The bland squamous cells of well-differentiated squamous cells demonstrate dyskeratosis characterized by early keratinization (keratin pearls in the basal and parabasal layers) and excessive cytoplasmic keratinization. This important feature has been largely overlooked by the surgical pathologists since it is shared by the perfect mimicker of squamous cell carcinoma and pseudoepitheliomatous hyperplasia. A careful survey of the adjacent tissue in the latter, however, usually reveals an underlying infection, inflammation, trauma, or nonsquamous neoplasm [1, 2].

With the exception of the exceedingly rare variant of esophageal verrucous squamous cell carcinoma, squamous cell carcinomas typically show an infiltrative pattern. The infiltrative fronts are usually narrow based with no surface connection. Pseudoepitheliomatous hyperplasia poses a diagnostic problem as in the skin and manifests an infiltrative front as a result of benign epithelial–mesenchymal transition induced by a predisposing lesion [1, 2]. Disruption of the basement integrity results from decreased production of laminin and type IV collagen rather than increased activities of the MMPs. The infiltrative front, however, has a jagged contour with sharp tips which are connected to a broad base. Usually, a connection of the elongated thick invasive projections to the surface can be traced (Fig. 7.2).

Caveat: The rare variant of esophageal verrucous squamous cell carcinoma has the characteristic morphological features as in the skin. It is composed of exophytic and endophytic components which possess different cytological and architectural changes. The cells have low proliferation activity which is restricted to the basal layer.

Esophageal adenocarcinomas are believed to follow the intestinal metaplasia–dysplasia–carcinoma pathway. The metaplastic glandular epithelium is lined by a layer of periglandular myofibroblastic cells which regulate many aspects of the epithelial cells like in the intestine (refer to the intestine section for more detailed discussion) [4]. Due to the presence of the periglandular myofibroblastic sheath, benign intestinal epithelial proliferations have a characteristic serrated appearance. Except for the rare serrated adenocarcinomas in the colon which possess other easily discernible characteristic features, genuine serration is rarely seen in gastrointestinal epithelial malignancies. This important feature can be very useful in the diagnosis of well-differentiated carcinomas where invasion is insidious and desmoplasia is not evident.

Corresponding to the lack of a serrated appearance in epithelial malignancies is the progressively decreasing presence of periglandular myofibroblastic cells along the spectrum of intestinal hyperplasia to dysplasia and adenocarcinoma [5, 6]. In the evaluation of periglandular myofibroblasts, the pathologist should be aware of a common phenomenon in Barrett’s esophagus: thickening of the muscularis mucosa and even the muscularis propria [7–9]. The hypertrophic muscle fibers can extend into the lamina propria or even form a new layer of muscular mucosa (duplication of muscularis mucosa). The upward extending muscle fibers, however, do not go around the glands in an accommodating fashion as the periglandular myofibroblasts. To help light up the periglandular myofibroblasts, immunostaining for SMA can be used. The other related issue is to avoid making an erroneous diagnosis of muscle invading adenocarcinoma when the hypertrophic muscle fibers are adjacent to benign or malignant glands in the lamina propria. Attention to the location of the hypertrophic muscle fibers is useful in this regard.

Most gastric adenocarcinomas resemble their esophageal counterparts in that they follow the same intestinal metaplasia to dysplasia to carcinoma pathway. Metaplastic glands also have a layer of periglandular myofibroblasts, and a similar progressively decreasing existence pattern for them exists along the metaplasia–dysplasia–carcinoma spectrum as in the intestine [13] (Figs. 7.5 and 7.6).

The fact that gastric foveolae, hyperplastic polyps, and adenomas share striking morphological similarities with their intestinal equivalents suggests that a similar type of periglandular myofibroblasts should be present around normal nonmetaplastic gastric glands. Our unpublished data indicate that this is the case even though it appears that the lower portions of glands have an attenuated layer of myofibroblast which surrounds several rather than individual glandular lumens. The rapidly proliferative upper portions have a sheath of periglandular myofibroblasts. This layer of myofibroblasts is supposed to play a vital role in the normal glandular morphogenesis and maintenance like their intestinal counterparts. Presumably, their presence in the upper portions of glands contributes to the foveolar appearance of normal and hyperplastic mucosa. This might explain why some benign gastric epithelial lesions such as oxyntic gland polyp/adenomas and pyloric gland adenomas do not have or have only a focal foveolar appearance. Fortunately, these two rare entities can be differentiated from well-differentiated adenocarcinomas by their overall lobular pattern and other cytological features.

