Clinical Findings

Table 1 summarizes the salient demographic characteristics and the preoperative and intraoperative findings of 6 patients with solitary lesions. All developed in the upper eyelids of adults with a median age of 68 years and an average of 62.5 years. The patients became aware of a lump in the eyelid and were symptomatic from 10 weeks to 3 years before the first excision. Three lesions were diagnosed as chalazia even after several recurrences; 1 lesion originally was suspected of being a melanoma, and 2 lesions were diagnosed correctly clinically from the outset. At the time of the last excision of their lesions, 2 patients were known to have an epithelial cystic lesion based on the results of an antecedent pathologic evaluation. One patient had 3 recurrences ranging from 9 months to 5 years after the first surgery, whereas a second had 4 recurrences at intervals between 5 months to 2 years. Only 1 patient had a decline of visual acuity to 20/80 with additional reports of pain and photophobia. The latter resulted from a tarsal surface irregularity that caused a superior marginal erosive keratitis. Otherwise, the patients described no discomfort or other ocular symptoms.

In 5 patients, careful external examination revealed a clearly visible noninflamed lump on the cutaneous aspect of the upper eyelid that felt firm and nonfluctuant and also was nonpainful ( Figure 1 , Top left and Top middle). In 1 patient ( Table 1 ; Patient 5), an externally discernible skin lesion was not present. The lesions measured from 3 to 9 mm in diameter. The overlying skin was of normal color, evidenced a healed scar where there had been earlier surgery ( Figure 1 , Top middle), and sometimes exhibited tenseness or tautness over the nodule. The externally palpable lesions were fixed to the tarsus; the overlying skin freely moved over the nodule. On eversion of the eyelids, the palpebral conjunctival surface was involved in all 6 cases (including the one without an obvious cutaneous lesion) and revealed signs of a subsurface intratarsal lesion. Three of these were clearly defined, upraised white or pale yellow bulges ( Figure 1 , Top right and Bottom left); 2 were gray to blue ( Figure 1 , Bottom middle), 1 of which had a vertical linear bluish extension that reached the eyelid margin to create a small notch; and the last appeared as a translucent area of the palpebral conjunctiva. None of the lesions was subjected to transillumination, nor was ultrasonography performed.

Photographs showing meibomian gland keratinous cysts. (Top left) A 76-year-old man ( Table 1 ; Patient 1) had a slightly whitened skin nodule with indistinct borders that had developed over 2 years. The lesion was fixed to the tarsus but the skin moved over it. There was no dysfunction of the eyelid. (Top middle) A 49-year-old woman ( Table 1 ; Patient 2) had experienced multiple recurrences before this last lesion appeared over a 9-month period. Note the cutaneous scar and the tautness of the overlying skin, lack of inflammation, and sharp margination. (Top right) Everted eyelid of the patient shown in the Top left panel revealing a smooth, white, elevated, circumscribed, noninflamed conjunctival component to the lesion. (Bottom left) A 31-year-old man ( Table 1 ; Patient 3) had been aware of a lump in his right upper eyelid for 2 to 3 years. Eversion disclosed an associated elevated, sharply demarcated yellow-white subconjunctival presentation. (Bottom middle) A 63-year-old woman ( Table 1 ; Patient 4) had a 3-month history of a palpable left upper eyelid nodule accompanied by a blue-gray palpebral nodule that was slightly elevated. (Bottom right) Incision of subconjunctival portion of the lesion ( Table 1 ; Patient 3) released milky, keratinous material that spread thinly and nasally toward the punctum.

Four surgical approaches initially were adopted. In 1 case (Patient 5), a full-thickness pentagonal central eyelid resection initially was performed that included the totality of the unruptured cyst. In 2 cases, there were sequential incisions and curettages until an anterior cutaneous incision through the eyelid crease led to complete removal. In 2 other cases, the first approach was an eyelid crease incision that included a partial tarsectomy; in only 1 patient was a transconjunctival partial tarsal excision attempted. During dissections, the cysts had a thick and well-developed wall or capsule that was always fused focally to the tarsus as an intrinsic part of it. During these dissections, the walls of all 5 of the locally excised cysts ruptured. Milky fluid ( Figure 1 , Bottom right), rather than semisolid, pasty, or cheesy material characteristic of a chalazion or a conventional epidermal cyst, was released in 4 of these 5 cases; dark bluish viscid material was extruded in the fifth. In each case, a full-thickness hole in the tarsus was left after a wide local excision was performed that included normal surrounding tarsus in the tarsectomy. Cautery was used in 3 cases to extirpate any potential remnants of the cysts’ epithelial lining after a correct histopathologic diagnosis ruling out a chalazion had been established.

