Current Pathology Model of Pancreatic Cancer

Simple Summary Pancreatic cancer is a very aggressive and lethal malignant neoplasm with overall 5-year survival rates below 10%. The field of pancreatic cancer research is rapidly evolving. Reports of newly revealed pathomechanisms of the nature of these tumors are published daily. Nevertheless, many aspects of a pathologic evaluation are still uncertain. It is crucial to be able to pull out practical information that impacts the diagnostic process, called a pathologic evaluation. In this review, we comprehensively summarize some of the recent papers from the pathologists’ and clinicians’ points of view. We specifically focus on pathology assessment and reporting, to make them meaningful for clinical and research purposes. Lastly, we highlight novel diagnostic and research approaches, point out some missing pieces in the field, and suggest further study directions. Abstract Pancreatic cancer (PC) is one of the most aggressive and lethal malignant neoplasms, ranking in seventh place in the world in terms of the incidence of death, with overall 5-year survival rates still below 10%. The knowledge about PC pathomechanisms is rapidly expanding. Daily reports reveal new aspects of tumor biology, including its molecular and morphological heterogeneity, explain complicated “cross-talk” that happens between the cancer cells and tumor stroma, or the nature of the PC-associated neural remodeling (PANR). Staying up-to-date is hard and crucial at the same time. In this review, we are focusing on a comprehensive summary of PC aspects that are important in pathologic reporting, impact patients’ outcomes, and bring meaningful information for clinicians. Finally, we show promising new trends in diagnostic technologies that might bring a difference in PC early diagnosis.


Introduction
Pancreatic cancer is one of the most aggressive and lethal malignant neoplasms, ranking in seventh place in the world in terms of the incidence of death [1] and it is projected to surpass breast, prostate, and colorectal cancers to become the second leading cause of cancer-related deaths by 2030 [2]. It was estimated that there were 466,003 new pancreatic-cancer-related deaths in 2020 worldwide [3]. Significant improvements in diagnosis and management have not improved 5-year survival rates, which remain below 10%. The reason for this fact, among others, is the tumor heterogeneity including molecular aberrations, but also the tumor nature, appearing as a wide spectrum of patterns involving cancer gland formation and tumor stroma composition, which is much underrated among pathologists. The knowledge about processes and interactions between cancer cells and the surrounding tumor stroma microenvironment is rapidly expanding. It is of crucial Figure 1. Main trends in PDAC pathology and research that are expected to improve survival. Poor PDAC patients' prognosis is multifactorial-no sensitive and specific early diagnostic methods is one of the reasons. Another is the resistance to available therapeutic options, which is caused, among other things, by the tumor's molecular and morphological heterogeneity. Detailed pathological reporting is crucial for targeted and personalized therapy. The development of new diagnostic methods, combined with a proper pathologic evaluation and spectroscopic profiling, leads to effective treatment. Altogether, this will increase PDAC patients' survival rates. PDAC, pancreatic ductal adenocarcinoma; LIF, leukemia inhibitory factor; IL-6, interleukin-6; cfDNA, cell-free DNA; FTIR, Fourier transform infrared spectroscopy; Raman, Raman spectroscopy; SERS, surface-enhanced Raman spectroscopy; NGS, next-generation sequencing; ICB, immune checkpoint blockers; NRF2, nuclear factor-erythroid 2-related factor 2; GEM, gemcitabine; PANR, PDAC-associated neural remodeling; CAFs, cancer-associated fibroblasts; CSCs, cancer stem cells; EGFR, epithelial growth factor receptor.

Contents Summary
This text is divided into sections. In the first two parts (Sections 3 and 4), we briefly outline the pathomechanisms of PDAC development via the process described as "pancreatic intraepithelial neoplasia" (PanIN) and "intraductal pancreatic mucinous neoplasm" (IPMN) and show already proposed molecular subtype classifications. Following (Section 5), the importance of proper histopathologic evaluation is shown, starting with the description of the standardized examination protocol (LEEPP-Leeds Pathology Protocol). Main trends in PDAC pathology and research that are expected to improve survival. Poor PDAC patients' prognosis is multifactorial-no sensitive and specific early diagnostic methods is one of the reasons. Another is the resistance to available therapeutic options, which is caused, among other things, by the tumor's molecular and morphological heterogeneity. Detailed pathological reporting is crucial for targeted and personalized therapy. The development of new diagnostic methods, combined with a proper pathologic evaluation and spectroscopic profiling, leads to effective treatment. Altogether, this will increase PDAC patients' survival rates. PDAC, pancreatic ductal adenocarcinoma; LIF, leukemia inhibitory factor; IL-6, interleukin-6; cfDNA, cell-free DNA; FTIR, Fourier transform infrared spectroscopy; Raman, Raman spectroscopy; SERS, surface-enhanced Raman spectroscopy; NGS, next-generation sequencing; ICB, immune checkpoint blockers; NRF2, nuclear factor-erythroid 2-related factor 2; GEM, gemcitabine; PANR, PDAC-associated neural remodeling; CAFs, cancer-associated fibroblasts; CSCs, cancer stem cells; EGFR, epithelial growth factor receptor. central regulator of redox, metabolic, and protein homeostasis, among others, by altering glucose and glutamine metabolism and increasing glutaminolysis [53][54][55][56]. Mukhopadhyay et al. (2020) analyzed NRF2 expression in KRAS-driven pancreatic cancer tissues and cell lines. They found that a high NRF2 expression level was associated with a poor clinical outcome. Moreover, the authors showed that NRF2 regulates the sensitivity of PDAC cells to gemcitabine (a standard chemotherapeutic for PDAC), and concluded that targeting the NRF2-induced glutaminolysis by glutaminase inhibitors might sensitize PDAC cells to gemcitabine [57]. More studies are required to properly assess NRF2 expression in PDAC tissues and its influence on prognosis and gemcitabine resistance.

