Updates in Ocular Surface Tumor Diagnostics

Abstract

Purpose

To evaluate technologies and techniques available for the diagnosis of ocular surface tumors.

Methods

A review of the literature from 1947 to 2017, through the PubMed Database, was conducted in order to evaluate current diagnostic methods for ocular surface tumors.

Results

Ocular surface squamous neoplasia, conjunctival melanoma, and conjunctival lymphoma are the three most common ocular surface malignancies. Technologies available to assist with diagnosis of these conditions, in addition to full thickness biopsy, include vital dyes, aspiration and impression cytology, in vivo confocal microscopy, ultrasound biomicroscopy, genetic testing, and anterior segment optical coherence tomography.

Conclusions

Histology remains the gold standard for diagnosis for all 3 of these malignancies. However, multiple diagnostic techniques are available to assist in making preliminary and early diagnoses, in differentiating between similar-appearing lesions, and in some cases, avoiding biopsy prior to initiating treatment. As imaging and technology continue to evolve, these adjunctive techniques will likely continue to play a greater role in clinical practice.

Ocular surface tumors represent a range of conditions, from benign to malignant lesions. They originate from a variety of cell types, forming epithelial, melanocytic, lymphoid, leukemic, fibrous, lipomatous, and other lesions. The most common malignant tumors of the cornea and conjunctiva are ocular surface squamous neoplasia (OSSN), conjunctival melanoma (CM), and conjunctival lymphoma (CL)¹. Each of these tumors arises from a similar-appearing pre-malignant lesion or can resemble a pre-malignant lesion. Differentiating between benign and malignant lesions, as well as between these three malignant conditions, can sometimes present a clinical challenge. However, accurate diagnosis is paramount because of the differences in treatment of these lesions. In this review, we will highlight the various modalities available for the diagnosis of conjunctival tumors and in particular for these three most common malignant tumors.

I. Ocular Surface Squamous Neoplasia

OSSN is an umbrella term encompassing a spectrum of disorders ranging from dysplasia to invasive squamous cell carcinoma, and is the most common malignant tumor of the ocular surface². Risk factors include ultraviolet light exposure, human immunodeficiency virus and other forms of immunosuppression, age, and male sex^3–7. OSSN generally presents as unilateral disease, but may be bilateral in immunosuppressed patients. Patients are often asymptomatic, but the most common presenting symptoms are redness, tearing, grittiness, or pain⁷.

OSSN can present on the cornea or conjunctiva in a variety of ways, including as a gelatinous mass, an elevated growth with overlying leukoplakia, a nodule, or when on the cornea, as a flat opalescent layer. The diagnosis can often be made clinically, based on a classic appearance. At times, however, the lesion can be subtle, and difficult to determine clinically. In addition, there is a wide breadth of lesions, both benign and malignant, that can masquerade as OSSN. These include pterygia, amelanotic melanoma, corneal pannus, corneal nodular degeneration, pyogenic granuloma, sebaceous cell carcinoma, and others^8,9. In order to differentiate between these growths, other diagnostic modalities besides clinical exam are sometimes required.

The gold standard for the diagnosis of OSSN involves excisional biopsy, with histology of the specimen. The advantage of excisional biopsy is that it allows one to obtain a histologic diagnosis for the entirety of the specimen. However, for large and diffuse lesions without clear margins, even intended excisional biopsy may not remove the entire lesion. Also, for large or recurrent lesions, excisional biopsies lead to increased risks of limbal stem cell deficiency, symblepharon, and scarring^10,11. Lastly, as medical options for treatment of OSSN increase in number and become more utilized, non-invasive methods of diagnosis become more important¹².