The patient with the full-thickness eyelid resection has not experienced a recurrence during 20 months of follow-up. Two patients with one local excision have had follow-ups of 5 and 12 months without recurrence. The other 2 patients with multiple recurrences have been disease free for 6 months and 14 months since the last surgery, which finally included a partial tarsectomy and cautery. One patient who had an initial partial tarsectomy has been followed up for a disease-free period of 4 months.

Histopathologic Findings

The specimen obtained from the full-thickness eyelid resection ( Table 1 ; Patient 5) contained 2 cysts, a larger one at mid-height in the tarsus and a smaller one nearer the eyelid margin ( Figure 2 , Top left). The cysts were positioned approximately in the center of the tarsus, with somewhat more collagen separating the posterior edge from the palpebral conjunctiva than from the anterior orbicularis muscle. The tarsus was blandly scarred without inflammation and was devoid of sebaceous lobules and ducts above the cysts. A few surviving elements of the acinar elements of the Meibomian glands were observed near the eyelid margin. The cysts were filled with compact keratin and were lined by a smooth, curvilinear, multilaminar squamous epithelium that was devoid of keratohyalin granules and a distinct cuticle ( Figure 2 , Top right). No hair structures were observed within the keratin. Hematoxylinophilic granular material was found between the keratinous lamellae ( Figure 2 , Top right) within the central cavity. The Brown-Hopps stain ( Figure 2 , Top right, outer inset) disclosed clusters of positive, blue-staining cocci without any associated inflammation; the von Kossa stain ( Figure 2 , Top right, inner inset) revealed the black staining of calcium; and the Prussian blue stain for iron demonstrated negative results.

Photomicrographs showing meibomian gland keratinous cysts. (Top left) Full-thickness eyelid resection (Patient 5) containing two adjacent intratarsal keratinous cysts. On the bottom left, note the uninflamed scarred tarsus devoid of Meibomian ducts and acini. The surviving Meibomian acini near the eyelid margin and next to the smaller cyst fail to exhibit squamous metaplasia (hematoxylin and eosin, ×40 magnification). (Top right) The larger cyst (Patient 5) displays trichilemmal keratinization without keratohyalin granular or cuticular layers. Hematoxylinophilic granular material is prominent within the intracavitary compactions of anuclear keratin. The outer inset demonstrates collections of bacteria; the inner inset reveals the presence of calcium. Prussian blue staining failed to reveal iron. A Meibomian secretory acinus with an inconspicuous outer basal germinal cell layer is present on the bottom right and shows preservation of centrally located, well-differentiated sebocytes without any evidence of squamous metaplasia (hematoxylin and eosin, ×100 magnification). (Middle left) Two levels through a single cyst (Patient 1). The component on the upper left includes a segment of the tarsus with embedded Meibomian glands displaying central lipidization (hematoxylin and eosin, ×30 magnification). (Middle right) The wall of the cyst is composed of poorly vascularized, tightly woven bundles of tarsal collagen without inflammation (Patient 1). The epithelial lining is undulating and the keratin in the cyst is string-like and loose. The inset highlights a delaminating inner cuticular layer surmounting the crenulated squamous epithelium that lacks a keratohyalin granular layer. Intracavitary string-like keratin has accumulated (hematoxylin and eosin, ×100 magnification; inset, ×400 magnification). (Bottom left) Collapsed serpiginous cyst (Patient 6) with a prominent epithelial lining and a thick, uninflamed fibrous wall (hematoxylin and eosin, ×40 magnification). (Bottom right) The epithelial lining is composed of 4 to 5 layers of corrugated squamous epithelium. Note the string-like, deeply eosinophilic keratin strands in the lumen. The fibrous wall is constituted by tarsal collagen. The inset discloses the thick keratin strands that have shed from the cuticle of the epithelial lining (hematoxylin and eosin, ×100 magnification; inset, ×400 magnification).

The other 5 specimens of the locally excised lesions showed remarkably similar histopathologic findings. If the cyst protruded anteriorly into the eyelid skin, it was still clearly attached to a portion of excised tarsus ( Figure 2 , Middle left). The 5 locally excised cysts exhibited poorly vascularized, thick collagenous walls ( Figure 2 , Middle right, Bottom left and right). No focal collections of mononuclear inflammatory cells or hemosiderin-laden macrophages were observed in the pericystic dense connective tissue. The epithelial linings were composed of squamous cells without keratohyalin granules or intermixed goblet or sebaceous cells ( Figure 2 , Middle right and Bottom right). The innermost cells were thrown into undulations or corrugations (crenulations) with an intensely eosinophilic cuticular layer bordering the lumen. The cuticle diffusely desquamated into the lumen as needle-like or string-like refractile, loose, anuclear, keratinous structures ( Figure 2 , Middle right, inset and Bottom right, inset), rather than resembling the more loosely woven keratin typical of epidermal cysts. Atrophic Meibomian sebaceous glands were found in the vicinity of the cysts in 2 specimens. One specimen of a subtotally removed cyst that was obtained via a transconjunctival approach caused some initial interpretive confusion. The conjunctival surface was covered by shaggy, nonkeratinizing epithelium that eventually was demonstrated to possess goblet cells by means of Alcian blue staining. It was distinguished readily from a nearby portion of the cyst’s lining, which exhibited larger cohesive eosinophilic squamous cells with undulations, an inner cuticular layer, and the absence of goblet cells.

