Graefe’s Arch Clin Exp Ophthalmol (2003) 241:907–913
C L I N I C A L I N V E S T I G AT I O N
DOI 10.1007/s00417-003-0748-z
Kaan Gündüz Carol L. Shields Ilhan Günalp Jerry A. Shields
Received: 5 February 2003 Revised: 25 June 2003 Accepted: 26 June 2003 Published online: 25 October 2003 © Springer-Verlag 2003
Presented in part as a free paper at the 10th International Congress of Ocular Oncology, Amsterdam, the Netherlands, 17–21 June 2001 K. Gündüz (✉) · I. Günalp Oncology Service, Department of Ophthalmology, Ankara University, Faculty of Medicine, Ankara, Turkey e-mail:
[email protected] Fax: +90-312-2291812 K. Gündüz G.M.K. Bulvari 116/3, 06570 Maltepe, Turkey C. L. Shields · J. A. Shields Oncology Service, Wills Eye Hospital, Thomas Jefferson University, Philadelphia, Pennsylvania, USA
Magnetic resonance imaging of unilateral lacrimal gland lesions
Abstract Purpose: To report the findings on magnetic resonance imaging (MRI) of various benign and malignant unilateral lacrimal gland lesions. Methods: This is a retrospective noncomparative interventional case series. Thirty-one consecutive patients with a unilateral lacrimal gland lesion were analyzed. The preoperative MRI findings were correlated with the pathology results. The main outcome measures were anatomic extent, configuration, margins, angulation, internal features on T1- and T2-weighted images (with respect to extraocular muscles and cerebral gray matter), contrast enhancement of the lesion and adjacent bone change on MRI. Results: Of the 31 patients, 21 had chronic dacryoadenitis, 3 had lymphoid tumors, and 7 had epithelial tumors including pleomorphic adenoma (3), adenoid cystic carcinoma (3), and pleomorphic adenocarcinoma (1). The results of the patients with chronic dacryoadenitis demonstrated involvement of the orbital lobe alone in 13 patients (62%), involvement of both orbital and palpebral lobes in 8 (38%), a molded configuration with ill-defined margins, sharp angles in 13 (62%), round angles in 8 (38%), lack of bone change, an isointense internal signal on T1-weighted images, a hypointense signal on T2weighted images, and moderate contrast enhancement. The patients with lymphoid tumors demonstrated in-
volvement of the orbital lobe, a molded configuration with ill-defined margins and sharp angles, lack of bone change, an isointense internal signal on T1-weighted images, an isointense signal on T2-weighted images, and moderate contrast enhancement. Those with epithelial tumors showed involvement of the orbital lobe, a well-circumscribed oval configuration, and round angles. Pleomorphic adenoma demonstrated smooth margins, bone expansion in two patients , and no bone change in one. Adenoid cystic carcinoma and pleomorphic adenocarcinoma showed irregular margins and bone destruction. All epithelial tumors demonstrated an isointense internal signal on T1-weighted images, a hyperintense signal on T2-weighted images, and moderate contrast enhancement. Conclusions: It is difficult to uniformly correlate the MRI features and histopathologic findings in lacrimal gland lesions. However, MRI seems to be useful in determining the etiology of a unilateral lacrimal gland lesion. Internal tissue features on T1- and T2-weighted images of MRI are most helpful in categorizing these lesions. Although the number of patients is small, our findings suggest that there are differences in orbital MRI findings of inflammatory lesions and lymphoid tumors as compared to benign and malignant epithelial tumors in the lacrimal gland fossa.