A. Vital Dyes

In addition to clinical exam alone, vital dye staining can assist in making the diagnosis of OSSN both inexpensively and quickly; these vital dyes include rose bengal, methylene blue, and toluidine blue. Rose bengal is a iodinated derivative of fluorescein that stains degenerating, devitalized, and dead epithelial cells in a bright pink color. Due to its ability to stain unhealthy and metabolically deranged cells, this dye highlights epithelial neoplasms on the eye¹³ (Figure 1). However, because it stains any devitalized tissue, it is not specific for OSSN and will also stain exposed epithelium along an elevated pterygium or unhealthy epithelium in dry eye syndrome. Methylene blue is an acidophilic dye that penetrates cells and has a selective affinity for nucleic acids, thus staining cells with a high metabolic rate. This dye is useful for detecting malignant lesions, as surface neoplasias pick up the stain at a high rate. Like rose bengal, however, methylene blue is not specific for OSSN and benign lesions can also show positive dye uptake¹⁴. Toluidine blue, another acidophilic dye, also stains cells with a high mitotic rate, as well as accumulating between cells and therefore staining tissues with poor cell-to-cell adhesion. Similar to methylene blue, the test is sensitive, meaning that OSSN lesions have a high rate of staining; however, benign lesions may also take up dye¹⁵. The staining of benign lesions with these acidophilic dyes is likely because of a similar causal pathway between OSSN and benign lesions such as pterygia and actinic keratosis. Vital dyes are therefore helpful adjuncts to clinical examination but their staining is not specific to malignant squamous lesions. Rose bengal can also cause some significant burning at the time of instillation, less so for the other agents.

Figure 1.

An 81 year old female with OSSN of the left eye. Rose bengal highlights devitalized and metabolically deranged epithelial cells, thus highlighting her limbal tumor.

B. Cytology

Another adjunctive diagnostic method involves cytology, either by aspiration or impression. In aspiration cytology (AC), cells are aspirated from the conjunctival surface with a fluid-filled syringe; a syringe without a needle is used rather than aspiration of cells from within the tumor.¹⁶ A small study of AC for OSSN has shown good histological correlation between this technique and subsequent biopsy¹⁶. However, limitations include lack of adequate sampling, both in terms of cell number and sampling of only superficial cells, thereby not enabling a differentiation between invasive and non-invasive disease. The technique also requires experience, both of the practioner performing the aspiration cytology and of the cytologist reading the specimen. Due to these limitations, AC has limited applicability¹⁶.

Impression cytology (IC) uses a technique of collecting superficial cells from the surface of the eye with filter paper; this paper is attached to a collecting device. The device is applied to the surface of the eye to adhere cells to the paper. Cells are then fixed, stained, mounted, and evaluated for atypia, based on features such as enlarged nuclei, irregular nuclear outline, coarse chromatin and presence of prominent nucleoli, and hyperchromasia¹⁷. Studies of IC, like AC, show overall good correlation, from 80 to 84%, with histology specimens from biopsy^17,18. Similar to AC and unlike biopsy, IC has the advantage of being minimally invasive and preserving limbal stem cells. It is quick and can be performed even on children in clinic. However, similar limitations to AC are present, including the technique being limited to superficial sampling of cells and the requirement of an experienced cytologist. Additionally, IC has decreased sensitivity when used for keratotic lesions, as the abundance of surface keratin can decrease the sampling of actual cells. Consecutive repeated applications of filter paper, in order to access the deeper epithelium and remove any superficial keratin, have been shown to increase sensitivity. However, invasive disease can still not be definitively identified¹⁹.

C. In Vivo Confocal Microscopy

In vivo confocal microscopy (IVCM) is a non-invasive technique that allows en face images of the ocular surface. It utilizes a point or slit light source and a point or slit detector with the same focal point; the source and detector use conjugate pinholes to increase optical resolution by decreasing interference from light contamination²⁰. Impressive resolution, down to the cellular level, is obtained, although only one plane can be imaged during any time point. The technique is safe and non-invasive and can be performed in clinic so the information is immediately available. However, although it is non-invasive, there are both contact and non-contact forms; the contact technique does have risks of corneal abrasion and infection.

Studies of IVCM for OSSN have had mixed conclusions. Some have shown several characteristic features of the tumor, including hyperreflective pleomorphic cells of varying shapes and sizes, large nuclei, bright dots within middle layer cells corresponding to prominent nucleoli, a basal cell layer with poorly defined cell borders and bright nuclei, and the apparent absence of subbasal nerves due to the high reflectivity of the OSSN cells causing difficulty in imaging (not due to actual degenerated nerves)^21–23. In small studies of IVCM, it has been shown that the technology has good correlation to histology; however, as these were all correlation studies, the sensitivity and positive predictive value of this technique is not yet known^21–23. One of these studies did show that IVCM is useful for predicting the grade of dysplasia and the extent of epithelial involvement as well as stromal invasion²³.

