Epithelial Lacrimal Gland Tumors: Pathologic Classification and Current Understanding

Objective  To apply the updated epithelial salivary gland classification scheme to a large cohort of lacrimal gland tumors so as to provide an updated lacrimal gland tumor classification scheme.

Methods  A retrospective multicenter cohort study of 118 cases of epithelial neoplasia was undertaken. Main outcome measures included pathologic analysis, subtyping, and survival.

Results  Of 118 cases, 17 (14%) were reclassified using the proposed expanded classification scheme based on the current World Health Organization classification of salivary gland tumors. The most frequent neoplasms were pleomorphic adenoma and adenoid cystic carcinoma, of which we highlight more unusual histologic features. Three tumors were found to be unclassifiable with the updated scheme, with 2 having histologically malignant features. Deficiencies and variations in pathologic assessment were noted. Variation in the histologic findings of pleomorphic adenoma and assessment of the extent of invasion of carcinoma ex pleomorphic adenoma were highlighted.

Conclusions  The use of the more histologically diverse classification of salivary gland tumors can be successfully applied to the epithelial lacrimal gland neoplasms. This expanded classification system led to reclassifying 14% of cases. Currently, there are no consistent pathologic standards for processing and evaluating these lesions.

Lacrimal gland lesions represent 5% to 25% of orbital tumors, and the proportion in the literature that are epithelial range from 23% to 70% of biopsied cases.^1-6 Despite our current understanding that rational clinical management of salivary and lacrimal gland tumors depends on specific histologic tumor typing,^3-5,7-13 there has been no official update of the lacrimal gland tumor classification since the World Health Organization (WHO) publication of 1980.¹⁴ The Armed Forces Institute of Pathology (AFIP) monograph on lacrimal gland tumors published in 1994 reviewed 39 cases but did not attempt a comprehensive reclassification.¹⁵ Our understanding of lacrimal gland tumors reflects the histologically similar but more prevalent salivary gland tumors, the classification of which has undergone several iterations since the introduction of the AFIP classification in 1953¹⁶ to include newly recognized tumor types correlating with biological behavior. In 2006 the AFIP monograph on lacrimal gland tumors¹⁷ showed an expanded classification based on the 1992 WHO classification of salivary gland tumors. It is also clear that the salivary gland classification has filtered into the lacrimal gland literature, which has described many tumors analogous to their salivary gland counterparts, including ductal carcinoma,^18-23 acinic cell carcinoma,^24-27 primary squamous cell carcinoma,^28-31 mucoepidermoid carcinoma,^20,32-38 oncocytic carcinoma,³⁹ polymorphous low-grade adenocarcinoma,^5,40 myoepithelial carcinoma,^41-43 lymphoepithelial carcinoma,^44,45 epithelial-myoepithelial carcinoma,⁴² cystadenocarcinoma,⁴⁶ primary sebaceous adenocarcinoma,^5,47-53 basal cell adenocarcinoma,⁵⁴ oncocytoma,^55,56 cystadenoma,⁵⁷ and myoepithelioma.^5,43,58-62

The purposes of our study are to review the standards of histologic analysis and propose an updated classification scheme based on a series of 118 cases from 4 contributing institutions reviewed by a head and neck pathologist (K.W.B.) well versed in salivary gland tumor pathology and classification.

Epithelial lacrimal gland neoplasms from 4 contributing institutions were reviewed for clinical profile, pathology, and outcome at the University of British Columbia, Vancouver, British Columbia, Canada. Only cases with pathology specimens available for examination were included, resulting in a total of 118 cases in the combined series.

