Iris melanoma

Iris melanoma

Iris Melanoma Risk Factors for Metastasis in 169 Consecutive Patients Carol L. Shields, MD,1 Jerry A. Shields, MD,1 Miguel Materin, MD,2 Eric Gershenb...

838KB Sizes 39 Downloads 140 Views

Iris Melanoma Risk Factors for Metastasis in 169 Consecutive Patients Carol L. Shields, MD,1 Jerry A. Shields, MD,1 Miguel Materin, MD,2 Eric Gershenbaum, MD,1 Arun D. Singh, MD,1 Andrew Smith, PhD3 Objective: To identify risk factors that predict distant metastases of iris malignant melanoma. Design: Retrospective case series. Participants: The participants included 169 consecutive patients with microscopically confirmed iris malignant melanoma managed on the Oncology Service at Wills Eye Hospital between 1974 and 1999. Main Outcome Measures: The main outcome measure was the development of distant tumor metastasis. Cox proportional regression models were used to calculate the risk of eventual metastatic spread. Results: Of 1054 patients referred with suspicious iris melanocytic tumors (rule out malignant melanoma) over a 25-year period, 169 patients (16%) had microscopically proven iris melanoma, and the remainder (84%) had clinically diagnosed iris nevus. Of the patients with iris melanoma, the mean age at the time of diagnosis was 43 years (median, 45 years; range, 1–90 years). All patients were Caucasian. The mean tumor base was 6 mm (median, 5 mm; range, 1–17 mm), and mean tumor thickness was 2 mm (median, 2 mm; range, 1– 4 mm). The mean number of clock hours of tumor involvement in the iris was four, tumor seeding on the iris was four, and tumor seeding into the anterior chamber angle was four. Extraocular extension was present in 10 eyes (6%). The tumor management consisted of local resection (iridectomy, iridocyclectomy, or iridocyclogoniectomy) in 102 patients (60%), enucleation in 51 (30%), plaque radiotherapy in 9 (5%), and observation in 7 patients (4%). Metastasis developed in nine patients (5%). Using Kaplan-Meier life table analysis, metastasis was found in 3% of patients at 5 years, 5% at 10 years, and 10% at 20 years. The clinical factors at initial evaluation predictive of eventual metastasis from iris melanoma included increasing age at diagnosis (P ⫽ 0.03), elevated intraocular pressure (P ⫽ 0.03), posterior tumor margin at angle or iris root (versus midzone) (P ⫽ 0.02), extraocular extension (P ⫽ 0.02), and prior surgical treatment of the tumor elsewhere before referral (versus observation) (P ⫽ 0.006). The method of management (resection, radiotherapy, or enucleation) did not have an impact on metastasis. Conclusions: Microscopically confirmed iris melanoma demonstrates distant metastasis in 5% of patients at 10 years follow-up. Metastases are more likely to develop in those patients who are older and show tumor features of iris root/angle location with elevated intraocular pressure and extraocular extension. Ophthalmology 2001;108:172–178 © 2001 by the American Academy of Ophthalmology. Uveal melanoma represents the most common primary intraocular malignancy.1,2 It involves the posterior segment of the eye (choroid and ciliary body) far more often than the anterior segment (iris). Jensen,3 in his review of malignant melanoma of the uvea in Denmark, found that 97% of all uveal melanoma involved the choroid and ciliary body, Originally received: February 11, 2000. Accepted: August 7, 2000. Manuscript no. 200071. 1 Oncology Service, Wills Eye Hospital, Thomas Jefferson University, Philadelphia, Pennsylvania. 2 Consultores Oftalmologicos, Buenos Aires, Argentina. 3 Medical Economics and Epidemiology Unit, Wills Eye Hospital, Thomas Jefferson University, Philadelphia, Pennsylvania. Presented in part as a poster at the annual meeting of the American Academy of Ophthalmology, Dallas, Texas, October 2000. Supported by the Paul Kayser International Award of Merit in Retina Research, Houston, Texas (J. Shields); Lions Eye Bank, Philadelphia, Pennsylvania; Macula Foundation, New York, New York (C. Shields); and the Eye Tumor Research Foundation, Philadelphia, Pennsylvania. Reprint requests to Carol L. Shields MD, Oncology Service, Wills Eye Hospital, 900 Walnut St., Philadelphia, PA 19107.

