GLANDULAR LESIONS OF THE UTERINE CERVIX

Robert H. Young, M.D., FRCPath

Professor of Pathology, Harvard Medical School, Boston, Massachusetts,

Pathologist and Director of Surgical Pathology,

James Homer Wright Pathology Laboratories,

Massachusetts General Hospital, Boston

In the first part of my talk benign glandular lesions are reviewed with emphasis on the propensity for many to be misdiagnosed as malignant. The potential for these lesions to cause diagnostic difficulty has been highlighted in many recent reports and they account for many problems in diagnosis based on our experience with consultation material.  Most of the processes reviewed are not even neoplastic and these pseudoneoplastic lesions (Table 1) are considered first before benign glandular neoplasms which may also be problematic, albeit less often.  The references provided are primarily from the last two decades but important, sometimes seminal, contributions from the earlier years will also be included because knowledge of some entities is incomplete if crucial older papers are not reviewed.  

Pseudoneoplastic Lesions

The extent to which this area has been the source of major discussion in the literature only relatively recently is quite striking.  If books from as recently as the late 1970s are reviewed, no or little mention is often made of most of the processes considered here.  Given the florid nature of many of them, it is hard to avoid the conclusion that in the past many were the cause of what we would now consider erroneous diagnoses of adenocarcinoma.  This is clearly documented in the case of one of the entities described earlier that most of those considered here, microglandular hyperplasia, as will be considered further under that heading.  Additionally, in one review of 13 cervical “adenocarcinomas” seen at one institution over a period of 14 years (1) five were considered benign upon review.  There are few studies that look at the frequency of these lesions in routine material.  The results from one such (2) study are presented in Table 2.  It should be noted that that study was performed before several entities (two in particular: lobular endocervical gland hyperplasia and diffuse laminar endocervical gland hyperplasia) were characterized.  In Table 1, pseudoneoplastic glandular lesions are subdivided into three broad categories:  endocervical gland hyperplasia, metaplasias and ectopias, and reactive and inflammatory lesions.  Deep glands and cysts are not hyperplasias but fit most neatly with them.  Endosalpingiosis and endocervicosis are of peritoneal derivation but when they involve the cervix, consideration with the closely related lesions, tubal metaplasia and endometriosis, is most logical. 

Table 1

Pseudoneoplastic Glandular Lesions of Uterine Cervix

Endocervical Gland Hyperplasias

Tunnel clusters

Deep glands and cysts

Lobular endocervical gland hyperplasia, not otherwise specified

Diffuse laminar endocervical gland hyperplasia

Microglandular hyperplasia

Mesonephric hyperplasia

Glandular hyperplasia, not otherwise specified

----------------------------------------------------------------

Metaplasias and Ectopias of Endocervical Glands

Tubal, tubo-endometrioid and endometrioid metaplasia

Endometriosis

Endosalpingiosis

Endocervicosis

Atypical oxyphilic metaplasia

Intestinal metaplasia

Arias-Stella reaction

Ectopic prostate

----------------------------------------------------------------

Reactive and Inflammatory Lesions            

Papillary endocervicitis

Infectious atypias

Radiation atypia

 Changes secondary to mucin extravasation

Table 2

Number of Benign Endocervical Glandular Lesions in a

Consecutive Series of Hysterectomy (473) and Cone Biopsy Specimens (33) (Ref 2)

Tubal Metaplasia and Related Lesions                                 84+

Type B Tunnel Clusters                                                          71

Type A Tunnel Clusters                                                             29

Microglandular Hyperplasia                                                   28

Deep Nabothian Cysts                                                            7

Mesonephric Remnants/Hyperplasia                                    7

Mucin Extravasation                                                             6

Endometriosis                                                                         5

+includes 10 cases of endometrioid metaplasia

ENDOCERVICAL GLANDULAR HYPERPLASIAS

Tunnel Clusters

           Fluhmann (3,4) introduced the designation "tunnel cluster" for this common lesion that is usually an incidental microscopic finding but may cause a striking gross abnormality.   Tunnel clusters were thought by him to represent an involutionary stage of endocervical glands (clefts) that had originally become hyperplastic under the influence of pregnancy.   He found tunnel clusters in 40 percent of women in the first trimester of pregnancy.  Although an association with parity is often stressed, it should be noted that 4 of the 30 patients in Fluhmann’s original report were nulligravida (4).  The lesions are usually found in women over 30 years.  Two types of tunnel clusters, types A and B, were characterized by Fluhmann, the former non-cystic and lined by columnar epithelium, the latter cystic and lined by cuboidal or flattened epithelium.  Small cysts are seen in some type A lesions and the two groups merge imperceptibly, so their separation is unnecessary for reporting purposes but the issues raised by them are somewhat different and it is appropriate to consider them separately here.  In our experience type B tunnel clusters are considerably more common than type A although Fluhmann (4) found the opposite.  This is probably due to differing criteria being used.   The relatively recent analysis presented in Table 2 found Type B lesions to be more than twice as common as Type A. 

            The only recent study of type B lesions is that of Segal and Hart (5), who identified them in 6 percent of hysterectomy specimens and 10 percent of cone biopsies.  Almost 80 percent of the patients had at least three previous pregnancies and all but one were multigravida.  In their series tunnel clusters were more common in the posterior lip.  Cysts were observed grossly in approximately 40 percent of the cases and significant distortion of the cervical wall occurred in one-third of this group. 

           Type B clusters are typically discrete, usually 0.5-5 mm (mean 2.4 mm) in diameter (but occasionally up to 20 mm), rounded foci composed of 20 to 50, usually oval to round but occasionally irregular, closely packed cystic glands.  Multiple foci are present in about 80 percent of the cases, and occasionally the tunnel clusters are confluent.  The cysts, which typically contain inspissated mucin, are separated by scanty connective tissue, and the tunnel cluster is surrounded by normal endocervical stroma. The mucin does not tend to contain the inflammatory cells so typical of microglandular hyperplasia.  Like other benign disorders of the endocervical glands, type B clusters may occasionally extend deeply into the cervical wall.  Individual very cystic units of a tunnel cluster are indistinguishable from nabothian cysts but others can still be seen to be part of the distinctive lobule that is the defining feature of a tunnel cluster.  Although an erroneous diagnosis of adenoma malignum ("minimal deviation adenocarcinoma") was made initially in two of the 29 cases of type B tunnel clusters in the study of Segal and Hart (5) and we have also seen this mistake, type A clusters in our opinion are more often the case of diagnostic difficulty.

            Although the mistake associated with Type B tunnel clusters is usually this benign lesion being considered malignant, adenocarcinomas with cystic glands can be thought to be tunnel clusters, particularly on low power.  We have seen this usually with the common endocervical-type adenocarcinoma which may be cystic but even cystic clear cell carcinoma has elicited a possible diagnosis of type B tunnel clusters in our experience.  Careful evaluation of cystic glands on high power is important cytologic features of malignancy are seen in the case of the cystic cancers and hobnail and clear cells are not a feature of tunnel clusters although it must be noted that cystic clear cell carcinoma may have a dominant lining of deceptively benign flattened cells; invasive characteristics are usually seen in these cases, however.

            Type A lesions can be problematic because the small closely packed glands of this variant can be very irregular and, most noteworthy, can show degenerative type nuclear atypia that can be alarming if one is not aware of the phenomenon.  We reported 14 cases of this type (6).  As is usual for tunnel clusters the patients were all over 30 years old.  All but one were multigravida.  Most of the lesions were well circumscribed but irregular borders created a pseudoinfiltrative appearance in four cases.  Six cases focally had a cellular, sometimes edematous, stroma.  The stromal change, pseudoinfiltration, and cytologic atypia, as well as a glandular architecture that was more irregular than had been emphasized by previous writers all combined to make some of these cases challenging interpretations.  Most type A lesions are associated with a type B component and this, as well as the lobulation and circumscription that is also a feature of the type A lesion, as it is the type B, should enable distinction from adenocarcinoma including adenoma malignum.  Tunnel clusters also lack the frequent desmoplasia of adenoma malignum and the glands of tunnel clusters are always closely packed and do not infiltrate and become widely separated as in adenoma malignum.  It should be emphasized that some adenocarcinomas may have striking cysts and simulate Type B clusters to a degree but the lining cells always show as least focal cytologic atypia in the cancers (7).

Deep Glands and Cysts

            Endocervical glands occasionally lie deep in the cervical stroma (8). However, they are usually beneath normal superficial endocervical glands, are usually few in number and normal in size and shape, lack cytologic atypia and are unassociated with any stromal alteration, indicating their benign nature.  Similar observations indicate the nature of the benign endometrioid glands that may be encountered deep in the cervical wall (9). 

            Nabothian cysts may also extend deep into the wall, a finding that may cause concern (10).  Gross examination of the cervix shows numerous mucin-filled cysts measuring up to 1 cm in diameter replacing the endocervical wall.  Histological examination shows cysts lined by benign columnar, cuboidal to flattened endocervical epithelium devoid of mitotic activity.  The differential diagnosis is with cystic adenocarcinoma, but the glands of cystic cancers are more irregularly distributed, are lined by cytologically malignant cells and may have an associated stromal reaction (7).

Lobular Endocervical Gland Hyperplasia, Not Otherwise Specified

            Occasionally the most striking architectural feature of endocervical gland hyperplasia is distinct lobulation  (but without the specific features of tunnel clusters) giving rise to the designation "lobular endocervical gland hyperplasia, not otherwise specified" (LEGH) (11).  A series of 13 examples of this process, many of which were problematic with regard to the distinction from carcinoma, has recently been reported by Nucci et al (11) who introduced the term used for this process here.  However, as could be said for all the “recently described” lesions discussed here examples have obviously been encountered in the past.  For example, cases 3 and 4 in a group of unusual endocervical lesions described by Fetissof et al (12) likely fall in this group.  Additionally, cases that appear to meet the criteria for the entity described by Nucci et al. were included for comparison purposes in a study in 1998 that investigated the immunohistochemical staining of adenoma malignum and possibly related lesions (13).  Furthermore, Mikami et al. (14) independently recognized the benign nature of LEGH and published their findings synchronously with Nucci and colleagues.

            The lesions have been encountered throughout the adult years and are most often an incidental microscopic findings but there is a gross abnormality more often than is the case with most of the other pseudoneoplastic glandular lesions being considered here.  For example, 3 of 13 cases were grossly abnormal in Nucci’s series (11) and all three in that of Mikami et al. (14) who include striking gross photographs of two of their cases.  Microscopic examination reveals a lobular proliferation of mostly moderate sized, rounded glands often centered around a large central gland.  The lobules are typically well demarcated and usually confined to the inner half of the cervical wall.  The intervening cervical stroma is unremarkable and the glands within the lobules are lined by columnar mucinous cells similar to those of the normal endocervix.  Although the lobulation of this lesion is a feature shared with tunnel clusters, and in occasional cases it may be difficult to know where to place a given lesion, there is not the particular cystic change of type B tunnel clusters or the small much more irregular glands of a type A tunnel cluster.  Features most helpful in the distinction from adenoma malignum, in addition to the striking orderly lobular arrangement of the glands, are a lack of irregular stromal infiltration, absence of a desmoplastic stromal response, and no evidence of even focal malignant cytologic features, the latter usually being seen in adenoma malignum except in some small biopsy specimens (15).  Some prefer to consider that LEGH represents adenoma malignum, possibly in an in situ phase, and this issue has engendered some controversy (16,17) but although a possible relation to adenoma malignum is intriguing and hard to disprove we and others feel LEGH is benign.  Even those who consider it in the adenoma malignum group found that their six cases were non-metastasizing (14).

            Japanese investigators have emphasized in their writings that LEGH exhibits pyloric gland metaplasia (13,14,16).  Ishii et al. (13) found this in the above mentioned study in which cases of apparent LEGH were analyzed along with cases of adenoma malignum and other entities.  Mikami et al. (14) elected to designate LEGH “pyloric gland metaplasia” because of this aspect.  We have not investigated our cases histochemically but have seen only one case in which by routine light microscopy the glands of LEGH appeared “pyloric”, being unable to distinguish the glands in others, based on their cytoplasm, from otherwise unremarkable endocervical glands in most instances.  Other benign glandular lesions have not been intensively studied for possible pyloric metaplasia.  We feel at this time the designation LEGH optimal but time and the results of additional studies may justify a modification.