Molecular Subtypes
Detailed molecular characteristics of PDAC go beyond the scope of this review. Nevertheless, we will briefly outline the main trends in molecular subtyping.
In 2011 Collisson et al. described three subtypes (named classical, quasimesenchymal, and exocrine-like) and defined gene signatures for them. That study showed significant differences in patient outcomes and therapy responses between defined PDAC subtypes [58]. Moffitt et al. [59] (2015) studied PDAC gene expression in primary and metastatic tumors. They distinguished normal, tumor, and stroma-specific gene expression signatures highlighting the role of stroma in pancreatic cancer and they emphasized the need for analyzing it separately from the cancer cells. Authors suggested that the molecular characterization of the "quasimesenchymal" subtype defined in the study by Collisson's team [58], was contaminated by the stroma of the tumor and the "exocrine-like" subtype by normal pancreatic tissue [59]. A total of four PDAC subtypes were defined with stroma-specific (normal and activated) and tumor-specific subgroups (basal-like and classical). A year after the report of Moffitt et al., Bailey et al. (2016) published results from a study that revealed four molecular subtypes: squamous, pancreatic progenitor, immunogenic, and aberrantly differentiated endocrine exocrine (ADEX), and described molecular pathways that are characteristic for each type [60].
Subtypes proposed in the studies briefly described above, seem to have overlapping features. Comparison of these was a subject of multiple studies that describe, in detail, the molecular nature of each group [15,61].
Recent studies have shown that PDAC tumors differ in the expression of immune features that are associated with the response to so-called immune checkpoint blockade (ICB) therapy. Liu et al. [62] analyzed immune signature gene sets including the activation of macrophage/monocytes, overall lymphocyte infiltration, TGF-β response, IFN-γ response (IFN-γ), and wound healing activity (wound healing) in the 383 pancreatic tumor samples. This allowed distinguishing between three subtypes (named C1-C3) of PDAC tumors, which differed in terms of the survival rates of patients. The authors highlight that a more personalized strategy should be considered when designing ICB treatment in PDAC patients [62].

Histopathologic Evaluation
Current guidelines for pathology reporting include recommendations from the International Collaboration on Cancer Reporting (ICCR) [19,63]. Since it is very important to follow these in daily pathologic workup, herein we particularly highlight some of them.
Multiple studies report that a standardized examination protocol for pancreatic tumors involving the head of the pancreas reveals a high positive margin (R1) rate (above 70%) [64][65][66][67]. It is important to evaluate all relevant surfaces including the anterior and posterior pancreatic surface, the surface of the superior mesenteric vein (SMV) groove, and the superior mesenteric arterial (SMA) dissection surface. A widely used standardized protocol for pancreatoduodenectomy specimens called LEEPP (Leeds Pathology Protocol) proposed by Verbeke et al. in 2006, relies on multicolor inking of all surfaces mentioned above and serial slicing of the whole pancreatic head specimen in an axial plane, perpendicular to the duodenum [68]. Additionally, trans-section margins including duodenal, stomach, bile duct, and pancreatic neck are recorded. Before the proposed standard, positive margins (R1) for pancreatoduodenectomy specimens were reported significantly less often, between 20% and 30% [69,70], and did not reflect proper prognostic value. Currently, a minimum clearance for R1 resection is considered 1 mm for trans-section margin and SMA/SMV dissection surfaces, whereas direct breaching of the surface is required for anterior/posterior pancreatic head involvement (0 mm) [19,71,72]. Patients with an R0 resection status have a significantly better prognosis but only when assessed with the LEEPP methodology [72].
The assumption of whether the incidence of low-grade or high-grade PanIN lesions in the trans-section margins can be considered a prognostic factor was disproved by Matthaei [73]. The risk of reoccurrence in R0 resected PDAC patients is not increased in such cases [22,73].
Reporting of the tumor histological subtype according to the WHO classification of tumors of the gastrointestinal tract, 5th edition, 2019 [16], is the core element in the ICCR guidelines for pathology reporting [19]. Other histological patterns and subtypes were shown to impact the patients' outcomes, including the "large-duct pattern", which could be easily misdiagnosed with IPMN-derived adenocarcinoma [74].

