The Use of High Resolution Optical Coherence Tomography (HR-OCT) in the Diagnosis of Ocular Surface Masqueraders

Abstract

Introduction:

Ocular surface masqueraders encompass any ocular surface lesion masquerading as another ocular surface lesion. High resolution optical coherence tomography (HR-OCT) has emerged as an adjunctive tool to clinical acumen. This study’s purpose is to evaluate the utility of HR-OCT images in guiding the diagnosis and management of those lesions.

Material and methods:

22 individuals with a clinically ambiguous ocular surface lesion with slit lamp photographs (SLP), HR-OCT images, and histopathological examination were included in the study. The presumptive clinical diagnosis based on SLP was compared to the diagnosis suggested by HR-OCT findings and to definitive diagnosis by histopathology. The main outcome of this study was the frequency in which HR-OCT findings guided the clinician to the correct diagnosis.

Results:

7 lesions were epithelial, 3 had an epithelial and a subepithelial component, and 12 were subepithelial. HR-OCT was most effective in discerning lesion location, successfully identifying the location in 100% of cases. Classic HR-OCT findings were detected in 68.2% of cases while suggestive features were detected in 31.8% of cases. The epithelial lesions’ mean epithelial thickness was 265.4±140.6 μm, the subepithelial lesions’ mean was 58.0±25.0 μm, and the combined lesions’ mean was 140.0±70.0 ±m. The epithelium was significantly thicker in epithelial lesions compared to subepithelial and combined lesions. By ROC analysis we identified that using a cut off of 156 μm, the sensitivity was 86% and the specificity was 93%.

Discussion:

HR-OCT can be a valuable diagnostic tool, assisting in the differentiation of ambiguous ocular surface pathologies by providing a cross-sectional, morphological image of the lesion.

1. Introduction:

The term “masquerade syndrome” was first used in 1967 by Theodore to characterize a conjunctival carcinoma manifesting as chronic blepharoconjunctivitis. [1] The term has since been expanded to encompass any lesion masquerading as another entity. Early recognition and differentiation of these mimicker lesions is of paramount importance as some masqueraders may exhibit a significant risk of morbidity and mortality and timely diagnosis can guide appropriate intervention.

In a retrospective series of 5002 conjunctival tumors by Shields et al, approximately half of the lesions were benign and half malignant. [2] A diagnostic challenge may occur as the clinical presentation of benign and malignant lesions is diverse and can overlap. [3] There are multiple case reports in the literature of conjunctival malignancies such as ocular surface squamous neoplasia (OSSN) and lymphoma initially misdiagnosed as scleritis or chronic conjunctivitis, thus delaying appropriate intervention. [4-6] Histopathological and cytologic analysis of incisional or excisional conjunctival specimens remains the gold-standard in establishing the diagnosis of such ocular surface malignancies. However, exciting advances in imaging techniques have revolutionized the diagnosis and management of numerous ocular surface pathologies, providing a form of “optical biopsy”. [7]

Anterior segment high resolution optical coherence tomography (HR-OCT) in particular, has been studied for its use as a non-invasive and non-contact imaging modality to diagnose corneal and conjunctival lesions by providing cross-sectional images of the ocular surface. HR-OCT has been successful in distinguishing different ocular surface pathologies including OSSN, pterygium and pinguecula, lymphoma and pigmented lesions through the identification and establishment of classic findings of each class of lesion. [3] Moreover, HR-OCT reliably identified OSSN in a series of 16 patients with OSSN and co-existent ocular surface pathology, whose presence clouded the clinical detection of malignancy. [8] To date however, there are no studies describing the utility of HR-OCT in aiding in the diagnosis of potentially masquerading lesions beyond OSSN. Given this knowledge gap, this study aims to evaluate the utility of HR-OCT in guiding the diagnosis of a wide variety of clinically ambiguous lesions.

2. Materials and Methods:

This study was approved by the University of Miami Institutional Review Board, was conducted in accordance with the principles of the Declaration of Helsinki and was compliant with the Health Insurance Portability and Accountability Act. A retrospective chart review of 220 patients with ocular surface lesions identified through a clinical dataset at the Bascom Palmer Eye Institute, University of Miami Miller School of Medicine between the years of 2015 and 2020 was conducted. All patients underwent a complete ophthalmological examination by an ocular surface specialist (CLK). Inclusion criteria included individuals with an ocular surface lesion which was deemed clinically ambiguous by the experienced ocular surface specialist. Cases with available slit lamp photographs, HR-OCT images, and histopathological examination were included in this retrospective study (n=22).

2.1. Devices

Three types of spectral domain OCT machines were used for imaging, the Optovue Avanti (Fremont, CA), the Optovue RT Vue (Fremont, CA) and the custom built ultra-high resolution (UHR) OCT. The Optovue Avanti has a central wavelength of 840 nm, axial resolution of 5 μm, transverse resolution of 15 μm, and performs 70,000 A-scans per second. The Optovue RT Vue has a central wavelength of 840 nm, axial resolution of 5 μm, transverse resolution of 8 μm, and performs 26,000 A-scans per second. The custom built UHR-OCT has a central wavelength of 840 nm, axial resolution of 3 μm and captures 32 frames per scan.

2.2. Methodology

Along the area of clinical interest multiple scans were obtained radially with approximately 1 mm of raster separation. In addition, multiple radial 360 degrees scans across the corneal limbus were obtained. These scans were performed by the same two trained technicians who have specialized on imaging of the anterior segment. Scans were repeated if the images were not of high-quality. All images were reviewed by an ocular surface specialist (CLK). In addition, slit lamps photographs and HR-OCT images were assessed independently by a second observer (ocular surface specialist) who was masked to the clinical history.