In contrast to these smaller studies, in the largest study to date of IVCM for OSSN did not find IVCM to be as helpful; this study included 120 patients, 52 of which underwent excisional biopsy with histopathology. A significant overlap was identified between IVCM findings for benign and malignant lesions, including hyperreflective and pleomorphic cells and a “starry night” appearance of the basal cell layer in both groups²⁴. In addition, some OSSN lesions showed no suspicious IVCM characteristics at all. While IVCM may be a useful adjuvant to histology, these findings indicate that it cannot replace biopsy. Limitations include the requirement for operator expertise in capturing and the interpretation of images, and decreased visualization in very thick/keratinized lesions that degrade the optical properties of the normally transparent conjunctival and corneal epithelium^22,23.

D. Anterior Segment Optical Coherence Tomography

The advent of optical coherence tomography (OCT) has allowed clinicians to obtain an in vivo, cross-sectional, high-resolution image of ocular tissue. The OCT, which was initially limited to the posterior segment, was first used for anterior segment imaging in 1994²⁵. It has recently been shown to be very helpful for ocular surface lesions as well^9,26–32.

OCT uses a Michelson interferometer, which splits a beam of light into a reference beam that is reflected from a mirror, and a measurement beam, which is reflected from the variably reflective tissue layers of the eye. These lights are then combined upon reflection to create an interference pattern, which is detected. Multiple interference patterns are created over the structure being imaged, and used to create a series of A-scans that are put together into a composite, cross-sectional image of the tissue.

With the original time-domain OCT (TD-OCT) technology, the images are created as the reference mirror moves in a linear fashion. In contrast, with the newer spectral domain OCT (SD-OCT), the detector arm of the interferometer uses a spectrometer instead of a single detector, allowing simultaneous detection of a broad range of depths; the entire A-scan is then produced using Fourier transformation of spectrometer analysis. As scanning can be performed much faster using SD-OCT, resolution is also better³³. Compared to 18 micron resolution with TD-OCT, resolution with SD-OCT is 3 microns with custom-built ultra-high resolution devices³⁰ and 5 microns with commercially-available high-resolution machines.

While an initial study using TD-OCT for imaging conjunctival tumors found that TD-OCT and UBM had a similar ability to image surface tumors²⁶ the improved resolution of SD-OCT improves the diagnostic ability of the OCT devices. Classic characteristics of anterior segment OCT (AS-OCT) for OSSN, using SD-OCT, include an abrupt transition between the normal and abnormal epithelium, as well as epithelial thickening and epithelial hyper-reflectivity in the area of the tumor^28,31 (Figure 2). SD-OCT for the anterior segment is useful in differentiating between OSSN and benign conditions such as pterygium, scarring, Salzmann’s nodular degeneration, and limbal stem cell deficiency as well as between OSSN and other malignant conditions such as melanoma^9,30,31.

Figure 2. A 63 year old male with OSSN of the left eye.

a. Slit lamp photograph of a sessile conjunctival lesion extending to the limbus.

b. AS-OCT reveals an abrupt transition (arrow) between normal and thickened hyperreflective (asterisk) epithelium

Advantages of OCT compared with other imaging for the anterior segment include that is that it is not only non-invasive but also non-contact. Imaging can be performed in clinic, so results are immediately available. And, unlike confocal microscopy, which requires an experienced user for interpretation, AS-OCT can be quickly taught to even novice users with moderate sensitivity but high specificity (unpublished data). However, limitations of this technology include difficulty in imaging the posterior portion of thick, keratinized, and pigmented lesions due to shadowing of the image³⁰. With thick tumors, imaging can be done at the edges to minimize shadowing, but it remains difficult to image the entire posterior border. There is also limited knowledge regarding the ability of the technique to detect invasion into the stroma.