Morphological Heterogeneity
Pancreatic cancer (PC) is well known to be very heterogeneous in its molecular and morphological phenotype ( Figure 2). It is one of the reasons for a poor patient prognosis, as current treatment options do not reflect the tumor heterogeneity and give insufficient results [18,75]. The mechanisms of failure are usually not well known, which compels researchers to look deeper into the molecular nature of these tumors. Numerous publications are explaining different aspects of genetic alterations in PC. There are, though, amazingly, few attempts to classify it in terms of morphological divergence yet this, from a pathological point of view, would give more practical information.
IPMNs are cystic tumors that can lead to invasive carcinomas. Generally, depending on the type of epithelial lining of the cystic structures, a gastric-type, intestinal-type, or pancreatobiliary-type are distinguished. Each has a different risk of progression into invasive PC [10,76].
Gastric-type IPMN is the most common (50-60%), with low-grade dysplastic epithelium and without MUC1 and MUC2 mucin expression on immunostaining (MUC1 negative, MUC2 negative). It progresses to invasive adenocarcinoma in 15% of cases. Intestinal-type IPMN accounts for about 20-30% of cases, 50% of them present with high-grade dysplasia, and immunostaining reveals MUC2 and CDX2 expression (MUC2 positive, CDX2 positive). The progression-to-PDAC rate reaches up to 40%. However, the vast majority progress to the invasive component in a form of colloid carcinoma, which has a slightly better prognosis. The third IPMN type, pancreatobiliary, is the rarest (10-15%), but most of them present as high-grade lesions with MUC1 expression (MUC1 positive). This type of IPMN is frequently invasive (60-70% of cases reveal invasive components) [10,16,76].
Another classification of IPMN tumors groups them by the place of origin, dividing them into "main-duct" and "branch-duct" lesions. Main-duct IPMNs present with highgrade dysplasia in 60% of cases and 45% of them are associated with invasive PC. Branchduct IPMNs are mostly low-grade (only 25% have high-grade dysplasia), and only 20% reveal invasive components [77]. In 40% of cases, tumors are found to be multicentric [78]. Muraki et al., in the single-institution study of 501 consecutive PDAC resected cases, pointed out that it is easy to misdiagnose IPMN-related carcinoma with IPMN-mimickers (pseudo-IPMN) [11]. As a pseudo-IPMN, the team described a secondary duct ectasia (retention cyst), large duct type PDAC, simple mucinous cyst, congenital cyst, paraduodenal wall cyst in grooves pancreatitis, and pseudocysts. Almost 3.8% of analyzed cases were classified as PDAC with pseudo-IPMN, compared to 6.2% with true IPMN (Table 1).
Pancreatic carcinoma may derive from an IPMN (PDAC derived from IPMN) or may develop apart from it (PDAC concomitant to IPMN). The histologic transition between Muraki et al., in the single-institution study of 501 consecutive PDAC resected cases, pointed out that it is easy to misdiagnose IPMN-related carcinoma with IPMN-mimickers (pseudo-IPMN) [11]. As a pseudo-IPMN, the team described a secondary duct ectasia (retention cyst), large duct type PDAC, simple mucinous cyst, congenital cyst, paraduodenal wall cyst in grooves pancreatitis, and pseudocysts. Almost 3.8% of analyzed cases were classified as PDAC with pseudo-IPMN, compared to 6.2% with true IPMN (Table 1).  [76][77][78] Pancreatic carcinoma may derive from an IPMN (PDAC derived from IPMN) or may develop apart from it (PDAC concomitant to IPMN). The histologic transition between IPMN and the invasive component should be revealed to determine the PDAC origin [63,80,81]. In the current WHO classification of pancreatic malignancies [16], there is no distinction between the two, and both should be reported as an "IPMN with associated carcinoma". Although the different origin of IPMN concomitant carcinomas suggests a less favorable outcome, similar to that of conventional PDAC (cPDAC), in the work by Yamaguchi et al., both (derived from and concomitant) had similar, significantly favorable biological behavior [80]. The authors pointed out that this might be due to an earlier diagnosis of such cystic lesions.
Colloid carcinoma of the pancreas is characterized by the presence of large extracellular mucin pools (in at least 80% of the mass of the neoplasm) containing suspended neoplastic cells [16]. In the vast majority of cases, it is derived from intestinal-type IPMN. Multiple studies show that colloid IPMC had significantly better outcomes than tubular IPMC or cPDAC.
Another type described by the WHO classification [16] is a medullary carcinoma. A rare tumor of the pancreas that is often associated with microsatellite instability (MSI)/ defective DNA mismatch repair (dMMR). Luchini and his team, in a meta-analysis of 34 studies, showed that the incidence rate of MSI/dMMR in PDAC is very low, ranging from 1-2%. It was significantly associated with medullary and colloid histological subtypes. Consequently, they suggested that cases of PDAC with medullary or colloid histology should routinely be examined in terms of MSI/dMMR, by use of immunohistochemistry [82].
Not many studies attempted to subtype a large group of "ductal adenocarcinoma, NOS", concerning histological and immunohistochemical features. Kalimuthu et al. (2020) distinguished the morphological patterns of PDAC by separating two groups depending on gland formation ("gland forming" and "non-gland forming") and correlated them with earlier described molecular subtypes ( Figure 3) [17]. The four morphological patterns included conventional, tubulo-papillary, squamous, and composite. Nevertheless, this study did not explain PDAC morphological variability. A more comprehensive approach was demonstrated in a study by Sántha et al. [18]. From 233 foci selected from 39 pancreatic ductal adenocarcinoma specimens, the team analyzed 26 features including morphological and immunohistochemical patterns. Four common subtypes (67% of cases in the studied series) with significant differences in the areas of cancer cell proliferation (Ki67) and migration (collagen fiber alignment, metalloproteinases-MMP14), cancer stem cells (CD44, CD133, ALDH1), extracellular matrix (total collagen, collagen I and III, fibronectin, hyaluronan), cancer-associated fibroblasts (αSMA), and cancer-stroma interactions (integrins α2, α5, α1; caveolin-1) were distinguished. The patterns described as periglandular (PP), tendon-like (TP), fascicular (FP), and chicken wire (CP) were assessable by standard hematoxylin and eosin staining (H&E) and characterized by distinct features including the cancer cells, gland formation, and stromal compartment ( Figure 3). There were significant differences between the subgroups including most of the features. What is important is that the study showed, among other things, heterogeneity in the stroma compartment composition that may affect different aspects of tumor growth, invasion potential, and resistance to therapy. The authors suggest that proper subtyping of pancreatic ductal adenocarcinoma may reveal these with better clinical outcomes and, secondly, allow the selection of subtypes that could benefit from new treatment options. It is important to give pathologists new ways of subgrouping "PDAC NOS" tumors that will be available for routine pathologic reporting and that will bring relevant and meaningful information to clinicians and patients.
Clear cell carcinoma of the pancreas is not a well-known entity and can be diagnostically misleading [83,84]. Kim et al., in the study of 84 pancreatic cancer specimens, reported that 24% of the analyzed cases contained significant clear cell components, and 14% of the studied cohort was defined as having clear cell carcinoma with over 75% of the tumor volume showing clear cell features [85]. In differentiating clear-cell-appearing tumors in the pancreas, other entities have to be taken into consideration, such as metastatic clear cell renal cell carcinomas, ovarian and adrenal carcinomas, and primary clear cell neuroendocrine tumors of the pancreas (frequently occurring in patients with von Hippel-Lindau disease) [86,87]. In the differential diagnosis, various histopathological stainings come in handy, including immunohistochemistry for carbonic anhydrase IX, HMB45, vimentin, PAX8, CD10, synaptophysin, or chromogranin [83]. Particularly useful for diagnosing clear cell pancreatic carcinoma of ductal origin (exocrine) is hepatocyte nuclear factor-1β (HNF1B), which significantly shows a stronger positivity more frequently in clear cell components compared to conventional ductal adenocarcinomas [85].
One more pattern called a "foamy gland pattern", shows some similarities with clear cell carcinomas [88,89]. It has a benign-appearing look with well-formed glands and subtle infiltration. There is no data on whether these two entities are of a similar origin or follow the same molecular pathways. To the best of our knowledge, to date, no studies have compared clear cell and foamy gland patterns.
for MUC6 [74,91,92], which might be helpful to differentiate it from gastric-type IPMN. The term "cystic papillary pattern" is sometimes distinguished from a large duct pattern, describing more complex or papillary structures [92]. However, some authors suggest that both represent the same entity, reflecting distinct evolutionary stages [90]. Interestingly the large duct pattern of PDAC might fit into the subgroup distinguished by Kalimuthu et al. called "tubulo-papillary", rather than the "gland forming" group [17].