All pathology specimens were evaluated by a single pathologist specializing in head and neck pathology (K.W.B.). Specimens were classified according to the most recent WHO salivary gland tumor classification.⁶³ The amount of material available varied, and initial review was based on hematoxylin-eosin–stained sections. Additional material, analysis with histochemical and immunohistochemical stains, and a more complete examination of the extent of the tumor were requested for any case in which the diagnosis was questionable. Initial review was performed with the pathologist blinded to the clinical history and original diagnosis. All of the cases with rare diagnoses, with atypical findings, or in which the review diagnosis differed from the original were submitted to a second review by a committee comprising the head and neck pathologist (K.W.B.) and 2 ophthalmic pathologists (V.A.W. and J.R.).

Linear and logistic regression was used to compare baseline characteristics between different tumor subtypes and histologic findings. Clinical outcomes for adenoid cystic carcinoma were analyzed using survival analysis in the form of the log-rank test. Statistical significance was defined as P < .05.

The pathology specimens of 118 cases from 4 institutions were classified as shown in Table 1. This study was coordinated by the Orbit Clinic, University of British Columbia. Forty cases from the University of British Columbia, 33 from the Department of Ophthalmology, University of Naples, Naples, Italy, 33 from the Department of Ophthalmology, University of Amsterdam, Amsterdam, the Netherlands, and 12 from the Department of Ophthalmology, Kaiser Permanente Medical Center, Harbor City, California, were enrolled. No interinstitution or intra-institution standards for specimen sampling or handling were found. The most frequent diagnosis was pleomorphic adenoma (PA), followed by adenoid cystic carcinoma (ACC) and carcinoma ex pleomorphic adenoma (CEPA). Three tumors were unclassifiable, 2 of which were considered to have histopathologically malignant features and the third of which was indeterminate. Of the 118 cases, 17 were reclassified in this study (Table 2).

According to the AFIP salivary gland tumor classification, the “essential diagnostic feature” of a PA, or benign mixed tumor, is that it is “composed of both epithelial and mesenchymal-like tissues.”⁶³ The epithelial cells form characteristic ductal structures with surrounding myoepithelial cells, which trail out gradually into myxomatous mesenchyme (Figure 1A). All but 2 of the 57 cases of PA in our series easily fit this description and were diagnosed as such in the original pathology reports.

Variable features typical of PA were systematically analyzed, including cellularity, presence of myxoid, chondroid, and hyaline stroma, and cellular characteristics including squamous metaplasia and presence of plasmacytoid cells. Other rare features were occasionally noted.

Typical and infrequent features are summarized in Table 3 and shown in Figure 1. The mean diameter was 24.5 mm, ranging from 10 to 40 mm. Mitotic figures were rarely detected. Tumor capsules varied from a few micrometers to a few hundred micrometers in thickness and were incomplete in some. In 31 cases in which we had adequate sections to analyze the full extent of the tumor, 27 were found to border or invade the capsule and 4 (including 3 not originally reported) extended beyond.

The review diagnosis differed from the original diagnosis in 2 cases. One case, originally diagnosed as noninvasive carcinoma in PA, was on review considered a PA with epithelial atypia and oncocytic metaplasia not warranting a diagnosis of carcinoma. The second case, originally diagnosed as acinic cell carcinoma, was considered a variant PA with atypical features.

The 3 clinically recurrent tumors were characterized histologically by multifocal and multinodular growth patterns. The recurrences occurred at 23, 25, and 30 years after the initial resection. In all 3 cases, the tumor formed numerous nodules generally composed predominantly of stroma with minor epithelial components. One tumor showed focal calcification and another showed focal ossification. There was no evidence of malignant transformation.

A single case of myoepithelioma was identified.⁶⁴ It was well circumscribed and composed of sheets of myoepithelial cells with rare ducts. Most of the cells were hyaline or plasmacytoid myoepithelial cells, but epithelioid and spindle cells were also present in small numbers. There was no specialized stroma present and mitotic figures were undetectable. Myoepithelial prominence not reaching the 90% to 95% required by this definition was found in 5 cases of PA in our series. A final tumor originally diagnosed as a Warthin tumor was on review considered an oncocytoma.

