The term “dysplasia” refers to “an unequivocal neoplastic epithelial alteration without invasive growth”. The term “intraepithelial neoplasia” often replaces “dysplasia” in World Health Organisation (WHO) guidance. Dysplasia is a precursor lesion of cancer and a marker for high cancer risk, offering a window of opportunity for early detection and cure of neoplasia. Most pathologists now classify columnar dysplasia as low grade (LGD) and high grade (HGD). The criteria for grading dysplasia include both cytological and architectural abnormalities. The diagnosis of dysplasia can be challenging in some clinical settings, especially when there is a background of active or resolving inflammation [e.g. in Barrett’s oesophagus (BO) or inflammatory bowel disease (IBD)] that may cause reactive epithelial atypia. Additionally, there is significant inter- and intra-observer variability for the diagnosis and grading of dysplasia. The variability may reflect the limitations of morphology-based criteria and has led to the development of adjunctive diagnostic methods such as immunohistochemistry. These methods, although promising, are controversial and require evaluation in further studies. This chapter describes the classification, microscopic features, and grading of dysplasia at different sites in the gastrointestinal (GI) tract.
The term ‘dysplasia’ refers to ‘an unequivocal neoplastic epithelial alteration without invasive growth’. The term ‘intraepithelial neoplasia’ often replaces ‘dysplasia’ in World Health Organization (WHO) guidance. Dysplasia is a precursor lesion of cancer and a marker for high cancer risk, offering a window of opportunity for early detection and cure of neoplasia. Most pathologists now classify columnar dysplasia as low grade (LGD) and high grade (HGD). The criteria for grading dysplasia include both cytological and architectural abnormalities. The diagnosis of dysplasia can be challenging in some clinical settings, especially when there is a background of active or resolving inflammation (e.g. in Barrett’s oesophagus [BO] or inflammatory bowel disease [IBD]) that may cause reactive epithelial atypia. In addition, there is significant inter- and intraobserver variability for the diagnosis and grading of dysplasia. The variability may reflect the limitations of morphology-based criteria and has led to the development of adjunctive diagnostic methods such as immunohistochemistry. These methods, although promising, are controversial and require evaluation in further studies. This chapter describes the classification, microscopic features, and grading of dysplasia at different sites in the gastrointestinal (GI) tract.
Definition of Dysplasia (Intraepithelial Neoplasia)
Carcinomas of the GI tract, whether squamous or columnar, constitute classic models of multistep carcinogenesis characterised by successive genetic alterations that transform a non-neoplastic epithelial cell into a cancer cell according to the sequence: non-invasive epithelial neoplasia and invasive carcinoma. The premalignant phase offers a window of opportunity for early detection and cure. Thus, pathologists frequently encounter biopsies of premalignant epithelial lesions of the GI tract (Practice Points 8.1).
The term ‘dysplasia’ refers to ‘an unequivocal neoplastic epithelial alteration without invasive growth’.
The term ‘intraepithelial neoplasia’ often replaces ‘dysplasia’ in the WHO recommendations and in the Vienna nomenclature system for gastrointestinal neoplasia.
Dysplasia is a precursor lesion of cancer and a marker for cancer risk.