Diffuse Laminar Endocervical Glandular Hyperplasia

            This term is used for cases of endocervical glandular hyperplasia characterized by a diffuse distribution of closely packed endocervical glands typically appearing as a discrete layer sharply demarcated from the underlying cervical stroma (18).  The seven examples initially described occurred in patients who ranged from 22 to 48 (mean 37) years of age; only one had a history of hormone intake.  All the lesions were incidental findings.  Microscopic examination revealed a proliferation of closely packed, moderately sized, endocervical glands within the inner-third of the cervical wall.  Reactive cytologic atypia was seen in some cases but significant atypia was absent.  A marked inflammatory response was present in five of the seven cases.  Focal stromal edema was present in six of the cases but a desmoplastic stromal reaction of the type often seen in cases of adenoma malignum was absent.  The latter feature, along with a lack of an irregular and deep stromal infiltration and an absence of focal malignant cytologic features, aid in the distinction of diffuse laminar endocervical gland hyperplasia from adenoma malignum.   One example has been reported under this designation since our initial report but on the basis of the illustrations, the glands were not as closely packed as in our series and as we require for this particular subcategorization (19).

Microglandular Hyperplasia

            This particular process was independently recognized as being benign by four groups in the 1960s (20-23), all of them highlighting an association with oral contraceptive use, two of them also noting an association with pregnancy (21,23) and one introducing the now familiar designation “microglandular hyperplasia” (21).  The potential for misdiagnosis was highlighted strikingly in two of the first reports.  Five cases in the Armed Forces Institute of Pathology Series were referred in with carcinoma favored or seriously considered (20) and in the University of Michigan report three cases were referred in for treatment of cervical adenocarcinoma (22).

            Microglandular hyperplasia (MGH) has been considered unique among the lesions being reviewed here over the years because of the associations with oral contraceptives and pregnancy (20-27) but a recent paper has questioned whether the association with oral contraceptives and pregnancy is statistically significant (28).   Further similar studies are indicated.   Certainly MGH may be seen in a patient receiving only estrogen or one who is not on hormones and is not pregnant.  

            The lesion typically occurs in young women but approximately 6 percent of the reported cases have occurred in postmenopausal women (20-27).  MGH is typically an incidental microscopic finding but occasionally the lesion is visible as an erosion, 60% of the cases in one of the early reports (23), is indistinguishable from a typical cervical polyp grossly, or, uncommonly, is friable and suspicious for carcinoma.  Indeed, five cervices in one of the early series were clinically felt to be “compatible with carcinoma” (22).  As expected, these lesions may cause significant symptomatology in the form of bleeding.

            On microscopic examination MGH may be polypoid or non-polypoid, or may involve a portion (sometimes almost all) of a pre-existent endocervical polyp.  Closely packed glands that vary from small and round to large, irregular, and cystically dilated are characteristic.  The glands often have subnuclear vacuoles.  They usually contain a basophilic or eosinophilic mucinous secretion that often contains many acute inflammatory cells.  There are typically many acute and chronic inflammatory cells in the intervening stroma.  There may be mucin extravasation into the stroma.  There is often associated squamous metaplasia that is often immature.   

            The features just summarized represent the well-known morphology, of the lesion as described in the several papers that defined this entity, essentially synchronously over 30 years ago.  Unusual features mentioned in some of those reports were:  reticular (20), solid cords (20), compact medullary masses (22), hyaline stroma (23), a trellis-like appearance (23), and trabeculae (22).  In 1984 Leslie and Silverberg (26) contributed a major paper on this topic.  Firstly, they emphasized a solid growth (seen in three of their cases).  Secondly, they highlight that MGH may be seen in endocervical tissue present in an endometrial curettage specimen.  Thirdly, they were the first to record microglandular hyperplasia-like changes in endometrial tissue proper.  Fourthly, they stressed that typical MGH may occur in postmenopausal patients.  Finally, they described three cases of endocervical adenocarcinoma with MGH-like appearances.  In a relatively recent paper some of these rare, but challenging other features and appearances were reemphasized by us (29) and a few others added.  The latter were:  1. small nests and larger aggregates of epithelial cells irregularly distributed in a myxoid stroma, imparting a pseudoinfiltrative pattern; 2.  Signet-ring cells; 3.  hobnail cells; 4.  moderate nuclear atypia and occasional mitotic figures.  The solid pattern present in four cases caused particular difficulty as it did in two cases in one of the original series (22).  One case in our series showed particularly striking stromal hyalization.  The cells in solid areas may have abundant eosinophilic cytoplasm or have pale cytoplasm and eccentric nuclei, simulating the signet-ring cells of an adenocarcinoma, the possibility for misdiagnosis as cancer being heightened by the fact that these cells may be mucin positive.  The hobnail cells in our experience are usually seen in reticular foci.  The nuclei of the cells of MGH are usually small, regular, but rarely they may be mildly or even moderately atypical and have hyperchromatic nuclei, particularly when hobnail-type cells.  Mitotic figures are generally rare but up to one mitotic figure per ten high power fields may be seen.  In occasional cases the appearance may be complicated by the co-existence of another non-neoplastic glandular lesion such as papillary endocervicitis. 

           

            The cervical adenocarcinoma most likely to be confused with MGH is clear cell carcinoma because both lesions may have tubular, cystic and solid patterns and a hyalinized stroma.  The solid foci of clear cell carcinoma usually consist of cells with abundant, clear, glycogen-rich cytoplasm whereas those in solid foci of MGH have never had conspicuous clear cytoplasm in our experience.  Although some clear cell carcinomas may have relatively bland cytologic features and low mitotic rates and, as noted, some cases of MGH may have some degree of atypia and mitotic figures, the atypia and mitotic rates in clear cell carcinoma always exceed those of MGH.  Distinction of MGH from rare endocervical adenocarcinomas of the usual (endocervical) type that have a microglandular pattern may be difficult but the cytologic criteria just mentioned are usually helpful (30).  One cervical adenocarcinoma we have seen that closely simulated MGH in areas occurred in a patient who was on oral contraceptives (30).  A commoner problem than those just considered is distinction of MGH from endometrial adenocarcinomas of endometrioid, mucinous, or mixed endometrioid and mucinous types.  These carcinomas may have a combination of focal microglandular patterns, extensive acute inflammation, and glands filled with mucin, creating a striking resemblance to MGH in some instances (26,30).  Fragments of an endometrial tumor of this type may be present in a specimen thought to be endocervical in origin, and the diagnosis of MGH should be made with caution in a postmenopausal patient, particularly if its features are not characteristic and cytologic atypia greater than usual.  Negative immunohistochemical staining for carcinoembryonic antigen (CEA) supports a diagnosis of MGH (31,32).  However, adenocarcinomas are occasionally negative or only focally positive for CEA studies.  Mesonephric adenocarcinomas are typically CEA-negative (18).   Accordingly, a negative stain for CEA should be cautiously interpreted but intense positive staining throughout the cytoplasm is unlikely in a benign lesion and strongly favors carcinoma.  Negative CEA staining is also seen with the other pseudoneoplastic glandular lesions considered here but experience with CEA staining of many of them is limited and a diagnosis should not be rendered solely on the basis of immunohistochemical staining results.  Normal endocervical glands and those of pseudoneoplastic lesions may exhibit a glycocalyceal pattern of CEA staining that contrasts with the diffuse cytoplasmic staining typically seen in cervical adenocarcinomas.

Mesonephric Hyperplasia 

            Uncommonly mesonephric remnants are so numerous that the diagnosis of adenocarcinoma is suggested, especially in cases accompanied by cytologic atypia and mitotic activity, and the designation mesonephric hyperplasia is indicated in such cases (33-36).  Although a location of mesonephric remnants in the lateral walls is often emphasized in practice, this is not practically relevant in cases of mesonephric hyperplasia.  In some instances it is obviously arbitrary whether a proliferation is placed in the hyperplasia category.  A size cut off of 6 mm. is recommended by Ferry and Scully (33) who divided their cases of mesonephric hyperplasia into three categories:  lobular mesonephric hyperplasia (31 cases), diffuse mesonephric hyperplasia (8 cases), and pure mesonephric ductal hyperplasia (2 cases).  Their lobular cases ranged from 4 – 22 mm. and their diffuse cases from 13-25 mm. in greatest dimension (15).  In the recent study of Seidman and Tavassoli (34) there were 40 cases of lobular and 11 cases of diffuse hyperplasia; no pure ductal hyperplasias were recorded in that series. Lobular cases have been discovered in a somewhat younger age group than diffuse cases.   The lesions are usually asymptomatic but in the series of 14 cases described by Jones et al. (35) they were thought to account for the symptoms in 2 cases.  In both of these the hyperplastic tubules reached the mucosa and were associated with ulceration, presumably explaining the symptoms.  In this same series, it was felt that in one case the process accounted for a gross abnormality in the form of a “submucosal nodular elevation."  A detectable mass was present in one of the cases of Ferry and Scully (33).  Gross abnormalities were described in 8% of the cases in another report but it is often impossible to be sure if this was actually due to the mesonephric proliferation (34).

            In cases of lobular hyperplasia, round to oval, often focally dilated, tubules are arranged in at least vaguely lobular aggregates.  The tubules characteristically contain bright pink, periodic acid-Schiff positive, hyaline-like material in their lumens but this is by no means an invariable finding being absent in 10% of the cases in one series (34).  In diffuse hyperplasia no lobular grouping is apparent, and in some of them the irregular distribution of the tubules, their particularly conspicuous number, and their presence deep in the cervical wall may create a very worrisome appearance.  In about one-third of both types of non-ductal hyperplasia, tubules extend to within 1 mm of the endocervical surface, and in some they may reach the mucosa as noted above.  The low power picture may suggest transmural involvement by adenocarcinoma, a consideration heightened when the tubules extend occasionally to the deep margin of a cone biopsy or hysterectomy specimen, or to the lower uterine segment or upper vagina (35). Other features described in some cases of mesonephric hyperplasia by Seidman and Tavassoli (34) have included interanastomosing slit-like (retiform) tubules (32% of their cases), an endometrioid-like appearance that was likened to epididymal differentiation (14% of their cases) and a resemblance to seminal vesicle epithelium with lipofuscin pigment (1 case). Three-quarters of cases that are predominantly composed of tubules have an additional component of ductal hyperplasia (33), which is also occasionally seen in pure form.

            Ductal hyperplasia is characterized by one or more large ducts that may be round or elongated depending on the plane of section and are lined by pseudostratified epithelium that often forms small papillae (33).  Although the tubular form of mesonephric hyperplasia is most likely to be misinterpreted as invasive adenocarcinoma, ductal hyperplasia, particularly when occurring in pure form, is usually misdiagnosed as a premalignant lesion such as endocervical glandular dysplasia or adenocarcinoma in situ.  The frequently elongated form of the ducts and a usual lack of an association with endocervical glands are initial clues to their nature, and high power scrutiny reveals no significant cytologic atypia.  Finally, the micropapillae seen in pure ductal hyperplasia are distinctive and usually not a feature of premalignant glandular lesions. 

            Mesonephric hyperplasia must obviously be distinguished from mesonephric adenocarcinoma, a very rare lesion that should be diagnosed only after florid mesonephric hyperplasia has been excluded.  In contrast to mesonephric hyperplasia, which is almost always an incidental microscopic finding, a mesonephric carcinoma is more likely to produce a grossly visible lesion.  On microscopic examination, mesonephric carcinomas are typically associated with irregular destructive invasion by the glandular elements, which are more crowded than in hyperplasia and often have an associated stromal reaction.  High degrees of nuclear atypicality and prominent mitotic activity favor a malignant lesion although these features have not been present in all the carcinomas.  Two groups did not find the mitotic rate in hyperplasia to exceed one per 10 high power fields (33,34).  Most cases of mesonephric carcinoma have been associated with mesonephric hyperplasia. 

