The correlation between clinical parameters and RT in pancreatic cancer patients
From the year 2004 until the end of the year 2015, 164035 patients with pancreatic cancer were enrolled in SEER database. In total, there were 2089 stage I patients and 12888 stage II patients underwent surgery operation were enrolled in this study. The median age at diagnosis was 67 years old (age ranging from 18 to 97). Among stage I patients, 88 patients received neoadjuvant RT and 388 patients received adjuvant RT. The proportion of patients receiving radiotherapy differed in year of diagnosis, sex, race, primary site, pathological grade, T stage and whether received chemotherapy.
544 patients received neoadjuvant RT and 4200 patients received adjuvant RT in stage II patients. The composition ratio of radiation therapy in stage II patients differed in age, year of diagnosis, race, primary site, pathological grade, T stage, N stage, whether received chemotherapy and surgical methods (P < 0.05). The characteristics of all patients are presented in Table 1.
Overall survival analysis of Stage I/II pancreatic cancer patients
14977 patients with pancreatic cancer were divided into three groups according to whether they received RT or not. Kaplan-Meier (KM) analysis showed that the median survival time of patients without RT was 20 months, neoadjuvant RT was 24 months and adjuvant RT was 23 months (p < 0.0001, Fig. 2A). We performed subgroup analysis according to different stages.
2089 cases of stage I pancreatic cancer patients were divided into three groups by radiotherapy. Comparison of the median survival time differences was conducted in the three groups. Median survival time of patients without RT was 56 months, significantly longer than the median survival time of those with RT regardless of neoadjuvant or adjuvant RT (no RT: 56 months vs adjuvant RT: 37 months vs neoadjuvant RT: 27 months, P = 0.0039, Fig. 2B). Similarly, all 12888 patients with stage II pancreatic cancer were divided into three groups according to whether they received radiotherapy or not. Median survival time of patients with neoadjuvant RT was 24 months, adjuvant RT 22 months, significantly prolonged than the median survival time of those without radiotherapy(neoadjuvant RT: 24 months vs adjuvant RT: 22 months vs no RT: 17 months, P༜0.0001, Fig. 2C).
Univariate analysis of clinical features affecting the prognosis
Univariate survival analysis was separately performed for stage I/II patients. For stage I patients, the unadjusted protective factor was tumor located in tail of pancreas. Radiotherapy including age (≥ 60 years old), male, pathology with worse differentiation, higher T stage were adverse diagnostic factors.
For patients with stage II pancreatic cancer, radiotherapy, chemotherapy and newly diagnosed pancreatic cancer after 2010 were the advantageous factors for the prognosis of patients. The harmful factors for the prognosis included age (≥ 60 years old), male, pathology with worse differentiation, higher T and higher N stage. The results are shown in Table 2.
Multivariate analysis of clinical features affecting the prognosis
Multivariate survival analysis of stage I pancreatic cancer showed that neoadjuvant RT (HR = 1.434, P = 0.023, 95% CI: 1.051–1.957), age (≥ 60 years old), male, pathology with worse differentiation, higher T stage were independent risk factors for prognosis, indicating a shorter survival period. Chemotherapy, tumor located in body and tail of pancreas, non-white/black Americans were favorable prognostic factors and related to longer survival period.
Also, we conducted multivariate survival analysis for stage II pancreatic cancer patients. Results showed that adjuvant RT (HR = 0.904, P < 0.001, 95% CI: 0.861–0.950) including chemotherapy, expanded surgical procedure, pancreatic duct tumors, diagnosed pancreatic cancer after 2010 and non-black Americans predicted better outcomes. On the contrary, age (≥ 60 years old), male, pathology with worse differentiation, higher T and N stage predicted worse outcomes, which means shorter survival time. Multivariate analysis results are shown in Table 3.
Competitive risk analysis for cancer-related death of I/II pancreatic cancer patients
The causes of death were divided into cancer-related death and non-cancer-related death. We analyzed the death outcomes by competitive risk model. The results showed that the 1-, 3-, and 5-year cancer-related mortality rate in stage I patients treated with neoadjuvant RT were 13.98%, 52.33% and 63.58% respectively. In adjuvant RT group, the 1-year, 3-year and 5-year cancer-related mortality rates were 10.95%, 43.16% and 48.94%, nevertheless these rates decreased to 11.54%, 32.02% and 39.51% respectively in patients without radiotherapy. The risk of cancer-related death caused by neoadjuvant RT was significantly higher than adjuvant RT and no-RT (p < 0.001; Fig. 3A) .
The 1-year, 3-year and 5-year cancer-related mortality rates in patients with stage II pancreatic cancer without radiotherapy were 32.70%, 66.78% and 74.16% respectively, In patients who received adjuvant RT, the 1-year, 3-year and 5-year cancer-related mortality rates were 21.16%, 63.47% and 73.38%. The rates were 17.57%, 55.30% and 68.49% in neoadjuvant RT group. The results showed the cancer-related death of II stage patients without radiotherapy was significantly higher than that of patients received radiotherapy(P < 0.001; Fig. 3B).