The morphology, thickness, and reflectivity of the epithelial and subepithelial layers were evaluated. Increased whiteness of the involved tissue compared to normal tissue was described as hyperreflectivity. Darkening of the tissue within the lesion was described as hyporeflectivity. In addition, the epithelial thickness of all lesions was measured at their thickest point. In the 18 patients scanned with a commercial device, the distance-measuring tool (built into the software of the OCT machine) was utilized. In the 4 cases scanned with the custom device, the thickness was calculated by comparing the thickest point to a normal cornea epithelium of 55 um, as there was no caliper built into the software.

Histopathological and cytological analysis of the conjunctival biopsy specimens were performed in all cases. Formalin fixed tissue samples were obtained in all cases and fresh tissue samples when lymphoproliferative disease was entertained. Biopsy specimens were dehydrated, embedded in paraffin blocks and sectioned at 5 μm. Hematoxylin-eosin, Congo-Red, periodic acid Schiff and other designated stains were performed, as appropriate. Blocks were visualized through a light microscope (Olympus Optical Co., Tokyo, Japan) and were photographed with the use of a digital system. In cases of lymphoproliferative disease, cytological analysis was performed to evaluate for monoclonal and polyclonal lymphocytic proliferation along with gene re-arrangement studies.

Demographic information, tumor characteristics (affected eye, clinical appearance of the lesion, size, location), and HR-OCT findings were summarized. Patient demographics including age, sex, race, ethnicity, risk factors, medical history and medication information were collected from the patients’ medical records as available. The presumptive clinical diagnosis based on slit-lamp examination was compared to clinical diagnosis by HR-OCT findings and ultimately to definitive diagnosis by histopathology. The main outcome of this study was the frequency in which HR-OCT findings assisted the clinician in correctly identifying a lesion or suggesting a diagnosis. The HR-OCT findings used to identify a lesion or suggest a diagnosis are detailed in Table 1. [8-20]

Table 1.

Classic and suggestive high resolution optical coherence tomography (HR-OCT) findings of common benign and malignant lesions

Malignant Lesions		Classic HR-OCT findings	Suggestive HR-OCT findings
Ocular Surface Squamous Neoplasia (OSSN)	Intraepithelial neoplasia-Carcinoma in situ	Thickened epithelium Hyperreflective epithelium Abrupt transition of the normal from the abnormal epithelium	2 out of the 3 classical findings
	Squamous cell carcinoma	Variably thickened and hyperreflective epithelium Hyperreflective mass invading the subepithelial tissue, often with visible areas of epithelium invading the subepithelial space, heterogeneous	Normal or thickened epithelium Hyperreflective mass invading the subepithelial tissue which may or may not connected to the epithelium
Lymphoma		Thin epithelium Subepithelial homogenous, discreet hyporeflective massith monomorphic dot-like infiltrates. Normal conjunctiva is displaced by the mass creating bands of hyperreflectivity surrounding the lesion	Thin epithelium Foci of variably homogeneous, hyporeflective, subepithelial infiltrates Absence of the displaced normal conjunctiva
Melanoma		Mildly thickened epithelium of variable hyperreflectivity Hyperreflective subepithelial mass, usually homogeneous	Thin epithelium Subepithelial mass of variable reflectivity
Benign Lesions
Amyloid		Normal, thin epithelium Subepithelial infiltrate of heterogeneous, dark lesions with irregular borders, hyperreflective linear opacities within the infiltrate	Normal, thin epithelium Subepithelial infiltrates of heterogeneous lesions, no definitive hyperreflective linear opacities
Pterygium		Normal to slightly thickened mildly hyperreflective epithelium Hyperreflectivity stringy subepithelial mass near the limbus	Normal to slightly thickened mildly hyperreflective epithelium Subepithelial mass of variable hyperreflectivity
Nevus		Normal thickness, mildly hyperreflective epithelium Subepithelial mass &presence of cysts	Normal thickness, mildly hyperreflective epithelium Subepithelial mass with no or rare cysts
Salzmann’s nodular degeneration		Thinner than normal epithelium Subepithelial, hyperreflective nodular material anterior to the Bowman’s layer	Normal thickness epithelium Subepithelial, hyperreflective lesion anterior to the Bowman’s layer flat, not nodular
Melanosis (primary or complexion associated)		Thin epithelium Linear hyperreflectivity of the basal epithelium	Thin epithelium Possible foci of hyperreflectivity with no definitive hyperreflectivity of the basal epithelium
Neuroma		Thin epithelium Well circumscribed subepithelial lobular areas of mixed reflectivity	Thin epithelium Subepithelial lobular areas of mixed reflectivity
Pannus		Normal epithelium Flat, highly hyperreflective lesion along Bowman’s layer	Normal epithelium Trace hyperreflectivity or nodularity along Bowman’s

^*Posterior shadowing often occurs in lesions over 400 um

HR-OCT, High resolution optical coherence tomography

2.3. Statistical analysis

IBM SPSS Statistics software package (version 25, IBM Corp., Armonk, NY, USA) was utilized for the statistical analysis. Descriptive statistics were used to summarize patient demographic and clinical information. Cases were split into three main groups (epithelial, sub-epithelial, and combined lesions) and epithelial thickness was compared between the groups using Analysis of variance (ANOVA) methodology followed by Fisher’s LSD post-hoc test to examine group wise comparisons. Receiver Operating Characteristic (ROC) analysis was used to examine cut-offs for epithelial thickness in categorizing a lesion as epithelial. P values less than 0.05 were considered statistically significant.

