Objectives To identify new serum biomarkers of pancreatic cancer (PaCa) by MALDI-TOF analysis. Methods Fifty-one PaCa, 38 chronic pancreatitis (ChrPa), 48 controls (24 with type II diabetes mellitus, DM) were studied. Fasting sera were purified by Sep-Pak C18 before MALDI-TOF anchorchip analysis. Results One-hundred and 76 features (1207-5374 m/z) were selected. Seven features were highly correlated (P<0.0001) with disease: two (m/z 3182 and 4009) with PaCa, one (m/z 2049) with ChrPa, four (m/z 1530, 1778, 2006, 2602) were less represented in PaCa and/or ChrPa as compared to controls. Ten-fold cross validation binary recursive partitioning trees for patient classification were obtained. The first tree, which included CA 19-9, age, m/z 2006, 2599, 2602, allowed to well discriminate controls (AUC=0.988) and ChrPa (AUC=0.988) from PaCa (AUC=0.974) (correct classification equal to 90.5%). The second tree (CA 19-9, age, m/z 2006, 2599, 2753, 4997), built considering only patients and controls with DM, allowed to distinguish DM (AUC=0.997) from ChrPa (AUC=0.968) and PaCa (AUC=0.980) (correct classification equal to 90.1%). While CA 19-9 alone did not discriminate localized from advanced PaCa (AUC=0.685), the tree including CA 19-9, 1550 and 2937 m/z features, achieved AUC=0.970. We obtained a successful fragmentation by MALDI-TOF/TOF analysis for three cancer-associated features (1530, 1550, 1778 m/z) which were found to be part of clusterin, apolipoprotein A1 and complement C3. ApoA1 and C3 were measured in all sera, and in a series of 76 new serum samples (26 type II DM, 50 PaCa). ApoA1 was significantly reduced in PaCa with respect to all the other groups (F=13.13, P<0.0001). At multivariate logistic regression analyses (predictors: age, gender, CA 19-9, ApoA1 and C3), only ApoA1 (P<0.0001) was confirmed to be strictly correlated with PaCa, thus further supporting its role as a potential biomarker for this tumour. Conclusion We demonstrated that new serum biomarkers identified using a proteomic approach, significantly enhance the diagnostic performance of CA 19-9.
Discovery of New Serum Biomarkers for Pancreatic Cancer Diagnosis by MALDI-TOF Analysis.
PADOAN, ANDREA;MOZ, STEFANIA;GRECO, ELIANA;FOGAR, PAOLA;PASQUALI, CLAUDIO;BOZZATO, DANIA;ZAMBON, CARLO-FEDERICO;PEDRAZZOLI, SERGIO;PLEBANI, MARIO;BASSO, DANIELA
2011
Abstract
Objectives To identify new serum biomarkers of pancreatic cancer (PaCa) by MALDI-TOF analysis. Methods Fifty-one PaCa, 38 chronic pancreatitis (ChrPa), 48 controls (24 with type II diabetes mellitus, DM) were studied. Fasting sera were purified by Sep-Pak C18 before MALDI-TOF anchorchip analysis. Results One-hundred and 76 features (1207-5374 m/z) were selected. Seven features were highly correlated (P<0.0001) with disease: two (m/z 3182 and 4009) with PaCa, one (m/z 2049) with ChrPa, four (m/z 1530, 1778, 2006, 2602) were less represented in PaCa and/or ChrPa as compared to controls. Ten-fold cross validation binary recursive partitioning trees for patient classification were obtained. The first tree, which included CA 19-9, age, m/z 2006, 2599, 2602, allowed to well discriminate controls (AUC=0.988) and ChrPa (AUC=0.988) from PaCa (AUC=0.974) (correct classification equal to 90.5%). The second tree (CA 19-9, age, m/z 2006, 2599, 2753, 4997), built considering only patients and controls with DM, allowed to distinguish DM (AUC=0.997) from ChrPa (AUC=0.968) and PaCa (AUC=0.980) (correct classification equal to 90.1%). While CA 19-9 alone did not discriminate localized from advanced PaCa (AUC=0.685), the tree including CA 19-9, 1550 and 2937 m/z features, achieved AUC=0.970. We obtained a successful fragmentation by MALDI-TOF/TOF analysis for three cancer-associated features (1530, 1550, 1778 m/z) which were found to be part of clusterin, apolipoprotein A1 and complement C3. ApoA1 and C3 were measured in all sera, and in a series of 76 new serum samples (26 type II DM, 50 PaCa). ApoA1 was significantly reduced in PaCa with respect to all the other groups (F=13.13, P<0.0001). At multivariate logistic regression analyses (predictors: age, gender, CA 19-9, ApoA1 and C3), only ApoA1 (P<0.0001) was confirmed to be strictly correlated with PaCa, thus further supporting its role as a potential biomarker for this tumour. Conclusion We demonstrated that new serum biomarkers identified using a proteomic approach, significantly enhance the diagnostic performance of CA 19-9.
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
