You are on page 1of 15

Food Additives and Contaminants

Vol. 29, No. 4, April 2012, 679–693

Modelling uncertainty estimation for the determination of aflatoxin M1 in milk by visual and
densitometric thin-layer chromatography with immunoaffinity column clean-up
K.L. Carvalhoa, G.A.A. Gonçalvesa, A.L. Lopesa, E.A. Santosa, E.A. Vargasa and W.F. Magalhãesb*
a
Laboratory of Quality Control and Safety Food – LACQSA/LANAGRO-MG/MAPA, Av. Raja Gabaglia, 245, Cidade Jardim,
CEP 30380-090 – Belo Horizonte, MG, Brazil; bChemistry Department, Universidade Federal de Minas Gerais – UFMG, Av.
Pres. Antonio Carlos, 6627, Campus Pampulha, CEP 31270-901 – Belo Horizonte, MG, Brazil
(Received 26 November 2010; final version received 7 November 2011)

The uncertainty of aflatoxin M1 concentration in milk, determined by thin-layer chromatography (TLC) with
visual and densitometric quantification of the fluorescence intensities of the spots, was estimated using the cause-
and-effect approach proposed by ISO GUM (Guide to the expression of uncertainty in measurement) following
its main four steps. The sources of uncertainties due to volume measurements, visual and densitometric TLC
calibration curve, allowed range for recovery variation and intermediary precision to be taken into account in the
uncertainty budget. For volume measurements the sources of uncertainties due to calibration, resolution,
laboratory temperature variation and repeatability were considered. For the quantification by visual readings of
the intensity of the aflatoxin M1 in the TLC the uncertainty arising from resolution calibration curves was
modelled based on the intervals of concentrations between pairs of the calibration standard solutions. The
uncertainty of the densitometric TLC quantification arising from the calibration curve was obtained by weighted
least square (WLS) regression. Finally, the repeatability uncertainty of the densitometric peak areas or of the
visual readings for the test sample solutions was considered. For the test samples with aflatoxin M1 concentration
between 0.02 and 0.5 mg l1, the relative expanded uncertainties, with approximately 95% of coverage probability,
obtained for visual TLC readings were between 60% and 130% of the values predicted by the Horwitz model.
For the densitometric TLC determination they were about 20% lower. The main sources of uncertainties in both
visual and densitometric TLC quantification were the intermediary precision, calibration curve and recovery. The
main source of uncertainty in the calibration curve in the visual TLC analysis was due to the resolution of the
visual readings, whereas in the densitometric analysis it was due to the peak areas of test sample solutions
followed by the intercept and slope uncertainties of the calibration line.
Keywords: chromatographic analysis; clean-up – affinity columns; statistical analysis; regression; measurement
uncertainty; aflatoxins; mycotoxins – aflatoxins; mycotoxins; milk

Introduction determined by thin-layer chromatography (TLC) with


One of the most important metrological characteris- visual and densitometric quantification. Within this
tics of a measurement result is its uncertainty. It is a methodology, this work addresses especially to the
worldwide consensus that a result is not complete following issues:
without an expression of its uncertainty (ABNT, . Evaluation of the uncertainty due to the
INMETRO 2003; BIPM et al. 1995) and its estima- resolution of the intensity of aflatoxin
tion is a requirement for testing laboratories accred- M1 in the TLC quantification by a visual
itation by the International Organization for reading.
Standardization (ISO) (2005). The present paper . Accounting for the heteroskedasticity of the
uses the methodology called the cause-and-effect instrumental response (fluorescence intensity)
approach (Ellison and Barwick 1998a, 1998b; in the densitometric TLC calibration by using
Barwick and Ellison 1998), or GUM approach weighted least squares (WLS).
(ABNT, INMETRO 2003; BIPM et al. 1995) or . Inclusion of an uncertainty contribution due
bottom-up approach to estimate the uncertainty of to the uncorrected bias within an allowed
aflatoxin M1 concentration in bovine milk which was range for the recovery variation.

*Corresponding author. Email: welmag@ufmg.br

ISSN 1944–0049 print/ISSN 1944–0057 online


ß 2012 Taylor & Francis
http://dx.doi.org/10.1080/19440049.2011.648959
http://www.tandfonline.com
680 K.L. Carvalho et al.

. Accounting for a within-laboratory reproduc- consumed feedstuffs contaminated with aflatoxin B1.
ibility or intermediary precision through the Recent reports of the International Agency for
data of the analysis of control samples for Research on Cancer (IARC) assessed the potential
quality control. carcinogenic risks to humans (World Health
Organization (WHO) and IARC 2002) of aflatoxin
Other approaches that estimate the uncertainty of
M1, and concluded that even very low levels of
an analytical measurement make use of data obtained
exposure to aflatoxins, i.e. 1 ng kg1 bw day1 or less
from a method validation study (Ellison et al. 2000;
contribute to a risk of liver cancer (Byrne 2000). Due it
Barwick and Ellison 2000a, 2000b; Barwick et al. 2000)
high human risks the European Union has established
or from interlaboratory studies through the results of
a very restrict maximum residue limit (MRL) or
collaborative trials (Adriaan et al. 1998; Barwick and
maximum permitted limit (MPL) for aflatoxin M1
Ellison 1998; Ellison 1998).
content in milk at 0.050 mg kg1 (Commission of the
Recent publications have addressed considerations
European Communities 1998, 2001, 2004, 2010).
to assess the uncertainty of aflatoxin M1 determination
European Union legislation (Commission of the
(Calaresu et al. 2006; Populaire and Giménez 2006).
European Communities 2006) states that the uncer-
Calaresu et al. (2006) used data obtained from preci-
tainty of the analytical measurement for aflatoxin M1
sion, trueness and ruggedness studies during method
contamination should be considered when reporting
validation to estimate uncertainty components.
and interpreting the analytical results (Commission of
However, the uncertainties due to the input quantities
the European Communities 2002, 2004, 2006). Besides,
appearing in their measurand equation and in their
European Union legislation (Commission of the
cause-and-effect diagram were not included in their
European Communities 2006) establishes the necessity
uncertainty budget. For instance, glassware tolerances
of declaring the correction of the aflatoxin M1
were used to estimate uncertainties and not the data
contamination by the recovery, or not, and what is
from the calibration certificates and laboratory tem-
the recovery. It also establishes the minimum perfor-
perature variation which appear in the cause-and-effect
mance criteria that an analysis procedure should meet
diagram. A very simple measurand equation was to be used in the analysis of aflatoxin M1 in milk that
presented where only three input quantities appear: was used when setting criteria for method perfor-
the volume of the sample taken for analysis, the final mance. MERCOSUR and Brazil (Agência Nacional de
volume of the sample solution, and the concentration Vigilância Sanitária (ANVISA) 2011; MERCOSUR/
of aflatoxin evaluated from the calibration curve. GMC/RES 2002) have established limits of 0.5 and
However, nothing was presented about the least square 5 mg kg1 for aflatoxin M1 in milk and powdered milk
procedure used to fit the calibration curve data, nor along sampling plans (Mercosur), but no method
about both the calibration curve range or the values of performance has been established, nor the necessity
the intercept, the slope of the calibration curve, and of reporting the recovery or uncertainty of the analyt-
their uncertainties and covariance. ical measurement. Recovery and measurement uncer-
Populaire and Giménez (2006) compared the tainty have been reported when demanded by
uncertainty estimation obtained from the bottom-up customers or according to institution policies as
method with the top-down method. They concluded defined by ISO/IEC 17025 (ISO 2005).
that the main sources of uncertainties in different The LACQSA/LANAGRO-MG/MAPA labora-
analytical methods were mainly due to the intermedi- tory is accredited according ISO/IEC 17025 by the
ary precision, accuracy/recovery and, in some cases, national metrology institute – the Instituto Nacional
calibration. They also concluded that the uncertainties de Metrologia Normalização e Qualidade Industrial
due to weighting and volume measurement were in (INMETRO) – to realise the analytical procedure
general negligible. However, no details of their calcu- for the determination of the herein reported ‘determi-
lations were given, also they used a poor calibration nation of aflatoxin M1 in milk by visual and densito-
design with only one standard calibration solution. metric thin-layer chromatography (TLC) with
The present paper presents a detailed uncertainty immunoaffinity column clean-up’.
budget to estimate the measurement uncertainty for
aflatoxin M1 determination in milk with immunoaffi-
nity column clean-up and thin-layer chromatographic
Materials and methods
(TLC) determination with visual and densitometric
quantification. Uncertainty estimation procedure
Aflatoxin M1 (4-hydroxyaflatoxin B1, (6aR-cis)- In the cause-and-effect approach for uncertainty esti-
2,3,6a,9a-tetrahydro-9a-hydroxy-4-methoxycyclopenta mation, or bottom-up approach, the detailed knowl-
[c]furo[30 ,20 :4,5]furo[2,3-h][l]benzopyran-1,11-dione) is edge of the chemical analytical procedure is necessary
a genotoxic carcinogenic hydroxylated metabolite of to enable the correct identification of the measurand
aflatoxin B1 found in the milk of animals that have and of its sources of uncertainties. In this paper the
Food Additives and Contaminants 681

