Detection and elimination strategies of matrix effects in quantitative multi-target LC-ESI-MS/MS analysis David Steiner, Michael Sulyok, Leonardo Mariño Repizo, Lidija Kenjerić, Rudolf Krska MultiCoop Training School, Prague,
Background routine laboratory analysis is undergoing a noticable change LC-TQMS, 2D-LC-MS/MS, LC-HRMS instrumental approaches gain traction single target analysis is increasingly replaced by multi-class concepts LC-MS/MS multi-class publications between 2008 and 2018 90 80 Number of Publications 70 60 50 40 30 20 10 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 MultiCoop Training School 2
Analysis Scheme Phenomenex Gemini C18 150 x 4.6 mm, 5 µm Extraction EIC in ESI pos > 1,000 analytes 5 g sample 20 ml ACN+H 2 O+HFo (79:20:1) v:v:v Agilent 1290 Infinity AB SCIEX QTRAP 5500 Dilution 1:1 with ACN+H 2 O+HFo (20:79:1) v:v:v Setup represents a compromise for LC conditions and extraction designed for robustness Analysis 5 µl injection of diluted raw extract 1 ml/min flow rate 2 injections pos/neg MultiCoop Training School 3
ESI and Ion Source Overview *Electrospray Ionisation Inlet (LC System) Uniformity MS Ion Source Region Matrix Effect (ME) combined effect of all components of the sample other than the analyte absolute ME increase or decrease in response between solvent standard and spiked pretreated sample http://www.lamondlab.com/msresource/lcms/massspectrometry/electrosprayionisation.php SSSE = Area of post extraction spike Area of neat standard RRRRRR oooo SSSSSS > 20% relative ME differences in response, accuracy and/or precision between different batches of the same matrix MultiCoop Training School 4
Legislation & Performance Criteria EC 32/2002 Undesirable substances EC 576/2006 Mycotoxins in feed EC 1881/2006 Mycotoxins in case of more than 20% signal suppression or enhancement, MEs need to be addressed in calibration EC 396/2005 Pesticides EC 37/2010 Pharmacologicals MultiCoop Training School 5
Evaluation of Matrix Effects silage alfalfa onion walnut oat apple herbal mix comprehensive evaluation of SSE% in 6 different matrices from different food & feed commodities effects were evaluated for 50 mycotoxins matrix effect potential from: carbohydrates & dietary fibre lipids & peptids polar co-eluting substances ionic species interferents with similar chemical structure MultiCoop Training School 6
Evaluation of Matrix Effects 80 70 60 Percentage of Analytes 50 40 30 20 10 0 Apple Walnut Onion Oat Silage Herbal Mix <60 60-80 80-120 120-140 >140 SSE in % MultiCoop Training School 7
Matrix Effects in Animal Feed 90 80 70 Percentage of Analytes 60 50 40 30 20 10 0 < 60 60-80 80-120 120-140 >140 SSE in % MultiCoop Training School 8
Animal Feed Validation Status quo Ingredient Validation data must be maintained for each feed group on at least one of the listed sample matrices Cereals Proteins Macrocomponents Microcomponents Amino Acids Wheat DDGS Vitamins Corn Fish Meal Colorants Barley Oil Cake Enzymes Matrix effect (± 20%) Repeatibility RSDr 20% variation on global compound feed market compositional uncertainty MultiCoop Training School 9
Complex Feed Matrices Absolute matrix effects 28 compound feed samples chicken, pig, fish, cattle 7 different lots of each sample type evaluation of SSE, EE, RA MultiCoop Training School 10
Relative Matrix Effects Frequency of analytes affected by intra matrix variation chicken < pig < fish < cattle 9% 20 % 24 % 32 % Signal suppressions for Zearalenone in cattle feed 36% 55% 23% 70% 59% 31% 53% 100% 90% 80% 70% 60% 50% 40% Composition of 7 compound cattle feed formulas 30% 20% 10% no uniformity regarding the composition compositional source of uncertainty up to 35 % 0% 1 2 3 4 5 6 7 Lucerne meal Broad beans Wheat bran Rye Sunflower cake Triticale Corn Corn meal Peas Barley Rest MultiCoop Training School 11
Feed Model Matrices Cattle Feed 100% 80% 60% 40% 20% 0% C1 C2 C3 C4 C5 C6 C7 Lucerne meal Broad beans Wheat bran Sunflower cake Triticale Corn Barley no compositional uncertainties use of blank single feed ingredients simulation of intra-matrix variation MultiCoop Training School 12
Results Model Matrices Fusarium Metabolites 100 90 80 Recovery in % 70 60 50 40 30 20 10 0 RA SSE RA SSE RA SSE Cattle Pig Chicken absolute matrix effects: SSE 20 33 % relative matrix effects: RSD 9 18 % Trichothecenes 15-Acetyldeoxynivalenol 3-Acetyldeoxynivalenol Diacetoxyscirpenol Fusarenon X HT-2 Toxin Monoacetoxyscirpenol Neosolaniol Nivalenol T-2 Toxin MultiCoop Training School 13
