What Is Dysplasia in the Gastrointestinal Tract?

Henry D. Appelman, MD

● Dysplasia in the gastrointestinal tract is considered both the skin and those in the salivary gland, the most prom- a carcinoma precursor and a marker of high cancer risk inent of which is the pleomorphic adenoma. These tumors for the site at which it is found. Dysplasia is defined as are composed of mature epithelium, and the risk of cancer unequivocally neoplastic epithelium, yet the specific cri- occurring within them is minute. It was no wonder that teria for making that determination are imperfectly de- the same name, adenoma, was used for polypoid epithelial fined. The current criteria actually include a mix of archi- proliferations, mainly in the colon, that were considered tectural and cytologic features, all of which occur in dif- neoplastic but not malignant. However, it was not known ferent intensities in different epithelia that are given the then that the things called adenomas in noncolonic sites same diagnosis. Gastrointestinal dysplasias are divided into were composed of mature epithelium, whereas the epi- 2 grades, but there are problem areas in diagnosis at the thelium in the colonic lesions of the same name was most- lower end where low-grade dysplasias overlap with regen- ly immature or at least less mature. In addition, at the time erating epithelia and in the middle where low- and high- of this naming, the concept of dysplasia as a cancer pre- grade dysplasias overlap. The diagnosis of dysplasia is too cursor was not well established, and the fact that the ep- subjective with less than optimal reproducibility to be as ithelium contained in those colonic adenomas was the useful a marker as needed. Pathologists need a dysplasia common colonic cancer precursor and identical to what stain or a whole set of new markers of high cancer risk, we now call dysplasia was not appreciated. Adenomas of presumably molecular and/or genetic, that are not depen- the gut are dysplasias and should be called by the dys- dent on pathologists’ diagnoses of dysplasia and their in- plasia name. However, the term adenoma for noninvasive herent subjectivity. gut neoplasms is so much a part of our jargon that it will (Arch Pathol Lab Med. 2005;129:170–173) take an act of Congress or a natural catastrophe to get rid of it. e have identified several models in the gastrointes- The answer to the second question, just what is this tinal tract in which there are postulated interme- dysplasia thing, is more complex and is the major subject W of this discussion. To be perfectly honest, the answer is diates between normal mucosa and invasive adenocarci- 1,2 noma. The most common of these models is typical or elusive and may not even be obtainable. Certainly, many ordinary colorectal carcinoma in which the intermediate people have tried to tell us what dysplasia is; there is a stage was given the name adenoma. Other such models are huge body of literature devoted to discussing and describ- adenocarcinomas that arise in the esophagus in Barrett ing dysplasia in various parts of the gut, and every text- mucosa, in the stomach in atrophic gastritis, and in the book of gastrointestinal pathology devotes plenty of space colon and rectum in chronic inflammatory bowel diseases, to it. There are so many references to dysplasia, that I have 3–9 mainly ulcerative colitis. In these cancer situations, the in- limited the reference list here to only a few examples. termediate stages, which look very much like histologic Dysplasia was defined as unequivocally neoplastic epithe- adenomas, are not called adenoma but instead are called lium by a group of supposedly experienced gastrointes- dysplasias. Why are there 2 different names for such his- tinal pathologists who were studying a large number of tologically similar adenocarcinoma intermediates, and just epithelial changes associated with chronic colitis, mostly 10 what is this dysplasia thing anyway? ulcerative colitis. The results of that study, copiously il- The answer to the first question is the easiest. The term lustrated, were published in 1983; by now, it is old stuff. adenoma had been used for benign nonsquamous epithelial The definition they provided sounds fabulous, but exactly neoplasms, such as those with adnexal differentiation in how do we know what is unequivocally neoplastic? This decision is made by pathologists. Those supposedly ex- perienced gastrointestinal pathologists never said who Accepted for publication October 7, 2004. was responsible for making the critical decisions in spe- From the Department of Pathology, University Hospitals, Ann Arbor, Mich. cific cases as to what changes constituted unequivocally Presented at a pathology symposium held in conjunction with the neoplastic epithelium. How did those or any other pa- 7th