Commutability of quality control materials: the way forward

Transcription

1 Commutability of quality control materials: the way forward Ferruccio Ceriotti Diagnostica e Ricerca S. Raffaele, Milano

2 Definition of EQA System for objectively checking laboratory results by means of an external agency. the main objective being the establishment of trueness. (ISO/REMCO N231, 1991) To accomplish this task we need: Commutable control materials Reference methods based target values F. Ceriotti - CIRME Meeting, Milan 30 Nov

3 Commutability The equivalence of the mathematical relationship among the results of different measurement procedures for an RM and for representative samples of the type intended to be measured. CLSI C53-A F. Ceriotti - CIRME Meeting, Milan 30 Nov

4 Model of Lab Measurements for an EQA material Y p = a p + b p X + d p,s Y p X = mean result of a peer group p = true (e.g. IDMS) concentration of the analyte a p, b p = calibration error of the peer group (obtained on native sera) d p,s = matrix bias (group p, material s) F. Ceriotti - CIRME Meeting, Milan 30 Nov

5 Uric acid: bias of group means from reference method value Uricase POD Uricase POD Beckman % Bias group mean/ Ref. 4% 2% 0% -2% -4% -6% -8% -10% -12% -14% Uric Acid (mg/dl) F. Ceriotti - CIRME Meeting, Milan 30 Nov

6 Uric acid: commutability verification of Beckman systems Beckman Cx - Ref. (mg/dl) Uric Acid (mg/dl) Pools Controls F. Ceriotti - CIRME Meeting, Milan 30 Nov

7 Assessing commutability of CM (according to CLSI C53-A) Select 20 single donor samples spanning the relevant concentration range Analyze both CM(s) and patients samples with the pair of methods (e.g. reference and routine method) trying to minimize the random errors (single run, adequate replication of measurements). Elaborate the data using regression analysis and calculate if the CM(s) fall within the 95% prediction interval defined by the patients samples F. Ceriotti - CIRME Meeting, Milan 30 Nov

8 F. Ceriotti - CIRME Meeting, Milan 30 Nov

9 AST commutability 40 fresh samples Modular - Ref (U/L) Fresh sera BioRad Lyoph Asahi material Reference Method (U/L) F. Ceriotti - CIRME Meeting, Milan 30 Nov

10 AST commutability 10 frozen pools Delta Modular - Ref (U/L) Frozen pools BioRad Lyoph. Asahi Reference Method (U/L) F. Ceriotti - CIRME Meeting, Milan 30 Nov

11 Non-commutability Undesired byproduct of materials preparation combined with the nonspecificity limitations of some routine clinical laboratory methods. (W.G. Miller) Depends upon an abnormal material method interaction F. Ceriotti - CIRME Meeting, Milan 30 Nov

12 Non - commutability Important to distinguish between non commutability: problems only when analyzing CMs non specificity of the method: problems also with patients sera When specificity problems exist non commutability is much more probable F. Ceriotti - CIRME Meeting, Milan 30 Nov

13 Bias GOD-PAP Ref. Meth. (mg/dl) High bilirubin samples Glucose Native sera CMs Reference method (mg/dl) F. Ceriotti - CIRME Meeting, Milan 30 Nov

14 Commutability verification experiment (hypothesis) 6 analytical systems 12 EQAS control materials 10 fresh frozen serum pools Triplicate analysis of both pools an control materials 20 common general chemistry analytes TOTAL: 7920 routine analyses, 1320 Reference method analyses F. Ceriotti - CIRME Meeting, Milan 30 Nov

15 Limits of commutability evaluation Can be applied only to homogenous analytical systems Impossible to check every analytical system Very expensive for immunochemical methods Complex organization, need for collaboration with manufacturers and / or many laboratories F. Ceriotti - CIRME Meeting, Milan 30 Nov

