The 66 previously reported morphologically pure ABL in the English literature are summarized in table 1. Cases associated with "separate malignancies" (vide infra) have been excluded. However, we elected to keep 3 cases with microinvasive carcinoma, because the latter only develops in the setting of high grade dysplasia – a nearly ubiquitous finding in these patients – and because their presence is not expected to alter the natural history of the ABL any more than the high grade dysplasia alone. For the 52 patients in which this information was itemized, patient age ranged from 21 to 91 years (mean 64.6; Median 66.5). Although some of the earlier series 2324 as well as those emanating out of South Africa 79 seemed to suggest a marked predilection for black females, others have found no such predilection 2021. However, ABL does seem to disproportionately affect non-white women. Approximately 80% of patients were asymptomatic at initial diagnosis, and 80–90% of them came to clinical attention due to an abnormal pap smear; the latter was a high-grade squamous intraepithelial lesion in >90% of cases. A pelvic examination fails to reveal an abnormality in almost all cases of pure ABL. Similarly, pathologic evaluation of the hysterectomy or conization specimens usually fails to reveal a mass lesion. Cervical ulceration, hemorrhage or eversion was identified in 7 cases, and a mass lesion was noted in 2 13.

Morphologically pure ABL are not macroscopically evident. As previously noted, >90% of cases have an associated high-grade squamous intraepithelial lesion. In the subepithelial compartment is typically the incidentally discovered epithelial proliferation of ABL. The latter is comprised of multiple, small nests of basaloid epithelial cells. Scanning magnification often reveals an apparently deeply invasive process (figure 1); nests extended down into the stroma for distances that ranged from 2 mm to 10 mm (mean 4.3 mm) in one study 10 and invaded to 50–90% of the cervical stroma in another 13. Based in part on the depth of invasion, 6 (46%) of 13 and 7 (54%) of 13 cases were International Federation of Gynecology and Obstetrics stage 1A and 1B respectively in one study 20. Extensions into the myometrium of the isthmus have also been reported 17. The nests are often arranged in a vague lobule-like configuration 10. Each nest is comprised of monomorphic, small cells with basaloid, round to oval nuclei, inconspicuous nucleoli and scant cytoplasms (figure 2 and 3). Rare examples may show prominent clear cell change 9 (figure 4). The cells often palisade around the peripheral zones of the nests, in a pattern reminiscent of cutaneous basal cell carcinoma 20. The nests are most commonly small (<100 cells) and solid, but may form glandular or even cribriform structures (figure 5). The latter were lined by cells with ciliary structures suggestive of tuboendometrioid differentiation in one case 1. In support of their true glandular nature, another case was found to be mucicarcmine positive in these foci 14. Secretory material may occasionally be identified within the lumen of the glandular structures (figure 6). Small basement membrane-like material may also be occasionally encountered (figure 7). Foci of squamous metaplasia may involve nests of ABL (figure 8), and such metaplasia may be so extensive as to cause the lesion to simulate an invasive squamous cell carcinoma (SCC). This potential pitfall is further magnified because the metaplastic nests are typically expanded and may contain cells showing atypia. In our experience, all cases showing this pattern have been involved by low-grade dysplasia and transitions from clefts with more clearly diagnostic foci were evident. Other authors 10 have noted that these nests typically have a faint, CAM 5.2 immunopositive rim of basaloid cells at their peripheries. These squamous nests may show prominent microcyst formation 413.

In their seminal study, Baggish and Woodruff emphasized that the nests immediately subjacent to the overlying epithelium seemed to be "dropping off" from it and were "well-demarcated from the adjacent stroma" 24. Those authors have also emphasized the absence of any significant stromal reaction to the nests (figure 2), in contrast to the stromal hyalinization that may be seen in ACC. Brainard and Hart 10, however, noted that in 50% of their 12 cases, foci of "stroma edema or loosening" were present around the ABL nests and a small lymphoplasmacytic infiltrate was present around the nests in all 12 cases. Jones et al 13 also noted that "focal stromal reactions" were present in the most deeply invasive nests of their tumors, and around those nests with squamous differentiation. Layton-Henry et al 17 also illustrated in their report an adenoid basal nest that displayed squamous differentiation and an inflammatory stromal reaction. Nests of ABL should not show necrosis, perineural invasion, lymphovascular invasion or a significant desmoplastic reaction 102088; the presence of one or more of the latter features should trigger a re-evaluation of a putative diagnosis.