3. Results:

Twenty-two individuals with a clinically ambiguous lesion as determined by an ocular surface specialist (CLK) with availability of slit lamp photography, HR-OCT and histopathologic examination were identified. Mean age of the patient population was 53 ±20 years, 59% of patients were male, 86% identified as white (non-Hispanic 79%, Hispanic 21%) and 14% as black.

Our study included 7 purely epithelial lesions, 12 subepithelial lesions and 3 lesions with an epithelial and a subepithelial component. Fifteen of the lesions were confirmed to be malignant by histopathological examination while 7 were confirmed to be benign. When evaluating the HR-OCT scans, there was 100% agreement between the two observers. We found that HR-OCT was most effective in assisting the clinician discern the location of the lesion (epithelial, subepithelial or combined), in 100% of cases. We were able to detect that the lesion was limited to the epithelium in all epithelial cases. Similarly, we were able to identify involvement of the epithelium as well as the presence of a subepithelial mass in all cases which had an epithelial and a subepithelial component. Specifically, the HR-OCT allowed the clinician to correctly identify the lesion based on classic features in 68.2% of the cases (n=15) and suggested neoplasia in the remaining 31.8% (n=7), based on the criteria outlined in Table 1. Interestingly, 12 out of the 22 cases (54.5%) were initially diagnosed as benign but were ultimately diagnosed as malignant on histopathology. Of the 22 referred with a benign diagnosis, 7 of the 12 lesions (58.3%) exhibited classic features of malignancy on HR-OCT images while 5 (41.6%) exhibited features suggestive of malignancy.

The mean epithelial thickness of the 7 purely epithelial lesions was 265.4±140.6 μm, the mean of the 12 subepithelial lesions was 58.0±25.0 μm, and the 3 combined epithelial and subepithelial lesions mean thickness was 140.0±70.0 μm. The epithelium was significantly thicker in epithelial lesions compared to sub-epithelial and combined lesions (p<0.0005). When examining the data by ROC analysis (using multiple OCT machines (Optovue Avanti (Fremont, CA), the Optovue RT Vue (Fremont, CA) and the custom built UHR OCT), we identified that using a cut off of 156 μm, the sensitivity was 86% and the specificity was 93%. When the goal is to maximize sensitivity, using a cut off of 72 μm, the sensitivity of HR-OCT for detecting a lesion as epithelial was 100% with a specificity of 60% (Figure 8).

Figure 8:

Receiver operating characteristics (ROC) curve and corresponding area under the curve (AUC) for prediction of a lesion as epithelial based on epithelial thickness. Area under the ROC curve=0.914.

3.3. Epithelial lesions:

Classic features were more frequently identified in epithelial lesions compared to subepithelial and combined lesions. Specifically, in epithelial lesions, classic HR-OCT features were noted in 6 of 7 cases. In two of the cases, pigmented lesions were referred for conjunctival malignant melanoma (CMM) but were diagnosed as pigmented OSSN based on HR-OCT images (case 5, case 6, table 2). One case of intraepithelial sebaceous carcinoma had suggestive but not classic features of malignancy by HR-OCT (case 7, table 2). This ambiguous case was of a 56-year-old female who was referred for management of her severe dry eye following external beam radiation for childhood retinoblastoma. She presented with gelatinous changes on the conjunctiva, severe meibomian gland disease, madarosis and trichiasis. It was uncertain if the changes were secondary to chronic inflammation, dry eye, the prior radiation, or new neoplastic changes. HR-OCT images revealed hyperreflective and thickened epithelium; however, a clear transition point was not seen. In addition, diffuse stromal granularity was noted potentially due to diffuse inflammatory changes and extensive posterior shadowing limited deeper visualization. Ultimately, these findings suggested possible malignancy and guided appropriate location for biopsy, and ultimate diagnosis of sebaceous carcinoma.

Table 2.

Patient demographics, risk factors and patient initial and final lesion diagnosis

Case	Age	Sex	Race	Risk Factors	Initial diagnosis	Final diagnosis	HR-OCT
1	60	F	W	Cervical dysplasia	Episcleritis	OSSN (CIN:CIS)	Classic
2	26	M	W	Chronic inflammation	Scarring	OSSN (CIN:CIS)	Classic
3	73	M	w	UV exposure (BCC/SCC skin)	Pannus	OSSN (CIN:CIS)	Classic
4	65	F	W	UV exposure (SCC/BCC/melanoma skin), cervicaldysplasia	Scleritis	OSSN (CIN)	Classic
5	65	M	W	Occupational UV exposure	CMM	PigmentedOSSN (CIN)	Classic
6	78	M	W	UV exposure	CMM	Pigmented OSSN (CIN)	Classic
7	56	F	W	Radiation, history of melanoma, chronic blepharitis	Chronic inflammation	Sebaceous carcinoma	Suggestive
8	66	M	B	Immunosuppressant medications	Scleritis	OSSN (invasive SCC)	Classic
9	68	F	W	UV exposure	Scleritis	OSSN (invasive SCC)	Suggestive
10	66	M	W	Occupational UV exposure, genital warts	Episcleritis	OSSN (invasive SCC)	Suggestive
11	57	M	W	UV exposure	Episcleritis	Amelanotic CMM	Suggestive
12	63	M	W	UV exposure	OSSN	Amelanotic CMM	Classic
13	35	M	W	Occupational radiation exposure	Nevus	CMM	Suggestive
14	58	F	W	UV exposure	CMM	Conjunctival nevus & PAM	Classic
15	8	M	W	None	Conjunctival lesion	Conjunctival nevus	Classic
16	37	F	W	History of retinal detachment, contact lens wear	CMM	Mycetoma	Suggestive
17	27	F	B	HIV+	Sarcoid	Lymphoma	Classic
18	74	M	W	None	Chronic conjunctivitis	Lymphoma	Classic
19	28	F	W	None	Episcleritis	Amyloid	Classic
20	64	M	W	None	Lymphoma	Amyloid	Suggestive
21	18	F	B	MEN	OSSN	Neuroma	Classic
22	69	M	W	UV exposure (BCC skin), cutaneous warts	OSSN	Salzmann’s nodular degeneration	Classic