II. Conjunctival Melanoma

CM is a rare but potentially life-threatening tumor of the ocular surface. It comprises 2% of all ocular malignancies, but similar to cutaneous melanoma, the incidence of CM is rising³⁴. CM is most common in non-Hispanic white populations, but it can occur in any group, including blacks, Hispanics, Asians, and Native Americans³⁵. CM generally arises from primary acquired melanosis (PAM), but can also arise from conjunctival nevi or de novo³⁶. The clinical presentation of CM is typically a pigmented and nodular lesion on the bulbar conjunctiva with surrounding feeder vessels. However, CM can also occur on the palpebral conjunctiva or caruncle, and may be minimally elevated^37,38. Additionally, these tumors may be amelanotic or minimally pigmented in up to one-fifth of cases, causing diagnostic difficulty and often, diagnostic delay³⁶. Risk of metastatic disease with CM varies with origin, and at ten years is 25–49%, while 10-year tumor related mortality is 9–35%³⁶. Risk factors for both metastatic disease and mortality include increased tumor thickness, local recurrence, and de novo tumor origin.^36,39–41

As with OSSN, the gold standard for diagnosis of CM is histology, after excision with cryotherapy is performed for complete removal of the lesion. Excisional surgery allows for removal of the entire lesion, so that all areas can be examined for invasive disease and so that margins can be evaluated for atypical cells. A total excision, although important for all tumors, has even greater implications for conjunctival melanoma, as incomplete excision leads to elevated risk of local recurrence, metastatic disease, and tumor-related death^42,43. Diagnostic imaging is therefore especially important in CM, to avoid an incisional biopsy and to proceed directly to full excisional procedures. The diagnosis of CM depends on differentiation not only between various malignant tumors but also between CM and its precursors. Diagnostics for CM precursors will therefore also be discussed in this section.

A. Impression Cytology

The first study of IC for lesions of melanocytic origin was published in 1992 and evaluated IC for PAM, nevi, and melanoma⁴⁴. Cytologic features of malignancy in melanocytic lesions include an increased nucleus to cytoplasmic ratio, enlarged nucleoli, irregular nuclear chromatin pattern, anisokaryosis, and mitoses; a high relative proportion of atypical cells with these features is consistent with a diagnosis of malignant melanoma. Correlation between impression cytology and histology for diagnosis in that study was 73%; errors included the diagnosis of melanoma as PAM in one eye⁴⁴. A second study with IC using similar grading criteria found a correlation with histology of 88%, with errors again including melanoma being diagnosed by IC as a pre-malignant condition rather than melanoma⁴⁵.

IC, as discussed previously, has advantages of being painless and minimally invasive. Additionally, it takes advantage of the ascent of atypical melanocytes to the epithelial surface that it is indicative of malignancy⁴⁴. A recent case report of an unusual amelanotic lesion on the cornea, in which IC provided the first indication of melanoma, demonstrates the utility of IC in challenging diagnostic cases and its use as an adjuvant to histologic diagnosis⁴⁶. However, with the risk of local and distant metastasis increasing as the tumor grows and thickens, the false negatives seen in both studies do raise concern. Also, as IC is limited to the superficial cells, it is not able to detect the atypia in early PAM that begins at the basal epithelium. IC is also limited by tumor location, as sampling is difficult for tumors in the fornices, palpebral conjunctiva, and caruncle^44,45.

II. In Vivo Confocal Microscopy

IVCM characteristics for PAM, nevi, and melanoma were developed by Messmer et al.⁴⁷ For conjunctival nevi, IVCM demonstrates nests or collections of medium-sized, uniform stromal cells as well as pseudocyst-like structures. For PAM with atypia, hyper-reflective cells and large dendritic cells are seen throughout the epithelium, while in PAM without atypia the hyper-reflective cells remain mostly in the basal epithelium and the dentritic cells are small. Conjunctival melanoma has the characteristic feature of large cells with prominent nuclei/nucleoli. Subepithelial highly reflective cells are also directly visible, confirming invasion. In vivo confocal microscopy for the diagnosis of CM has a sensitivity of 89% and a specificity of 100%⁴⁷. While this indicates that IVCM cannot replace histology as the gold standard of diagnosis, it does suggest that the need for excisional biopsy can be guided by IVCM findings. Similar results have also been found using a handheld dermatologic confocal device in place of the floor-mounted ophthalmic devices, which allows for easier alignment with the region of interest, including more difficult to image locations such as the caruncle⁴⁸.