World Congress of the International Organization for Specialized thologists learn how to make that decision? They were Studies on Diseases of the Esophagus, UNESCO Headquarters, Paris, able to make the decision because of several factors. The France, September 3, 2003. most important factor is their training. Perhaps someone The author has no relevant financial interest in the products or com- older and wiser taught them the truth during their resi- panies described in this article. Reprints: Henry D. Appelman, MD, Department of Pathology, Uni- dencies, and maybe they took postgraduate courses from versity Hospitals, 1500 E Medical Center Dr, Room 2G332, Ann Arbor, alleged experts or sent cases to such experts. After all, the MI 48109-0054 (e-mail: [email protected]). reason we have experts is to teach us stuff, to give courses, 170 Arch Pathol Lab Med—Vol 129, February 2005 Dysplasia in the Gastrointestinal Tract—Appelman to act as consultants, and to write papers such as this one. a sequence that begins with native epithelium altered by Otherwise, why would we need them? Second, personal a combination of genetic and environmental factors. We experience plays a role, including clinical follow-up, which preach that this in turn leads to the development of first gave these pathologists information as to what happened low-grade dysplasia, which then gives rise to high-grade to people they labeled as having this unequivocally neo- dysplasia and finally to invasive carcinoma. Perhaps much plastic epithelium. A third factor, which is not subject to of this supposition is based on the fact that epithelia di- debate, is the issue of guesswork, meaning that we all agnosed as low-grade dysplasia at a particular site are less make diagnostic guesses as best we can about many likely to be accompanied by invasive carcinoma than are things with which we are either unfamiliar or have limited epithelia diagnosed as high grade. Furthermore, when familiarity, and we offer these guesses as true diagnoses. low-grade dysplasia is found, the future risk of developing The next logical question is, how do we know whether invasive carcinoma at the specific site is less than that for this training, experience, and diagnostic guesswork are high-grade dysplasia, and it takes longer for carcinomas applicable to the particular problem? The answer is, we to develop. However, there is virtually no information on don’t! As a result, all surgical pathologists who have var- what happens over time to a specific focus of low-grade iable training and variable experience are given the re- or high-grade dysplasia. Thus, dysplasias are definitely sponsibility of making determinations, and they have risk indicators for carcinoma, but there is little proof that done the best they can with little help because there are they are precursors. no clear sets of diagnostic criteria for each change they see Pathologists recognize many histologic alterations that in biopsies, even in the publication that summarized the are considered dysplastic changes. In the figures present- analyses of that distinguished group of supposedly ex- ed here, Barrett mucosa is used as a model. The alterations perienced gastrointestinal pathologists. Therefore, the def- are both architectural and cytologic. The architectural inition of dysplasia must include the fact that it is the in- changes involve differences from normal structure, such vention of pathologists, and it is both our responsibility as clustering and crowding of tubules, an exaggerated vil- and our fault. One of the most truthful definitions of dys- lous surface pattern, or the appearance of bridges of cells plasia comes from one of my colleagues: ‘‘dysplasia is like that cross lumens. This last feature results in the classic pornography; I know it when I see it’’ (J. K. Greenson, cribriform pattern of growth, a form of architectural com- MD, Professor of Pathology, University of Michigan, oral plexity. Other architectural changes involve the relation of communication, July 2003). the cells and their nuclei to other cells and are recognized Although the original dysplasia definitions and illustra- as nuclear and cellular stratification (Figures 1 through 4). tions were based on changes associated with inflamma- The cytologic features are more complex and involved tory bowel diseases, mainly ulcerative colitis, these defi- both cytoplasm and nuclei. The cytoplasmic features most- nitions have been applied to other settings, such as colonic ly involve loss of maturation or specialization, such as de- adenomas, Barrett mucosa, and atrophic gastritis, but the crease in or loss of mucus content. The nuclear changes carcinomas and their intermediates evolve in remarkably include enlargement with increase in the ratio of nucleus different settings. Colonic adenomas