Mortality After Diagnosis of Small Melanocytic Lesions of the Choroid

CLINICAL SCIENCES Mortality After Diagnosis of Small Melanocytic Lesions of the Choroid

Anne Marie Lane, MPH; Kathleen M. Egan, ScD; Ivana K. Kim, MD; Evangelos S. Gragoudas, MD

Objective: To evaluate the risk of dying of metastatic (median, 64 years). A total of 15 deaths due to ocular choroidal melanoma in patients with small, indetermi- melanoma were ascertained (median follow-up of sur- nate, pigmented lesions of the uveal tract. vivors, 8.2 years), 13 in the melanoma group and 2 in the indeterminate lesion group; actuarial tumor- Methods: A cohort of 1063 consecutive patients were specific death rates at 10 years after evaluation were 5% evaluated in the Ocular Oncology Clinic of the Massa- (95% confidence interval, 3%-8%) and 1% (95% confi- chusetts Eye and Ear Infirmary between January 1976 and dence interval 0%-3%), respectively. No deaths due to June 1996 with definite choroidal nevus (n=256), inde- ocular melanoma occurred in the nevus group. terminate lesions (n=334), or small melanoma (n=373). Deaths occurring up to December 1998 were identified Conclusions: These data document the very low malig- through active follow-up or by a search of the National nant potential of most indeterminate melanocytic le- Death Index. Cumulative death rates were compared sions of the choroid and support the current practice of among diagnostic groups using the Kaplan-Meier method. monitoring these tumors, with treatment provided when growth and other signs of malignant transformation are Results: Mean lesion diameter was 4.6 mm in the nevi, observed. 7.0 mm in the indeterminate lesions, and 8.1 mm in the small melanomas. Patients ranged in age from 3 to 95 years Arch Ophthalmol. 2010;128(8):996-1000

HERE IS CONSIDERABLE EVI- Uveal nevi are generally flat, slate-gray dence that intraocular mela- lesions without sharply demarcated mar- nomas arise from benign gins; their size is limited to about 6 mm in pigmented precursor le- diameter.14 However, there is considerable sions (nevi) in the uveal overlap in size distributions of nevi and in- tract. In histopathological sections of uveal determinate lesions compared with small T 15 melanomas, nevoid cells can be identified melanomas, and the differential diagno- at the base of many tumors,1 and there are sis can be especially difficult in borderline documented cases of uveal melanomas aris- cases with characteristics of both lesions.16 ing from preexisting nevi.2 Also, persons While most studies have evaluated the with large numbers of cutaneous3-7 and iris6-8 growth potential of nevi and indetermi- nevi are at heightened risk of developing nate pigmented lesions of the uveal tract, melanomas in the posterior uveal tract (ie, the eventual mortality associated with such the choroid and ciliary body). lesions is not known. One study evaluated Benign nevi of the choroid are com- risk factors for metastasis in 1329 small cho- mon in white populations. In a survey of roidal lesions of 3 mm or less in height and 3654 subjects (99% white), the nevus found that 3% of patients developed me- prevalence rate was estimated at 8.9% in tastasis.17 Another study of patients with in- women and 8.3% in men,9 based on fun- determinate choroidal lesions was re- dus photographs. These figures likely un- stricted to those who were subsequently derestimate the prevalence, as small, less diagnosed with small melanomas; the 5-year pigmented nevi are often missed clini- mortality rate after treatment with plaque cally.10 In comparison, melanomas of the radiotherapy was 3.9%.18 In this study, we uveal tract are uncommon, with an an- followed up a large series of patients diag- nual incidence in the United States of about nosed with choroidal nevi or indetermi- 6 to 7 cases per million persons.11 An es- nate lesions to determine long-term sur- timated 1 in 8845 choroidal nevi will trans- vival experience after initial presentation of Author Affiliations: form to a malignant tumor per year in the the uveal tumor. We compared their out- Massachusetts Eye and Ear 12 Infirmary, Boston (Drs Kim and white US population. The tumors me- comes with those of patients diagnosed with Gragoudas and Ms Lane); and tastasize primarily to the liver, and up to small melanomas and treated to determine Moffitt Cancer Center, Tampa, 50% of patients may die of the disease whether an observational approach to man- Florida (Dr Egan). within 10 years of diagnosis.13 aging indeterminate lesions is appropriate.