F, Female; M, Male; W, White; B, Black; HR-OCT, High resolution-optical coherence tomography; CIN, Conjunctival/corneal intraepithelial neoplasia; CIS, Carcinoma in situ; SCC, Squamous cell carcinoma; BCC, Basal cell carcinoma; OSSN, Ocular surface squamous neoplasia; CMM, Conjunctival malignant melanoma; PAM, Primary acquired melanosis; UV, Ultraviolet; HIV, Human immunodeficiency virus; MEN, Multiple endocrine neoplasia

3.2. Subepithelial lesions:

Classic features by HR-OCT were noted in 8 of 12 sub-epithelial lesions and suggestive features were seen in the remaining 4 cases. Certain entities such as lymphoma (n=2), Salzman’s nodular degeneration (n=1), and neuroma (n=1) displayed classic features as described in Table 1 which allowed for easy identification of the lesions. One case referred for OSSN was noted to have a subepithelial mass on HR-OCT images with a hyperreflective epithelium of essentially normal thickness (case 12, table 2). This finding ruled out an epithelial lesion (OSSN) and correctly identified an amelanotic CMM which was confirmed by histopathology (See figure 4). However, in 2 cases, the HR-OCT images displayed features which were not classic, but rather suggestive of a melanocytic malignancy. For example, one patient presented with a history of a long-standing amelanotic conjunctival lesion with recent enlargement (case 13-figure 5, table 2) and HR-OCT images revealed a hyperreflective subepithelial mass overlying a subepithelial mass with cysts. OSSN was ruled out and the hyperreflective sub-epithelial lesion was suggestive of a nevus or malignancy but a definitive diagnosis could not be made based on HR-OCT images alone. Excisional biopsy was performed and histology ultimately diagnosed CMM in the setting of a nevus.

Figure 4: Amelanotic CMM masquerading as OSSN/pannus.

a. Slit lamp photography of the left eye demonstrating gelatinous lesion encroaching on the cornea from 3:30 to 6 o’clock associated with hairpin loop vessels and trace pigment. b. High resolution optical coherence tomography (HR-OCT) demonstrates a subepithelial hyperreflective mass (white star) with an overlying hyperreflective and essentially normal thickness epithelium (white arrows). Posterior shadowing is seen. Inset shows scan location. c. Atypical melanocytic cells are present within the substantia propria (star). Tumor cells are located within the basal aspect of the epithelium (arrowheads) with minimal overlying acanthosis (between arrows). (Hematoxylin-eosin, original magnification x 200).

Figure 5: Conjunctival malignant melanoma (CMM) masquerading within a longstanding pinguecula/pterygium.

a. Slit lamp photography of the left eye of a 35-year-old white male shows a gelatinous lesion on the conjunctiva and encroaching on the cornea from 2 to 4 clock hours accompanied by a feeder vessel and with a pigmented edge on the corneal margin. Melanosis inferiorly to the lesion was seen. b. High resolution optical coherence tomography (HR-OCT) shows a subepithelial hyporeflective mass (asterisk) with some associated cysts (thin arrow), most likely representing long-standing nevus. Adjacent to this lesion, there is a hyperreflective subepithelial mass (thick arrowheads) and hyperreflective thin epithelium, suggestive of conjunctival melanoma. Inset shows location of the scan. c. A large nest of morphologically benign nevus cells is present within the substantia propria (asterisk). Atypical pleomorphic melanoma cells are present within the epithelium and superficial substantia propria (arrows). (Hematoxylin-eosin, original magnification x 100).

3.3. Combined epithelial and subepithelial lesions:

Classic features were identified in 1 of 3 combined epithelial and subepithelial lesions, which were all histologically diagnosed as invasive squamous cell carcinoma (SCC). Classic features of SCC were found in 1 patient with diffusely thickened, hyperreflective epithelium connected to a mass invading into the subepithelial space (case 8-figure 2, table 2). In the second case, a diffusely thickened, hyperreflective epithelium with an abrupt transition from normal to abnormal epithelium was noted in addition to a subepithelial mass in some cuts however no clear connection between the two was noted (case 10, table 2). In the third case, HR-OCT images displayed scattered areas of thickening in the epithelium as well as a subepithelial mass, however no clear connection between the two was identified (case 9-figure 3, table 2). In the two latter cases, findings were suggestive of neoplasia but not conclusive.