four steps presented elsewhere were strictly followed Step 2: Identifying the sources of uncertainties
(Carvalho, Santos, et al. Forthcoming). As proposed elsewhere (Carvalho, Santos, et al.
Forthcoming) the cause-and-effect, or Ishikawa, dia-
gram was constructed by considering the input quan-
Step 1: Measurand specification tities that appear in the measurand equation as the
Measurand specification is realised through the mea- primary sources of uncertainties. Other sources of
surand equation and the measurement procedure uncertainties considered are the allowed range of
(ABNT, INMETRO 2003; BIPM et al. 1995; Ellison uncorrected bias due to the allowed variation of the
et al. 2000) or analytical procedure, which is normally recovery ratio (RR), and the intermediary precision or
documented through the standard operation proce- internal reproducibility or intra-laboratory reproduc-
dures. The analytical procedure for aflatoxin M1 ibility obtained from the analysis of control samples
determination using TLC is summarised below. for quality control.
Complete measurand equations, also presented
below, show in detail all the direct measurements
realised during the analytical procedure using TLC Volumes
with densitometric quantification (Equation 2) or with For the measurements of volume, four basic sources of
visual quantification (Equation 3). uncertainties were considered: instrumental uncer-
tainty (Joint Committee for Guides in Metrology
(JCGM/WG 2) et al. 2008; INMETRO 2009) due to
Analytical procedure the formal calibration of the instrument of volume
The analytical procedure for aflatoxin M1 determina- measurement; uncertainty due to resolution for grad-
tion was adapted from a published immunoaffinity uated volume instruments; uncertainty due to the
clean-up method (Dragacci et al. 2001). It can be variation of the laboratory temperature around the
summarised in the following steps: mean laboratory temperature; and uncertainty due to
volume measurement repeatability.
(1) A volume of 100 ml (Vs) of defatted milk was
taken after centrifugation and filtered by using
folded filter paper.
(2) Clean-up was done by using an immunoaffinity Calibration
column (VICAN or R-Biopharm), washing For TLC calibration, six standard calibration solutions
with 5 ml of water, followed by the elution of were used labelled P1, P2, . . . , P6 with concentrations
aflatoxin M1 with a solution of 2.5 ml of near to 1.0, 0.5, 0.3, 0.2, 0.1 and 0.05 mg ml1,
acetonitrile and methanol (2:1, v/v) and 2.5 ml respectively. As presented by Carvalho, Santos, et al.
of pure methanol. The sample eluate was (Forthcoming), the resolution of the visual calibration
evaporated at 40 C under nitrogen bubbling was modelled by considering the different intervals
just before complete dryness (critical step). (D’s) among the concentrations of the calibration
(3) The dried extract was re-dissolved with the solutions. This model considers that when reading,
addition of 100 ml (Vr) of toluene–acetonitrile by visual comparison, the analyst can state that the
(9:1 v/v) and by homogenisation with sonica- fluorescence of the sample spot is equal to the
tion and vortexed mix. fluorescence of one of the calibration standard solu-
(4) Application of 20 ml (Va) of the sample extract tions, Pi, or that its fluorescence is found between two
obtained in step 3 and of 10 ml (Vp) of the successive solutions Pi–Piþ1. This model is represented
calibration aflatoxin M1 standard solutions (six in Figure 1. From this model the uncertainty of the
spots) was made on the TLC plate, followed by visual quantification due to calibration resolution
elution with ether–methanol–water (96:3:1 v/v/
v).
(5) Detection and quantification step: the fluores- ••••• Δ(P4-P5) Δ(P3-P4) Δ(P2-P3) Δ(P1-P2)

cence intensities of the spots for the test


P6 P5 P4 P3 P2 P1
samples and the calibration solutions were
read under ultraviolet light ( ¼ 365 nm) by ••••• P3P4 P2P3 P1P2
visual comparison between sample spots and
••••• Δ(P2P3-P3P4) Δ(P1P2-P2P3)
calibration standard solutions spots or by using
a densitometer (dual wavelength flying spot
Figure 1. Variations (D) between any pair Pi–Piþ1 of
scanning densitometer; model CS-9301PC, standards and between half readings PiPiþ1–Piþ1Piþ2 used
Shimadzu, São Paulo, Brasil) at a wavelength to model the uncertainty due to the resolution of the visual
of 365 nm. calibration of the fluorescence readings in the TLC.
682 K.L. Carvalho et al.

(uVCresol) was estimated for different fluorescence underestimation of this source of uncertainty. This
readings as a function of the standard calibration model does not consider the intrinsic Poisson uncer-
solution concentration, C(Pi) by using the rectangular tainty of the densitometric peak area of each individual
probability density function (PDF) of the following spot, which is negligible compared with the other two
equations (1a) and (1b): sources of uncertainties considered.
   
CðPi Þ þ CðPiþ1 Þ CðPi Þ  CðPiþ1 Þ
uVCresol ¼ pffiffiffi ð1aÞ Recovery
2 2 3
The analytical method presents a recovery ratio
The uncertainty obtained by Equation (1a) is variable from day to day, but no correction is applied
attributed to the averaged concentration of the stan- to the results. Therefore, the allowed recovery ratio
dard solutions Pi and Pi þ 1: range was used to estimate a Type B uncertainty, as
½CðPi Þ þ CðPi þ1Þ=2 described by Carvalho, Santos, et al. (Forthcoming).
  
CðPi Þ þ CðPiþ1 Þ CðPiþ1 Þ þ CðPiþ2 Þ
uVCresol þ 2
2 2 Intermediary precision
 
CðPi ÞþCðPiþ1 Þ CðPiþ1 ÞþCðPiþ2 Þ Although the uncertainty of repeatability was included
 2  2 
¼ pffiffiffi ð1bÞ in the volume and calibration uncertainty sources, it
2 3 also included an intermediary precision to account for
The uncertainty obtained by Equation (1b) is other sources of uncertainties from bath-to-bath, day-
attributed to the concentration: to-day, sample preparation-to-sample preparation, etc.
variations.
f½CðPi Þ þ CðPiþ1 Þ=2 þ ½CðPiþ1 Þ þ CðPiþ2 Þ=2g=2
For each visual reading, the modelled resolution
Step 3: Estimating (quantifying) the standard
uncertainty was combined with the reproducibility
uncertainties of each source of uncertainty
standard deviation due to the two or three readings of
the TLC spots fluorescence intensities of different The uncertainties due to volume measurements,
analysts. visual and densitometric TLC calibration and
For the densitometric TLC calibration, the areas of quantification were estimated as described by
the calibration peaks, associated with the fluorescence Carvalho, Santos, et al. (Forthcoming).
of each standard calibration solution spots on the TLC
plate, were fitted against the standard calibration Recovery
solution concentration using the weighted least
When recovery is not used to correct the results, its
square (WLS). To estimate the uncertainties on the
uncertainty is obtained from the allowed variation of
densitometric TLC peak areas, the results of the
the recovery of the quality control samples analysed
calibration curves obtained over 8 months were used.
with each bath of analysis. For the samples with
The areas of the densitometric peaks of each of four
aflatoxin M1 contamination from 0.01 to 0.05 mg l1,
sets of calibration standard solutions were measured
the allowed recovery ratio range is from 60% to 120%;
on at least 3 days to obtain a model for the sample
above this contamination this allowed range becomes
standard deviations (square root of the variances) of
from 70% to 110% (Commission of the European
these areas as a function of the standard calibration Communities 2006). Each of these ranges defines a
solution concentration. This model was used to weight total amplitude 2a of the rectangular PDF for a Type B
each densitometric peak area in the WLS fitting of the estimation of the uncertainty.
calibration curves (see the last column of Table 2).
With this approach, the uncertainties of the intercept
and slope of the calibration curves take into account Intermediary precision
the sources of uncertainties due to the intermediary The uncertainties of intermediary precision of the
precision of the densitometric peak areas as well as, densitometric and visual TLC methods were obtained
and implicitly, that due to the preparation of the from the straight line model fitted to the data of the
standard calibration solutions, as required by standard deviation of the results of analysis of the
European Union norms (Commission of the fortified blank samples with 0.02, 0.05 and 0.5 mg l1
European Communities 2002, 2004). Although these for quality control, which was carried out over a period
European laws do not specifically apply to mycotoxin of 11 months. This is a Type A estimation; and as only
analysis, the use of the above approach to estimate three points were used to fit the straight line, the
calibration uncertainty is consistent with statistical degrees of freedom of this uncertainty component
concepts and is recommended in order to avoid the was only 1.
Food Additives and Contaminants 683