Results Model Matrices Aspergillus Metabolites 100 90 80 Recovery in % 70 60 50 40 30 20 10 0 RA SSE RA SSE RA SSE Cattle Pig Chicken absolute matrix effects: SSE 34 37 % relative matrix effects: RSD 7 13 % Aflatoxin B1 Aflatoxin B1 Aflatoxin B2 Aflatoxin G1 Aflatoxin G2 Aflatoxin M1 Averantin Averufin Sterigmatocystin Versicolorin A MultiCoop Training School 14
Real Sample vs. Model Matrix 100 90 SSE in % - Real Samples 80 70 60 50 40 30 20 10 0 0 10 20 30 40 50 60 70 80 90 100 SSE in % - Model Samples chicken cattle pig MultiCoop Training School 15
Prediction of Matrix Effects Chicken Feed Corn Soy DDGS Rapeseed 100 ME = 100 - nn (p FSSE) ME matrix effect n number of single feed ingredients p percentage SSE contribution FSSE signal suppression/enhancement factor Matrix Effect in % 80 60 40 20 0 ME calculated ME model ME real Aflatoxin B2 Alternariol Sterigmatocystin Diacetoxyscirpenol MultiCoop Training School 16
Method Performance s 0 is the estimated standard deviation of m single result s 0 is the standard deviation used for calculating LOD and LOQ n is the number of replicate observations avaraged when reporting results Limit of Quantification in µg/kg 30,0 25,0 20,0 15,0 10,0 5,0 0,0 1,6 0,7 25 8 Aflatoxin B1 Fumonisin B1 Ochratoxin A T-2 Toxin 11 4 12 2,5 Directive 2002/32/EG Analyte ML in µg/kg Aflatoxin B1 5 Fumonisin B1 5000 Ochratoxin A 50 more realistic estimation of the methods performance by using self designed compound feed formulas Compound Feed (Cattle) Single Feed (Maize) MultiCoop Training School 17
Matrix Reduction Strategies cattle feed dilution QuEChERS deep freezing dilution of extract (1:10 and 1:100) anhydrous MgSO 4 and NaCl + C18 (Discovery DSC-18) + PSA (PSA-Silica Sigma) anhydrous MgSO 4 and NaCl + freezing of extract at -20 C for 16h MultiCoop Training School 18
Reduction Efficiency 120% 100% 10% 6% dilution of extract reduces both, relative and absolute matrix effects SSE & RSD in % 80% 60% 40% 20% 24% 54% 85% 99% 20% 21% 20% 47% 49% 47% tenfold dilution steps reduction of relative matrix effects by a factor of 2 matrix reduction with QuEChERS extraction followed by an unspecific clean up (PSA, C18) is less efficient 0% Dil 1:1 Dil 1:10 Dil 1:100 PSA C18 Frozen Absolute Matrix Effects Relative Matrix Effects loss of Fumonisins during PSA step MultiCoop Training School 19
Dilute & Shoot on trial Simple dilution of the samples proved to be an efficient and fast way to reduce the signal suppressing/enhancing matrix effects provided by the matrices. [Eilfeld and Czapiewski 2013] 8 different herbal mix samples undiluted 1 µl 1:10 1:20 1:10 dilution 10 µl Analyte Other Components 1:20 dilution 20 µl MultiCoop Training School 20
Matrix Protectants 140 120 100 SSE (%) 80 60 40 20 0 Fumonisin B1 HT-2 Toxin Ascochlorin Ascofuranone Norsolorinic acid undiluted - 1 µl 1:10 dilution - 10 µl 1:20 dilution - 20 µl Equisetin MultiCoop Training School 21
What do high absolute and relative matrix effects mean? Is the method fit for purpose? MultiCoop Training School 22
Extraction Efficiencies 100 90 Relative Frequency of Analytes in % 80 70 60 50 40 30 20 10 0 < 70 % 70-80 % 80-90 % 90-100 % EE in % MultiCoop Training School 23
Proficiency Tests 140 PT results for regulated mycotoxins in animal feed obtained within 8 years >93 % of submitted results in the satisfactory range Compilaton of z-scores obtained by our method in routine proficiency testing (BIPEA) 3,00 2,00 1,00 z.score 0,00-1,00 Zearalenone Aflatoxin G2 Aflatoxin B1 Nivalenol -2,00-3,00 MultiCoop Training School 24
Summary matrix effects (ME) are a major limitating factor for LC-ESI-MS/MS multi-class methods dilution of extracts is a straightforward solution for a decisive reduction of ME degree of dilution has to be considered in terms of sensitivity and protective mechanisms complex matrices like animal feedstuff represents an additional challenge in terms of relative matrix effects validation scheme should take intra-matrix variation into account feed model matrices solves the compositional uncertainty better estimation for method performance MultiCoop Training School 25
Acknowledgement Gerd Schatzmayr Andreas Gschaider Zoltan Balla Eduard Schneeberger Leonardo Mariño Repizo Lidija Kenjerić Rudolf Krska, Michael Sulyok MultiCoop Training School 26