Figure 2: Squamous cell carcinoma (SCC) masquerading as infectious sclerokeratitis.

a. Slit lamp photograph of the right eye of a 66-year-old black male with diffuse conjunctival injection, midstromal corneal infiltrates & pigmented keratic precipitates (white arrow), 2+ cell in the anterior chamber, a temporal thickening of the conjunctiva (white arrowhead) b. High resolution optical coherence tomography (HR-OCT) reveals a thickened, hyperreflective epithelium (thin white arrows) with a mass invading into the subepithelial space (thick white arrow). Shadowing precludes clear visualization of the scleral tissue. c. Squamous cell carcinoma consisting of glands and cords of malignant epithelial cells located within the substantia propria (thick black arrow). The overlying epithelium contains a mild chronic inflammatory cell infiltrate and is minimally thickened (between small black arrows). (Hematoxylin-eosin, original magnification x 200).

Figure 3: Squamous cell carcinoma masquerading as sclerokeratitis.

a. Slit lamp photography of the left eye of a 68-year-old white Hispanic female with presumptive herpetic sclerokeratitis. Image shows diffuse corneal opacification, injected feeder vessels, lipid keratopathy, and a gelatinous area from 1 to 3 o’clock. Black arrow shows location of high resolution optical coherence tomography (HR-OCT) scan. b. HR-OCT reveals a thickened hyper reflective epithelium (asterisk) along with a diffuse, subepithelial mass (white arrows) without an obvious connection. These features were suggestive of neoplasia. c. Squamous cell carcinoma with foci of dyskeratosis (black arrows) are present within the substantia propria. The overlying epithelium (asterisk) appears unremarkable (hematoxylin-eosin, original magnification x 200).

Below, we illustrate eight cases to illustrate the role of HR-OCT in diagnosing clinically ambiguous lesions.

3.4. Cases:

Case 1: OSSN masquerading as chronic episcleritis

A 60-year-old white Hispanic female was referred to our clinic for episcleritis of the right eye. She reported progressive right eye discomfort for about 10 months. The patient was seen by an outside optometrist one month prior and was treated with a 2-week course of topical difluprednate 4 times daily, however no improvement was noted. The patient had a history of hysterectomy for cervical dysplasia but no history of skin cancers. Best corrected visual acuity of the right eye was 20/30. Slit lamp examination revealed localized hyperemia (Figure 1a). HR-OCT was performed, revealing a thickened, hyperreflective epithelium with an abrupt transition, classic findings of OSSN (Figure 1b) which led the physician to the diagnosis of OSSN. Histopathologic analysis of an incisional biopsy confirmed conjunctival intraepithelial neoplasia (Figure 1e). She was treated with 4 one-week cycles of 5 fluorouracil (5 FU) 1% with excellent clinical resolution of her tumor (Figure 1c). A HR-OCT post 5 FU treatment revealed normalized conjunctival epithelium (Figure 1d). Post treatment map biopsies confirmed the HR-OCT findings of tumor resolution at all sites.

Figure 1: Ocular surface squamous neoplasia (OSSN) masquerading as chronic episcleritis.

a. Slit lamp photography of the right eye of a 61-year-old white Hispanic female referred for a 10-month history of episcleritis (white arrow). b. High resolution optical coherence tomography (HR-OCT) demonstrates hyperreflective thickening of the epithelium (asterisk) classic for OSSN. Inset shows scan location. Biopsy confirmed the HR-OCT diagnosis of conjunctival intraepithelial neoplasia, carcinoma in situ. c. Slit lamp photography shows dramatic clinical resolution of the lesion after treatment with 4 one-week cycles of 5-fluorouracil 1% eye drops. d. HR-OCT reveals normalization of the epithelium (white arrows). Inset shows scan location. Post treatment biopsies confirmed the HR-OCT findings of tumor resolution. e. Thickened epithelium (between arrows) that demonstrates faulty maturational sequencing which extends up to full thickness is present. (Hematoxylin and eosin, original magnification x 100).

Case 8: SCC masquerading as infectious sclerokeratitis

A 66-year-old black male was referred for infectious sclerokeratitis of the right eye. The patient complained of a seven-day history of pain, conjunctival erythema and almost complete loss of vision in the right eye. Past medical history was notable for renal transplant and immunosuppressant medication. Past ocular history was significant for pterygium excision in the right eye 25 years prior and subsequent “re-growth of the pterygium” associated with a gradual decrease in visual acuity. Slit lamp examination of the right eye demonstrated diffuse conjunctival injection, a subepithelial nodule located temporally at the limbus, midstromal infiltrate with feathery borders, pigmented keratic precipitates, 2+ cell in the anterior chamber and iris nodules (Figure 2a). Intraocular pressure (IOP) was noted to be elevated to 38 mm Hg.

The differential for the presentation included infectious sclerokeratitis (herpetic, microbial, fungal), pterygium recurrence, or neoplasm. A culture and anterior chamber tap were negative for organisms. B-scan ultrasonography demonstrated thickened conjunctiva, however no scleritis or vitreous membranes were identified. A HR-OCT was performed which revealed a thickened, hyperreflective epithelium with a mass invading into the subepithelial space (Figure 2b), findings classic of SCC. Map biopsies confirmed mucoepidermoid variant of SCC with extension to multiple margins was identified on histopathologic analysis. The patient ultimately underwent enucleation, and SCC was found in the anterior chamber, iris, and angle. (Figure 2c)

Case 9: SCC masquerading as sclerokeratitis

A 68-year-old white Hispanic female was referred to the clinic for longstanding (~2 years) herpetic sclerokeratitis of the left eye that was unresponsive to topical prednisolone and oral acyclovir therapy. The patient endorsed epiphora, photophobia, pain and conjunctival redness. Best corrected visual acuity of the left eye was 20/400. On slit lamp examination, the clinical picture revealed diffuse corneal haze and thickening, large feeder vessels, and a low gelatinous area from 1 to 3 o’clock at the limbus. Within the haze, white infiltrates were present thought to be lipid keratopathy. The adjacent sclera was red and painful to the touch (Figure 3a). B scan ultrasound was negative for vitreous membranes and no microbial growth was identified on culture. HR-OCT demonstrated slightly thickened slightly hyper reflective epithelium (Figure 3b star) along with a subepithelial lesion (Figure 3b arrows). However, no obvious connection between the epithelium and the subepithelial mass could be identified. These HR-OCT findings were not definitive, but rather suggestive of a mass from squamous cell carcinoma, and this prompted an incisional biopsy. Histology revealed subepithelial nests of atypical squamous cells (Figure 3c) leading to a final diagnosis of invasive SCC.