arise in a background to cytoplasm, pleomorphism (ie, variation in size and of histologically normal, although presumably genetically shape), hyperchromatism (ie, increase in nucleic acid con- altered, colorectal mucosa. Barrett esophagus is a meta- tent causing the nuclei to stain more darkly than normal), plasia in which normal squamous mucosa in the lower and increase in the number of mitotic figures and unusual esophagus is replaced by columnar mucosa, as a result or strange looking mitoses (Figures 1 through 8). These (we are told) of chronic gastroesophageal or duodenogas- histologic changes are accompanied by numerous genetic troesophageal reflux. Atrophic gastritis is also accompa- variations, but at the moment, identification of such chang- nied by metaplastic changes, although they are not as dra- es is not useful in routine diagnosis. matic as the Barrett replacement of squamous by columnar Dysplasias can be separated into as many histologic epithelium. In the gastric case, new types of cells, includ- grades or subsets as pathologists are able to recognize. As ing intestinal and mucus gland, replace the normal gastric an example, the squamous dysplasias in the uterine cervix, epithelium, usually as a result of chronic Helicobacter pylori also known as cervical intraepithelial neoplasia, were di- infection. In ulcerative colitis, the mucosa is not normal vided into 3 grades. Gastrointestinal dysplasias involving nor is it metaplastic. Instead, it is altered by many years columnar mucosa, including the metaplastic columnar of inflammation, with repeated bouts of damage followed mucosa of Barrett mucosa in the esophagus, have been by healing, although this sequence of events is not the divided into 2 grades, low and high, each of which leads constant background for the cancers. Because these set- to specific clinical responses for management of the pa- tings are so different, it may not be totally appropriate to tient. Thus, in the gastrointestinal tract dysplasia nomen- apply colitic issues to noncolitic situations, but that is ex- clature, there is no such thing as mild, moderate, or severe actly what we do. dysplasia. The clinicians know this, and the pathologists Dysplasia is said to be a cancer precursor, but how do must comply with this rigid nomenclature. To complicate we know that? We believe that it is a cancer precursor things, these mucosae are subject to many punishments, because we find invasive carcinomas in fields of dysplasia, which induce inflammation, epithelial damage, and con- mainly the high-grade type. Dysplasia is also a marker of current regeneration. Regenerating epithelium shares high cancer risk, but how do we know that? We have lots some characteristics with dysplastic epithelium, including of data indicating that once dysplasia develops, the like- loss of maturation and increased proliferation, which lihood of cancer occurring in that site in greatly increased means that regenerative and dysplastic epithelia can look over that for normal tissue. In addition, when we identify very much alike microscopically. The difference between dysplasia, we find cancers not just at the sites of dysplasia them is physiologic. Regenerating epithelium undergoes but elsewhere in the same organ. There is a widespread maturation and proliferation is reduced once it has accom- belief that carcinomas throughout the body are the end of plished its goal of repairing the damage. Dysplastic epi- Arch Pathol Lab Med—Vol 129, February 2005 Dysplasia in the Gastrointestinal Tract—Appelman 171 Figure 1. Dysplastic mucosa in which the tubules are more crowded toward the sur- face than at the base, and there is great var- iation in size and shape of the tubules (he- matoxylin-eosin, original magnification ϫ100). Figure 2. There is little difference in lo- cation of the crowded tubules, but they are even more architecturally complex (hema- toxylin-eosin, original magnification ϫ100). Figure 3. Villous architecture with cover- ing epithelium that has cytoplasmic matu- ration, as indicated by apical mucin. The nuclei are hyperchromatic but not pleo- morphic. The nuclear stratification is in the basal half to two thirds of the cells, except for a few foci where it approaches full thick- ness (hematoxylin-eosin, original magnifi- cation ϫ200). Figure 4. Villous architecture with cover- ing epithelium that has only focal cytoplas- mic maturation. The nuclei are hyperchro- matic as in Figure 3, but the stratification is almost all full thickness. However, there is little stratification (hematoxylin-eosin, orig- inal magnification ϫ200). Figure 5. Complex tubular architecture with lining epithelium that has almost no cytoplasmic maturation. The nuclei are stratified full thickness, hyperchromatic, and pleomorphic (hematoxylin-eosin, orig- inal