BACKGROUND: There is ongoing debate on how variations in surgical technique affect outcomes in pancreatic cancer. This population-based study examines current surgical practice and outcomes for cancer of the pancreatic head. METHODS: All patients 18 to 85 years old diagnosed with nonmetastatic adenocarcinoma of the pancreatic head from 1998 through 2003 were identified from the Surveillance, Epidemiology and End Results (SEER) Program registry. Multivariable regression was used to elucidate factors associated with the type of pancreaticoduodenectomy performed, extent of lymph node (LN) assessment, early mortality, and late survival. RESULTS: Overall, 2111 patients were included in the study, with 83.7% treated with a standard Whipple pancreaticoduodenectomy (PD). However, there was marked regional variation in the use of pylorus-preserving pancreaticoduodenectomy (PPPD; range, .03%-32.0%; P < .0001) and total pancreatectomy (TP; range, .04%-19.5%; P < .0001). TP was associated with significantly higher early mortality (odds ratio, 2.6; 95% confidence interval, 1.6 to 4.1; P < .0001), but late survival did not differ significantly between TP, PPPD, and PD (P = .69). Significant variation was also seen in the number of LN assessed (range across SEER regions, 7.3-13.5; P < .0001). Decreased LN assessment reduced the odds of diagnosing a patient as node positive and was associated with worse late survival. CONCLUSIONS: In this population-based study, we found marked clinically important variability in the surgical treatment of adenocarcinoma of the pancreatic head, with respect to the use of TP, PPPD, or PD, and the extent of LN assessment. Further research is warranted to elucidate the underlying reasons, and to clarify the role of adequate lymphadenectomy.
BACKGROUND: There is ongoing debate on how variations in surgical technique affect outcomes in pancreatic cancer. This population-based study examines current surgical practice and outcomes for cancer of the pancreatic head. METHODS: All patients 18 to 85 years old diagnosed with nonmetastatic adenocarcinoma of the pancreatic head from 1998 through 2003 were identified from the Surveillance, Epidemiology and End Results (SEER) Program registry. Multivariable regression was used to elucidate factors associated with the type of pancreaticoduodenectomy performed, extent of lymph node (LN) assessment, early mortality, and late survival. RESULTS: Overall, 2111 patients were included in the study, with 83.7% treated with a standard Whipple pancreaticoduodenectomy (PD). However, there was marked regional variation in the use of pylorus-preserving pancreaticoduodenectomy (PPPD; range, .03%-32.0%; P < .0001) and total pancreatectomy (TP; range, .04%-19.5%; P < .0001). TP was associated with significantly higher early mortality (odds ratio, 2.6; 95% confidence interval, 1.6 to 4.1; P < .0001), but late survival did not differ significantly between TP, PPPD, and PD (P = .69). Significant variation was also seen in the number of LN assessed (range across SEER regions, 7.3-13.5; P < .0001). Decreased LN assessment reduced the odds of diagnosing a patient as node positive and was associated with worse late survival. CONCLUSIONS: In this population-based study, we found marked clinically important variability in the surgical treatment of adenocarcinoma of the pancreatic head, with respect to the use of TP, PPPD, or PD, and the extent of LN assessment. Further research is warranted to elucidate the underlying reasons, and to clarify the role of adequate lymphadenectomy.