Generation and Impact of Neural Invasion in Pancreatic Cancer

Pancreatic cancer (PCa) as one of the most aggressive malignancies of mankind has an unparallelled propensity to invade intrapancreatic nerves. This “neural invasion” is therefore one of the most frequent routes of spread in PCa in addition to lymphatic and vascular paths. The major clinical relevance of neural invasion (NI) has triggered intense research efforts to understand its pathomechanims, and the findings derived from all these studies show how multi-faceted this peculiar route of cancer invasion in PCa is. This chapter is devoted to a thorough description of the characteristics, the pathomechanism and the clinical impact of NI in PCa, with the discussion of the most important pathways which may be future targets for therapeutic intervention.


Introduction
Pancreatic cancer (PCa) as one of the most aggressive malignancies of mankind has an unparallelled propensity to invade intrapancreatic nerves. This "neural invasion" is therefore one of the most frequent routes of spread in PCa in addition to lymphatic and vascular paths. The major clinical relevance of neural invasion (NI) has triggered intense research efforts to understand its pathomechanims, and the findings derived from all these studies show how multi-faceted this peculiar route of cancer invasion in PCa is. This chapter is devoted to a thorough description of the characteristics, the pathomechanism and the clinical impact of NI in PCa, with the discussion of the most important pathways which may be future targets for therapeutic intervention.

What is "neural invasion" in pancreatic cancer?
NI is a relatively new and more comprehensive term for the traditionally used description "perineural invasion". In several malignancies, e.g. in the prostate, head and neck, but also several gastrointestinal malignancies, cancer cells are commonly encountered around nerves (Liebig et al., 2009). The frequent presence of cancer cells along the perineurium, the protective sheet around neural fascicles, has hence made pathologists adopt the term "perineural invasion". Classically, cancer cells which penetrate through the epineurium come to lie between the epineurium and the underlying perineurium and "push" on the nerve fascicles within the constrained intraneural area. The earliest report on perineural invasion stems from Cruveilheir in 1835 where he noticed that cancer cells can actually extend along the invaded nerves . Interestingly, although pathologists frequently observed perineural invasion in several types of tumors, its role except for serving as an additional path of cancer spread has not been genuinely investigated and understood until 1990s. In his pioneering article on the ultrastructural features of perineural invasion in PCa, Dale Bockman from Augusta, Georgia, USA performed an electron microscopic analysis of invaded nerves in PCa (Bockman et al., 1994). There, he noted that, in contrast with the traditional assumption, PCa cells penetrate through the perineurium and become intimately associated with the interior of nerve fascicles, i.e. axons and Schwann cells (Bockman et al., 1994). These observations made by Bockman during mid-1990s laid the foundation for our understanding and the subsequent research on NI in PCa into the present time.