magnification ϫ200). Figure 6. Variably complex tubular archi- tecture with virtually no cytoplasmic mat- uration. There is also high variability in de- gree of nuclear stratification and hyperchro- matism. Compare the highly stratified, uni- form nuclei in the tubule at the lower left with the much less stratified but more pleo- morphic and vesicular nuclei in the tubule at the right edge of the field (hematoxylin- eosin, original magnification ϫ200). Figure 7. Tubules on the left and right sides have a similar shape, but the lining epithelia are different. The epithelium in the left tubule has more apical mucus and more pleomorphic, more vesicular, less hyper- chromatic, and less stratified nuclei, where- as in the right tubule there is no apical mu- cus, and the nuclei are more uniform, more hyperchromatic, and more stratified (he- matoxylin-eosin, original magnification ϫ200). Figure 8. High variability in this single field of apical cytoplasmic mucin, vesicular nuclei, pleomorphism, hyperchromatism, and stratification. Two tubules with apical mucus in all the cells are not dysplastic, but are all the other cells dysplastic or are some of them regenerative, such as the syncytial epithelium with vesicular nuclei in the mid- dle of the field (hematoxylin-eosin, original magnification ϫ200)? thelium, in contrast, never matures. This overlap in his- histopathology. However, it is impossible to define this tologic features between regenerating and dysplastic epi- entity because, as the name indicates, it has no definition, thelium was recognized and led to the concept of epithe- and it is equally impossible to illustrate and teach it, be- lium that defied classification as to whether it was cause the diagnosis of indefinite for dysplasia lacks con- regenerative or dysplastic; such epithelium was named in- sistency. When I photograph some epithelium I think is definite for dysplasia, one of the most honest inventions in indefinite for dysplasia and then look at the photographs 172 Arch Pathol Lab Med—Vol 129, February 2005 Dysplasia in the Gastrointestinal Tract—Appelman a few days later, that epithelium now looks either negative of histologic features that vary in intensity from one case or low grade. to the next and are therefore not as reproducible as we These cytologic and architectural changes do not all oc- would like. The diagnosis of dysplasia, especially low cur at the same rate in all epithelia. When a pathologist grade, is just plain hard to make! In real life daily diag- decides whether an epithelium is dysplastic and the grade nostic pathology practice, how do we decide what is low- of dysplasia, the decision is made not on individual fea- grade dysplasia? We evaluate all the architectural and cy- tures but on the aggregate of features. Therefore, not all tologic features and decided whether they are intense low-grade dysplasias have the same changes, and the enough for the low-grade diagnosis. Sometimes, we con- same is true for all high-grade dysplasias. Low-grade dys- sult other people to see whether they agree. How do we plasia, in simplest terms, has a greater degree of change diagnose high grade? Again, we try to determine whether than is acceptable for either normal or regenerative epi- those same architectural and cytologic features look worse thelium, but the changes are less intense than those as- than those seen in cases of low-grade dysplasia. Some- sociated with high-grade dysplasia. Changes associated times we consult other people to see whether they agree. with high-grade dysplasia are more intense than those as- If the diagnosis of dysplasia is fraught with so much sub- sociated with low-grade dysplasia. Because all low-grade jectivity and so much inconsistency, isn’t there some way and high-grade dysplasias contain epithelia with changes to alleviate the difficulties and inconsistencies? The best of variable intensity, the epithelia in both types are not solution would be for us to have a dysplasia stain or a homogeneous, but each type contains a group of different specific dysplasia antibody, but these are not available. epithelia that have been given the same names. What we are really looking for is some definable, consis- There are some consequences inherent in this set of con- tent, objective change that informs us that the risk of both clusions. First, because regenerating and low-grade dys- concurrent and future carcinoma is high enough for some plastic epithelia have so much in common, because the kind of clinical intervention. This is what we hoped dys- category of indefinite for dysplasia recognizes this over- plasia would be, but the inconsistencies get in our way. So lap, and because there are so many histologic criteria in- we need a substitute, a surrogate for dysplasia. This sub- volved in the diagnoses of