Step 4: Calculating the combined and expanded recovery; RR is the recovery ratio measured through
uncertainty of the measurand the analyses of a spike sample of quality control during
To calculate the combined standard uncertainty of the the bath of analysis; CResol is a null correction applied
concentration of aflatoxin M1 in the test sample, the to the average reading LVm due to the resolution of
classical uncertainties propagation law was used. To the visual calibration (by introducing this null correc-
obtain the expanded combined uncertainty, the stan- tion the not null uncertainty due to the visual reading
dard combined uncertaity is multiplied by the coverage resolution was computed); Cprecint is the null correction
factor to a coverage probability of 95%, as described due to the intermediary precision; and LVm is the
by Carvalho, Santos, et al. (Forthcoming). The cover- mean visual reading for the test sample concentration
age factor is obtained from the student PDF point of (mg ml1 or ppm) obtained by the average of three
probabilities, depending on the effective degree of individual readings of three different analysts for the
freedom calculated by the Welch-Satterthwaite equa- fluorescence intensity of the test sample spot compared
tion (INMETRO 2009; Joint Committee for Guides in with the fluorescence of the six spots of the aflatoxin
Metrology (JCGM/WG 2) 2008; Carvalho, Santos, M1 standard calibration solutions.
et al. Forthcoming). Note that CSAA is obtained from the intercept and
slope of the calibration curve by the equation:
APDA  a
CSAA ¼ ð4Þ
Results and discussion b
Step 1: Measurand specification Therefore, the uncertainty of CSAA, u(CSAA), will
After a detailed analysis of the standard operation be dependent on the uncertainties and covariance of a
procedures the proposed measurand equations were: and b, but also on the repeatability uncertainty of the
For densitometric TLC quantification: instrumental response for the test sample solution
applied to the TLC plate, u(APDA).
Vp  CSAA  Vr
CaflaM1 ¼  CFrecup þ Cprecint
Va  Vs
Vp  ðAPDA  aÞ  Vr 1
¼  þ Cprecint Step 2: Identifying the sources of uncertainties
b  Va  Vs RR
ð2Þ The cause-and-effect (Ishikawa) diagrams, obtained
for densitometric and visual methods, are showed in
For visual TLC quantification: Figures 2 and 3, respectively. All the input quantities
Vp  ðLVm þ CResol Þ  Vr in Equations (2) and (3) are present in these
CaflaM1 ¼ cause-and-effect diagrams. For Figures 2 and 3 some
Va  Vs
symbols are defined as above; other symbols are as
 CFrecup þ Cprecint ð3Þ
follows: CP, concentration of the calibration standard
where CaflaM1 is the aflatoxin M1 concentration solutions; APDP, area of the densitometric peak for
(mg l1 or ppb) quantified in the milk sample; CSAA the spots of the calibration standard solutions; Calibr,
is the concentration (mg ml1 or ppm) of the test sample calibration; Resol, resolution of the measurement
solution applied in the TLC plate; Vp is the volume (ml) instrument; Repet, repeatability; and Lab Temp Var,
of the standard solutions of different concentrations variation of the laboratory temperature around its
applied to the TLC plate to quantify the aflatoxin mean value.
contamination by using a densitometer calibration
curve or by visual comparison of the fluorescence
intensity related to aflatoxin M1 from the standards
Step 3: Estimating (quantifying) the standard
and the test sample spots; Vr is the volume (ml) of the
uncertainties of each source of uncertainty
solvent used to re-dissolve the sample after drying the
test sample extract obtained after the sample clean-up Volume
through the elution of aflatoxin M1 from the The three Type B uncertainties of the volume
immunoaffinity column; Va is the volume (ml) of the measurements due to the resolution, uResol(V), the
test sample solution applied to the TLC plate; Vs is the calibration, uCalib(V), and the total laboratory
test sample volume (ml) of the milk injected onto temperature variation, DT, of 5 K around the mean
the immunoaffinity column to be purified; APDA is the laboratory temperature, uVartemp(V), as well as the
area for the densitometric peak for the test sample Type A uncertainty due to the repeatability, uRep(V),
solution applied in the TLC plate; a and b are, were obtained as shown elsewhere (Carvalho, Santos,
respectively, the intercept and slope of the densitom- et al. Forthcoming).
eter calibration curve; CFrecup ¼ 1/RR is the correction The measurement function (INMETRO 2009; Joint
factor of the unitary value due to the uncorrected Committee for Guides in Metrology (JCGM/WG 2)
684 K.L. Carvalho et al.

Calibration
APDP
CSAA Allowed recovery variation
CP
a range

APDA uc(CaflaM1)
and
Purification U(CaflaM1)
Resol
Experimeters
Volumes
Lab Temp Var
Different days
Calibr

Repeat
Intermediary precision or internal
reproducibility or intralaboratorial Vp Vr Vs Va
precision

Figure 2. Cause-and-effect (Ishikawa) diagram to represent the sources of uncertainties to estimate the uncertainty of the
aflatoxin M1 content u(CaflaM1) for densitometric TLC quantification.

Allowed recovery variation


Resolution Repeatabilty LVm
range

Calibration

u(CaflaM1)
and
Purification
U(CaflaM1)
Experimeters Resol
Volumes
Lab Temp Var
Different days
Calibr

Intermediary precision or internal Repeat


reproducibility or intralaboratorial Vp Vr Vs Va
reproducibility

Figure 3. Cause-and-effect (Ishikawa) diagram to represent the sources of uncertainties to estimate the uncertainty of the
aflatoxin M1 content u(CaflaM1) for visual TLC quantification.

2008) or measurand equation of the volume came from the uncertainties u(a) ¼ sa and u(b) ¼ sb, and
quantity is: from the covariance u(a,b) ¼ cov(a, b) of the intercept
and of the slope of the calibration straight line
V ¼ Vnominal þCResol þ CCalib þ CVartemp þ CRep ð5Þ
(Carvalho, Gonçalves, et al. Forthcoming, Carvalho,
By applying the law of uncertainty propagation on Santos, et al. Forthcoming).
Equation (5) the combined uncertainties of each of the To quantify a test sample, a set of standard
four measured volumes, V (Vs, Vr, Va or Vp), were calibration solutions with the concentrations presented
obtained by the equation: in Table 2 was used. The same set of calibration
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi standard solutions was used for visual and densito-
uðVÞ ¼ u2Resol ðVÞ þ u2Calib ðVÞ þ u2Vartemp ðVÞ þ u2Rep ðVÞ metric TLC quantification. The peak areas of these
ð6Þ calibration standard solutions for a typical calibration
Table 1 shows the metrological characteristics of are also shown in the Table 2.
the instruments used for volume measurement: their The uncertainties (sample standard deviations) of
standard uncertainties due to instrument resolution, the APDP in the last column of Table 2 and the
calibration, repeatability and laboratory temperature uncertainties bars in Figure 4 clearly show the
variation, as well as the combined uncertainty of heteroskedasticity of the instrumental responses, and
volume measurement u(V) as calculated by were experimentally obtained as presented above
Equation (6). through the linear model:

uðAPDPÞ ¼ 10:80037 þ 165:56443  CP ð7Þ


Calibration
The uncertainty of the concentration of the aflatoxin By fitting a straight line to the data of the Table 2
M1 test sample solution, u(CSAA), applied to the TLC using the WLS (the continuous line in Figure 4), the
plate determined by the densitometric quantification intercept a ¼ (7.83  17.00), the slope b ¼ (1767.18 
Food Additives and Contaminants 685

Table 1. Metrological characteristics of the instruments used for volume measurements inp TLC
ffiffiffi quantification of aflatoxin M1:
scale division (d), scale space (SS), kd is the divisor in the equation: uResol(V) ¼ SS/[kd x 3], uncertainties due to resolution,
calibration laboratory temperature variation and repeatability, as well as the combined standard uncertainty for graduated
cylinder (Vs), automatic pipette (Vr) and microsyringe (Va and Vp).