172

© 2001 by the American Academy of Ophthalmology Published by Elsevier Science Inc.

whereas only 3% involved the iris. Others have confirmed the relatively low incidence of iris melanoma.4,5 In our series of 8800 consecutive patients with uveal melanoma, 8631 (98%) originated in the choroid or ciliary body, whereas only 169 (2%) originated in the iris. Most prior reports on iris melanoma have focused on the clinical features,6 –9 differential diagnosis,10 management,11–15 and pathologic findings.4,16 –18 Few clinicians have managed enough patients with iris melanoma to make reliable observations on the incidence of metastasis.4,19,20 In 1958, Rones and Zimmerman4 reported information from the files of the Armed Forces Institute of Pathology and found that only 6% of 72 cases of iris melanoma treated with iridectomy died of metastasis by 10-year follow-up. In 1980, Sunba and coworkers21 found that 3.5% of patients with iris melanoma documented in the pathology department at the Institute of Ophthalmology in London had metastatic foci develop at 5 years. Later, Geisse and Robertson22 provided a review of the literature on iris melanoma and confirmed the low incidence of metastasis at 3%. Despite these informative reports, there has been no statistiISSN 0161-6420/00/$–see front matter PII S0161-6420(00)00449-8

Shields et al 䡠 Iris Melanoma Metastasis cally analyzed publication on the clinical features predictive of metastasis from iris melanoma. This is partly due to the rarity of this tumor and partly because many of the previous studies emanated from pathology laboratories with little clinical information available. The purpose of our study was to evaluate our comprehensive 25-year clinical experience with iris melanoma with regard to the incidence of metastasis using Kaplan-Meier analysis and report the clinical factors predictive of metastasis using Cox proportional hazards regressions.

ables were analyzed as discrete variables except for patient age, intraocular pressure, tumor base, tumor thickness, and number of clock hours affected by tumor or seeds, which were analyzed as continuous variables. The variables that were significant on a univariable level (P ⬍ 0.05) were entered into a stepwise regression analysis. For variables that showed a high degree of correlation, only one variable from the set of associated variables was entered at a time in subsequent multivariate models. A final multivariable model fitted variables that were identified as significant predictors (P ⬍ 0.05) in the stepwise model and variables deemed clinically important for the outcome of metastasis. Kaplan-Meier survival estimates were used to analyze the development of melanoma metastasis as a function of time.

Patients and Methods The clinical records of all patients with the diagnosis of iris malignant melanoma managed on the Ocular Oncology Service at Wills Eye Hospital, Philadelphia, Pennsylvania, between June 1974 and June 1999 were reviewed. Only those patients with microscopically confirmed (histopathologically and/or cytopathologically) melanoma that arose in the iris with or without minor extension to the adjacent structures were included in this study. Tumors that arose in the ciliary body with secondary iris extension were excluded. Clinical data were gathered regarding features of the patient, the eye, and the tumor, as well as clinical management and follow-up. The clinical data were then analyzed with regard to the outcome of tumor metastasis. The data gathered included patient features at initial examination such as age, race (African American, Hispanic, Asian, Caucasian), and gender (female, male). The ocular data included affected eye (right eye, left eye), best-corrected Snellen visual acuity, intraocular pressure (mmHg), iris color (blue, green, brown), and ocular melanocytosis (present, absent). The tumor data included predominant quadrant location (superior, nasal, inferior, temporal, diffuse), clock hour meridian (1 o’clock–12 o’clock), number of clock hours affected by tumor (1–12 hours), anteroposterior epicenter and margins (pupil, midzone, root, angle, diffuse), configuration (nodule, flat), basal dimensions (mm), thickness (mm), color (brown, yellow), and surface features (irregular, smooth). The associated features included clock hour extent of adjacent iris seeds (0 –12 hours), tumor satellite (present, absent), ectropion iridis (present, absent), corectopia (present, absent), episcleral sentinel vessel (present, absent), feeder iris vessel (present, absent), intrinsic iris vessels (present, absent), clock hour extent of anterior chamber angle seeds (present, absent), anterior chamber features (normal, hyphema, inflammation), secondary cataract (present, absent), and extraocular extension (present, absent). The tumor management before referral to us and after referral (observation, fine needle aspiration biopsy [FNAB], incisional biopsy, iridectomy, iridocyclectomy, iridogoniocyclectomy, plaque radiotherapy, enucleation) was assessed. The presence of previously documented growth was recorded. The microscopic results of melanoma cell type (spindle, mixed, epithelioid) and the number of mitoses per 40 high power fields were collected. Follow-up data were gathered regarding evidence of tumor recurrence (present, absent), systemic metastasis (present, absent), and death (present, absent).