            Mesonephric hyperplasia may be confused with other adenocarcinomas including adenoma malignum, clear cell adenocarcinoma and rarely metastatic adenocarcinoma.  The glands in adenoma malignum are much more irregular in size and shape than the tubules of mesonephric hyperplasia.  In addition they are lined, at least in areas, by tall columnar mucinous cells in contrast to the cuboidal non-mucinous epithelial cells of mesonephric tubules and are often surrounded by a desmoplastic stromal reaction that is absent in mesonephric hyperplasia.  The tubules in mesonephric hyperplasia might be confused with the tubular glands that are present in many clear cell adenocarcinomas of the cervix, and the tubules in mesonephric hyperplasia may undergo cystic dilatation imparting a low power appearance, which may mimic that of the tubulocystic pattern of clear cell carcinoma.  In contrast to mesonephric hyperplasia, however, at least some of the cells lining the glands and cysts of clear cell carcinoma have conspicuous clear, glycogen-rich cytoplasm or are of hobnail type.  Additionally, clear cell carcinomas with a pure tubular pattern are uncommon; solid sheets of clear cells and papillary patterns incompatible with a diagnosis of mesonephric hyperplasia, are also usually present at least in minor amount.  Occasional metastatic adenocarcinomas, for example from the breast, may consist of small tubular glands and have an appearance somewhat reminiscent of mesonephric hyperplasia.  The carcinomas will usually focally contain cords of cells and single cells, and cytologic features of carcinoma inconsistent with mesonephric hyperplasia.  

            Distinction of mesonephric hyperplasia from other benign glandular lesions is an academic exercise and usually not difficult. A combination of the low cuboidal non-mucinous, non-stratified nature of the epithelium and frequent luminal hyaline-like material distinguish it from the various other pseudoneoplastic glandular lesions.  Perhaps the most realistic problem may result in some cases of mesonephric hyperplasia in which the glands may appear endometrioid and simulate those of endometrioid metaplasia but an association with more typical mesonephric tubules is usually helpful.

Glandular Hyperplasia, Not Otherwise Specified

            The endocervical glands may be hyperplastic, sometimes floridly so, without exhibiting any of the specific aforementioned histologic patterns.  Lack of deep invasion, an orderly, sometimes lobulated arrangement of the glands, a well demarcated margin with the adjacent cervical stroma, the usual lack of a stromal reaction, and bland nuclear features are all features seen to varying extents in individual cases that indicate that these lesions are non-neoplastic.  In some cases diagnostic problems result from the presence of irregularly shaped glands, an appearance exaggerated by poor orientation and artifactual distortion, emphasizing that although markedly irregular gland contours are characteristic of adenoma malignum they are not pathognomonic of it.  In other cases, a worrisome appearance results from reserve cell hyperplasia and a cribriform glandular pattern.  There is overlap of this pattern with that of microglandular hyperplasia.

METAPLASIAS AND ECTOPIAS

Tubal, Tubo-Endometrioid and Endometrioid Metaplasia

                An alteration of the usual endocervical epithelium to one of tubal, endometrioid, or tubo-endometrioid cell type is common but only the source of intense investigation recently.  In tubal metaplasia, the endocervical epithelium is replaced by a single layer of ciliated cells, non-ciliated cells, and peg cells, as seen in the normal tubal mucosa whereas in endometrioid metaplasia the cells have a uniform endometrioid appearance.  The tubal and endometrioid cell types overlap and often co-exist so it is not meaningful to consider them separately here.  In 1986 Wells and Brown (37) commented on the propensity for tubal metaplasia to be confused with adenocarcinoma and this was followed a few years later by a detailed original report of 11 cases by Suh and Silverberg (38). Although it is not the focus of this review, it is worth noting that two years earlier Pacey et al. (39) had focused attention on the problems caused by tubal metaplasia in cytology smears, a now widely recognized phenomenon (40). 

            Two of the 11 cases of Suh and Silverberg were initially thought by others to be precancerous glandular lesions, diagnoses suggested by the hyperchromatic appearance on low power.   The gland architecture was normal in all their cases, cytologic atypia was absent and there were no stromal alterations.  Surface involvement, involvement of deeply located glands and partial gland involvement were documented as were involvement of a tunnel cluster (1 case) and coexistent cervical endometriosis (1 case). Tubal metaplasia was subsequently found in 31 percent of 108 cone biopsy and hysterectomy specimens (41).  There was no relation to age.  An even greater frequency, 80%, was found in a series in which all the specimens were hysterectomies (42).  In that series the lesion was always present in the upper one-third of the endocervix but also involved the transformation zone in one-quarter of the cases.    Tubal metaplasia is most common in the deeper glands of the upper endocervix but involves the surface epithelium in approximately one-third of the cases. 

            Oliva and colleagues (43) have emphasized that the glands involved by tubal metaplasia may show some architectural irregularity, which should not lead to confusion with adenocarcinoma, nor should the associated stromal cellularity, which those authors also drew attention too.  Pertinent findings in her study, some reinforcing prior observations were:  a gross abnormality due to prominent cystic glands in one case, hypercellular (64%) or edematous to myxoid stroma (40%), glandular branching (52%), prominent cystic dilatation (36%), glandular crowding (16%) and glands in the outer third of the cervical wall (12%).  She also observed apical cytoplasmic snouts in the metaplastic glands in some of her cases.  Although the hypercellular fibrotic stroma seen in some cases of tubal metaplasia differs from that of endometriosis with its classic naked-nucleus appearance and conspicuous arterioles, the two stromal appearances merge imperceptibly in some instances.  Benign endometrioid glands, interpreted as ectopic endometrial glands rather than metaplastic endocervical glands, were found deep in the wall in 1 percent of cervices in one series (40). 

            In 1991 Ismail (44) made an important contribution to this area pointing out the relation of this process in many cases to a recent cone biopsy and, additionally, its association with endometriosis.  In her series the cone biopsy site in 26% of hysterectomy specimens showed tubo-endometrioid metaplasia.  In an additional 43% of the cases the presence of endometrial-type stroma, often only focal, around otherwise similar glands warranted the diagnosis of endometriosis.  The majority of the cases of endometriosis also had glands without stroma (the close relation of tubo-endometrioid metaplasia and endometriosis is referred to in the next section).  Mitotic activity was described in the epithelium in the cases of tubo-endometrioid metaplasia.  She concluded that tubo-endometrioid metaplasia and endometriosis were metaplastic sequelae of injury caused by cone biopsy.  Furthermore, she surmised that the shorter interval postcone biopsy in the pure glandular metaplasia group might indicate that this process induced endometrial stromal differentiation in the adjacent stroma. 

             

            The admixture of cell types, lack of atypia, absence of invasive characteristics, and the usual absence of immunoreactivity for CEA are clues to the correct diagnosis in cases of tubal metaplasia.  It should be noted, however, that CEA positivity was found in 39% of cases of tubal metaplasia in one series (45).  That study found tubal metaplasia to be vimentin positive in 78% of cases in contrast to adenocarcinoma in situ.  Additionally, as with all epithelial proliferations of the cervix whether squamous or glandular, reactive atypia must be cautiously interpreted.  Ismail, for example, described papillary hyperplasia and hobnail cells metaplasia in one of her cases.  These and other reactive changes placed on the background of the preexistent glandular abnormalities reviewed here can produce a great range of often perplexing appearances.

            It is obvious that the epithelium of glands involved by tubal metaplasia is not immune from premalignant or even malignant change and indeed Oliva et al (39) included a cautionary comment to that end in the discussion of their paper on tubal metaplasia.  Furthermore, a recent paper has specifically looked at this issue by describing 11 cases with atypical tubal metaplasia or adenocarcinoma in situ with tubal features (46).  Cases should be placed in such categories with caution as the hyperchromatic appearance of conventional tubal metaplasia causes it to look a little more “atypical” than non-tubal endocervical epithelium.  Unquestionable moderate to severe atypia with easily found mitoses does warrant an atypical or potentially in-situ cancer diagnosis.  Apoptotic bodies, when frequent, have been shown in a recent study to be typical of adenocarcinoma in situ, compared to various benign lesions (47). 

Endometriosis

            Cervical endometriosis may be superficial (also referred to as primary) or deep (also referred to as secondary)  (48).  This disease in both its forms encountered in the cervix was well discussed in the older literature and selected excellent older papers are referenced (49-52).  Overall, these papers noted that the superficial form was not rare, was easiest to diagnosis at the time of menstruation, was associated with a history of cervical trauma such as a prior cone biopsy and rarely mimicked carcinoma grossly (49).  The papers did not emphasize a propensity for superficial endometriosis to mimic premalignant glandular lesions but this was addressed in a recent study (53).  Superficial endometriosis usually involves the inner-third of the cervical wall but may extend into the middle third, this being seen in one of 20 cases in the recent series (53).  Superficial endometriosis is found in approximately 1-2 percent of patients in the reproductive age group and unlike deep endometriosis is usually unassociated with pelvic endometriosis.  In this regard, it is linked to tubo-endometrioid metaplasia and indeed the more we have seen of cases in this overall category, the more we have formed the opinion that there is no clear distinction in many cases between tubo-endometrioid metaplasia with cellular stroma and endometriosis and perhaps one evolves into the other as suggested by Ismail (44).

             The patients are usually in the reproductive years with a mean age of 37 years in the study of Baker et al. (53).   Superficial endometriosis may be misinterpreted as endocervical glandular dysplasia or adenocarcinoma in situ, particularly when ulceration and edema obscures the stromal component.  The latter was difficult to recognize in 35% of Baker’s cases. As many as half the cases in her series were initially considered potentially to be pre-malignant or malignant and the mitotic activity that is not unexpected in endometriosis in the reproductive years contributes to this problem.  However, the mitotic rate is generally much higher in cases of adenocarcinoma in situ.  Changes according to the menstrual cycle such as secretory, may be seen as potentially may pre-malignant and even malignant change.  Apoptotic bodies were seen in one of 20 cases in Baker’s series indicating that, although more characteristic of adenocarcinoma in situ, they are not diagnostic. Just as tubal metaplasia may cause problems in cytologic material, so also may superficial endometriosis (54). 

Endosalpingiosis

            This peritoneal based process of mullerian glandular type rarely involves the uterine cervix extensively.  Four examples have recently been described (55).  Three of these patients were in the reproductive age group but one was 74-years-old.  The gross manifestations may be striking.  In two of the four reported cases multiple cysts involved the serosal surfaces of the uterus and adnexa in one case and the colon and rectosigmoid, pelvic sidewalls and cul-de-sac in the other.  The other two cases had cysts confined to the cervix and lower corpus but were striking in each of them.  This process is similar in the distribution of the mullerian-type glands to endocervicosis although in our experience to date this process has been more extensive.  Florid endosalpingiosis may potentially be misinterpreted as minimal deviation adenocarcinoma of endometrioid type if the fact that the process is centered in the outer cervical is not appreciated and if there is continuity with overlying glands exhibiting tubal metaplasia. 

            The microscopic appearance in these cases has been fundamentally similar to that of endosalpingiosis in general although the degree of cystic change exceeds that seen in many cases and accounts for the particularly dramatic gross appearance of some cases.  The glands and cysts are lined by the typical tubal-type epithelium that characterizes the lining of endosalpingiotic glands in general.  The glands are typically scattered haphazardly but occasionally they may form lobulated aggregates.  They vary markedly in size and shape, often having highly irregular contours due to complex infoldings or branching.  Rarely stromal papillae may project into the lumens.  The glands are separated by normal cervical stroma.  In one of our cases potential confusion was caused by a focus in which the glands were closely apposed to a nerve simulating perineural “invasion”.

            This process is distinguished from tubal metaplasia on the basis of the dominant location of the glands and cysts.  The designation tubal metaplasia is reserved for endocervical glands that have the usual superficial distribution and configuration of normal endocervical glands although occasionally the glands may be somewhat deeper. 

