Ulceration caused by mechanical abrasion is usually shallow and transient. Like other portions of the alimentary system, shallow ulcers heal within hours by flattening and sliding of adjacent epithelium followed by replacement of mucus cells generated by a transient increase in mitotic activity within the isthmus zone at the base of the foveolae. The regenerating epithelium is flatter and more basophilic, with little mucus and slight irregularity in orientation of the nucleus (Fig. 29-3). Shallow ulcers heal completely within just a few hours. The adjacent superficial lamina propria will usually have at least some edema and perhaps even hemorrhage, but the leukocytic reaction is usually minimal. A few neutrophils may be found in the very superficial lamina propria or intermingled with some fibrin and mucus filling in the ulcer. The paucity of leukocytes is in part because the injured tissue is sloughed into the intestinal lumen so that there is little chemical stimulus for leukocyte recruitment, and partly because the stomach has almost no resident bacteria to contaminate the wounds. With deeper or more persistent ulceration, there will be more obvious hyperplasia and basophilia within the isthmus region reflecting a greater need for mitotic replacement of injured epithelium. Deeper ulcers that have destroyed the isthmus and portions of the lamina propria will take longer to heal and will require stromal proliferation to build a scaffold for effective epithelial healing, but remodeling is extremely effective and it is rare to see any permanent residual scarring. Regardless of the severity of the original lesion, the usual outcome is histologic normalization (Fig. 29-4).

Figure 29-3 Uremic gastropathy, dog. Note the typical horizontal band of acute epithelial necrosis affecting mostly the parietal cells in the middle third of the mucosa. In severe cases there is mineralization of degenerate smooth muscle in the muscularis mucosa and submucosal blood vessels.

Figure 29-4 Normal feline stomach with a substantial amount of fibrous tissue between the base of the glands and the muscularis mucosa, including the very distinctive lamina densa (arrow).

Causes of ulceration other than mechanical abrasion are numerous and include histamine-induced ulceration in dogs with mast cell tumor,^3,⁴ accidental chemical ingestion, ulcers caused by steroidal and nonsteroidal antiinflammatory (and occasionally other) drugs,⁵^–⁷ and exercise-induced ulcers that are best documented in racing sled dogs.⁸ In general, the histologic character of the ulceration provides no clue as to the pathogenesis. Careful estimation of the age of the lesions (distance of epithelial sliding, restoration of the normal columnar epithelial shape and normal cytoplasmic eosinophilia, and the maturity of subepithelial fibrosis) will often allow the clinician to identify the causative drug or event. Progression to perforation is rare and usually requires continued application of the injurious stimulus (e.g., continued administration of ulcerogenic drugs or persistent histamine production by a mast cell tumor).

Gastric Necrosis Other Than Ulceration

Epithelial necrosis not typically associated with superficial ulceration is seen with uremia and sometimes with mucosal ischemia secondary to thromboembolic disease, submucosal vasculitis, or gastric dilation/volvulus.

Uremic gastropathy occurs almost exclusively in dogs and its exact pathogenesis remains unknown. The distinctive lesions include mineralization and necrosis of the parietal cells occupying the middle third of the gastric mucosa, mineralization of basement membranes of the glands and of the submucosal blood vessels, and mineralization of degenerate smooth muscle within the muscularis mucosa or even tunica muscularis (Fig. 29-5). Mineralization of submucosal blood vessels may become extensive, involving the full thickness of the vessel wall and accompanied by medial necrosis and endothelial destruction.⁹

Figure 29-5 Canine fundic stomach with glandular atrophy, glandular nesting, and lamina propria in the absence of any significant infiltration of leukocytes. This is a common lesion, but with no proof that it is medically significant.

With ischemic injury related to vasculitis or vascular occlusion, the character of the lesion depends on the duration and completeness of obstruction. In gastric volvulus, venous obstruction is complete and the intensely congested gastric wall dies following infarction. The mucosa dies first, undergoing coagulation necrosis that quickly becomes lost amid diffuse hemorrhage. In cats with vasculitis caused by feline infectious peritonitis (FIP) or in dogs with disseminated intravascular coagulation (DIC), the necrosis is often patchy.