Statistical Analysis The effect of each individual clinical variable on the development of tumor-related metastasis was analyzed by a series of univariate Cox proportional hazards regressions.23 The correlation among the variables was determined by using Pearson correlations. All vari-

Results Over the 25-year period of this study, there were 1054 patients referred to the Ocular Oncology Service with a suspicious iris melanocytic tumor (rule out malignant melanoma). Of these, 169 have been confirmed microscopically to be iris melanoma, and 885 are being managed with observation with a presumptive diagnosis of iris nevus. The mean age at initial diagnosis elsewhere was 43 years (median, 45 years; range, 1–90 years). All 169 patients were Caucasian. There were 85 (50%) men and 84 (50%) women. The referral diagnosis was iris melanoma in 57 patients (34%), nevus in 46 (27%), nonspecified tumor in 60 (36%), metastatic tumor in 2 (1%), cyst in 1 (⬍1%), and Cogan-Reese syndrome in 1 (⬍1%). Management of the iris tumor elsewhere was observation in 151 (89%) patients, incisional biopsy in 10 (6%), local resection in 7 (4%), and enucleation in 1 (⬍1%). Referral to the Ocular Oncology Service was for general management opinion in 92 (54%) and specific opinion regarding tumor growth in 46 patients (27%), associated hyphema in 10 (6%), and associated secondary glaucoma in 21 (12%). After referral to us, the mean patient age was 48 years (median, 52 years; range, 8 –90 years). The right eye was affected in 93 patients (55%) and the left eye in 76 (45%). The visual acuity was 20/20 to 20/40 in 145 cases (86%), 20/50 to 20/100 in 13 (8%), and 20/200 or worse in 11 (6%). The mean intraocular pressure was 20 mmHg (median, 18 mmHg; range, 9 – 65 mmHg). In 50 patients (30%) the intraocular pressure was 23 mmHg or greater, and in 29 patients (17%) the pressure was 30 mmHg or greater. The iris color was blue in 143 eyes (85%), green in 17 (10%), and brown in 9 (5%). Ocular melanocytosis was present in 3 eyes (2%). The epicenter of the tumor was located in the superior quadrant of the iris in 5 eyes (3%), nasal quadrant in 36 (21%), inferior quadrant in 81 (48%), and temporal quadrant in 31 (18%). In 16 cases (10%), the iris involvement was diffuse with relatively uniform involvement of all four quadrants. The clock-hour meridian of the tumor epicenter was 1:00 in 0 cases (0%), 2:00 in 0 (0%), 3:00 in 15 (9%), 4:00 in 21 (12%), 5:00 in 22 (13%), 6:00 in 38 (22%), 7:00 in 20 (12%), 8:00 in 16 (9%), 9:00 in 12 (7%), 10:00 in 3 (2%), 11:00 in 1 (1%), 12:00 in 4 (2%), and diffuse with relatively uniform involvement of all clock hours in 16 (10%) (Fig 1). Thus, the tumor epicenter was located at the horizontal meridian or below in 144 cases (85%). The number of clock hours involved by each tumor was 1 hour in 31 cases (18%), 2 hours in 63 cases (37%), 3 hours in 25 cases (15%), 4 hours in 14 cases (8%), 5 hours in 4 cases (2%), 6 hours in 7 cases (4%), 7 hours in 2 cases (1%), 8 hours in 3 cases (2%), 9 hours in 3 cases (2%), 10 hours in 1 case (1%), 11 hours in 0 cases (0%), and 12 hours in 16 cases (10%) (Fig 2). The mean number of clock hours affected by

173

Ophthalmology Volume 108, Number 1, January 2000 Table 1. Final Tumor Management in 169 Patients with Microscopically Documented Iris Melanoma Management

Number of Patients

(%)

Enucleation Local resection* Plaque radiotherapy Observation

51 102 9 7

(30) (60) (5) (4)

* Iridectomy, iridocyclectomy, or iridogoniocyclectomy.