What are the specific histological characteristics of NI in PCa?
A review of scientific literature on NI in PCa reveals that NI has mostly been perceived as the presence of PCa on and along the perineurium (Kayahara et al., 2007;Liebig et al., 2009). Recent studies, however, revealed that PCa cells are readily encountered in the endoneural area, i.e. between nerve fascicles, as originally observed by Bockman . Consequently, subsequent studies applied a scoring system to describe the degree of penetration of PCa cells into intrapancreatic nerves, classifying NI into "no invasion" (score of 0/zero), "perineural invasion" (score of I/one) and "endoneural invasion" (score of II/two) (Ceyhan et al., 2006;Ceyhan et al., 2011). Importantly, the presence of PCa cells in the interior of nerves should not necessarily be perceived as the invasion of the "endoneurium" which is the connective tissue within nerve fascicles of a nerve. Similarly, invasion around nerves does not directly imply invasion of the "perineurium" as the connective tissue layer encircling nerve fascicles. Rather, "endo-" and "perineural" stand for "between" or "around" nerve fascicles ( Figure 1) Ceyhan et al., 2011).
Careful examination of PCa tissue specimens also reveals that the presence of NI is not independent of the localization of the nerves within the pancreatic tissue. Particularly, we could demonstrate that NI in ductal adenocarcinoma of the pancreas is detected more frequently in areas with severe desmoplasia . The reasons for this association between desmoplasia and NI are so far not known. However, it is assumed that the extracellular matrix is a rich source of growth factors which may be trophic upon nerves (Zhu et al., 1999;. Still, NI should not be assumed to be limited to desmoplastic areas. In a histopathological study on the normal pancreatic regions of PCa patients who underwent pancreatic resection, NI was also encountered in normal pancreatic areas which are distant from the actual tumor (Takahashi et al., 1997). In the original study, this type of NI termed "nex" was found in more than 50% of resected pancreatic specimens and correlated to the grade of intrapancreatic neural invasion or the presence of extrapancreatic neural plexus invasion (Takahashi et al., 1997). Moreover, its presence was also found to correlate to worse survival after removal of the tumor (Takahashi et al., 1997). Therefore, the presence of NI in the supposedly normal regions of the pancreas implies that NI is a rapidly progressive process where PCa cells grow very early along intrapancreatic nerves.
The different histological appearances of NI in PCa have also been employed to understand its pathomechanism and spread pattern. In a study by Kayahara et al. (Kayahara et al., www.intechopen.com 2007), the investigators analyzed consecutive sections of surgically resected PCa specimens in order to elucidate the main patterns of cancer cell growth along nerves: (1) direct invasion of the nerves, (2) continuous tumor cells growth in the perineural space, (3) branching of the growing tumor mass along neural branches, (4) formation of a foremost growth cone of tumor cells, and (5) direct invasion of contiguous lymph nodes (Kayahara et al., 2007). Hence, the authors could provide an anatomical mechanism for the manifestation of NI and particularly extrapancreatic neural plexus invasion. Importantly, their study proved the continuous growth of PCa cells along intrapancreatic nerves towards the extrapancreatic neural plexus (Kayahara et al., 2007).

Why is neural invasion so deciding in the course of PCa?
There are several factors which make NI a crucial aspect of PCa and an attractive field of research. First, NI has an utmost high prevalence in PCa, varying between 88% to 100% (Takahashi et al., 2001;Liu&Lu, 2002). Interestingly, according to a study by Kayahara et al., NI in PCa is significantly more common than in cancers which originate from direct anatomical neighbours of the pancreas, e.g. cancers of the distal bile duct or carcinoma of the papilla of Vater Kayahara et al., 1993;Kayahara et al., 1994;Kayahara et al., 1995;Kayahara et al., 1996). Unfortunately, intrapancreatic NI is nearly always accompanied by invasion to the extrapancreatic neural plexus: In their series, Nakao et al detected intrapancreatic NI in 116 out of 129 (90%) patients, of whom 80 (69%) showed extrapancreatic nerve plexus involvement (Nakao et al., 1996). Based on these findings, it seems that clinicians should assume the presence of NI in every patient with PCa even if the pathology report does not include a statement regarding this histopathological feature.
In the face of such a high prevalence, NI is at the same time one of the foremost reasons for local tumor recurrence after curative tumor resection Nagakawa et al., 1991;Kayahara et al., 1995;Kayahara et al., 1996;Nagakawa et al., 1996;Ozaki et al., 1999). In a study by Kayahara et al., the investigators analyzed the mode of recurrence in 30 patients who had originally undergone macroscopically curative resection (Kayahara et al., 1993;Liu&Lu, 2002). They showed that the rate of local retroperitoneal recurrence, i.e. the prevalence of extrapancreatic NI was 80%, of hepatic metastasis 66%, of peritoneal dissemination 53%, and of lymph node recurrence 47%, an observation which was confirmed by further antemortem studies (Kayahara et al., 1993;Kayahara et al., 1995;Liu&Lu, 2002). Among the several parts of the extrapancreatic neural plexus which demonstrate local tumor recurrence, pancreatic head plexus and splenic plexus are the most common sites of tumor recurrence (Liu&Lu, 2002). Based on the frequency of NI towards retropancreatic neural plexus, several surgeons advocated routine extended resections (including celiac plexus) or at least more aggressive surgery in the surgical treatment of PCa (Hiraoka et al., 1986;Nagakawa et al., 1991;Nagakawa et al., 1996;Imamura et al., 1999). However, subsequent clinical studies confirmed that, while extended resection -even in combination with radiotherapy-can contribute to local tumor control, there is no survival benefit for patients due to the early spread pattern of PCa (Bachmann et al., 2006;Takamori et al., 2008;Yokoyama&Nagino, 2011). However, these studies mostly concentrated on more extensive lymphadenectomy rather than plexus resection as the actual measure to reduce NI. In further newer studies Sperti et al., ;Kondo et al., 2001;Hirano et al., 2007;Chakravarty et al., 2011), the feasibility and safety of an en bloc resection including celiac artery, plexus and ganglia was demonstrated, but the actual survival beneift from this Fig. 1. Severity of neural invasion (NI) in pancreatic cancer (PCa). Examination of intrapancreatic nerves with NI reveals that PCa cells demonstrate varying degrees of interaction with the nerves. In many cases, PCa cells surround intrapancreatic nerves without breaching the epineural barrier (A, also termed epineural association). On the other hand, several nerves demonstrate a lack of the epineural barrier where pancreatic cancer cells surround the fascicles along their perineurium (B, perineural invasion). In most severe cases, PCa cells are encountered between nerve fascicles, along their endoneurium (C, endoneural invasion). There is a significant association between the severity of NI and the degree of pain sensation among PCa patients (please refer to the main text for the respective references). All images at 200x magnification.
radical operation remains to be demonstrated. Importantly, a common denominator of these studies on "en bloc" resection of retropancreatic neural plexus is the pronounced pain relief as a result of the resection of celiac plexus (Kondo et al., 2001;Hirano et al., 2007). This deciding association between extrapancreatic neural plexus and pain sensation builds up the link to the concept of "pancreatic neuropathy" in PCa which encompasses NI and several other neural alterations in PCa , as explained in the following section.