Gastric Mucosal Fibrosis and Atrophy

Varying degrees of mucosal fibrosis and glandular atrophy are common in biopsy samples of the gastric mucosa. Such lesions have received almost no attention in the veterinary literature, but are common sources of confusion for pathologists attempting to interpret endoscopic biopsies. Some have referred to this as “atrophic gastritis,” with no proof that the pathogenesis of the fibrosis is linked to previous inflammatory disease. In the only study to specifically address this question, noninflammatory atrophy and/or fibrosis was seen in 128 of 482 vomiting dogs. Similar lesions were also seen in five of 19 clinically healthy dogs.¹⁰ It is not clear whether the fibrous tissue represents a true increase in the volume of collagen or simply a condensation of resident collagen subsequent to glandular atrophy. The two lesions (mucosal fibrosis and glandular atrophy) are almost always concurrent.

There remain numerous questions related to gastric atrophy and fibrosis, including these:

1. What is the normal amount of fibrous tissue within the lamina propria in different portions of the stomach?

2. Does it increase with age?

3. How commonly is such fibrosis a marker of previous inflammatory disease?

4. Is it permanent?

5. Is it functionally significant, significant only as a diagnostic marker, or not significant at all?

Normally, the glands within the fundic mucosa lie against each other with almost no intervening fibrous tissue. There is a small amount of loose connective tissue among the foveolae, and there can be substantial fibrous tissue separating the base of the glands from the muscularis mucosa. The amount of fibrous tissue in this deep location is greater in cats than in dogs. In cats there is also a distinct, very dense band of hyalinized fibrous tissue just superficial to the muscularis mucosa, known as the lamina densa (Fig. 29-6). Within the pyloric antrum there is substantially more fibrous tissue throughout the lamina propria, so any diagnosis of mucosal fibrosis in the pyloric stomach should be viewed with skepticism.

Figure 29-6 Superficial mucosal eosinophilic gastritis, dog. This microscopic field corresponds to Southorn’s “mucosal unit.” In this instance, it has approximately twice the allowable number of leukocytes, and five times the normal number of eosinophils. Despite the severity of the infiltration, note that the overlying epithelium (as usual) remains virtually normal.

Dogs and cats with chronic gastritis often have persistent mucosal edema that matures into fibrosis. Concurrent inflammatory injury to the gastric glands is followed by regeneration that creates “nests” of hyperplastic glands embedded within the fibrous tissue (see “Atrophic Gastritis” section). It is also common to see the combination of fibrosis and glandular nesting with no evidence of active inflammation and no clinical history of gastric disease (Fig. 29-7). The glands within these nests usually have at least some degree of atrophy of the parietal cell mass with a relative increase in the prominence of the columnar mucus-producing epithelial cells. The combination of glandular atrophy, dysplastic repair, and mucus (intestinal) metaplasia represents a strong risk factor for progression to gastric carcinoma in people,¹¹ an association that has not been proven in dogs.

Figure 29-7 Transmucosal gastritis, dog. There is a threefold increase in mononuclear leukocytes throughout the lamina propria, with some mild glandular nesting accentuated by the inflammatory edema and fibrosis. The lack of architectural effacement distinguishes this, even at low magnification, from lymphoma.

Gastritis

The gold standard for the histologic diagnosis of gastritis is an increase in mucosal leukocytes accompanied by other evidence of inflammation such as hyperemia, edema, and bystander injury to the adjacent structural components of the mucosa (e.g., epithelial injury or reparative fibrosis). That gold standard is met in a proportion of the cases, but not in all. The lack of uniformity in the criteria used to diagnose gastritis impacts on the credibility of virtually all of the published literature on the prevalence, classification, etiology, and treatment of gastric inflammatory disease. To many, gastritis is defined simply as an increase in the number of leukocytes within the lamina propria beyond the normal range. The most obvious problem with this definition is that we do not know the normal range. Less obvious but even more fundamental is the question of whether an increase in leukocytes within a tissue designed to respond to antigenic challenge by increasing the number and activity of resident leukocytes should be interpreted as “inflammation.” An increase in lymphocytes or plasma cells in a lymph node is usually assumed to reflect an appropriate and purposeful response to antigenic stimulation. In the intestinal lamina propria, the largest lymphoid organ in the body, a similar increase in lymphocytes and plasma cells is generally viewed as pathologic and therefore deserving of therapeutic intervention. The distinction between adaptive and pathologic lymphoid proliferation lies not just in the number of leukocytes, but also in the phenotype and cytokine production profile of the cells. These complexities are beyond routine histopathology and immunohistochemistry, but should not be ignored.