Figure 1. Clock-hour location of tumor epicenter in 169 patients with microscopically proven iris melanoma.

each tumor was four (median, 2 hours; range, 1–12 hours). The anteroposterior tumor epicenter was at the pupillary margin in 3 eyes (2%), midzone of iris in 105 (62%), iris root in 24 (14%), anterior chamber angle in 6 (4%), and diffuse with relatively homogeneous involvement of all anteroposterior zones in 31 (18%). The tumor anterior margin was at the pupillary margin in 111 eyes (66%), midzone in 49 (29%), iris root in 8 (5%), and anterior chamber angle in 0 (0%). The posterior tumor margin was at the pupillary margin in 0 (0%), midzone in 19 (11%), iris root in 72 (43%), and anterior chamber angle in 77 (46%). The mean tumor base was 6 mm (median, 5 mm; range, 1–17 mm), and the mean tumor thickness was 2 mm (median, 2 mm; range, 1– 4 mm). The predominant tumor color was brown in 124 eyes (73%) and yellow in 45 eyes (27%). The tumor configuration was nodular in 121 cases (72%) and flat in 47 (28%). The tumor surface was irregular in 157 eyes (93%) and smooth in 12 (7%). Tumor seeds onto the adjacent iris were present in 94 eyes (56%) and involved 1 to 3 clock hours in 40 eyes (24%), 4 to 6 clock hours in 13 eyes (8%), 7 to 9 clock hours in 1 eye (1%), and 10 to 12 clock hours in 40 eyes (24%). The mean number of clock hours of tumor seeding onto the iris was four (median, 1; range, 0 –12). Secondary seeding led to a satellite nodule of melanoma separate from the primary site in 18 eyes (11%). Tumor seeding into the anterior chamber angle was present in 96 eyes (57%) and involved 1 to 3 clock hours in 24 eyes (14%), 4 to 6 clock hours in 25 eyes (15%), 7 to 9 clock hours in 10 eyes (6%), and 10 to 12 clock hours in 37 eyes (22%). The mean number of clock hours affected with angle seeding was four (median, 1 clock hour; range, 0 –12 clock hours. Extraocular extension of iris melanoma was detected in 10

Figure 2. Number of clock hours of iris involvement in 169 patients with microscopically proven iris melanoma.

174

eyes (6%), and in all of those cases there was iris root and anterior chamber angle involvement. Ectropion iridis was found in 75 eyes (44%), corectopia in 104 (62%), prominent episcleral vessels in 43 (25%), feeding iris vessels in 42 (25%), visible intrinsic tumor vessels in 72 (43%), hyphema in 16 (9%), anterior chamber inflammation in 7 (4%), and secondary cataract in 23 (14%). Ultrasound biomicroscopy was performed in 14 cases and revealed an echodense mass in 13 cases and an echolucent mass in 1 case. The final tumor management included observation (patient preference) in 7 eyes (4%), iridectomy in 50 (30%), iridocyclectomy in 28 (17%), iridogoniocyclectomy in 24 (14%), plaque radiotherapy in 9 (5%), and enucleation in 51 (30%) (Table 1). In those eyes managed with observation or plaque radiotherapy, microscopic confirmation of the tumor diagnosis had been obtained, usually with FNAB or incisional biopsy.24 FNAB was used in 21 cases, usually followed by enucleation with histopathologic confirmation. Only three cases had cytologic confirmation with FNAB alone. The microscopic diagnosis was iris melanoma with spindle cell type in 92 cases (55%), mixed cell type with predominant spindle cells in 23 (14%), mixed cell type in 40 (24%), mixed cell type with predominant epithelioid cells in 9 (5%), and epithelioid cell type in 4 (2%). The mean number of mitoses per 40 high power fields was 1 (median, 0; range, 0 –9 mitoses) in the 52 cases in which mitoses were recorded. The final visual acuity in the affected eye was 20/20 to 20/40 in 76 cases (45%), 20/50 to 20/100 in 24 (14%), and 20/200 to no light perception in 18 (11%). In 51 (30%) patients, enucleation was performed. Over the mean follow-up of 9.4 years (median, 8.4 years), distant melanoma metastases were detected in nine patients (5%). Metastasis was detected at a mean of 5.0 years (median, 3.8 years) after the diagnosis of iris melanoma was microscopically established. Death ensued a mean of 1.4 years (median, 0.6 years) after metastases were detected. Using Kaplan-Meier life table analysis, metastasis was found in 3% of all patients at 5 years, 5% at 10 years, and 10% at 20 years (Fig 3). The accuracy of the Kaplan-Meier estimate at 20 years is uncertain because of the small number of patients followed at 20 years or beyond. Using univariate analysis, the clinical factors at the time of initial examination predictive of subsequent metastasis from iris melanoma included increasing age at diagnosis (P ⫽ 0.02), elevated intraocular pressure (P ⫽ 0.03), tumor margin at angle or iris root (versus midzone) (P ⫽ 0.02), extraocular extension (P ⫽ 0.02), and prior surgical intervention elsewhere before referral (versus observation) (P ⫽ 0.006) (Figs 4 and 5) (Table 2). The method of tumor management did not affect metastasis. There was no factor or group of factors found to be significant in the multivariate analysis as predictors of iris melanoma metastasis.