Case 12: Amelanotic CMM masquerading as OSSN/pannus

A 63-year-old white male was referred for a conjunctival lesion in the left eye suspicious for pannus or OSSN which was present for approximately 2 years. The patient reported no history of skin cancers; however, he reported a history of significant sun exposure. On physical examination there was no evidence of enlarged preauricular, submandibular or supraclavicular lymph nodes. Best corrected visual acuity of the left eye was 20/25. On slit lamp examination, a gelatinous lesion was noted encroaching onto the cornea from 3:30 to 6 o’clock, associated hairpin loop vessels, and tiny traces of pigment at the leading edge of the lesion and lightly scattered along the bulbar conjunctiva (Figure 4a). HR-OCT was performed to better evaluate the lesion, which revealed a subepithelial, hyperreflective mass inconsistent with OSSN. The HR-OCT findings were not consistent with OSSN and instead were classic for conjunctival melanoma. Thus, wide surgical excision with cryotherapy was advised. The histopathological analysis revealed a conjunctival malignant melanoma arising from PAM with severe atypia (Figure 4c). Six years later the patient died of metastatic disease of an uncertain primary.

Case 13: CMM masquerading as longstanding pinguecula/pterygium

A 35-year-old white male presented with a conjunctival lesion of the left eye. The patient stated that the lesion had been present for 20 years and thought to be a pinguecula or pterygium; however, he noticed an increase in the size within the last 3 years. He denied significant sun exposure and history of cutaneous malignancies but reported occupational radiation exposure. On slit lamp examination, an elevated gelatinous conjunctival lesion with a feeder vessel was seen temporally on the conjunctiva. The lesion encroached the cornea from 2 to 4 clock hours with a faint pigmented edge. Melanosis was also seen on the conjunctivae inferior to the lesion and on the inferior lid margin (Figure 5a). HR-OCT demonstrated one area of the mass had subepithelial cysts in some of the cuts, suggestive of a long-standing conjunctival nevus, but adjacent to this was a hyperreflective subepithelial mass with no evident cysts. The overlying epithelium was hyperreflective and thin. These finding were suggestive of possible malignant transformation of a chronic nevus (Figure 5b). Wide surgical excision with cryotherapy and intraoperative mitomycin was advised and tissue analysis demonstrated conjunctival melanoma in-situ in association with a subepithelial compound nevus (Figure 5c). In retrospect, the longstanding lesion on his eye was likely an amelanotic nevus, which transformed into a conjunctival melanoma.

Case 17: Lymphoma masquerading as sarcoidosis

A 23-year-old black female patient presented with a conjunctival lesion of the right eye. The patient reported that the lesion was rapidly enlarging over the last month. On slit lamp examination, a gelatinous mass was located on the bulbar surface and the inferior bulbar conjunctiva on the right eye (Figure 6a). The top differential was sarcoidosis, but also included pseudotumor, IgG-4 disease, lymphoma, and infection. HR-OCT identified a homogenous, hyporeflective, subepithelial mass with monomorphic dot-like infiltrates bordered by a band of hyperreflectivity superiorly and a thin overlying epithelium (Figure 6b), findings classic for conjunctival lymphoma. An incisional biopsy was performed, and histopathologic analysis demonstrated high grade non-germinal center large B-cell lymphoma of the conjunctiva extending to the orbit (Figure 6c). An MRI of the orbits reported an enhancing ring of soft tissue surrounding the right orbit which was inseparable from the globe and demonstrated mild postseptal extension. Upon sharing the diagnosis, the patient disclosed that she was HIV positive, information she had withheld. The patient had CT-PET scan for staging and was found to be stage IV.

Figure 6: Lymphoma masquerading as sarcoidosis.

a. Slit lamp photography of the right eye of a 23-year-old black female shows diffuse conjunctival thickening and a whitish nodular mass located on the inferior bulbar conjunctiva. b. High resolution optical coherence tomography (HR-OCT) reveals a subepithelial, homogenous, hyporeflective subepithelial mass (asterisk) with monomorphic dot-like infiltrates bordered by a band of hyperreflectivity superiorly (thin arrows) and a thin epithelium. Extensive posterior shadowing is noted. Inset shows scan location. c. A diffuse infiltrate of atypical large basophilic lymphoma cells is present within the substantia propria (arrows). (Hematoxylin-eosin, original magnification x 40).