Historically, the term designating non-invasive neoplastic epithelium in the digestive system is ‘dysplasia’. The definition offered by Riddell in 1983, now widely accepted, is purely morphological, i.e. ‘unequivocal neoplasia of the epithelium confined to the basement membrane, without invasion into the lamina propria’.1 According to the standardised classification system proposed by Riddell in 1983, initially for grading dysplasia in the setting of IBD, there are three main categories: (1) negative for dysplasia (a category reserved for non-dysplastic epithelium, whether normal or regenerating); (2) indefinite for dysplasia; and (3) positive for dysplasia (either low grade or high grade).1 Modification of this scheme now forms the basis for histological evaluation of dysplasia in the entire tubal GI tract, including BO, chronic gastritis, IBD, and colorectal adenomas. In addition, it is applicable to squamous neoplastic lesions of the oesophagus.2
Pathologists worldwide use the Riddell classification. However, there are some differences in approach, particularly between Japanese and Western/North American pathologists. In Western countries, the architectural features are very important for a diagnosis of dysplasia, whereas in Japan the cytological features often receive greater weight. The Vienna nomenclature, proposed by an international panel of pathologists, represents an attempt to address these discrepancies.3
The Vienna grading system, which is now in use by many pathologists worldwide, denotes five diagnostic categories: (1) negative for neoplasia; (2) indefinite for neoplasia; (3) non-invasive low grade neoplasia; (4) non-invasive high grade neoplasia (including non-invasive carcinoma in situ and suspicion of invasive carcinoma); and (5) invasive neoplasia (including intramucosal carcinoma and submucosal carcinoma or beyond). Thus, in the Vienna system, the term ‘non-invasive neoplasia’ replaces the term ‘dysplasia’ and the category ‘suspicion of invasive carcinoma’ describes lesions that show equivocal cytological or architectural features of tissue invasion. A comparison of Vienna and Riddell’s classifications is shown in Table 8.1.4 The WHO classification has proposed replacement of the term dysplasia by ‘intraepithelial neoplasia’.5 In fact, pathologists now use both terms, often interchangeably, according to institutional and national convention. However, the term ‘intraepithelial neoplasia’ is generally preferred to ‘dysplasia’ for anal lesions.
|1. Negative for neoplasia/dysplasia||Negative for dysplasia|
|2. Indefinite for neoplasia/dysplasia||Indefinite for dysplasia|
|3. Non-invasive low-grade neoplasia (low-grade adenoma/dysplasia)||Low-grade dysplasia|
a Not described in Riddell’s classification.
Diagnosis and Classification of Dysplasia (Fact Sheet 8.1)
The microscopic features that allow a diagnosis of dysplasia are both architectural and cytological. Abnormal growth patterns indicate faulty control of cellular proliferation, and include glandular crowding, tubular or villiform architecture, and the absence of normal base-to-surface epithelial maturation.
Grading of dysplasia uses a two-tier system: low grade or high grade. This method is demonstrably more reproducible than the three-tier system (mild, moderate, and severe), and has replaced the latter for descriptions of dysplasia of the columnar mucosa of the GI tract.6
In the intestinal columnar-lined GI tract, the distinction between low- and high-grade dysplasia depends on (1) nuclear features and (2) abnormal growth patterns. Nuclear features include the distribution of nuclei within the cells. In low-grade dysplasia, the nuclei remain in the basal half of the cells and in high-grade dysplasia there is haphazard distribution of nuclei between the basal and apical halves of the cells. In addition, high-grade dysplasia shows a greater degree of cytological atypia with more markedly enlarged, pleomorphic, hyperchromatic nuclei, and with loss of nuclear polarity (Figures 8.1 and 8.2). Table 8.2 summarises the features of low- and high-grade dysplasia.
Figure 8.1 Low-grade dysplasia. There is mild architectural disarray. The adenomatous epithelial cells show mild hyperchromasia with uniform, cigar-shaped nuclei that occupy the basal half of the cell cytoplasm. Basal polarisation of the nuclei is retained.
Figure 8.2 High-grade dysplasia. Cytologically, high-grade dysplastic epithelium shows pronounced nuclear stratification with mucin depletion and loss of nuclear polarity. Nuclei usually reach the surface of the cells. Architectural abnormalities are more severe than in low-grade dysplasia, often including cribriform areas as in this example.
|Surface maturation and transition from normal to atypical||Architecture||Cytology|
|Regenerative epithelial changes||Preserved|
|Low-grade dysplasia||Absent or only mild abnormalities including irregular shapes, rare crypt branching|
|High-grade dysplasia||Complex glandular architecture, including glandular crowding, cribriforming, back-to-back pattern|
In oesophageal and anal squamous mucosa, dysplasia again shows both architectural and cytological abnormalities. Architectural abnormalities include disorganisation of the epithelium, loss of cell polarity, overlapping nuclei, and irregular budding of neoplastic cells into the underlying lamina propria. Cytological abnormalities include nuclear enlargement, nuclear hyperchromasia, nuclear pleomorphism, an increased nuclear:cytoplasmic ratio, and increased mitotic activity. Mitotic figures may be present beyond the normal regenerative basal zone of the squamous epithelium.7 Individual cell keratinisation and dyskeratosis are common, particularly when dysplasia is associated with human papilloma virus (HPV) infection as in the anal canal (Figure 8.3). Dysplasia associated with HPV infection also shows koilocytic changes. In carcinoma in situ, analogous to full-thickness dysplasia, atypical cells extend through the full thickness of the epithelium without evidence of surface maturation. Acanthotic epithelial buds occur in both low-grade and high-grade dysplasia. Grading of squamous dysplasia currently varies between two-tier and three-tier systems but in general is moving towards two-tier.
