Simple Clefts, Complex Problems: Reflections on Glandular Lesions of the Uterine Cervix

This issue of the Journal contains another contribution on the immunoprofile of endocervical adenocarcinoma, with special reference to its role in the distinction of that neoplasm from endometrial adenocarcinoma (1). This falls hard on the heels of two other recent contributions to the Journal on that topic (2,3), as well as recent symposia organized by our society on glandular lesions of the uterine cervix at the 18th Congress of the European Society of Pathology in Berlin in September and the recent meeting of the US–Canadian Academy of Pathology (USCAP) in Chicago. These are very obvious indications of the ongoing issues associated with this important category of lesions. The Editor-in-Chief of the Journal has asked me to make some comments on this area to accompany the original contribution of Kamoi and colleagues (1) in this issue. Broadly speaking, glandular lesions of the uterine cervix can be considered in four categories: tumor-like lesions, precursor lesions in the category of the nebulous entity endocervical glandular dysplasia and the better defined entity of adenocarcinoma in situ, typical invasive adenocarcinoma, and, finally, variants of cervical neoplasia with a glandular, as well as non-glandular, component. Anyone who has a busy consultation practice in gynecologic pathology sees many cases in which the issue revolves around whether a peculiar endocervical glandular proliferation is benign or malignant. From the historical perspective, it is quite remarkable the extent to which tumor-like lesions have not attracted much attention overall until recently. However, it should be noted that in his wonderful book of almost a century ago, Cancer of the Uterus, Dr. Thomas S. Cullen included among his four chapters on adenocarcinoma of the cervix one on differential diagnosis. His first illustration is of what we might today call microglandular hyperplasia. He noted with his characteristic certainty that he was “at a loss to know how this condition is produced, but have not the slightest hesitancy in saying that it is benign” (4). That contribution notwithstanding, if one excludes the flurry of interest in the late 1960s when microglandular hyperplasia was described as an entity by several groups and named by one of them (5), it is only since 1990, with the study of tubal metaplasia by Suh and Silverberg (6), that this area of gynecologic pathology has really seen a major surge of interest. Another study on this area in the following year by Ismail (7) further expanded upon this topic by noting that in some such cases the epithelium had an endometrioid as much as a tubal appearance. She additionally pointed out that there was an association between “tuboendometrioid” metaplasia of the uterine cervix and a history of a recent procedure, just as there is for endometriosis, as has long been known. A subsequent study by Baker et al. (8) was devoted to endometriosis of the cervix alone and highlighted its potential confusion with adenocarcinoma in situ and that subtle cases may be confused with endocervical glandular atypia-dysplasia. The well-known entity of tunnel clusters, although perfectly described by Dr. Frederic Fluhmann (9,10) in 1961, was somewhat overlooked by many pathologists until being “rediscovered” by Segal and Hart (11). Two enigmatic tumor-like lesions are recently described: lobular endocervical gland hyperplasia (12–16) and “ectopic” prostate (17,18) in the uterine cervix. The former is thought by some Japanese workers (who consider a pyloric nature of the epithelium a distinctive feature) to be a precursor of adenoma malignum. The occurrence in the cervix of glandular tissue that on routine and immunohistochemical evaluation is indistinguishable from prostate tissue is even more intriguing. When the first case was encountered by us, it was deemed so unusual that we did not envision we would ever see a case again and only rather timidly referred to it in a book (19). When another group reported a case (17) and others came our way, an original report seemed indicated, and in due course four such cases were described by Nucci et al. (18) in 2000. Our current experience is now double that which was documented at that time. Although we elected to report our cases as “ectopic” prostatic tissue, I now believe it more likely represents an intriguing metaplastic change, although no proof exists in either direction. This lesion is associated with a differential diagnosis, including the adenoid basal carcinoma or adenoid basal epithelioma (see below), because of a somewhat basaloid appearance to the squamous metaplasia that may be encountered in such cases, and small acini that may resemble the small tubules of the neoplasm just noted. Glandular hyperplasia, not otherwise specified, or conceivably even some peculiar form of well-differentiated adenocarcinoma, could be considerations in other cases of a more florid proliferation of prostatic-type tissue. One wonders what some of our eminent predecessors, such as Dr. Rupert Willis (20), who took such a particular interest in ectopias, would make of this phenomenon! Just as the tumor-like lesions are largely a matter of interest in recent times so also are the precursor lesions of adenocarcinoma. In 1953, Friedell and McKay (21) documented the first case of adenocarcinoma in situ and there was little activity in this area until the late 1970s, but since that time there have been numerous papers on the topic. Issues associated with this lesion have included its distinction from early invasive adenocarcinoma. In some cases, early invasion is seen in the form of single cells and small clusters of cells invading from a gland showing in situ neoplasia. In many cases, however, invasion is manifest by subtle architectural features best appreciated on low power. The diagnosis of adenocarcinoma in situ rests on the neoplastic cells being within glands that are consistent with normally arranged and located endocervical glands, whereas in invasive adenocarcinoma, their architectural disposition is indicative of stromal