Clin Chem Lab Med 2007;45(7):928–929  2007 by Walter de Gruyter • Berlin • New York. DOI 10.1515/CCLM.2007.134 2006/56 Article in press - uncorrected proof Letter to the Editor Could standard ISO 15189 be useful for the conundrum of CA 19-9/GI Monitor measurement? Figure 1 Mountain plot of the comparison between DxI and Liaison analyzers. Romolo M. Dorizzi*, Sandra Meneghelli, Maria Rocca and Anna Ferrari Laboratorio Analisi Chimico-Cliniche ed Ematologiche, Azienda Ospedaliera di Verona, Verona, Italy Keywords: CA 19-9; GI Monitor; ISO 15189. The letter by Fuentes-Arderiu (1) and the paper by Passerini et al. (2) published in the January 2007 issue of Clinical Chemistry and Laboratory Medicine pro- voke some comments. The accreditation process appears to be increasingly intertwined with the daily activity of clinical laboratories of all sizes. The sub- clauses pointed out by Fuentes-Arderiu (1) regarding the criteria to consider when acquiring a measure- ment system are very relevant. Nevertheless, the eight steps indicated by him are not easy to imple- ment in many cases, as demonstrated by Passerini et al. (2). The authors, using Passing-Bablok regression and a mountain plot, showed very effectively that some analyzers yield very similar results and others are much less interchangeable in measuring CA 19-9. The matter becomes even more muddled when we extend the comparison to systems that measure CA 19-9 using antibodies different from 1116NS-19-9 (Mab 19-9), such as clone C192:22:5 used in the Access GI Monitor assay, which is a paramagnetic particle, chemiluminescent immunoassay for the quantitative determination of CA 19-9 antigen levels in human serum and plasma using Access Immuno- assay Systems and the UniCel DxI platform marketed by Beckman Coulter. These systems, approved by the FDA as ‘‘substan- tially equivalent’’ to other devices marketed for CA 19- 9 quantitation, have been reported to provide clinically reliable results, even if, of course, they are less comparable to other analyzers employing Mab 19-9 (3). We recently had the chance to compare results obtained for GI Monitor measurement on a Unicel DxI 800 analyzer (Beckman Coulter, Fullerton, CA, USA) (DxI) and CA 19-9 measurement on a Liai- son system (Diasorin, Saluggia, Italy) (Liaison) in 53 *Corresponding author: Dr. med. Romolo M. Dorizzi, Laboratorio Analisi Chimico-Cliniche ed Ematologiche, Azienda Ospedaliera di Verona, Piazzale Stefani 1, 37126 Verona, Italy Phone: q39-045-8073045, Fax: q39-045-74431187, E-mail: romolo.dorizzi@azosp.vr.it serum samples collected form patients suffering from pancreatic cancer (4). Both systems are fully auto- mated, high-throughput chemiluminescence analyz- ers with a very wide panel of immunoassays, including several tumor markers. Measurements were carried out in duplicate and statistical calculations were carried out using MedCalc release 9.2 (MedCalc, Mariakerke, Belgium) and Mountain Plot software kindly provided by Dr. Krouwer. Liaison, the routine method used in our laboratory, and DxI yielded the following respective results: lowest value, 1 and 0.8 kU/L; highest value, 445.4 and 667.8 kU/L; mean, 68.5 and 80.4 kU/L; SD, 85.57 and 105.95 kU/L; and median, 37.4 and 52 kU/L. We calculated Passing- Bablok regression as DxIs4.27 w 95% confidence inter- val (CI): y0.3761 to 8.5372xq1.16 (95% CI 1.026– 1.376) Liaison, with a correlation coefficient of rs 0.911 (95% CI 0.850–0.948). Figure 1 shows a moun- tain plot of CA 19-9 concentrations and Figure 2 shows a difference plot of the concentrations after logarithmic transformation. The mountain plot con- firms relevant dispersion of the results for the two analyzers, while the bias is greater than the values reported by Passerini et al., probably due to the dif- ferent origin of the antibody used by the DxI and Liai- son systems. The difference plot after logarithmic transformation confirms systematic bias without par- ticular trends. Our results uphold the notion that using the same or very similar cutoff values for assays when bias is so relevant and comparability is so low can misguide laboratorians and clinicians. Pas- serini et al. (2) stated that CA 19-9 results obtained with different platforms are no more interchangeable in 2007 than in 2001 (5). Laboratory professionals, manufacturers and regulatory agencies should pro- actively implement their recommendations to note the method/platform used for CA 19-9 in the reports of clinical laboratories. There is no doubt that manu- facturers have been slow in their efforts to harmonize and standardize immunoassays. Therefore, laborato-