Our series included 38 cases of ACC (32%), which was the most common malignant neoplasm. The mean histologic diameter was 27.7 mm, with the diameter ranging from 16 to 42 mm and no statistically significant difference when compared with PA (P = .09). Most tumors had combinations of 2 or 3 histologic patterns (Table 4). Comedonecrosis was present in 10 cases, in both cribriform and solid patterns, but was most prominent in areas of the solid pattern.

Mitotic activity ranged from undetectable to 25 mitotic figures per 10 high-power fields. In general, the lowest activity was found in tumors with tubular and cribriform patterns (median, 2.9 mitotic figures per 10 high-power fields), while tumors with at least some component of solid pattern had higher mitotic activity (median, 10.1 mitotic figures per 10 high-power fields) (P < .001).

The ACCs were nonencapsulated with infiltration of surrounding fat, periorbita, and muscle. Invasion into the sclera or optic nerve was not identified in any case. Perineural invasion was seen in 13 cases, although caution in interpreting this is necessary because of the variable sectioning patterns by the different institutions. Perineural invasion did not correlate with mitotic activity (P = .90), particular histologic pattern noted on analysis (cribriform, P = .50; tubular, P > .99; solid, P = .49), or predominant pattern (cribriform, P = .38; tubular, P = .51; solid, P > .99). Perineural invasion was described in 9 original pathology reports, including 2 cases in which it was not found on review. Inconsistent sampling, orientation, and pathologic preparation of bone samples were noted. Of the 11 cases in which bone was sampled, 7 were positive and 4 were negative for tumor. The absence of a cribriform pattern (P = .01) and the presence of a solid pattern (P = .02) were significantly associated with death.

In 2 cases, an area of high-grade carcinoma was found within the ACC: a poorly differentiated adenocarcinoma in one and undifferentiated carcinoma in the other (Figure 2). The undifferentiated carcinoma was not mentioned in the original pathology report. The ACC with adenocarcinoma had originally been diagnosed as epidermoid carcinoma. The adenocarcinoma was too poorly preserved to determine mitotic activity, while the mitotic rate in the undifferentiated carcinoma (6 mitotic figures per 10 high-power fields) was clearly higher than that in the surrounding ACC (2 mitotic figures per 10 high-power fields). A third case had originally been diagnosed with ACC with undifferentiated carcinoma but was found on review to be a solid-pattern ACC.

The second most common malignant tumor in our series and the literature is CEPA, or malignant mixed tumor, of which we had 9 cases. The malignant components included 7 adenocarcinomas, 1 mucoepidermoid carcinoma, and 1 biphasic, carcinosarcomatous tumor. In most, the carcinomas were poorly differentiated, with the exception of 1 well-differentiated papillary adenocarcinoma and the carcinosarcoma that had distinct, well-differentiated adenocarcinomatous and spindle cell sarcomatous components (Figure 3). The mucoepidermoid carcinoma showed high-grade cytologic atypia. The adenocarcinomas displayed high mitotic activity, with a mean of 8 mitotic figures per 10 high-power fields and a range from less than 1 to 26 mitotic figures per 10 high-power fields. The sarcomatous component of the biphasic tumor had a mitotic rate of 6 mitotic figures per 10 high-power fields, and the adenocarcinomatous component had a mitotic rate of 30 mitotic figures per 10 high-power fields.

The extent of the malignant component and its relationship to the PA capsule varied between cases. Tumors could be classified into 3 main groups: those confined within PA (noninvasive; 3 cases) (Figure 4), those that invaded less than 1.5 mm from the capsule (minimally invasive; 1 case) (Figure 5), and those that had spread more than 1.5 mm beyond the capsule into surrounding tissue (invasive; 5 cases) (Figure 6). Two adenocarcinomas arose as minor components (approximately 30% and <50%) completely surrounded by typical PA with intact capsules. One of these was mainly a cellular PA with a prominent myoepithelial component, and the other had both cellular and stromal areas. We called the malignant transformation in these cases noninvasive as it was completely contained within the surrounding PA. In a third case, the adenocarcinoma composed more than half of the tumor but remained within the preexisting PA, and the malignant component appeared to be replacing the ductal epithelium of the adenoma.