(REPRINTED) ARCH OPHTHALMOL / VOL 128 (NO. 8), AUG 2010 WWW.ARCHOPHTHALMOL.COM 996

No. (%)

Definite Indeterminate Small Indeterminate Nevus Lesion Melanoma Lesion Characteristic (n=256) (n=334) (n=373) (n=334) Age at diagnosis, mean (IQR), y 62 (54-73) 62 (52-73) 59 (49-70) 62 (52-73) P value .64 Ͻ.001 Sex Male 88 (34) 116 (35) 197 (53) 116 (35) Female 168 (66) 218 (65) 176 (47) 218 (65) P value .93 Ͻ.001 Eye involved Right 133 (53) 158 (47) 200 (54) 158 (47) Left 119 (47) 176 (53) 173 (46) 176 (53) Both 4 0 0 0 P value .19 .06 Symptoms Yes 101 (39) 137 (41) 219 (59) 137 (41) No 155 (61) 197 (59) 154 (41) 197 (59) P value .70 Ͻ.001 Diameter, mean (IQR), mm 4.7 (3-6) 7.0 (6-9) 8.1 (7.5-9) 7.0 (6-9) NR 74 (29) 64 (19) 0 64 (19) P value Ͻ.001 Ͻ.001 Height, mean (IQR), mm 0.1 (0-0.5) 1.3 (0.8-1.8) 2.7 (2-3.2) 1.3 (0.8-1.8) NR 127 (50) 37 (11) 0 37 (11) P value Ͻ.001 Ͻ.001

Abbreviations: IQR, interquartile range; NR, not recorded in the medical record.

METHODS Index (NDI), maintained by the US National Center for Health Statistics. The Social Security Death Index is available online, with free access (http://www.ancestry.com), and contains the date and Patients were referred to the Ocular Oncology Clinic of the Mas- residence at the time of death but not the cause of death. The sachusetts Eye and Ear Infirmary (MEEI) between January 1976 NDI records deaths in all 50 states, Puerto Rico, and the Virgin and June 1996 for evaluation of a pigmented lesion of the cho- Islands, beginning in 1979. Qualified researchers can, for a fee, roid that, based on funduscopic and ultrasound findings, was di- request searches of NDI data files for a prescribed range in years agnosed either as a definite choroidal nevus (n=256) or an in- based on combinations of identifiers including name, date of birth, determinate lesion (eg, nevus vs small melanoma; n=334). In maiden name, and social security number; any near matches are Ͻ general, lesions were categorized as nevi if they were flat ( 0.5 identified, and the date and cause of death as recorded on the death mm) and less than 6 mm in diameter. Lesions were categorized certificate are provided. The NDI has consistently been found to as indeterminate if they were less than 2 mm thick and did not have high sensitivity and specificity related to death ascertain- exhibit known risk characteristics such as orange pigment, sub- ment.20,21 In general, we used the Social Security Death Index to retinal fluid, or symptoms at presentation. Another 373 patients screen for deaths and to identify social security numbers (when Ͻ were diagnosed with a small melanoma (lesions 10 mm in di- not available in medical records), which increase NDI specific- Ͻ ameter and 5 mm in height). Small melanomas were treated at ity. The NDI was used to confirm deaths and to identify cause. MEEI by proton irradiation, while nevi and indeterminate le- We calculated cumulative death rates of ocular melanoma ac- sions were observed for changes indicating malignant transfor- cording to diagnostic subgroup using the Kaplan-Meier approach. mation. If malignant transformation occurred and a subsequent The data were censored at December 31, 1998, the date through diagnosis of small melanoma was made, the tumor was treated which NDI had complete records at the time the search request with proton irradiation. Because melanomas of the iris have low was submitted to the National Center for Health Statistics. Expected 19 malignant potential, lesions involving the iris were not in- numbers of death from cutaneous and ocular melanoma were es- cluded. All patients were residents of the United States and were timated based on mortality data (1993 to 1997) from the National evaluated by a single physician (E.S.G.). Patient characteristics Center for Health Statistics. For ocular melanoma, these estimates including age, sex, race, eye involvement, and state of residence were based on tumors of the eye and orbit combined. at the time of evaluation were recorded for all patients. Informa- tion on tumor size including largest basal diameter of the lesion by ophthalmoscopy, tumor height by ultrasound (measured to RESULTS the back of the tumor), and whether the lesion was detected on screening examination or the patient developed symptoms, gen- Table 1 presents patient and lesion characteristics, con- erally available in the medical record, were also recorded. trasting nevi and indeterminate lesions (eg, nevus vs small Patients treated by proton irradiation and approximately one- melanoma) managed by observation, and definitive small quarter of the patients with nevi or indeterminate lesions were Ͻ Ͻ followed up actively through regular examination at the MEEI melanomas ( 10 mm in diameter and 5 mm in height) or through annual contacts with the referring ophthalmologist. treated at our institution by proton irradiation. Lesions in Of the remaining patients, including approximately 40% seen the nevus category ranged from 0.5 mm to 12 mm in di- only once in consultation, vital status was ascertained through ameter (mean diameter, 4.7 mm) and tended to be flat or searches of the Social Security Death Index and the National Death minimally elevated. Indeterminate lesions (Figure 1) were