Quantity Value d SS kd uResol (V) uCalib (V) uVartemp (V) uRep (V) u (V)

Vs 100 ml 1 ml 1.6 mm 4 0.14 ml 0.51 ml 0.030 ml 0.31 ml 0.61 ml


Vr 100 ml 1 ml –  0.29 ml 0.26 ml 0.030 ml 0.001 ml 0.39 ml
Va 20 ml 0.5 ml 1.5 mm 4 0.072 ml 0.085 ml 0.0061 ml 0.001 ml 0.11 ml
Vp 10 ml 0.5 ml 1.5 mm 4 0.072 ml 0.080 ml 0.0030 ml 0.001 ml 0.11 ml

Table 2. Calibration data of the TLC densitometric represent the uncertainty of any instrumental response
quantification.
within the calibration concentration range. The WLS
Standard CP (mg ml1) APDP u (APDP) and OLS lines cross each other at the crossing point
(0.268 mg ml1) marked in the Figure 4 with a signal
P6 0.036802 71.019 16.893 plus (þ). At concentrations lower than 0.268 mg ml1,
P5 0.101913 188.972 27.674 before the crossing point in Figure 4, the WLS line is
P4 0.184010 342.176 41.266
P3 0.331218 584.259 65.638 below the OLS line; after this concentration the WLS
P2 0.496827 920.293 93.057 line is above. This implies that the use of the OLS fit in
P1 0.993655 1696.175 175.314 the present case generates a proportional negative bias
for milk samples contaminated with aflatoxin M1 at
Note: CP, concentration of the calibration standard solution;
APDP, area of the densitometric peak for the spots of the levels of contamination lower than 0.268/
calibration standard solutions. 2 ¼ 0.134 mg ml1 and a positive bias above this con-
tamination. Figure 4 also shows the upper, UPL
(dashed line), and lower (dotted line), LPL, prediction
limits (lines) for 95% of confidence as obtained by the
WLS. Due the linear heteroskedasticity they are not
symmetrical around the centroid (median point, arith-
metic average), as occurs in the OLS, but are around
the barycentre (weighted centroid, weighted average),
where the prediction limit lines are the nearest to the
WLS fitted straight line. The region around the
barycentre of the calibration line is of better precision.
The area of the densitometric peak for a test sample
solution was 201.082, leading, according to Equation
(4) and the WLS calibration fitting parameters, to
CSAA ¼0.1094 mg ml1 and, according to Equation
(2), to an aflatoxin M1 content of 0.05468 mg l1. This
content on the basis of the OLS fit is 5.4% less than the
Figure 4. Calibration curve of the densitometric TLC content obtained from the WLS statistical method.
quantification as obtained by WLS (continuous line) and The barycentre and the centroid, which are marked
OLS fitting (dotted-dashed line). UPL and LPL are the upper
and the lower prediction limit curves for 95% of confidence. in Figure 4, are the points with coordinates
(0.0929 mg ml1, 172) and (0.3574 mg ml1, 634),
respectively.
115.91) ml mg1 and the covariance between the inter- Using the curves of the limit of prediction, as
cept and the slope cov(a, b) ¼ 1248.1 ml mg1 were recommended by the standard ISO 11843 (ISO 1997,
obtained. The intercept standard uncertainty of 2000) and described by Carvalho, Santos, et al.
17 ml mg1 shows that it is statistically equal to zero. (Forthcoming), the decision limit CC ¼
For comparison, the intercept and the slope of the 0.1171 mg ml1 (see CC in Figure 4) was obtained,
calibration line obtained from the ordinary least square corresponding, according to Equation (2), to aflatoxin
(OLS) fit (dash and dotted line in Figure 4) were M1 contamination CC ¼ 0.059 mg l1. Using the sim-
a ¼ 24.73 and b ¼ 1704.2 ml mg1. The uncertainties of plified methodology recommended in European
these OLS fitted parameters were not reported because Commission Decision 657 (Commission of the
they do not have any statistical meaning once the European Communities 2002), a decision limit of
residual standard deviation of the OLS fit does not CC ¼ 0.1132 mg ml1 corresponding to an aflatoxin
686 K.L. Carvalho et al.

M1 contamination of CC ¼ 0.055 mg l1 was found. Table 3. Modeling the typical uncertainty of resolution of
Note that these decision limits are nearly 10–20% the visual TLC calibration.
larger than the MPL of the European Union.
Standard CP
However, these calculated CC have only taken into solution (mg ml1) DCP Rr uResol (LVm)
account the calibration curve uncertainty. By including
the within-laboratory reproducibility they will become P1 1.000 – 0.787 0.227
higher. According to European legislation concerning Mean of P1P2 0.750 0.500 0.513 0.148
P2 0.500 0.350 0.294 0.0849
residues in animals and animal products for human
Mean of P2P3 0.400 0.200 0.222 0.0641
consumption, samples with residue levels above the P3 0.300 0.150 0.160 0.0462
limit of decision will be considered as being non- Mean of P3P4 0.250 0.100 0.132 0.0381
compliant (Commission of the European Communities P4 0.200 0.100 0.106 0.0306
2002, 2004; ISO 1997). Once, even taking into account Mean of P4P5 0.150 0.100 0.082 0.024
P5 0.100 0.075 0.061 0.018
the measurement uncertainty, such samples exceed the
Mean of P5P6 0.075 0.050 0.051 0.015
maximum residue limit (MRL) beyond reasonable P6 0.050 – 0.042 0.012
doubt (Commission of the European Communities
2002, 2004). Nowadays the decision limit is not applied Note: CP, typical standard concentration; DCP, concentra-
to decide about the compliance of the products tion variation between neighbouring calibration solutions;
Rr ¼ 2a is the range of the rectangular PDF. In the second
contaminated with aflatoxins. However, due to its column, values shown in bold are the concentrations of real
statistical consistency, it is expected that in the near standard calibration solutions; other values correspond to
future the decision limit will also apply to mycotoxins half readings.
as well as to all other products and contaminants.
In the visual quantification the readings of three
analysts for the fluorescence intensity of a sample spot Concentration variation between neighbours
standard solutions
were equivalent to those spots of the standard P5, 0.60
between the standards P5 and P6 (mean concentration
of P5 and P6), and the standard P5 again. These 0.50
y = 0.672x – 0.023
readings lead to a mean standard concentration R² = 0.951
LVm ¼ 0.0911 mg ml1 and standard deviation of 0.40
repeatability ¼ 0.0108 mg l1. This LVm corresponds,
ΔCP

