MEASUREMENT OF PLASMINOGEN-ACTIVATOR INHIBITOR TYPE-1 ANTIGEN - COMPARISON OF TINTELIZE(TM) AND IMUBIND(TM) METHODS

Our specific aim was to compare three plasminogen activator-inhibitor type 1 (PAI-1) antigen ELISA kit assays (the Biopool AB, Ltd, TintEliz e(TM) PAI-1 Strip-Well Format; the American Diagnostica, Inc., Imubind (TM) 822/1; and the second generation Imubind(TM) 822/1S). Within-run coefficients of variation (n = 6) for the TintElize, Imubind 822/1 and Imubind 822/1S methods were 5.5%, 5.9% and 6.8%, respectively. Betwee n-run coefficients of variation for six aliquots per run were 2.9% for TintElize, 3.8% for Imubind 822/1, and 3.5% for Imubind 822/1S. Compa rison of the average of duplicate aliquots from hyperlipidaemic patien ts demonstrated intraclass correlations of 0.75, 0.79 and 0.95 for Tin tElize us Imubind 822/1 (n = 39), TintElize us Imubind 822/1S (n = 39) , and Imubind 822/1 us 822/1S (n = 84), respectively. Lower 95% confid ence interval limits of the intraclass correlation were 0.55, 0.48 and 0.93, respectively. Mean PAI-1 antigen values (n = 39) were 12.1, 15. 8, 15.8 and 16.0 ng/ml, respectively, for TintElize, TintElize without using the quenching well, Imubind 822/1, and Imubind 822/1S. All thre e methods were easily performed and exhibited high correlation and rep roducibility. A significant systematic bias (P < 0.006) existed betwee n TintElize and TintElize without using the quenching well, Imubind 82 2/1, and Imubind 822/1S. However, there was no significant bias when T intElize without using the quenching well is compared with Imubind 822 /1 (P > 0.8) and to 822/1S (P > 0.8) nor is there significant systemat ic bias between Imubind 822/1 and 822/1S (P > 0.3). By convention, int erchangeability between assay methods suggests that the lower limit of the 95% intraclass correlation confidence interval be greater than 0. 75. Thus, the three assay methods for measuring PAI-1 antigen are very similar and directly interchangeable when quenching well correction i s not employed.