Endocervicosis

            This process, which has been highlighted relatively recently in genitourinary tract pathology in general as a distinct process involving the bladder, may also involve the uterine cervix with four cases recently described in detail (56).  The patients were 29 to 45 years of age and two had a history of caesarian section.  In three a gross abnormality of the outer aspect of the cervix was noted at the time of hysterectomy and in the fourth at the time of gross pathologic examination.  The process involved the anterior wall of the cervix in each case.  The lesions were firm rubbery masses 1 to 2.5 cm. in greatest dimension with cysts seen on sectioning in two.  Microscopic examination disclosed a dominant population of glands of variable size and shape, some of them cystically dilated, lined by mucinous endocervical-type epithelium that ranged from columnar to flattened.  The cytologic features were bland.  Although there was a periglandular stromal reaction related to mucin extravasation in two cases, there was no desmoplastic stromal reaction or other indication of aggressive infiltration.  A clue to the diagnosis of this entity is that the process is centered in the outer cervical wall and often involves the paracervical connective tissue.  Furthermore, although the glands of the lesion may, if it is particularly extensive, abut eutopic endocervical glands causing concern for deeply invasive endocervical adenocarcinoma of the adenoma malignum type, there is usually a zone of uninvolved connective tissue between the glands of this lesion and the overlying endocervical glands.  Alertness to this finding and awareness of this process may be crucial in preventing an erroneous diagnosis of cancer.  This process is distinguished from an adenomyoma by the circumscription and intervening benign myomatous muscle of the latter.

Atypical Oxyphilic Metaplasia

            In some cases the endocervical cells have, on an apparent idiopathic basis, abundant eosinophilic cytoplasm and atypical nuclei that are hyperchromatic and ‘smudgy’.  A lesion can only be packed in this category when there is no history of radiation which can causes a similar alteration as noted in the next section.  The six examples of oxyphilic metaplasia we described occurred in adults from 41 to 62 years of age (57).  One was on oral contraceptives and one on tamoxifen but there is no evidence of these being significant associations.  The lesion is an incidental microscopic finding and the only additional comment merited on its appearance as presented at the start of this section is to note that the nuclei may be multilobated, nucleoli are often prominent and mitoses are lacking.  Although rare apical snouts are present and the cells are consistent morphologically with apocrine cells immunostains have not supported such a nature. 

Intestinal Metaplasia

            Intestinal metaplasia is the rarest form of benign glandular metaplasia seen in the uterine cervix and is characterized by the focal presence of goblet and argentaffin cells within the endocervical glands (58).  Intestinal metaplasia in the cervix is seen more commonly in cases of adenocarcinoma in situ and occasionally in benign neoplasms and in adenocarcinomas.  As intestinal metaplasia unassociated with any degree of dysplasia is rare, the diagnosis should only be made after careful evaluation has ruled out any cellular atypicality. 

Arias-Stella Reaction

            The Arias-Stella reaction is seen within endocervical glands in about 10 percent of gravid uteri (59,60).  It is typically focal, involving only one or two glands in each case.  Involvement of superficial glands is more common than that of deep glands and polyps may also be affected (62).  The cytologic features are as seen in the endometrium, stratified cells with abundant vacuolated cytoplasm and enlarged, pleomorphic and hyperchromatic nuclei, imparting a hobnail appearance to some cells.  Optically clear nuclei may also be seen.  Mitoses are rare or absent.  These changes may be mistaken for adenocarcinoma in situ or clear cell carcinoma, particularly in a biopsy or cytology specimen; such diagnoses should therefore be made with caution in a pregnant patient.  Clear cell carcinoma, however, will often be associated with a mass and histologic examination will reveal features inconsistent with the Arias-Stella reaction, including the presence of invasion and tubular and solid patterns.  Adenocarcinoma in situ, in contrast to the Arias-Stella reaction, usually exhibits uniformly atypical nuclei and relatively frequent mitotic figures, and usually lacks marked cytoplasmic vacuolation, hobnail cells, and optically clear nuclei.

Ectopic prostate

            Five examples of this unusual phenomenon have been reported in recent years (61,62).  The oldest patient was 77-years-old and this is the only case in which the prostate tissue apparently formed a mass.  It measured about 4 cm.  All the other lesions occurred in patients who were much younger, ranging from 22 to 38 years.  In these four cases the lesion was always an incidental finding in a cone biopsy specimen performed because of dysplasia.  The microscopic features in all five cases has been remarkably similar.  The prostatic-type glands have been found both superficially and deep in the cervical wall and have shown cribriform and papillary patterns recognizable as of prostatic type although, for obvious reasons, the diagnosis may not readily come to the mind of the pathologist due to the exotic nature of the lesion and its only recent highlighting in the literature.  All five cases have shown squamous metaplasia but without keratin formation.  High power examination shows the typical two cell layers of secretory cells and basal cells that typify normal prostate and the secretory cells have always been positive with immuno stains for PSA and PSAP and the basal cells have been highlighted by the high molecular weight cytokeratin stain.  The explanation of this tissue in the uterine cervix is not clear at this time and nothing can be added here to the considerations presented in the two papers on this topic.  Available follow-up in the reported cases has been unremarkable.

            As our personal experience with this process has increased, we have been impressed with degree to which the picture looks rather typical from case to case, particularly because of the invariable presence of cribriform and papillary patterns with squamous differentiation.  When this picture makes prostate tissue come to mind, the differential diagnosis is limited.  Lobular endocervical gland hyperplasia may be suggested because of the tendency for the glands to be closely packed but the two cell layers of secretory and basal type cells contrasts with the single layer of tall columnar mucinous endocervical type cells lining the glands of lobular endocervical glandular hyperplasia, and the latter typically does not show striking squamous differentiation.  The only neoplasm that we think might rarely be a reasonable consideration is an exceedingly low grade variant of adenoid basal carcinoma that some like to designate “adenoid basal tumor”.  Although one may see squamous differentiation in the latter, their basaloid appearance contrasts with the cases of ectopic prostate encountered to date and there is not the characteristic double layer of secretory and basal cells in that neoplasm.  

REACTIVE AND INFLAMMATORY LESIONS

Papillary Endocervicitis

Although this is probably the commonest of the lesions considered in this review, it exceptionally causes diagnostic problems for the pathologist.  From the clinical perspective there may be an erythematous cervix because of the intense inflammation that is usually present and the process may be biopsied because of significant clinical concern for neoplasia.  This lesion is characterized by papillae that are usually relatively regular, contain chronic inflammatory cells and are covered by a single layer of benign endocervical columnar epithelium.  The papillae are typically of modest breadth, rarely being filiform on the one hand or edematous on the other.  The appearance may potentially be confused with that of endocervical adenocarcinoma with a villoglandular pattern but the tumors exhibit focal cellular stratification and more cytologic atypia than seen in papillary endocervicitis, although the latter, given its inflammatory nature may show reactive atypia.

Infectious Atypias

            Cytomegalovirus (CMV) infection, which is usually an incidental histologic finding within the endocervix, is associated with large basophilic intranuclear inclusions within the affected endocervical epithelial cells, and in some cases, endothelial cells.  Immunohistochemical staining for CMV antigen may be confirmatory.  The intranuclear inclusions, as well as associated granular cytoplasmic inclusions, are sufficiently characteristic that confusion with adenocarcinoma in situ is unlikely.  In additional contrast to the latter, the cells showing the CMV-related changes are singly disposed within normal endocervical columnar cells. 

            Brown and Wells (37) described five examples of bizarre multinucleated giant cells within endocervical glands.  In each case, there was evidence of HPV infection elsewhere in the cervix, and the authors concluded the changes were secondary to the viral infection.  Herpes virus infection of the uterine cervix is quite common (63).  In typical cases the clinician will often suspect the diagnosis but it is not rare for the lesion to be overlooked clinically.  On microscopic examination the characteristic ground glass intranuclear inclusions should establish the diagnosis with relative ease in most cases although occasionally a misdiagnosis as a dysplastic process is made. 

Radiation Atypia

            Radiation to the cervix can cause markedly atypical nuclear changes that can be mistaken for a preneoplastic or neoplastic glandular process, particularly if the pathologist is unaware of the history.  Changes in endocervical glandular epithelium have only been the subject of one study, a recent one by Lesack et al. (64) who described the alterations in 10 patients, all of whom had received at least 3600 rads to the pelvis from 6 weeks to 17.5 years previously.  The therapy was external beam alone in five and external beam and intracavitary in five.  Gross alterations due to the radiation included fibrosis and stenosis but it is on microscopic examination that the potential for diagnostic errors exists.  The problematic changes are cytologic and are:  nuclear elargement, nucleolar prominence and loss of polarity.  Despite the nuclear enlargement, increased cytoplasmic volume results in no significant alteration in the nuclear/cytoplasmic ratio.  Nuclear hyperchromasia was uncommon in the Lesack study.  The usually prominent nucleoli were typically eosinophilic.  Features helpful in indicating the nature of the process were dense, often abundant eosinophilic cytoplasm, lack of mitoses and architectural and stromal features.  The architectural feature that is typical is the presence of fewer glands than normal and frequent dilated glands.  Stromal changes include fibrosis, that may be hyaline, and radiation fibroblasts although the latter were, perhaps surprisingly, present in only one of 10 cases in the Lesack study.  Stromal edema and myxoid change are sometimes seen as are the vascular changes of radiation injury, initial thickening being seen in 70% of the cases in the Lesack series.  The most marked stromal changes were seen in the patients with the longest post-treatment interval in Lesack’s study.  In that report CEA staining and DNA content analysis were not found helpful in diagnosis.

           

Changes Secondary to Mucin Extravasation

            Although the lack of a stromal response often supports the benign nature of an endocervical glandular lesion, the stroma adjacent to benign glands may exhibit reactive changes, usually, but not always, in response to mucin extravasated from a ruptured gland.  A foreign-body giant cell reaction is sometimes seen in these cases and in some of the cases foamy histiocytes are conspicuous.  Mucin extravasation may be seen in association with some of the above mentioned lesions such as microglandular hyperplasia.   In one case we saw gland rupture was associated with dissection of mucin into the stroma and vascular lumens, the appearance raising initial concern about a primary or metastatic adenocarcinoma.  Michael and her colleagues (1) described another case of a pseudoneoplastic glandular lesion with an unusual appearance interpreted by them as being due, at least in part, to mucin extravasation.

Benign glandular neoplasms

            It is appropriate to consider certain benign neoplastic processes here, not only because the focus of  this review is the important distinction between non-neoplastic and neoplastic, meaning in most instances by the latter carcinoma, but also because some of the entities considered in this section, particularly endocervical-type adenomyomas, may occasionally be confused with some of the non-neoplastic processes already considered. 

           

Polyps

           

            Occasional, otherwise typical, endocervical polyps may exhibit florid hyperplastic and metaplastic changes which can cause diagnostic confusion particularly if they are first encountered in a fragmented curettage specimen.  We have seen cases in which this has been the case and only when the resected lesion has been examined and the overall architecture has been viewed, has the issue been confidently clarified.  The epithelium of polyps may exhibit florid squamous metaplasia, papillary hyperplasia or microglandular hyperplasia.   In curettage material one occasionally sees detached fragments of more conventional non-proliferative endocervical polyp, which may be a clue to an exuberant problematic proliferation being a focal finding in the polyp. 

Adenomyomas

            Some endocervical polyps may have a myomatous stromal component which, if limited in amount, is not worthy of consideration in the designation of the lesion.  When conspicuous the designation adenomatous polyp may be used, provided the lesion is otherwise a typical “polyp”.   We prefer the designation adenomyoma in cases that have a gross appearance different from that of the usual polyp, forming a solid or solid and cystic mass, albeit typically exophytic.  Endocervical adenomyomas may be of endocervical type (65), having a mucinous epithelial component as expected given the nature of the native endocervical epithelium.  Less common are endometrioid adenomyomas which may be of typical type as recently presented in detail in the literature by Gilks and colleagues (66) or alternatively they may fall in the category of the atypical polypoid adenomyoma as defined by Mazur (67,68). 