0.30
according to Equation (3), to an aflatoxin M1 con-
tamination of 0.0455 mg l1. 0.20
This contamination, determined by the visual TLC
quantification, was 16.7% lower than that of the 0.10 y = 0.447x2 + 0.315x + 0.025
R² = 0.969
densitometric TLC quantification. For other contam-
0.00
ination levels the differences between these methods is
0.00 0.20 0.40 0.60 0.80
nearly between 15% and 20% and the signal of these
differences alternates randomly, and both analytical CP = Standard Concentration
procedures can be considered to be equally accurate Figure 5. Linear and the parabolic mathematical functions
concerning its trueness. modelling the aflatoxin M1 concentration variation, DCP,
The uncertainty of the mean read of the spot between successive calibration solution concentrations
fluorescence of the test sample in the visual TLC and between successive half calibration solution concentra-
tions (mean between two standards) used to estimate the
quantification has two components: repeatability and
uncertainty of resolution of the visual TLC calibration.
resolution. The first was obtained as the standard
deviation of three readings. To obtain a mathematical
model for the uncertainty due to the resolution of Commission 2010). The fourth column of Table 3
the visual TLC calibration according to Figure 1 and shows the calculated standard concentration variation
the strategy presented by Carvalho, Santos, et al. around each possible reading of the visual calibration
(Forthcoming), Table 3 was constructed. according to the parabolic model, which is used as the
Figure 5 shows the linear and parabolic (the range of the rectangular PDF (Rr) to estimate the
numerator in Equation 8) functions fitted to the data uncertainty of resolution of the mean visual reading,
of DCP against CP of Table 3. Owing to the coefficient uResol(LVm):
of determination, R2, the parabolic function is only
Rr 0:447CP2 þ 0:315CP þ 0:025
slightly better than the linear one. However, the uResol ðLVmÞ ¼ pffiffiffi ¼ pffiffiffi ð8Þ
parabolic function fits the data better at low concen- 2 3 2 3
trations, near the standard concentration of where CP is the concentration of a real or hypothetical
0.1 mg ml1, which correspond to the maximum per- standard calibration solution at the mean visual
mitted contamination level of 0.05 mg l1 (European reading.
Food Additives and Contaminants 687

Recovery For the visual TLC quantification:


As mentioned above, the uncertainty due to the
recovery variation assumes two different values uðprecintÞ ¼ 0:0006 þ 0:2706  CaflaM1ðmg=kgÞ ð14Þ
depending on the contamination of the test sample.
The correction factor due to the recovery (CFrecup) is
equal to the inverse of the recovery ratio. For samples Step 4: Calculating the combined and expanded
with contamination in the range 0.01–0.05 mg l1, uncertainty of the measurand
for which the allowed range of the recovery ratio To combine the standard uncertainties of the input and
was 60–120%, the standard uncertainty of the recov- influence quantities, u(), the particular case of the
ery ratio u(RR) was estimated by (Carvalho, Santos, uncertainty propagation law by propagating the rela-
et al. Forthcoming, Carvalho, Gonçalves et al. tive uncertainties was not used, as often occurs (Ellison
Forthcoming): et al. 2000; Barwick and Ellison 2000a, 2000b;
1:20  0:60 Armishaw 2003; Calaresu et al. 2006), because the
uðRRÞ ¼ pffiffiffi ¼ 0:1732 ð9Þ present measurand equations do not only have multi-
2 3
plications and division operations. For densitometric
As CFrecup ¼ 1/RR, the uncertainty of the correc- TLC quantification the uncertainty propagation law
tion factor of recovery u(CFrecup) is (Carvalho, takes the form:
Gonçalves et al. Forthcoming, Carvalho, Santos,
et al. Forthcoming): uðcaflaM1Þ
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

uðRRÞ 1:20  0:60 uh i2 h i2
u @CaflaM1
u CFrecup ¼ ¼ pffiffiffi ð10Þ u @Vp uðVpÞ þ @CaflaM1 @CSAA uðCSAAÞ
RR2 2 3  RR2 u
u
u h i h i
u þ @CaflaM1 uðVrÞ 2 þ @CaflaM1 uðVaÞ 2
While for samples with contamination above u @Vr @Va
u
0.05 mg l1, for which the allowed range of recovery ¼u h
u i
was 70–110%, the standard uncertainty of the recovery u þ @CaflaM1 uðVsÞ 2
u @Vs
ratio u(RR) was estimated by (Carvalho, Gonçalves u
u
et al. Forthcoming, Carvalho, Santos, et al. t h
i2 h@CaflaM1
i2
Forthcoming) þ @CaflaM1
@CFrecup u CFrecup þ @Cprecint u Cprecint

1:10  0:70 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi


uh i2 h i2
uðRRÞ ¼ pffiffiffi ¼ 0:1155 ð11Þ u @CaflaM1 @CaflaM1
2 3 u u ð Vp Þ þ u ð APDA Þ
u @Vp @APDA
u
u h i2 h i2

uðRRÞ 1:10  0:70 u @CaflaM1 @CaflaM1
u CFrecup ¼ ¼ pffiffiffi ð12Þ uþ u ð a Þ þ u ð b Þ
RR2 2 3  RR2 u @a @b
u
u h i2 h i2
u @CaflaM1
u þ @Vr uðVrÞ þ @CaflaM1 u ð Va Þ
¼u
As the recovery ratio varies significantly even @Va
u
within a batch of analysis, the median value 0.90 of uh i2
u @CaflaM1
the allowed range of recovery ratio variation was used u @Vs uðVsÞ
u
for both the recovery ratio range in Equationspffiffi(10)
ffi u
u h
i2 h@CaflaM1
i2
and (12), leading to values ofpffiffi0.740741/(2
ffi 3) ¼ u @CaflaM1
u þ @CFrecup u CFrecup þ @Cprecint u Cprecint
0.21383 mg l1 and of 0.493827/(2 3) ¼ 0.14256 mg l1 u
t
as standard uncertainties for the correction factor of
þ2 @CaflaM1 @a
@CaflaM1
@b uða, bÞ
the recovery ratio.
ð15Þ

Intermediary precision The first and second forms of the above equation
The straight line fitted to three standard deviations of are based on the respective forms of the measurand
the spiked samples blanks with aflatoxin M1 standard equation presented in Equation (2). Here the partial
solutions for 0.02, 0.05 and 0.5 mg l1 aflatoxin M1 derivatives (differential quotient or differential coeffi-
contaminations, measured many times during the 11 cient) are the sensitivity coefficients given by the
months, leads to the following models for uncertainty following equations:
due to intermediary precision or intra-laboratory
@CaflaM1 ðAPDA  aÞ  Vr CaflaM1
reproducibility: ¼  CFrecup ¼
For the densitometric TLC quantification: @Vp b  Va  Vs Vp
ð16Þ
uðprecintÞ ¼ 0:2262  CaflaM1ðmg=kgÞ ð13Þ
688 K.L. Carvalho et al.

@CaflaM1 Vp  ðAPDA  aÞ  Vr where the partial derivatives (differential quotient or


¼  CFrecup
@Va b  Va2  Vs differential coefficient) are the sensitivity coefficients
CaflaM1 given by:
¼ ð17Þ
Va
@CaflaM1 ðLVm  CResol Þ  Vr
¼  CFrecup
@Vp Va  Vs
@CaflaM1 Vp  ðAPDA  aÞ
¼  CFrecup CaflaM1
@Vr b  Va  Vs ¼ ð26Þ
CaflaM1 Vp
¼ ð18Þ
Vr
@CaflaM1 Vp  ðLVm  CResol Þ  Vr
¼  CFrecup
@CaflaM1 Vp  ðAPDA  aÞ  Vr @Va Va2  Vs
¼  CFrecup CaflaM1
@Vs b  Va  Vs2 ¼ ð27Þ
CaflaM1 Va
¼ ð19Þ
Vs @CaflaM1 Vp  ðLVm  CResol Þ
¼  CFrecup
@Vr Va  Vs
@CaflaM1 Vp  ðAPDA  aÞ  Vr CaflaM1 CaflaM1
¼ ¼ ¼ ð28Þ
CFrecup b  Va  Vs CFrecup Va
ð20Þ
@CaflaM1 Vp  ðLVm  CResol Þ  Vr
¼  CFrecup
@CaflaM1 @Vs Va  Vs2
¼1 ð21Þ CaflaM1
@Cprecint ¼ ð29Þ
Vs
@CaflaM1 Vp  Vr CaflaM1 @CaflaM1 Vp  ðLVm  CResol Þ  Vr CaflaM1
¼  CFrecup ¼ ¼ ¼
DAPDA b  Va  Vs APDA  a @CFrecup Va  Vs CFrecup
ð22Þ
ð30Þ