Judicious use of the two morphological features is particularly useful in the distinction of well-differentiated gastric adenocarcinomas from most benign proliferations in small biopsy specimens where invasion is not obvious. In the evaluation of periglandular myofibroblasts, a strict criterion should be applied as in the esophagus and intestine. The spindle cells should closely embrace the contours of the corresponding glands to be considered periglandular myofibroblasts. This is to avoid mistaking tumor stromal myofibroblasts for periglandular myofibroblasts since the former can get close to the malignant glands. The tumor stromal cells, however, do not accommodate their arrangement to the shape of the glands and they lack a connection to the muscularis mucosa. Like in the Barrett’s esophagus, many benign gastric polyps and reactive lesions have a characteristic smooth muscle component in their lamina propria. These muscle fibers can be traced down to the muscularis mucosa [14]. They might get close to the benign glands; however, they do not show the characteristic embracing pattern. Instead, a perpendicular relationship to the mucosa can be appreciated. In the case of gastritis cystica polyposa/funda, the muscle fibers can be traced up to the muscularis mucosa, and due to their entrapped nature, they can become disorganized. Pancreatic heterotopia also has hypertrophic, disorganized smooth muscle fibers emanating from the muscularis mucosa. The disorganized bundles often mix with pancreatic acini and duct. The importance of appreciating the smooth muscle components and their presenting features lies mainly in avoiding a diagnosis of muscle invading adenocarcinoma especially when reactive changes are present in the epithelial component.

The so-called gastric-type well-differentiated adenocarcinomas have been slowly recognized, particularly in Europe and Japan [15]. They consist of cuboidal to columnar cells with mucinous cytoplasm. Characteristically, papillary and villous projections are prominent in the upper portion and irregularly branching and fusing glands are present in the middle and bottom portions of the tumor. This branching and fusion pattern has been interpreted as stromal invasion. In addition to the unique architectural feature, the tumor cells lack surface maturation and show cytological atypia (small nuclei with nucleoli, lack of apical mucinous cap).

A very small fraction of gastric adenocarcinomas are extremely well differentiated [16–18]. The gastric phenotype is easily confused with hyperplastic polyps or even normal gastric mucosa, whereas the intestinal phenotype is more likely mistaken for intestinal metaplasia. The presence of a foveolar appearance in some of the glands indicates that there probably exists at least partially operative interaction between the malignant epithelium and periglandular myofibroblasts. The tumor cells are generally benign looking, slowly proliferative, and p53 negative. However, the malignant nuclei are obviously larger and more hyperchromatic at least focally when comparison with the adjacent normal gastric mucosal cells is made. The only definitive sign of their malignancy is deep location and/or metastasis (Fig. 7.7). In the exercise of identifying deep invasion, we propose that attention be paid to the relationship of the beguilingly benign glands with the adjacent smooth muscle fibers as benign gastric polypoid lesions often contain prominent hypertrophic smooth muscle fibers. The invasive glands should be splitting the horizontally arranged smooth muscle fibers. An emanating relationship with the muscularis mucosa is nonexistent since the invaded fibers are part of the muscularis propria.

Well-differentiated colorectal adenocarcinomas are characterized by predominantly well-formed glands with smooth and regular lumens. Except for the rare entity of serrated adenocarcinoma which has its characteristic cytological and architectural features, genuine serration is lacking in colorectal adenocarcinomas. Serrations are not to be confused with papillations or pseudo-serrations in necrotic complex tubular or cribriform structures. They should contain epithelial tufts composed of only epithelium or epithelium and basement membrane material [24].

The serrated appearance of benign colorectal proliferations is believed to result from a decreased apoptosis/proliferation ratio and delayed cell migration which still follows the normal bottom to surface route [25–27]. Whereas the pericryptal fibroblasts might very well provide some trophic factors for the epithelial growth, their physical presence along with the basement membrane as a component of a complex framework constitutes a powerful restraining force in preventing a full-blown expansion of the crypts and stratification.

