The comparison of prognosis and risk factors after radical resection for pancreatic ductal adenocarcinoma between China Pancreas Data Center and SEER

Abstract

Objectives: To compare the prognosis of patients underwent radical resection for pancreatic ductal adenocarcinoma(PDAC) in Surveillance, Epidemiology, and End Results(SEER) and China Pancreas Data Center(CPDC), and to compare the prognostic factors for PDAC in both databases. Methods: The data of patients underwent radical resection for PDAC in CPDC database from January 2016 to December 2017 and SEER database from January 2014 to December 2015 were retrospectively analyzed. The prognosis of patients in both databases was analyzed by the Kaplan-Meier method, Log-rank method, and propensity score matching, and the Cox proportional hazard regression was used to analyze the independent prognosis factors for PDAC. Results: There were 1 977 cases and 2 220 cases of pancreatic cancer that underwent radical resection from CPDC and SEER, respectively. There were more male patients(60.90%,1 204/1 977) than female patients(39.10%,773/1 977) in CPDC, while nearly 1∶1 ratio(male:1 112 cases,female:1108)was observed between male and female in SEER(χ²=48.977,P<0.01). The proportion of patients under 45 years old was the smallest in both databases, and the age group with the most significant proportion was 60 to 74 years old. The ratio of patients over 75 years old in the SEER(24.28%,539/2 220) was higher than that of CPDC(7.89%,156/1 977)(χ²=202.090,P<0.01), while the proportion of patients between 45 and 59 years old in CPDC(33.69%,666/1 977) was higher than that in SEER(19.77%,439/2 220)(χ²=103.640,P<0.01). There were more pancreatic head cancers than body and tail cancers in both databases, and no statistical difference was found in tumor size between the two databases (W=2 181 502,P=0.740). More positive and examined lymph nodes were found in SEER patients (W=3 265 131,W=2 954 363,all P<0.01); and the proportion of patients who had at least 15 lymph nodes dissected was higher in SEER(63.24%,1 404/2 220)(χ²=532.130,P<0.01). There were more patients without neoadjuvant or adjuvant therapy in CPDC(57.16%,1 130/1 977) than that in SEER(24.91%,553/2 220)(χ²=451.390,P<0.01). After propensity score matching, the overall survival for CPDC was better than that for SEER(Log-rank test:χ²=4.500,P=0.034), and the median overall survival was 24 months and 23 months respectively. Cox regressional analysis showed the common independent prognosis factors in both databases were ≥75 years old, pancreatic head cancer, poorly differentiated and undifferentiated tumors, T stage, N stage(All P<0.05). Neoadjuvant or adjuvant therapy was a protective factor in both databases(CPDC:Wald=27.082;SEER:Wald=212.285, all P<0.01) and 45 to 59 years old was protective factor in the SEER database(Wald=5.212,P=0.020). Conclusions: The data in both databases have a good consistency. However, in terms of data quality, examined lymph nodes count, and neoadjuvant/adjuvant therapy rate, the CPDC differs greatly from the SEER.