penetration. Gland confluence, irregularity of the arrangement of the glands one to another, and a cribriform pattern, to varying degrees in individual cases, attest to the presence of invasion. There may be little or no stromal reaction in these cases, which can add to the diagnostic difficulties. Other issues associated with early invasive adenocarcinoma are largely of a clinical nature with regard to the prognosis and the influence of the depth of invasion on the surgical procedure to be performed. These issues will be considered in Dr. Zaino's review of precursor lesions and early invasive adenocarcinoma, which, along with the other contributions from the symposium of the Society held at the 2002 USCAP meeting, will appear in the next issue of the Journal and accordingly will not be considered here. Whether we can confidently recognize dysplastic changes that are below those of adenocarcinoma in situ remains a controversial topic. Indeed, at the same recent USCAP meeting, the Papanicolaou Society held a seminar (also to be published in the next issue) that included speakers for (Dr. Silverberg) and against (Dr. Lee) the issue as to whether pathologists should diagnose endocervical preneoplastic lesions “less than” adenocarcinoma in situ. That two senior pathologists who have taken such an interest in the area can feel quite differently about the topic exemplifies the controversy surrounding it. To say that more work needs to be done in this area by the pathology community collectively is an understatement. It is within the area of invasive adenocarcinoma that I may have some views that contrast with those of some others who have taken a major interest in this area, and I certainly would not be so bold to pronounce that my own views are necessarily correct, although they are based on significant personal experience. Although the most common form of endocervical adenocarcinoma is often referred to as “mucinous adenocarcinoma,” approximately 80% of cervical adenocarcinomas that I encounter, which can be considered the “usual form,” do not have a conspicuous mucinous nature on hematoxylin and eosin evaluation (22). They do, however, have a rather characteristic appearance, with cells containing eosinophilic cytoplasm and exhibiting brisk mitotic activity with frequent apoptotic bodies. Historically they have had the descriptor “mucinous” applied, presumably because of their cell of origin. I have no great personal experience with mucin stains of this lesion because it is not indicated in practice, but the few I have stained have been negative or have shown no more cytoplasmic mucin than is present in a number of other adenocarcinomas of the genital tract that are not classified as mucinous. There is, as is so often the case with neoplasia in general, a spectrum of cases from the most common form of endocervical adenocarcinoma that is mucin poor, to tumors that have varying proportions of mucin-poor and mucin-rich cells, to tumors that are overtly mucinous with hematoxylin and eosin and mucin stains. These latter tumors are often histologically differentiated but are obviously malignant given their combined architectural and cytologic features. In contrast, one notorious subtype of mucinous adenocarcinoma is the adenoma malignum (23), a tumor that has an abundant component of glands that, viewed in isolation, are cytologically indistinguishable from normal and may not be obviously malignant in areas given their architectural features. This is one of the most treacherous entities in surgical pathology and certainly seems to be the driving force behind the submission of many benign glandular lesions in consultation. Without doing a specific tally, I would hazard that for every case of adenoma malignum received in consultation, we probably see 30 cases of benign glandular proliferations that have suggested the diagnosis of adenoma malignum. That observation indicates the rarity of adenoma malignum, and indeed the good prognosis in some earlier reports in the literature of this entity make one wonder if review of those cases in the light of current knowledge might lead to reclassification of some of them as one of the recently described benign glandular lesions. It is helpful that in most cases of adenoma malignum in which a cone biopsy or a hysterectomy specimen is available for review, there is focal atypicality as well as features indicative of unequivocal invasion. If adenoma malignum is a relatively “old” entity in the classification of the endocervical adenocarcinomas, a much newer comer to the area is the well-differentiated villoglandular papillary adenocarcinoma. It is associated with its own problems. We highlighted the lesion because of the good prognosis in the 13 cases we initially encountered (24). Furthermore, in several of our cases, a hysterectomy subsequent to a cone biopsy showed no residual neoplasia, suggesting that perhaps the cone biopsy might have been sufficient treatment and the often young patients in these cases could have retained their uterus. Others subsequently have also shown that this tumor is associated with a good prognosis and can often be managed conservatively (25,26). However, it is crucial to remember that adenocarcinomas of the usual type (including deeply invasive tumors) may have a villous papillary component, and thus the diagnosis of villoglandular papillary adenocarcinoma should be reserved for lesions that are unassociated with an underlying component of conventional adenocarcinoma. It is best not to diagnose villoglandular papillary adenocarcinoma unequivocally (but only suggest the diagnosis) in biopsy material and require at least a cone biopsy specimen to use the term definitively because of the clinical implications associated with its usage. It is also a term best restricted to tumors that have exclusively grade 1 nuclear features. Moreover, there is a tendency to erroneously call some endocervical adenocarcinomas of the usual type well differentiated, most of which are uniformly gland forming, even though they are briskly mitotic and have at least moderate nuclear atypicality. The “well differentiated” label