Neural invasion as part of "pancreatic neuropathy" in PCa
The pancreas is one of the most densely innervated visceral organs (Bradley&Bem, 2003). The extrinsic component of its innervation is composed of nerve fibers running within the vagal and splanchnic nerves which originate from vagal nuclei or DRGs, respectively. Like the intestine, it also has an intrinsic innervation which is represented by intrapancreatic neurons. Importantly, enteric and intrapancreatic neurons are embryologically closely related: intrapancreatic neurons develop from a subgroup of neural crest-derived enteric nervous system (ENS) precursors and thus belong to the ENS (Kirchgessner&Gershon, 1990;Kirchgessner&Gershon, 1991). Moreover, there exists a direct innervation of the pancreas from the duodenum termed "entero-pancreatic innervations", as evidenced by the entrance of nerve fibers directly from the duodenal ENS into intrapancreatic ganglia (Kirchgessner&Gershon, 1990;Kirchgessner&Gershon, 1991).
In the currently most comprehensive systematic analysis of NI in PCa, our group aimed at the study of nerve morphology in 546 patients with different pancreatic tumors, including ductal adenocarcinoma, neuroendocrine tumors, intraductal papillary mucinous neoplasms (IPMN), serous and mucinous cystadenoma and other neoplasms of the pancreas ). In the mentioned study, we could demonstrate that ductal adenocarcinoma of the pancreas exhibits the highest degree of NI in comparison to all other pancreatic tumors . Interestingly, ductal adenocarcinoma (PCa) also harbored an unparalleled degree of nerve alterations among all these tumors ). In particular, PCa was characterized by a prominently increased neural density, a pronounced neural hypertrophy and neural inflammatory cell infiltration ("pancreatic neuritis") (Ceyhan et al., 2006;. Moreover, we could also detect a key link between the severity of NI in PCa and the extent of intrapancreatic neuroplastic alterations: The more nerves and neural hypertrophy were present, the higher was the extent/severity of NI in PCa ).
This association between pancreatic neuroplasticity and NI gained a further dimension in a subsequent study where the pancreatic "innervation quality" in PCa was studied and compared to normal human pancreas (NP) ). Interestingly, not only had nerves in PCa tissue fewer sympathetic nerve fibers than in NP, but nerves with NI had at the same time reduced amounts of both sympathetic and cholinergic nerve fibers ). This "neural remodeling" in PCa implies that PCa cells may not be arbitrarily invading intrapancreatic nerves but also aiming at specific fiber qualities for so far unknown reasons. Overall, these neural alterations which seem to be specific for PCa, i.e. pancreatic neuroplasticity, neural remodeling and the high degree of NI, are the three hallmarks of socalled "pancreatic neuropathy" in PCa .
While the mechanisms of these neuroplastic alterations are not completely understood, there is increasing evidence that these neuropathic alterations in PCa can in part be attributed to the neurotrophic character of the tumor microenvironment in PCa . In a novel in vitro neuroplasticity assay, we could demonstrate that tissue extracts of PCa, PCa cell supernatants and supernatants of human pancreatic stellate cells as main generators of desmoplasia can all induce axonal sprouting, increased neurite density and perikaryonal hypertrophy of neurons isolated from dorsal root ganglia or myenteric plecus under in vitro conditions . In a very recent study, Li et al. added a novel dimension to our understanding of neural alterations in PCa: In accordance with their former hypothesis (Li&Ma, 2008), patients with hyperglycemia demonstrate more pronounced neural hypertrophy and increased neural density than normoglycemic patients (Li et al., 2011). Hence, it is to be expected that research on pancreatic neuroplasticity and especially NI in PCa may take a direction towards increased investigation of the impact of impaired glucose metabolism upon pancreatic neuropathy in PCa.