Figure 8.3 Anal squamous intraepithelial neoplasia (AIN), characterised by immature cells above the basal layer that sometimes show nuclear pleomorphism. Also present are prominent human papilloma virus (HPV)-related cytological changes including koilocytosis (irregular nuclei with a perinuclear space). HPV changes may be difficult to distinguish reliably from dysplasia and may also cause difficulties with grading of AIN.
Challenges Related to the Diagnosis of Dysplasia
Distinguishing Non-neoplastic Regenerative Epithelial Atypia from Dysplasia
The interpretation of biopsies taken to assess dysplasia of columnar mucosa, particularly in endoscopic surveillance programmes, can be challenging, especially when there is a background of active or resolving inflammation (e.g. in BO or IBD). Ulceration may make the interpretation even more difficult. In these settings, reactive changes may overlap considerably with the changes of dysplasia. Regenerating epithelium may show features that resemble those of dysplasia, e.g. enlargement, hyperchromasia, stratification, pleomorphism, and variation in size of nuclei, prominent nucleoli, and an increase in the number of mitotic figures. A hallmark of a regenerative process is ‘maturation’ of epithelial cell nuclei, i.e. loss of their atypical features towards the luminal surface. Dysplastic epithelium, in contrast, extends uniformly along the crypt axis and onto the surface epithelium with little or no surface maturation.8 In addition, loss of nuclear polarity and complexity of glandular architecture are unusual in regenerative epithelium. Furthermore, gradual transition from atypical epithelium to non-atypical epithelium, particularly in association with a decrease in inflammation, favours regeneration over dysplasia (Figure 8.4). Microscopic features that help distinguish dysplasia from regenerative/reactive epithelial changes are summarised in Table 8.2.
Figure 8.4 Regenerative columnar mucosa with epithelial maturation gradient.
In the oesophageal squamous epithelium, basal cell hyperplasia, defined histologically as involvement of more than 15% of the thickness of the epithelium by basal cell-type appearance, can be observed in response to inflammation and may resemble dysplasia. According to one definition, the upper limit of the basal cell layer is the level above which the nuclei are separated by a distance greater than the nuclear diameter.9
Recognition of the overlap between the histological features of regenerative and dysplastic epithelium by Riddell et al. led to the creation of the designation ‘indefinite for dysplasia’ (category 2 in the Vienna classification). The ‘indefinite for dysplasia’ category is appropriate for cases in which a definite distinction between non-neoplastic changes and neoplasia is not possible, because of factors such as severe inflammation and regeneration, ulceration, or technical artefacts (too small, too superficial, maloriented, fragmented, poorly fixed, suboptimally stained, and/or marked cautery artefact) (Figures 8.5 and 8.6). In such cases, additional biopsy specimens are necessary, preferentially after administration of anti-inflammatory treatment. The category ‘indefinite for dysplasia’ includes lesions that are ‘probably negative’ or ‘probably positive’ for dysplasia. Although these terms might be helpful to the gastroenterologist when considering management options, most pathologists do not use them currently.
Figure 8.5 Indefinite for dysplasia. In these rectal biopsies of a patient with longstanding Crohn’s disease, poorly controlled by medical therapy, both crypts and surface are lined by cells showing enlargement, variably sized nuclei, nuclear hyperchromasia, and no base-to-surface maturation gradient.
Figure 8.6 Indefinite for dysplasia. Actively inflamed mucosa from an ileoanal pouch showing epithelial nuclear atypia with hyperchromaticity, pleomorphism, and little evidence of maturation towards the surface. In the context of an inflamed pouch these worrying appearances were labelled as indefinite for dysplasia rather than as dysplasia. The abnormalities have persisted for more than 15 years, suggesting (with hindsight) that they are regenerative rather than dysplastic.