Normal Gastric Mucosal Cellularity

Only three studies address the critical question of the reference range for leukocyte types in the normal canine gastric mucosa. The first reports findings from examination of endoscopic biopsies from 20 clinically normal young adult dogs (Table 29-1).¹² These dogs had no history of vomiting or diarrhea for at least 2 months prior to entering the study, and were all fed a standardized diet and had the same husbandry conditions. Leukocytes were identified with light microscopy and immunohistochemistry, and were counted separately in both superficial and deep lamina propria in “mucosal units” of 250 µm² (corresponding to approximately one-half of a ×40 microscopic field). Within the superficial lamina propria of the fundic mucosa, this corresponds to a rectangle of tissue incorporating three adjacent foveolae and the associated lamina propria, and extending vertically to the level of the first parietal cell. Within the pyloric antrum, the “mucosal unit” corresponds to an area of lamina propria bordered by two adjacent foveolae, with its deep border capturing the most superficial portion of the gastric glands.

Table 29-1 Leukocyte Numbers in the Superficial Gastric Mucosa of Healthy Dogs

Cell Type	Superficial Fundic*	Superficial Pyloric^†
Lymphocytes (CD3 and CD79)	8.4 (1.5-27.0)^‡	22.9 (11.5-32.0)^§
Plasma cells	1.6 (0-5.8)^‡	6.8 (0.5-15.5)^§
Eosinophils	0.5 (0-2.0)^‡	2.7 (0-6.0)^§
Total leukocytes**	10.4 (3.6-26.2)^‡	32.6 (20.0-46.5)^§

^* Number of cells within a rectangle of lamina propria encompassed horizontally by three foveolae, and vertically to the level of the first parietal cell.

^† Number of cells in a square of lamina propria encompassed by two adjacent foveolae with the deep border at the foveolar–glandular junction.

^‡ Mean (range) of three cell counts in each of 20 dogs.

^§ Mean (range) of two cell counts in each of 8 dogs.

^** Sum of CD3⁺ and CD79⁺ lymphocytes, plasma cells, and eosinophils, but excluding mast cells and unidentified cells.

Source: Adapted with permission from Southorn EP. An improved approach to the histologic assessment of canine chronic gastritis. DVSc Thesis, University of Guelph, Ontario, Canada, 2004.

The other two studies were not specifically designed to provide data about normal gastric mucosal cellularity or other features relevant to the diagnosis of gastritis. Nonetheless, both provide a series of pictorial templates of normal and diseased stomachs in an attempt to reduce interobserver variation in interpretation of gastric histopathology. Neither study provided any information about how the photographic templates identified as “normal” were selected.

The first of these studies was based on 18 client-owned dogs with GI disease and eight dogs euthanized for nonenteric disease. None of the dogs had been treated with antibiotics, corticosteroids, or antacids in the months prior to sampling. Lesions within the fundic mucosa alone were graded by two pathologists using pictorial templates adapted from the human Sydney system that depicts various severities of atrophy, fibrosis, and cellular infiltration. The results from the eight “normal” dogs were not separately identified when tabulating the results (presented as the mean grading results from all 26 dogs).¹³

The second and much larger study provided not only a series of pictorial templates, but also a brief text to guide the grading of GI biopsies, including those from stomach. It included pictorial templates for leukocyte accumulations, fibrosis, ulceration, and other parameters within the fundic and antral mucosa.¹⁴

In both of these latter studies in which the pictorial templates were specifically designed to reduce interobserver variation, the results showed poor correlation among grading scores assigned by different pathologists. Painful though it is, we must collectively confront the reality that it is essentially impossible to compare results from different studies claiming to evaluate causation or therapy of “gastritis.” In most published studies of gastritis in either dogs or cats, the descriptions of leukocyte numbers or the published photographs of lesions claimed to represent mild and even moderate gastritis would be considered “normal” using criteria in the three studies listed previously. This is particularly true of the many studies attempting to establish the pathogenicity of Helicobacter spp. for dogs and cats in which any histologic changes were described as mild increases in superficial mucosal cellularity.¹⁵^–²³

Histologic Classification of Gastritis

The previously mentioned problems notwithstanding, there is no doubt that genuine gastritis does exist and is quite common. Most cases are part of more generalized GI inflammatory disease,²⁴ but some are purely or predominantly gastric. There is no evidence that histologic classification based on the distribution or type of leukocytes has relevance to pathogenesis, treatment, or prognosis, nor is there evidence to the contrary. Thus it is appropriate to use (at least for the moment) a classification system that reflects histologic changes, with the expectation that at least some of these differences will eventually prove to have etiologic, therapeutic, or prognostic significance.

