Importance: Neoadjuvant therapy is increasingly used in localized pancreatic carcinoma, and survival is correlated with carbohydrate antigen 19-9 (CA19-9) levels and histopathologic response following neoadjuvant therapy. With several regimens now available, the choice of chemotherapy could be best dictated by response to neoadjuvant therapy (as measured by CA19-9 levels and/or pathologic response), a strategy defined herein as adaptive dynamic therapy.
Objective: To evaluate the association of adaptive dynamic therapy with oncologic outcomes in patients with surgically resected pancreatic cancer.
Design, setting, and participants: This retrospective cohort study included patients with localized pancreatic cancer who were treated with either gemcitabine/nab-paclitaxel or fluorouracil, leucovorin, irinotecan, and oxaliplatin (FOLFIRINOX) preoperatively between 2010 and 2019 at a high-volume tertiary care academic center. Participants were identified from a prospectively maintained database and had a median follow-up of 49 months. Data were analyzed from October 17 to November 24, 2020.
Exposures: The adaptive dynamic therapy group included 219 patients who remained on or switched to an alternative regimen as dictated by CA19-9 response and for whom the adjuvant regimen was selected based on CA19-9 and/or pathologic response. The nonadaptive dynamic therapy group included 103 patients who had their chemotherapeutic regimen selected independent of CA19-9 and/or tumoral response.
Main outcomes and measures: Prognostic implications of dynamic perioperative therapy assessed through Kaplan-Meier analysis, Cox regression, and inverse probability weighted estimators.
Results: A total of 322 consecutive patients (mean [SD] age, 65.1 [9] years; 162 [50%] women) were identified. The adaptive dynamic therapy group, compared with the nonadaptive dynamic therapy group, had a more pronounced median (IQR) decrease in CA19-9 levels (-80% [-92% to -56%] vs -45% [-81% to -13%]; P < .001), higher incidence of complete or near-complete tumoral response (25 [12%] vs 2 [2%]; P = .007), and lower median (IQR) number of lymph node metastasis (1 [0 to 4] vs 2 [0 to 4]; P = .046). Overall survival was significantly improved in the dynamic group compared with the nondynamic group (38.7 months [95% CI, 34.0 to 46.7 months] vs 26.5 months [95% CI, 23.5 to 32.9 months]; P = .03), and on adjusted analysis, dynamic therapy was independently associated with improved survival (hazard ratio, 0.73; 95% CI, 0.53 to 0.99; P = .04). On inverse probability weighted analysis of 320 matched patients, the average treatment effect of dynamic therapy was to increase overall survival by 11.1 months (95% CI, 1.5 to 20.7 months; P = .02).
Conclusions and relevance: In this cohort study that sought to evaluate the role of adaptive dynamic therapy in localized pancreatic cancer, selecting a chemotherapeutic regimen based on response to preoperative therapy was associated with improved survival. These findings support an individualized and in vivo assessment of response to perioperative therapy in pancreatic cancer.