should be used cautiously because its inappropriate usage may give a favorable connotation to a lesion that, if significantly invasive, may be a major threat to the patient. Indeed, it is the size of the tumor, its depth of invasion, and the presence or absence of vascular invasion that are as important, perhaps more important, than its grade. Another perplexing aspect of pure adenocarcinomas pertains to the great variation in the frequency of endometrioid adenocarcinoma reported by different workers. As an overt example of that, in a recent editorial on the topic of endocervical adenocarcinoma and its distinction from endometrial adenocarcinoma, someone with great experience in this area indicated that in his material endometrioid adenocarcinomas accounted for about 50% of endocervical adenocarcinomas (27). On the other hand, in a recent combined series from the Brigham and Women's Hospital and the Massachusetts General Hospital, endometrioid carcinomas accounted for only 7% of the cases (28)! These greatly discrepant results, which are represented by other studies in the literature as well, can only be due to the inherent subjectivity in subclassifying adenocarcinomas that have more in common with each other than they have differences. Almost certainly, what we would consider a “typical” or “usual” endocervical adenocarcinoma with no or little cytoplasmic mucin (as noted above) would be considered endometrioid by others. I have rarely diagnosed endometrioid adenocarcinoma of the cervix, but whether I am undercalling it or others are overcalling it is a matter on which I am not the appropriate one to judge. Nonetheless, it is yet another area for us all to collectively formulate criteria that will lead to a more uniform approach to the topic. The good news is that our differing labels are not likely to be causing much harm to the patient, if indeed any. The contribution by Kamoi and co-authors (1) in this issue of the Journal, which in part prompted this editorial, represents yet another exploration of the role of immunohistochemistry in this area. I will not comment in detail on the subject if only because I claim no expertise in it and leave it for the reader to compare the findings in this study with the other recent papers in the Journal on this topic. Towards the end of their contribution, Kamoi et al. (1) take a rather provocative approach when, with knowledge of the mere frequency of the various types of adenocarcinoma in the endometrium and endocervix, they surmise that they could predict the correct diagnosis in just over 80% of the cases! Indeed, when I recalculated using what I believe is the frequency of endometrioid carcinoma of the cervix (generously 10%), the accuracy was just over 90%! Not very scientific but interesting nonetheless. I encourage others to explore this area by immunohistochemically evaluating their own material in conjunction with routine morphology, although I personally feel that, as Kamoi and co-authors (1) indicate, the latter is more beneficial in this area, and indeed the aid provided by immunohistochemistry is not matched by the time and expense in using it. That is certainly the case given that the great majority of endometrial adenocarcinomas are endometrioid, and the great majority of endocervical adenocarcinomas (in my opinion) are of the usual endocervical type (as discussed above), and the two look distinctly different on routine evaluation. When the coloscopic and hysteroscopic findings, as well as the presence of associated atypical endometrial hyperplasia in some cases of endometrial carcinoma and associated adenocarcinoma in situ in some cases of endocervical adenocarcinoma, are factored in, the number of cases in which immunohistochemistry is of major help are few. The other main variants of endocervical adenocarcinoma generally parallel in cell type their counterparts in the uterine corpus and ovary, but there are important differences in their frequency. For example, serous carcinoma of the cervix is exceedingly uncommon (29) and, intriguingly, based on a small number of cases, does not appear to be as aggressive as its counterpart in the uterine corpus. The mesonephric subtype of adenocarcinoma is largely restricted to the uterine cervix, and, at this site, recent work has led to an expansion of its morphologic spectrum (30,31). Although many tumors have a component of tubular glands with luminal eosinophilic material (a characteristic feature of mesonephric remnants and mesonephric hyperplasia), other tumors may have a prominent endometrioid-like glandular pattern or an appearance easily mistaken for serous carcinoma. Finally, an interesting subgroup of endocervical carcinomas are those that have a glandular component and a non-glandular component, usually squamous but rarely of another morphologic type. One of the most important of these is the lesion that has most often been referred to as “adenoid basal carcinoma” (32), but the good prognosis of this entity has led one group to prefer the designation “adenoid basal epithelioma” (33) to help avoid inappropriately aggressive therapy. As I hope will be self evident from the above remarks, the area of cervical glandular pathology represents one of the most diagnostically challenging areas for the gynecologic pathologist and one of those areas associated with some noteworthy differences with regard to the frequency with which various tumors are diagnosed, and even the terminology applied. It is a fruitful area for ongoing research with immunohistochemical and molecular studies, but I also hope that we will subject these lesions to intense conventional morphologic analysis. I think we still have much to learn from that time-honored approach that still remains, and in my opinion and more notably in the opinions of two giants in the field of surgical pathology (34,35), will remain, the most reliable, cost effective, and timely manner of evaluating tumors for the lifetime of all who read this, and likely their heirs. As the first president of our Society has eloquently noted elsewhere, recently the “good-old” H and E slide has indeed still much to offer (35).