            Review of the literature discloses only rare cases of what may have been endocervical-type adenomyomas.  For example, the first two cases in the report of Fetissof et al mentioned earlier (12) may fall in this group.  Only one sizable series of endocervical-type adenomyomas has been reported.  They typically occurred in reproductive age females with a mean age of 40 years (65).   Most of the lesions were asymptomatic.  The lesions in that series were typically exophytic (80% of them) with the remaining two being mural lesions that projected into the pelvis from the outer aspect of the cervix.  The exophytic tumors ranged up to 8 cm.  The mural neoplasms measured 11 and 23 cm.  These lesions are typical well circumscribed and gray-white and range from solid to solid and cystic with mucin-filled cysts sometimes being conspicuous.  By definition, microscopic examination shows a composition of glands and cysts lined by a single layer of endocervical type mucinous epithelium separated by a stroma that is predominantly smooth muscle in character.  The glands may be small and simple, sometimes arranged in lobules or more irregular in shape with papillary epithelial infolding.  Minor foci of tubal-type epithelium or endometrioid-type epithelium may be present within the glandular component.  Gilks and colleagues (65) reported their cases because they found that these lesions often raised the issue of adenoma malignum in the mind of the contributing pathologist.  The polypoid gross appearance of many of the adenomyomas is exceptional for adenoma malignum, as would their characteristic gross circumscription.  On microscopic examination the myomatous mesenchymal component without a desmoplastic stromal reaction, the latter being typically present at least focally in adenoma malignum, and the frequent lobular arrangement of the glands which are lined by cells devoid of significant atypia, in aggregate, should facilitate the differential diagnosis with adenoma malignum or any form of adenocarcinoma.  Although it is only an academic distinction, we have occasionally found it difficult to distinguish between an endocervical-type adenomyoma and lobular endocervical gland hyperplasia because the glands of the former often are lobulated and may be present in a fragmented curettage specimen.  When it is know that there is a discrete mass lesion, the suspicion should be for an adenomyoma and on microscopic examination the striking smooth muscle between the benign glands and sharp circumscription should enable this distinction to be made.  

            Adenomyomas of endometrioid type, whether typical or atypical, are relatively uncommon and less often an issue in differential diagnosis with carcinoma in our own personal experience.   The attention the atypical polypoid adenomyoma has received in recent years and its typical features with frequent morular-type squamous type differentiation generally cause it to be recognized.  Nonetheless, in both this lesion, and typical endometrioid adenomyomas, a somewhat irregular distribution of glands in a stroma that may be cellular and atypical, particularly in the atypical polypoid adenomyoma, may raise the issue of an invasive adenocarcinoma such as the minimal-deviation type of endometrioid carcinoma of the cervix.  Features generally similar to those mentioned in the distinction of the endocervical-type adenomyoma from adenoma malignum are again helpful. 

Villous and Villoglandular Adenomas

            These are included for completeness sake because such lesions rarely occur but as they are often associated with underlying invasive adenocarcinoma we believe the diagnosis of “villous adenoma” should be made with great caution in the cervix (69).   It should be made only after an entire lesion has been removed with a negative margin and thoroughly examined microscopically to exclude focal invasion indicative of adenocarcinoma.  Many endocervical adenocarcinomas have a surface papillary component that may be potentially misinterpreted as a “villous adenoma”. 

Mullerian Papilloma

            These rare benign lesion covered by glandular epithelium of mullerian type occur almost exclusively in children in the first decade of life (70).   Fine branching papillae are typically lined by a single layer of columnar to cuboidal to flattened epithelium.  Although there may be stromal inflammatory cells with edema, the process has the appearance of a benign neoplasm rather than an inflammatory process such as papillary endocervicitis, which may be in the differential diagnosis.

MALIGNANT GLANDULAR LESIONS

ENDOCERVICAL GLANDULAR DYSPLASIA, ADENOCARCINOMA IN SITU, AND MICROINVASIVE (EARLY INVASIVE) ADENOCARCINOMA

         AIS typically occurs in the transformation zone (71), and appears to begin at or near the squamocolumnar junction.  AIS typically involves both the surface columnar and underlying glandular epithelium; less commonly it is confined to the glands, which are typically admixed with normal glands.  Very rarely, only the surface columnar epithelium is affected.  AIS almost always involves the necks of the glands, and in some cases, deeper portions of the glands as well.  Deep gland involvement in the absence of superficial gland involvement is rare (18).  AIS is unusually unicentric, but multifocality has been documented in approximately 15 percent of the cases.  Glands involved by AIS may be covered by normal, metaplastic, or dysplastic squamous epithelium.

         In AIS the involved glands usually have a normal distribution and a relatively normal configuration but glands that are somewhat abnormal in contour, or cystically dilated, may also be involved.  Papillae with fibrous cores that project into involved glands or that replace the surface epithelium may be seen.  A cribriform pattern is occasionally present, and in some cases solid sheets of cells obliterate gland lumens.  AIS may involve and rarely be confined to an endocervical polyp.

          AIS is usually recognizable on low-power examination because the characteristic cellular stratification and nuclear atypia make it appear darker than the normal glandular epithelium.  When AIS involves only a portion of an endocervical gland it is usually sharply demarcated from the normal epithelium.  AIS is characterized by pseudostratified or stratified columnar cells with at least focal malignant nuclear features. The lesional cells have enlarged fusiform nuclei oriented perpendicular to the lumen, are variably hyperchromatic, and contain fine to coarse chromatin and nucleoli that are typically small; macronucleoli are seen in a minority of cases.  Occasionally, the glands are lined by more pleomorphic large polygonal cells.  Mitotic figures are typically frequent and may be abnormal.  Individual cell necrosis with intraepithelial nuclear debris (apoptosis) is relatively frequent (47).  The malignant cells occasionally exfoliate into the gland lumens.

         In the typical or “endocervical” form of AIS, the lesional cells have moderate amounts of predominantly juxtaluminal cytoplasm that stains variably positive for mucin, although the latter is diminished in amount compared to the mucin content of normal endocervical cells.  The presence of mucin-rich goblet cells (and less commonly, argentaffin cells and Paneth cells) are characteristic of the “intestinal” form of AIS.  The “endometrioid” variant of AIS is characterized by an absence of stainable mucin.  The frequency of the endocervical, intestinal, and endometrioid subtypes in one series of AIS was 95.8 percent, 29 percent, and 16 percent, respectively; the two less common subtypes were almost invariably admixed with the endocervical subtype (74).

         In a study of 21 cases, Ostor et al (73) found a mean longitudinal extent of 7 mm (range 0.5 - 30 mm) for AIS.  In that study lateral involvement of AIS in cone biopsy specimens ranged from 0.5 mm to 25 mm (mean, 12 mm).  In one study, the depth of AIS was 2 mm or less in 81 percent of the cases, 3 mm in 9.5 percent, and 3.5 to 5 mm in 9.5 percent (20).  There is a strong association between AIS and squamous CIN.  A co-existent microinvasive or invasive squamous cell or adenosquamous carcinoma is present in occasional cases. 

MICROINVASIVE (EARLY INVASIVE) ADENOCARCINOMA

         Microinvasive adenocarcinoma (MIA) may appear as irregular bud-like projections, small glands, or uncommonly, solid nests of cells that arise from foci of AIS (75).  The invasive foci may be associated with a stromal reaction that consists of a fibroblastic response with a chronic inflammatory cell infiltrate.  The invasive cells typically resemble those of AIS but occasionally are squamoid with abundant eosinophilic cytoplasm and enlarged, rounded nuclei with cleared chromatin and prominent nucleoli.  The presence of these squamoid cells within the glands of AIS should suggest the possibility of early stromal invasion in another plane of section, and deeper sections should be obtained. 

         In later stages of MIA, the invasive focus is measurable in two or three dimensions, corresponding to FIGO Stage IA2 microinvasive squamous cell carcinoma. Low-power examination reveals an aggregate of malignant glands that lack the orderly arrangement of normal endocervical glands or those involved by AIS.  The invasive glands are typically separated by cervical stroma but may be crowded or even confluent, and are typically haphazardly disposed.  The invasive focus in such cases, however, is usually still within the zone of endocervical stroma occupied by normal glands.  The finding of malignant glands deep to this zone, an obvious desmoplastic stromal response, the finding of lymphatic invasion, or combinations thereof, provide confirmatory evidence of invasion.  The depth of the glands, their focal confluence and irregular architecture, to varying extents are helpful in individual cases although the assessment of invasion in some cases is exceptionally difficult.

ENDOCERVICAL GLANDULAR DYSPLASIA

         The endocervical columnar epithelium may exhibit a spectrum of precancerous changes that are less severe than the changes that characterize AIS.  A number of histologic studies have made attempts to define these changes, which have been referred to as “endocervical glandular dysplasia” (EGD) (76-78).

        

         Gloor and Hurlimann (76) found foci of EGD (their cervical intraepithelial glandular neoplasia [CIGN] grades I and II of both endocervical and intestinal types) in 22 of 23 cases of AIS (their CIGN grade III).  The cells of CIGN I had slightly hyperchromatic nuclei without significant stratification; occasional mitotic figures were present.  CIGN II had pseudostratified, crowded hyperchromatic nuclei with more frequent mitotic figures and diminished intracellular mucin.  Brown and Wells (37) identified 17 cases of EGD in 105 cases of squamous CIN III within cone biopsy and hysterectomy specimens.  These investigators subclassified their cases of pure EGD into low-grade EGD (nine cases) and high-grade EGD (seven cases).  The latter lesion was characterized by nuclear stratification (greater than two-thirds of the epithelial height), enlarged, elongated, hyperchromatic, occasionally vesicular nuclei, pleomorphic nuclei with densely clumped chromatin, and an increased nuclear: cytoplasmic ratio.  Mitotic figures were rare (<1 per glandular profile).  Most of the glands had an abnormal profile, with irregular branching and budding and intraluminal cellular tufts and low papillae that lacked stromal cores.  Low-grade EGD exhibited a spectrum of appearances ranging from glandular epithelium with minimal abnormalities to changes similar to but less severe than those of high-grade EGD.  Nuclear enlargement, elongation, and hyperchromasia were noted, but vesicular nuclei were not as common as in the high-grade lesions, and stratification was limited to the basal two-thirds of the epithelium.  Mitotic figures were not seen.  Glandular profiles were abnormal and similar to those of high-grade EGD.  Both low and high-grade lesions had decreased intracellular mucin, and subnuclear vacuoles were a common finding. 

         Jaworski (71) has suggested that all lesions regarded as less than AIS be presently categorized as EGD (without subdivision into grades).  The minimal criteria used by Jaworski (1) for EGD include nuclear atypia and some evidence of cellular turnover, specifically apoptotic bodies and occasional mitotic figures (2 or fewer per gland). 

INVASIVE ADENOCARCINOMA

         A histologic classification of these tumors is presented in Table 3.  The various microscopic subtypes are discussed separately.  Those in the mixed group  are not discussed specifically as the features of the individual components are similar histologically to when seen in pure form.  A tumor is placed in the mixed group when, a neoplasm has more than one component of a specific subtype and the smaller component accounts for at least 10 percent of the neoplasm.

ADENOCARCINOMA, ENDOCERVICAL TYPE 

         Adenocarcinomas of typical or endocervical type account for approximately 70 percent of cervical adenocarcinomas(79-89).  Most of them are moderately to well differentiated mucinous adenocarcinomas characterized by glands of medium size that are lined by atypical stratified columnar cells some of which contain cytoplasmic mucin.  In some cases, however, the glands are very small and at the other extreme may be large and cystically dilated, sometimes containing abundant basophilic mucin or, less commonly, an eosinophilic secretion.  The arrangement of the glands is very variable, some being widely spaced, others forming closely packed aggregates with a cribriform pattern.  Rarely, a microglandular pattern is present  (30) or large cysts are present (7).  The tumors may contain solid areas that cannot be distinguished from poorly differentiated squamous cell carcinomas if viewed in isolation.  Papillae may be present both on the surface and within gland lumens but are uncommonly conspicuous.  The tumors may have little or no stroma, a desmoplastic stroma, or occasionally a prominent fibromatous stroma.  There is a synchronous association with premalignant squamous lesions in as many as 43 percent of the cases.  

 

        

VARIANTS OF ADENOCARCINOMA, ENDOCERVICAL TYPE

Adenoma Malignum (Minimal Deviation Adenocarcinoma)       

        

         This term is used for cervical adenocarcinomas characterized by mucinous glands, the great majority of which have a deceptively benign histological appearance (15,90-94).  On gross examination, the cervix is often firm or indurated and the mucosal surface may be hemorrhagic, friable or mucoid.  Sectioning of the wall typically discloses yellow or tan-white tissue; cysts are occasionally prominent.  Microscopic examination discloses closely to widely spaced glands that vary from small to large and are often irregular in size and shape.  They are occasionally cystically dilated, and may exhibit papillary infolding.  Most of the glands are lined by innocuous appearing mucin-containing columnar epithelial cells with basal nuclei.  Thorough sampling of the tumor in a hysterectomy specimen, however, almost always shows at least occasional glands that are lined by obviously malignant epithelium and in some cases, foci of less well differentiated adenocarcinoma are found.  A desmoplastic stromal response is usually present around at least some of the glands in most cases and may be a prominent feature. Vascular invasion is seen in approximately half the cases and perineural invasion in approximately one-sixth of them.  In most cases, the tumor is deeply invasive of the cervical wall, and transmural extension or spread to the parametrium is seen in approximately 40 percent of the cases; the myometrium is involved in a similar proportion of cases.