Superficial Mucosal Gastritis

Most examples of gastritis in dogs and cats involve leukocyte infiltration of the superficial 20% of the mucosa. In almost all cases, the surface epithelium remains normal. In most cases, lymphocytes and plasma cells predominate, but there are almost always at least some eosinophils. Neutrophils are rarely present, but their presence almost always means nearby or recent ulceration. There may be a concurrent increase in intraepithelial lymphocytes, and in cats there can be an increase in intraepithelial large granular lymphocytes (formally known as globular leukocytes) that are of unknown significance. Edema accompanies the leukocytes and there is often activation of resident fibroblasts and blood vessels (Figs. 29-8 and 29-9).

Figure 29-8 Focal gastric pyloric mucosal hypertrophy, dog. The hypertrophy is limited to the foveolae and surface epithelium. The elongation of the foveolae may be so marked that a routine endoscopic biopsy will capture only the surface and foveolar epithelium and none of the underlying pyloric glands. This justifies a presumptive diagnosis of mucosal hypertrophy in an endoscopic biopsy.

Figure 29-9 Gastric polyp within the fundic mucosa, cat. In contrast to pyloric mucosal hypertrophy, the proliferation here is by hyperchromatic germinal epithelium with a little mucus production and parietal cell differentiation. Cystic glandular dilation is common. The lack of invasion into the lamina propria distinguishes this from papillary gastric carcinoma (which would be exceedingly rare in a cat).

It has been traditional to describe the predominant cell type (lymphocytic, plasmacytic, eosinophilic) as part of the classification system, but there is no evidence that this is useful.^7,^14,²⁴ First, there is no uniformity between laboratories in how well eosinophils, mast cells, or even plasma cells are stained. Second, pathologists differ in defining what proportion of cells is required before that cell type is considered “predominant.” Some will diagnose a lesion as “eosinophilic” if they believe that the eosinophils are the most numerous cells entering the tissue on an hourly basis, even if they are outnumbered by lymphocytes or plasma cells that have accumulated over weeks and months. Others use a more traditional approach and name the lesion based on the cell type that is most numerous within the tissue at the time of assessment. That means that almost any chronic gastritis will be classified as “lymphoplasmacytic.” Until everyone uses the same convention for naming such lesions, subclassification based on the predominant cell type will remain unreliable. It makes more sense to simply list each identifiable cell type as a percentage of the total infiltrate, so that such information can later be analyzed to see if it influences therapeutic response or other clinical parameters.

Transmucosal Gastritis

Approximately 30% of samples identified as having gastritis by conservative histologic criteria have an increase in leukocytes with variable degrees of edema and/or fibrosis throughout the full thickness of the lamina propria, causing separation of the gastric glands from each other and from the underlying muscularis mucosa (Figs. 29-10 and 29-11). There usually is no obvious necrosis within the glands or along the surface. Lymphocytes and plasma cells predominate, but eosinophils are often conspicuous and sometimes even predominant.

Figure 29-10 Early gastric carcinoma in the fundic mucosa, dog. In early cases, the diagnosis can be difficult and is based on detecting invasion and replacement of lamina propria by monotypic mucus-producing epithelial cells. Formation of tubules is not common within the intramucosal portion of such tumors.

Figure 29-11 Gastric carcinoma, dog. There is abundant reactive fibrosis induced by transmural invasion of a mucus-producing tubular carcinoma. The large lakes of mucin within the tunica muscularis are often more obvious than the tumor cells themselves.

Even though it is unusual to see actual necrosis, there often is some glandular atrophy so that the glands appear in isolated nests separated by a mixture of leukocytes, edema, and fibrosis. This is probably the precursor for the lesion usually referred to as atrophic gastritis, an extremely frequent but controversial microscopic entity that usually has no clinical counterpart.

