血漿早期膵がんマーカーの多施設共同検証

抄録

We previously identified a plasma biomarker set that was able to distinguish pancreatic cancer patients from healthy controls with high accuracy (Honda et al., Cancer Res., 65:10613-22, 2005). Here we report the preliminary results of an ongoing validation study using 549 plasma samples [pancreatic cancer (n=227), gastric cancer (n=68), colorectal cancer (n=31), other cancers (n=2), benign pancreatic disorders (n=21), other benign disorders (n=8), and healthy controls (n=192)] collected at seven medical institutions. All the samples were blinded and processed randomly using a robotic magnetic bead-based hydrophobic interaction chromatography system. The processed samples were randomly allocated to 16 wells of 384-well MALDI plates, and MS spectra were obtained within the range 1000-30,000 m/z using a high-resolution hybrid quadrupole MALDI-TOF-MS. As a result of the highly repetitious measurement protocol, the correlation coefficient (CC) and coefficient of variance (CV) values for all the corresponding MS peaks (>2200 peaks per sample) were constantly kept within 0.993 and 0.0392, respectively, throughout the experiments. We identified the amino acids sequence of previously two reported biomarkers (17252 m/z and 8765 m/z). In addition, we confirmed that the intensities of two biomarker proteins combination (17252 m/z and 8765 m/z) differed between pancreatic cancer patients (n=227) and healthy controls (n=186) with high statistical significance [Mann-Whitney U test, P=1.11X10E-42 and AUC (area under curve) value >0.888]. In order to analyze a sufficient number of plasma samples and establish the clinical utility of the biomarkers, prospective collection of well-characterized clinical samples will be necessary. A collaborative study involving 7 medical institutions throughout Japan is now underway. This collaborative study is part of the "Third-Term Comprehensive Control Research for Cancer" conducted by the Ministry of Health, Labor and Welfare of Japan.