Shields et al 䡠 Iris Melanoma Metastasis

Figure 3. Kaplan-Meier estimates showing the proportion of patients free of metastases over time with microscopically proven iris malignant melanoma.

Discussion Diagnosis Most melanocytic tumors of the iris are benign nevi or freckles, and these lesions tend to remain stable.1,2 Territo and associates25 reported the experience from our department with 175 suspicious melanocytic iris lesions referred because of suspected melanoma and managed by us with cautious observation. We found that only 5% enlarged during a 5-year period. In these cases, tumor growth was not always equivalent with malignant degeneration into melanoma, because three of the five tumors that grew and were resected demonstrated melanoma, whereas two growing lesions showed only benign nevus cells. Thus, the presence of growth in iris melanocytic tumors is not always a sign of malignancy. For this reason, we chose to evaluate, in this major study on iris melanoma, only those patients with histopathologically or cytologically proven iris melanoma and to exclude those that were managed with observation or plaque radiotherapy without microscopic confirmation of the diagnosis. With this approach, our results would likely not be diluted with presumed iris melanoma that had not been confirmed microscopically. This is important because of the 1054 patients referred to us with the diagnosis of possible iris melanoma, only 169 (16%) had tumors with features suspicious enough to warrant microscopic confirmation of iris melanoma. The suspicious clinical features that prompt excisional biopsy, in our experience, include large tumor size, prominent tumor vascularity, tumor seeding, elevated intraocular pressure, and tumor-related ocular complications such as hyphema. If all of the 1054 patients with clinically suspicious iris-pigmented lesions were included in this analysis, the results would have been immensely diluted, a concern expressed by others.18 Several prior studies have been done on iris melanoma in which the authors stress the typical clinical and histopathologic features of this tumor. Rones and Zimmerman,4 in one of the earlier studies, presented data on iris melanoma

metastasis, and they found 1% metastasis at 5 years and 6% metastasis at 10 years on the basis of data from the Armed Forces Institute of Pathology. Our findings of 3% metastasis at 5 years, 5% metastasis at 10 years, and 10% metastasis at 20 years, from a major ophthalmic oncology center, further corroborate their results. In addition, we have further contributed information regarding clinical features of patients at risk for metastasis on the basis of the initial patient examination. These features include older patient age, elevated intraocular pressure, tumor involvement in the iris root or angle, extraocular extension, and prior surgical intervention elsewhere before referral. These risk factors should be kept in mind when counseling patients with melanocytic lesions of the iris. Most iris melanomas are detected when their size is substantially smaller than most choroidal or ciliary body melanomas. If we compare the frequency of metastasis from iris melanoma on the basis of this report to that of similarsized ciliary body and choroidal melanoma,26,27 the prevalence of metastasis is comparable at 3% to 10%. Davidorf28 recognized the similarity in prognosis of both anterior and posterior uveal melanoma and emphasized that iris melanoma is discovered when tumor volume is only an average of 55 mm3, whereas choroidal melanoma is generally discovered when it has a mean volume of 300 mm3. This difference in tumor burden at the time of diagnosis may explain the difference in prognosis of only 3% metastasis with iris melanoma and approximately 15% metastasis for choroidal melanoma.28 Data from animal research suggest that factors other than tumor size, such as biochemical properties, anatomic differences, tumor vascularization, and host immune response, may be involved in the disparate rate of metastasis of iris versus ciliary body and choroidal melanoma.29 There are some similarities and differences among clinical risk factors for metastasis from iris melanoma and ciliary body or choroidal melanoma. Older patient age and extraocular extension have been identified as important risks for metastasis from both anterior and posterior uveal melanoma. Elevated intraocular pressure and iris root/angle invasion are risks more unique to iris melanoma. Elevated intraocular pressure is found in approximately 7% to 30% of patients with iris melanoma and only 2% of patients with choroidal melanoma.8 The mechanisms of glaucoma differ in these two groups. Patients with iris melanoma have elevated pressure develop usually from tumor infiltration of the trabecular meshwork with blockage of outflow channels, whereas those with choroidal melanoma have elevated pressure develop from iris neovascularization (56%) or anterior displacement of the lens/iris diaphragm (34%), leading to angle closure without tumor access to trabecular outflow channels.8 The presence of elevated intraocular pressure in an eye with iris melanoma is thus an important risk for melanoma metastasis. In a separate study, we have analyzed in greater detail the clinical factors that lead to glaucoma and the management of glaucoma in eyes with iris melanoma (Shields CL, Shields JA, Materin M, et al. Factors associated with elevated intraocular pressure in eyes with iris melanoma. Submitted for publication). The most important

