EVALUATION OF PRECISION AND RELATIVE ACCURACY OF A NEW AUTOMATED LUMINESCENCE IMMUNOASSAY FOR TSH AND FT4

The chemiluminescence immunoasssay system LIA-mat S 300 for TSH and fT 4, developed by Byk-Sangtec Diagnostica (Dietzenbach, Germany), has be en evaluated in a multicenter collaborative study among 5 laboratories , using control materials previously circulated in a national External Quality Assessment (EQA) program. LIA-mat S 300 is an open, fully aut omated system in which all the analytical steps (pipetting of samples and reagents, mixing, incubation, washing, luminescent measurement and data reduction) are controlled by a programmable computer. TSH is mea sured by a noncompetitive technique using two different monoclonal ant ibodies. In fT4 assay the free hormone in the sample and a fixed amoun t of T4 conjugate -coated on the tube-compete for binding to a monoclo nal anti T4 antibody labeled with an isoluminol derivative. The workin g range of LIA-mat TSH (computed from the precision profile, as the co ncentration range measurable with a within-assay precision better than 10 CV%) extended from 0.09 to 80 mu IU/mL; the sensitivity was 0.017 mu IU/ mL. The mean inprecision in the concentration range 0.25-43 mu IU/mL, were 6.0 CV% (between-assay) and 8.7 CV% (between-lab). The com parison of the results from LIA-mat and those obtained by other method s/kits on EQA control samples indicated a satisfactory agreement. The working range of fT4 extended from 0.37 to 8.96 ng/dL and the sensitiv ity was 0.07 ng/dL. The mean between-assay variability was 8.8 CV% for samples at concentration below 1.0 ng/dL and 5.7 CV% for higher sampl es; the between-lab variability was 10.2 CV%. The relative accuracy of LIA-mat fT4 has been evaluated by comparison with the mean of the met hods/kits more used in the EQA; we observed that the agreement between the different methods, including LIA-mat fT4, is satisfactory for con trol samples in the normal and in the hyperthyroid range. Moreover the measurement of LIA-mat fT4, as theoretically predicted, is not affect ed by dilution, it produces a result in the lower normal range for a p regnancy control pool and finds in a FDH (Familial Disalbuminemic Hype rthyroxinemia) pool a value in the borderline normal-hyperthyroid rang e.