@CaflaM1 Vp  Vr CaflaM1 @CaflaM1


¼  CFrecup ¼  ¼1 ð31Þ
@a b  Va  Vs APDA  a @Cprecint
ð23Þ
@CaflaM1 Vp  Vr CaflaM1
¼  CFrecup ¼ ð32Þ
@CaflaM1 Vp  ðAPDA  aÞ  Vr @LVm Va  Vs LVm  CResol
¼  CFrecup
@b b2  Va  Vs
CaflaM1 @CaflaM1 Vp  Vr CaflaM1
¼ ð24Þ ¼  CFrecup ¼
b @CResol Va  Vs LVm  CResol
The only covariance taken into account was that ð33Þ
between the intercept and the slope, u(a, b) ¼ cov(a, b), Tables 4 and 5 summarise the uncertainty
of the calibration curve of the densitometric TLC combination for a blank sample spiked with aflatoxin
quantification obtained from WLS fitting. M1 to a content of 0.05 mg l1, which is the European
For visual TLC quantification, the uncertainty Union maximum permitted value, and determined as
propagation law takes the form: 0.04553 mg l1 by visual TLC quantification and as
0.05468 mg l1 by densitometric TLC quantification.
uðcaflaM1Þ The standard uncertainties of the quantities in
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Tables 4 and 5, where an asterisk (*) is shown after the
uh i h i
u @CaflaM1 uðVpÞ 2 þ @CaflaM1 uðLVmÞ 2
u @Vp @LVm
quantity symbol, are obtained through the uncertainty
u
u combination of the uncertainties at the lines immedi-
u h i2 h i
u þ @CaflaM1 uðVrÞ þ @CaflaM1 uðVaÞ 2 ately before it. For the case of the volumes, this
u @Vr @Va
u combination is realised according to Equation (6). The
¼u h i2 h i2
u graduated cylinder used to measure Vs was the unique
u þ @CaflaM1 uðVsÞ þ @CaflaM1 uðC Þ
u @Vs @CResol Resol volume measurement instrument not formally cali-
u
u brated. However, the data for internal quality control
t h
2 h@CaflaM1
i
i2
þ @CaflaM1
@CFrecup u CF recup þ @Cprecint u C precint
of a set of such cylinders showed a maximum bias of
0.8834 ml. In a conservative way, this bias was used as
ð25Þ the half range of the rectangular PDF (see the
Table 4. Calculation of the combined uncertainty for densitometric TLC quantification of aflatoxin M1 in milk.

Input or influence quantities Probability density Uncertainty contribution


sources of uncertainties function (PDF) (distribution) ui(y) or u(y;xi)

Standard Sensitivity Degrees of


Symbol Value, xi Interval Units Type Name Divider, k uncertainty u(xi) coefficient, ci ui(y) or u(y;xi) freedom i or eff [ui(y)]4/i

CResol 0 0.125 ml B Rectangular 1.73205 0.072169 0.005468 0.000395 1.0E þ 99 2.4E–113


CVartemp 0 0.00525 ml B Rectangular 1.73205 0.003031 0.005468 0.000017 1.0E þ 99 7.5E–119
CCalib 0 0.16 ml B Normal 2 0.08 0.005468 0.000437 1.0E þ 99 3.6E–113
CRep 0 0.0003 ml A Normal 1 0.0003 0.005468 0.000002 9 8.04E–25
Vp* 10 ml 0.107785 0.005468 0.000589
CResol 0 0.5 ml B Rectangular 1.73205 0.288675 0.000547 0.000158 1.0E þ 99 6.2E–115
CVartemp 0 0.0525 ml B Rectangular 1.73205 0.030311 0.000547 1.66E–05 1.0E þ 99 7.5E–119
CCalib 0 0.51 ml B Normal 2 0.255 0.000547 0.000139 1.0E þ 99 3.7E–115
CRep 0 0.00055 ml A Normal 1 0.00055 0.000547 3.01E–07 9 9.08E–28
Vr* 100 ml 0.386364 0.000547 0.000211
CResol 0 0.125 ml B Rectangular 1.73205 0.072169 0.002734 0.000197 1.0E þ 99 1.5E–114
CVartemp 0 0.0105 ml B Rectangular 1.73205 0.006062 0.002734 1.66E–05 1.0E þ 99 7.5E–119
CCalib 0 0.17 ml B Normal 2 0.085 0.002734 0.000232 1.0E þ 99 2.9E–114
CRep 0 0.0006 ml A Normal 1 0.0006 0.002734 1.64E–06 9 8.04E–25
Va* 20 ml 0.111671 0.002734 0.0003053
CResol 0 0.25 ml B Rectangular 1.73205 0.144338 0.000547 7.89E–05 1.0E þ 99 3.8E–116
CVartemp 0 0.0525 ml B Rectangular 1.73205 0.030311 0.000547 1.66E–05 1.0E þ 99 7.5E–119
CCalib 0 0.8834 ml B Rectangular 1.73205 0.510031 0.000547 0.000279 1.0E þ 99 6.0E–114
CRep 0 0.3056 ml A Normal 1 0.3056 0.000547 0.000167 4 1.94E–16
Vs* 100 ml 0.612597 0.000547 0.000335
APDA 201.082 28.9060 1 A Normal 1 28.90599 0.000283 0.008179 16 2.79E–10
a 7.83 17.0011 1 A Normal 1 17.001138 0.000283 0.00481 4 1.33E–10
b 1767.18 115.911 ml mg1 A Normal 1 115.911 3.09E–05 0.003586 4 4.13E–11
Cov(a,b) –1248.1 –1248.1 ml mg1 A Normal 1 –1248.1 8.75E–09 –1.09E–05 – –
CSAA* 0.1094 mg ml1 0.018004 0.5 0.009002
CSAA* 0.1094 mg ml1 0.020287** 0.5 0.010143
CFrecup 1 0.493827 1 B Rectangular 3.46410 0.142556 0.045529 0.006491 1.0E þ 99 1.7E–108
1
Cprecint 0 0.01237 mg l A Normal 1 0.012370 1 0.01237 1 2.34E–08
Sum 2.387E–08
Summary
Relative Relative
Quantity Combined Effective Coverage Expanded standard expanded
nominal Estimated standard degrees of probability Coverage uncertainty uncertainty uncertainty
value Y value y uncertainty freedom eff P (%) factor k U(y) RSU REU
Food Additives and Contaminants

uc(y)

0.05 0.0547 0.016637 3.21 95 3.1824 0.052945 30.4 96.8


689

Notes: Only the correlation between the intercept and the slope of the calibration line was considered according to Equation (34).
**Value for the standard uncertainty of CSAA obtained without the covariance between the intercept and slope of the calibration straight line.
Table 5. Calculation of the combined uncertainty for visual TLC quantification of aflatoxin M1 in milk.

Input or influence quantities sources Probability density function Uncertainty contribution 690
of uncertainties (PDF) (distribution) ui(y) or u(y;xi)

Standard Sensitivity Degrees of


Symbol Value, xi Interval Units Type Name Divider, k uncertainty U(xi) coefficient, ci ui(y) or u(y;xi) freedom i or eff [ui(y)]4/i

CResol 0 0.125 ml B Rectangular 1.73205 0.072169 0.004553 0.000329 1.0E þ 99 1.2E–113


CVartemp 0 0.00525 ml B Rectangular 1.73205 0.003031 0.004553 1.380E–05 1.0E þ 99 3.6E–119
CCalib 0 0.16 ml B Normal 2 0.08 0.004553 0.000364 1.0E þ 99 1.8E–113
K.L. Carvalho et al.