175

Ophthalmology Volume 108, Number 1, January 2000 Table 2. Univariate Cox Proportional Hazards Analysis of Clinical Factors at Initial Patient Examination Predictive for Metastasis in 169 Eyes with Microscopically Documented Iris Melanoma

Clinical Feature Treatment elsewhere Surgical intervention vs. observation* Extraocular extension Present vs. absent* Posterior tumor margin Angle or iris root vs. midzone* Age at diagnosis Increasing Intraocular pressure Raised vs. normal*

P Value

Relative Risk

95% Confidence Interval

Discrete Variable, Continuous Variable

0.0063

1.5649

1.1348, 2.1579

Discrete

0.0214

6.9147

1.3314, 35.9125

Discrete

0.0220

2.9899

1.1712, 7.6328

Discrete

0.0284

1.0491

1.0051, 1.0951

Continuous

0.0321

4.2779

1.1321, 16.1648

Discrete

* Reference variable.

Figure 4. Iris melanoma in a patient with no risk factors for metastasis. A, This 54-year-old patient had an iris melanoma extending from the pupillary margin to midzone of the iris without iris seeding. B, After wide margin iridectomy, the pupil was repaired and the patient has healed well. The tumor proved to be spindle cell–type malignant melanoma.

Figure 5. Iris melanoma in a patient with risk factors for melanoma metastasis. A, This 56-year-old patient had an extensive, documented growing iris melanoma with two risk factors for metastasis, including posterior tumor margin at iris root and elevated intraocular pressure. B, Gonioscopy demonstrates diffuse seeding in the trabecular meshwork. Enucleation was performed, and the tumor proved to be a mixed-cell type melanoma with predominance of epithelioid cells.

176

Shields et al 䡠 Iris Melanoma Metastasis factor by multivariate analysis is the extent of tumor seeding in the anterior chamber angle. Less important factors include poor visual acuity at presentation, greater tumor size and seeding on the iris, location of tumor in the angle, and flat (rather than nodular) tumor shape. Flat, diffuse iris melanoma tends to be discohesive and leads to tumor dispersion on the adjacent iris and into the angle.30 Of the five clinical factors predictive of iris melanoma metastases, the most commonly encountered risk factor that might be controlled is elevated intraocular pressure, which is found in 30% of patients with iris melanoma. We cannot alter the age at which the patient is initially seen or the location of the tumor within the iris. By identifying patients with extensive tumor seeding on the iris and in the angle and those with large tumor size and flat configuration, we can closely follow the patient for eventual rise in intraocular pressure. Perhaps earlier intervention in patients with these features is warranted before glaucoma develops or the tumor progresses further. Avoidance of glaucoma-filtering procedures is advised, because this could seed tumor into an extraocular compartment, leading to a greater risk for metastasis. One interesting risk factor for metastasis from iris melanoma was surgical intervention before referral to us. It is difficult to interpret this finding, but comment should be made regarding the 18 patients who had prior surgical intervention. Ten patients had incisional biopsy, seven had resection (technique unknown), and one had enucleation. Three of the 18 patients had metastases develop after incisional biopsy, iridocyclectomy, and enucleation in single cases each. It is unclear whether the surgical intervention or unidentified factors that may have contributed to the need for surgical intervention were related to the metastases.

Management Our present strategy for the management of suspected iris melanoma, especially those with documented growth, begins with a careful inspection of all anterior segment structures, including the episcleral tissue, cornea, anterior chamber angle, lens, and iris. The preoperative assessment is critical for the planning of management. The goal of management is to treat the lesion completely with resection or radiotherapy without manipulating or disrupting the tumor itself. If a tumor is circumscribed in the iris with no seeding or involvement of the peripheral portion of the iris, we would approach this surgically with partial lamellar scleroiridectomy.1,2,31 We prefer a scleral flap incision to minimize postoperative astigmatism. However, it is important to make a large limbal incision, often encompassing five or more clock hours of limbus, to allow for tumor removal without brushing the tumor cells on the adjacent tissue. At least 1 to 2 mm of tumor-free margins and “no touch” technique (without handling the tumor tissue directly) are planned for all surgical resections. For circumscribed tumors with iris root involvement but without seeding, partial lamellar iridocyclectomy is performed with a more posterior scleral flap, generous incision, and wide tumor margins with excision of the entire iris mass surrounded by normal iris and ciliary body. When involvement of the anterior cham-