The role of neural invasion in the pain due to PCa
Decreased survival and local tumor recurrence are undoubtedly among the leading factors which make NI into a highly relevant clinical subject. However, within the true clinical impact of NI, its role in pain sensation occupies a special place. It has long been accepted that the extension of PCa along the intrapancreatic nerves towards extrapancreatic neural plexus may be a causal factor in the generation of pain in advanced Pca Nakao et al., 1996;Kayahara et al., 2007). Bockman also postulated a significant role for NI in pain generation in PCa (Bockman et al., 1994), but the actual pioneering study in this context came from Zhu et al. who for the first time demonstrated the correlation between the intrapancreatic expression of the nerve growth factor (NGF) (Zhu et al., 1999), the frequency of perineural invasion and the degree of pain sensation in PCa patients, an observation which was later also discovered for the expression of NGF receptor TrkA (Zhang et al., 2005;Dang et al., 2006). Owing to this study and its successors, it became increasingly clear that the extent of NI in the pancreas affects pain sensation, where nervederived molecules like NGF play a key role in both pain sensation and potentially in the attraction of PCa cell to nerves . The resulting interest in such nervederived mediators, especially in neurotrophic factors, have inaugurated the era of research on the molecular biological mechanisms of NI which last until the present time . Moreover, the identified cross-link between NI, pancreatic neuroplastic alterations and pain sensation by PCa patients revealed the potential involvement of "neuropathic" pain mechanisms in PCa . Hence, researchers and clinicians have recently and increasingly understood that damage to nerves within the pancreas may be the actual pain-triggering mechanism in PCa .
One can assume that the blockade of pain transmission via the damaged nerves from the pancreas may be of major benefit to treat pain due to PCa. As the celiac plexus contains a large portion of the afferent nerve fibers from the pancreas, several studies have tested the efficieny of celiac plexus blockade/neurolysis in the treatment of pain due to PCa. In all these studies, patients had significant pain relief (Wong et al., 2004;Stefaniak et al., 2005;Yan&Myers, 2007) following this intervention. While the efficiency of this "denervation" www.intechopen.com technique does not necessarily prove the neuropathic character of pain in PCa, it underlines the deciding contribution of nerves and the transmitted signals in the generation of the pain syndrome in PCa  Considering the neuropathic character of pain in PCa, one can assume that neuropathic analgesics may be of benefit to treat PCa-associated pain. As of today, the impact of neuropathic analgesic regimens to treat of patients with advanced PCa has not yet been systematically investigated.