In a fourth case, considered minimally invasive, the malignant component invaded through the capsule of a mixed cellular and stromal PA but not more than 0.1 mm beyond. In a fifth case, the PA comprised a small nodule of intermediate cellularity with a prominent myoepithelial component in a densely hyalinized stroma, completely surrounded by an invasive adenocarcinoma extending up to 18 mm beyond its border. The sixth case was similar except that the remnants of the cellular components of the PA were almost unrecognizable in the central hyaline nodule of the high-grade mucoepidermoid carcinoma.

Three additional tumors diagnosed as CEPA had no recognizable cellular elements of a PA but shared the common feature of a circumscribed hyalinized nodule within a malignant tumor. One nodule had a scattering of cells of benign appearance within it, and the other 2 had no significant cellular elements at all. The original pathology report of one of these gave a diagnosis of CEPA. Given the continuum of changes described for the first 6 cases, we diagnosed these last 3 cases as likely CEPA as well. Other tumors, such as 2 examples of ACC and 1 of mucoepidermoid carcinoma, had some hyalinization within the stroma but lacked a circumscribed nodule and therefore were not considered to have arisen in PA.

We had 3 cases of primary adenocarcinoma. One had sheets of very high-grade anaplastic cells with prominent gland formation, extensive necrosis, and a very high mitotic rate (85 mitotic figures per 10 high-power fields). A second had lower-grade pleomorphism, highly differentiated gland formation throughout, and a very low mitotic rate of less than 1 mitotic figure per 10 high-power fields. The third had variable gland formation combined with nests of clear cells and areas resembling solid-pattern ACC but had nuclei more pleomorphic than is typical of ACC. The mitotic rate of this tumor was 6 mitotic figures per 10 high-power fields, and apoptosis was prominent. This third tumor had originally been diagnosed as ACC, whereas the first 2 were originally diagnosed as adenocarcinomas.

We had 2 cases of mucoepidermoid carcinoma (Figure 7). One was highly cellular with a large proportion of intermediate cells (approximately 50%), poorly formed glandular elements with intracellular mucin, and a minor epidermoid component. The other tumor was composed of islands of cystic mucin-filled glands and epidermoid nests with few intermediate cells embedded in a dense desmoplastic reaction. Both were grade 3 (high-grade) mucoepidermoid carcinomas.⁶⁵

Ductal carcinoma, defined by its resemblance to mammary ductal carcinoma, has only recently been reported in the lacrimal gland, and to our knowledge 1 of our 2 cases was the first reported in the lacrimal gland.¹⁸ In this case, the tumor was arranged in lobules composed largely of solid epithelial nests, some of which showed central necrosis resembling comedonecrosis of mammary ductal carcinoma in situ. Mitotic figures were common, with a rate of 9 mitotic figures per 10 high-power fields. The second case had similar lobules but comedonecrosis was more prominent, and the epithelium had a micropapillary architecture (Figure 8). The mitotic rate was less than 1 mitotic figure per 10 high-power fields. The original pathology report diagnosed this as high-grade papillary adenocarcinoma.

We had 1 case each of myoepithelial and squamous cell carcinoma. The myoepithelial carcinoma had nests of myoepithelial cells with nuclear atypia, a mitotic rate of 1 mitotic figure per 10 high-power fields, and no gland formation (Figure 9). Immunohistochemical staining results were positive for keratin, irregularly positive for S-100 protein, and weakly positive for actin. The review diagnosis differed from the original diagnosis, which was “malignant epithelial neoplasm, most consistent with ACC.” The single case of squamous cell carcinoma had areas of moderately well-differentiated epithelial cells with occasional keratin pearls, and other foci of more poorly differentiated carcinoma, embedded in a desmoplastic stroma. No goblet cells or mucin production were seen. There was a high mitotic rate of 15 mitotic figures per 10 high-power fields. The tumor was infiltrative with no capsule. The review diagnosis concurred with that of the original.