these changes, the reproduc- stitute cannot be judged against our current diagnoses of ibility for diagnosing regenerative, indefinite, and low- dysplasia, or it will suffer from the same lack of objectiv- grade dysplastic epithelia is poor. In a number of studies ity. This substitute must be judged strictly by clinical out- of both colitis and Barrett mucosa, this hypothesis has come, and it will probably be molecular and/or genetic, been supported.11–14 However, because the changes in- incorporating reproducible measurements that are not crease in intensity as the grade of dysplasia increases, in- subject to individual interpretation.15 Meanwhile, because cluding the appearance of many features common to car- we do not have such markers, we must plod along, doing cinomas, there is better reproducibility in the diagnosis of the best we can, using old-fashioned histologic interpre- high-grade dysplasia. Second, because only 2 grades of tation, and in the process we hope that we do not do too dysplasia are recognized and each grade is a collection of much harm to the patients. epithelia with changes occurring at different rates in dif- References ferent epithelia of the same diagnosis, then there must be 1. McKenna BJ, Appelman HD. Dysplasia of the gut: the diagnosis is harder some overlap between the collection called low grade and than it seems. J Clin Gastroenterol. 2002;34:111–116. that called high grade, and the distinction is not always 2. McKenna BJ, Appelman HD. Dysplasia can be a pain in the gut. Pathology. 2002;34:518–528. going to be easy. This situation is exactly what occurs in 3. Guarner J, Herrera-Goepfert R, Mohar A, et al. Diagnostic yield of gastric practice. Some epithelia seem to have a mixture of low- biopsy specimens when screening for preneoplastic lesions. Hum Pathol. 2003; and high-grade features. Unfortunately, the classification 34:28–31. 4. Geboes K, Van Eyken P. The diagnosis of dysplasia and malignancy in Bar- of dysplasia into 2 grades (low and high) does not allow rett’s oesophagus. Histopathology. 2000;37:99–107. for these mixed features, so we try to put such epithelia 5. Haggitt RC. Barrett’s esophagus, dysplasia, and adenocarcinoma. Hum into one grade or the other. Pathol. 1994;25:982–993. 6. Lauwers GY, Riddell RH. Review: gastric epithelial dysplasia. Gut. 1999;45: To complicate this issue even more, when we read any 784–790. article that deals with dysplasia of any grade, we must 7. Rugge M, Correa P, Dixon MF, et al. Gastric dysplasia: the Padova inter- recognize that these dysplasias are not single entities but national classification. Am J Surg Pathol. 2000;24:167–176. 8. Schlemper RJ, Riddell RH, Kato Y, et al. The Vienna classification of gastro- rather groups of entities, all of which have been given the intestinal epithelial neoplasia. Gut. 2000;47:251–255. same diagnosis. What the pathologist authors of one study 9. Skacel M, Petras RE, Gramlich TL, et al. The diagnosis of low-grade dysplasia have diagnosed as low- or high-grade dysplasia, regard- in Barrett’s esophagus and its implications for disease progression. Am J Gastroen- less of the experience and/or reputation of those pathol- terol. 2000;95:3383–3387. 10. Riddell RH, Goldman H, Ransohoff DF, et al. Dysplasia in inflammatory ogists, will not necessarily be diagnosed the same way by bowel disease: standardized classification with provisional clinical applications. the pathologist readers. Furthermore, pathologist authors Hum Pathol. 1983;14:931–968. of different studies of the same diagnostic category are 11. Dixon MF, Brown LJR, Gilmour HM, et al. Observer variation in the as- sessment of dysplasia in ulcerative colitis. Histopathology. 1988;13:385–398. not always talking about the same epithelia. Therefore, all 12. Melville DM, Jass JR, Morson BC, et al. Observer study of the grading of articles dealing with dysplasia of any grade are not nec- dysplasia in ulcerative colitis: comparison with clinical outcome. Hum Pathol. essarily discussing the same entities, and such studies 1998;20:1008–1014. 13. Montgomery E, Bronner MP, Goldblum JR, et al. Reproducibility of the cannot be considered comparable. Thus, meta-analyses in diagnosis of dysplasia in Barrett’s esophagus: a reaffirmation. Hum Pathol. 2001; this area may not be useful. 32:368–378. Now we can finally complete the definition of dysplasia. 14. Reid BJ, Haggitt RC, Rubin CE, et al. Observer variation in the diagnosis of dysplasia in Barrett’s esophagus. Hum Pathol. 1988;19:166–178. It is unequivocally neoplastic epithelium, an invention of 15. Berman JJ, Henson DE. The precancers: waiting for a classification. Hum pathologists, the diagnosis of which depends on a group Pathol. 2003;34:833–834.

Arch Pathol Lab Med—Vol 129, February 2005 Dysplasia in the Gastrointestinal Tract—Appelman 173