Analytical conditions in a system for capillary zone electrophoresis (Beckman Paragon CZE 2000) were originally selected to allow serum protein separation into five discrete protein zones, corresponding to those of conventional clinical electrophoresis. To improve the system's performance, new analytical conditions have been made available. We compared the two sets of conditions ("new" = y; "old" = x) for possible variations of results caused by the change. One hundred thirteen serum samples, covering wide intervals of values, were assayed on two twin instruments working under the old and the new conditions; results were assessed statistically and graphically. Possible clinical significance of differences was checked by comparison with the biological variation-based quality specifications for bias. Statistically significant (y-x) differences were observed for the alpha1-, alpha2- and beta-globulin zones; clinically significant differences were observed for all the zones, with the exception of the gamma-globulin zone. Therefore, old/new regression equations were calculated, whose reliability was assured by the wide interval of values, by the large sample size, and by the low dispersion of single values around the mean concordance estimates. Such equations may be used to convert "old" into "new" reference values, and for the intercomparison of patient results obtained under different analytical conditions.

Keywords: artificial neural network; capillary electrophoresis; monoclonal component; serum protein.Serum protein electrophoresis (SPE) is mostly used to screen for the presence of immunoglobulin monoclonal components (MCs) indicative of abnormal clonal expansion of plasma cells. The incidence of monoclonal gammopathies of undetermined significance (MGUS) is estimated to be about 3% in persons older than 70 years (1) and 1.25-1.7% in persons older than 50 years (2, 3). Moreover, long-term follow-up studies (1) reported an incidence of malignant progression of 17% after 10 years and of 21% after 20 years from the diagnosis of MGUS (1). As a result, the risk of progression to plasma cell dyscrasia is considerably higher for MGUS patients compared to the general population, with a 25-fold increase for multiple myeloma, 46-fold for macroglobulinemia and 8fold for primary amyloidosis (1). In medium-large laboratories in Italy, about 100,000 SPE analyses/year are carried out (approx. 5% of total number of tests/ year), but the presence of an MC is evident or suspected in only a small percentage of electropherograms (5-10% in our experience). Visual inspection by a well-trained operator of large numbers of samples per day is therefore required. The inspection aims to identify all suspect samples, at the same time avoiding unnecessary second-level tests, such as serum and urine immunofixation. Moreover, visual inspection is time-consuming and highly affected by intra-and inter-observer variability. The latter is definitely difficult to monitor and quantify.