Evidence Based Commutability: Bias 2 Study Janice Gill Manager RCPAQAP Chemical Pathology Adelaide SA
Australian Bias Studies conducted by Gus Koerbin, ACT Pathology on behalf of AACB Harmonisation Committee Bias 1 Study Aim to assess between method bias to see whether it would be possible to use a common reference interval for an analyte Conducted in 2011 12 23 Australasian labs analysed 33 serum samples for 27 common analytes Na K Cl urea creatinine Ca Mg PO 4 ALT AST CK LD GGT Na, K, Cl, urea, creatinine, Ca, Mg, PO 4, ALT, AST, CK, LD, GGT, ALP, total protein, albumin, urate, cholesterol, triglycerides, HDL cholesterol, CRP, iron, transferrin, lipase, glucose, total bilirubin, bicarbonate
Bias 1 Study 8 major chemistry platforms Abbott Architect, Roche Modular, Roche Integra, Beckman Coulter (Dx Series), Beckman Coulter (Olympus), Siemens Advia, Siemens Dimension, OCD Vitros Provided reagent manufacturer & lot no., QAP method code, calibrator manufacturer, lot no. traceability, analyser age Labs asked kdto assay the appropriate QAP samples concurrently However the return on the QAP data was too low to analyse
Bias 1 Study Patient Samples Results reviewed using RCPAQAP Allowable Limitsof Performance (ALP) Analytes aytescass classified edusing the eta traffic light tsystem Green All results fall within the ALP Regression line does not cross the ALP within the current RI Amber No more than 4 results fall outside the ALP Regression line does not cross the ALP within the current RI Red Greater than 4 results fall outside the ALP Regression crosses the ALP within the current RI
Bias 1 Study Sodium Cholesterol Albumin Bicarbonate Potassium Triglycerides ALT Total Bilirubin Chloride HDL AST GGT Creatinine Glucose CRP (Vitros LOD) LD Iron Urea (DxC) Lipase Total protein Adjusted Ca Mg CK Transferrin (units) Urate (RxL) PO4 (Vitros) ALP (Vitros, AU2700)
Bias 2 Study An extension of Bias1 Post 2012 Harmonisation Workshop To provide evidence on someofof the questions raised by the workshop 30Labs, 9 frozen serum samples with extended concentration ranges where possible Plus 5 levels of RCPAQAP General Serum Chemistry samples 22 analytes, including ferritin, TSH, Free T4, Total PSA
Bias 2 Study Sodium (DXc) Cholesterol Albumin (BCP/BCG, cobas) Total Bilirubin (samples) Potassium Ti Triglycerides id ALT (P 5 P) P) GGT (methods) ( serum/plasma) Chloride HDL AST (P 5 P) LD (methods) Creatinine Glucose CRP (Vitros LOD) Lipase (methods) Iron Urea (DxC) Ferritin Total protein Urate (RxL) TSH Adjusted Ca (calculation) Mg CK Transferrin (units) PO4 (Vitros) ALP (Vitros, AU2700) Bicarbonate (Architect) t) FT4
Bias 2 Study An opportunity to compare patient vs QAP samples to investigate whether the QAP material is commutable Comparison against combined data of Bias 1 & 2
Sodium - Patient Samples 160.0 155.0 150.0 Sodium DXc slightly lower than other instruments Ob btained (mmo ol/l) Siemens ADVIA 145.0 Abbott Architect Beckman Coulter AU 140.0 Beckman Coulter DxC Roche Integra Roche Cobas 135.0 Siemens Dimension Vitros Fusion Low ALP 130.0 High ALP Linear (Low ALP) 125.0 Linear (High ALP) 125 130 135 140 145 150 155 160 Average (mmol/l)
Sodium QAP Samples 165.0 (mmo140.0 Ob btained ol/l) 160.0 155.0 150.0 145.0 140.0 135.0 130.0 125.0 120.0 115.0 110.00 105.0 DXc within ALP Vitros >145mmol/L Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Cobas Vitros Fusion Low ALP High ALP Linear (Vitros Fusion) Linear (Low ALP) Linear (High ALP) 100.0 100 105 110 115 120 125 130 135 140 145 150 155 160 165 Median (mmol/l)
Urea Patient Samples Urea 14 12 Obtain ned (mmol/l) 10 8 6 4 2 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Integra Roche Cobas Siemens Dimension Vitros Fusion Low ALP High ALP 2 4 6 8 10 12 14 Average (mmol/l) DxC slightly lower than other instruments at lower concentrations
Urea QAP Samples 25.0 20.0 Ob btained (mmo ol/l) 15.0 10.0 5.0 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Cobas Vitros Fusion Low ALP High ALP 0.0 0 5 10 15 20 25 Median (mmol/l) DXcwithin ALP
Creatinine Patient Samples 280.0 260.0 240.0 220.0 (umo180.0 Ob btained l/l) 200.0 180.0 160.0 140.0 120.00 100.0 80.0 60.00 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Integra Roche Cobas Siemens Dimension Vitros Fusion Low ALP High ALP 40.0 20.0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 Median (umol/l)
