Molecular classification as prognostic factor and guide for treatment decision of pancreatic cancer

Abstract

Clinico-pathological factors fail to consistently predict the outcome after pancreatic resection for pancreatic ductal adenocarcinoma (PDAC). PDACs show a high level of inter- and intra- tumor genetic heterogeneity. A molecular classification should help sort patients into less heterogeneous and more appropriate groups regarding the metastatic risk and the therapeutic response, with the consequences of better predicting evolution and better orienting the treatment. PDAC can be classified based on mutational subtypes and 18gene alterations. Whole-genome sequencing identified mutational signatures, mutational burden and hyper-mutated tumors with specific DNA repair defects. Their overlap/similarities allow the definition of molecular subtypes. DNA and RNA classifications can be used in prognosis assessment. They are useful in therapeutic choice for they allow the design of approaches that can predict the respective drug sensitivity of each molecular subtype. This review provides a comprehensive analysis of available molecular classifications in PDAC and how this can help guide clinical decisions.

Introduction

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive human cancers [1] with a 5-year survival inferior to 10% and an increasing incidence [2]. Complete surgical removal of the tumor followed by adjuvant chemotherapy is the only curative treatment. However, <20% of patients are candidate to surgery. Recently, adjuvant combination of gemcitabine and capecitabine following pancreatic resection for PDAC (ESPAC-4) showed a median overall survival of 28 months (95% CI 23·5–31·5) for patients who received gemcitabine plus capecitabine compared to 25·5 months (22·7–27·9) for gemcitabine alone [3]. Patients with borderline resectable or locally-advanced tumor are often treated with neoadjuvant chemotherapy (FOLFIRINOX) to obtain a margin-negative resection with a disease-free survival and overall survival better than upfront resected patients (DFS: 29.1 vs 13.7, P < 0.001; OS: 37.7 vs 25.1 months from diagnosis, P = 0.01) [4].

Only few chemotherapy agents (gemcitabine with or without nab-paclitaxel, FOLFIRINOX regimen combining 5-FU, leucovorin, oxaliplatin and irinotecan) and one targeted therapy (erlotinib, a tyrosine kinase inhibitor, notably for the epidermal growth factor receptor (EGFR)) show some degree of efficacy in unselected patients but with a limited clinical impact. Only 31.6%, 23% and <10% of patients respond to FOLFIRINOX, gemcitabine/nab-paclitaxel and gemcitabine, respectively [5,6]. Thus, it is crucial to both develop novel drugs and specifically identify patients most likely to benefit from one or another treatment.

Currently, the AJCC TNM staging is the only prognostic factor used in clinical practice to assess the survival of a resected PDAC and guide treatment decision. However, this clinicopathological factor fails to consistently predict the outcome after pancreatic resection. Characterization of molecular alterations may help develop techniques for early detection, identify new molecular targets and design novel targeted therapies. In breast and lung cancers, targeting drugs to tumor molecular subtypes improves treatment response and outcome [7,8]. Like these tumors, PDAC is a heterogeneous molecular disease. To solve this inter-tumor heterogeneity, high-throughput molecular studies have been applied to PDAC samples. They revealed molecular subtypes associated with different genetic alterations and prognosis, as well as gene expression signatures associated with survival [[9], [10], [11], [12]].