Atrophic Gastritis

Atrophic gastritis is a purely descriptive name for chronic gastritis that combines the elements of chronic inflammation, glandular atrophy, regenerative glandular nesting, and mucosal fibrosis. Lymphofollicular hyperplasia may accompany these changes. There is reduction in overall mucosal thickness and in most cases an obvious reduction in the number of parietal cells with a corresponding absolute or relative increase in the proportion of mucus-producing epithelial cells within the glands. If there are functional consequences to the reduction in glandular mass, they are rarely recognized clinically. Pathologists are not uniform in the criteria used to make this diagnosis, so it is difficult to assess the scattered literature related to pathogenesis and clinical significance. Atrophic gastritis may be just the residual lesion of transmucosal gastritis described previously. Atrophic gastritis has also been used as the name for the parietal cell atrophy and neuroendocrine cell hyperplasia that precedes the development of anaplastic and neuroendocrine gastric carcinomas in Norwegian Lundehund dogs with a distinctive, familial protein-losing enteropathy.²⁵

Etiologic Classification of Gastritis

The majority of cases classified as gastritis by histologic criteria have no identified cause. The list of potential causes includes infectious agents, chemical irritants, toxic plants, and immune-mediated diseases.⁷ Most of the chemical and plant irritants cause transient gastric ulceration rather than true gastritis, and the inclusion of various allergic or other immune-mediated pathogeneses is based on assumptions rather than proof. Because the stomach is not a site of any significant protein absorption, assumptions of immune-mediated disease may be more appropriate for small intestine than stomach.

Instances of gastritis with a known cause are almost always related to specific infections. These represent a very small proportion of all examples of gastritis, and for most of these agents infection is not usually linked to clinical signs of gastritis or to significant histologic lesions. Parasitic agents include the nematodes Physaloptera, Gnathostoma, and Cylicospirura in dogs and cats, and Ollulanus in cats alone. Infrequent but devastating examples of infection with the fungus Pythium insidiosum cause mucosal and submucosal suppurating granulomas in the stomach and elsewhere in the GI tract of dogs. The stomach is an occasional victim in the course of systemic fungal diseases (e.g., blastomycosis, cryptococcosis). Finally, the crusade to establish Helicobacter as a cause for clinically significant gastritis or gastric ulceration in dogs or cats has triggered more descriptions of alleged gastric lesions than has any other proposed agent of gastric disease.

Parasitic Gastritis

Among the nematode parasites that may infect the stomach and cause histologic lesions, the only ones usually considered of potential significance are Physaloptera and Ollulanus. Gnathostoma and Cylicospirura may cause small submucosal suppurating granulomas, but they do not cause clinical signs. Infection with Physaloptera has been reported to cause persistent vomiting in dogs, and elimination of the worms following treatment resulted in reversal of clinical signs.²⁶ Descriptions of histologic lesions are few and vague.

Ollulanus is sometimes listed as a cause of vomiting in cats. It has been described as a cause for fibrosis, lymphofollicular hyperplasia, and increased intraepithelial large granular lymphocytes in cats, but in the original report there was no clinical disease associated with the infection.²⁷ In some cats the worms can be seen embedded within the foveolae, with no histologic changes. The association between Ollulanus infection and apparent increases in intraepithelial large granular lymphocytes remains unproven. The latter is a frequent observation in the stomach and small intestine of cats with GI disease and is of unknown significance.

Cryptosporidiosis has been reported in the stomach of cats and dogs, but with no proof of pathogenicity.²⁸

Pythiosis

P. insidiosum is an aquatic oomycete fungus found primarily in tropical and warm temperate climates. Most reported cases in dogs have been from the southeast United States, but the disease is seen worldwide. Lesions in the stomach have the same histologic character as those seen elsewhere: destructive coalescing necrotizing granulomas found randomly within submucosa, tunica muscularis, or within the serosa.^29,³⁰ The diagnosis depends on identifying characteristic fungal hyphae. These are often difficult to see without special stains (e.g., periodic acid-Schiff [PAS] or silver stain), but the necrotizing lesions throughout the stomach wall are not easily mistaken for anything else.

