B-093 Establishing measurement uncertainty for tumor markers: impact of analytical variation in quality control material vs patient serum matrix

Abstract

The College of American Pathologists (CAP) recommends estimating measurement uncertainty (MU) during validation (typically by analyzing replicates of quality control [QC] material over 5 days); and re-evaluating MU estimates in response to a clinical inquiry (typically by a top-down approach using extended length of QC data). Since the results obtained on QC material are often not commutable with patient serum test results for tumor markers, this study was conducted to compare precision (a component of MU) observed in patient serum matrix and the QC material. Five to 10 consecutive patients serum samples were pooled to prepare 3 groups of 15 pools (PPS): 5 around the lower end of analytical measurement range (AMR) (level 1), 5 near the reference interval/cutoff (level 2), and 5 around the upper end of AMR (level 3) for each of 3 tumor markers. PPS were centrifuged to remove any debris, aliquoted and stored in capped tubes in refrigerator. Each PPS was tested for cancer antigens 19-9 (CA 19-9), 27.29 (CA 27.29), and carcinoembryonic antigen (CEA) on Siemens ADVIA Centaur XP analyzer (Siemens Healthcare Diagnostics, Tarrytown, NY). Testing for each PPS was repeated 5 times over a period of 5 days. Three levels of QC materials near (low AMR, cutoff, high AMR) were also tested in the same fashion. The coefficient of variation (CV) was calculated to compare the precision of QC materials to PPS. The average CV observed for 5 replicates of QC and PPS was calculated for comparison. Table 1 shows the precision data of QC materials and PPS. For CA 19-9, QC levels 1, 2, and 3 demonstrated CVs of 6.51%, 6.36%, and 4.57%; PPS levels 1, and 3 demonstrated CVs of 9.24%, 2.92%, and 2.84%, respectively. For CA 27.29, QC CVs were 7.15%, 2.85%, and 2.41%; PPS CVs were 3.34%, 5.10% and 1.52%. For CEA, QC CVs were 6.34%, 2.78%, and 2.17%; PPS CVs were 2.40%, 1.73% and 1.87%. When comparing the average CV of QC materials with that of the 5 PPS over a 5-day period, 7 of 9 (78%) QC samples exhibited higher imprecision than the PPS. Because the QC materials demonstrated greater imprecision than the PPS in 78% of cases, these findings suggest that MU for the tumor markers CA 19-9, CA 27.29, and CEA may need to established using patient-based matrices rather than commercial controls.