The pericryptal fibroblastic sheath in the intestinal tract is a unique syncytium of specialized fibroblasts which extend laterally into the lamina propria to connect with other fibroblasts and downward to associate with the muscularis mucosa [6, 28–32]. Through gap and tight adherence, they are even connected to the overlying epithelial cells. Thus, formed complex network provides the framework for the normal cryptal formation, maintenance, and function. In fact, not only are they a very important component of the stem cell niche located at the bottom of the crypts; the pericryptal fibroblasts also play an essential role in the proliferation, maturation, and upward migration of the epithelial cells. Furthermore, this epithelial–mesenchymal (pericryptal fibroblast) interaction appears to be bidirectional [33]. An excellent example in highlighting this interactive nature is the so-called perineuroma-like proliferation in which the pericryptal fibroblasts are believed to undergo perineural differentiation and demonstrate their characteristic pericryptal whirling growth pattern. Some focal perineuroma-like proliferations are associated with sessile serrated adenomas.

The pericryptal fibroblasts are likely to play an important role in intestinal epithelial tumorigenesis. For instance, pericryptal fibroblasts of colorectal adenomas express high concentrations of cyclooxygenase (COX-2) [34]. Disturbance in their cell adhesion molecule expression has been associated with concurrent neoplasia in patients with ulcerative colitis [35]. More importantly, corresponding to the colorectal hyperplasia–dysplasia–carcinoma spectrum is the progressively decreasing presence of the pericryptal fibroblasts with absence or rarity of pericryptal fibroblast around carcinomatous glands [28, 36, 37]. While it is still premature to conjecture how changes in the pericryptal fibroblasts contribute to epithelial tumorigenesis, the following statement holds true: this is total disruption of a healthy, harmonious nature of the epithelial–mesenchymal interaction with resultant absence or rarity of pericryptal fibroblasts around malignant glands. The diagnostic importance of this observation is analogous to that of the absence of basal cells in prostatic adenocarcinomas and lack of myoepithelial cells in mammary malignancies. It would prove to be very helpful in surgical pathology sign-outs (particularly small biopsies) when other signs of malignancy are not salient.

The nature of dysplasia in serrated lesion differs significantly from that of the conventional adenomas in that apoptosis is largely inhibited and the increased proliferative activity is still restricted to the lower portion of the crypt [25, 26]. Presumably, the epithelial–mesenchymal interaction in serrated adenomas and even serrated adenocarcinomas is better preserved than in their conventional counterparts. The serrated neoplastic cells have a low nuclear/cytoplasmic ratio with preservation of cell polarity and abundant eosinophilic cytoplasm, and they still retain their cell-to-cell adhesion with stratification restricted largely to the tufted areas [25, 26, 38]. This is in contrast to the conventional adenomas where proliferation activity is no longer restricted to the crypt bottom. Instead, they manifest a diffuse, top heavy proliferation pattern. We speculated that a change (disturbance of cell adhesion molecules) similar to that of ulcerative colitis associated pericryptal fibroblasts might happen here, allowing for epithelial stratification and loss of polarity.

It is believed that serrated adenocarcinoma occurs as a result of loss of inhibition on the apoptotic pathways [25–27, 39]. Published diagnostic criteria for differentiating it from non-serrated adenocarcinomas include serrated morphology, mucinous differentiation, eosinophilic cytoplasm, vesicular nuclei, and absence of dirty necrosis. In well-differentiated serrated adenocarcinomas, genuine serration with well-polarized cells is evident. However, information is lacking on the status of periglandular fibroblasts in serrated tumors.

As in the evaluation of gastric and esophageal periglandular fibroblasts, strict criteria must be followed so that tumor stromal myofibroblasts and hypertrophic smooth muscle fibers are not confused for pericryptal fibroblasts. The spindle cells should closely embrace the contours of the corresponding glands/crypts, and a connection to the muscularis mucosa is usually possible.

Finally, a small fraction of inflammatory bowel disease-associated colorectal adenocarcinomas present as the so-called very well-differentiated adenocarcinoma (low-grade tubuloglandular adenocarcinoma) of the colon [40]. They are composed of cells with low-grade nuclei. Some of the glands have round or tubular profiles, while some have branching glands. They lack both serration and pericryptal fibroblasts.