CRep 0 0.0003 ml A Normal 1 0.0003 0.004553 1.366E–06 9 3.87E–25


Vp* 10 ml 0.107785 0.004553 0.000491
CResol 0 0.5 ml B Rectangular 1.73205 0.288675 0.000455 0.000131 1.0E þ 99 3.0E–115
CVartemp 0 0.0525 ml B Rectangular 1.73205 0.030311 0.000455 1.380E–05 1.0E þ 99 3.6E–119
CCalib 0 0.51 ml B Normal 2 0.255 0.000455 0.000116 1.0E þ 99 1.8E–115
CRep 0 0.00055 ml A Normal 1 0.00055 0.000455 2.504E–07 9 4.37E–28
Vr* 100 ml 0.386364 0.000455 0.000176
CResol 0 0.125 ml B Rectangular 1.73205 0.072169 0.002276 0.000164 1.0E þ 99 7.3E–115
CVartemp 0 0.0105 ml B Rectangular 1.73205 0.006062 0.002276 1.380E–05 1.0E þ 99 3.6E–119
CCalib 0 0.17 ml B Normal 2 0.085 0.002276 0.000194 1.0E þ 99 1.4E–114
CRep 0 0.0006 ml A Normal 1 0.0006 0.002276 1.366E–06 9 3.87E–25
Va* 20 ml 0.111671 0.002276 0.000254
CResol 0 0.25 ml B Rectangular 1.73205 0.144338 0.000455 6.571E–05 1.0E þ 99 1.9E–116
CVartemp 0 0.0525 ml B Rectangular 1.73205 0.030311 0.000455 1.380E–05 1.0E þ 99 3.6E–119
CCalib 0 0.8834 ml B Rectangular 1.73205 0.510031 0.000455 0.000232 1.0E þ 99 2.9E–114
CRep 0 0.3056 ml A Normal 1 0.3056 0.000455 0.000139 4 9.37E–17
Vs* 100 ml 0.612597 0.000455 0.000279

LVmRepet 0 0.010850779 mg ml1 A Normal 1 0.010851 0.5 0.005425 2 4.33E–10


LVmResol 0 0.035885331 mgm l1 B Rectangular 1.732051 0.020718 0.5 0.010359 1.0E þ 99 1.2E–107
LV mg ml1 0.023388 0.5 0.011694
CFrecup 1 0.740740741 1 B Rectangular 3.46410 0.213833 0.04553 0.009736 1.0E þ 99 9.0E–108
Cprecint 0 0.012920313 mg l1 A Normal 1 0.012920 1 0.012920 1 2.79E–08
Sum 2.83E–08
Summary
Combined Effective Relative Relative
Quantity standard degrees of Coverage Coverage Expanded standard expanded
nominal Estimated uncertainty freedom probability factor uncertainty uncertainty uncertainty
value Y value y uc(y) eff P (%) k U(y) RSU REU

0.05 0.045 0.01997 5.62 95 2.5706 0.051340 36.5263 93.8938

Note: No possible correlations were considered.


Food Additives and Contaminants 691
0.02
almost equal to the uncertainties of the intermediary
Densitometric
precisions alone. This result corroborates the proposed
Uncertainty contributions

0.015 Visual
use of the intermediary precision as a gross estimation
of the analytical method’s combined uncertainty
0.01
(Populaire and Giménez 2006). The higher contribu-
0.005
tions of the combined uncertainty come from the
intermediary precision or within-laboratory reproduc-
0 ibility, followed by the uncertainties due to the
Comb Unc CU-prec Int Prec calib Recov Volumes calibration process and by the uncertainty due to the
allowed recovery variation range. The two last have
Figure 6. Contributions for the combined uncertainty of the
visual and densitometric TLC quantification of aflatoxin M1 nearly the same contribution. The uncertainties due to
in milk for a sample blank sample spiked to a content of the volume measurements are almost negligible and
0.05 mg l1 of aflatoxin M1. For volume measurement, the even the combination of their four contributions is
bars show the combined contribution of the four measured very low, 0.00144 and 0.00120 for densitometric and
volumes. visual quantification, respectively (see the last pair of
bars in Figure 6), corresponding to less than 10% of
the analytical methods’ combined uncertainties. Within
calibration component of Vs in Tables 4 and 5) to these uncertainties’ sources the most important is the
estimate its calibration uncertainty. contribution due the volume measurement of 10 ml of
The standard uncertainty of the concentration of the calibration standard solution, Vp, which is respon-
the sample solution applied in the TLC plate, sible for 41% of the total volume uncertainties in both
u(CSAA) ¼ 0.018004 mg ml1, was obtained from the methods.
WLS parameters of the calibration straight line
through Equation (34) (see also Equation E.3.3 in
Ellison et al. 2000 and Equation 26 in Carvalho,
Conclusions
Santos, et al. Forthcoming):
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
This paper showed in a detailed and conceptually
u 2
u u ðAPDAÞ þ u2 ðaÞ þ CSAA2  u2 ðbÞ consistent manner the estimation of the combined
u
t þ2  CSAA  covða, bÞ uncertainty of the aflatoxin M1 content in milk,
uðCSAAÞ ¼ determined by visual and densitometric quantification
b2
with immunoaffinity column clean-up, through the
ð34Þ
cause-and-effect methodology or bottom-up approach
As the co-variance (and correlation) between the recommended by the ISO through the ISO GUM
intercept and the slope of the calibration straight line is (ABNT, INMETRO 2003; BIPM et al. 1995). From
always negative, u(CSAA) ¼ 0.020287 mg ml1 (see the these estimations it seems that the uncertainty of the
value with a double asterisk in the second line with the densitometric TLC quantification is approximately
label CSAA in Table 4) without this correlation is between 10% and 30% lower than that of the visual
higher than u(CSAA) ¼ 0.018004 mg ml1 (see the value TLC quantification, but their trueness is the same
in the first line with the label CSAA in Table 4) when (INMETRO 2009; Joint Committee for Guides in
this correlation is taken into account. Metrology (JCGM/WG 2) 2008).
The combined standard uncertainties of aflatoxin The main source of uncertainty in both the visual
M1 content as determined by the densitometric and and the densitometric methods, as generally happens,
visual TLC quantification were 0.017 and 0.020 mg l1, was the intermediary precision, which alone is a
respectively, for a blank sample of milk spiked with reasonable estimation of the total (combined) analyt-
0.05 mg l1 (Tables 4 and 5). These combined uncer- ical uncertainty (Populaire and Giménez 2006), which
tainties without the contribution of the intermediary justifies the use of the top-down approach to estimate
precision are reduced to 0.011 and 0.015 mg l1 (see the the analytical uncertainty. However, the systematic but
bars labelled CU-prec in Figure 6), respectively, almost strenuous cause-and-effect (bottom-up) approach for
equal to the uncertainties of the intermediary preci- the estimation of analytical method uncertainty
sions alone: 0.0124 and 0.0129 mg l1, respectively. enabled the authors to understand the most important
Figure 6 summarises the calculations in Tables 4 sources of uncertainties, which, if reduced, lead to an
and 5, showing the combined uncertainties and their efficient analytical method precision improvement.
contributions to the densitometric and visual TLC Excepting the intermediary precision, the uncertainties
quantifications of a 0.05 mg l1 aflatoxin M1 spiked due to the calibration process and the allowed recovery
blank sample milk. For comparison, the values of the ratio range are equally the most important sources of
combined uncertainties without the intermediary pre- uncertainties of the TLC aflatoxin M1 analysis, while
cision contribution (see CU-Prec in Figure 6) are the uncertainty contributions due to the volume
692 K.L. Carvalho et al.