(REPRINTED) ARCH OPHTHALMOL / VOL 128 (NO. 8), AUG 2010 WWW.ARCHOPHTHALMOL.COM 997

Figure 1. Fundus photograph taken during evaluation at Massachusetts Eye and Ear Infirmary. Stable lesion measuring 1.35 mm (A) at baseline evaluation and 1.4 mm (B) after 12 years of observation.

A B

Figure 2. Fundus photograph taken during evaluation at Massachusetts Eye and Ear Infirmary showing flat nevus (A) with subretinal fluid, taken 3 years after initial presentation. B, Five Years after initial presentation, the lesion measured 2.0 mm on ultrasound, and melanoma diagnosis was made.

larger, on average, than the nevi but smaller than melano- To determine if any clinical factors routinely available mas. Compared with the other groups, patients with small in medical records predicted lesion progression, we com- melanomas were significantly younger and more likely pared patients followed up at MEEI who progressed dur- to be symptomatic. Patients with melanoma were bal- ing observation and were subsequently treated by proton anced regarding sex, while those in the benign and in- irradiation with patients whose lesions remained stable dur- termediate subgroups were almost twice as likely to be ing observation at MEEI, and found that tumor height female (PϽ.001). greater than 1.5 mm and presentation with symptoms sig- A total of 39 patients with indeterminate lesions and 3 nificantly elevated risk of progression (data not shown). nevi subsequently had treatment for the lesion with pro- A total of 4 patients died of metastatic cutaneous mela- ton irradiation after signs of growth and/or other changes, noma and 2 of ocular melanoma. Both patients who died from 2 months to 18 years (median, 4 years) after initial of ocular melanoma were originally diagnosed with an in- observation at MEEI. Growth may have included expan- determinate lesion and were later treated with proton ir- sion laterally as well as in height, and other changes in- radiation 10 and 20 months (Figure 3) after initial pre- cluded the development of orange pigment, subretinal fluid, sentation at MEEI. The patients died of metastatic disease and symptoms. The patients with nevi that progressed dur- 5.5 and 12.8 years, respectively, after presentation at MEEI ing observation (Figure 2) were subsequently treated by (4.7 and 11.1 years after receiving treatment). proton irradiation 3.4, 7.2, and 9.4 years after initial evalu- Of the patients with an indeterminate lesion or ne- ation. In each case, in the baseline examination, at least 1 vus (n=590), the observed number of deaths due to mela- dimension of the lesion was larger than the average com- noma exceeded that expected in the general population pared with other nevi in the series. Each of these lesions (Table 2). For cutaneous malignant melanoma, death was also symptomatic, and 2 were associated with the pres- rates were 13 times higher than expected in 4682 person- ence of subretinal fluid. years based on cancer-specific mortality rates in the United

(REPRINTED) ARCH OPHTHALMOL / VOL 128 (NO. 8), AUG 2010 WWW.ARCHOPHTHALMOL.COM 998

Figure 3. Fundus photograph taken during evaluation at Massachusetts Eye and Ear Infirmary showing a suspicious lesion (A) that underwent malignant transformation (B) 20 months after the initial evaluation. The patient was treated with proton therapy and subsequently developed metastasis 11 years after treatment.