908
Table 1 Magnetic resonance imaging of unilateral lacrimal gland lesions: anatomic extent, configuration, margins and angulation in the lesion Pathologic diagnosis
Anatomic extent Orbital lobe
Chronic dacryoadenitis (n=21) Lymphoid tumors (n=3) Pleomorphic adenoma (n=3) Adenoid cystic carcinoma (n=3) Pleomorphic adenocarcinoma (n=1)
13 (62%) 3 (100%) 3 (100%) 3 (100%) 1 (100%)
Configuration
Orbital Molded and palpebral lobes
Oval
8 (38%) – – – –
– – 3 (100%) 3 (100%) 1 (100%)
21 (100%) 3 (100%) – – –
Margins
Angulation in the lesion
Ill-defined Well-circumscribed
Sharp
Round
13 (62%) 3 (100%) – – –
8 (38%) – 3 (100%) 3 (100%) 1 (100%)
21 (100%) 3 (100%) – – –
Smooth contour
Irregular contour
– – 3 (100%) – –
– – – 3 (100%) 1 (100%)
Introduction
Materials and methods
Lacrimal gland tumors are classified as being either epithelial or non-epithelial. In 1956, Reese reported that approximately 50% of lacrimal gland lesions are of epithelial origin and 50% are of non-epithelial origin [11]. In a later report, Shields and associates found that nonepithelial lesions accounted for 78% of lacrimal gland masses that came to biopsy [13]. Inflammatory (dacryoadenitis) and lymphoid lesions were the most common non-epithelial tumors, accounting for 64% and 14% of all lacrimal gland tumors respectively [13]. Of the epithelial tumors, 12% were benign pleomorphic adenomas, 6% were dacryops and 4% were malignant tumors, including adenoid cystic carcinoma and pleomorphic adenocarcinoma [13]. A number of rare malignant tumors, including monomorphic adenocarcinoma and mucoepidermoid carcinoma, have also been reported to occur in the lacrimal gland area [14]. Patient history and clinical features are important in differentiating lacrimal gland lesions from each other [18]. However, since there may be an overlap in the clinical findings produced by lacrimal gland tumors, orbital imaging is necessary to assist in their differentiation. Computed tomography (CT) has been routinely used in the evaluation of lacrimal gland lesions because it is more widely available and less expensive than magnetic resonance imaging (MRI). One advantage of CT over MRI is that it is not affected by the presence of ferromagnetic material in the body. Other advantages of CT include detection of calcification and detailed evaluation of bone. On the other hand, the advantages of MRI compared to CT include better soft tissue contrast and lack of ionizing radiation. The use of a surface coil and fat suppression techniques have further improved the quality of orbital MRI [2]. Based on the superior soft tissue characteristics provided by MRI compared to CT, we aimed to analyze the MRI findings in unilateral lacrimal gland lesions and correlate the MRI findings with the pathologic diagnosis.
This study is a retrospective noncomparative interventional case series. We analyzed the preoperative orbital MRI findings of 31 consecutive patients with a unilateral lacrimal gland lesion who underwent an orbital biopsy (K.G.) at our center in Ankara between October 1998 and January 2003, and correlated the MRI findings with the histopathologic diagnosis. The MRI findings were evaluated with respect to seven parameters: anatomic extent (orbital lobe, palpebral lobe, or both), configuration (molded, round, or oval), margins (ill-defined, well-circumscribed regular, or well-circumscribed irregular), angulation in the mass (sharp or round), internal features (on T1- and T2-weighted images with respect to extraocular muscle and cerebral gray matter), contrast enhancement after gadolinium-DTPA (diethylenetriamine penta-acetic acid) injection (mild, moderate, extreme), and adjacent bone change (no change, bone expansion or bone destruction). Although data on demographics and clinical findings was collected, a detailed description of the demographics and clinical findings of the patients is beyond the scope of this paper which concentrates on the correlation of MRI findings with the histopathologic diagnosis. All the patients underwent a superolateral skin crease incision for orbitotomy using an extraperiosteal approach, and an incisional or excisional biopsy of the lacrimal gland lesion was performed depending on the clinical findings. An informed consent was obtained from all patients.