Helicobacteriosis

Helicobacter spp. are normal inhabitants of the surface and foveolae of dogs and cats. In dogs, most infections are with Helicobacter bizzozeronii or Helicobacter heilmannii, while in cats H. heilmannii and Helicobacter felis predominate. These large spiral bacteria are easily seen in routine endoscopic biopsies, and their detection is enhanced by the use of silver stains. More elaborate and sensitive detection methods (e.g., polymerase chain reaction [PCR] or tissue urease activity) are research tools rarely if ever used in a diagnostic setting. In the past 15 years, there have been at least nine studies specifically examining the relationship between Helicobacter infection, development of histologic lesions, and clinical disease. None of these studies has proven any causal relationship between infection and disease.^13,¹⁵^–²⁰ All but one, however, have claimed that infection is associated with an increase in gastric mucosal leukocytes, fibrosis, and lymphoid follicles. The problem is that the increased leukocytes claimed to represent “gastritis” invariably fall within the normal range for gastric mucosal cellularity as established by Southorn and as subsequently adopted by the World Small Animal Veterinary Association (WSAVA) Gastrointestinal Standardization Group.^12,¹⁴ Most of those studies had no uninfected control against which to measure the claims for increased cellularity. Even if one were to accept the hypothesis that colonization by Helicobacter causes a small increase in mucosal leukocytes (including lymphofollicular hyperplasia), the most plausible explanation is that the Helicobacter simply represent an increased antigenic challenge to the resident mucosal leukocytes that respond in an entirely appropriate fashion, analogous to what occurs when gnotobiotic animals are first introduced to a “normal” bacterial flora.³¹

Gastric Proliferative Disease

Proliferative changes within the stomach include focal adenomatous hyperplasia in the form of gastric polyps, more diffuse but rare canine hypertrophic gastropathy, and genuine neoplasms.

Focal Adenomatous Hyperplasia

Focal adenomatous hyperplasia occurs in two forms. The more prevalent is focal pyloric mucosal hypertrophy in which the foveolae of the pyloric antrum become greatly exaggerated and form villus-like protrusions covered by mature and otherwise normal epithelium.³² The lamina propria sometimes contains an increase in eosinophils, but it is unknown whether those examples with eosinophils have a different pathogenesis than those without. It is unknown whether the eosinophils are involved in the development of the lesion, are the result of the lesion, or are unrelated.

The prevalence of these lesions is directly related to the frequency of endoscopy and many are clinically silent. They become significant only if they grow large enough and if they are located where they may cause pyloric obstruction. The diagnosis is difficult to confirm in endoscopic samples. If orientation and depth are both perfect, then the diagnosis is based on seeing the biopsy filled with superficial and foveolar epithelium with no glands because the increased thickness of the foveolar region occupies the entire capacity of the biopsy forceps (Fig. 29-12).

Figure 29-12 Large cell “lymphoblastic” lymphoma, cat. Typically, these form a solitary discrete mass originating within the lamina propria. This is traditional lymphoma with architectural obliteration by a diffuse sheet of large monotypic lymphocytes (in a cat, virtually always B cells), here leaving only a few lingering remnants of preexistent glands.

Less prevalent than pyloric mucosal hypertrophy is focal adenomatous hyperplasia creating polyps in the gastric body or fundus. These are about three times more prevalent in cats than in dogs. In contrast to the proliferation of mucus-laden superficial epithelium typical of pyloric mucosal hypertrophy, these gastric polyps are formed primarily by papillary proliferation of hyperchromatic glandular epithelium forming branching and sometimes cystic tubules. Most of the epithelium is hyperchromatic and primitive columnar epithelium with no mucus production, but with occasional maturation into parietal and chief cells. They differ from papillary adenocarcinomas by not showing any invasion across the basement membrane and into the lamina propria or submucosa. The lack of invasion is the only reliable criterion distinguishing gastric polyp from a well differentiated papillary adenocarcinoma. There is no proof that these lesions progress to neoplasia.

Gastric Neoplasia

Although a wide variety of neoplasms are reported within the stomach,^35,³⁶ the majority fall into one of three groups: gastric carcinoma, lymphoma, and smooth muscle tumors.

Gastric Carcinoma

Mucus-producing, scirrhous tubular adenocarcinoma is the most common tumor of the canine stomach, but gastric carcinomas are almost nonexistent in cats. In dogs, they usually occur as a regional plaque-like thickening with ulceration in the distal body or within the pyloric antrum. Ulcerated tumors may be confused both macroscopically and microscopically with chronic ulcers that have stimulated abundant reactive fibrosis and dysplastic epithelial repair.