measurements are practically negligible. The inclusion Barwick VJ, Ellison SLR. 2000a. The evaluation of
of the analytical method uncertainty in the calculation measurement uncertainty from method validation studies.
of the decision limit for product compliance, according Part 1: Description of a laboratory protocol. Accred Qual
European legislations, increases the importance of a Assur. 5:47–53.
Barwick VJ, Ellison, SLR. 2000b. Protocol for uncertainty
realistic and conceptually consistent estimation of the
evaluation from validation data. January, LGC/VAM/
combined and expanded uncertainties of the analytical 1998/088. Valid Analytical Measurement – VAM,
procedures. By increasing the analytical precision, Department of Trade and Industry, Laboratory of
reducing its uncertainty, the consumer and vendor Government Chemists – LGC, Tedington, UK.
risks concerning the compliance decision are simulta- Barwick VJ, Ellison SLR, Rafferty MJQ, Gill RS. 2000. The
neously decreased. evaluation of measurement uncertainty from method
This work shows that the use of the OLS to fit the validation studies. Part 2: The practical application of a
data of measurement instrument calibration, when laboratory protocol. Accred Qual Assur. 5:104–113.
they are the subject of heteroskedasticity, leads to BIPM, IEC, IFCC, ISO, IUPAC, IUPAP, OIML. 1995.
Guide to the expression of uncertainty in measurement.
proportional bias and to incorrect estimation of the
Geneva (Switzerland): International Organization for
measurement uncertainty. The OLS lower and upper
Standardization (ISO).
prediction limit lines for the heteroskedastic calibra- Byrne D. 2000. Written question E2343/99 by Robert Evans
tion data presented in this work are far from the fitted (PSE) to the Commission. Aflatoxin fungus.
calibration straight line at the barycentre and MPL Official Journal of the European Union C225 E,
when compared with the WLS lower and upper 08/08/2000, vol. 43, pp. 0109-0110. [cited 2012 Jan 15].
prediction limit lines. This implies that the uncertain- Available from: http://eur-lex.europa.eu/LexUriServ/
ties of the analyte content at these levels, obtained LexUriServ.do?uri¼OJ:C:2000:225E:0109:0110:EN:PDF.
from the OLS fit, is much higher than the uncertainty Calaresu G, Piras P, Carta G, Galarini R, Chessa G. 2006.
obtained from the WLS fit. As a consequence, the Estimation of measurement uncertainty for the determina-
tion of aflatoxin M1 in milk using immunoaffinity clean-up
decision limit (CC) of the OLS is also higher than the
procedures. Accred Qual Assur. 11:10–16.
CC of the WLS. This can be a great risk for consumer Carvalho KL, Gonçalves GAA, Neres CS, Almeida NG,
health when using the decision limit as a compliance Almeida MI, Santos EA, Vargas EA, Magalhães WF.
criterion. Forthcoming. Modeling uncertainty estimation for myco-
toxins analysis at the MAPA/LANAGRO-MG/LACQSA,
within the ISO-GUM framework – Part III:
Determination of aflatoxin BG using immunoaffinity
References column purification step and quantification by high
performance liquid chromatography (HPLC). Food
ABNT, INMETRO. 2003. Guia para a Expressão da Addit Contam. To submit.
Incerteza de Medição. Terceira edição brasileira. Rio de Carvalho KL, Santos EA, Vargas EA, Magalhães WF.
Janeiro (Brazil): ABNT, INMETRO. Forthcoming. Modeling uncertainty estimation for myco-
Adriaan MH, van der Veen A, Broos JM, Alink A. 1998. toxins analysis at the MAPA/LANAGRO-MG/LACQSA,
Relationship between performance characteristics within the ISO-GUM framework – Part I: General
obtained from an interlaboratory study programme and approach. J Braz Chem Soc. To submit.
combined measurement uncertainty: a case study. Accred Commission of the European Communities. 1998.
Qual Assur. 3:462–467. Commission Directive 98/53/EC, laying down the sam-
Agência Nacional de Vigilância Sanitária (ANVISA). 2011. pling methods and the methods of analysis for the official
Resolução RDC n 07, de 18 de fevereiro de 2011; control of the levels for certain contaminants in foodstuffs.
D.O.U. – Diário Oficial da União; Poder Executivo, de Off J Eur Comm. L201:93–101.
18 de março de 2011, que dispõe sobre os limites máximos Commission of the European Communities. 2001.
tolerados (LMT) para micotoxinas em alimentos. Commission Regulation (EC) No. 466/2001, setting
ANVISA, Brası́lia, Brasil. [cited 2012 Jan 15]. Available maximum levels for certain contaminants in foodstuffs.
from: http://bvsms.saude.gov.br/bvs/saudelegis/anvisa/ Off J Eur Comm. L77:1–13.
2011/res0007_18_02_2011_rep.html Commission of the European Communities. 2002.
Armishaw P. 2003. Estimating measurement uncertainty in Commission Decision 2002/657/EC, Implementing
an afternoon. A case study in the practical application of Council Directive 96/23/EC concerning the performance
measurement uncertainty. Accred Qual Assur. 8:218–224. of analytical methods and the interpretation of results. Off
Barwick VJ, Ellison SLR. 1998. Estimating measurement J Eur Comm. L221:8–36.
uncertainty using a cause and effect and reconciliation Commission of the European Communities. 2004.
approach. Part 2: Measurement uncertainty estimates Commission Directive 2004/43/EC, amending Directive
compared with collaborative trial expectation. Anal 98/53/EC and Directive 2002/26/EC as regards sampling
Commun. 35:377–383. methods and methods of analysis for the official control of
Food Additives and Contaminants 693

the levels of aflatoxin and ochratoxin A in food for infants from: http://www.inmetro.gov.br/infotec/publicacoes/
and young children. Off J Eur Comm. L113:14–16. VIM_2310.pdf/
Commission of the European Communities. 2006. European International Organization for Standardization (ISO). 1997.
Regulation (EC) No. 401/2006 of 23 February 2006. ISO 11843-1:1997(E/F), Capability of detection, Part 1:
Laying down the methods of sampling and analysis for the Terms and definitions. Geneva (Switzerland): ISO.
official control of the levels of mycotoxins in foodstuffs. International Organization for Standardization (ISO). 2000.
Off J Eur Union. L70:12–34. ISO 11843-2:2000(E), Capability of detection, Part 2:
Dragacci S, Grosso F, Gilbert J. 2001. Immunoaffinty Methodology in the linear calibration case. Geneva
column clean-up with liquid chromatography for determi- (Switzerland): ISO.
nation of aflatoxin M1 in liquid milk: collaborative study. International Organization for Standardization (ISO). 2005.
J AOAC Int. 84(2):437–443. ISO/IEC 17025:2005, General Requirements for the
Ellison SLR. 1998. ISO uncertainty and collaborative trial competence of calibration and testing laboratories.
data. Accred Qual Assur. 3:95–100. Geneva (Switzerland): ISO.
Ellison SLR, Barwick VJ. 1998a. Estimating measurement Joint Committee for Guides in Metrology (JCGM/WG 2),
uncertainty reconciliation using a cause and effect BIPM, IEC, IFCC, ILAC, ISO, IUPAC, IUPAP and
approach. Accred Qual Assur. 3:101–105. OIML, JCGM 200:2008 (E/F). 2008. International voca-
Ellison SLR, Barwick VJ. 1998b. Using validation data for bulary of metrology – basic and general concepts and
ISO measurement uncertainty estimation, Part 1. associated terms (VIM). Vocabulaire international de
Principles of an approach using cause and effect analysis. métrologie – Concepts fondamentaux et généraux et
Analyst. 123:1387–1392. termes associés (VIM). Document produced by Working
Ellison SLR, Rosslein M, Williams A. 2000. EURACHEM/ Group 2 of the Joint Committee for Guides in Metrology
CITAC guide quantifying uncertainty in analytical chem- (JCGM/WG 2). [cited 2012 Jan 15]. Available from: http://
istry. 2nd ed. EURACHEM;CITA. [cited 2012 Jan 15]. www.bipm.org/en/publications/guides/vim.html/
Available from: http://www.eurachem.org/guides/pdf/ MERCOSUR/GMC/RES. 2002. N 25/02 Regulamento
Técnico Mercosul Sobre Limites Máximos de
QUAM2000-1.pdf.
Aflatoxinas Admissı́veis no Leite, Amendoim e Milho
European Commission. 2010. Commission Regulation (EU)
(revogação da RES. GMC N 56/94). XLVI GMC –
No. 165/2010 of 26 February 2010, amending Regulation
Buenos Aires, 20/VI/02.
(EC) No. 1881/2006 setting maximum levels for certain
Populaire S, Giménez EC. 2006. A simplified approach to the
contaminants in foodstuffs as regards aflatoxins. Off J Eur
estimation of analytical measurement uncertainty. Accred
Comm. L50:8–12.
Qual Assur. 10:485–493.
Instituto Nacional de Metrologia Normalização e Qualidade
World Health Organization (WHO) and International
Industrial (INMETRO). 2009. Vocabulário Internacional
Agency for Research on Cancer (IARC). 2002.
de Metrologia – Conceitos fundamentais e gerais e termos
IARC Monographs on the Evaluation of Carcinogenic
associados (VIM 2008). Primeira Edição Brasileira do
Risks to Humans, Vol. 82: Some traditional herbal
VIM (Tradução autorizada pelo BIPM da terceira edição medicines, some mycotoxins, naphthalene and styrene
internacional do VIM – International vocabulary of summary of data reported and evaluation, Last updated: 4
metrology – basic and general concepts and associated December 2002. Available from: http://monographs.
terms – JCGM 200:2008); [cited 2010 Apr 21]. Available iarc.fr/ENG/Monographs/vol82/volume82.pdf/

You might also like