RAFA 2009, 3 Nov 2009, Prague, Czech Republic. 1. Current approaches to determine limit of detection and limit of quantification. P. López, Sz. Szilágyi, ...
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
Joint Research Centre (JRC)
Current approaches to determine limit of detection and limit of quantification P. López, Sz. Szilágyi, D. Lerda, T. Wenzl IRMM - Institute for Reference Materials and Measurements Geel - Belgium
http://irmm.jrc.ec.europa.eu/ http://www.jrc.ec.europa.eu/
1
Outline RAFA 2009, 3 Nov 2009, Prague, Czech Republic
1. Introduction ¾ Legislative references ¾ Definitions ¾ Procedures
2. Practical example – GC/MS ¾ Ion to use for LOD: quantifier or qualifier? ¾ Comparison of procedures ¾ Calibration requirements
3. Summary: points of concern
2
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
INTRODUCTION
3
Legislative requirements RAFA 2009, 3 Nov 2009, Prague, Czech Republic
4
Regulation (EC) No 882/2004 LOD and LOQ are criteria to characterize the methods of analysis
Commission Regulation (EC) No 333/2007 LOD/LOQ definitions
Benzo[a]pyrene (BaP) LOD = 0.3 µg/kg LOQ = 0.9 µg/kg
Commission Regulation (EC) No 1881/2006 Maximum levels of BaP (µg/kg wet weight) Oils and fats Smoked meats (products) Muscle meat of smoked fish (products) Muscle meat of fish Infant formula
2.0 5.0 5.0 2.0 1.0
Limit of detection. Definitions RAFA 2009, 3 Nov 2009, Prague, Czech Republic
5
Commission Regulation (EC) No 333/2007 LOD, smallest measured content, from which it is possible to deduce the presence of the analyte with reasonable statistical certainty. The limit of detection is numerically equal to three times the standard deviation of the mean of blank determinations (n > 20)
IUPAC
(IUPAC: Orange Book, p. 5)
The limit of detection, expressed as the concentration, cL , or the quantity, qL, is derived from the smallest measure, xL, that can be detected with reasonable certainty for a given analytical procedure. The value of xL is given by the equation
xL = x blank + k ∗ sblank
Where xblank is the mean of the blank measures, sblank is the standard deviation of the blank measures, and is a numerical factor chosen according to the confidence level desired.
ISO 11843:1 The true value of the net state variable, X, in the actual state that will lead with a probability of (1-ß) to the conclusion that the system is not in the basic state
Limit of detection. Definitions RAFA 2009, 3 Nov 2009, Prague, Czech Republic
6
α-error (type I): false positive ß-error (type II): false negative Default value at LOD: pα = pß = 0.05
LC = critical level LD = limit of detection
α=ß=0.05, at LD Commission Decision Figures from R. Boque and Y.V. Heyden, LCGC Europe,2002/657/EC 2009 (2) 82-85
Limit of detection. Definitions RAFA 2009, 3 Nov 2009, Prague, Czech Republic
7
European Pharmacopeia
2H S N= h
*
Background noise shall be determined at position of peak from blank material Width of noise area shall be 20 time peak width at half height Figures from R. Boque and Y.V. Heyden, LCGC Europe, 2009 (2) 82-85
Limit of quantification. Definitions RAFA 2009, 3 Nov 2009, Prague, Czech Republic
8
Commission Regulation (EC) No 333/2007 Limit of quantification (LOQ)
lowest content of the analyte which can be measured with reasonable statistical certainty. If both accuracy and precision are constant over a concentration range around the limit of detection, then the limit of quantification is numerically equal to six or 10 times the standard deviation of the mean of blank determinations (n > 20)
IUPAC minimum quantifiable value (LQ)
