S. Jeong1, M. Aldakkak1, K. Ahn2, C. Huang3, K. K. Christians1, B. A. Erickson4, P. S. Ritch5, B. George5, D. B. Evans1, S. Tsai1 4Medical College Of Wisconsin,Radiation Oncology,Milwaukee, WI, USA 5Medical College Of Wisconsin,Hematology Oncology/Dept Of Medicine,Milwaukee, WI, USA 1Medical College Of Wisconsin,Surgery,Milwaukee, WI, USA 2Medical College Of Wisconsin,Biostatistics,Milwaukee, WI, USA 3University Of Wisconsin, Milwaukee,Biostatistics/Joseph J. Zilber School Of Public Health,Milwaukee, WI, USA
Introduction: American Joint Committee on Cancer (AJCC) TNM staging system provides population based estimates of survival based on pathologic variables. For patients with pancreatic cancer (PC), survival estimates were generated from patients who have undergone surgery with or without postoperative therapy. Increasingly, preoperative (neoadjuvant) therapy followed by surgery is being utilized in patients with PC in recognition of the high risk of disease recurrence and the inability to consistently deliver adjuvant therapy after pancreatectomy. Whether the AJCC staging system accurately predicts survival among patients who have received neoadjuvant therapy is unclear. We assessed the survival discrimination of the AJCC staging system for patients with PC who have received neoadjuvant therapy and surgery, and developed a novel prognostic nomogram using clinical variables.
Methods: Clinical data and survival outcomes from patients with PC who completed neoadjuvant therapy and surgery at a single institution were collected. Survival at 1-, 2-, and 3-years from the date of restaging after neoadjuvant therapy and surgery were used for the purpose of nomogram construction. Concordance index (c-index) and calibration plots were used to assess predictive accuracy. The nomogram was developed using multivariable Cox proportional hazards model. Clinical stage was defined at the time of diagnosis and patients were categorized as having resectable or borderline resectable disease. Postoperative carbohydrate antigen 19-9 (CA19-9) was measured at the first restaging visit after surgery.
Results: Data was available for 167 patients with resectable and borderline resectable PC. The concordance indices (c-index) for 1-, 2-, and 3- years using the AJCC staging system were 0.57, 0.55, and 0.56, respectively. Clinical stage (HR:2.31; 95%CI:1.48-3.63) and postoperative CA19-9 levels (HR: 2.14; 95%CI:1.38-3.34) were the strongest prognostic factors. A parsimonious nomogram (Figure 1) including clinical stage, postoperative CA19-9, and age predicted 1-, 2-, and 3- year survival with c-indices of 0.66, 0.66, and 0.66, respectively. Calibration plots showed good fitness between observed and predicted probabilities. A combined nomogram using clinical stage, postoperative CA19-9, age, and AJCC stage demonstrated c-indices for 1-, 2-, and 3-years of 0.68, 0.67, and 0.67, respectively.
Conclusion: AJCC staging system poorly discriminates survival for patients who have received neoadjuvant therapy and surgery. A prognostic nomogram utilizing clinical stage and postoperative CA19-9 levels provides more accurate survival prediction than the AJCC model. External validation will be performed to assess the generalizability of the nomogram.