         The deceptively benign histologic appearance of adenoma malignum results in frequent diagnostic difficulties, particularly when only small biopsy specimens are available for examination and the diagnostic features in hysterectomy specimens may be absent.   In distinguishing the glands of adenoma malignum from abnormally-shaped benign endocervical glands, the typical stromal response associated with the former may be helpful in indicating the infiltrative nature of the glands, although this change is not seen around most neoplastic glands, is often absent in small biopsy specimens, and is sometimes seen in association with benign glandular proliferations.  The finding of glands lined by clearly malignant epithelium, as well as vascular or perineural invasion, establishes the diagnosis.  In the absence of diagnostic features on biopsy, adenoma malignum should be considered when large, abnormally-shaped glands are present.  In some superficial biopsy specimens, especially if there is artifactual distortion, the abnormalities may be insufficient to warrant a diagnosis of carcinoma, but any worrisome proliferation as well as a clinical suspicion of carcinoma should lead to additional biopsies.  The latter should be as deep as feasible because diagnostic features may be more apparent in deeper portions of the tumor. 

Villoglandular Papillary Adenocarcinoma

         The cardinal feature of this tumor (95-97) is a surface papillary component of variable thickness with papillae that are usually tall and thin, but occasionally short and broad, and that have a fibromatous stromal core.  Occasional small cellular papillae may bud off the surfaces of the larger papillae, but cellular budding of the extent seen in serous papillary carcinomas is absent.  The invasive portion of the tumors is typically composed of elongated branching glands separated by a fibromatous stroma similar to that present in the papillae , but occasionally the stroma is desmoplastic or myxoid at the advancing margin of the tumor.  The papillae and glands are usually lined by stratified non-mucinous columnar cells, but some of them are lined by a single layer of mucinous cells, a single layer of non-mucinous cells, or both.  The tumor cells typically exhibit mild to moderate nuclear atypicality and contain scattered mitotic figures.  Lymphatic or vascular invasion is rarely observed.  The adjacent cervical glandular epithelium often shows adenocarcinoma-in-situ.  A tumor should not be placed in this group if any ominous feature is present as that may lead to undertreatment of a potentially ominous lesion (98).

ENDOMETRIOID ADENOCARCINOMA

         When adenocarcinomas of endocervical type contain relatively little mucin, as they sometimes do, a superficial resemblance to endometrioid adenocarcinoma often results but careful search in these cases, sometimes with the additional help of mucin stains, usually discloses endocervical features, warranting the diagnosis of adenocarcinoma of endocervical type.  In our experience, true endometrioid carcinomas of the cervix are uncommon.  They have the characteristic histology of typically large tubular glands, sometimes with villous papillae as seen in villoglandular endometrioid carcinomas in the uterine corpus.  One endometrioid adenocarcinoma of the cervix arose from cervical endometriosis (99).   Recently it has been recognized that occasional endometrioid adenocarcinomas of the cervix have a deceptively benign appearance meriting the diagnosis minimal deviation adenocarcinoma (100,101).  These tumors are diagnosable as carcinoma, despite their bland cytologic features, on the basis of a distribution of glands and architecture of glands which is incompatible with any benign process.

CLEAR CELL ADENOCARCINOMA

         Microscopic examination reveals three basic patterns, tubulocystic, solid and papillary (102-104).  In the tubulocystic pattern, tubules and cysts of varying sizes are lined by hobnail, flat  or clear cells.  The cysts often contain mucin, and rarely intracytoplasmic mucin is present in occasional cells.  The solid pattern is characterized by nests and sheets of cells containing abundant, clear, glycogen-rich cytoplasm.  The least common of the three patterns, the papillary, is characterized by numerous papillae extending into the tubules or cysts.  The papillae vary from small and delicate with inconspicuous fibrovascular cores, to larger with more prominent fibrovascular cores that are often extensively hyalinized.  The stroma of non-papillary tumors is also occasionally prominently hyalinized.  

        

SEROUS PAPILLARY ADENOCARCINOMA

         The microscopic features are identical to serous papillary carcinomas of the ovary and uterine corpus being characterized by cellular papillae and prominent, slit-like glandular spaces, and usually moderate to severe cytologic atypia (105).  The diagnosis of primary serous carcinoma of the cervix should be made only after spread from the ovary, fallopian tube or endometrium has been excluded.

MESONEPHRIC ADENOCARCINOMA

         Mesonephric adenocarcinoma is one of the rarest subtypes of cervical adenocarcinoma with less than 20 reported cases (106).  Most of the "mesonephric carcinomas" reported in the older literature are examples of clear cell adenocarcinoma.  Microscopic examination reveals a variety of appearances, but a tubuloglandular pattern usually predominates.  The tubules and glands, which may be closely apposed, or separated by stroma, are usually small and round but are occasionally larger and resemble endometrioid glands or are cystic.  The lumens may contain bright pink or red hyaline material.  Rarely the tubuloglandular pattern merges with a solid, undifferentiated growth or slit-like spaces with a "retiform" appearance or serous carcinoma-like look.  Mesonephric carcinomas usually  extensively invade the cervical wall and also usually extend close to the overlying endocervical mucosa, sometimes eroding it.  Mesonephric hyperplasia is usually present at the periphery of the tumor.         

 

INTESTINAL-TYPE ADENOCARCINOMA

         Rare adenocarcinomas of the cervix are lined by cells with an intestinal appearance, including goblet cells, and rarely tumors in this category have the features of a colloid adenocarcinoma (107). 

SIGNET-RING CELL ADENOCARCINOMA

        

         Signet-ring cells may be seen within otherwise typical cases of cervical adenocarcinoma and within adenosquamous carcinomas as discussed below.  Pure or almost pure signet-ring cell carcinomas are exceedingly rare (108-109), and are less common than metastatic signet-ring cell carcinoma, which must be excluded before a primary cervical signet-ring cell carcinoma is diagnosed.  Microscopic examination shows cells with eccentric nuclei and pale mucin-rich cytoplasm growing singly and in clusters.

ADENOSQUAMOUS CARCINOMA

         The glandular component in adenosquamous carcinomas (110-111) is almost always of endocervical type.  If the glands are of endometrioid type, the tumor should be placed in the endometrioid category but such tumors are rare.  Occasional cervical adenosquamous carcinomas contain bland squamous morules but the squamous component in adenosquamous carcinomas is usually moderately to severely atypical.  In poorly differentiated tumors, the coexistence of squamous and glandular elements may be appreciated only on close scrutiny, and in some cases, mucin stains may be essential to demonstrate poorly-formed glands or intracellular mucin.  In occasional cases, the mucinous component of the tumor takes the form of a signet-ring cell carcinoma. 

         The definition of adenosquamous carcinoma is to some extent arbitrary, but in keeping with World Health Organization criteria for histologic typing of tumors, it is appropriate to confine the diagnosis to tumors that contain mucin or glands in addition to squamous elements that are recognizable without the use of special stains. 

GLASSY CELL CARCINOMA

         Glucksmann and Cherry (112) considered the entity that they designated "glassy cell carcinoma" the most undifferentiated form of adenosquamous carcinoma of the cervix.  Microscopic examination reveals sheets of large cells with abundant eosinophilic or amphophilic, ground-glass or finely granular cytoplasm, prominent cell borders, large nuclei with prominent nucleoli, a high mitotic rate and often a stromal inflammatory infiltrate composed predominantly of eosinophils and plasma cells (113,114).  Rare foci of squamous or glandular differentiation and intracellular mucin may be present.  Tumors with significant components of glassy cell carcinoma and typical adenocarcinoma should be regarded as mixed carcinomas. 

ADENOID BASAL CARCINOMA

         The cervix is normal on pelvic and gross pathologic examination in most of these cases (115-117).  Microscopic examination reveals widely separated or occasionally closed packed small, round, oval, or lobulated nests of uniform basaloid cells with peripheral palisading.  Lumens may form in the centers of the nests and may be cystically dilated.  The spaces can be lined by mucinous epithelium, cells with clear cytoplasm, basaloid cells, or flattened cells; some of the nests may show squamous or transitional cell differentiation.  There is rarely more than an occasional mitotic figure and there is typically no stromal response to the nests of tumor.  A premalignant squamous abnormality or microinvasive squamous cell carcinoma is usually present in the adjacent cervix, and is responsible for the associated abnormal cytologic smears.  In some cases, adenoid basal carcinoma merges with typical squamous cell carcinoma.  Some prefer to call these lesions adenoid basal epitheliomas because of their excellent prognosis (118).

“ADENOID CYSTIC” CARCINOMA

         These patients usually have an obvious cervical mass, which varies greatly in size, and may be exophytic or endophytic.  Microscopic examination shows nests of cells often with a focal cribriform pattern resembling that seen in adenoid cystic carcinoma of the salivary glands, as well as sheets, trabeculae and cords.  The glandular lumens may contain hyaline or mucinous material.  There is usually at least focal palisading of cells at the periphery of tumor nests.  The neoplastic cells are larger than those of adenoid basal carcinoma and have more pleomorphic nuclei.  The mitotic rate is generally high, and necrosis is typically present and may be extensive.  An additional difference from adenoid basal carcinoma is the presence of a stromal response, which may be myxoid, fibroblastic or hyaline.  In approximately 20 percent of adenoid cystic carcinomas, solid foci of undifferentiated carcinoma are present, giving rise to the use of the designation “solid variant” of adenoid cystic carcinoma by some (119).  Small foci of squamous differentiation are seen in some cases.   In approximately 20 percent of adenoid cystic carcinomas, a minor component of small nests and cords of basaloid cells, similar to those of adenoid basal carcinoma, is present. 

ADENOCARCINOMA ADMIXED WITH CARCINOID-SMALL CELL CARCINOMA

         Although most cervical tumors of neuroendocrine type occur in pure form, occasionally they contain a component of adenocarcinoma.  The small cell component usually predominates. 

METASTATIC ADENOCARCINOMA

Metastatic adenocarcinomas, usually primary in the gastrointestinal tract or breast, may involve the uterine cervix (120-123) and rarely may be the presenting manifestation of disease.  The morphology of these tumors in the cervix is similar to that seen at their primary sites and the diagnosis of a metastasis is aided by clinical information, an unusual histologic appearance for a primary cervical adenocarcinoma (e.g. signet-ring carcinoma), an absence of preneoplastic change, and the presence in many cases of prominent lymphatic and vascular invasion, a feature of metastatic tumors to the genital tract in general.


TABLE 3

Classification of Premalignant and Malignant Glandular Lesions

and Other Tumors with a Glandular Component

1.         Endocervical Glandular Dysplasia

2.         Adenocarcinona in situ

3.         Microinvasive adenocarcinoma

4.         Invasive Adenocarcinoma

         a.         Endocervical type

                  Variants

                          (i)  Adenoma malignum

                               (minimal deviation adenocarcinoma)

                          (ii) Villoglandular

         b.         Endometrioid

                  Variant

                          Minimal deviation adenocarcinoma

         c.         Clear cell

         d.         Serous

         e.         Mesonephric

         f.         Intestinal-type

         g.         Signet-ring cell

5.         Adenosquamous Carcinoma

                  Variant

                          Glassy Cell Carcinoma

6.        Adenoid Basal Carcinoma

7.         “Adenoid Cystic” Carcinoma

8.        Adenocarcinoma, mixed

9.        Adenocarcinoma and “carcinoid”/small cell carcinoma

10.        Metastatic adenocarcinoma


References

1.                  Michael H, Grawe L, Kraus FT.  Minimal deviation endocervical adenocarcinoma:  Clinical and histologic features, immunohistochemical staining for carcinoembryonic antigen, and differentiation from confusing benign lesions.  Int J Gynecol Pathol 3:261-276, 1984.

