Pancreatic cancer is one of the most lethal cancers worldwide, associated with poor survival rates due to frequently delayed diagnosis and limited treatment options
[1]. Among pancreatic cancers, pancreatic ductal adenocarcinoma (PDAC) represents the most common histological subtype, accounting for more than 90% of all cases
[2]. Its incidence is dramatically increasing in the Western world, for yet widely unknown reasons. With survival rates only marginally improving, the 5-year survival rate is still appallingly low around 9%
[3]. Thereby, PDAC exhibits the worst prognosis among all solid tumors
[4], currently ranking as third leading cause of cancer-related deaths in the US
[5]. By 2030, it is expected to become the second leading cause of cancer-related deaths in Western societies
[4][6][7]. Approximately 10% of all PDAC cases are based on hereditary genetic predispositions
[8]. In addition, several lifestyle factors have been shown to significantly increase the risk of developing PDAC. Besides smoking, chronic pancreatitis and diabetes mellitus, obesity represents one of the most significant risk factors
[9][10].
Overweight and obesity have previously been reported as risk factors for a variety of chronic and metabolic diseases such as type 2 diabetes mellitus, hypertension, cardiovascular disease, and metabolic syndrome
[26][27]. In addition, there is a clear link between obesity and an increased risk for numerous malignancies
[28][29], including pancreatic cancer
[9]. Obesity is the most important avoidable risk factor for cancer
[30], being responsible for 14% of cancer deaths in men and 20% of cancer deaths in women worldwide
[31]. In Germany, it has been estimated that in 2018 around 7% of all newly diagnosed cancer cases were caused by obesity
[32]. There is mounting evidence that both incidence and mortality of pancreatic cancer are significantly increased among obese individuals
[9][33]. In line with this, it has been shown that obese people are already at a higher risk of developing pancreatic precancerous lesions
[34]. The link between obesity and cancer seems to be multifactorial. In addition to the influence of proinflammatory cytokines such as IL-6 or TNF-alpha, growth-stimulating effects of various obesity-associated hormones such as leptin, estrogen, or insulin have been well described
[19][35][36][37]. In terms of pancreatic cancer, coherences and mechanisms of obesity-driven carcinogenesis have been reviewed previously
[38][39][40][41][42].
However, the underlying molecular mechanisms linking obesity to PDAC development and progression remain largely unknown. Therefore, it is crucial to develop realistic and physiologically accurate models of obesity-induced pancreatic carcinogenesis. The following contents aims to introduce current in vitro and in vivo models of PDAC and obesity and shed light on the newest generation of preclinical models to investigate obesity-driven pancreatic carcinogenesis (Figure 1).
Preadipocytes (blue) can be differentiated in vitro into adipocytes (yellow) with similar characteristics compared to in vivo rose adipocytes. Otherwise, mature adipocytes can be isolated out of the fat tissue and cultured for a couple of days (ceiling culture) or up to two weeks (membrane mature adipocyte aggregate cultures = MAAC) until dedifferentiation.
Pancreatic cancer cell lines (red) or isolates can be cultured in 2D and 3D models. Cancer associated fibroblasts (yellow) can be added for increased physiological relevance.
Common murine obesity models are based on a genetically engineered deficit in Leptin signaling (ob/ob and db/db mouse) or are the result of a high caloric diet (high fat or western diet). Pancreatic cancer in mice can arise from genetically engineered pancreas-specific mutations or induced by xenograft implantation of pancreatic cancer cells, tumor chunks, as well as organoids.
The combination of in vitro and in vivo models allows the creation of models to study obesity-driven pancreatic carcinogenesis. Individual advantages and disadvantages of the chosen models should be considered with regard to the specific scientific question. The figure was created by using BioRender (BioRender.com, accessed on 14 August 2022)
2. Review of Current Methodologies
2.1. Murine/Human Adipocyte In Vitro Models
The systemic impact of obesity is highly complex, with adipocytes interacting with multiple other cell types directly or indirectly via secreted factors
[17]. In obesity-associated cancer, the crosstalk between adipocytes and immune cells is instrumental in modulating carcinogenesis and tumor progression
[36]. Adipocytes account for 90% of the volume, but only for 20–40% of the total cell number in adipose tissue
[17][19]. The majority of non-adipocyte cells in adipose tissues are immune cells
[19]. Compared to normal-weight individuals, immune cell composition is markedly different in obese persons
[20][21][25]. Considering the importance of the different immune components in obesity-driven pancreatic carcinogenesis, exploring the dynamic interaction between the adipose tissue and resident and/or infiltrating immune cells during tumor development and progression would provide further insight into the pathogenesis and possibly open new therapeutic avenues. Therefore, appropriate in vitro and in vivo models recapitulating obesity-driven pancreatic carcinogenesis and tumor progression are urgently required.
Because of their functional relevance and high prevalence in obesity, these contents focus on white adipocytes. Dufau et al. have previously published a detailed overview on different rodent and human adipose cell models
[43]. Generally, in vitro differentiated adipocytes must be distinguished from isolated mature primary adipocytes. In vitro differentiation is feasible both for murine embryonic fibroblast cell lines and primary isolated preadipocytes.