ABL may occasionally be associated with conventional malignancies within the same specimen. Morphologic transitions have been identified, albeit inconstantly, between the pure adenoid basal nests and their associated malignancies 3. The latter may also be a focal finding within an otherwise morphologically pure adenoid basal lesion 9. Some authors 9 have interpreted this phenomenon as examples of "divergent epithelial differentiation" in ABL. For the purpose of this review, we would refer to these as "separate" malignancies. Although this determination is undoubtedly limited by a selection bias inherent in the reported cases, ABL appears to be associated with these separate malignancies at a frequency that is much higher than would be predicted for high-grade dysplasia alone. This frequency is also much higher than would be predicted for any pseudoneoplastic proliferation of the cervix 90, which some authors have suggested ABL represents 108588. The precise frequency with which ABL is associated with separate malignancies is difficult to estimate. In their 1971 series of adenoid basal "lesions", 2 of the 5 cases reported by Baggish and Woodruff were associated with an ACC and an "epidermoid" carcinoma 23. The recent study from Parwani et al 3 perhaps provides the best insight into this issue: these investigators searched the routine and consultation files of a large academic center (Baltimore, MD, USA) for the period between 1984 and 2004. The precise distribution of consultation versus routine cases was not outlined. Thirty-three potential cases were identified. The authors excluded from their study "typical low grade adenoid basal tumors" (n = 19), which presumably represented cases unassociated with separate cancers. Of the remaining 14 cases, 10 were retrievable and all were associated with a carcinoma of non-ABL differentiation. The associated carcinomas in 8 of the 10 cases were invasive squamous cell carcinomas (1 microinvasive), 1 was a mixture of SCC and ACC and 1 was a mixture of SCC and small cell neuroendocrine carcinoma. However, in what remains the largest series of ABL in the literature, Ferry and Scully 20, did not identify an associated carcinoma in 13 of their 14 cases (opinions differ as to whether their 14^th case could be considered a true ABL, vide infra). In the aforementioned study of Parwani et al 3, all 10 cases showed morphologic transitions between the ABL component and the associated conventional malignancy. The latter were distinguished from the former by virtue of their greater degree of nuclear atypia, increased mitotic activity and the presence of a stromal desmoplastic reaction. However, the invasive component, which as noted previously was most commonly SCC, also showed various degrees of "adenoid basal" differentiation ("characterized by cells in the nests with scanty cytoplasm, peripheral palisading, and focal gland formation") 3.

The relationship between ACC and ABL, separate from their aforementioned historical co-evolution, is worthy of mention. These tumors share some morphologic features, are frequently associated with cervical dysplasia, and both disproportionately affect the non-white, postmenopausal population 20. However, their differences are more significant, and the distinction between these 2 neoplasms is of critical importance given the biologic differences between them. In contrast to patients with pure ABL, the majority of ACC patients present with postmenopausal bleeding and an apparent mass lesion 92021. Microscopically, the nests of ACC are larger, show necrosis, a higher mitotic rate and nuclear pleomorphism. ACC is also typically associated with a significant stromal reaction 92021.

The aforementioned differences notwithstanding, there is at least a morphologic relationship between ACC and ABL. Varying amounts of ACC may exist in ABL and vice-versa 39217982858789, and morphologic admixtures and transitions have been observed between the 2 areas in tumors that are comprised of both. It has been stated that small ABL-like nests are present in up to 20% of cervical ACC 8689. Grayson and colleagues 9 have championed the viewpoint that ABL and ACC exist along a morphologic spectrum of basaloid neoplasms of the cervix, and that ABL may even represent a precursor to ACC (see "histogenesis and pathogenesis" below).

In addition to ACC and SCC, ABL also appears to be associated with cervical carcinosarcomas, usually admixed with an SCC, at a disproportionately high rate given its overall frequency 5791. ABL formed a major part of the epithelial component in 25% of the 8 carcinosarcomas of the cervix reported by Grayson et al 7. Similarly, 2 of the 9 cases reported by Clement et al 91 had ABL as an epithelial component, albeit minor. These 2 studies represent the largest series of cervical carcinosarcomas in the literature.

Teramoto et al 2 recently reported 7 examples of adenosquamous cell carcinoma of the cervix showing ABL-like areas. Grayson et al 9 had previously reported similar findings, but with the adenoid basal component predominating. Small ABL-like nests have also been identified adjacent to cervical basaloid carcinomas, ie squamous cell carcinomas with basaloid features 10

We attempted to itemize all cases in the literature in which an ABL or ABL-like lesion was identified adjacent to a separate conventional malignancy. Almost 30 cases could be retrieved 235792391. The aforementioned selection bias notwithstanding, given that only 66 pure ABL have been reported, this figure appears to be substantial. However, we emphasize that the precise frequency is not readily quantifiable at present time.