Immunostaining
Currently, there is no immunostaining marker that would be recommended for the routine pathological diagnostic workup of PDAC. Nevertheless, there have been multiple reports suggesting that p53 [93][94][95] and insulin-like growth factor-II mRNA-binding protein 3 (IMP3/IGF2BP3/KOC) [96][97][98][99][100][101][102][103] could significantly help to avoid misdiagnosis. Overexpression of these markers correlates with patient prognosis. Liu et al. reported 90% of PDAC cases to be positive for IMP3, Maspin, and S100 calcium-binding protein P (S100P) expression [104]. Another study utilized IMP3, Maspin, S100P, and von-Hippel-Lindau gene protein (pVHL) for comparison between autoimmune pancreatitis (AIP), PDAC, and normal pancreas specimens [105]. The authors showed that although weak and focal expression was seen in AIP and a normal pancreas, PDAC was characterized by strong overexpression of Maspin, IMP3, and S100P in 95%, 75%, 75%, respectively, whereas weak and no expression was seen in 0%, 25%, 10% of PDAC cases, respectively. Furthermore, 100% of studied PDAC samples were negative for pVHL. Recently, Senoo et al. [106] and Mikata et al. [107] conducted retrospective and prospective studies evaluating the usefulness of p53 and IMP3 in endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) specimens of pancreatic tumors. They reported that none (0%) of the benign lesions analyzed expressed IMP3 nor p53, whereas, for malignant lesions, IMP3 Another, non-WHO pattern of PDAC called "large duct pattern" (also called "cystic papillary pattern"), can be misdiagnosed as an IPMN [74]. In this pattern, carcinomatous ducts are enlarged over 0.5 mm. If at least 50% of the tumor glands show such dilatation, the name "large duct carcinoma" is used. In some cases, an elastic fiber staining (i.e., orcein stain) helps to distinguish between non-malignant (ducts dilated due to occlusion) and malignant ducts. Large duct carcinoma is frequently accompanied by perineural invasion (88%) [90]. There are no mucin pools and signet ring cells, in contrast to colloid carcinoma. A total of 73% of large duct PDACs stain with MUC1 positively and over 55% for MUC6 [74,91,92], which might be helpful to differentiate it from gastric-type IPMN. The term "cystic papillary pattern" is sometimes distinguished from a large duct pattern, describing more complex or papillary structures [92]. However, some authors suggest that both represent the same entity, reflecting distinct evolutionary stages [90]. Interestingly the large duct pattern of PDAC might fit into the subgroup distinguished by Kalimuthu et al. called "tubulo-papillary", rather than the "gland forming" group [17].
Of note, as mentioned above, hepatocyte nuclear factor-1β (HNF1B) shows a strong positive expression in clear cell pattern ductal adenocarcinoma of the pancreas [85].
CDX2 is known to be a driver for gastric-to-intestinal type progression in IPMN [47]. In 95% of analyzed intestinal-type IPMN cases, CDX2 showed strong positive nuclear staining in more than 90% of cells. Xiao et al. reported a loss of CDX2 expression during PanIN progression from low-grade to high-grade lesions, and only one-third of PDAC samples showed weak CDX2 expression [108]. It is believed that colloid carcinoma of the pancreas develops almost exclusively through the intestinal-type IPMN progression pathway [10]. Further research would be required to assess the prevalence of CDX2 expression strictly in pancreatic colloid carcinoma.
Leukemia inhibitory factor (LIF) and interleukin-6 (IL-6) are potentially promising biomarkers for early diagnosis of PDAC, that can be detected in patients' serum [12,13]. Aside from the LIF serum levels, a raised LIF concentration was observed in pancreatic cancer tissue samples compared to chronic pancreatitis or benign lesions (PanIN) and inversely correlated with the tumor differentiation level [13,[111][112][113].

