Experiments, showed that point mutations in tumor samples up to 5 tumor content were detectable. This provided confidence that our inclusion of tumor samples, only if those had at least 10 tumor content (n = 171), would more than adequately enable the detection of mutations. Another 
TA-01 site criterion applied for mutation detection was reproducibility. Mutations were scored only when band shifts were reproducible in at least two independent experiments. Repeat experiments using SSCP followed by DNA sequencing were used for confirmation and identification of mutations (Figure S2). We also obtained independent confirmation of KRAS mutations in a random sub-set (n = 6) analyzed blindly in the reference laboratory of the Institute of Pathology, University Hospital of Heidelberg. In the KRAS gene, we detected 134 mutations in 171 tumors (78 ), with 131 mutations in exon 2 and 3 mutations in exon 3 (Table 1). Mutations in exon 2 in all tumors were localized to codon 12. Out of 131 tumors that carried mutation at codon 12, 61 tumors had GGT.GAT (G12D, 80 of 131) mutation, followed by GGT.CGT (G12R, 23 of 131, 18 ), GGT.GTT (G12V, 22 of 131, 17 ), GGT.TGT (G12C, 4 of 131, 3 ), GGT.GCT (G12A, 1 of 131) and GGT.GTC (G12V, 1 of 131). Three tumors carried mutations in exon 3 that were confined to codon 61 featuring the Q61H mutation due to CAA.CAC base change. The mutation frequency in ductal adenocarcinomas was 82 (117 of 143) including adenosquamous and order Tetracosactide anaplastic undifferentiated tumors. All 4 of the ampulla of Vater tumors showed KRAS mutation, while 7 of 9 IPMN-malignant types harbored mutation (Table 1 and Table S3). A total of 43 tumors (25 ) showed 16574785 aberrations in the CDKN2A gene. Of the CDKN2A alterations in 43 tumors, 9 carried point mutations and the remainder showed deletion at the locus. All the point mutations in the gene were located in exon 2. Two tumors carried mutation at codon 80 (CGA.TGA, R80*), 3 at codon 83 (CAC.TAC, H83Y), followed by solitary tumors with mutations at codon 58 (CGA.TGA, R58*), codon 129 (TAC.TAA, Y129*), codon 130 (CTG.CAG, L130Q) and one tumor had 2 base pair insertion of GG at codon 78 (CTC.CGGTC). Deletions at the 9p21 locus were detected with varying frequency with 17?20 in the CDKN2A (p16INK4a) and 26?8 within the promoter associated with exon 1b of p14ARF transcript. Univariate analyses showed that among clinico-pathological factors, only tumor grade significantly affected 
overall survival in the studied cohort (Table 1). Presence of KRAS mutations tended to shorten survival of patients in general (n = 150; P = 0.07) and inall studied sub-categories (except tumor stage T4), however without reaching statistical significance (Table S2). In 150 patients with malignant exocrine tumors, the activating KRAS mutations were associated with reduction in median survival time nearly by half (17 vs 30 months, Kaplan-Meier method with log-rank test P = 0.07; Figure S3A). The presence of KRAS mutations was associated with poor survival in tumor stage III (HR = 1.94, P = 0.03; Table S2). Risk factors such as smoking, alcohol consumption or diabetes had no effect on patient survival either with or without KRAS mutations. A multivariate Cox regression model that included age, gender, TNM, tumor grade and tumor histology as co-variants confirmed KRAS mutational status as a potential independent prognostic marker with a hazard ratio (HR) of 1.87 (95 CI 0.99?.51, P = 0.05; Table 2). Analysis with specific types of KRAS mutati.Experiments, showed that point mutations in tumor samples up to 5 tumor content were detectable. This provided confidence that our inclusion of tumor samples, only if those had at least 10 tumor content (n = 171), would more than adequately enable the detection of mutations. Another criterion applied for mutation detection was reproducibility. Mutations were scored only when band shifts were reproducible in at least two independent experiments. Repeat experiments using SSCP followed by DNA sequencing were used for confirmation and identification of mutations (Figure S2). We also obtained independent confirmation of KRAS mutations in a random sub-set (n = 6) analyzed blindly in the reference laboratory of the Institute of Pathology, University Hospital of Heidelberg. In the KRAS gene, we detected 134 mutations in 171 tumors (78 ), with 131 mutations in exon 2 and 3 mutations in exon 3 (Table 1). Mutations in exon 2 in all tumors were localized to codon 12. Out of 131 tumors that carried mutation at codon 12, 61 tumors had GGT.GAT (G12D, 80 of 131) mutation, followed by GGT.CGT (G12R, 23 of 131, 18 ), GGT.GTT (G12V, 22 of 131, 17 ), GGT.TGT (G12C, 4 of 131, 3 ), GGT.GCT (G12A, 1 of 131) and GGT.GTC (G12V, 1 of 131). Three tumors carried mutations in exon 3 that were confined to codon 61 featuring the Q61H mutation due to CAA.CAC base change. The mutation frequency in ductal adenocarcinomas was 82 (117 of 143) including adenosquamous and anaplastic undifferentiated tumors. All 4 of the ampulla of Vater tumors showed KRAS mutation, while 7 of 9 IPMN-malignant types harbored mutation (Table 1 and Table S3). A total of 43 tumors (25 ) showed 16574785 aberrations in the CDKN2A gene. Of the CDKN2A alterations in 43 tumors, 9 carried point mutations and the remainder showed deletion at the locus. All the point mutations in the gene were located in exon 2. Two tumors carried mutation at codon 80 (CGA.TGA, R80*), 3 at codon 83 (CAC.TAC, H83Y), followed by solitary tumors with mutations at codon 58 (CGA.TGA, R58*), codon 129 (TAC.TAA, Y129*), codon 130 (CTG.CAG, L130Q) and one tumor had 2 base pair insertion of GG at codon 78 (CTC.CGGTC). Deletions at the 9p21 locus were detected with varying frequency with 17?20 in the CDKN2A (p16INK4a) and 26?8 within the promoter associated with exon 1b of p14ARF transcript. Univariate analyses showed that among clinico-pathological factors, only tumor grade significantly affected overall survival in the studied cohort (Table 1). Presence of KRAS mutations tended to shorten survival of patients in general (n = 150; P = 0.07) and inall studied sub-categories (except tumor stage T4), however without reaching statistical significance (Table S2). In 150 patients with malignant exocrine tumors, the activating KRAS mutations were associated with reduction in median survival time nearly by half (17 vs 30 months, Kaplan-Meier method with log-rank test P = 0.07; Figure S3A). The presence of KRAS mutations was associated with poor survival in tumor stage III (HR = 1.94, P = 0.03; Table S2). Risk factors such as smoking, alcohol consumption or diabetes had no effect on patient survival either with or without KRAS mutations. A multivariate Cox regression model that included age, gender, TNM, tumor grade and tumor histology as co-variants confirmed KRAS mutational status as a potential independent prognostic marker with a hazard ratio (HR) of 1.87 (95 CI 0.99?.51, P = 0.05; Table 2). Analysis with specific types of KRAS mutati.