The normal adult hepatic tissue is ideally crafted for optimal substance transport between the circulation, bile canaliculi, and hepatocytes. The hepatic cords are normally one-cell layer thick with sinusoid vessels lining on each side. Two-cell thick plates and even rosette formations are visible only in regenerative conditions. The normal hepatic structure is maintained by a delicate network of reticulin lining the hepatic sinusoids as a result of normal epithelial–mesenchymal (stellate cells) interaction [42, 43]. Importantly other fibrogenic activity of the organ is kept to the minimum in the functional units with visible fibrous tissue limited to the portal tracts, bile duct and accompanying vascular bundles, and the capsule.

In most hepatocellular carcinomas, the antifibrogenic feature of the hepatoblasts is largely conserved (Fig. 7.13). While reticulin production is still operational or even increased, it seems to be outstripped by the rapidly proliferating tumor cells [43, 44]. Thus [44], thick cords (three or more cells across) lined by attenuated reticulin fibers become evident. This important feature can be appreciated even in the rare fibrolamellar and scirrhous variants in which abundant fibrotic tissue is present. The increased fibrogenic activity in these two variants indicate their cholangiolar differentiation (fibrolamellar variant cells: positive for CK7, EMA, and Hepar1; cirrhotic variant cells: negative for Hepar1 and positive for CK7) [45–49]. The pseudoglandular variant can also be viewed as having thickened plates if one can mentally cut the lumen open and spread the cells out. The normal canaliculi are lined by only two to three hepatocytes, and when spread out, the cords are no more than three cells thick. The imaginary cords in hepatocellular carcinomas seem to be more cellular and more diffuse than those derived from rosettes in hepatic adenomas and hepatitis.

In evaluating reticulin stain, it is important to pick areas uncompromised by steatosis. It is known that steatosis can impair the production of reticulin to an extent comparable to that of the hepatocellular carcinoma [50]. Moreover, hepatocellular carcinomatous cells can have steatotic changes.

Another important feature of hepatocellular carcinomatous cells is that they maintain the capability to induce brisk angiogenesis and obtain abundant blood supply like their hepatoblast counterparts [51–53]. The difference is, however, that the former tend to derive their nutrients from the arterioles, while the latter is supplied by both portal veins and arterioles. Thus, the tumor microvessels are subjected to higher perfusion pressures leading to capillarization. The normal sinusoids are negative for CD34 and CD31. Capillarized microvessels become positive for the two markers. This important feature has also been employed widely in the daily surgical pathology sign-out.

In the utility of both the reticulin and CD34/CD31 stains for the differentiation of hepatic nodules, it is advisable that they are considered in tandem. Rare cases of hepatocellular carcinoma with increased reticulin production have been reported and the increased reticulin production presents in the form of monolayer trabecular pattern [44]. Disturbance in the blood supply is common in benign hepatic nodules, and focal capillarization of the microvessels can occur. When the two stains seem to conflict with each other, careful attention to the distribution of CD34/CD31 positivity and finding a non-steatotic area can be very helpful. For instance, in cirrhotic nodule, the CD34/CD31 positivity is largely restricted to the periphery while focal nodular hyperplasia has its positivity near the fibrous septa [54, 55]. Variable staining patterns have been reported for hepatic adenomas. However, they have intact or only focally reduced reticulin framework with cords less than three cells across [54, 55].

The pancreatic exocrine morphogenesis/organogenesis follows the classic paradigm for glandular formation with a resultant lobular appearance. In chronic pancreatitis, the most common mimicker of well-differentiated ductal adenocarcinomas, this lobular pattern is still largely preserved even though the benign ducts are often atrophic and deprived of acinar structures. In contrast, carcinomatous glands are nonlobular and randomly distributed.

Reviewing this vital organ of the human body, one would not help but wonder at the ingenuity of Mother Nature in devising various safety mechanisms in order to prevent and limit the dreadful event of activating or leaking destructive digestive enzymes to adjacent tissues or organs. First of all, rather than obtaining its blood flow from a single source, the pancreas derives its arterial supply from rich anastomoses around the organ with many contributing components [59, 60]. Second, benign pancreatic ductal structures containing digestive enzymes are structurally separated from muscular vessels and are surrounded by acini and fibrous tissue [61, 62]. In doing so, the risk of lethal hemorrhage due to corrosion into the large arteries by digestive enzymes is reduced. This vessel shunning property of ductal epithelium has been illustrated by developmental biology research showing that early endothelial cells are essential for the onset of pancreatic budding and endocrine specification, but later on they inhibit further branching and exocrine differentiation [63–66].