The performances of three enzyme linked immunosorbent assays (ELISAs)
and one radioimmunoassay (RIA) in 17 alpha-19-nortestosterone (17 alph
a-19-NT) analysis of bovine urine have been evaluated. Sample preparat
ion was performed by enzymic deconjugation and solid-phase extraction.
The main object of the study was the evaluation of the matrix effect,
which is generally great in 19-NT immunoassays. Fifty-seven bovine ur
ine samples have been analysed by immunoassay and by gas chromatograph
y-mass spectrometry (GC-MS). The defection limit of GC-MS analysis was
1 mu g l(-1). The mean sample blank values in urine analysis were clo
se to 1.5 mu g l(-1) for one ELISA, and 0.4-0.8 mu g l(-1) for the oth
er assays (n = 55, samples coming from several breeds in different geo
graphic areas). The detection limits, calculated in compliance to the
EEC criteria, were 2.67, 1.97, 2.05 and 5.01 mu g l(-1) for RIA, ELISA
s 1, 2 and 3, respectively. The percentage of fate positive results ad
opting this criteria were: 1.8, 3.6, 1.8 and 3.6%. The probability of
the occurrence of false negative results was evaluated by adding 2 mu
g l(-1) of 19-NT to a negative sample, was high in each assay, Two;sam
ples from experimentally treated animals gave positive results both in
GC-MS and in immunoassays. The results obtained confirm that be emplo
ying immunoassay and solid-phase extraction as sample clean-up, it is
not possible to apply the action level (2 ppb) suggested by the Europe
an Union (EU).