The volume-outcome relationship has been repeatedly demonstrated for pancreatectomy, but identifying underlying reasons for this association has been challenging. Some have suggested that differences in surgical technique may affect longterm survival, but it is unknown whether margin-positive resection rates vary by hospital volume. Our objective was to evaluate the effect of hospital pancreatectomy volume on margin status.
Patients who underwent pancreaticoduodenectomy for localized pancreatic adenocarcinoma were identified from the National Cancer Data Base (1998 to 2004). Regression modeling adjusting for patient, tumor, and hospital factors was used to assess predictors of margin involvement and to evaluate the effect of margin status on survival. Volume quintiles were based on average annual hospital pancreatectomy volume.
Of 12,101 patients, 24.4% had positive resection margins (14.6% microscopic/R1; 9.8% macroscopic/R2). From 1998 to 2004, there was not a significant change in margin-positive resection rates (p = 0.43). On multivariable analysis, patients were more likely to have a margin-positive resection if they had a higher T classification or nodal involvement, were uninsured or living in lower-income areas, or underwent resection at lowest-volume hospitals compared with highest-volume hospitals (25.9% versus 22.6%, p < 0.0001; odds ratio, 1.21; 95% confidence interval, 1.01 to 1.43). On multivariable analysis, margin involvement was associated with a higher risk of longterm mortality compared with margin-negative resections (p < 0.0001).
Involved resection margins are a poor prognostic factor after a pancreaticoduodenectomy. Patients undergoing pancreaticoduodenectomy at low-volume centers are more likely to have margin-positive resections. Standardization of pathologic evaluation for pancreatectomy specimens is needed.