correlation with all routine assays Passing-Bablok analysis indicates overlap of the 95% confidence intervals of the regression slopes of patient samples and all pools indicating that all the serum pools are commutable for all routine assays PCA also indicates pools are commutable
Data analysis of ctni harmonisation Data analysis of ctni harmonisation Slope correction was determined for each assay using Passing-Bablok regression analysis against mean ctni for 17 assays for 90 patient samples Mathematical recalibration/recalculation was applied correction factor (CF) determined as [1/regression slope] recalculated ctni = measured ctni x CF Between-method agreement (CV) post recalibration for: all 17 assays 16 assays (1 assay excluded)
ctni post recalibration ctni post recalibration 16AxSYM ctni after recalibration 12 8 4 Mean ctni 17 assays 0 0 4 8 12 16 Passing-Bablok regression N = 88 Slope : 1.011 [ 0.964 to 1.045 ] Intercept : -0.0112 [ -0.0327 to 0.0074 ] 16 AccuTnI after recalibration 12 8 4 Mean ctni 17 assays 0 0 4 8 12 16 Passing-Bablok regression N = 88 Slope : 1.034 [ 0.996 to 1.070 ] Intercept : -0.0024 [ -0.0254 to 0.0146 ] 0.2 0.1 0.0-0.1-0.2 Log Difference (AxSYM minus mean ctni) Mean ctni -0.3-1.6-1.2-0.8-0.4 0.0 0.4 0.8 1.2 Logarithmic difference plot N = 88 Mean difference : -0.00726 [ -0.0234 to 0.00887 ] 0.2 0.0-0.2-0.4 Log Difference (AccuTnI minus mean ctni) Mean ctni -0.6-1.2-0.8-0.4 0.0 0.4 0.8 1.2 Logarithmic difference plot N = 88 Mean difference : 0.000835 [ -0.0168 to 0.0185 ]
Assays with same antibody specificity Assays with same antibody specificity 0.6 0.4 0.2 0.0-0.2 Log Difference (Dimension VISTA minus mean ctni) L9 L23 M27 M25 Mean ctni -0.4-1.2-0.8-0.4 0.0 0.4 0.8 1.2 Logarithmic difference plot N = 88 Mean difference : 0.00918 [ -0.0155 to 0.0338 ] Log Difference 0.4 (Dimension RxL minus mean ctni) L23 M27 0.3L9 M25 H15 0.2 0.1 0.0-0.1-0.2 Mean ctni -0.3-1.2-0.8-0.4 0.0 0.4 0.8 1.2 Logarithmic difference plot N = 88 Mean difference : 0.00838 [ -0.0105 to 0.0272 ] Log Difference 0.8(Dimension EXL minus mean ctni) L9 0.6 L23 M27 0.4 M25 0.2 0.0-0.2 Mean ctni -0.4-1.2-0.8-0.4 0.0 0.4 0.8 1.2 Logarithmic difference plot N = 88 Mean difference : 0.0577 [ 0.0327 to 0.0826 ] Log Difference 0.4(Stratus CS minus mean ctni) L19 0.3 H15 0.2 0.1 0.0-0.1-0.2 L12 L33 Mean ctni -0.3-1.6-1.2-0.8-0.4 0.0 0.4 0.8 1.2 Logarithmic difference plot N = 88 Mean difference : -0.0095 [ -0.0283 to 0.00932 ]
ctni harmonisation post recalibration ctni harmonisation post recalibration Between-method variation (%) post recalibration 120.0 100.0 80.0 60.0 40.0 20.0 0.0 Low ctni Patient Samples L16 & 21 L9 L23 Overall mean variation ~26% 0.00 0.20 0.40 0.60 Mean ctni for 17 assays (µg/l) Between-method variation (%) post recalibration 50.0 40.0 30.0 20.0 10.0 0.0 High ctni Patient Samples H15 Overall mean variation ~14% 0.00 5.00 10.00 15.00 Mean ctni for 17 assays (µg/l) Between-method variation (%) post recalibration 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Medium ctni Patient Samples M23 M25 M27 Overall mean variation ~18% 0.00 1.00 2.00 3.00 4.00 Mean ctni for 17 assays (µg/l) Between-method variation (%) after recalibration 35.0 30.0 25.0 20.0 15.0 10.0 5.0 A G F D D & B Serum Pools 0.0 0.00 2.00 4.00 6.00 8.00 10.00 E C Mean ctni for 17 assays (µg/l)