There were 2 cases of carcinoma that did not clearly match the features of any entity in the recent lacrimal and salivary gland tumor classification schemes. One was a biphasic tumor with a hypercellular area containing cells with pleomorphic nuclei and a high mitotic rate of 12 mitotic figures per 10 high-power fields forming multiple squamous eddies as well as a hypocellular area with cells arranged in nests within a hyaline stroma. Neither area had features to support the original diagnosis of ACC. The second was a tumor with sheets of eosinophilic, epithelioid cells with pleomorphic nuclei and a high mitotic rate of 6 mitotic figures per 10 high-power fields. There was no gland formation and no intracellular or extracellular mucin detected by histochemical analysis. A recurrence of the tumor excised 2 years later had sheets of cells with a more anaplastic appearance. The original diagnosis of mucoepidermoid carcinoma was not supported by our review.

The WHO's classification scheme for lacrimal gland epithelial neoplasia¹⁴ has provided the context for our understanding of them and included 2 broad categories of “other adenomas” and “other carcinomas.” An updated pathologic classification is needed to include the recognition of the advances in salivary gland histologic classification and the identification of several new entities. In addition, the behavior of various benign and malignant lacrimal gland tumors has been further elucidated.^10,28,66 Since the expanded salivary gland classifications were published,^67-69 the use of the salivary gland classification for lacrimal gland tumors has been reported.^5,18-62 In 2006 the AFIP monograph on lacrimal gland tumors¹⁷ showed an expanded classification based on the 1991 WHO classification of salivary gland tumors. Our study reviewed 118 cases from 4 institutions, successfully applying the updated salivary gland classification scheme to lacrimal gland tumors and resulting in a revision of the diagnosis in 14% of cases.

Table 5 lists our proposed histologic classification for lacrimal gland epithelial neoplasia based on our series and recently published series (Table 6) and case reports.^{5,12,13,15,20,24-26,28,44-46,54,55,70} The most significant aspect of this listing is the increased diversity of diagnoses and the inclusion of subtypes within the diagnostic categories that may influence clinical prognosis. Furthermore, review of the pathologic specimens revealed no consistent sampling, orientation, or pathologic preparation of samples within and between the various institutions. In response to these findings, Rootman and White⁷¹ have given recommendations on sampling, orientation, and preparation of pathologic specimens of lacrimal gland tumors.

The 2 major factors determining the prognosis of PA are its likelihood of recurrence and evidence of malignant transformation. Recurrence rates as high as 30% have been reported in the older literature and literature from mainland China, increasing with the length of follow-up.^5,8,72 In contrast, the chance of recurrence is negligible, if not eliminated, by proper initial surgical management including en bloc excision.^{6,8,10,12,70,73-75} With a mean (SD) follow-up of 4.47 (4.58) years, no recurrences occurred in our series for PA handled in this manner.

The likelihood of malignant transformation of PA appears to be related to tumor age,^8,76 with reports of transformation in recurrent tumors 40 years or longer after initial excision.^70,74,77 The salivary gland literature reports up to a 9% incidence of transformation over 15 years.⁷⁶ In our series, none of the 3 recurrent tumors had transformed, but patients presenting with CEPA were significantly older than those with PA (median age difference, 20 years). Tumor hyalinization has been shown to correlate with malignant transformation in the salivary gland.⁷⁸ Extensive hyalinization has been reported previously for lacrimal gland CEPA^50,74 and was a prominent feature in 5 of the 9 cases of CEPA in our series. If hyalinization is taken as a sign of tumor age, then this histologic marker supports the association between malignant transformation and patient age.