analyte (true) value that will produce estimates having a specified relative standard deviation (RSD), commonly 10%
Procedures RAFA 2009, 3 Nov 2009, Prague, Czech Republic
9
Approaches for the estimation of LOD and LOQ Calibration line method (ISO definition) According to ISO 11843-2 or the German standard DIN 32645 From the residual standard deviation of the calibration line Calculation from the 95% estimation interval of the calibration line
Blank procedures (IUPAC definition) DIN 32645
s LOD = 6 ∗ blank b s LOQ = 9 ∗ blank b
Regulation 333/2007 n=10 α=ß=0.01
LOD = 3 ∗ sblank
n=20 α=ß=0.05
LOQ = 6 ∗ sblank
S/N ratio (European Pharmacopeia definition)
sblank: standard deviation of the measurements at the blank level b: slope of the calibration curve
LOD = 3 x S/N LOQ = 10 x S/N
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
PRACTICAL EXAMPLES
10
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
11
Meat Spiking PAHs)
(13C-lab
Extraction (PLE) Evaporation Water elimination Filtration (5.0 µm)
GPC Evaporation
SPE Evaporation Addition injection standard
GC-MS (SIM)
Ion to use for LOD: quantifier or qualifier? RAFA 2009, 3 Nov 2009, Prague, Czech Republic
12
GC/MS in SIM (single ion monitoring)
Quantifier
Quantification Confirmation criterion: Ratio
Qualifier
Identification 252.0
Abundance
Benzo[a]pyrene
A126 = 13.1% ± %tolerance A252
9000 8000 7000 6000 5000
126.0
4000 3000 2000 1000
50.0
29.0 0 m/z--> 20
40
60
75.0 80
152.0
100.0 100
120
140
160
176.0 180
200.0 200
224.0 220
240
260
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
Guidance given by Commission Decision (EC) No 2002/657/EC
Identification criterion has to be fulfilled
13
Ion to use for LOD: quantifier or qualifier? RAFA 2009, 3 Nov 2009, Prague, Czech Republic
14
b) DhA (14.1%; 0.3 µg/kg) A
Abundance 3500
=
139
3000
13 . 5%
Ion 278.00
A
2500 2000
278
1500 1000 500 Time
|
|
-- >
36.00
36.10
36.20
36.30
36.40
Abundance
36.50
36.60
36.70
36.80
36.90
37.00
36.60
36.70
36.80
36.90
37.00
Ion 139.00
3500 3000
|
2500
|
2000 1500 1000 500 Time
-- >
36.00
36.10
36.20
36.30
36.40
36.50
c) CHR (10.9%; 2.5 µg/kg) A
= 0%
113 Abundance 20000
Ion 228.00
A 228
10000 0 Time -- >
18.40
18.60
18.80
19.00
19.20
Abundance
19.40
19.60
19.80
20.00
20.20
20.40
20.60
20.80
19.80
20.00
20.20
20.40
20.60
20.80
Ion 113.00
70000 60000 50000 40000 30000 20000 10000 0 Time -- >
18.40
18.60
18.80
19.00
19.20
19.40
19.60
Ion to use for LOD: quantifier or qualifier? RAFA 2009, 3 Nov 2009, Prague, Czech Republic
15
What if identification criterion is ignored?
Quantifier
µg/kg DIN 32645
Qualifier
Combination
LOD
LOQ
LOD
LOD
LOQ
Benzo[a]anthracene
BaA
0.11
0.19
0.23
0.23
0.23
Chrysene
CHR
0.11
0.20
>2.45
-
-
Benzo[a]pyrene
BaP
0.15
0.25
0.28
0.28
0.28
Dibenzo[a,h]anthracene
DhA
0.12
0.22
0.24
0.24
0.24
too optimistic
Comparison of procedures RAFA 2009, 3 Nov 2009, Prague, Czech Republic
16
Benzo[a]pyrene S/N ratio
LOD (µg/kg)
LOQ (µg/kg)
European Pharmacopeia
0.17
0.48
Blank procedure (IUPAC)
LOD (µg/kg)
LOQ (µg/kg)
0.15
0.47
LOD (µg/kg)
LOQ (µg/kg)
0.26 0.15 0.10 0.15
0.25 0.32 0.22
Blank standard deviation Calibration line ISO 11843:2 DIN 32645 Residuals 95% confidence interval
Comparison of procedures RAFA 2009, 3 Nov 2009, Prague, Czech Republic
17
S/N ratio approach
&
'
Ease of implementation Peak shape assumptions Subjective evaluation: manual integration Instability of the baseline over time
Blank procedures (blanks or fortified samples)
'
Inexistence of real blank matrices Dependence on spiking concentration
LOD/LOQ µg/kg