Cancer-Stroma Interactions
PDAC invasion is characterized by an extensive, dense, desmoplastic stroma that is not only a silent actor but plays a crucial role in the tumor growth, maintenance, invasion, metastatic potential, and chemoresistance [130][131][132][133][134][135][136][137][138][139]. The complicated relationship between cancer cells and the stroma was metaphorically, yet very vividly envisaged by Adamek and Stoj (2014). They have proposed a figurative concept of cancer as a form of the "mafia" within the body, in which the cancer cells "corrupt" non-neoplastic cells and, as a result, aid and abet them in "the crime of cancer". What is more, the "criminal cells" may even cunningly change their properties and mislead researchers, altering the study results and efficacy of treatment [140]. This holistic approach to cancer biology explains some difficulties in the cancer-stroma interplay research.
The explanation of the complex interactions between pancreatic cancer cells and the stroma compartment will undoubtedly be a milestone in the development of PDAC therapy strategies. Currently, multiple trials are investigating possible options in targeting stromal compartment mechanisms or cancer-to-stroma interaction pathways with varied preliminary reports [9,136,[141][142][143][144][145][146]. There are strong clinical implications of the presence of different stromal compartment composition features. Some authors suggest that reporting them should be a part of the routine pathological workup [18].
The research on PDAC stromal nature, particularly the cancer-associated fibroblasts (CAFs) that are the main component of the tumor microenvironment (TME), faces many difficulties, caused by, among other things, inconsistencies in key definitions. In 2019, a Banbury Center meeting of international researchers and clinical scientists gathered in New York (USA) and agreed on a consensus statement where they summarized good practice advice and described the recommended methodology for CAF research [147].
The catabolic mechanisms of CAFs modulate cancer cell metabolism and fuel cancer, with energy sources like amino acids and other nutrients, sustaining the tumor growth [172][173][174][175] in a process with the suggested name "reversed Warburg effect" [176][177][178]. From one side, CAFs produce a collagen-rich ECM, from the other, cancer cells stimulate CAF autophagy that "produces" a substantial dose of alanine [179].
Multiple studies recognized CAFs within the pancreatic cancer stroma to be a heterogeneous population of cells that present divergent phenotypes and seem to have different roles. Three main groups are myofibroblastic (myCAFs), inflammatory (iCAFs) and antigenpresenting CAFs (apCAFs) [150,[180][181][182][183]. MyCAFs are located in the periglandular region of the cancer site and express high levels of α-SMA and low levels of IL-6. Activation (PSCs to myCAFs) is believed to be rendered via the TGFβ/SMAD signaling pathway. More distally to the glands, iCAFs are found. They express high levels of IL-6 (low α-SMA), due to IL-1/LIF/JAK/STAT pathway activation [135]. The third subtype, called antigenpresenting CAFs, shows high expression of MHC II family genes but their activation mechanisms and specific features are yet to be determined [180].
Classic biomarkers used for CAF detection include α-SMA, fibroblast activation protein (FAP), fibroblast-specific protein 1 (FSP1/S100A4), platelet-derived growth factor receptors (PDGFRα, PDGFRβ), or podoplanin (PDPN/gp38). However, some studies recognize the heterogeneity of CAF expression profiles [184,185]. The Banbury Center meeting [147] highlighted the importance of determining specific subtypes of CAFs in the tumor stroma with the use of new immunohistochemistry methods but, currently, there is not enough evidence to state a consensus in this field.
Stroma composition was a part of the molecular subtyping by Moffitt et al. [59]. Authors distinguished stroma-specific from the tumor-specific gene expression and revealed two molecular subtypes regarding the stroma compartment, namely "activated" and "normal". Stroma evaluation played a substantial part in the morphological subtyping of PDAC specimens by Sántha et al. [18]. They studied such components as an extracellular matrix with collagen composition and arrangement, fibronectin and hyaluronan deposition, cancer-associated fibroblasts (α-SMA staining and collagen density with the use of the so-called activated stroma ratio [114,115]), and cancer-stroma interactions (integrins α2, α5, β1, metalloproteinases-MMP14 and caveolin-1). The biological role of all of these features was documented substantially in publications prior to that study. Proposed morphologi-cal subtypes that presented with significantly different phenotypes regarding the above, showed the importance of stroma compartment evaluation in pathological practice.
Cancer stem cells (CSCs) represent a small number of cancer cells within the tumor that define its potential to grow and propagate. CSCs have stem cell properties and tend to be self-renewable, and multipotent. Some studies showed that CSCs might be responsible for tumor initiation, rapid growth, resistance to therapy, recurrence, and metastases [186][187][188][189]. The induction of CSCs is related to the epithelial-to-mesenchymal transition (EMT) regions [190][191][192][193]. In pancreatic cancer, CSCs express CD24, CD44, CD133, aldehyde dehydrogenase 1 (ALDH1), and epithelial-specific antigen (ESA) [186,188,193]. Pancreatic stromal stellate cells (PSCs) cooperate in a paracrine manner with the CSCs to increase their invasiveness and self-renewal properties via the Nodal/Activin signaling pathways [194,195]. For some authors though, the idea of CSCs is controversial [196].
Detection of CSCs in pathology reporting might be of importance due to potential therapy options [195] and correlation with clinical prognosis [189,193]. Sántha et al. [18] studied the expression of CD44, CD133, and ALDH1 in pancreatic cancer cells as part of a morphological subtyping scheme and found significant differences in marker expression between proposed subtypes that correlated with other morphological and immunohistochemical features.