Our series included 5 cases of PA with cellular atypia; all were predominantly cellular tumors and 3 had prominent myoepithelial components, 1 of which had originally been diagnosed as CEPA. High cellularity or myoepithelial prominence was a feature in 3 of the 9 cases of CEPA in our series. It is unknown whether high cellularity, atypia, and myoepithelial prominence may increase the likelihood of eventual malignant transformation, and we thus recommend retaining these descriptions in examining PA.

The extent of invasion of the malignant component of CEPA in the salivary gland has been shown to be strongly correlated with subsequent behavior. A number of studies of salivary gland neoplasms have demonstrated that tumors in which the malignant component is confined within the capsule or shows minimal invasion (variably defined as extension 8 mm,⁷⁹ 5 mm,⁸⁰ or 1.5 mm⁸¹ beyond the PA capsule) are associated with a benign course. There is a single report of cervical lymph node metastases from a parotid gland CEPA that was apparently well sampled and showed no capsular invasion.⁸² Despite this isolated case, we feel that there is sufficient evidence to support the contention that the extent of invasion is an important prognostic finding. We suggest adopting the WHO terminology of noninvasive, minimally invasive, and invasive to recognize these differences in outcome. In our series, 3 patients with noninvasive carcinoma and 1 patient with minimally invasive carcinoma (extension <1.5 mm beyond the PA capsule) had no recurrence, whereas only 1 of 4 patients with invasive carcinoma remained disease-free in follow-up.

We required frank cytologic and/or architectural evidence of malignancy to render a diagnosis of noninvasive CEPA. Cases that exhibited nuclear enlargement and pleomorphism without evidence of significantly increased mitotic activity or necrosis were considered atypical PAs. We adopted a conservative approach to the diagnosis of noninvasive CEPA to prevent overtreatment as noninvasive CEPA has a prognosis similar to that of PA, with possible rare exceptions.⁸²

Several studies have shown that solid-pattern ACC portends a worse prognosis than the cribriform or tubular pattern for both salivary^83-86 and lacrimal^87-89 gland tumors. Our series shows a similar tendency as patients with tumors with a solid component had a lower survival rate. The absence of a cribriform pattern was also associated with poor survival. We recommend retaining histologic pattern descriptions within the ACC category.

The phenomenon of dedifferentiation in ACC is a recently described rare event in tumors of minor and major salivary glands.^90-94 Most commonly, the malignant neoplasm arising within ACC is a high-grade adenocarcinoma. A small number of undifferentiated carcinomas have been described. In 1 case, the malignant component was a sarcomatoid neoplasm with focal myoepithelial features.⁹⁰ In 2 of 3 dedifferentiated ACCs, molecular analysis demonstrated mutation of the p53 gene in only the dedifferentiated component, suggesting a role for this oncogene in the pathogenesis.^91,93 In most of the reported cases, the finding of dedifferentiation has been associated with poor outcome.^90,93,94

Two cases of dedifferentiated ACC, to our knowledge the only known cases of dedifferentiated lacrimal gland ACC, were seen in our series. In one case, the patient has remained disease-free for more than 10 years of follow-up; in the other, the patient died of lung metastases 7.3 years after exenteration. Our classification includes this subtype of ACC as its prognostic significance may become evident as additional cases are reported.

Myoepithelioma. Lacrimal gland myoepithelioma has been reported rarely,^5,43,58-62 but it is well described in the salivary gland literature and is widely held to represent one end of the spectrum of benign tumors showing epithelial and myoepithelial differentiation. Although some studies suggest it may be aggressive,^1,95 others have indicated an excellent outcome.^96,97 We include it as a separate entity because of its unusual microscopic appearance that may make diagnosis difficult.

