Development and Validation of a Prognosis Prediction Model for Re-Irradiation Indicated for Highly Selected Patients with Pancreatic Cancer

Abstract

Despite advances of management of pancreatic cancer, disease recurrence is still the predominant cause of death. A significant number of patients may develop local recurrence within the primary regions. However, these patients may not be amenable for surgery or chemotherapy due to its poor response rates and toxicities. Therefore, radiotherapy may be appropriate in the setting of re-irradiation of pancreatic cancer. Stereotactic body radiation therapy (SBRT) has previously been taken as the option for re-irradiation of pancreatic cancer. Therefore, the aim of the study was to develop a clinical point score for selecting patients who are candidates for re-irradiation. The training cohort included re-irradiation group (25 patients) and control group (50 patients) from 2012 to 2016, who all had biopsy-proven pancreatic cancer. Only patients receiving two courses of SBRT due to recurrence at the same irradiated region (the primary lesion) were included. Additionally, the progression free survival of eligible patients in the control group were more than 12 months. The validation group included 260 patients with biopsy-proven pancreatic cancer undergoing SBRT between 2015 and 2016. CA19-9 response was defined as a CA19-9 decrease by 50% from baseline levels of ≥74 U/ml. Hence, three groups were formed for analysis: CA19-9 levels ≥74 U/ml with response vs. CA19-9 levels ≥74 U/ml with no response (including CA19-9 levels within the normal range before SBRT while increased after SBRT) vs. CA19-9 levels <74 U/ml (before SBRT and during follow-up). Baseline factors of re-irradiation group and control group were assessed by the log-rank test. Thereafter, parameters before the second SBRT of re-irradiation group and those determined at the initial progression after SBRT or at the last follow-up without progressions of control group were also compared by the log-rank test. Factors with a P-value <0.05 in the multivariate analysis were chosen to form the scoring system. The regression coefficients of the factors were multiplied by 2 and rounded in order to be easily applied in clinical practice. No differences of baseline characteristics were found between training and control group. After multivariate analysis, baseline ECOG (P=0.013), history of previous treatment (P=0.003), CA19-9 response (P=0.021) and tumor response (P<0.001) were included to form the prediction model: CAPER score (named after acronyms of enrolled factors). In the training cohort, patients with the score of 0-7 points had a better 2-year overall survival (OS) rate (P<0.001). More patients with the score of 0-7 points received the re-irradiation (P<0.001). In the validation cohort, the median OS was longer in patients with the score of 0-7 points than those with the score >7 points (11.2 months vs. 6.6 months, P<0.001). Patients with a CAPER-score of 0-7 points have a better prognosis, indicating that re-irradiation with SBRT is worth considering and may provide survival benefits.