III. Anterior Segment Optical Coherence Tomography

AS-OCT can be helpful in differentiating between various tumors of melanocytic origin. As first noted by Shields et al., AS-OCT of conjunctival nevi demonstrates intrinsic cysts that infer a benign lesion⁴⁹. Using high-resolution AS-OCT, intrinsic cysts may be seen even when not visible by clinic exam; in addition, with high-resolution OCT, nevi demonstrate hyperreflective basal epithelial layers^30,31. For AS-OCT of primary acquired melanosis, multiple studies have demonstrated a uniform hyper-reflective band along the basal epithelium, with normal overlying epithelium and an absence of cysts. However, AS-OCT, even at ultra-high resolution, is unable to differentiate between PAM with and without atypia, as these devices are unable to evaluate cellular details and there is no difference between hyper-reflective band thickness between lesions with and without atypia^30–32. For CM, both pigmented and amelanotic, AS-OCT reveals a hyper-reflective epithelium of variable thickening overlying a hyper-reflective sub-epithelial mass, and an absence of cysts^30,31.

AS-OCT, in the diagnosis of CM, is principally helpful in differentiating it from other types of tumors. The distinct differences between AS-OCT for CM and OSSN allow differentiation of these tumors, even in difficult cases such as amelanotic melanomas (Figure 3) or pigmented OSSN⁹. The technology does have limitations in the diagnosis of pigmented lesions, however. Optical shadowing is a concern for AS-OCT for all pigmented lesions, as it prevents calculations of depth. Additionally, unlike cytology and IVCM, atypia cannot be identified at a cellular level.

Figure 3. A 56 year old male with amelanotic melanoma of the left eye.

a. Slit lamp photograph of a conjunctival lesion, extending to the limbus. The mass is predominately amelanotic with the exception of small area of pigment at the limbus.

b. AS-OCT showing a thin epithelium overlying a large and elevated subepithelial mass with shadowing of the underlying tissue. The OCT is consistent with the appearance of a conjunctival melanoma.

c. Atypical basophilic cells with prominent nucleoli present within the substantia propria (Original magnification x40)

d. Melan-A with red chromogen highlights the tumor cells within the substantia propria (Original magnification x100), consistent with conjunctival melanoma

IV. Ultrasound Biomicroscopy (UBM)

UBM uses sound waves to analyze structures; the frequency of waves used is higher than classic ocular ultrasound, which increases resolution but decreases penetration into tissue. The resolution of UBM is 42 microns and penetration is 4–5mm⁵⁰. UBM is able to penetrate opaque tumors, allowing the posterior margin of CMs, even heavily pigmented melanomas, to be visualized. This ability improves detection of tumor invasion²⁶. A small study of patients with CM also showed good correlation between tumor thickness on UBM and histology, which is useful for prognostication⁵¹. However, while UBM is able to delineate tumor margins and extent, the resolution of internal tumor details is limited. Other disadvantages of UBM are that it requires contact with the ocular surface and is generally only available in large centers.

V. Genetic markers

The genetics of melanoma, while not strictly used for diagnosis, are important in the diagnostic process because of their potential ability to guide therapy. The mitogen-activated protein kinase (MAPK) pathway is one of the main regulatory pathways involved in CM development, particularly through mutations in BRAF, NRAS, and KIT^52–54. BRAF mutations, seen in up to 50% of CM, are common in cutaneous melanoma but rare in mucosal and uveal melanoma^43,52,55. BRAF mutations appear to be early events in the development of CM, as they are seen in conjunctival nevi as well^43,55. CM with BRAF mutations tends to occur in younger patients and be associated with CM of nevus origin⁴³. Mutations in NRAS, which comprise up to 18% of mutations in CM, do occur in cutaneous melanoma also, but are more common in mucosal melanoma; NRAS mutations are very rare, however, in uveal melanomas. KIT mutations, which comprise the least common CM mutation in the MAPK pathway, occur in both cutaneous and mucosal melanomas.

For cutaneous melanoma, targeted therapies to BRAF (BRAF-inhibitors) have extended life expectancy in patients with those mutations⁵⁶. For CM, while no intralesional or topical therapy exists at this time targeted to genetic mutations, systemic targeted medications can be useful in patients with metastatic disease⁵⁷. Additionally, understanding these genetic mutations opens the door in the future for new and better local treatments.