In vitro models
Researchers and clinicians have long puzzled about why PCa cells are frequently encountered around intrapancreatic nerves. Early reports had claimed that PCa cells enter nerves through the perineurium at its weakest points, i.e. along neural lymph vessels(di Mola&di Sebastiano, 2008), which, however, could not be confirmed in later studies. In later studies, investigators suggested that PCa cells grow along the path of least resistance after entering nerves, which was thought to be the perineural space (Rodin et al., 1967;Bockman et al., 1994;di Mola&di Sebastiano, 2008). Indeed, a higher proliferative index and decreased apoptosis in the perineural space could previously be shown for prostate cancer cells invading nerves (Ayala et al., 2004). However, newer studies could demonstrate that limiting PCa cells' presence around nerves to the local physical circumstances may be an oversimplification of the utmost frequent NI in Pca . In particular, the development of novel in vitro research tools to study NI in PCa has enabled the discovery of a true cancer-nerve affinity as an important biological mechanism in Pca (Zhu et al., 1999;Zhu et al., 2002). Especially, we know today that peripheral nerves in the tumor microenvironment can serve a source of tumor-trophic factors and cancer-attracting molecules (Zhu et al., 1999;Gil et al., 2010). This "biological cancer-nerve affinity" in PCa is today one of the cardinal pathomechanistic concepts in our understanding of NI in Pca (Zhu et al., 1999;Gil et al., 2010).
This increased appreciation of cancer-nerve affinity in PCa was largely possible owing to increased efforts to develop novel advanced in vitro models of NI in PCa. These models generally employ heterotypic co-cultures of neurons and PCa cells as in of the earliest models by Dai et al (Dai et al., 2007). In their study, the investigators co-cultivated the human PCa cell line MiaPaCa-2 with neurons from mouse dorsal root ganglia (DRG). In accordance with the hypothesized trophic effect of nerves, PCa cells which were co-cultured with DRG exhibited stronger growth than non-co-cultured control PCa cells and overexpressed prosurvival genes like MALT1 and TRAF (Dai et al., 2007). As a frequent observation also made by other current models, also PCa cells supported the growth of the neurons, as evidenced by their increased neurite growth (Dai et al., 2007).
This mutual trophic effect gained a further dimension in a recent study by our group where we presented another in vitro model which allows a precise spatiotemporal monitoring of NI by PCa cells . As shown in the original article, different PCa cell lines were co-cultured together with rat DRG or myenteric plexus (MP) cells in a threedimensional (3D) extracellular matrix (ECM)-based migration assay .
The presented assay offers several advantages: First, it initiates from a clear-cut physical separation of PCa cells and neurons, as it is the case under in vivo conditions. Therefore, the model allows exact monitoring of cell behavior from the very beginning. Second, the model includes a pre-defined migration path for PCa cells, i.e. defined ECM-bridges, which allows the generation of a chemical gradient for any chemotactic factor . Using this novel assay, we could demonstrate that PCa cells react to the presence of neurons with a characteristic morphological alteration including cell flattening, grouping, colony formation and spike-like cellular polarization directed towards neurites. Following their targeted migration towards neurons, PCa cells established physical contact with neurites along which they were guided in their migration . Similar to the findings by Dai et al., we also observed increased neurite growth from DRG neurons towards PCa cells when compared to non-co-cultured DRG neurons . Moreover, the presented 3D-migration assay for the first time included a neuronal subtype which represents the intrinsic pancreatic neurons, i.e. neurons of the enteric nervous system (ENS).
The key role of neurotrophins which was initially demonstrated by Zhu et al. for NGF could be confirmed by means of this novel assay where we monitored the quantitative alterations in the expression of neurotrophins, their receptors and the members of the glial-cell-derived neurotrophic factor (GDNF) family (Zhu et al., 1999;. As opposed to several members of the GDNF family, the nerve growth factor (NGF) increased continuously throughout the migration process of 120 hours . Certainly, this novel 3D migration assays is at the same time a novel tool to investigate the contribution of numerous molecular factors from different cellular sources to NI in PCa. A very similar model, though without pre-defined paths for chemical gradient generation, has been recently reported by Gil et al. where the investigators could demonstrate a potent chemotactic effect of GDNF from DRG neurons upon PCa cells (Gil et al., 2010). In a further study, our group showed the potent enhancer effect of the neurotrophic factor artemin, a member of the GDNF family of neurotrophic factors, upon invasiveness of PCa cells (Ceyhan et al., 2006).
In the presence of an increasing number of studies on the role of neurotrophic factors in NI in PCa, there is only a limited of reports on the role of chemokines in NI in PCa (Marchesi et al., 2008;. In one of the first studies where the potential role of chemokines was recognized, Marchesi et al could show that the neural immunoreactivity for the receptor of the chemokine fractalkine, i.e. CX3CR1, was significantly higher in perineural invasive lesions of PCa (Marchesi et al., 2008). By using CX3CR1-overexpressing PCa cells for an in vivo implantation model, they could also demonstrate that CX3CR1-overexpressing PCa cells exhibited a more pronounced infiltration of peripheral nerves (Marchesi et al., 2008). This study stands out in the literature due to the seminal investigation of chemokines in the generation of NI in PCa and pointed to the CX3CR1-CX3CL1 as a potential therapeutic target.
Interestingly, beyond molecules with chemoattractive potential, other nerve-derived molecules have also been the focus of recent research on NI in PCa. From these, in a mouse perineural invasion and orthotopic transplantation model, the stable knockdown of synuclein gamma (synuclein-γ) via by short hairpin RNA significantly reduced the incidence of perineural invasion and liver and lymph node metastasis (Hibi et al., 2009). In another study where the investigators provided a novel perspective on NI in PCa, Swanson www.intechopen.com et al. showed that Schwann cells of peripheral nerves express myelin-associated glycoprotein (MAG) which can serve as a receptor for the transmembrane mucin MUC1 on PCa cells (Swanson et al., 2007). Hence, based on this study, it seems that nerves in PCa tissue not only chemo-attract PCa cells but also can undergo direct physical contact, as initially observed by Bockman (Swanson et al., 2007;. It is conceivable that the presented in vitro models in the literature would enable the identification of a gene set which would reflect the differentially upregulated genes in NI in PCa. Accordingly, in a recent study, Abiatari et al. aimed at obtaining a transcriptome signature of NI in PCa by means of an in vitro / ex vivo model . Specifically, they confronted human PCa cell lines with explanted rat vagus nerves and quantified the differentially regulated genes in highly versus less nerve-invasive PCa cells . Interestingly, the differentially regulated genes which were identified by this study were primarily related to cell motility, including kinesin family member 14 (KIF14) and Rho-GDP dissociation inhibitor beta (ARHGDIbeta), a gene set which they could expand in a subsequent by two molecules, i.e., the microtubule-associated protein MAPRE2 and the nuclear protein YPEL2 . Based on these important observations, the investigators underlined the importance of increased PCa cell motility in the generation of NI in PCa as additional molecular biological mechanism .