Creatinine QAP Samples 430.0 380.0 330.0 l/l) Ob btained (umo 280.0 230.0 180.00 130.0 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Cobas Vitros Fusion Low ALP High ALP 80.0 30.0 30 80 130 180 230 280 330 380 430 Median (umol/l) Architect, cobas at higher concentrations
ALT Pyridoxal-5-Phosphate Activator Positive bias with methods using P 5 P P as an activator Bias less pronounced with ALT concentrations >30 IU/L (15 32%) 60.0% 48.0% Differrence e (%) Platform - Average 36.0% 24.0% 12.0% 0.0% -12.0% -24.0% 12 17 22 27 32 37 42 ALT (U/L) Assays using P-5-P P activation Assays not using P-5-P activation
ALT Patient Samples 350.0 300.0 250.0 Siemens ADVIA Obtained (IU/L L) 200.0 150.0 100.0 Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Cobas Architect P5P Vitros P5P 50.0 0.0 0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350 Average (IU/L)
ALT QAP Samples 200.0 180.0 160.0 140.0 Obtained (IU/L L) 120.00 100.0 80.0 60.0 40.0 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Cobas Architect P5P Linear (High ALP) 20.0 0.0 0 50 100 150 200 Median(IU/L)
Calcium Patient Samples 2.80 2.70 2.60 Ob btained (mmo ol/l) 2.50 240 2.40 2.30 2.20 2.10 2.00 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Integra Roche Cobas Siemens Dimension Vitros Fusion Linear (ALP Low) 1.90 1.80 1.80 2.00 2.20 2.40 2.60 2.80 Average (mmol/l)
Calcium QAP Samples 3.40 3.20 3.00 2.80 Ob btained (mmo ol/l) 260 2.60 2.40 2.20 2.00 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Cobas Vitros Fusion 1.80 1.60 1.40 1.40 1.90 2.40 2.90 3.40 Average (mmol/l)
GGT Patient Samples 350.0 325.0 300.0 Obtained (U/L L) 275.0 250.0 225.0 200.0 175.0 150.0 125.0 100.0 75.0 50.0 25.0 0.0 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Integra Roche Cobas Siemens Dimension Vitros Fusion Low ALP High ALP Linear (Low ALP) Linear (High ALP) 0 25 50 75 100 125 150 175 200 225 250 275 300 325 350 Average (U/L) Glutamyl 4 nitroanilide Substrate
GGT QAP Samples 250.0 200.0 Obta ained (IU/L) 150.00 100.0 50.0 Siemens ADVIA Abbott Architect Beckman Coulter AU Beckman Coulter DxC Roche Cobas Vitros Fusion Low ALP 0.0 0 50 100 150 200 250 Median (IU/L)
Commutability Do the QAP samples behave in different methods with the same numerical relationship as is seen with patient samples? Not a simple yes/no answer Yes for some analytes eg GGT DxClow in patient and QAP samples For some analytes, the majority of methods behave in the same manner but 1 method may not eg DxC sodium low in patient samples but not in QAP RCPAQAP Chemical Pathology
Commutability How to evaluate this data? Evaluate using the same criteria as Bias Study Allowable Limits of Performance If a method is outside the ALP in patient samples; is it also outside the ALP in QAP samples? And vice versa Classify according to the traffic light system Green = agreement Orange = 1 method different Red = >1 method different RCPAQAP Chemical Pathology
QAP Samples Commutability Potassium Urate (DxC) Sodium (Vitros,DxC) Chloride Urea (DxC in) Albumin (BCP vs BCG) Bicarbonate ALP (DxC) Phosphate (Vitros, DxC) Creatinine ALT (P5P) Iron (DxC, Vitros, AU) Calcium AST (P5P) LD Magnesium Total Protein GGT Ferritin Transferrin ft4 TSH PSA Not Performed T Bilirubin (low pt samples) Glucose CK Cholesterol Triglyceride HDL Lipase CRP
Lyophilised QAP Samples Advantages Increased concentration ranges Linearly related End of cycle statistics Ease ofstorage Stable samples Transportation globally Disadvantages Not commutable for all analytes
Fresh Patient Samples Advantages Commutable material Disadvantages Limited concentration ranges Statistical summaries difficult Collection challenges Volume challenges Globaltransport difficulties
The Way Forward Lyophilised material Supplemented with fresh patient samples General Serum Chemistry + Liquid Serum Chemistry Glycohaemoglobin Lyophilised + Fresh Whole Blood Immunosuppressant Lyophilised + Fresh Whole Blood
Acknowlegements Gus Koerbin AACB Harmonisation Committee All the participating laboratories RCPAQAP Chemical Pathology