2.                  Combalia N, Ferreres JC, Cabezuelo A, Bella MR, Andreu J, Rey M.  Pseudotumoral epithelial lesions of uterine cervix.  Abstract No. 438.  Presented at International Academy of Pathology, Madrid, Spain, 1992.

3.                  Fluhmann CF.  The cervix uteri and its diseases.  W.B. Saunders Co. Philadelphia, 1961.

4.                  Fluhmann CF.  Focal hyperplasia (tunnel clusters) of the cervix uteri.  Obstet Gynecol 17:206-214, 1961

5.                  Segal GH, Hart WR.  Cystic endocervical tunnel clusters.  A clinicopathologic study of 29 cases of so-called adenomatous hyperplasia.  Am J Surg Pathol 14:895-903, 1990

6.                  Jones MA, Young RH.  Type A tunnel clusters with cytologic atypia.  A report of 14 cases.  Am J Surg Pathol 20:1312-1318, 1996.

7.                  Tambouret R, Bell DA, Young RH.  Cystic endocervical adenocarcinomas:  A report of eight cases emphasizing the potential for misdiagnosis as a benign lesion.  Am J Surg Pathol 24:369-374, 2000.

8.                  Daya D, Young RH.  Florid deep glands of the uterine cervix:  another mimic of adenoma malignum.  Am J Clin Pathol 103:614-617, 1995.

9.                  Noda K, Kimura K, Ikeda M, Teshima K.  Studies on the histogenesis of cervical adenocarcinoma.  Int J Gynecol Pathol 1:336-346, 1983

10.              Clement PB, Young RH.  Deep Nabothian cysts of the endocervix.  A possible source of confusion with minimal-deviation adenocarcinoma (adenoma malignum).  Int J Gynecol Pathol 8:340-348, 1989

11.              Nucci MR, Clement PB, Young RH.  Lobular endocervical gland hyperplasia.  A report of 13 cases and comparison with adenoma malignum.  Am J Surg Pathol, 23:886-891, 1999.

12.              Fetissof F, Heitzman A, Machet M-C, Lansac J.  Unusual endocervical lesions with endocrine cells.  Path Res Pract 189: 928-939, 1993.

13.              Ishii K, Hosaka N, Toki T, Momose M, Hidaka E, Tsuchiya S, Katsuyama T.  A new view of the so-called adenoma malignum of the uterine cervix.  Virchows Arch 432:315-322, 1998.

14.              Mikami Y, Hata S, Fujiwara K, Imajo Y, Kohno I, Manabe T.  Florid endocervical glandular hyperplasia with intestinal and pyloric gland metaplasia:  Worrisome benign mimic of “adenoma malignum”.  Gynecol Oncol 74:504-511, 1999.

15.              Gilks CB, Young RH, Aguirre P et al.  Adenoma malignum (minimal deviation adenocarcinoma) of the uterine cervix:  A clinicopathological and immunohistochemical analysis of 26 cases.  Am J Surg Pathol 13:717-729, 1989

16.              Tsuda H, Okada S, Kasamatsu T.  Is pyloric gland metaplasia different from adenoma malignum?  (Letter) Gynecol Oncol 77:341-342, 2000.

17.              Mikami Y.  Reply (to letter of Tsuda et al).  Gynecol Oncol 77:342-343, 2000.

18.              Jones MA, Young RH, Scully RE.  Diffuse laminar endocervical glandular hyperplasia:  A report of seven cases.  Am J Surg Pathol 15:1123-1129, 1991.

19.              Maruyama R, Nagaoka S, Terao K, Honda M, Koita H.  Diffuse laminar endocervical glandular hyperplasia. Pathology International 45:283-286, 1995.

20.              Taylor HB, Ivey NS, Norris HJ.  Atypical endocervical hyperplasia in women taking oral contraceptives.  JAMA 202:185-187, 1967

21.              Kyriakos M, Kempson RL, Konikov NF.  A clinical and pathologic study of endocervical lesions associated with oral contraceptives.  Cancer 22:99-110, 1968

22.              Candy MD, Abell MR.  Progestogen-induced adenomatous hyperplasia of the uterine cervix.  JAMA 203:323-326, 1968

23.              Govan ADT, Black WP, Sharp JL.  Aberrant glandular polypi of the uterine cervix associated with contraceptive pills:  Pathology and pathogenesis.  J Clin Path 22:84-89, 1969

24.              Nichols TM, Fidler HK.  Microglandular hyperplasia in cervical cone biopsies taken for suspicious and positive cytology.  Am J Clin Pathol 56:424-429, 1971

25.              Wilkinson E, Dufour DR.  Pathogenesis of microglandular hyperplasia of the cervix uteri.  Obstet Gynecol 47:189-195, 1976

26.              Leslie KO, Silverberg SG.  Microglandular hyperplasia of the cervix:  Unusual clinical and pathological presentations and their differential diagnosis.  Prog Surg Pathol 5:95-114, 1984

27.              Chumas JC, Nelson B, Mann WJ et al.  Microglandular hyperplasia of the uterine cervix.  Obstet Gynecol 66:406-409, 1985

28.              Greeley C, Schroeder S, Silverberg SG.  Microglandular hyperplasia of the uterine cervix:  A true “pill” lesion?  Int J Gynecol Pathol 14:50-54, 1995.

29.              Young RH, Scully RE.  Atypical forms of microglandular hyperplasia of the cervix simulating  carcinoma:  A report of five cases and review of the literature.  Am J Surg Pathol 13:50-56, 1989

30.              Young RH, Scully RE.  Uterine carcinomas simulating microglandular hyperplasia.  A report of six cases.  Am J Surg Pathol 16:1092-1097, 1992.

31.              Speers WC, Picaso LG, Silverberg SG.  Immunohistochemical localization of carcinoembryonic antigen in microglandular hyperplasia and adenocarcinoma of the endocervix.  Am J Clin Pathol 79:105-107, 1983

32.              Steeper TA, Wick MR.  Minimal deviation adenocarcinoma of the uterine cervix ("adenoma malignum").  An immunohistochemical comparison with microglandular endocervical hyperplasia and conventional endocervical adenocarcinoma.  Cancer 58:1131-1138, 1986.

33.              Ferry JA, Scully RE.  Mesonephric remnants, hyperplasia and neoplasia in the uterine cervix:  A study of 49 cases.  Am J Surg Pathol 14:1100-1111, 1990.

34.              Seidman JD, Tavassoli FA.  Mesonephric hyperplasia of the uterine cervix:  A clinicopathologic study of 51 cases.  Int J Gynecol Pathol 14:293-299, 1995.

35.              Jones MA, Andrews J, Tarraza H.  Mesonephric remnant hyperplasia of the cervix: A clinicopathologic analysis of 14 cases.  Gynecol Oncol 49:41-47, 1993.

36.              Lang G, Dallenbach-Hellweg G.  The histogenetic origin of cervical mesonephric hyperplasia and mesonephric adenocarcinoma of the uterine cervix studied with immunohistochemical methods.  Int J Gynecol Pathol 9:145-157, 1990.

37.              Brown LJR, Wells M.  Cervical glandular atypia associated with squamous intraepithelial neoplasia:  A premalignant lesion?  J Clin Pathol 39:22-28, 1986.

38.              Suh K-S, Silverberg SG.  Tubal metaplasia of the uterine cervix.  Int J Gynecol Pathol 9:122-128, 1990

39.              Pacey F, Ayer B, Greenberg M.  The cytologic diagnosis of adenocarcinoma in situ of the cervix uteri and related lesions.  III.  Pitfalls in Diagnosis.  Acta Cytol 32:325-330, 1988.

40.              Novotny DB, Maygarden JJ, Johnson DE, Frable WJ.  Tubal metaplasia.  A frequent potential pitfall in the cytologic diagnosis of endocervical glandular dysplasia on cervical smears.  Acta Cytol 36:1-10, 1992.

41.              Jonasson JG, Wang HH, Antonioli DA Ducatman BS.  Tubal metaplasia of the uterine cervix:  A prevalence study in patients with gynecologic pathologic findings.  Int J Gynecol Pathol 11:89-95, 1992.

42.              Al-Nufussi A, Rahilly M.  The prevalence of tubo-endometrial metaplasia and adenomatoid proliferation.  Histopathology 22:177-179, 1993.

43.              Oliva E, Clement PB, Young RH.  Tubal and tubo-endometrioid metaplasia of the uterine cervix.  Unemphasized features that may cause problems in differential diagnosis:  A report of 25 cases.  Am J Clin Pathol 103:618-623, 1995.

44.              Ismail SM.  Cone biopsy causes cervical endometriosis and tubo-endometrioid metaplasia.  Histopathology 18:107-114, 1991

45.              Marques T, DeAngelo, Andrade AL, Vassallo J.  Endocervical tubal metaplasia and adenocarcinoma in situ:  role of immunohistochemistry for carcinoembryonic antigen and vimentin in differential diagnosis.  Histopathology 28:549-550, 1996.

46.              Schlesinger C, Silverberg SG.  Endocervical adenocarcinoma in situ of tubal type and its relation to atypical tubal metaplasia.  Int J Gynecol Pathol 18:1-4, 1999

47.              Biscotti CV, Hart WR.   Apoptotic bodies.  A consistent morphologic feature of endocervical adenocarcinoma in situ.  Am J Surg Pathol 22:434-439, 1998.

48.              Clement PB.  Pathology of endometriosis.  Pathol Annu 25(Pt. 1):245-295, 1990

49.              Overton, DH Jr., Wilson RB, Dockerty MB.  Primary endometriosis of the cervix.  Am J Obstet Gynecol 79:768-779, 1960

50.              Branscomb L.  Habitual premenstrual spotting following electrocauterization of the cervix:  a newly observed phenomenon.  Am J Obstet Gynecol 79:16-23, 1960

51.              Williams GA.  Endometriosis of the cervix uteri – a common disease.  Am J Obstet Gynecol 80:734-741, 1960

52.              Wolfe SA, Mackles A, Greene HJ.  Endometriosis of the cervix.  Classification and analysis of 17 cases.  Am J Obstet Gynecol 81:111-123, 1961.

53.              Baker PM, Clement PB, Bell DA, Young RH.  Superficial endometriosis of the uterine cervix:  A report of 20 cases of a process that may be confused with endocervical glandular dysplasia or adenocarcinoma in situ.  Int J Gynecol Pathol 18:198-205, 1999.

54.              Symonds DA, Reed TP, Didolkar SM, Graham RR.  AGUS in cervical endometriosis.  J Reprod Med 42:39-43, 1997.

55.              Clement PB, Young, RH.  Florid cystic endosalpingiosis with tumor-like manifestations.  Am J Surg Pathol, 23: 166-175, 1999.

56.              Young RH, Clement PB.  Endocervicosis involving the uterine cervix.  A report of four cases of a benign process that may be confused with deeply invasive endocervical adenocarcinoma.  Int J Gynecol Pathol.  19:322-328, 2000.

57.              Jones MA, Young RH.  Atypical oxyphilic metaplasia of the endocervical epithelium.  A report of nine cases.  Int J Gynecol Pathol 16:99-102, 1997.

58.              Trowell JE.  Intestinal metaplasia with argentaffin cells in the uterine cervix.  Histopathology 9:551-559, 1985

59.              Cove H.  The Arias-Stella reaction occurring in the endocervix in pregnancy.  Recognition and comparison with adenocarcinoma of the endocervix.  Am J Surg Pathol 3:567-568, 1979

60.              Schneider V.  Arias-Stella reaction of the endocervix.  Frequency and location.  Acta Cytol 25:224-228, 1981

61.              Larraza-Hernandez O, Molberg KH, Lindberg G, Albores-Saavedra J.  Ectopic prostate tissue in the uterine cervix.  Int J Gynecol Pathol 16:291-293, 1997.

62.              Nucci MR, Ferry JA, Young RH.  Ectopic prostate tissue in the uterine cervix:  A report of four cases and review of ectopic prostatic tissue.  Am J Surg Pathol. 24:1224-1230, 2000.

63.              Josey WE, Nahmias AJ, Naib ZM.  Viral and Virus-like infections of the female genital tract.  Clin Obstet Gynecol 12:161-178, 1969.