According to most authors, true dysplasia lacks surface maturation and must involve the mucosal surface.6, 10 However, a subgroup of lesions, ‘basal crypt dysplasia’, is characterised by dysplasia limited to the basal portions of the crypts without involvement of the upper (luminal) half of the crypts and surface epithelium. Some authors state that crypt dysplasia may involve any (or all) of the length of the crypt and may not be limited to the base. In either circumstance, the intercrypt surface epithelium should show no dysplasia. This phenomenon may occur in BO, gastric epithelial dysplasia, and IBD.11–14 Although basal crypt dysplasia is sometimes included in categories such as regenerative lesions or indefinite for dysplasia, it appears to have neoplastic potential. In addition, it harbours the proliferative and molecular abnormalities that also occur in conventional dysplasia.11
Crypt dysplasia and conventional low-grade dysplasia both require nuclear pseudostratification for diagnosis. Occasionally, crypt dysplasia may show high-grade cytological features, including markedly enlarged nuclei with increased nuclear/cytoplasmic ratio, eosinophilic cytoplasm, irregularity of the nuclear membranes, and loss of nuclear polarity. High-grade changes are much less common than in conventional dysplasia, but should not deter pathologists from making the diagnosis of crypt dysplasia.15
Interestingly, biopsies from the majority of patients with crypt dysplasia also show areas of conventional LGD or HGD.11, 16 Therefore, when pathologists suspect basal crypt dysplasia, multiple levels through all of the mucosal biopsies to determine the highest grade of dysplasia and to seek conventional dysplasia are advisable. Finally, active inflammation in the lamina propria or within the crypt epithelium in the affected area precludes a diagnosis of crypt dysplasia.15
Distinction between low-grade and high-grade dysplasia is important, because of the consequences for patient management. The diagnosis of dysplasia depends on assessment of a slide stained with haematoxylin and eosin (H&E). Because dysplasia represents a continuum of changes without defined histological cut-off points, the distinctions between low-grade and high-grade dysplasia are subjective. In BO and IBD, most studies show significant inter- and intraobserver variability. Variability is especially high for low-grade versus indefinite for dysplasia or reactive, while agreement levels are better for the two extremes of the spectrum, i.e. negative for dysplasia and HGD.6, 17–19 To address this problem, some schemes recommend confirmation of dysplasia and its grade by a pathologist with expertise in GI pathology, even though the problem of poor reproducibility persists among ‘experts’.20, 21 The variations in grading of dysplasia may also reflect pathologist experience and place of work. Compared to specialists, non-specialist or general pathologists tend to assign a higher grade of dysplasia. In a Dutch registry study, re-review by three expert pathologists of 70 patients initially diagnosed with flat IBD-related LGD in ulcerative colitis confirmed flat LGD in 21 (30%). However, the diagnosis was downgraded to indefinite for dysplasia in 29 patients (41%) and to no dysplasia in 17 (24%), and was changed to non-IBD dysplasia in 3 (5%).22 Differences also persist in the interpretation of histopathology samples in BO between US and European/UK pathologists and in the criteria for dysplasia between Western and Japanese pathologists.23
Adjunctive methods may help improve interobserver variability in the detection and grading of dysplasia, notably in BO and IBD. p53 mutations can stabilise the inactivated protein causing p53 overexpression that is then detectable by immunohistochemistry. Truncating p53 mutations or epigenetic silencing may also result in protein inactivation and subsequent loss of p53 expression (the ‘absent pattern’) rather than overexpression. This absent pattern, initially described in neoplastic lesions in BO,24 also occurs in IBD and is also considered abnormal.25
p53 immunostaining in practice may help distinguish dysplasia from regenerative alterations but this is a controversial topic that divides pathologists, including experts.26 Aside from its possible value for diagnosis, p53 expression may also help predict the progression of IBD-associated neoplasia, and there may be a correlation between p53 overexpression and severity of IBD-associated neoplasia.22, 27 In patients with BO, aberrant p53 protein expression (and especially loss of p53 protein expression) is associated with an increased risk of neoplastic progression. In some reports, it is a more powerful predictor than low grade dysplasia of neoplastic progression.24 This has led the British Society of Gastroenterology (BSG) to recommend routine use of p53 staining in addition to histopathological assessment in order to improve the reproducibility of a diagnosis of dysplasia.28 However, the BSG also proposes expert pathologist consensus as the best way to ensure a correct diagnosis.