Results Results of pathologic examinations revealed that 21 of the 31 patients had chronic dacryoadenitis, 3 had lymphoid tumors (lymphoid hyperplasia), and the remaining 7 had epithelial tumors, including pleomorphic adenoma (3), adenoid cystic carcinoma (3), and pleomorphic adenocarcinoma (1). The MRI features of these 31 lesions are listed in Tables 1 and 2. The results of the patients with chronic dacryoadenitis showed involvement of the orbital lobe alone (Fig. 1) in 13 patients (62%), and involvement of both orbital and palpebral lobes (Fig. 2) in 8 (38%). Chronic dacryoadenitis displayed a molded configuration with ill-defined margins in 21 patients (100%), sharp angles (Fig. 1) in 13 (62%), and round angles in 8 (38%). Round angles were noted in cases where there was extensive orbital involvement by the inflammatory tissue extending from
909
Table 2 Magnetic resonance imaging of unilateral lacrimal gland lesions: internal features, contrast enhancement, and adjacent bone change Pathologic diagnosis
Chronic dacryoadenitis (n=21) Lymphoid tumors (n=3) Pleomorphic adenoma (n=3) Adenoid cystic carcinoma (n=3) Pleomorphic adenocarcinoma (n=1)
Internal MRI features T1-weighted T2-weighted image* image*
Contrast Adjacent bone change enhancement No bone Bone Bone change expansion destruction
Isointense Isointense Isointense Isointense Isointense
Moderate Moderate Moderate Moderate Moderate
Hypointense, homogeneous Isointense, heterogeneous Hyperintense, heterogeneous Hyperintense, homogeneous Hyperintense, heterogeneous
21 (100%) 3 (100%) 1 (33%) – –
– – 2 (67%) – –
– – – 3 (100%) 1 (100%)
*The intensities on T1- and T2-weighted images are with respect to the extraocular muscle and cerebral gray matter
Fig. 1a, b Chronic dacryoadenitis. T1- and T2-weighted axial MR images showing chronic dacryoadenitis involving the orbital lobe of the lacrimal gland. This tumor displays a molded configuration conforming to the shape of the globe and orbital bone with ill-defined margins and sharp angles. There is no adjacent bone change. a T1-weighted axial image showing chronic dacryoadenitis isointense with respect to the extraocular muscle and cerebral gray matter (arrow). b T2-weighted axial image demonstrating chronic dacryoadenitis hypointense with respect to the extraocular muscle and cerebral gray matter (arrow)
the lacrimal gland. Chronic dacryoadenitis demonstrated an isointense internal signal on T1-weighted images (Fig. 1a) in 21 patients (100%), a homogeneous hypointense signal on T2-weighted images (Fig. 1b) in 21 (100%), and moderate contrast enhancement in 21 (100%). There was no adjacent bone change (Fig. 1a-b) in any of the 21 patients (100%). The results of the patients with lymphoid tumors showed involvement of the orbital lobe (Fig. 3a-b) in 3 patients (100%), a molded configuration with ill-defined margins in 3 (100%), and sharp angles (Fig. 3a-b) in 3 (100%). Lymphoid tumors demonstrated an isointense internal signal on T1-weighted images (Fig. 3a) in 3 patients (100%), a heterogeneous isointense signal on T2-
Fig. 2 Chronic dacryoadenitis. T1-weighted sagittal MR image with gadolinium contrast showing chronic dacryoadenitis with involvement of both the orbital and palpebral lobes of the lacrimal gland (arrows)
910
Fig. 3a, b Lymphoid tumor. T1- and T2-weighted axial MR images showing lymphoid tumor involving the orbital lobe of the lacrimal gland. This tumor demonstrates a molded configuration conforming to the shape of the globe and orbital bone with ill-defined margins and sharp angles. There is no adjacent bone change. a T1weighted axial image showing lymphoid tumor isointense with respect to the extraocular muscle and cerebral gray matter (arrow). b T2-weighted axial image demonstrating lymphoid tumor heterogeneously isointense with respect to the extraocular muscle and cerebral gray matter (arrow)
Fig. 4a, b Pleomorphic adenoma. T1- and T2-weighted axial MR images showing pleomorphic adenoma occurring in the orbital lobe of the lacrimal gland. This tumor displays an oval appearance, regular margins, round angles, and adjacent bone expansion. a T1-weighted axial image showing pleomorphic adenoma isointense with respect to the extraocular muscle and cerebral gray matter (arrow) with adjacent bone expansion (arrowhead). b T2weighted axial image demonstrating pleomorphic adenoma heterogeneously hyperintense to the extraocular muscle and cerebral gray matter (arrow) with adjacent bone expansion (arrowhead)
weighted images (Fig. 3b) in 3 (100%), and moderate contrast enhancement in 3 (100%). There was no adjacent bone change (Fig. 3) in any of the 3 patients (100%). The results of the patients with pleomorphic adenoma showed involvement of the orbital lobe (Fig. 4) in 3 patients (100%), and an oval configuration with smooth, regular margins and round angles (Fig. 4) in 3 (100%). The tumor displayed an isointense internal signal on T1weighted images (Fig. 4a) in 3 (100%), a heterogeneous hyperintense signal on T2-weighted images (Fig. 4b) in