The earliest recognizable histologic lesion is focal obliteration of mucosal glandular architecture by proliferating polygonal epithelial cells that violate the basement membrane of the glands/crypts to proliferate as individualized cells within the lamina propria. These cells, which may not be easily recognized as epithelial cells, may be found only deep within the lamina propria. They may be difficult to capture in endoscopic biopsies especially if the tumor is arising in pyloric mucosa (the thickness of the pyloric mucosa makes it difficult to capture the deep third of the mucosa). There will be times in which full-thickness samples are required to confirm the diagnosis, but that is the exception rather than the rule. The tumor cells invade the submucosa and then the tunica muscularis. In most cases, the initial invasion is in the form of slender cords or tubules of mucin-producing epithelium within or surrounding lymphatics. They penetrate the serosa and exfoliate into the peritoneal cavity to seed the omentum and mesentery. In most cases the tumor cells are well differentiated and form easily recognizable tubules, but some examples comprise only lakes of mucin within the tunica muscularis, or nodules of reactive fibrous tissue on the mesentery or omentum. In such cases, the use of a PAS stain will accentuate the mucus within the cytoplasm of the anaplastic epithelial cells that will be around the mucus or buried within the fibrous tissue.

It is traditional to subclassify gastric carcinomas based on histologic pattern (e.g. papillary, tubular, scirrhous),³⁵ but most examples exhibit substantial regional variation that includes more than one subtype. In addition, there is no proven prognostic or therapeutic significance to these patterns so it is difficult to justify subclassification.

Leiomyoma and Leiomyosarcoma

Any published estimates of the prevalence of smooth muscle tumors must be viewed with skepticism. In part this is because smooth muscle tumors are usually small and affect tunica muscularis only, so the majority do not cause clinical signs and cannot be detected endoscopically. Estimates of prevalence therefore differ in studies based on clinical signs as contrasted with postmortem studies.^42,⁴³ Further confusion arises from the fact that many of those tumors initially classified as leiomyosarcomas based on traditional histopathology have been reclassified as gastrointestinal stromal tumors (GISTs) on the basis of immunohistochemistry (positive for CD117, negative for smooth muscle actin).^44,⁴⁵ Those papers addressed small intestinal, cecal, and colonic tumors rather than specifically looking at those in stomach, and in addition specifically addressed smooth muscle tumors initially diagnosed as poorly differentiated leiomyosarcomas rather than more common, mature leiomyomas. It is therefore a mistake to assume that all tumors previously diagnosed as well-differentiated smooth muscle tumors are at risk of having that diagnosis “overturned” by immunohistochemistry.

Smooth muscle tumors usually occur as focal, solitary, discrete masses arising within the tunica muscularis. They represent a continuum from benign to malignant, with no precise histologic division between leiomyoma and leiomyosarcoma. Those with discrete, purely expansile growth and formed of histologically and cytologically mature smooth muscle cells are called leiomyomas. At the other end of the spectrum are primitive stromal tumors with invasive growth, increased mitosis, and only vague resemblance to smooth muscle cells that have traditionally been classified as leiomyosarcomas. Many examples lie in the middle of that continuum, and pathologists are not uniform in how they classify these. Leiomyomas are formed by cells virtually indistinguishable from normal smooth muscle, except that they grow in interlacing bundles to create a disorganized nodule clearly different from the surrounding, laminar arrangement of the tunica muscularis. These tumors may protrude outward through the serosa or inwardly to cause bulging of the mucosa with subsequent ulceration.

The poorly differentiated tumors are comprised of pleomorphic spindle cells with the usual criteria of malignancy, including regional invasion. Any decision to refer to these as smooth muscle tumors rather than GISTs is subjective and should be confirmed with immunohistochemistry.

There is poor correlation between histologic appearance and biologic behavior. As a group, the smooth muscle tumors have a relatively good postoperative prognosis with prolonged postoperative tumor-free intervals and survival times regardless of histologic appearance.⁴⁵ For those tumors that behave aggressively, the initial histologic appearance (i.e., leiomyoma vs. leiomyosarcoma) is not predictive.