Case 19: Conjunctival amyloidosis masquerading as episcleritis

A 28-year-old white female presented with a one-month history of irritation and conjunctival injection in the right eye. The patient reported no prior medical history. Topical corticosteroids 4 times a day were started by an outside physician, and the injection improved. She was then told by an outside physician that she had conjunctival “cancer”. On slit lamp examination, an area of localized hyperemia was noted in the inferior bulbar conjunctiva (Figure 7a). HR-OCT allowed detection of a normal, thin epithelium overlying heterogeneous dark subepithelial lesion with irregular borders. Linear subepithelial, hyperreflective infiltrates were identified (Figure 7b), findings classic for conjunctival amyloidosis. To confirm the diagnosis, excisional biopsy and cryotherapy was performed. The histopathological specimen stained with Congo red and demonstrated apple green birefringence (Figure 7c), confirming the diagnosis of conjunctival amyloidosis. A work-up did not identify systemic causes of amyloidosis and thus a diagnosis of primary localized amyloidosis was established.

Figure 7: Conjunctival amyloidosis masquerading as episcleritis.

a. Slit lamp photography of the right eye of a 28-year-old white female shows a hyperemic, elevated area on the inferior bulbar conjunctiva (arrowheads). b. High resolution optical coherence tomography (HR-OCT) reveals thin epithelium overlying a subepithelial heterogeneous, dark lesion with irregular borders. Linear subepithelial, hyperreflective infiltrates were present in the mass(white arrows). Inset shows scan location. c. Congo red positive amyloid is present within the substantia propria (asterisk), (Congo red; original magnification x 100). d. Apple green birefringence is present within the substantia propria (asterisk), (Congo red; original magnification x 400).

4. Discussion:

While many conjunctival lesions can be diagnosed clinically through identification of classic clinical features, the diagnosis can be challenging in lesions that do not exhibit characteristic findings or in eyes with multiple ocular surface pathologies. HR-OCT is a valuable diagnostic tool that can assist to guide in the differentiation of ocular surface lesions by providing detailed cross-sectional images with morphological patterns. [3, 13] In this paper, we focused on clinically ambiguous lesions and assessed the effectiveness of HR-OCT as an adjunctive tool in guiding the diagnosis and management of 22 individuals.

In this study, we included only lesions without a clear diagnosis on slit-lamp examination. Our focus was the HR-OCT images evaluation and how it guided the diagnosis. The slit lamps photographs and HR-OCT images were available to the clinician (CLK) and then assessed by another observer who was masked to the clinical history. When evaluating the HR-OCT scans, there was 100% agreement between the two observers. In a future study, it would indeed be interesting to have clinicians evaluate the clinical images and the OCTs independently and assess their responses.

We found that HR-OCT was most helpful in identifying the lesion location, correctly identifying lesions as epithelial, subepithelial, or combined lesions in 100% of cases. Overall, in these challenging and clinically ambiguous cases, the HR-OCT images guided the clinician to correctly identify the lesion in 68.2% of the cases (n=15) based on classic findings and suggested the diagnosis in the remaining 31.8% (n=7). By ROC analysis, a cut-off of 156 μm using multiple OCT machines (Optovue Avanti (Fremont, CA), the Optovue RT Vue (Fremont, CA) and the custom built UHR OCT) provided an 86% sensitivity and 93% specificity in identifying a lesion as epithelial. This number is comparable to our and other previously published data. [9, 13, 21] Using the custom built UHR-OCT, a cut-off value of 142μm provided a sensitivity of 94% and specificity of 100% in differentiating OSSN (an epithelial lesion) from pterygium (a sub-epithelial lesion). [13] In another series using a spectral domain OCT (RTVue, Optovue, Fremont, CA), a cutoff value of 120 μm provided a sensitivity of 100% and specificity of 100% in differentiating OSSN from pterygium. [9] In another study, using a cutoff value of 141 μm (Spectralis SD-OCT, Heidelberg, Germany) provided a 100% sensitivity and specificity in differentiating cornea-conjunctival intraepithelial neoplasia from pterygium. [21]

In addition, classic features were most often detected in epithelial lesions, and were found in 6 of 6 OSSN cases. In a case of sebaceous carcinoma images displayed features suggestive of malignancy as the epithelium appeared abnormal however there was no abrupt transition of the normal from the abnormal epithelium. Classic features were found in 66.6% of subepithelial tumors. In particular, certain lesions including lymphoma, Salzmann nodular degeneration, and neuroma had classic features in all cases. Other lesions with a subepithelial mass including amyloid, SCC and CMM were less definitive and more likely to display only suggestive features on HR-OCT images. In these cases, posterior shadowing limited full visualization of the lesions; however, HR-OCT findings still guided the clinician towards the right diagnosis.

HR-OCT has been well studied in relationship to OSSN [3, 8, 10, 13, 15, 22], which can range from dysplasia to invasive SCC and can have varied appearances.[23] Given this spectrum, OSSN has been reported to masquerade as chronic blepharoconjunctivitis [24], superior limbic keratoconjunctivitis [4], corneal ulcer [25], pterygium [26], sclerokeratitis [27] and necrotizing scleritis [28], among others. In this study, the clinician was able to detect OSSN based on classic HR-OCT features in all six lesions referred for episcleritis (n=1), scleritis (n=1), pannus (n=1), scarring (n=1) or CMM (n=2).