3 (100%), and moderate contrast enhancement in 3 (100%). There was bone expansion in 2 patients (67%) (Fig. 4), and no adjacent bone change in 1 patient (33%). The results of the patients with adenoid cystic carcinoma showed involvement of the orbital lobe (Fig. 5a-b) in 3patients (100%), and an oval configuration with irregular margins and round angles (Fig. 5a-b) in 3 (100%). The tumor demonstrated an isointense internal signal on T1-weighted images (Fig. 5a) in 3 (100%), a homogeneous hyperintense signal on T2-weighted imag-
911
▲
es (Fig. 5b) in 3 (100%), and moderate contrast enhancement in 3 (100%). There was irregular adjacent bone destruction (Fig. 5) in all 3 patients (100%). The patient with pleomorphic adenocarcinoma showed involvement of the orbital lobe (Fig. 6), and an oval configuration with irregular margins and round angles (Fig. 6). The tumor demonstrated an isointense inFig. 5a, b Adenoid cystic carcinoma. T1- and T2-weighted axial MR images showing adenoid cystic carcinoma affecting the orbital lobe of the lacrimal gland. This tumor displays an oval appearance, irregular margins with bone destruction, and round angles. a T1-weighted axial image showing adenoid cystic carcinoma isointense with respect to the extraocular muscle and cerebral gray matter (arrow). There is irregular bone destruction (arrowhead). b T2-weighted axial image demonstrating adenoid cystic carcinoma homogeneously hyperintense to the extraocular muscle and cerebral gray matter (arrow) Fig. 6a–c Pleomorphic adenocarcinoma. T1- and T2-weighted axial MR images showing pleomorphic adenocarcinoma involving the orbital lobe of the lacrimal gland. This tumor displays an oval appearance, irregular margins, and round angles. There is irregular bone destruction. a T1-weighted axial image showing the pleomorphic adenocarcinoma isointense with respect to the extraocular muscle and cerebral gray matter (arrow). b T1-weighted sagittal image demonstrating pleomorphic adenocarcinoma with irregular bone destruction (arrow). c T2-weighted coronal image showing pleomorphic adenocarcinoma heterogeneously hyperintense to the extraocular muscle and cerebral gray matter (arrow). The tumor contains hyperintense cystic areas (arrowhead)
912
ternal signal on T1-weighted images (Fig. 6a), a heterogeneous hyperintense signal on T2-weighted images (Fig. 6c), and moderate contrast enhancement. There was irregular adjacent bone destruction (Fig. 6b).
Discussion We analyzed the orbital MRI findings of unilateral lacrimal gland lesions with respect to seven parameters: anatomic extent (involved lobe), configuration, margins, angulation in the mass, internal features, contrast enhancement, and adjacent bone change discussed below. Anatomic extent, configuration, margins and angulation In our study, palpebral lobe involvement was seen only in chronic dacryoadenitis. Lymphoid tumors can also affect the palpebral lobe of the lacrimal gland [8]. Epithelial tumors, on the other hand, generally involve only the orbital lobe of the lacrimal gland [12, 14]. In rare cases, the palpebral lobe of the lacrimal gland can also be affected by an epithelial or metastatic tumor [9, 16, 17]. Non-epithelial tumors (chronic dacryoadenitis and lymphoid tumors) had a molded configuration conforming to the shape of the globe and orbital bone with ill-defined margins and usually sharp angles. The sharp angulation feature of non-epithelial lesions is probably due to the spillover of lacrimal gland inflammation or infiltration into adjacent fat limited by orbital tissue planes. However, if the inflammation or infiltration is severe, it may extend beyond the limiting tissue planes, leading to more extensive orbital involvement and round angulation. Occasionally, a lymphoid tumor may present with a well-circumscribed oval configuration on axial scans simulating an epithelial tumor [5, 15]. In contrast to chronic dacryoadenitis and lymphoid tumors, benign and malignant epithelial tumors of the lacrimal gland had an oval appearance with well-circumscribed margins and round angles. While pleomorphic adenoma had regular margins, adenoid cystic carcinoma and pleomorphic adenocarcinoma had irregular margins. Mafee and associates, in their review article, pointed out that irregular margins displaying nodularity and infiltration into adjacent orbital fat are highly suggestive of a malignant tumor [8]. However, in rare cases a malignant tumor can have regular margins [8]. Jakobiec and coworkers reported that all nine malignant epithelial tumors in their series demonstrated irregular margins with nodularity and infiltration into the adjacent orbital fat using CT [5]. MRI is similar to CT in that it provides information on anatomic extent, configuration, margins, and angulation features of a lacrimal gland fossa mass [5]. In this regard, MRI adds little to our knowledge but is still useful in evaluating a patient with a lacrimal gland lesion.