Table 2. Observed and Expected Numbers of Deaths Table 3. Actuarial Death Rates of Ocular Melanoma Due to Ocular and Cutaneous Melanoma Among 590 Metastasis in Indeterminate Lesions and Small Melanomas Patients With a Definite Pigmented Choroidal Lesion Cumulative Death Rates Cause of Death Observed Expecteda O:E Deaths Due by Years of Follow-up, to Ocular % (95% CI) Ocular melanoma 2 0.015 133.3 Patients, Melanoma, Cutaneous melanoma 4 0.31 12.9 Diagnosis No. No. 51015 Indeterminate 334 2 0 1 (0-3) 3 (1-15) Abbreviation: O:E, observed vs expected. lesions a Expected numbers are based on mortality data (1993 to 1997) provided by the US National Center for Health Statistics; for ocular melanoma, tumors Small 373 13 2 (1-4) 5 (3-8) 7 (4-12) of the eye and orbit are combined. melanomas

Abbreviation: CI, confidence interval. States (1985-1995). For ocular melanoma, 2 deaths were observed, while 0.015 were expected (observed vs ex- sis of melanoma is more likely if several of these charac- pected,133); based on these numbers, 1 death due to ocu- teristics are present in addition to growth. lar melanoma would be expected per year in 2341 pa- Some argument can be made that any lesion suspi- tients with indeterminate lesions or nevi [(2/4682)-1]. cious for intraocular melanoma should be treated promptly Fifteen patients died of melanoma, 2 (0.6%) in the inde- to avoid the possibility of metastasis.29 However, the determinate lesion group and 13 (3.5%) in the small melanoma cision to intervene in tumors of the eye is complicated group. Death rates for metastasis at 5, 10, and 15 years were by the fact that any intervention will potentially impair 0%, 1%, and 3%, respectively, for indeterminate melanocytic vision and may risk loss of the globe. The current data lesions (n=334) and 2%, 5%, and 7%, respectively, for defi- suggest that the malignant potential of small melano- nite small melanomas (n=373) (Table 3). No patients with cytic lesions in the eye is very low, with only 1 death ex- a presumed benign nevus died of melanoma metastasis in pected owing to ocular melanoma in 2340 patient-years the period of follow-up (mean, 8.4 years). after detection (combining nevus and indeterminate lesions). However, about 60% of these patients would be COMMENT expected to lose useful vision in the eye (to worse than 20/200) if treated by proton irradiation, applying rates The management of small elevated pigmented lesions of observed in the small melanomas (data not shown). It is the choroid is controversial in ophthalmology, and the usual not clear that early intervention would prevent every case practice currently is to withhold intervention until there of metastasis,29 whereas studies suggest that a period of is documented evidence of tumor activity.22 Tumor growth, observation prior to intervention is not likely to increase which occurs within several years in more than a third of death rates when compared with immediate treat- uveal lesions suspected for melanoma during observa- ment.30 However, a randomized trial would be necessary tion,16,23,24 is considered the most important sign of malig- to definitively address this question. nancy. However, tumor growth is not always a reliable In this series, patients with a nevus or an indetermi- marker of malignant potential. Benign choroidal nevi may nate lesion were significantly older and more likely to enlarge under observation,25 while uveal melanomas may be female when compared with patients with a definite be stable for long periods of time without progression.26 melanoma. The 2-fold excess in women was present in Other characteristics that have been identified as signs of the symptomatic lesions as well as those detected by malignant transformation include orange pigment, sub- screening (data not shown). It is possible that women retinal fluid, and the presence of symptoms.27,28 A diagno- have a lower threshold for screening examination or for