ber angle is encountered preoperatively, we plan partial lamellar iridogoniocyclectomy, in which the trabecular meshwork structures are removed intact with the main tumor mass in one specimen without cutting through tumor tissue. Seeding of the tumor on the adjacent iris or into the anterior chamber angle, as well as remotely on the lens or corneal endothelium, poses special problems for tumor management that need to be addressed. When diffuse seeding from iris melanoma is present, we often perform a FNAB24 of the main tumor to confirm the diagnosis and rule out simulating lesions such as benign melanocytoma.32 Once the diagnosis is confirmed cytologically, enucleation or, in special circumstances, plaque radiotherapy14 is used. If the seeding is localized to within one or two clock hours of the mass, cautious resection of the entire tumor and seeding components as one specimen is performed, generally followed by plaque radiotherapy to the resection site. In other instances, plaque radiotherapy to the entire site is delivered without prior resection. Plaque radiotherapy is also important in postoperative consolidation treatment for an eye harboring a high-grade resected iris melanoma or one with tumor cells that reach the margin of the resection specimen. We prefer to avoid incisional biopsy because of the risk of tumor dispersion. However, in cohesive diffuse tumors, in which enucleation may be planned and needle biopsy fails to yield cellularity, incisional biopsy through clear corneal approach may be beneficial to confirm the diagnosis before enucleation. Our results should be interpreted with caution. In this report we did not assess all pigmented iris tumors, but we selected only those with microscopically proven iris melanoma. Therefore, we cannot comment on factors predictive of growth of pigmented lesions, nor can we prove what is the best management of pigmented iris tumors. However, by selecting those with microscopically proven melanomas, we provided expected frequency and risks for metastasis and did not dilute our results with cases of clinically diagnosed melanomas without microscopic verification of the diagnosis. It should be understood that our data might be biased toward more difficult and advanced cases, because we are a tertiary referral center for eye cancers. Thus, the results may be skewed toward a greater frequency of metastasis from iris melanoma than expected. On the other hand, by selecting only microscopically proven melanomas, we may have missed some low-grade melanomas that were showing no growth. Hence, the frequency of metastasis from iris melanomas may be lower than suggested by our study. Nevertheless, our carefully planned study may provide the best information currently available on risk factors and frequency of metastasis for iris melanoma. In summary, our results show that microscopically proven iris melanoma metastasizes in 3% of patients at 5 years, 5% of patients at 10 years, and 10% of patients at 20 years follow-up. Factors that predict metastasis include older patient age, iris root/angle involvement, elevated intraocular pressure, and extraocular tumor extension. Factors that may signal the development of glaucoma include increasing tumor size and seeding on the adjacent iris or in the anterior chamber angle, location of tumor in the angle, flat

177

Ophthalmology Volume 108, Number 1, January 2000 growth pattern, and poor visual acuity at presentation. Those patients with large tumors, especially those located in the iris root or angle and with iris and angle seeding, generally require active treatment rather than observation, because they pose the greatest risk for the development of glaucoma and ultimate tumor metastasis. The clinician should be cognizant of these factors when counseling a patient with a melanocytic iris lesion.