III. Conjunctival lymphoma

Lymphomas are a group of neoplasms derived from clonal proliferations of lymphocytes. Approximately one-quarter of ocular adnexal lymphomas are confined to the conjunctiva⁵⁸. Ninety-eight percent of conjunctival lymphomas (CL) are derived from B-cell lineage, and most of these are derived from four major subtypes: 1) extranodular marginal zone lymphoma, 2) follicular lymphoma, 3) diffuse large B-cell lymphoma, and 4) mantle cell lymphoma. CL tends to present in an elderly population, but can occur at any age. It generally presents as a pink, or salmon-colored lesion, which is asymptomatic⁵⁹. It also has a follicular clinical form, which may be subtle.⁶⁰

CLs, despite being malignant tumors, often present without classically malignant features of feeder vessels or rapid growth. No clear clinical difference can be identified between CL and reactive lymphoid hyperplasia, the latter of which is benign. In general, malignant lesions, when compared to benign lesions, tend to occur in older patients, in the fornix, and tend to be larger and bilateral⁶¹; however, none of these characteristics definitively characterize a mass as malignant. As mentioned above, CL can present as a chronic follicular conjunctivitis, an entity that has many etiologies, most of which are also benign⁶⁰. For a lesion suspicious for CL, the gold standard of diagnosis is a conjunctival biopsy, which is then sent for both histology and flow cytometry and gene rearrangement. The mainstay of treatment of CL involves orbital radiation if the disease is local or systemic chemotherapy if the disease is metastatic⁶². Rarely, excision with cryotherapy or intralesional chemotherapy or immunotherapy is utilized^63–65. Approximately 17% of CLs with unilateral involvement and 47% of CLs with bilateral involvement manifest systemic disease, either at the time of diagnosis or that develops thereafter⁶³.

A. In Vivo Confocal Microscopy

Limited work using IVCM has been performed for lymphoid lesions. The first demonstration, in 2008, of IVCM for CL showed small, hyper-reflective, tightly-packed cells in nests or cyst-like spaces⁶⁶. Cellular atypia, to differentiate these cells from those seen in inflammatory conditions, however, cannot be identified⁴⁸. IVCM may therefore be helpful in differentiating CL from other malignant tumors, but given its similarity in appearance to inflammatory lesions, cannot supplant histology for a definitive diagnosis.

B. Anterior Segment Ocular Coherence Tomography

AS-OCT for CL is characterized by a homogenous, hypo-reflective mass, with shadowing of the underlying tissue in thick lesions (Figure 4). On UHR-OCT, the dark mass appears to be made up of small, hyper-reflective stippled dots, while on lower resolution OCT, these dots may not be visualized^30,31. Furthermore, on HR-OCT, a hyper-reflective band of tissue can be seen overlying the tumor³¹. This likely represents conjunctival tissue displaced upward by the subconjunctival infiltrate. No study of AS-OCT to date, however, has demonstrated a comparison between conjunctival lymphoma and benign reactive hyperplasia, so it is unclear whether AS-OCT can be used for differentiation between CL and its benign masquerading counterparts.

Figure 4. A 57 year old male with conjunctival lymphoma of the left eye.

a. Slit lamp photograph of the left eye showing a nasal, fleshy, salmon patch lesion.

b. AS-OCT shows a dark, monomorphic, hyporeflective subepithelial lesion (arrow) with normal overlying epithelium.

IV. Summary

Ocular surface squamous neoplasia, conjunctival melanoma, and conjunctival lymphoma are the most common malignant tumors of the ocular surface. Each has a classic clinical appearance, but may also present in a subtle or atypical fashion, thus requiring more than slit lamp examination for diagnosis. The gold standard for diagnosis of each of these lesions remains histology. However, numerous modalities including cytology, IVCM, AS-OCT, and UBM can assist with prompt diagnosis. Understanding the genetics of these tumors is also important, as it improves the development of targeted therapies. As technology continues to improve, we will likely see a host of new diagnostic modalities in the future and an expansion of their use for ocular surface pathologies.