(REPRINTED) ARCH OPHTHALMOL / VOL 128 (NO. 8), AUG 2010 WWW.ARCHOPHTHALMOL.COM 999

©2010 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/02/2021 visiting a doctor with symptoms. However, in a large sur- REFERENCES vey, the prevalence of choroidal nevi was also higher in 9 women. In ocular melanoma, incidence rates are com- 1. Yanoff M, Zimmerman LE. Histogenesis of malignant melanomas of the uvea II: parable in men and women, with a suggestion in some relationship of uveal nevi to malignant melanomas. Cancer. 1967;20(4):493-507. series of slightly higher rates in men.11 Why women would 2. Eskelin S, Kivela T. Mode of presentation and time to treatment of uveal mela- be at higher risk of the presumed precursor lesion, though noma in Finland. Br J Ophthalmol. 2002;86(3):333-338. 3. Tucker M, Hartge P, Shields J. Epidemiology of intraocular melanoma. Recent not the cancer itself, is not clear. However, death rates Results Cancer Res. 1986;102:159-165. of these tumors are slightly higher in men, which ap- 4. Holly EA, Aston DA, Char DH, Kristiansen JJ, Ahn DK. Uveal melanoma in relation to pears to be related to a protective effect of childbearing ultraviolet light exposure and host factors. Cancer Res. 1990;50(18):5773-5777. in women,31 and it may be that immunologic factors in- 5. Seddon JM, Gragoudas ES, Glynn RJ, Egan KM, Albert DM, Blitzer PH. Host factors, UV radiation, and risk of uveal melanoma: a case-control study. Arch fluence progression at various stages of the disease. Ophthalmol. 1990;108(9):1274-1280. There are several issues to consider when evaluating 6. Bataille V, Sasieni P, Cuzick J, Hungerford JL, Swerdlow A, Bishop JA. Risk of these data. The diagnosis and treatment of small mela- ocular melanoma in relation to cutaneous and iris naevi. Int J Cancer. 1995; nomas was based on clinical judgment alone, without the 60(5):622-626. aid of a biopsy for definitive diagnosis. Therefore, it is 7. Weis E, Shah CP, Lajous M, Shields JA, Shields CL. The association of cutaneous and iris nevi with uveal melanoma: a meta-analysis. Ophthalmology. 2009; possible that a small number of treated tumors were not 116(3):536-543, e2. ocular melanomas. This was not a study of the natural 8. Wilder HC. Relationship of pigmented cell clusters in the iris to malignant mela- history of choroidal pigmented lesions. Many of the pa- noma of the uveal tract. Annl N Y Acad Sci. 1948;4:137-143. tients in this series were referred to MEEI for consulta- 9. Sumich P, Mitchell P, Wang JJ. Choroidal nevi in a white population: the Blue Mountains Eye Study. Arch Ophthalmol. 1998;116(5):645-650. tion only, and we lacked data on the clinical course and 10. Naumann G, Yanoff M, Zimmerman LE. Histogenesis of malignant melanomas eventual treatment of their lesions. Thus, an unknown of the uvea I. Arch Ophthalmol. 1966;76(6):784-796. proportion of the patients may have had some interven- 11. Egan KM, Seddon JM, Glynn RJ, Gragoudas ES, Albert DM. Epidemiologic as- tion at a later date, and the influence of treatment on the pects of uveal melanoma. Surv Ophthalmol. 1988;32(4):239-251. 12. Singh AD, Kalyani P, Topham A. Estimating the risk of malignant transformation current results cannot be evaluated. Also, the NDI search of a choroidal nevus. Ophthalmology. 2005;112(10):1784-1789. does not include deaths that occur outside the United 13. Gragoudas E, Li W, Goitein M, Lane AM, Munzenrider JE, Egan KM. Evidence- States, and such deaths go undetected. However, the NDI based estimates of outcome in patients irradiated for intraocular melanoma. Arch has been shown to have high sensitivity and specificity Ophthalmol. 2002;120(12):1665-1671. for determining deaths and attributing causes.20,21 In the 14. Ganley JP, Comstock GW. Benign nevi and malignant melanomas of the choroid. Am J Ophthalmol. 1973;76(1):19-25. present series, the higher-than-expected rates of death 15. Augsburger JJ, Correa ZM, Trichopoulos N, Shaikh A. Size overlap between be- due to cutaneous melanoma in a selected series of pa- nign melanocytic choroidal nevi and choroidal malignant melanomas. Invest Oph- tients with melanocytic lesions in the eye tends to vali- thalmol Vis Sci. 2008;49(7):2823-2828. date the method. Finally, though follow-up was pro- 16. Char D, O’Brien J. Clinical management of patients with indeterminate pigmented choroidal lesions. Semin Ophthalmol. 