Internal tissue features on T1- and T2-weighted images All non-epithelial lesions and epithelial tumors were isointense with respect to the extraocular muscle and cerebral gray matter on T1-weighted images. While chronic dacryoadenitis was hypointense on T2-weighted images, lymphoid tumors were isointense, and epithelial tumors (pleomorphic adenoma, pleomorphic adenocarcinoma and adenoid cystic carcinoma) were hyperintense on T2weighted images. The low signal intensity of chronic dacryoadenitis on T2-weighted images is probably due to tissue fibrosis [1]. Lymphoid tumors are isointense or slightly hyperintense on T2-weighted images due to the cellular density as well as to the water content in these lesions [10]. The cellular epithelial tumors show greater signal intensities on T2-weighted images because of their higher water content [10]. Pleomorphic adenoma and pleomorphic adenocarcinoma were heterogeneously hyperintense on T2-weighted images. Adenoid cystic carcinoma, on the other hand, was homogeneously hyperintense on T2weighted images. MRI provides important information on the internal tissue features on T1- and T2-weighted images that cannot also be assessed with CT [1, 5, 7, 10]. Contrast enhancement Chronic dacryoadenitis, lymphoid, and epithelial tumors showed moderate contrast enhancement after gadolinium-DTPA injection. Therefore, enhancement after contrast injection does not seem to be a reliable sign in differentiating lacrimal gland lesions from one another [1]. Adjacent bone change In our study, chronic dacryoadenitis and lymphoid tumors did not produce bone change in any of the patients. Bone expansion was present in patients with pleomorphic adenoma, and bone destruction was noted in patients with malignant epithelial tumors. The presence of bone destruction strongly favors the diagnosis of a malignant lacrimal gland tumor [8]. However, not all malignant lacrimal gland lesions produce bone destruction. Jakobiec and associates reported that two of the nine malignant epithelial tumors in their series did not produce marked bone change using CT [5]. Furthermore, in long-standing cases benign pleomorphic adenoma can lead to bone destruction [4]. A number of other benign and malignant lesions that can occur in the lacrimal gland fossa (including schwannoma, dermoid cysts, chronic idiopathic orbital inflammation and metastatic carcinoma) can also lead to bone destruction [3, 5]. We realize that MRI is not the imaging method of choice to investigate bone change. CT can provide more detail
913
on bone features [5]. However, MRI can still reveal general information on the status of the adjacent bone [1]. Similarly, calcification within the mass, which is suggestive of a malignant lacrimal gland tumor, is not depicted well by MRI and is better identified on CT. A search of the literature on comparative MRI of lacrimal gland lesions revealed a preliminary report [6] and a review article [8] on the subject. In the preliminary study of Lemke and associates, MRI was compared with CT in the evaluation of 15 lacrimal gland lesions [6]. The authors found that MRI was superior to CT in the differentiation of non-epithelial lesions from epithelial tumors of the lacrimal gland [6]. CT was only found to be useful in the differentiation of pleomorphic adenomas with a heterogeneous internal structure, and in the evaluation of bony structures. Mafee and coworkers reviewed the MRI and CT scans of several lacrimal gland lesions and discussed their imaging features [8]. However, they did not specifically address the issue of differentiating various lacrimal gland lesions with MRI [8]. Although our study only involves a small number of patients, it is important as it is the first study to compare the MRI findings in patients with unilateral lacrimal gland lesions. The current study has limitations that should be addressed. First, the results are based on the data from only 31 patients. A larger number of patients would provide more definitive information, but it should be realized
that lacrimal gland lesions are uncommon. Second, a number of other lesions that can occur in the lacrimal gland area were not represented in our study. Therefore, the results from this group of patients should not be applied to patients with other non-epithelial or epithelial lesions. Third, some of the studies were performed without a surface coil and fat suppression providing suboptimal tissue resolution. Fourth, it is difficult to uniformly correlate MRI features with histopathologic findings. As mentioned above, different lacrimal gland tumors may show overlap with respect to several MRI features. In summary, MRI seems to be useful in determining the etiology of a unilateral lacrimal gland lesion. Anatomic extent, configuration, margins, and angulation features on MRI are important in evaluating a patient with a lacrimal gland lesion. However, internal tissue features on T1- and T2-weighted images of MRI are most helpful in categorizing these lesions. Contrast enhancement, on the other hand, is not helpful in determining the etiology of lacrimal gland lesions. Although inferior to CT imaging, MRI can detect adjacent bone change. The number of patients in our study is too small to permit the drawing of broad conclusions. However, our findings suggest that there are differences in orbital MRI findings between inflammatory lesions and lymphoid tumors, on one hand, and benign and malignant epithelial tumors of the lacrimal gland fossa, on the other.
