Systems Biology of Pancreatic Cancer Stem Cells

Introduction

Pancreatic cancer (PC) kills ∼300,000 individuals worldwide, and each year in the United States there are an estimated 43,920 new diagnoses with an annual mortality of 37,390 . It is one of the most difficult cancers to treat, with a five-year survival (<5%) due in part to the high degree of treatment resistance leading to failure of most of the available therapies. The etiology of the disease is mostly unknown. It is suggested that PC is correlated with tobacco use, diabetes, obesity, and chronic pancreatitis . There also is a genetic basis to acquiring the disease, since 5–10% of patients have a family history of PC. A major stumbling block is that PC is diagnosed at a very late stage of the disease where invasion or metastatic spreads (both micro- and macro-metastases) have already occurred, contributing to a poor prognosis. In spite of the increased understanding of the mechanisms of PC development, there is no standardized diagnostic methodology for its early detection. Many of the symptoms initially described by patients go unnoticed by physicians due to their nonspecific and variable nature, resulting in misdiagnoses. This delay in identification after the onset of some symptoms is crucial for the future prospects of PC patients. Sadly, at the time of actual diagnosis, nearly half of the patients exhibit metastasis. Typically >80% have advanced or locally advanced disease, and less than 15% of patients have organ-confined disease at the original PC site. This is partly due to the long duration of time between disease onset and diagnosis. Frequently, PC can be identified during unrelated CT scans of patients, and several studies have attempted to derive structural similarities that may alert radiologists to a malignant presence before diagnosis. Some patients were identified as having PC retrospectively about 18 months prior to diagnosis while asymptomatic. The chance observations call for urgent innovations in the areas of early diagnosis and novel development of cytotoxic agents, as well as agents for “personalized medicine”. Emerging advances would likely help in early diagnosis, for example, identification of circulating tumor cells related to PC metastasis and epithelial–mesenchymal transition (EMT), as well as identification of the PC stem-like cells (CSLCs or CSCs) that are increasingly being recognized to play a role in sustaining the heterogeneous and resistant nature of the disease.

The Complexity of Pancreatic Cancer

Like many other cancers, PC can be defined in terms of accumulating genetic mutations in key tumor activator and inhibitor genes such as K-ras and a myriad other regulatory molecules . Analyses of upstream regulators in PC progression models have identified members of the sonic hedgehog signaling pathway (SHH and Smo) as being aberrantly expressed in the precursor PanIN lesions, which highlights the role and degree of K-ras and HER2/neu mutations and their significance in the initial appearance of PanIN lesions . The significance of the SHH pathway in PC was emphasized when it was demonstrated that inhibition of this pathway strengthened the efficacy of chemotherapy in vivo . Related to K-ras is the delicate balance maintained through TGF-β family molecules with dual functions, both as oncogenic and as a tumor suppressor. Smad7 was found to inhibit the function of TGFβRI from phosphorylating Smad2 or Smad3, resulting in an increased tumorigenic potential in PC by preventing nuclear localization of Smad4 . Disruption of Smad4 function has been realized as an indicator of decreased survival, yet some ambiguity of results has posed problems in accurately ascertaining the role of Smad4 dependency or TGFβ resistance in PC cells . Moreover, studies have shown that inhibition of hedgehog signaling had a measurable effect on the EMT that is thought to be crucial for PC invasiveness and metastasis, mediated by downstream inhibition of Snail and Slug transcription factors (the role of these important signaling molecules will be further clarified when we present the PC stem cell work following). Hedgehog has also been implicated in the development of stem cells, which are normally part of the gastrointestinal (GI) tissue development . Generally, progression of PC involves mutations or abnormal expression of Pdx1, hedgehog signaling, K-ras, p16, p53, DPC4, and BRCA2, as well as their interacting molecules, resulting in carcinoma . Figure 13.1 is the Ingenuity Pathway Analysis (IPA)–derived pathway depicting the involvement of various signaling pathways during the different stages of PC development. In addition to frequently cited mutations and aberrant expression of the DNA sequence, epigenetic factors have been implicated whether they are methylations or dysregulations of microRNAs (miRNAs), notably a detectable hypermethylation in developing PanINs .

Even with extensive knowledge gained over the vast array of molecular mechanisms involved in PC, early stage identification remains problematic and at times incidental. Under ideal circumstances, the majority of patients do not possess those traits. The resectability of a PC tumor represents the initial line of defense for the treatment of PC. Detection at early stages is key, yet even with resection under ideal circumstances, the five-year survival postresection holds at 20% . Mortality postresection has been a point of debate with different attempts being made to measure markers of success or failure . Problems in risk stratification and availability of the surgery have contributed to many patients not being offered this course of treatment, and despite resection, adjuvant therapies such as 5-FU, FOLFIRINOX, and gemcitabine are still the standard line of care for both unresectable and resectable tumors . There is high desirability of testing for PC presence and treatment progress through serum biomarkers, yet one such strategy, the testing for biomarker CA19-9, a molecule frequently observed in PC patients, has shown low sensitivity and specificity . Other molecules such as HCGβ and CA-72-5 have shown higher sensitivity than CA19-9 in the laboratory setting; however, these and other markers continue to be under investigation for further clinical efficacy as efficient biomarkers in PC . Interestingly, some groups have used gemcitabine as a filter for resistance or sensitivity to treatments, and at least one study was able to correlate treatment success by testing for the hENT1, a molecule that regulated gemcitabine uptake . Using such a strategy, a more personalized approach was used in determining future treatment choice. The early models for studies have been aimed at precisely identifying the genetic basis of occurrence, especially of familial PC, by establishing a database to pinpoint what classifies the at-risk individuals . A more generalized approach has been the use of traditional family medical histories, including diagnostic data, compared with data-based repositories to identify at-risk individuals through factors also outside genetics, such as risky behavior or environmental determinants . Such a model is not new; however, it hedges a bet on the development of precise risk factors to ascertain the likelihood of PC development, some of which epidemiologically have seen a strong correlation to disease development . The databases are a key step but are still in their infancy, with results anticipated in the future. Many other molecules have been characterized as possibly having high prognostic, diagnostic, or druggable values, although different studies performed have so far provided conflicting information as to their true nature in the management of PC . This provides a unique problem that is only now being realized where single targets or even a small subset of related targets may not confer the expected outcome when targeted or used as biomarkers in PC.

The traditional strategies in designing treatment have centered around combination therapies as well as single-agent targeted therapies but have slowly shifted toward a multitargeted approach . The identification of multiple pathways and effective targets has produced several approved inhibitors and multitarget drugs, yet a proper sequence of delivery remains a question due to the lack of comparative clinical studies . One of the difficulties in the use of combination therapies is in part due to an unclear picture of precise molecular interactions between the different drugs within the used combination. In a previous study, we were able to demonstrate the delicate tug-of-war where oxaliplatin, a key component of FOLFOX and FOLFIRINOX combination therapies, participates in the activation of tumorigenic mechanisms . PC is an inherently complex disease, with multiple factors influencing tumor growth, and contains very intricate genetic composition . The key to better targeting and treatment depends on proper validation in the appropriate preclinical and clinical stages where accurate identification of biomarkers becomes vital in order to identify a patient population, predict efficacy in their treatment, and assess the mechanism of treatment failure .