The X2 test and the Fishers exact test were used to test associations between genotypes and categorical vari ables describing the clinico pathologic features of the study population. Survival curves were plotted seriously using the Kaplan Meier and compared using the log rank test. The Cox proportional hazards model was used for multivariate analysis to estimate and test demographic characteristics, clinical and genetic features for their associations with OS. In this exploratory study, no formal correction for mul tiple comparisons was adopted. However, all the following variables were included in multivariate Cox model age, sex, ECOG performance status, weight loss, anemia, albu min level, CEA level, tumor grading, histologic subtype according to Laurens classification, tumor location, liver involvement, presence of peritoneal carcinomatosis, number of metastatic sites and response to first line chemotherapy.

Assuming a 20% lowest frequency for an unfavorable genotype, 157 events would allow to detect an Hazard Ratio of 1. 75 associated with this group. All results were considered significant at two sided p. 05 value. All analyses were performed by using the MedCalc software version 11. 1. Results Characteristics of patients and genotyping One hundred sixty one patients were analyzed. All of them received first and second line chemotherapy and died after gastric cancer progression. First line chemother apy was oxaliplatin or cisplatin plus a fluoropyrimidne in 150 patients, or bolus/infusional 5 Fluorouracil in 11 patients.

Second line chemotherapy was 5 Fluorouracil coupled with cisplatin or oxaliplatin in 48 patients, with CPT 11 in 45 patients, with anthracycline in 33 patients, with paclitaxel or docetaxel in 25 patients, with VP 16 in 10 patients. Median survival time in the whole group was 9. 4 months. Carriers of the rs1800795 G/G, G/C and C/C geno types were 74, 68 and 19, respectively. Carriers of the rs8192284 A/A, A/C and C/C genotypes were 58, 73 and 30, respectively. These frequencies did not show deviation from Hardy Weinberg equilibrium and they are comparable with frequencies commonly observed in Caucasian populations. Details of the characteristics of enrolled patients to gether with their distribution according to rs1800795 and rs8192284 genotypes are shown in Table 1. No significant association was observed except for liver involvement and rs8192284 genotypes.

In particular, rs8192284 C/C car riers were prevalent in patients with liver metastases, while rs8192284 A/A carriers were prevalent in patients without liver metastases. Survival analyses Survival Dacomitinib curves of carriers of the rs1800795 and rs8192284 genotypes are shown in Figure 1. In carriers of the rs1800795 G/G, G/C and C/C genotypes, median survival times were 8. 4, 11 and 12. 6 months, respectively. In carriers of the rs8192284 A/A, A/C and C/C genotypes median survival times were 11. 7, 10. 1 and 8.