Database
We used the SEER database to analyze the data of patients with pancreatic ductal adenocarcinoma (PDAC), the most common type of pancreatic cancer [11]. The database includes patients of 18 registries in the USA from 1973–2013, encompassing approximately 28% of the USA population. All the malignant cases were followed-up annually to determine vital status.
Patient Population
All the variables’ definitions are encoded in the SEER database. To identify the PDAC cases, site codes (C25 pancreas, C25.0-C25.9) and histology codes (8140 adenocarcinoma, 8500 infiltrating duct carcinoma) based on the International Classification of Diseases for Oncology, Third Edition (ICD-O-3) were used [12]. Only cases that underwent PD and microscopically confirmed were included.
Outcome Variables
We only included those PDAC patients that underwent PD with precise data available for the following variables: age at diagnosis, year of diagnosis, gender, race, tumor site, tumor size, regional nodes positive, regional nodes examined, grade, stage, vital status, and survival months.
In order to compare the long-term survival between young and elderly patients, all cases were divided into two group age < 80 years and age ≥ 80 years. Since 2004, the AJCC 6th stage has been used in the SEER database. Thus, the diagnosis years of the cases included in our study ranged from 2004–2013. To analyze the median overall survival between different time periods, we divided the year of diagnosis into two groups: 2004–2008 year and 2009–2013. Furthermore, we also divided the tumor size into three groups: ≤ 2 cm, 2–4 cm, and > 4 cm. The lymph node ratio (LNR) is considered a robust prognostic factor after resection of pancreatic cancer and was estimated using regional nodes positive divided by regional nodes examined [13]. LNR was then categorized into three groups: 0%, 1–50%, > 50%.
Data Analysis And Statistics
The patients’ clinical and pathology characteristics were summarized with descriptive statistics. Normally distributed continuous variables were presented as mean ± standard deviation and Student’s t-test was used to evaluate these variables. In the case of non-normal distribution, the continuous variables were expressed as median (interquartile range, IQR) and analyzed using Mann–Whitney U test. The categorical and ordinal variables were presented as frequencies and proportions. Chi-square or Mann–Whitney U tests were used independently for categorical or ordinal variables, respectively, to investigate the differences between the two groups. Survival was analyzed using the life-table curve, and Gehan-Wilcoxon test was used to compare these curves. Univariate and multivariate Cox proportional hazards model (enter method) was used to identify the independent factors associated with prognosis in elderly patients. All statistical analyses were performed using SPSS version 19.0 (IBM Corporation. Armonk, NY, USA). P-value ≤ 0.05 was considered statistically significant.