Aucun titre de diapositive

1
Results of a European proficiency test for the
detection of streptomycin/dihydrostreptomycin,
gentamicin and neomycin in milk by ELISA and
Biacore methods
CRL-AFSSA Fougères
EURORESIDUE V,Noordwijkerhout (The Netherlands),
10-12 May, 2004 Valérie GAUDIN, N. CADIEU, Pascal
SANDERS Community Reference Laboratory for
Veterinary Drug Residues, AFSSA, BP 90203, 35302
Fougères, France e-mail v.gaudin_at_fougeres.afssa.
fr
Aminoglycoside antibiotics are commonly used in
veterinary medicine for treatment of bacterial
infections. Neomycin (NEO), gentamicin (GTM),
streptomycin (STR) and dihydrostreptomycin (DHS)
are included in Annex I of Council Regulation
(EEC) 2377/90 1 and their respective MRLs in
milk are 1500, 100, 200 and 200 µg/kg. In
2003, an interlaboratory study was proposed to
the European Community and Third countries
National Reference Laboratories (NRL) for the
analysis of aminoglycoside residues in bovine
milk by ELISA. This test was intended to allow
the participants to control their aminoglycoside
ELISA methods when used routinely and also to
compare the performance of various ELISA kits for
the detection of neomycin, gentamicin,
streptomycin and DHS in bovine milk.
ORGANISATION OF THE PROFICIENCY TEST
Test materials  12 random coded frozen samples
were sent in dried ice (including 2 blank samples
and 10 spiked milk samples) STR at 100 ng/ml
(0.5MRL) and at 300 ng/ml (1.5MRL) in blind
duplicate, DHS at 100 ng/ml (0.5MRL) and at 300
ng/ml (1.5MRL), GTM at 50 ng/ml (0.5MRL) and at
150 ng/ml (1.5MRL), NEO at 75 ng/ml (0. 5MRL)
and at 225 ng/ml (0.15MRL). It is important to
underline first that the NEO samples were 10
times less concentrated than it was planned. So
all the samples were well below the MRL.
Instructions to the participants  Each sample
had to be analysed in triplicate (that means 3
different extractions for each sample) with the
ELISA kits of their choice (commercial or
in-house), if possible looking for STR/DHS, GTM
and NEO. If not, it was recommended to test only
STR/DHS twice, with two different batches of the
same kit. It was underlined that each sample may
contain or not aminoglycosides in the following
list STR, DHS, GTM, NEO. Homogeneity and
stability studies  The 10 spiked samples were
tested according to the reference publications
2,4. It was concluded that all materials were
sufficiently homogenous and were stable until the
analyses deadline.
RESULTS AND DISCUSSION    
Quantitative evaluation A quantitative
exploitation was also performed by calculating
the assigned value for each material 2. Then
the evaluation of laboratorys performance was
carried out by calculating accuracy and
repeatability z-scores 4.
The impact of batch kits and laboratories The
different ELISA kits used by the 14 participants
for their analyses
Examples of the graphical representation of the
individual lab results
Combined accuracy and repeatability z-scores STR
300 µg/kg
Accuracy z-scores GTM 150 µg/kg
4.8
6.2
In grey colour the labs which used the same batch
of kits.
Satisfactory z-scores are included between 2 and
2 ( ).
A statistical evaluation was performed to
evaluate the quantitative effect of different
parameters batches, laboratories and various
kits. The mean calculated concentrations were
compared (t-test (Student test) or one-way
analysis of variance).
YES effect of the parameter NO No effect of the
parameter
Performing the analyses of STR/DHS with different
batches of the same kit in one lab did not
produce significantly different quantitative
results in term of concentration. Moreover the
decision (compliant or non-compliant result) was
the same in all labs. Furthermore the decision
taken by different labs was the same for all
aminoglycosides and all kits suppliers, whichever
was the batch (except in lab Z).

• Whichever was the manufacturer of the ELISA kit
  or the Biacore kit, questionable or
  unsatisfactory z-scores were found.
• High variability of the quantitative results (16
  labs) for STR/DHS.
• The assigned values for the STR/DHS materials
  were very near to the real concentration of the
  spiked samples.
• Regarding the individual laboratory results, the
  STR/DHS tests can not be considered as
  quantitative tests. Some laboratories always give
  the lowest concentrations (for example F and Z)
  and other laboratories the highest results
  (laboratory H). However accuracy z-scores for
  most of the labs (except labs H and Z mainly)
  were satisfactory.
• The repeatability of the results for STR/DHS was
  very satisfactory whichever was the kit.
• With the GTM kits, concentrations far from the
  assigned values were only obtained by lab H
  (Eurodiagnostica kit as other labs). The assigned
  values were very satisfactory.
• Considering NEO kits, the variability was very
  low. All the 6 laboratories found concentrations
  lower than the MRL (1500 µg/l) ?Satisfactory

• Qualitative evaluation
• Two steps should be considered in a screening
  test
• the performance of the kit and its capacity to
  detect one analyte at concentrations upper than a
  threshold value
• the decision of the lab concerning the presence
  or absence of the analyte (non-compliant or
  compliant) in the sample.
• Then depending on the laboratory strategy a
  non-compliant decision is followed or not by a
  confirmation by physico-chemical methods.
• Definitions
• A false compliant sample a sample declared
  compliant while it contains the aminoglycoside
  (s) detected by the concerned ELISA kit
  (whichever was the real concentration).
• A false non-compliant sample a sample declared
  non-compliant whereas it does not contain any
  aminoglycoside nor the aminoglycoside (s)
  detected by the concerned ELISA kit (whichever
  was the real concentration).
• Then according to the Mac Clure publication 3
  for the validation of screening qualitative
  methods, four parameters were calculated False
  non-compliant and false compliant rates,
  sensitivity and specificity.

CONCLUSION
This inter-laboratory testing study on ELISA kits
for the screening of aminoglycosides in milk was
globally satisfactory. The global false compliant
rates (0.0 to 1.0 ) were lower than 5 at the
screening step whichever was the kit used. The
problem is only at the decision step because some
laboratories took a wrong decision. The global
false non-compliant rates were included between
14 and 45 . So a confirmation step is needed to
take the right decision. Finally these kits could
be considered as satisfactory tests for screening
purposes (with qualitative results).
Acknowledgements Many thanks to all the
participants to this inter-laboratory study G.
Suhren, M. Brandtner, I. Shwaiger, Y. Govaert, W.
Rybroeck, U. Pertilla, N. Cadieu, A.M. Ferrini,
C. Arts, S. Stead, L. Lynas, A. Honzlova and V.
Titajevs.
References

• Council Regulation n2377/90 of 26 June 1990,
  L251, 1-8
• 2. ISO /DIS 13528 working document Statistical
  methods for proficiency testing by
  interlaboratory comparisons.
• 3. McClure, F.D. (1990) Design and analysis of
  qualitative collaborative studies minimum
  collaborative program. JAOAC Int., 73 (6),
  953-960
• 4. Thompson, M. WOOD, R. (1993) International
  harmonised protocol for proficiency testing of
  analytical laboratories. JAOAC Int., 76 (4),
  926-940