Pancreatic cancer is the third leading cause of cancer-related death in the United States. Nearly half of patients with locoregional pancreatic cancer do not undergo potentially curative surgery, primarily because of nonclinical factors, according to the results of an analysis of data from the National Cancer Institute database by Amy T. Cunningham, MPH, a doctoral candidate at Thomas Jefferson University in Philadelphia, and colleagues.
Of 16,676 patients eligible for surgery, only 8524 (51.1%) had pancreatic resections. The likelihood that a patient would have surgery decreased with increasing age, female sex, nonwhite race or ethnicity, and unmarried status, the researchers suggested in a presentation at the 2016 International Society for Pharmacoeconomics and Outcomes Research annual meeting.
The most common (79.40%) reason for not undergoing resection was a lack of recommendation for surgery.
“Nearly half of patients with locoregional pancreatic cancer do not receive surgery; the reason why is often unclear. Understanding and addressing these disparities could increase pancreatic cancer resection rates and improve survival,” Ms Cunningham and colleagues concluded.
The findings came from a review of the Surveillance, Epidemiology, and End Results (SEER) Program database from 2004 to 2012. The investigators queried the system to identify all patients with newly diagnosed locoregional stage I or II pancreatic cancer involving the head, body, or tail of the pancreas. The tumor stage was determined based on the American Joint Committee on Cancer criteria.
Patients were excluded if they were identified by death certificate or autopsy, or if they had incomplete survival or surgical data. The primary outcome was the proportion of eligible patients undergoing pancreatic cancer resection.
The initial search of the database identified 197,864 patients with a primary diagnosis of pancreatic cancer. After applying the exclusion criteria, the total study population was 16,676 patients.
The primary analysis showed that 8152 (48.9%) of eligible patients did not undergo pancreatic cancer resection. When comparing patients who had surgery with those who did not, the investigators found that 48% of patients aged <65 years underwent resection. This number decreased to 31.1% for patients aged 65 to 74 years and to 21% for those aged ≥75 years (P <.001). Men underwent surgery more often than women (52.7% vs 49.6%, respectively; P <.001).
An analysis of surgery status by race or ethnicity showed that 52.2% of white or non-Hispanic patients had resection versus 45.9% of black patients and 48.9% of Hispanic patients (P <.001). Married patients had a substantially higher likelihood of having surgical resection than did unmarried patients (56.2% vs 44.5%, respectively; P <.001). The rate of surgical resection also varied significantly in the SEER registry.
Patients with stage I tumors had a significantly lower rate of resection than patients with stage II tumors (39.7% vs 54.6%, respectively; P <.001). Tumor location in the tail of the pancreas was associated with a higher rate of surgical resection (74.3%) than tumors located in the head (49.8%) or body (41.0%) of the pancreas.
SEER records for the 8152 patients who did not undergo surgical resection showed that surgery was not recommended in 79.4% of cases. An additional 10% of patients had contraindications to surgery because of other conditions, and in 4.23% of cases the patient or guardian refused surgery.
By multivariate analysis, the factors that were significantly associated with not undergoing surgery were older age (odds ratio [OR], 0.70 for patients 65-74 years; OR, 0.28 for patients ≥75 years), male sex (OR, 0.91), nonwhite race or ethnicity (OR, 0.69-0.86), and unmarried status (OR, 0.71).
The researchers noted several topics where additional research is needed, including analysis of regional care patterns and the impact of high-volume hospitals on pancreatic resection rates and interviews with physicians, patients, and caregivers to learn more about the decision-making process associated with resection.
The investigators acknowledged several limitations of the study, including inherent limitations of observational data, no knowledge about resection-associated variables that were not captured by the SEER database, and lack of details about the resection decision-making process.
