Walloon Agricultural Research Center Walloon Agricultural Research Center (CRA-W) Agriculture and Natural Environment Department (D3) – Valorisation of Agricultural products (D4) Agricultural systems, Territory and Information Technologies Unit (U11) – Authentification and traceability Unit (U16) Léon Lacroix Building - Rue de Liroux, 9 -– Henseval Building – Chaussée de Namur, B–5030 GEMBLOUX (Belgium) Tel : + 32 (0) Fax : + 32 (0) Tel : + 32 (0) Fax : + 32 (0) Setting of the threshold in the middle of the linear part in a logarithmic view Inverse regression for the determination of the cycle cut-off of a real-time PCR method for the detection of bovine tissues in feedingstuffs Viviane PLANCHON, Robert OGER Aline MARIEN, Gilbert BERBEN, Olivier FUMIERE AGROSTAT 2010 February 23-26, 2010 Benevento, Italy References Draper N. and Smith H. (1980) Applied regression analysis (Second edition). New York: John Wiley & Sons. Fumière O., Dubois M., Baeten V., von Holst C., Berben G. (2006). Effective PCR detection of animal species in highly processed animal by-products and compound feeds. Analytical and Bioanalytical Chemistry, 385, Fumière O., Veys P., Boix A., von Holst C., Baeten V., Berben G. (2009). Methods of detection, species identification and quantification of processed animal proteins in feedingstuffs. Biotechnologie, Agronomie, Société et Environnement, 13(S), Prado M., Berben G., Fumière O., Van Duijn G., Mensinga-Kruize J., Reaney S., Boix A., von Holst C. (2007). Detection of Ruminant Meat and Bone Meals in Animal Feed by Real-Time Polymerase Chain Reaction: Result of an Interlaboratory Study. Journal of Agricultural and Food Chemistry, 55, Exponential amplification phase Linear scale Logarithmic scale Ct ~ 22 cycles Depending of the operator: 20 < Ct < 25 Amplification : + No amplification : - Clear-cut signals Cut-off Late signals Aknowledgments EU Commission – DG Research and DG SANCO CRL-AP : Community Reference Laboratory for the detection of animal proteins in feedingstuffs ( ) A percentage of 95 % of blocks of triplicates detected corresponds to a level between 1 and 2 copies of the target Results Distribution of the cut-off values calculated Percentages of blocks (of 3 replicates) detected as positive The cut-off of the platforms are distributed on a wide range of Ct values < cut-off < Design of the inter-laboratory study 19 laboratories: 17 from the European Union, 1 from Japan, 1 from Australia 6 plates with 6 calibrations on each plate Precision of cut-off estimation and european inter-laboratory study Statistical aspects of inverse regression Background Statistical aspects of inverse regression and cut-off estimation Application to cut-off estimation Conclusions Concept of threshold cycle Real-Time PCR test 640 copies/5 µl 320 copies/5 µl160 copies/5 µl80 copies/5 µl40 copies/5 µl10 copies/5 µl Initially, calibrations with 28 points : 7 levels and 4 replicates/level Based on bias, variance and practical aspects decision for calibrations with 9 points : 3 levels, 3 replicates/level Evolution of the mean and the variance of X U Number of runs to evaluate X U ? CRA-W Inter thermocycler precision Intermediate precision  The cut-off determination was solved statistically using calibrations curves with plasmids carrying the PCR target and the application of inverse regression (Draper and Smith,1998) between the logarithm of the copy number and the Ct  The cut-off value is defined as the upper value (X U ) of the confidence interval for the Ct values corresponding to 1 copy of the target  X U is specific of a PCR platform X0X0 Inverse regression from ? and for a specific value Y(X 0 )= Y 0 known value unknown value Inverse regression for the determination of the cut-off for Y 0 =0 Y-axis :dependant variable predicted by X X-axis : independant variable allow to predict Y SAFEED-PAP : Detection of presence of species-specific processed animal proteins in animal feed ( ) In 1987 : bovine spongiform encephalopathy (BSE) epidemic. Most probable dissemination way of the disease = feeding with meat and bone meals (MBM) In 2000 : TOTAL BAN on the use of processed animal proteins (PAPs) in feed decided by European Commission (EC) in order to eradicate BSE in Europe (Council decision 2000/766/EC, Regulation 2001/999/EC, 2002/1774/EC) Progressive lift of the ban could be considered by the EC, if there is NO DANGER for the health or the policy for eradication of BSE (TSE Road Map 2005) Need of reliable analytical methods and tools for the species-specific detection and the quantificationof PAPs in the feedingstuffs : SAFEED-PAP European project. Objective of the project : to develop and validate a suitable PCR kit for the species-specific detection of bovine proteins in compound feed BUT the test is qualitative and requires a criterion to determine if a result is positive or negative : CUT-OFF DERTERMINATION PCR can sometimes give late signals that are not significant and such results have to be considered as negative. To that purpose, a cut-off value must be determined. It is generally expressed in terms of cycles to be compared to the Ct of an amplification curve. This Ct is the threshold cycle which means the number of cycles required for that amplification to reach a given fluorescence threshold. If the Ct of an amplification curve is higher than the cut-off, the result is considered as negative; however the Ct is a relative concept that is dependent on the thermocycler, the reagents used and the way to set the fluorescence threshold. At CRA-W, the cut-off was initially determined empirically to correspond to 40 cycles with the specific PCR conditions and the way the fluorescence threshold is set. Concept of cut-off Objectives To define a scientifically sound way to find out rapidly what is the cut-off value of any other PCR platform (thermocycler, reagent, laboratory environment) To evaluate the repeatability and inter thermocycler variability (intermediate precision) and reproducibility of the cut-off estimated through an inter-laboratory study Classical linear regression Confidence intervals of an estimated value X 0 XUXUXUXU ~ 10 calibrations (run) needed to set a robust platform cut-off value for 2, 4, 8, 10 and 16 runs data are grouped Number of copy levels and replications to evaluate X U ? Repeatability A scientifically sound way to find out rapidly what is the cut-off value of any other PCR platform (thermocycler, PCR reagent) has been determined based on inverse regression. Accuracy has been evaluated through evaluation of trueness and precision during an inter-laboratory study. The protocol designed for determination of the cut-off value is fit for purpose bias variance Real-time PCR = enzymatic reaction for the amplification of DNA fragment with a fluorescent detection. The production of fluorescence at each cycle depend on the concentration of targeted DNA in the reaction. unknown value known value