Clinicopathologic data for 66 previously reported pure ABL are summarized in Table 1. The distribution of diagnostic and/or therapeutic measures that were administered for the 61 patients in which this data was available and/or sufficiently itemized, was as follows: conization, loop electrosurgical procedure (LEEP), or cervical excision alone (n = 11, 18%), conization or LEEP or cervical excision followed by radiotherapy (n = 2, 3.3%), conization or LEEP followed by hysterectomy ± unilateral or bilateral salpingoo-oophorectomy with or without lymph node sampling (n = 13, 21.3%), hysterectomy ± unilateral or bilateral salpingoo-oophorectomy alone with or without lymph node sampling (n = 30, 49.18%, 2 patients in this group also received chemotherapy that was unrelated to cervical pathology), hysterectomy ± unilateral or bilateral salpingoo-oophorectomy and radiotherapy with or without lymph node sampling (n = 3, 5%), Conization, bilateral salpingoo-oophorectomy, lymph node sampling and radiation (n = 1, 1.6%), dilatation and curettage (n = 1, 1.6%). Follow-up information was available for analysis in 58 (89%) of the 66 previously reported patients. 3 (5.08%) of these 59 patients died of other causes. Of the remaining 56 patients, follow-up information was itemized on each individual patient in 47 cases. None of these patients had experienced tumor recurrence, metastases or tumor-related death, with follow-up periods ranging from 1 to 156 months (mean 44.7; median 28.).

One of the cases reported by Ferry and Scully 20 has garnered a lot of attention as the only example of an alleged adenoid basal lesion that metastasized and killed the patient (within 3 months). As the authors themselves noted, the morphologic features of this tumor differed significantly from the others in their series: "the tumor cells grew slender cords that penetrated deeply into a stroma showing striking myxoid change, analogous to the morphea type of basal cell carcinoma of the skin" 20. Hart 85 has noted that this case was "probably better classified as a high-grade invasive carcinoma with adenoid basal-like features". Ferry 88, in an editorial 9 years later, wrote that "the term ABC, without qualification, should be used only in cases with typical morphology when the tumor is composed solely of ABC (with or without associated squamous dysplasia)". We concur with the latter statement, and since we do not consider their case 14 20 a morphologically pure ABL, it was excluded from our analysis above and elsewhere in this paper.

In summary, patients that reportedly have been diagnosed with morphologically pure ABL have had excellent outcomes, irrespective of the modality of treatment. When adenoid basal nests are associated with separate conventional malignancies, it can be anticipated that the overall prognosis will be largely dependent on the stage, grade, histologic subtype and other clinicopathologic parameters of the associated malignancy.

There has been considerable controversy with respect to the precise nomenclature applied to ABL, a likely product of the continued absence of a complete understanding of their true nature and biologic potential. As noted previously, the term "adenoid cystic basal carcinoma" was initially applied by Moss and Collins 82 to describe a cervical neoplasm that would probably be considered, using contemporary criteria, a mixed ACC and ABC. The seminal 1966 report by Baggish and Woodruff 24 was captioned "adenoid-basal carcinoma of the cervix" and served to formally introduce the term ABC into the medical lexicon. However, the 3 lesions were described throughout the text as "adenoid-basal lesion(s)" and the authors even stated at the end of their text that "it would seem wise to apply the term "adenoid-basal cell tumor" to this unusual cervical neoplasm" 24. The term ABC was only used in the title of their report. In a report of 5 additional cases in 1971, the aforementioned authors entitled their report "adenoid basal lesions of the cervix" but this time used the term ABC in their text to describe not only the new cases but the cases they previously reported 23. These 5 cases were comprised in part of 2 cases of ABC associated with a cylindroma and an epidermoid carcinoma. The authors designated 2 of the remaining 3 cases "adenoid basal hyperplasia" whereas the 3^rd an ABC. Adenoid basal hyperplasia (ABH) was not precisely defined, and the only clinicopathologic difference between the 2 cases of ABH and the 1 case of ABC appeared to be that the basaloid proliferation invaded deep into the cervical stroma in the latter and was superficial in the former. The term ABH largely remained dormant in the medical literature until it was resurrected again in a 1985 report of van Dinh and Woodruff 21. In that report, the authors noted that in association with "many" of their cases of ABC were areas of ABH, "consisting of buds and irregular proliferations of basal cells attached to the epithelium" and that in "no case did the basal cell buds extend more than 5 mm into the stroma" 21. They further noted that this phenomenon was seen on 11 occasions amongst all the hysterectomies that they evaluated over a 12-year period. Three additional examples of ABH were reported by Brainard and Hart 10 in 1998. The cases of ABH were described as resembling ABC except that "the epithelial nests were smaller, and many appeared to remain attached to the adjacent squamous mucosa or endocervical clefts". In our opinion, the distinctions that have been made between ABH and ABC are so nebulous as to be nearly arbitrary. There are no solid criteria (or a solid basis for such criteria) for determining at what level of stromal invasion use of either term would be appropriate. There are no differences in clinical presentation, patient age, and certainly biologic behavior between cases of ABC and ABH 1023. We therefore suggest that the term ABH be discarded altogether. It serves no clinical purpose and only serves to contribute to the nomenclatural incoherence that has plagued this diagnostic entity.