Some studies show that restriction of the proliferation marker Ki67 staining to the basal third of crypts excludes a diagnosis of dysplasia.29, 30 Therefore, immunohistochemical detection of Ki67 within the superficial parts of crypts and in the surface epithelium may help strengthen a diagnosis of dysplasia in difficult cases. In HPV-associated lesions, there is a strong positive correlation between p16 immunostaining and precancerous disease, according to some reports.31 p16 immunostaining appears to reduce interobserver variability in the diagnosis of precancerous disease. Furthermore, diffuse strong p16 staining is associated with a positive test result for HPV-16 and other high-risk HPV types. In addition, a diffuse, strong, block-like positivity for p16 immunohistochemistry may help confirm a diagnosis of high-grade intraepithelial neoplasia (see the section on anal canal neoplasia).32 A p16-positive result is defined as strong and diffuse nuclear, or nuclear plus cytoplasmic, staining of the basal cell layers, involving at least one-third of the epithelial thickness. Isolated cytoplasmic staining represents a negative result.
Negative for dysplasia
Indefinite for dysplasia
Positive for dysplasia (either low grade or high grade)
Negative for dysplasia refers to non-dysplastic epithelium, either normal or regenerative
Indefinite for dysplasia is appropriate when a definite distinction between non-neoplastic changes and neoplasia is not possible, regardless of the reasons
Categorisation as indefinite for dysplasia warrants early repeat biopsy of the same area
Grade of dysplasia is determined by evaluating a combination of cytological and architectural abnormalities
Significant inter- and intraobserver variability is consistently reported for the diagnosis and grading of dysplasia, especially low-grade versus indefinite
Many guidelines recommend confirmation of dysplasia and its grade by an expert gastrointestinal pathologist, especially in the setting of inflammatory bowel disease or Barrett’s oesophagus
Adjunctive diagnostic methods such as immunohistochemistry (p53 immunostaining in the columnar-lined GI tract and p16 in human papilloma virus–associated neoplasms) may reduce interobserver variability, but the value of these methods requires further evaluation
Gastric Epithelial Dysplasia
The development of gastric adenocarcinoma represents the culmination of an inflammation – atrophic gastritis – intestinal metaplasia – dysplasia – carcinoma sequence (the ‘Correa cascade’).33 The precursor lesions predispose to intestinal-type gastric adenocarcinoma. Helicobacter pylori infection plays a pivotal role in gastric carcinogenesis.
Macroscopically, there are flat, polypoid, or depressed forms of gastric dysplasia. Most lesions are non-polypoid. Notably, European and North American pathologists restrict the term ‘adenoma’ to a discrete, protruding dysplastic lesion, whereas in Japan the same term includes all macroscopic forms of dysplasia.
The 2010 WHO classification of tumours of the digestive system identifies two subtypes of gastritis-associated dysplasia: (1) adenomatous (type 1) and (2) foveolar (type 2) (Table 8.3).45 Adenomatous (or intestinal-type or type I) dysplasia is the most common. This type, which resembles colonic adenomas and the dysplastic lesions complicating IBD, typically develops on a background of intestinal metaplasia and is probably the precursor of most intestinal-type gastric adenocarcinomas. Microscopically, grading of adenomatous dysplasia uses a two-tier system of low grade or high grade that depends on the severity of both architectural and cytological abnormalities (as in other parts of the GI tract). The second histological variant, less common and less well recognised, is foveolar (gastric) dysplasia (type II). This develops in non-metaplastic gastric epithelium and is associated with the subsequent development of poorly differentiated intestinal-type adenocarcinoma. Microscopically, the dysplastic cells are typically cuboidal or low columnar and have clear or eosinophilic cytoplasm, loss of nuclear polarity, and hyperchromatic vesicular nuclei with nucleoli. The low-grade form of this type of dysplasia is difficult to differentiate from regenerative changes. Most cases of intestinal and foveolar dysplasia show full-length pit and surface epithelial involvement, without maturation (Figure 8.7).