Prognosis
Summarized information about histomorphologic features of PDAC is presented in Table 3. The overall 5-year survival rate for pancreatic ductal adenocarcinoma is less the 10% [1]. The majority of patients present with inoperable and non-curable tumors. The survival rate, among other aspects, depends on the stage of the tumor at the time of diagnosis. A total of 10% of patients had T1-T2 disease with a 5-year survival rate reaching 32%, while the rate dropped to 12% for T3 tumors. More than half of the patients had T4 stage tumors with a 5-year survival rate of 3% [1].
The prognosis for patients with cancers derived from intraductal papillary mucinous neoplasm (IPMC), compared to conventional PDAC (cPDAC), has been in debate for a long time, sometimes with conflicting results. It is important when analyzing those results to bear in mind that often the distinction between "derived from" vs. "concomitant to" was not properly addressed [197][198][199][200][201]. Poultsides et al. reported significantly better survival only for tumors that did not reveal known adverse prognostic factors, such as poor differentiation, involved surgical margins (R1), or vascular (LV1), or perineural invasion (PNI) [202]. Okabayashi et al. [203] published results showing that invasive carcinomas derived from branch-duct IPMNs might be more aggressive with poorer patient outcomes compared to those derived from main-duct IPMNs. Notwithstanding, the risk of malignant transformation in main-duct IPMN is higher than in branch-duct IPMN. Additionally, SMAD4 and TGFβ expression was significantly increased in the carcinomas derived from branch-duct IPMNs. The question regarding a better prognosis of IPMC vs. cPDAC, regardless of the stage of the tumor, remains unanswered. What is certain is that IPMC patients are diagnosed earlier, which gives them a better start.
The patient prognosis in large duct carcinoma seems to be similar to that in tubular carcinoma or slightly better, probably because of good differentiation occurs more frequently [74,90,92]. Nevertheless, it is important to distinguish it from other cystic lesions, like IPMNs.
Surprisingly, unlike in other locations (colorectal, gastric, duodenal, or ampullary cancers) [207], medullary carcinoma of the pancreas is not correlated with a better prognosis [82], but there is a very small amount of data available. The KEYNOTE-158 clinical trial, including patients with MSI/dMMR PDAC tumors, showed objective responses in only four out of 22 patients (one complete, three partial) included in the trial [208].

Lymphatic and venous invasion
• both should be reported separately because they represent different biological processes-lymph node metastasis and distant, blood-borne spread [19] • considered as a non-core element due to possible difficulties in distinction; elastin staining might be helpful [19] Other features Lymph node status • the mechanism of lymph node involvement should be recorded, as direct or metastatic [ Kalimuthu et al. suggested that morphological classification is a better prognostic factor than the standard three-tiered grading system, as most of the studied cases were classified as moderately differentiated. They grouped specimens into two groups, defined as group A-which showed less than 40% "non-gland forming" components, and group B-with more than 40%. Group A had significantly better overall survival than group B [17].