References 1. Frohman LP, Kupersmith MJ, Lang J, Reede D, Bergeron RT, Aleksic S, Trasi S (1986) Intracranial extension and bone destruction in orbital pseudotumor. Arch Ophthalmol 104:380–384 2. Hornblass A, Friedman AH, Yagoda A (1981) Erosion of the orbital plate (frontal bone) by a benign tumor of the lacrimal gland. Ophthalmic Surg 12:737–743 3. Jakobiec FA, Yeo JH, Trokel S, Abbott GF, Anderson R (1982) Combined clinical and computed tomographic diagnosis of primary lacrimal fossa lesions. Am J Ophthalmol 94:785–807 4. Lemke AJ, Hosten N, Grote A, Felix R (1986) Differentiation of lacrimal gland tumors with high resolution computerized tomography in comparison with magnetic resonance tomography. Ophthalmologe 93:284–291 5. Mafee MF (1996) Eye and orbit. In: Som PM, Curtin HD (eds) Head and neck imaging, 3rd edn. Mosby, St. Louis, pp 1059–1128 6. Mafee MF, Edward DP, Koeller KK, Dorodi S (1999) Lacrimal gland tumors and simulating lesions. Clinicopathologic and MRI imaging features. Radiol Clin North Am 37:219–239
7. Parks SL, Glover AT (1990) Benign mixed tumors arising in the palpebral lobe of the lacrimal gland. Ophthalmology 97:526–530 8. Polito E, Galieni P, Leccisotti A (1996) Clinical and radiologic presentation of 95 orbital lymphoid tumors. Graefes Arch Clin Exp Ophthalmol 234:504–509 9. De Potter P, Dolinskas C, Shields CL, Shields JA (1995) Tumors of the lacrimal gland. In: De Potter P, Shields JA, Shields CL (eds) MRI of the eye and orbit, JB Lippincott, Philadelphia, pp 227–235 10. De Potter P, Flanders AE, Shields JA, Shields CL, Gonzales CF, Rao VM (1994) The role of fat suppression technique and gadopentate dimeglumine in magnetic resonance imaging evaluation of intraocular tumors and simulating lesions. Arch Ophthalmol 112:340–348 11. Reese AB (1956) The treatment of expanding lesions of the orbit: with particular regard to those arising in the lacrimal gland. The Seventh Arthur J. Bedell Lecture. Am J Ophthalmol 41:3–11 12. Rose GE, Wright JE (1992) Pleomorphic adenoma of the lacrimal gland. Br J Ophthalmol 76:395–400
13. Shields JA (1989) Diagnosis and management of orbital tumors. Saunders, Philadelphia, pp 259–274 14. Shields JA (1989) Diagnosis and management of orbital tumors. Saunders, Philadelphia, pp 316–340 15. Shields CL, Shields JA, Eagle RC, Rathmell JP (1989) Clinicopathologic review of 142 cases of lacrimal gland lesions. Ophthalmology 96:431–435 16. Shields JA, Shields CL, Eagle RC Jr, Singh AD, Armstrong T (2001) Metastatic renal cell carcinoma to the palpebral lobe of the lacrimal gland. Ophthalmic Plast Reconstr Surg 17:191–194 17. Vangveeravong S, Katz SE, Rootman J, White V (1996) Tumors arising in the palpebral lobe of the lacrimal gland. Ophthalmology 103:1606–1612 18. Wright JE, Stewart WB, Krohel GB (1979) Clinical presentation and management of lacrimal gland tumors. Br J Ophthalmol 63:600–606