Immunohistochemical Findings

Table 2 summarizes the results of the immunohistochemical studies. The immunolabeled antigens that most sharply distinguished the intratarsal Meibomian gland keratinous cysts from the cutaneous epidermal cysts were cytokeratin (CK) 17 ( Figure 3 , Top left and right), CEA, and EMA. This staining profile paralleled that of the normal ducts (short lateral ductules and main vertical excretory ducts) and the obstructed and secondarily ectatic Meibomian ducts. CK-5/6 and CK-17 stained only the outer basal cells of the normal Meibomian acini. With respect to the Meibomian gland cysts, CK-5/6 ( Figure 3 , Middle left) stained the basal and suprabasal region along with the cuticle and luminal keratin strands, whereas CK-14 ( Figure 3 , Middle right) stained most layers, but not the cuticle or keratin. CEA strongly stained the cuticle, and EMA did so to a lesser degree ( Figure 3 , Bottom left). The staining characteristics of the intracavitary keratin contents revealed by the various probes were not helpful in differential diagnosis. The epidermal cysts were best distinguished by negative immunostaining for CK-17, CEA, and EMA ( Figure 3 , Bottom right). AE1/AE3 consistently was positive only in the duct cysts and unreliably in the epidermal cysts; CK-7 was of no value in differentiating the 2 types of cysts.

Photomicrographs showing meibomian gland keratinous cysts. (Top left). Antibodies against cytokeratin 17 (CK-17) immunostain the cyst’s epithelial lining as well as the intracavitary cuticular keratinous material (immunoperoxidase method, diaminobenzidine chromogen, ×35 magnification). (Top right) CK-17 cytoplasmic positivity is exhibited by basilar, suprabasilar, and parabasilar cells, but not by the apical cells. Note that the cuticle also stains, as do the shedding keratinous strands piling up in the lumen (immunoperoxidase method, diaminobenzidine chromogen, ×400 magnification). (Middle left) Antibodies against CK-5/6 strongly immunoreact with the basilar and suprabasilar layers of the epithelial lining and the intracystic keratinous contents. The inset reveals that the cuticle is stained moderately and scrolls off into the lumen. S = stroma of cyst wall (immunoperoxidase method, diaminobenzidine chromogen, ×40 magnification). (Middle right) CK-14 immunolabels the epithelial lining but not the cystic contents. The inset demonstrates absent immunostaining of the cuticle and the shed keratin. S = stroma of cyst wall (immunoperoxidase method, diaminobenzidine chromogen, ×40 magnification; inset, ×200 magnification). (Bottom left) Anti–carcinoembronic antigen (CEA) antibodies vividly label the cuticle and the strand-like intraluminal keratin it generates. The inset reveals that epithelial membrane antigen (EMA) is less intensely positive. CEA and EMA immunostaining supports a ductal origin for the cysts (immunoperoxidase method, diaminobenzidine chromogen, ×400 magnification; inset, ×200 magnification). (Bottom right) Anti-CK-17 antibodies fail to highlight the epidermis, the wall of the epidermal cyst and its keratin contents. Only portions of vellus hairs immunoreact positively (immunoperoxidase method, diaminobenzidine chromogen, ×40 magnification).

The different cytokeratins were located preferentially in varying cellular layers: 1) basal and suprabasal, 2) parabasal or mid-level, 3) apical or adlumenal, and 4) cuticular. For the Meibomian cysts, CK-5/6 was identified in the basal and suprabasal layers, and CK-14 and CK-17 were detected in the basal, suprabasal, and parabasal layers. CK-14 stained the cytoplasm more intensely than CK-5/6 and CK-17. Whereas CEA and EMA both stained the cuticular and adlumenal layers of the cysts, CEA also stained the keratin within the duct cysts but EMA did not. For the cutaneous epidermal cysts, CK-17 was totally negative in all layers, CK-5/6 was observed in the basal and suprabasal layer, and CK-14 was observed in the basal, suprabasal, and parabasal layers.

Positive Ki-67 absolute counts (average of immunolabeled nuclei in 3 high-power ×400 microscopic fields) were: 18 for 2 separate examples of secondarily ectatic Meibomian ducts; 24 for the 6 Meibomian cysts; 22 for the germinal layer of normal epidermis in 3 specimens; and 14 for 2 epidermal cysts. In 1 specimen ( Table 1 ; Patient 5) with a preserved Meibomian gland secretory lobular architecture, the average Ki-67–positive cell count was 40.3 and was restricted to the outer compressed rim of basal germinal cells.