Spiked blank at 0.3 µg/kg
Non-spiked blank samples
BaA
0.16 / 0.48
0.05 / 0.15
Cyclopenta[cd]pyrene CPP
0.41 / 1.23
0.26 / 0.78
5-Methylchrysene
5MC
0.22 / 0.67
0.08 / 0.24
Dibenzo[a,l]pyrene
DlP
0.23 / 0.69
0.06 / 0.20
Benzo[a]anthracene
Comparison of procedures RAFA 2009, 3 Nov 2009, Prague, Czech Republic
18
Calibration line methods
&
More rigorous approach More accurate estimations of LOD/LOQ
' Computationally and experimentally more complex Comparison DIN 32645 vs ISO 11843-2 Matrix
Linearity
Homoscedascity
DIN 32645
X
X
X
ISO 11843-2
X
X
X
Heteroscedascity
Replicates I, J, L
X
I: Reference states used in calibration (3-5) J: Number of preparations for each reference state (J=2) K: Number of preparations for the actual state L: Number of repeated measurements per preparation
I, J, K, L
Calibration requirements RAFA 2009, 3 Nov 2009, Prague, Czech Republic
19
Calibration curve and blanks should be prepared in matrix LOD/LOQ (µg/kg) DIN 32645 Blank procedure
Matrix
Solvent
0.15 / 0.25 0.15 / 0.47
0.08 / 0.15 0.01 / 0.04
Individual preparations / pooled matrix extract LOD/LOQ (µg/kg) DIN 32645
Individual
Pooled extract
0.15 / 0.25
0.06 / 0.20
Calibration range close to LOD/LOQ values LOD/LOQ (µg/kg) DIN 32645
(0 - 2.5 µg/kg)
(2 – 15 µg/kg)
0.15 / 0.25
0.77 / 1.24
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
SUMMARY
20
Summary: points of concern RAFA 2009, 3 Nov 2009, Prague, Czech Republic
21
¾ LODs/LOQs: Differences due to estimation methodology methodology SHOULD be mentioned in the report ¾ LODs/LOQs MUST be determined in MATRIX and as INDIVIDUAL PREPARATIONS ¾ LODs/LOQs are likely to change over time Method has to be UNDER CONTROL ¾ Variability of blank signal frequently underestimated more than 1 sample more than 1 replicate
¾ LODs/LOQs should be verified in praxis ¾ Reporting LOD/LOQs: NEVER rounded down
Summary: points of concern RAFA 2009, 3 Nov 2009, Prague, Czech Republic
22
¾ Importance of LOD/LOQs: Fit for purpose: method feasibility to reach maximum levels of PAHs (BaP) in food (1 µg/kg) Risk assessment: estimation of exposure data
Thank you very much for your attention
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
23
Joint Research Centre (JRC)
Comparison of different approaches to estimate the limit of detection and limit of quantification of the15+1 EU priority polycyclic aromatic hydrocarbons (PAHs) in meat products P. López, Sz. Szilágyi, D. Lerda, T. Wenzl IRMM - Institute for Reference Materials and Measurements Geel - Belgium
http://irmm.jrc.ec.europa.eu/ http://www.jrc.ec.europa.eu/
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
Blank procedures
5g meat (n = 10 samples) 200 µL 35 ng/mL IS PAHs  1.4 µg/kg 200 µL 5.6 ng/mL PAHs  0.3 µg/kg ASE + GPC + SPE + GC/MS 1 injection/preparation Data treatment: DIN 32645 (software MVA)
24
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
25
Calibration procedures (5 g meat per preparation) Level 0
Level 1
200 µL
0
Level 2
Level 3
Level 4
1
2
3
4
5
IS PAHs 1.4 µg/kg
1.4 µg/kg
1.4 µg/kg
1.4 µg/kg
1.4 µg/kg
1.4 µg/kg
µg/kg
0.3 µg/kg
1.0 µg/kg
1.6 µg/kg
2.0 µg/kg
2.5 µg/kg
PAHs 0
Level 5
2 individual preparations per calibration level ASE + GPC + SPE + GC/MS 2 injection/preparation Data treatment: Excel macro (ISO 11843:2), Validata® (DIN 32645)
RAFA 2009, 3 Nov 2009, Prague, Czech Republic
26
S/N ratio Preparations used for the calibration approach ChemStation software (Agilent)
LOD (S/N = 3) LOQ (S/N = 10)
RMS (Pk-Pk) S/N (height)
1000
800
600
400
200
0 0.00
0.20
0.40
0.60
0.80
1.00
µg/kg BaP
1.20
1.40
1.60
1.80