In vivo models
The increasing number of efforts to elucidate the pathomechanism of NI in PCa necessitated the creation of in vivo models which better mimic NI in human PCa. The common characteristic of these models is that they involve implantation of PCa cells as xenograft tumors in one out of several locations, e.g. into the pancreas, under the skin or in the proximity of large-diameter peripheral nerves to allow NI by PCa cells. In the first one of these models, Eibl et al. aimed at creating a model to simulate the high rates of local recurrence and NI after curative resection (Eibl&Reber, 2005). For this purpose, they performed complete surgical resection of the tumor at 4, 6, and 8 weeks after orthotopic implantation of the PCa cell lines MiaPaCa-2 (undifferentiated) and Capan-2 (welldifferentiated) in nude mice pancreas. Six weeks after tumor implantation, local tumor recurrence with extensive retroperitoneal nerve invasion and distant organ metastasis were observed in nude mice who had received MiaPaCa-2 cells (Eibl&Reber, 2005). Astonishingly, although the investigators achieved a successful simulation of the local recurrence and NI associated with PCa in this murine model, there has since been no application of this model to identify further pathomechanistic features of NI in PCa. Certainly, the model is probably not suitable to study the initial, early events leading to PCa, however, it can possibly be employed to examine the therapeutic potential of different agents on NI in PCa in an animal model of PCa. In another model, Koide et al. subcutaneously (s.c.) implanted different PCa cell lines with or without human peripheral nerves in nonobese diabetes/severe combined immunodeficient mice and analyzed the frequency of NI by these different cell lines (Koide et al., 2006). Furthermore, they performed an oligonucleotide microarray to obtain the expression profiles of high and low perineurally invasive cell lines. Interestingly, only two well-differentiated cell lines (Capan-1 and Capan-2) demonstrated invasion of mouse s.c. nerves. In these invasive cell lines, they identified over-expression of CD74 in the specifically perineural invasive cells, which they confirmed in human PCa tissue specimens (Koide et al., 2006). Despite the implantation of PCa cells under skin and not the actual organ of origin (i.e. the pancreas), this model sticks out owing to its uncomplicated performance. Still, it has to be underlined that none of these in vivo models has so far found widespread application.

