The effects of common matrices for assay standards on performance of 'ultra sensitive' immunometric assays for TSH. Report of a joint WHO/IFCC collaborative study.

This report describes the results of a collaborative study organized by a joint working group of IFCC and WHO and involving nine manufacturers of TSH immunometric assay kits. The study was designed to determine whether a calibrator with a common matrix gives better between-laboratory agreement for calibration of serum samples than the various kit calibrators, and to assess various materials for their suitability for use as common matrices. Kit calibrators or calibrators consisting of the IRP for TSH made up in two common matrices, (1) serum from patients with untreated thyrotoxicosis or (2) serum taken from subjects treated with suppressive doses of triiodothyronine, gave similar results for the between-laboratory variation of estimates of TSH concentration for a range of serum samples. Dose-response curves for the two calibrators in 'common' matrices were similar to one another and to those for the kit calibrator. However, the occurrence of non-specific serum effects is shown by the comparison of results for these calibrators with results for calibrators made up in a third common matrix, serum treated with wheat germ lectin. Dose-response curves for this calibrator were dissimilar to those for the other calibrators and between-laboratory variation for estimates in terms of this latter calibrator showed a substantial increase. Moreover, although the between-laboratory variances for estimates of the TSH concentration in terms of each of these calibrators (except those made up in serum treated with in the wheat germ lectin) were similar for any one sample from five hyperthyroid patients, the variances were not consistent between samples, even for samples with similar mean TSH concentrations. These results suggest that a major factor in the between-laboratory variation, especially in the region near 'zero dose', is sample-related, and is caused by particular samples interacting differently with different assay systems. In general, it would appear that for the well controlled 'ultra-sensitive' TSH immunometric assay kits, included in this study, between-laboratory agreement of estimates of the TSH concentration in serum samples is not likely to be substantially improved by use of a common matrix for the standards.