- Torphy, Robert J;
- Wang, Zhen;
- True-Yasaki, Aisha;
- Volmar, Keith E;
- Rashid, Naim;
- Yeh, Benjamin;
- Anderson, Judy M;
- Johansen, Julia S;
- Hollingsworth, Michael A;
- Yeh, Jen Jen;
- Collisson, Eric A
Desmoplastic stroma is a cardinal feature of primary pancreatic ductal adenocarcinoma (PDAC), but its effects on the biology, prognosis and therapeutic outcomes are not known. We developed an automated method to assess tumor stroma density (TSD) and investigated computed tomography (CT)-correlates of stroma in PDAC. We collected PDAC samples from rapid autopsy and resection series and digitally annotated samples to quantify TSD. A series of resected patients also underwent preoperative multiphasic CT. Automated and manual assessments of TSD were highly correlated (ρ= 0.65, P < 0.001). Solid organ metastases had a lower median TSD than primary tumors (P < 0.001). Patients with high TSD enjoyed prolonged recurrence free survival (RFS) (P = 0.003; HR = 0.51) and overall survival (P = 0.008, HR = 0.57). In another independent dataset, patients with high TSD had decreased risk for recurrence (P = 0.003, HR = 0.03) and death (P = 0.003, HR = 0.03) at time of resection, however the protective effect diminished over time. Patients with normalized portovenous phase CT tumor enhancement ratio ≥0.40 had a longer RFS following resection (P = 0.020). Normalized portovenous phase CT tumor enhancement ratio was significantly correlated with TSD (P = 0.003). Objective quantitative assessment of stroma in PDAC revealed several clinically relevant observations. Firstly, stroma was less abundant in metastatic PDAC, the cause of most PDAC mortality. Secondly, high stromal content correlates with favorable outcome in resected cases, implying a protective effect of stroma and suggesting careful consideration of active stromal depletion therapies. Finally, standard multiphase CT imaging correlates with stroma content as well as clinical outcome, indicating that non-invasive assessment of stroma is a feasible sensitivity enrichment approach in PDAC.