References 1. Shields JA, Shields CL. Intraocular Tumors: A Text and Atlas. Philadelphia: WB Saunders, 1992;61– 84. 2. Shields JA, Shields CL. Atlas of Intraocular Tumors. Philadelphia: Lippincott Williams & Wilkins, 1999;17– 42. 3. Jensen OA. Malignant melanomas of the human uvea. Recent follow-up of cases in Denmark, 1943–1952. Acta Ophthalmol (Copenh) 1970;48:1113–28. 4. Rones B, Zimmerman LE. The prognosis of primary tumors of the iris treated by iridectomy. Arch Ophthalmol 1958;60:193– 205. 5. Raivio I. Uveal melanoma in Finland. An epidemiological, clinical, histological and prognostic study. Acta Ophthalmol Scand Suppl 1977;133:1– 64. 6. Cleasby GW. Malignant melanoma of the iris. Arch Ophthalmol 1958;60:403–17. 7. Kersten RC, Tse DT, Anderson R. Iris melanoma. Nevus or malignancy? Surv Ophthalmol 1985;29:423–33. 8. Shields CL, Shields JA, Shields MB. Prevalence and mechanisms of secondary intraocular pressure elevation in eyes with intraocular tumors. Ophthalmology 1987;94:839 – 46. 9. McGalliard JN, Johnston PB. A study of iris melanoma in Northern Ireland. Br J Ophthalmol 1989;73:591–5. 10. Shields JA, Sanborn GE, Augsburger JJ. The differential diagnosis of malignant melanoma of the iris. A clinical study of 200 patients. Ophthalmology 1983;90:716 –20. 11. Reese AB, Cleasby GW. The treatment of iris melanoma. Am J Ophthalmol 1959;47:118 –25. 12. Reese AB, Jones IS, Cooper WC. Surgery for tumors of the iris and ciliary body. Am J Ophthalmol 1968;66:173– 84. 13. Makley TA Jr. Management of melanomas of the anterior segment [review]. Surv Ophthalmol 1974;19:135–53. 14. Shields CL, Shields JA, De Potter P, et al. Treatment of non-resectable malignant iris tumours with custom designed plaque radiotherapy. Br J Ophthalmol 1995;79:306 –12. 15. Char DH, Crawford JB, Kroll S. Iris melanomas. Diagnostic problems. Ophthalmology 1996;103:251–5.

178

16. Zimmerman LE. Clinical pathology of iris tumors. Ward Burdick Award contribution. Am J Clin Pathol 1963;39:214 –28. 17. Arentsen JJ, Green WR. Melanoma of the iris: report of 72 cases treated surgically. Ophthalmic Surg 1975;6:23–37. 18. Jakobiec FA, Silbert G. Are most iris “melanomas” really nevi? A clinicopathologic study of 189 lesions. Arch Ophthalmol 1981;99:2117–32. 19. Jensen OA. Malignant melanoma of the iris. A 25-year analysis of Danish cases. Eur J Ophthalmol 1993;3:181– 8. 20. Batioglu F, Gunalp I. Malignant melanomas of the iris. Jpn J Ophthalmol 1998;42:281–5. 21. Sunba MSN, Rahi AHS, Morgan G. Tumors of the anterior uvea. I. Metastasizing malignant melanoma of the iris. Arch Ophthalmol 1980;98:82–5. 22. Geisse LJ, Robertson DM. Iris melanomas. Am J Ophthalmol 1985;99:638 – 48. 23. Lee ET. Statistical Methods for Survival Data Analysis. 2nd ed. New York: Wiley, 1992; 258. Wiley Series in Probability and Mathematical Statistics. 24. Shields JA, Shields CL, Ehya H, et al. Fine needle aspiration biopsy of suspected intraocular tumors. The 1992 Urwick Lecture. Ophthalmology 1993;100:1677– 84. 25. Territo C, Shields CL, Shields JA, et al. Natural course of melanocytic tumors of the iris. Ophthalmology 1988;95: 1251–5. 26. Diener-West M, Hawkins BS, Markowitz JA, Schachat AP. A review of mortality from choroidal melanoma. II. A metaanalysis of 5-year mortality rates following enucleation, 1966 through 1988. Arch Ophthalmol 1992;110:245–50. 27. Shields CL, Shields JA, Kiratli H, et al. Risk factors for growth and metastasis of small choroidal melanocytic lesions. Ophthalmology 1995;102:1351– 61. 28. Davidorf FH. The melanoma controversy. A comparison of choroidal, cutaneous, and iris melanomas. Surv Ophthalmol 1981;25:373–7. 29. Grossniklaus HE, Wilson MW, Barron BC, Lynn ML. Anterior vs posterior intraocular melanoma. Metastatic differences in a murine model. Arch Ophthalmol 1996;114:1116 –20. 30. Rones B, Zimmerman LE. The production of heterochromia and glaucoma by diffuse malignant melanoma of the iris. Trans Am Acad Ophthalmol Otolaryngol 1957;61: 447– 63. 31. Shields JA, Shields CL. Surgical approach to lamellar sclerouvectomy for posterior uveal melanomas: The 1986 Schoenberg Lecture. Ophthalmic Surg 1988;19:774 – 80. 32. Fineman M, Eagle RC Jr, Shields JA, et al. Melanocytomalytic glaucoma in eyes with necrotic iris melanocytoma. Ophthalmology 1998;105:492– 6.