1993;8(4):230-233. longed in many patients, it should be noted that malignant 17. Shields CL, Shields JA, Kiratli H, De Potter P, Cater JR. Risk factors for growth changes in precursor lesions may require decades to evolve and metastasis of small choroidal melanocytic lesions. Ophthalmology. 1995; and thus the ultimate prognosis associated with early me- 102(9):1351-1361. lanocytic lesions in the eye would require longer fol- 18. Sobrin L, Schiffman JC, Markoe AM, Murray TG. Outcomes of iodine 125 plaque low-up than was available in this series. Despite these limi- radiotherapy after initial observation of suspected small choroidal melanomas: a pilot study. Ophthalmology. 2005;112(10):1777-1783. tations, the results of this study revealed low mortality 19. Jakobiec FA, Silbert G. Are most iris “melanomas” really nevi? a clinicopatho- rates in patients with these tumors and support the cur- logic study of 189 lesions. Arch Ophthalmol. 1981;99(12):2117-2132. rent management practice of close serial examinations 20. Curb JD, Ford CE, Pressel S, Palmer M, Babcock C, Hawkins CM. Ascertainment without therapeutic intervention until growth is observed. of vital status through the national death index and the social security administration. Am J Epidemiol. 1985;121(5):754-766. 21. Sesso HD, Paffenbarger RS, Lee IM. Comparison of national death index and world wide web death searches. Am J Epidemiol. 2000;152(2):107-111. Submitted for Publication: June 17, 2007; final revision 22. Augsburger JJ. Is observation really appropriate for small choroidal melanomas. received December 18, 2009; accepted December 23, 2009. Trans Am Ophthalmol Soc. 1993;91:147-175. Correspondence: Anne Marie Lane, MPH, Retina Ser- 23. Gass JD. Observation of suspected choroidal and ciliary body melanomas for evidence of growth prior to enucleation. Ophthalmology. 1980;87(6):523-528. vice, 243 Charles St, Boston, MA 02114 (alane@meei 24. Augsburger JJ, Schroeder RP, Territo C, Gamel JW, Shields JA. Clinical para- .harvard.edu). meters predictive of enlargement of melanocytic choroidal lesions. Br J Ophthalmol. Author Contributions: Ms Lane had full access to all of 1989;73(11):911-917. the data in the study and takes responsibility for the in- 25. MacIlwaine WA IV, Anderson B Jr, Klintworth GK. Enlargement of a histologi- cally documented choroidal nevus. Am J Ophthalmol. 1979;87(4):480-486. tegrity of the data and the accuracy of the data analysis. 26. Gundersen T, Smith TR, Zakov N, Albert DM. Choroidal melanocytic tumor observed Financial Disclosure: None reported. for 41 years before enucleation. Arch Ophthalmol. 1978;96(11):2089-2092. Funding/Support: This study was supported by Re- 27. Butler P, Char DH, Zarbin M, Kroll S. Natural history of indeterminate pigmented search to Prevent Blindness (Dr Kim). choroidal tumors. Ophthalmology. 1994;101(4):710-716. Previous Presentations: Presented in part at the 28th an- 28. Shields CL, Cater J, Shields JA, Singh AD, Santos MC, Carvalho C. Combination of clinical factors predictive of growth of small choroidal melanocytic tumors. nual Macula Society Meeting; February 23-26, 2005; Key Arch Ophthalmol. 2000;118(3):360-364. Biscayne, Florida; the 2005 Annual Meeting of the Retina 29. Shields CL, Shields JA, Kiratli H, De Potter P, Cater JR. Risk factors for growth Society; September 14-18, 2005; Coronado, California; and and metastasis of small choroidal melanocytic lesions. Trans Am Ophthalmol the American Academy of Ophthalmology Subspecialty Day; Soc. 1995;93:259-275. 30. Augsburger JJ, Vrabec TR. Impact of delayed treatment in growing posterior uveal November 9-10, 2007; New Orleans, Louisiana. melanomas. Arch Ophthalmol. 1993;111(10):1382-1386. Additional Contributions: The authors wish to thank Molly 31. Egan KM, Quinn JL, Gragoudas ES. Childbearing history associated with improved Beals, BA, for technical assistance and data management. survival in choroidal melanoma. Arch Ophthalmol. 1999;117(7):939-942.

(REPRINTED) ARCH OPHTHALMOL / VOL 128 (NO. 8), AUG 2010 WWW.ARCHOPHTHALMOL.COM 1000