Standard mouse cell lines include 3T3-L1, 3T3-F442A, and C3H10T1/2 cells
[44][45][46]. After reaching confluence, those fibroblasts can be differentiated into adipocytes by using distinct hormonal differentiation stimuli
[47]. The use of cell lines offers a highly reproducible in vitro model, sparing the need to isolate primary adipose tissue. On the other hand, the cell lines used are immortalized and therefore only partly representative for primary adipocytes. In addition, several factors can influence the cell line’s capacity to differentiate in vitro, including confluence, cell passage number, serum source and lot number, contamination with mycoplasma, as well as reagent stability
[46][48][49][50], creating difficulties for comparison among different labs.
For the study of primary preadipocyte cells, the most frequently used method is the isolation of stromal vascular fraction (SVF) from the rodent adipose tissues for which several protocols have been established
[51]. The stromal vascular fraction contains heterogeneous cells, including adipose-derived stem cells (ADSCs), endothelial and mesenchymal progenitor cells, immune cells
[52] and epithelial cells, which may limit the initial purity of the preparations. The proportion of those cell populations might vary between isolations and is affected by several factors like age, sex and nutritional stage
[53][54]. In addition, different protocols used for in vitro differentiation have been shown to affect the phenotype and molecular profile of the differentiated adipocytes
[49]. Nevertheless, the primary isolation of SVF from genetically modified mice enables adipocyte-specific studies on the impact of specific genetic alterations. Compared to cell line-based in vitro differentiated adipocytes, the biology and metabolism of SVF-based adipocytes are closer to that of primary mature adipocytes
[55].
Compared to primary SVF preadipocytes, isolation and culture of primary mature adipocytes is experimentally challenging: these cells have a short ex vivo life span and are fragile, thereby handling can be demanding
[47]. Additionally, the high lipid content causes floating of the cells, necessitating a special ceiling culture
[43], for which flasks are completely filled with media and floating adipocytes attach to the upper plastic surface. A caveat of culturing mature adipocytes is their rapid dedifferentiation into fibroblast-like cells
[56]. To extend the time span for culturing and decrease dedifferentiation, Harms et al. developed a new method called membrane mature adipocyte aggregate cultures (MAAC), in which mature adipocytes are cultured under a transwell membrane, thereby preventing dedifferentiation up to two weeks
[56]. While this method works sufficiently for human mature adipocytes, murine mature adipocytes are even more challenging to culture
[56].
Other methods to culture mature adipocytes are tissue explant cultures, which are often used to investigate adipose tissue-derived inflammation and metabolic activity
[43]. As it is the case for most primary cells, these cultured adipocytes also change their phenotype after a few days ex vivo
[56].
The availability of murine adipose tissue compared to human primary material is apparently much simpler, and murine adipose cell models have traditionally been most commonly utilized. However, translating results from murine-based experiments to humans also requires in vitro models using human cells. To this extent, a handful of human cell lines are available. Yet, those cell lines result from artificial immortalization or are based on pathological conditions of the donor, which might affect the generalizability of results
[43]. As described for mice, adipose-derived stem cells (ASCs) can also be isolated from human adipose tissues and differentiated in vitro into adipocytes
[57][58]. However, results obtained with human adipocytes might be significantly affected by interindividual differences between the different donors
[57]. In addition to ASCs, isolation and culture of mature adipocytes is also feasible but underlies similar challenges as in mice
[56].
Three-dimensional (3D) culture of in vitro differentiated adipocytes enables higher differentiation rates and unilocular lipid storage
[43]. Disadvantages of this technique are the underlying experimental challenges as well as the higher culturing costs.
Taken together, all in vitro models have inherent limitations, most prominently the missing complex interaction of adipocytes with other organs and cell compartments. Therefore, animal models are still necessary to investigate the effects of obesity and get a better understanding of the pathological changes.
2.2. Murine In Vivo Obesity Models
Since mice are the most widely used in vivo models, these contents focus on murine obesity models. In general, either genetically modified or diet-induced mouse models have been commonly used to study the impact of obesity on a broad variety of diseases. In comparison, surgical (e.g., by inducing hypothalamic lesions) or drug-induced models play a minor role
[59]. Lutz et al.
[59], as well as Suleiman et al.
[60], reviewed different obesity mouse models in great detail. In brief, the most commonly used genetic mouse models are based on modifications in leptin, its receptor or downstream signaling. These mice develop obesity due to increased food intake and reduced energy expenditure
[61][62]. Limitations of these models are obesity-independent leptin effects on several other cell types. In particular, leptin has a significant influence on the immune response
[36][63][64] which can impact the phenotype of these mouse models, especially when studying the impact of obesity on carcinogenesis
[63].
