Abstract 1642P
Background
Pancreatic ductal adenocarcinoma (PDAC) is marked by a desmoplastic stroma and poor prognosis. Current pathological or molecular classifications are discordant and not applied in routine clinical practice. In order to predict relapse-free survival (RFS), we aimed to develop an immunohistochemical (IHC) signature involving immune, fibroblastic, tumor markers in a cohort of 217 resected PDAC patients.
Methods
Tumors were collected for tissue microarrays construction and characterized by digital pathology with 13 IHC stains at the tumor epithelial, immune and stromal compartments. Cox proportional hazard models were performed to estimate hazard ratio and CI95% for factors associated with RFS.
Results
Multivariate Cox analysis showed that loss of membrane localization of β-catenin on tumor cells (HR=1.54, 95%CI=1.08-2.20, p=0.02), a low density of tumor infiltrating CD8+ T-cells (HR=1.72, 95%CI =1.23-2.41, p<0.01) and a high percentage of HSP47-positive Cancer-Associated Fibroblasts (CAF) (HR=1.41, 95%CI =1.03-1.96, p=0.03) were independent predictors of reduced RFS. Based on the expression of these three markers, we constructed a scoring system to classify resected PDAC into three distinct prognostic groups (p<0.001): 0/1+ (mRFS=8.4, 95%CI=6.7-9.9), 2+ (mRFS=13.1, 95%CI=10.7-16.4) and 3+ (mRFS=22.5, 95%CI=17.1-131.7). The discriminative ability of the three-group model was confirmed in overall survival analysis. Furthermore, with the exception of neoadjuvant therapy (p=0.03) and the presence of lymphovascular invasions (p=0.04), the score developed was not related to other clinicopathological data.
Conclusions
We propose a novel RFS prediction model based on three independent prognostic IHC markers evaluating tumor epithelial, immune and fibroblastic compartments. These results highlight considerable heterogeneity in the survival of patients with resected PDAC. Moreover, this is the first demonstration of the prognostic character of HSP47 by IHC in PDAC.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.
Resources from the same session
1569P - The CODRP model for predicting drug sensitivity in patient-derived 3D gastric cancer cells
Presenter: Dong Woo Lee
Session: Poster session 22
1571P - Exploration of immune and metabolism gene signature for prognosis of esophageal carcinoma and establishment of a combined prediction model
Presenter: Hao Wu
Session: Poster session 22
1572P - Impact of HER2 and PD-L1 co-expression in Claudin18.2 positive resectable gastroesophageal cancers
Presenter: Antonella Cammarota
Session: Poster session 22
1573P - Involved field and elective nodal irradiation presented similar treatment efficiency in concurrent chemoradiation for locally advanced ESCC
Presenter: Baosheng Li
Session: Poster session 22
1575P - Factors associated with uptake of adjuvant nivolumab in a nationwide esophageal cancer patient cohort
Presenter: Rob Verhoeven
Session: Poster session 22
1577P - Prior antibiotic administration disrupts outcomes of PD-1 blockade in advanced gastric cancer by altering gut microbiome and systemic immune response
Presenter: Chang Gon Kim
Session: Poster session 22
1578P - Effect of immune checkpoint inhibitors in metastatic gastric cancer: A real-world evidence study
Presenter: Francesco Puccetti
Session: Poster session 22