Immunohistochemical Prognostic Factors
Loss of p16 (CDKN2A) expression in cancer cells was significantly associated with lymphovascular invasion and metastatic disease [95,128].
The prognostic value of SMAD4 aberrations is ambiguous. Some reports associate the loss of SMAD4 with poor survival and early metastasis [95,122,123], but some did not achieve similar results with significant relevance [124][125][126][127].
Song et al., in a study of 62 PDAC cases, showed that Janus kinase 2 (JAK2) immunostaining is an independent poor survival factor [121] though, there is a limited amount of data regarding this pathway (JAK/STAT) biomarker expression. There are some new potential treatment options and patients might benefit from JAK/STAT biomarker reporting in the future [219][220][221][222].
Some studies have shown that podoplanin-positive CAFs in the PDAC stroma compartment were associated with poor prognosis, aggressive behavior, and larger tumor size [118][119][120].
One report suggested, that Meflin-positive CAFs in PDAC stroma prevents poor differentiation of the tumor and are markers of a favorable outcome [117]. Another study by Ikenaga and colleagues showed that CD10-positive CAFs were associated with positive lymph node metastasis and a shorter survival time [116]. The ratio of the α-SMAstained area to the collagen-stained area was defined as the activated stroma index (ASI). Erkan et al. [114] differentiated fibrolytic (high α-SMA/low collagen), fibrogenic (low α-SMA/high collagen), inert (high α-SMA/high collagen), and dormant (low α-SMA/low collagen) pancreatic stroma composition. Significant differences in patient outcomes and progression-free survival between these composition types have been reported [114,115].
Last, but not least, a strong expression of hepatocyte nuclear factor-1B (HNF1B) in PDAC cases was correlated with a worse prognosis regardless of the morphological features [85]. See Table 2 for summarized info about immunohistological features of PDAC with prognostic relevance.

Perineural Invasion
Pancreatic cancer is characterized by early and extensive perineural invasion (PNI). Studies have shown that PNI in PDAC is an independent poor prognostic and an early recurrence factor [215][216][217][218]. It is considered a core element in pathology reporting [19]. Recent studies highlight the active role that nerves play to facilitate tumor spread [223][224][225][226][227][228][229]. Nervous cells interplay with cancer cells and the stromal compartment cells (CAFs, PSCs, and tumor-associated macrophages-TAMs) [230]. PDAC-associated neural remodeling (PANR) is a proposed term describing the alterations in the nerve compartment caused and facilitated by PDAC tumors, and resulting in higher nerve densities in PDAC due to peripheral nerve fiber infiltration and axonogenesis [112]. Bressy with colleagues [112] showed that PANR was supported by leukemia inhibitory factor (LIF). LIF influences Schwann cells and dorsal root ganglia (DRG) neurons via modulation of the JAK/STAT signaling pathways and facilitates their migration and differentiation. They suggested that the use of LIF-inhibitors might suppress PANR, limit tumor spread, and increase patients' quality of life [112].
More studies are required to comprehensively assess the PNI prognostic influence and to explain the divergence of PNI levels among PDAC tumors.

Lymph Node Metastasis
Regional lymph node status is a well-known poor prognostic factor in PDAC and it is considered a core element in the ICCR guidelines for pathology reporting [19]. However, the 8th edition of the TNM classification of the American Joint Committee on Cancer (AJCC) does not distinguish the mechanism of lymph node invasion (LNI). In 2015 Williams et al. [214] examined PDAC specimens regarding the mechanism of LNI (distinguishing true "metastatic" spread and "direct" LNI- Figure 4) and compared patients' survival rates. They concluded that "direct"-only LNI patients had a similar overall survival to those with node-negative disease. Recently, other authors reported similar results [211][212][213].
Cancers 2022, 14, x FOR PEER REVIEW 17 of 33 However, the 8th edition of the TNM classification of the American Joint Committee on Cancer (AJCC) does not distinguish the mechanism of lymph node invasion (LNI). In 2015 Williams et al. [214] examined PDAC specimens regarding the mechanism of LNI (distinguishing true "metastatic" spread and "direct" LNI- Figure 4) and compared patients' survival rates. They concluded that "direct"-only LNI patients had a similar overall survival to those with node-negative disease. Recently, other authors reported similar results [211][212][213].

Early Diagnostic Options
There are no efficient early diagnosis tools for pancreatic carcinoma. Late-stage disease at diagnosis is certainly a major issue that partially leads to overall poor survival rates.
Serum carbohydrate antigen (Ca19-9) is used in the diagnostic work-up of patients being diagnosed with pancreatic tumors, but it is neither specific nor sensitive for malignant lesions [231,232]. Some studies showed better usefulness of assessing IL-6 serum levels in differentiating PDAC patients from chronic or acute pancreatitis [233][234][235]. Recently, LIF was reported to be a promising serum biomarker of pancreatic malignancy [112], as much as a metastatic disease predictor for PDAC patients [236] and a therapy response monitor [13,111]. Moreover, LIF was shown to be a good biomarker for immune checkpoint blocker (ICB) therapy efficacy, which is a novel immunotherapeutic option for patients with solid tumors [237]. Loriot et al. identified elevated LIF serum levels as a poor prognostic factor for ICB-receiving patients [238].
The development of new diagnostic technologies might shed new light on the PDAC early diagnostic field. Raman spectroscopy (RS) and surface-enhanced Raman and more than an upper two-thirds (B) are occupied by solid PDAC growth, which directly invades the remnants of a lymph node (asterisks). This form of NI in PDAC has a different prognostic significance compared to a true metastatic spread (arrows in (C,D)). See the text (Section 12) for further details on the topic. (H&E stain, original magnification ×4).