Given the absence of a single case of tumor recurrence, metastasis or tumor-related death in an unequivocally pure ABL, use of the "carcinoma" designation has been questioned, most forcefully by Brainard and Hart 1085. These authors noted that ABL-like lesions showing atypical histologic features (i.e those associated with a separate malignancy or having otherwise atypical areas) should not be considered pure adenoid basal lesions. They proposed the term "adenoid basal epithelioma" (ABE) to describe cases previously designated ABH and histologically pure ABC 10. The term ABE, they argued, "adequately described the composition of epithelial cells and the mixture of basaloid and glandular elements, but it avoids designation as carcinoma" 10. They further argued that morphologically pure lesions are expected to have a clinically benign behavior, thus, applying the term ABE instead of ABC will prevent the potential over-treatment associated with a "carcinoma" diagnosis and is therefore preferable.

Grayson and Cooper 8792, in contrast, have argued for retention of the ABC term for morphologically pure ABT. These authors argue that the aforementioned integration of high-risk HPV types 312, p53-related genetic changes 13, and their potential to be the dominant epithelial component of carcinosarcomas 791 are arguments in support of the neoplastic nature of ABL. In these authors' opinion, the term ABE "underemphasizes the lesion's potential for aggressive behavior" 87. They suggest the term ABE be used only as a synonym for ABH. The ABC term continues to be firmly entrenched in the literature and is used in the most recent classification from the World Health Organization 93

Our position on the nomenclature for ABL is closer to the one proposed by Brainard and Hart 10. Although we agree with Grayson and Cooper 8792 that the presence of high-risk HPV types in an integrated pattern as well as p53 changes argues for the neoplastic nature of ABL, it certainly does not argue for their malignant nature that would justify use of the term carcinoma. Although pure ABL may represent precursor lesions to other malignancies, there is no evidence to suggest that morphologically pure lesions are, in of themselves, malignant. We know of no other situation in human oncology in which the term carcinoma (without an in-situ suffix) is applied for a tumor that has not been shown to recur, metastasize or cause death. In our opinion, there is insufficient evidence at present time to expose patients with morphologically pure ABL to the ominous implications – social, psychological, medical, financial – of a "carcinoma" diagnosis. We concur with use of the designation ABE for morphologically pure ABL. As noted by Brainard and Hart 10, the term ABE acknowledges the potentially neoplastic nature of ABT, is descriptively accurate of their morphology, and most importantly, avoids use of a term that implies inherent malignant potential. However, we further recommend that the term ABE be only used if the entirety of the proliferation is evaluable, e.g in a hysterectomy specimen or in a cervical excision with negative margins. Given the aforementioned difficulties in determining the precise frequency with which ABL are associated with separate malignancies, we suggest use of the simple term "adenoid basal tumor" for lesions that extend to the margins of an excisional biopsy or which broadly involves an incisional biopsy. This may be followed by a comment explaining this description and stating that the possibility of an associated malignancy is probably low but cannot be unequivocally quantified at present. This allows the patient and her physician to make an informed decision between a hysterectomy that eliminates the risk of an unsampled associated malignancy and retention of the uterus for reproductive or other purposes. This approach, in our opinion, strikes an appropriate balance between the need to prevent over-treatment of pure lesions on one hand, and the need to ensure that the lesions are indeed pure on the other. If an ABL is associated with a separate malignancy, then all management options and primary diagnoses should be based on the stage, grade, histologic subtype and other clinicopathologic parameters of the associated malignancy. Using this approach, there is no use for the term ABC, and we agree with Brainard and Hart 10 that this designation should no longer be used.