ctni between-method agreement ctni between-method agreement Between-method variation (%) 70 60 50 40 30 20 10 0 PRE recalibration of 17 assays POST recalibration of 17 assays POST recalibration of 16 assays 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Serum Pools A to G (duplicate series)
ctni Pilot Study: CONCLUSIONS ctni Pilot Study: CONCLUSIONS Serum pools behave better than most of the patient samples with lower inter-assay variability All serum pools are commutable with all routine assays Some assays correlated to the mean value better than other assays A high between-assay correlation for some assays from same manufacturer After calibration differences are removed method agreement was ~8 to 15 %CV in range 1-8 µg/l ctni
Next Steps Next Steps Production of SRM for ctni Value assignment and commutability testing of SRM Uncertainty budget determined for SRM Value transfer to manufacturers master calibrators Phase 3: harmonisation testing in a round robin
Production of SRM 2922 for ctni Production of SRM 2922 for ctni Minimum of 20 patient serum samples (min. vol 20 ml each) ctni in range 5-20 µg/l Stored at -70 C Prepare a serum pool from 20 patient sera (min. vol 610 ml) and Dilute 5-fold with normal pooled human serum (min. vol 2,440 ml) Aliquot diluted serum pool (0.5 ml) into 2 ml PP vials to be stored at -70 C 6,000 vials to be stored at NIST
Consensus value assignment for ctni Consensus value assignment for ctni Method harmonisation consensus approach using all commercial ctni assays mean or weighted mean value Use another panel of individual patient samples to confirm correlation at the time of valueassignment measurements similar to the pilot study but scaled down fewer patient samples and narrower concentration range Also use calibrant samples prepared from dilutions of SRM 2921 in ctni negative serum to re-calibrate the manufacturers data sets of the patient serum panel
Performance criteria for ctni: measurement uncertainty Performance goal Imprecision goal Bias goal Total error goal * Minimum 7.3% 21.6% 36% Desirable 4.9% 14.4% 24% Optimum 2.4% 7.2% 12% CVintraindividual 9.7%; CVinterindividual 56.8% * TE = Bias goal + 1.96xCVa Panteghini M. In: Laboratory and Clinical Issues Affecting the Measurement and Reporting of Cardiac Troponins: A Guide for Clinical Laboratories. Alexandria: AACB; 2012. p. 53-61.
Value transfer to manufacturers calibrators Value transfer to manufacturers calibrators Compare with value transfer of cystatin C ERM- DA471/IFCC Consensus method process uses a standardised value transfer RMP consisting of dilutions of master calibrator for ctni and candidate SRM Within and between day runs Number of replicates to depend on a predetermined precision goal Phase 3 Round Robin: Harmonisation testing using patient samples
Name J Tate (Chair) (AU) J Barth (UK) D Bunk (US) R Christenson (US) A Katrukha (FI) M Panteghini (IT) R Porter (UK) J Noble (UK) H Schimmel (BE) L Wang (US) I Young IFCC WG Standardization of Troponin I WG-TNI Membership Affiliation IFCC ACB NIST AACC HyTest Ltd. CIRME NPL IRMM NIST IFCC SD Liaison Labs that participated in ctni Pilot Study ABBOTT DIAGNOSTICS BECKMAN COULTER BIOMERIEUX MITSUBISHI CHEMICAL MED CO ORTHO-CLINICAL DIAGNOSTIC ROCHE DIAGNOSTICS GmbH SIEMENS DIAGNOSTICS NIST NPL