Efforts of controlling neural invasion in PCa
Among all studies in the literature, there are so far two in vivo models of NI in PCa where a primarily therapeutic goal was pursued: In the first study, Gil et al. aimed at treating NI by means of an attenuated, replication-competent, oncolytic herpes simplex virus which inhabits nerves (Gil et al., 2007). After injection of PCa cell lines into the perineurium of the sciatic nerve of athymic mice, they monitored limb function for 9 days after injection. Excitingly, a single injection of the oncolytic herpes simplex virus 7 days after PCa cell injection effectively eradicated NI without compromising physiologic nerve function (Gil et al., 2007). The same group utilized this model in a subsequent study for intraoperative diagnosis of NI: By using enhanced green fluorescent protein (eGFP)-expressing oncolytic herpes virus, they could detect invaded nerves following intrasciatic implantation of PCa cell lines via intraoperative fluorescent stereoscopic imaging (Gil et al., 2008). Thereby, they proposed a novel tool for enhanced diagnosis and therapy of NI in PCa and for facilitated detection of invaded nerves in cases where an extended resection may be considered (Gil et al., 2008). In a third study, the group applied PCa cell injection onto mouse sciatic nerves and subsequently treated the mice with pyrazolopyrimidine-1, a tyrosine kinase inhibitor targeting the RET pathway (Gil et al., 2010). Strikingly, systemic therapy with this agent diminished nerve invasion toward the spinal cord and prevented limb paralysis (Gil et al., 2010). Still, the sensitivity and true effectiveness of their method remains to be confirmed in future studies.

Future directions in research on NI in PCa
Today, research on NI in PCa is in an era of "data collection" and "expansion of knowledge". The increasing number of in vitro and in vivo models, together with the rapidly growing scientific interest in NI, create the best possible conditions to learn and discover about this peculiar histopathological phenomenon. However, we are convinced that future studies should aim at the generation of models with an increasingly therapeutic intention, because there is urgent need to employ additional, novel tools to treat PCa. Furthermore, the main pathomechanistic hypothesis for the generation of NI, i.e. the neuro-affinity of PCa cells, has to be carefully reviewed. It should certainly be considered that PCa cells may not be responsible for every aspect of NI, but rather be reacting to the signals coming from the nerves. The increasing number of nerve-derived molecules like neurotrophic factors or neuronal chemokines which are continuously shown to contribute to NI should serve as a motivation to delve deeper into the involvement of neuronal molecules during NI.
While novel in vitro models of NI in PCa are being steadily developed, the currently available in vivo models still exhibit major deficits. In particular, these models lack: 1. Tumors that directly originate from the pancreas 2. Histopathological confirmation of the tumor phenotype as ductal adenocarcinoma 3. The confirmation of the presence of NI even at early stages of tumor development and progression 4. The specific extension of NI towards the extrapancreatic neural plexus 5. Accompanying neuropathic and desmoplastic alterations 6. Neuropathic pain sensation Therefore, future in vivo models of NI in PCa should be superior to the current ones in the above-mentioned aspects, especially because they should increasingly be employed to deduce therapeutic targets and strategies.

Summary and conclusion
Neural invasion in PCa bears a unique importance in the biology of this disease due its impact on patient survival, local tumor reccurence and neuropathic pain sensation. Higher interest in NI has paved path for increased research on the biology of NI and accelarated the development of numerous experimental models. The discussed in vitro and in vivo models which shall help to eludicate the pathomechanisms of NI in PCa may provide novel tools to control and to reduce NI in this highly aggressive human malignancy. Considering the dismal average prognosis associated with PCa, one may wonder about the actual benefit of reducing the specific invasion of nerves in this tumor entity. Here, it should be underlined that reduction of NI can be regarded as one of several possibilities to control tumor growth, just as adjuvant therapy as an oncological therapy regimen aims at reaching microscopic tumor presence and reducing the systemic tumor burden. The control of NI, however, bears a further special importance since NI is not only the probably most common mode of spread for PCa, but also because nerves represent the most frequent site of local tumor recurrence in PCa. Moreover, limitation of NI is likely to have a considerable impact upon the neuropathic pain syndrome and thus quality of life of patients with PCa. Therefore, NI may find increased attention in the future as an additional therapeutic target for increased survival, enhanced postoperative outcome and improved quality of life among all patients with this dreadful malignancy.

Acknowledgements
The authors are grateful to Dr. Matthias Maak for his assistance with the figures.