64.              Lesack D, Wahab I, Gilks CB.  Radiation-induced atypia of endocervical epithelium:  A histological, immunohistochemical and cytometric study.  Int J Gynecol Pathol 15:242-247, 1996.

65.              Gilks CB, Young RH, Clement PB, Hart WR, Scully RE.  Adenomyomas of the uterine cervix of endocervical type:  A report of ten cases of a benign cervical tumor that may be confused with adenoma malignum.  Mod Pathol 9:220-224, 1996.

66.              Gilks CB, Clement PB, Hart WR, Young RH.  Uterine adenomyomas excluding atypical polypoid adenomyomas and adenomyomas of endocervical type:  A clinicopathologic study of 30 cases of an underemphasized lesion that may cause diagnostic problems with brief consideration of adenomyomas of other female genital tract sites.  Int J Gynecol Pathol 19:195-205, 2000.

67.              Mazur MT.  Atypical polypoid adenomyomas of the endometrium.  Am J Surg Pathol 5:473-82, 1981.

68.              Young RH, Treger T, Scully RE.  Atypical polypoid adenomyoma of the uterus:  a report of 27 cases.  Am J Clin Pathol 86:139-145, 1986.

69.              Michael H, Sutton G, Hull MT, Roth LM.  Villous adenoma of the uterine cervix associated with invasive adenocarcinoma:  A histologic, ultrastructural, and immunohistochemical study.  Int J Gynecol Pathol 5:163-169, 1986.

70.              Smith YR, Quint EH, Hinton EL.  Recurrent benign mullerian papilloma of the cervix.  J Pediatr Adolesc Gynecol 11:29-31, 1998.

71.              Jaworski RC.  Endocervical gland dysplasia, adenocarcinoma in situ and early invasive (microinvasive) adenocarcinoma of the uterine cervix.  Semin Diagn Pathol 7:190-204, 1990.

72.              Qizilbash AH.  In-situ and microinvasive adenocarcinoma of the uterine cervix.  A clinical, cytologic and histologic study of 14 cases.  Am J Clin Pathol 64:155-170, 1975.

73.              Ostor AG, Pagano R, Davoren RAM, Fortune DW, Chanen W, Rome R.  Adenocarcinoma in situ of the cervix.  Int J Gynecol Pathol 3:179-190, 1984.

74.              Jaworski RC, Pacey NF, Greenberg ML, Osborn RA.  The histologic diagnosis of adenocarcinoma in situ and related lesions of the cervix uteri.  Adenocarcinoma in situ.  Cancer 61:1171-1181, 1988.

75.              Ostor AG.  Early invasive adenocarcinoma of the cervix.  Int J Gynecol Pathol 19:29-38, 2000.

76.              Gloor E, Hurlimann J.  Cervical intraepithelial glandular neoplasia (adenocarcinoma in situ and glandular dysplasia).  A correlative study of 23 cases with histologic grading, histochemical analysis of mucins, and immunohistochemical determination of the affinity for four lectins.  Cancer 58:1272-1280, 1986.

77.              Lee KR, Sun D, Crum CP.  Endocervical intraepithelial glandular atypia (dysplasia):  a histopathologic, human papillomavirus, and MIB-1 analysis of 25 cases.   Hum Pathol 31:656-664, 2000.

78.              Goldstein NS, Ahmad E, Hussain M, Hankin RC, Perez-Reyes N.  Endocervical glandular atypia.  Does a pre-neoplastic lesion of adenocarcinoma in situ exist?  Am J Clin Pathol 110:200-209.

79.              Abell MR, Gosling JRG.  Gland cell carcinoma (adenocarcinoma) of the uterine cervix.  Am J Obstet Gynecol 83:729-755, 1962.

80.              Gallup DG, Abell MR.  Invasive adenocarcinoma of the uterine cervix.  Obstet Gynecol 49:596-603, 1977.

81.              Dallenbach-Hellweg G.  On the origin and histological structure of adenocarcinoma of the endocervix in women under 50 years of age.  Path Res Pract 179:38-50, 1984.

82.              Korhonen MO, Adenocarcinoma of the uterine cervix.  Prognosis and prognostic significance of histology.  Cancer 53:1760-1763, 1984.

83.              Haggard JL, Cotten N, Dougherty CM, Mickal A.  Primary adenocarcinoma of the cervix.  Obstet Gynecol 24:183-193, 1964.

84.              Hurt WG, Silverberg SG, Frable WJ, Belgrad R, Crooks LD.  Adenocarcinoma of the cervix: Histopathologic and clinical features.  Am J Obstet Gynecol 129:304-315, 1977.

85.              Shingleton HM, Gore H, Bradley DH, Soong SJ.  Adenocarcinoma of the cervix.  I. Clinical evaluation and pathologic features.  Obstet Gynecol 139:799-814, 1981.

86.              Fu YS, Reagan JW, Hsiu JG, Storaasli JP, Wentz WB.  Adenocarcinoma and mixed carcinoma of the uterine cervix. 1. A clinicopathologic study.  Cancer 49:2560-2570, 1982.

87.              Ireland D, Hardiman P, Monaghan JM.  Adenocarcinoma of the uterine cervix: A study of 73 cases.  Obstet Gynecol 65:82-85, 1985.

88.              Mober PJ, Einhorn N, Silfversward C, Soderberg G.  Adenocarcinoma of the uterine cervix.  Cancer 57:407-410, 1986.

89.              Hopkins MP, Schmidt RW, Roberts JA, Morley GW.  Gland cell carcinoma (adenocarcinoma) of the cervix.  Obstet Gynecol 72:789-795, 1988.

90.              McKelvey JL, Goodlin RR.  Adenoma malignum of the cervix.  Cancer  16:549-557, 1963.

91.              Silverberg SG, Hurt WG.  Minimal deviation adenocarcinoma ("adenoma malignum") of the cervix.  Am J Obstet Gynecol 123:971-975, 1975.

92.              Kaku T, Enjoji M.  Extremely well-differentiated adenocarcinoma  ("adenoma malignum") of the cervix.  Int J Gynecol Pathol 2:28-41, 1983.

93.              Kaminski PF, Norris HJ.  Minimal deviation carcinoma (adenoma malignum) of the cervix.  Int J Gynecol Pathol 2:141-153, 1983.

94.              Steeper TA, Wick MR.  Minimal deviation adenocarcinoma of the uterine cervix ("Adenoma Malignum").  Cancer 58:1131-1138, 1986.

95.              Young RH, Scully RE.  Villoglandular papillary adenocarcinoma of the uterine cervix.  A clinicopathological analysis of 13 cases.  Cancer 63:1773-1779, 1989.

96.              Hopson L, Jones MA, Boyce CR, Tarraza HM.  Papillary villoglandular carcinoma of the cervix.  Gynecol Oncol 39:221-224, 1990.

97.              Jones MW, Silverberg SG, Kurman RJ.  Well differentiated villoglandular adenocarcinoma of uterine cervix:  A clinico-pathological study of 24 cases.  Int J Gynecol Pathol 12:1-7, 1993.

98.              Kaku T, Kamura T, Shigematsu T, et al.  Adenocarcinoma of the uterine cervix with predominantly villoglandular papillary growth pattern.  Gynecol Oncol 64:147-152, 1997.

99.              Chang SH, Maddox WA.  Adenocarcinoma arising within cervical endometriosis and invading the adjacent vagina.  Am J Obstet Gynecol 110:1015-1017, 1971.

100.          Rahilly MA, Williams ARW, Al-Nafussi A.  Minimal deviation endometrioid adenocarcinoma of cervix:  A clinicopathological and immunohistochemical study of two cases.  Histopathol 20:351-354, 1992.

101.          Young RH, Scully RE.  Minimal deviation endometrioid adenocarcinoma of the uterine cervix.  A report of five cases of a distinctive neoplasm that may be misinterpreted as benign.  Am J Surg Pathol 17:660-665, 1993.

102.          Hart WR, Norris HJ.  Mesonephric adenocarcinomas of the cervix.  Cancer 29:106-113, 1972.

103.          Fawcett KJ, Dockerty MB, Hunt AB.  Mesonephric carcinomas and adenocarcinomas of the cervix in children.  J Pediatr 69:104-110, 1966.

104.          Kaminski PF, Maier RC.  Clear cell adenocarcinoma of the cervix unrelated to diethylstilbestrol exposure.  Obstet Gynecol 62:720-727, 1983.

105.          Zhou C, Gilks CB, Hayes M, Clement PB.  Papillary serous carcinoma of the uterine cervix.  A clinicopathologic study of 17 cases.  Am J Surg Pathol 22:113-120, 1998.

106.          Clement PB, Young RH, Keh P, Ostor AG, Scully RE.   Malignant mesonephric neoplasms of the uterine cervix.  A report of eight cases, including four with a malignant spindle cell component.  Am J Surg Pathol 19:1158-1171, 1995

107.          Fox H, Wells M, Harris M, McWilliam LJ, Anderson GS.  Enteric tumours of the lower female genital tract: a report of three cases.  Histopathol 12:167-176, 1988.

108.          Haswani P, Arseneau J, Ferenczy A.  Primary signet ring cell carcinoma of the uterine cervix:  A clinicopathologic study of two cases with review of the literature.  Int J Gynecol Cancer 8:374-379, 1998.

109.          Mayora M, Garcia-Valtuille A, Fernandez F et al.  Adenocarcinoma of the uterine cervix with massive signet-ring cell differentiation.  Int J Surg Pathol 5:95-100, 1997.

110.          Dougherty CM, Cotten N.  Mixed squamous cell and adenocarcinoma of the cervix.  Cancer 17:1132-1143, 1964.

111.          Glucksmann A, Cherry CP.  Incidence, histology and response to radiation of mixed carcinomas (adenoacanthomas) of the uterine cervix.  Cancer 9:971-979, 1956.

112.          Cherry CP, Glucksmann A.  Histology of carcinomas of the uterine cervix and survival rates in pregnant and non-pregnant patients.  Surg Gynecol Obstet 111:763-776, 1961.

113.          Littman P, Clement PB, Henriksen B, Wang CC, Robboy SJ, Taft PD, Ulfelder H, Scully RE.  Glassy cell carcinoma of the cervix.  Cancer 37:2238-2246, 1976.

114.          Talerman A, Alenghat E, Okagaki T.  Glassy cell carcinoma of the uterine cervix.  APMIS 23:119-125, 1991.

115.          van Dinh T, Woodruff JD.  Adenoid cystic and adenoid basal carcinomas of the cervix.  Obstet Gynecol 65:705-708, 1985.

116.          Baggish MS, Woodruff JD.  Adenoid basal lesions of the cervix.  Obstet Gynecol  37:807-819, 1987.

117.          Ferry JA, Scully RE.  "Adenoid cystic" carcinoma and adenoid basal carcinoma of the uterine cervix.  A study of 28 cases.  Am J Surg Pathol 12:134-144, 1988.

118.          Brainard JA, Hart WR.  Adenoid basal epitheliomas of the uterine cervix.  A reevaluation of distinctive cervical basaloid lesions currently classified as adenoid basal carcinoma and adenoid basal hyperplasia.  Am J Surg Pathol 22:965-975, 1998.

119.          Albores-Saavedra J, Manivel C, Mora A, Vuitch F, Milchgrub S, Gould E.  The solid variant of adenoid cystic carcinoma of the cervix.  Int J Gynecol Pathol 11:2-10, 1992.

120.          Korhonen M, Stenback F.  Adenocarcinoma metastatic to the uterine cervix.  Gynecol Obstet Invest 17:57-65, 1984.

121.          Lemoine NR, Hall PA.  Epithelial tumors metastatic to the uterine cervix.  A study of 63 cases and review of the literature.  Cancer 57:2002-2005, 1986.

122.          Yazigi R, Sandstad J, Monoz AK.  Breast carcinoma metastasizing to the uterine cervix.  Cancer 61:2558-2560, 1988.

123.          Imachi M, Tsukamoto N, Amagase H, Shigematsu T, Amada S, Nakano H.  Metastatic adenocarcinoma to the uterine cervix from gastric cancer.  A clinicopathologic analysis of 16 cases.  Cancer 71:3472-3477, 1993.