Early Diagnostic Options
There are no efficient early diagnosis tools for pancreatic carcinoma. Late-stage disease at diagnosis is certainly a major issue that partially leads to overall poor survival rates.
Serum carbohydrate antigen (Ca19-9) is used in the diagnostic work-up of patients being diagnosed with pancreatic tumors, but it is neither specific nor sensitive for malignant lesions [231,232]. Some studies showed better usefulness of assessing IL-6 serum levels in differentiating PDAC patients from chronic or acute pancreatitis [233][234][235]. Recently, LIF was reported to be a promising serum biomarker of pancreatic malignancy [112], as much as a metastatic disease predictor for PDAC patients [236] and a therapy response monitor [13,111]. Moreover, LIF was shown to be a good biomarker for immune checkpoint blocker (ICB) therapy efficacy, which is a novel immunotherapeutic option for patients with solid tumors [237]. Loriot et al. identified elevated LIF serum levels as a poor prognostic factor for ICB-receiving patients [238].
The development of new diagnostic technologies might shed new light on the PDAC early diagnostic field. Raman spectroscopy (RS) and surface-enhanced Raman spectroscopy (SERS) [239,240] methods were used for detecting earlier untraceable amounts of biomarkers in PDAC patients' serum [241,242]. SERS with a plasmonic gold nanohole array was used for the detection of DNA methylation aberrations [243]. These innovative methods might increase the sensitivity of aberrant methylation marker detection in the circulating cell-free DNA (cfDNA) and circulating tumor cells (CTC) in PDAC patients [244,245] ( Table 4).

Molecular Characteristics of Malignant Pancreatic Tissues
A growing number of scientific articles confirm an important role of molecular spectroscopy in the characteristics of the chemical structure and composition of various malignant tissues [246][247][248][249] Due to high chemical selectivity, both Raman and infrared spectroscopies can become efficient tools supporting the molecular screening of pancreatic tissue sections. This methodology provides information about the content of various biologically significant molecules and functional groups, including phospholipids and triglycerides, proteins, nucleic acids, phosphates, and carbohydrates. The results indicate differences in the metabolic pathways typical for various neoplasms. The main advantage of the molecular spectroscopic approach is achieving information about samples in a label-free and noninvasive manner. The research potential of spectroscopic methods has not yet been fully explored in the investigation of pancreatic cancer ( Figure 5) [250,251].
Tissues are complex systems and to achieve a complete overview of their molecular structure, hyperspectral mapping, which provides full spectral information from each pixel, is applied. To reduce data dimensionality and extract the most important information-marker bands of molecular pathologies, from the acquired data, various methods of multivariate data analysis are used [252]. K-means clustering (KMC), and principal component analysis (PCA) are commonly performed to explore spectral variation in maps acquired from tissue sections [253][254][255].
Molecular spectroscopy coupled with multivariate data analysis support standard immunohistochemical and histological staining-based procedures. The comprehensive approach may increase the effectiveness of proper diagnoses of pancreatic and ampullary cancer and their subtyping [251]. Cancers 2022, 14, x FOR PEER REVIEW 19 of 33 Figure 5. Spectroscopic mapping of ampullary adenocarcinoma: A hematoxylin-eosin slide image of ampullary cancer tissue with superimposed FTIR and Raman hyperspectral maps treated with hierarchical cluster analysis, and averaged spectra with corresponding second derivatives from each cluster; spectroscopic maps cover both cancerous (red circle) and noncancerous-stroma (green circle) tissue fragments. Reproduced with permission from Szymonski et al., Clinical Spectroscopy, published by Elsevier B.V, 2021 [251]. FTIR, Fourier transform infrared spectroscopy.

Conclusions
The field of pancreatic cancer research is rapidly evolving. Reports of newly revealed pathomechanisms of the nature of these tumors are published daily. Nevertheless, many aspects of pathologic evaluation are still uncertain (Table 5). Although it is sometimes hard to stay on top of things with such a dynamically increasing amount of knowledge, it is crucial to be able to pull out practical information that impacts the diagnostic process, called a pathologic evaluation. Hopefully soon, we will witness a great change in PDAC patients' prognosis, whether through the development of new early screening methods or new therapeutic options. In the meantime, pathologists should do whatever they can to make the pathologic reporting meaningful for clinical and research purposes. Table 5. Future research proposals in PDAC pathology. HNF1B, hepatocyte nuclear factor-1β; PANR, PDAC-associated neural remodeling; PNI, perineural invasion; AVAC, ampulla Vater adenocarcinoma; IHC, immunohistochemistry; NRF2, nuclear factor-erythroid 2-related factor 2.

Topic Aims
Foamy gland vs. clear cell patterns