Serum neuron specific enolase (S-NSE) reference interval evaluation by time-resolved immunofluorometry compared with a radioimmunoassay.

The serum neuron specific enolase (S-NSE; EC 4.2.1.11) reference interval was evaluated by DELFIA (Wallac) in 161 healthy blood donors and the method compared with the S-NSE RIA assay (Pharmacia). The DELFIA assay total analytical variation coefficient (CV%) was between 3.7% and 6.6%., the RIA CV% 7.6% to 13.1%. Late centrifugation (after hours) increased the variation as a result of contamination with blood cells. Log transformation into a gaussian distribution was selected by Box-Cox analysis and tested by two models: the gauss-distribution and the Refval transformation. The 95% reference intervals and corresponding 90% confidence intervals were: female 2.9-9.6 micrograms/l (2.6-3.2 and 8.5-10.7) micrograms/l and male 3.4-11.7 micrograms/l (3.0-3.8 and 10.2-13.2 micrograms/l). Mean values were significantly different (P < 0.001), female 5.3 (4.9-5.6), male 6.3 (5.8-6.7) micrograms/l. The serum NSE levels were analysed with both methods in a population of 110 patients. The results were significantly correlated (coefficient, 0.9896; r, 0.99; P < 0.0001-two tailed). For high S-NSE values (> 150.0 micrograms/l) differences between the methods exceeded the mean difference + 2S.D., while low concentrations were interconvertible. Maximal diagnostic efficacy was 0.91 for both assays, in DELFIA 17.2-23.9 micrograms/l and for RIA 17.2-21.9 micrograms/l. Identical sensitivity, specificity, discriminative power score, and likelihood ratio were found. The two methods are consequently interconvertible.