Invasive SCC lies on the spectrum of OSSN and is comprised of an epithelial tumor connected to a mass invading the subepithelial space. In all 3 SCC cases, HR-OCT images demonstrated a subepithelial mass, but classic epithelial findings as well as a connection of the epithelial tumor to the mass were only seen in 1 of 3 of the cases. Nevertheless, identification of a mass in the two cases led to suspicion of a malignancy and subsequent biopsy (case 9-figure 3, case 10, table 2). Interestingly, epithelial findings in SCC are sometimes not as classic as in non-invasive OSSN, with variable hyper-reflectivity and often an absence of an abrupt transition. [29] In addition, a connection between the abnormal hyperreflective epithelium and the subepithelial mass cannot consistently be located. In some locations of the tumor, the epithelium may be normal, as the tumor spreads subconjunctivally. Further challenges in diagnosing SCC on HR-OCT include the presence of posterior shadowing and stromal granularity that can limit visualization of the lesion. Last, HR-OCT cannot detect invasion in subtle cases and may miss lesions in tumors which are not imaged in their entirety. These findings highlight the importance of obtaining and examining multiple images around the tumor in a standardized fashion to ensure that all lesion characteristics are adequately visualized. In this way, false negative imaging can be minimized, and diagnostic accuracy can be increased. In addition, repeatability between studies can be ensured.

Conjunctival lymphoma and conjunctival amyloidosis are ocular surface conditions that may have a very similar clinical presentation classically consisting of a focal salmon patch or yellow, pink or red waxy lesions. [17] However, at times their clinical presentations can be atypical, posing a diagnostic challenge. Amyloidosis commonly masquerades as another lesion as demonstrated by a paper which reported a diagnosis of amyloidosis after biopsy in 6 individuals referred for lymphoma (n=2), allergic conjunctivitis (n=1), lacrimal gland tumor (n=1) and subconjunctival hemorrhage (n=2). [30] In a similar manner, conjunctival lymphoma can also masquerade as another lesions, and is often initially diagnosed as a bilateral chronic conjunctivitis [6, 31, 32] and scleritis [33]. Fortunately, their appearance on HR-OCT is usually classic and, in our study, these findings aided in the diagnosis of amyloid in two cases and lymphoma in two cases.

Conjunctival melanoma is a highly aggressive and potentially life-threating tumor. [18] Interestingly, about 20% of conjunctival malignant melanomas are amelanotic, and this presentation can be confused with OSSN, lymphoma, or a benign process such as episcleritis and pyogenic granuloma. [2, 10] Misdiagnosis can lead to a delay in treatment, with local spread or distant metastasis, or to an incorrect treatment, as incisional biopsy can lead to tumor seeding. [18] In our study, HR-OCT images displayed classic findings of CMM in 1 case of an amelanotic conjunctival lesion and suggestive findings of CMM in a patient with an amelanotic conjunctival lesion and another with a long-standing nevus. In nevus, the presence of cysts within the lesion helped the clinician identify nevi in two of our patients. As always, HR-OCT findings must be interpreted considering the clinical scenario. For example, PAM (Primary-associated melanosis) and CAM (Complexion-associated melanosis) have similar appearances on HR-OCT. However, PAM most often occurs unilaterally in individuals with fair complexion while CAM is often seen bilaterally in pigmented individuals.

Our findings are particularly important as correctly identifying an ambiguous lesion directly affects therapy. In our study, management was changed in 10 of 15 lesions that had classic findings for a specific pathology on HR-OCT images. This included 8 lesions initially suspected to be benign but had classic OCT features of OSSN (n=4), invasive SCC (n=1), CMM (n=1), and lymphoma (n=2). Conversely, two lesions suspected to be CMM had classic features of OSSN on HR-OCT and thus topical chemotherapy with 5-fluorouracil (5-FU) was initiated with complete resolution of the lesions. In lesions with suggestive findings of a specific pathology (n=7), HR-OCT images helped guide biopsy and subsequent management. This subgroup included 4 cases which were initially thought to be benign but the HR-OCT had features suggestive of invasive SCC (n=2) and CMM (n=2). For the SCC cases, it guided biopsy location, and for the CMM, led to immediate surgical excision.

Nevertheless, even though HR-OCT images can provide valuable clues and can guide the physician reach a diagnosis, it certainly does not eliminate the need for a biopsy and ultimately histopathological diagnosis. In many cases, HR-CT images can guide the biopsy location ensuring a good specimen is sent for analysis, especially if there is a suspicion of a false negative biopsy. More recently, OCTA has emerged as a new tool allowing for characterization and quantification of vascular networks of benign and malignant ocular surface lesions. [34, 35] Hopefully, in the future HR-OCT along with OCTA will assist the physician in further evaluating ocular surface lesions.

As with all studies, our findings need to be considered in light of the study limitations which included a retrospective design, a select patient population, and the use of various HR-OCT devices. In addition, technological challenges of HR-OCT include low penetration depth, posterior shadowing of thick lesions, inability to evaluate tissue on a cellular level, and difficulty in interpreting images from the fornix and caruncle. [3, 36] A question remains on the diagnostic accuracy of this technology when image interpretation is performed by less experienced physicians. It has been previously reported that the mean frequency of correctly identifying a lesion was 70% in a group of 34 novice clinicians at the Bascom Palmer Eye Institute. This was improved to 84% after a short 20-minute lecture. [37] Despite the above-mentioned limitations, HR-OCT technology is non-invasive, does not require technical expertise for scanning, and image interpretation can be performed by clinicians of all levels after appropriate training.

To conclude, this paper highlights the utility of HR-OCT images in providing valuable clues and guiding management of clinically ambiguous lesions. The HR-OCT findings assisted the clinician in identifying or suspecting the lesion location in all cases. In 68.2% of cases, the findings were classic and led to the diagnosis even prior to biopsy. In the remaining 31.8% of cases, the HR-OCT images were suggestive of the diagnosis, and the scans directed the location of biopsy. Ultimately, HR-OCT seems to be a technology that helps the physician as well as the patient. Future, larger studies in diverse patient populations using various imaging devices are needed to further validate our findings.