Given these limitations, diet-induced obesity (DIO) mouse models are most commonly used, especially since they readily recapitulate the most common, hyperalimentation-induced cause of obesity
[59]. By chronic exposure to a high-calorie diet, mice gain weight and develop obesity
[65]. One limitation of these models is the fact that the various diets in use differ in their nutritional content. The most commonly applied diet is a high-fat diet, in which 20–80% of its calories are based on fatty acids
[65][66][67]. Due to differences in the typical human diet in the Western world, which is predominantly carbohydrate-based, some researchers use a Western diet which more closely reflects the human dietary habits in developed Western countries
[68]. A limitation of all DIO models is the uncertainty if effects are caused by obesity directly or by other factors like nutritional content or obesity-associated stress
[65]. All mouse models allow the study of complex metabolic effects in vivo. Although animal models of obesity and related metabolic illnesses provide valuable insights, it must be kept in mind that their transferability to the human situation is limited due to variations in metabolism and physiology between mice and humans
[57][59]. An example in this context is the basal metabolic rate, which is seven times higher in mice than in humans, which causes differences, e.g., in senescence
[69].
3. Pancreatic Ductal Adenocarcinoma (PDAC)
PDAC is the most common histological type of pancreatic cancer
[2]. Approximately 90% of all PDACs in humans are characterized by activating mutations in the proto-oncogene Kras as key driver
[2][70], among them 98% exhibiting missense mutation in one of the three mutational hot-spots: glycine-12 (G12), glycine-13 (G13) or glutamine-71 (Q61), all causing a permanent activation of Kras
[71]. Kras mutation is one of the earliest genetic events in PDAC carcinogenesis but is insufficient to drive PDAC development alone. Therefore, several additional genetic or epigenetic hits are required
[72]. PDAC usually develops via different pancreatic precursor lesions, including mucinous cystic neoplasms (MCN), intraductal papillary mucinous neoplasms (IPMN) and pancreatic intraepithelial neoplasias (PanIN). Most PDACs develop from microscopic PanINs, which cannot be detected by conventional imaging methods
[73]. Based on their histological appearance PanIN can be categorized into PanIN grades 1–3
[74][75][76], with PanIN 1 lesions already exhibiting Kras mutations
[72]. During progression to invasive PDAC, additional inactivating mutations in tumor suppressors such as CDK2N2A, SMAD4 or TP53 are frequently acquired
[70][72]. Suitable in vitro and in vivo models have been developed to recapitulate human pancreatic carcinogenesis and characterize the underlying molecular driver events in detail.
Several integrated genomic studies provided molecular PDAC classifications and correlated the probability of treatment response and survival to those categories
[70][77][78]. Among them, the two major categories have been termed “classical epithelial” and “basal-like” (also called quasi-mesenchymal or squamous) subtypes
[77][79]. Human tumors and PDAC cell lines frequently represent a heterogeneous continuum of subtypes rather than a constant state
[79]. Interestingly, chemotherapy treatment may trigger shifts between subtypes
[79]. Knowledge of the respective subtype of a certain cell line is important for interpreting in vitro results. For example, the most common human PDAC cell lines Panc-1 and MiaPaca2 are classified as basal-like subtypes, whereas Capan2 and HAPFII are classified as classical epithelial subtypes
[79]. In humans, the molecular subtypes have gained increasing attention as predictive tools for selecting molecularly guided (neo)adjuvant or palliative treatment regimens
[80].
In addition to a complex and heterogeneous genetic background, the tumor microenvironment in PDAC exerts an important, yet still controversial, impact on cancer development and chemoresistance, comprising up to 90% of the tumor volume
[81][82][83][84][85]. Stromal components include immune cells, cancer-associated fibroblasts (CAF), endothelial and nerve cells as well as numerous extracellular matrix (ECM) components
[83]. ECM is mainly produced by CAFs
[86], but also by cancer cells themselves. Collagens, integrins, proteases, and proteoglycans are the predominant components of ECM
[85]. It still remains inconclusive under which exact spatial and temporal circumstances ECM can support or suppress cancer progression
[87]. Targeted depletion of ECM components has been shown to increase intratumoral chemotherapy concentrations in murine PDAC models
[88]. However, in contrast to the expectations, pharmaceutical depletion of ECM has resulted in a more aggressive disease in clinical trials underlining the complexity of this interaction
[87][89]. CAFs are usually derived from pancreatic stellate cells (PSC)
[90]. Based on their secretory and local functions, they can be classified as myofibroblastic CAFs (myCAF) and inflammatory CAFs (iCAF). MyCAFs mediate direct juxtacrine interactions with cancer cells and therefore are frequently located in direct tumor cell contact
[82]. They are characterized by a high expression level of alpha-smooth muscle actin (α-SMA)
[82]. In contrast, iCAFs are spatially distant from cancer cells, but their induction depends on secreted cancer cell-derived mediators
[82]. In turn, iCAFs can induce STAT3 signaling in PDAC
[82] by producing pro-inflammatory cytokines, especially IL-6
[80] which is known to also cause several systemic effects of PDAC like cachexia
[91] and decreased immunotherapy response
[84][92].
All in all, there is a complex interaction between PDAC and its microenvironment. Relevant preclinical models and clinical trials must recapitulate this complex interplay, providing preclinical in vivo platforms to evaluate combinatorial targeting approaches of both tumor cell autonomous and non-autonomous components.