Oncocytoma. The few cases of oncocytoma reported in the lacrimal gland literature, including 1 case in our series, have shown no evidence of recurrence throughout follow-up.^55,56 Our case was initially reported as a Warthin tumor; thus, with this review, Warthin tumor in the lacrimal gland remains an unreported entity.⁹⁸ Although the tumor had a prominent lymphocytic infiltrate suggestive of a Warthin tumor, the oncocytes themselves were not found in the highly ordered bilayered arrangement typical of Warthin tumors.⁶⁸

Adenocarcinoma. Adenocarcinoma was the third most common lacrimal gland malignant neoplasm, accounting for 5% of carcinomas in our series and 4% to 13% in other recent series.^{5,12,15,20,28,55} Generally, lacrimal gland adenocarcinoma carries a dismal prognosis (fatality ranging from 50%-80%)^8,12,28 and an average survival of 1.5 years.¹² It is therefore important to distinguish adenocarcinoma from low-grade polymorphous adenocarcinoma or mucoepidermoid carcinoma (grades 1 and 2), both of which carry a better prognosis.

Mucoepidermoid Carcinoma. The prognosis of mucoepidermoid carcinoma varies with tumor grade, with almost no recurrence for grades 1 and 2 tumors and greater than 50% mortality from grade 2 tumors.³⁵

Ductal Carcinoma. This series includes the first reported case of lacrimal ductal carcinoma to our knowledge¹⁸ and a second, originally unrecognized case diagnosed as adenocarcinoma. Both of these cases were disease-free after excision and radiotherapy. Of the 5 other cases reported in the literature, 3 fared well after a single operation and adjunct radiation.^18-23 A fourth patient has had local recurrences with surgery alone,¹⁹ and the fifth had lymphatic metastasis detected prior to scheduled radiotherapy.²³

Myoepithelial Carcinoma. Our case of myoepithelial carcinoma was initially misdiagnosed and is one of few reported cases.^41,43 With so few cases, it is impossible to draw conclusions regarding clinical behavior. Our patient died within months of diagnosis without definitive treatment. In the salivary glands, there is a reported recurrence rate greater than 50%.^99,100

Squamous Cell Carcinoma. Including our single case, there are 8 cases of squamous cell carcinoma of the lacrimal gland reported in the literature.^28-31 No clinical history is available for our case, and no recurrences were noted in the other published reports.

Carcinosarcoma. Carcinosarcoma, a biphasic tumor composed of both carcinoma and sarcoma, is a rare tumor of the salivary gland. Approximately 33 cases have been reported in the English-language literature, both as a primary malignant neoplasm and as CEPA.⁶⁹ A single case is listed without description in a large lacrimal gland tumor series.⁵ No follow-up data were available for our case of carcinosarcoma that arose in a PA and had spindle cell sarcoma and adenocarcinomatous components.

Malignant Tumor Types Not Seen in Our Series. Isolated cases of several other tumor types have been reported in the lacrimal gland, including polymorphous low-grade adenocarcinoma,^5,40 acinic cell carcinoma,^24-27 basal cell adenocarcinoma,⁵⁴ epithelial-myoepithelial carcinoma,⁴² and cystadenocarcinoma,⁴⁶ all of which appear to be low grade, as well as primary sebaceous adenocarcinoma^5,47-53 and lymphoepithelial carcinoma,^44,45 both of which are considered high grade. Since completing this review, we have seen a single case each of epithelial-myoepithelial carcinoma, acinic cell carcinoma, and myoepithelial carcinoma in the Orbit Clinic, University of British Columbia.

This study and review of the literature evokes 3 important conclusions. First, the use of a more diverse classification can be successfully applied to epithelial lacrimal gland neoplasms. Second, the expanded classification system led to reclassifying 14% of cases in our review. Finally, there are currently no pathologic standards for processing and evaluating these lesions.

Correspondence: Jack Rootman, MD, 2550 Willow St, Vancouver, BC V5Z 3N9, Canada (jrootman@interchange.ubc.ca).

Submitted for Publication: January 30, 2009; final revision received April 2, 2009; accepted April 6, 2009.

Financial Disclosure: None reported.