Systems and Network Pharmacology Strategies for Pancreatic Ductal Adenocarcinoma Therapy : A Resource Review

Introduction

According to the World Cancer Research Fund (WCRF) there were 279,000 cases of pancreatic cancer diagnosed worldwide in 2008 ( http://www.wcrf.org/cancer_statistics/ ). The estimated five-year prevalence of people living with pancreatic cancer is projected at 3.5% per 100,000 and it is the 13th most common cancer in the world. Pancreatic cancer is almost always fatal and is the eighth leading cause of cancer-related deaths in the world. The most common form of pancreatic cancer is the pancreatic ductal adenocarcinoma (PDAC). In the United States it is the fourth leading cause of cancer-related deaths . PDAC kills ∼39,000 patients each year , which translates to two American deaths every 30 min. The disease is presented at a very late stage and the symptoms are very complex, oftentimes overlapping with that of other disorders, leading to misdiagnoses. Aggressive and therapy resistant PDAC is refractory to any of the currently available treatment modalities. Gemcitabine and its combination with platinum-based compounds have minimal impact and improve the survival by a mere few weeks. Recently, gemcitabine-nab-paclitaxel (Abraxane) has been investigated, and that too has shown very nominal benefits. These morbid statistics and lack of effective drugs indicate that modern approaches to understanding the disease as well as novel molecularly targeted therapies are urgently needed in order to advance the current state of PDAC therapy.

The genetics of therapy resistant PDAC is highly complex . The refractory tumors harbor multiple aberrations in oncogenic and tumor suppressor signaling pathways . PDAC signaling networks are highly intertwined and robust, meaning that they resist changes such as that induced by targeted therapies. According to the most accepted PDAC progression model, normal duct epithelium progresses to infiltrating cancer through a series of histologically defined precursors (PanINs). The overexpression of HER-2/ neu and point mutations in the K- ras gene occur early, inactivation of the p16 gene at an intermediate stage, while the inactivation of p53 , DPC4 , and BRCA2 occur relatively late . It has previously been proposed that PDAC robustness stems from “passing of the baton” between genes responsible for driving the various stages of the tumor, and therefore an effective therapy should likely target an entire set of genes across a succession of stages to break such robustness . It is interesting to note that to date the moderately successful agents against PDAC have been the nucleoside analogs such as gemcitabine, 5FU, platinum drugs such as cisplatin or oxaliplatin, and recently nab-paclitaxel. While nucleoside analogs and platinum are considered DNA intercalators and nab-paclitaxel was originally discovered as an antimicrotubule agent , studies have shown that most of these agents have multiple network pharmacology type of effects on cancer cells . For example, the nucleoside analog gemcitabine has been shown to interact with a wide range of proteins such as P8 , and oxaliplatin (a component of FULFIRNOX) was shown to form 10 times more protein adducts than its originally designated DNA adduct forming anticancer mechanisms . Similarly the permutations and combinations of targets of individual components in FULFIRINOX have not been fully elucidated. Although stagnated at six to eight months overall survival benefit for the past 40 years, the multitargeted therapeutics mentioned previously have remained the most widely used agents for the treatment of PDAC. There is no doubt that these chemotherapeutic drugs work through network pharmacology principles. In this chapter, we present the case for using systems and network biology sciences to better understand PDAC complexity and provide examples of how network pharmacology strategies can be used to design superior network targeted single agent and combination strategies against PDAC leading to better treatment outcome.

Need for Revisiting the Progression Model of PDAC: Departure from Genes to Network

There is general consensus over the heterogenous nature of PDAC. Advances in genomic assessment tools have highlighted the intricate signaling network complexity that emanates from the interactions between the different components in PDAC tumor microenvironment. Differential gene expression (DE) analysis has been the traditionally adopted model to identify driver genes. This is the criteria used in the Hruban’s progression model where a set of driver gene mutations have been attributed to each specific stage in PDAC development. While these analyses manage to capture several major genes that show noticeable changes in expression/mutation, there are many more important genes that often do not display such drastic changes (do not fall under the DE criteria or cutoff values). These genes are not identifiable through their own behavior, rather, their changes are only quantifiable when evaluated in conjunction with other genes within their vicinity (i.e., through their role in the networks) . However, traditional molecular biology cannot sieve through this complexity, and systems and network level investigations that take a holistic view are needed . In this direction, Srihari and colleagues have utilized PDAC expression datasets comparing 39 paired normal vs. tumor samples to track the progression based on protein–protein (PPI) and gene interactions. Their analysis utilized a novel algorithm, MIN FLIP (FLIP are genes that are flipped in response to stage transition or perturbations). Their analysis shows that serine/threonine kinase are the major genes that act as ON/OFF switches regulating cell cycle progression during the PDAC differentiation process . The MIN FLIP resulted in the discovery of genes that were not marked in the traditional DE system.

Logically, the approaches that could break down the PDAC signaling complexity into smaller easily decipherable fragments should make rational design of drugs or their combination easier. There are a number of different network visualizing tools available that can be used to evaluate PDAC expression datasets to obtain sub network information both for diagnostic and therapeutic purposes. With the advent of high-throughput analyses systems, diseases such as PDAC are routinely investigated at multiple levels such as genomics, transcriptomics, and interactomics (metabolomics) levels. These large-scale analyses systems can further be benefited by applying network interaction visualization tools that are readily and publicly available. Table 18.1 lists some of the major network visualization tools that are freely available to evaluate expression datasets such as that derived from PDAC cell line, animal models, human primary tumors, and patient samples. These network visualization tools give a fairly good amount of insight into the interactions between the defined set of differentially identified genes (for example, differentially expressed genes) within a given expression dataset.

Defining Biological Networks

The selection of important network positions as drug targets faces major hurdles. An ideal target in cancer network has to be important enough to influence the disease, however, such network position must not be so critical for normal cell physiology that targeting it leads to outward toxicity. The answer to this problem can come from detailed knowledge on the structure and dynamics of complex cancer related networks that are presented in the following discussion. A biological network is composed of nodes and edges . Nodes can be either amino acids, genes, microRNAs, or proteins, and edges are the interactions between two nodes. Depending on the threshold of detection limits, the edges can connect more than two nodes. Even though these simplistic definitions hold true, there are exceptions, such as in case the node is a protein that is secretory in nature, and then definitions of nodes become diffused. There are a number of excellent reviews that detail the network identification methods and dilemmas associated with defining certain molecular networks . This chapter will not rereview the existing knowledge but will instead focus on how these existing resources will be applied for PDAC biomarker identification and drug discovery.