Institutional members access full text with Ovid®

Conditional Survival Probabilities for Patients With Resected Pancreatic Adenocarcinoma

Mishra, Mark V. MD*; Champ, Colin E. MD*; Keith, Scott W. PhD; Showalter, Timothy N. MD*; Anne, Pramila R. MD*; Lawrence, Yaacov Richard MD*; Bar-Ad, Voichita MD*

American Journal of Clinical Oncology: April 2014 - Volume 37 - Issue 2 - p 107–111
doi: 10.1097/COC.0b013e31826c62b7
Original Articles: Gastrointestinal

Background: The purpose of this study is to evaluate conditional survival probabilities for patients with resected pancreatic adenocarcinoma (PC).

Methods: Patients with resected PC from 1998 to 2008 were identified from the Surveillance, Epidemiology and End Results Database. Data on patient, tumor, and treatment characteristics were extracted. Overall survival (OS) rates were calculated using the Kaplan-Meier method. A multivariable analysis at different time points from survival was performed to determine independent prognostic factors associated with all-cause mortality hazard ratios using Cox proportional hazards models.

Results: A total of 4883 patients with resected PC were identified. The 1-, 3-, and 5-year survival estimates for patients at diagnosis were 67%, 29%, and 21%, respectively. The probability of surviving an additional 1-, 3-, or 5-year conditional upon already surviving 5 years after diagnosis were 89%, 76%, and 71%, respectively. Prognostic factors significantly correlated with improved OS at the time of diagnosis on multivariable analysis include: earlier stage, younger age, later year of diagnosis, white ethnicity, female sex, and residence in a high income district (P<0.05). Among those already surviving 3 years after diagnosis, younger age was the only prognostic factor statistically significantly correlated with improved OS (P<0.05).

Conclusions: Conditional survival estimates provide additional prognostic information that may be used to counsel PC patients on how their prognosis may change over time. Further research using prospectively collected data is warranted to help determine recommended follow-up intervals and benchmarks for future clinical trials.

*Department of Radiation Oncology, Jefferson Medical College

Division of Biostatistics, Thomas Jefferson University, Philadelphia, PA

Research in this publication includes work carried out by the Kimmel Cancer Center, which is supported in part by NCI Cancer Center Support Grant P30 CA56036.

The authors declare no conflicts of interest.

Reprints: Mark V. Mishra, MD, Thomas Jefferson University, Bodine Cancer Center, 111 S. 11th Street, Philadelphia, PA 19107. E-mail: mishramv@gmail.com

© 2014 by Lippincott Williams & Wilkins, Inc