16 Commutability assessment in a twin study Pairing laboratories two by two. Every pair of laboratories is asked to split six fresh patient samples with concentrations covering the analyte measurement range, to exchange them with the partner laboratory, and to assay them the next day (within 24 h after collection). Until analysis, specimens are stored at 4 C. In total, 12 patient specimens are assayed in duplicate by each laboratory in a single analytical batch with the Control materials randomly interspersed between the fresh patient specimens. F. Ceriotti - CIRME Meeting, Milan 30 Nov

17 Causes for non-commutability Matrix Turbidity ph Higher or lower viscosity Presence of exogenous substances Absence of trace elements Analyte Enzymes / proteins of animal origin Unusual isoenzyme composition Partially denaturated proteins Non glycosilated proteins F. Ceriotti - CIRME Meeting, Milan 30 Nov

18 Alternative ways to assess commutability Matrix problems. Check the methods Influence of: Turbidity Bilirubin ph of the sample Stabilizers Etc. Analyte problems (enzymes) Check the CM Km Effect of activators inhibitors ph optimum Buffer type / conc. Substrate type / conc F. Ceriotti - CIRME Meeting, Milan 30 Nov

19 GGT (% of activity) γgt ph optimum ph Pool CM F. Ceriotti - CIRME Meeting, Milan 30 Nov

20 ALP effect of the ion buffer With - without ion buffer ALP (U/L) without ion buffer Pools CRM F. Ceriotti - CIRME Meeting, Milan 30 Nov

21 How to obtain commutability? Frozen materials collected according to CLSI C37-A Specialized materials dedicated to small groups of analytes (e.g. lipids, enzymes etc.) Use of recombinant enzymes / proteins Factors to correct the matrix bias F. Ceriotti - CIRME Meeting, Milan 30 Nov

22 Model of Lab Measurements for an EQA material Y p = a p + b p X + d p,s Y p X = mean result of a peer group p = true (e.g. IDMS) concentration of the analyte a p, b p = calibration error of the peer group (obtained on native sera) d p,s = matrix bias (group p, material s) F. Ceriotti - CIRME Meeting, Milan 30 Nov

23 Usefulness of the Matrix Bias Correction Factor Measurement of a CM Matrix bias correction Result affected both by matrix bias and calibration error Result affected only by calibration error F. Ceriotti - CIRME Meeting, Milan 30 Nov

24 1. Analysis of CMs and 10 fresh frozen serum pools both with Reference and routine methods 2. Calculation of MBCF (material-peer Group specific) according to the following formula: MBCF = [(b p C L )+ a p ] /Y/ pl a p, b p = parameters of regression line vs. Ref. Meth (fresh frozen serum pools) C L = true value of control serum (Ref. meth) = mean value of control serum for P group Y pl Calculation of the Matrix Bias Correction Factor (MBCF) F. Ceriotti - CIRME Meeting, Milan 30 Nov

25 30 Cholesterol: commutability System A verification 20 System B System A - Ref. (mg/dl) Pools Controls System B - Ref. (mg/dl) Reference method (mg/dl) Reference method (mg/dl) F. Ceriotti - CIRME Meeting, Milan 30 Nov

26 Peer group Bias from the reference 6.0% 3.0% 0.0% -3.0% -6.0% -9.0% Cholesterol: behavior of two analytical systems in a PT scheme Frozen Level 1 Level 2 System A System B Level 1 Lyo mg/dl Frozen mg/dl Level 2 Lyo mg/dl Frozen mg/dl Lyo. Original Lyo. MBCF Frozen Lyo. Original Lyo. MBCF F. Ceriotti - CIRME Meeting, Milan 30 Nov

27 Conclusions Without commutable materials EQAS have very limited utility and Reference Method values are useless or even dangerous (wrong conclusions). The evaluation of commutability of the CMs requires a relevant effort, but can be performed. Fresh frozen pools are expensive to distribute and usually have narrow concentration ranges. Matrix bias correction factors can be an intermediate solution while developing better CMs F. Ceriotti - CIRME Meeting, Milan 30 Nov

28 Amylase: method comparison Vitros (U/L) Sera CRM Saliva Reference method (U/L) F. Ceriotti - CIRME Meeting, Milan 30 Nov