Figure. Kaplan-Meier plots with Kaplan-Meier estimate, 95% CI, and patients at risk (PaR) at 2, 3, and 5 years. A, Relapse-free survival. B, Eye preservation. C, Radiation retinopathy. D, Radiation opticopathy. E, Rubeotic glaucoma. F, Overall survival.
Suesskind D, Scheiderbauer J, Buchgeister M, et al. Retrospective evaluation of patients with uveal melanoma treated by stereotactic radiosurgery with and without tumor resection. JAMA Ophthalmol. Published online March 14, 2013. doi:10.1001/jamaophthalmol.2013.697.
eText: Details about recurrences
eFigure: Dose distribution
Daniela Suesskind, Jutta Scheiderbauer, Markus Buchgeister, Michael Partsch, Wilfried Budach, Karl U. Bartz-Schmidt, Rainer Ritz, Salvatore Grisanti, Frank Paulsen. Retrospective Evaluation of Patients With Uveal Melanoma Treated by Stereotactic Radiosurgery With and Without Tumor Resection. JAMA Ophthalmol. 2013;131(5):630–637. doi:10.1001/jamaophthalmol.2013.697
Author Affiliations: Center for Ophthalmology (Drs Suesskind, Partsch, Bartz-Schmidt, and Grisanti) and Departments of Radiation Oncology (Drs Scheiderbauer, Budach, and Paulsen) and Neurosurgery (Dr Ritz), Eberhard Karls University Tuebingen, Tuebingen; Department of Mathematics, Physics, and Chemistry, Beuth University for Applied Sciences Berlin, Berlin (Dr Buchgeister); and Department of Ophthalmology, University of Luebeck, Luebeck (Dr Grisanti), Germany.
Importance The present study intended to analyze the suitability of single-dose stereotactic radiotherapy in the treatment of uveal melanoma that cannot be handled with ruthenium-brachytherapy and therefore is a challenge for ophthalmologists concerning local tumor control, as well as preservation of the eye and visual function.
Objectives To evaluate local tumor control, eye preservation, visual course, radiation complications, metastases, and death after single-dose stereotactic radiotherapy (SDRT) applied exclusively or combined with tumor resection in uveal melanomas that are neither suitable nor favorably located for ruthenium brachytherapy.
Design Retrospective, observational case series.
Setting Primary care center.
Participants Seventy-eight patients with uveal melanoma were treated.
Intervention Between June 3, 2003, and March 18, 2008, patients with uveal melanoma received SDRT monotherapy (group 1, 60 patients) or SDRT combined with tumor resection (group 2, 18 patients). Radiotherapy was performed with a tumor-surrounding dose of 25 Gy on a linear accelerator.
Main Outcome Measures Local tumor control, eye preservation, visual results, and radiation complications.
Results Within a median follow-up of 33.7 months (range, 0.13-81.13 months), 6 recurrences occurred in group 1; none recurred in group 2. The Kaplan-Meier estimate for local control was 85% at 3 years in group 1 and 100% in group 2 (P = .22). Eye preservation rate was 77% vs 87% at 3 years (groups 1 and 2, respectively) (P = .82). Visual acuity decreased with a median loss of −18 Snellen lines (group 1) and −22 Snellen lines (group 2). More retinopathies (P = .07), opticopathies (P = .27), and rubeotic glaucomas (P = .10) occurred in group 1. No significant difference was observed in the development of metastases (P = .33). The groups differed in overall survival because of 2 deaths occurring shortly after surgery in group 2 for unexplained reasons (P = .06).
Conclusions and Relevance Survival analysis suggested that SDRT with combined tumor resection might be associated with increased tumor control and fewer radiation complications than SDRT as monotherapy. Both groups had similar eye retention rates and were comparable concerning the decrease in visual function in most eyes. However, the protocol was stopped after 3 unexplainable deaths after surgery.
In the past, enucleation was the only treatment for uveal melanoma (UM) in addition to observation. Eye-preserving treatment strategies (surgery, radiotherapy, and laser treatment) have now become popular.1- 9 Local tumor control rates currently reach more than 90%.2,4- 6,8,9Small and medium-sized UMs are usually treated with transpupillary thermotherapy or plaque brachytherapy. However, large UMs are still a therapeutic challenge. Tumor resection can be performed as an eye-preserving modality. However, because of the concerns of potential tumor cell dissemination during surgery and residual melanoma cells in the scleral bed, resection has been combined with irradiation.10,11 Radiotherapy of large tumors seems to be associated with increased long-term complications, such as proliferative retinopathy with secondary glaucoma, toxic tumor syndrome, or persistent exudative retinal detachment leading to visual loss or enucleation.12,13
Therefore, we planned to investigate the effectiveness of a combined radiotherapeutic and surgical treatment of UM not suitable for treatment with ruthenium plaques owing to their size or location. Because single-dose radiation of the OM431 choroidal melanoma cell line resulted in a higher percentage of cell death than did fractionated irradiation,14 we chose single-dose stereotactic radiotherapy (SDRT) as our approach. Resection of the irradiated tumor was carried out on the basis of the hypothesis that removing the tumor mass would reduce complications. To our knowledge, this is the first report on SDRT performed with a linear accelerator in the treatment of UM.
Originally, we planned an uncontrolled observational study including consecutive patients with UM suitable for SDRT and tumor resection for whom enucleation was the only reasonable therapeutic alternative in our department (UM ≥6 mm thick, touching the optic disc). Our observation that ruthenium plaques (130-Gy apical dose) in tumors touching the optic disc resulted in an increased risk for local recurrence and opticopathy led us to seek experience with external beam radiotherapy in UM with this unfavorable location. After the unexpected and unexplainable deaths of 2 patients on the day of surgery, the study was discontinued. Thereafter, patients desiring eye preservation received SDRT monotherapy.15 Tumor resection was added only in cases of severe adverse effects associated with SDRT. After a third unexplainable death shortly after surgery and more than 1 year after SDRT, the whole treatment concept was abandoned. This work is a retrospective report about all patients treated with SDRT in Tuebingen, Germany. The work was conducted according to the Declaration of Helsinki as revised in Tokyo and Venice and approved by the local ethics committee.
Seventy-eight consecutive patients with primary UM were treated between June 2003 and March 2008. Sixty patients underwent SDRT (group 1). However, in this group, tumor resection was performed in some patients during the course of the follow-up because of toxic tumor syndrome (n = 4), tumor disintegration (n = 2), or local recurrence (n = 2) or as a medical indication (n = 5). In 18 patients, radiotherapy was combined with tumor resection according to the original study protocol after informed consent was obtained from each patient (group 2).
The SDRT was performed on a 6-MV photon-linear accelerator (Mevatron; Siemens) by rotational arcs with an invasive stereotactic localizer (head ring; Brown-Roberts-Wells). The initially used retrobulbar anesthesia was changed to a functional eye immobilization device formed by a modified swimming mask with a carbon fiber rod assembly to position the light spot of a red light–emitting diode (SFH 756; Avago Technologies) via an optical fiber for active fixation by the patient.16 Magnetic resonance imaging and computed tomography images (Siemens) achieved with this device were matched in the treatment planning system (XKnife; Radionics). An infrared camera fixed on the treatment couch permitted monitoring of the eye during irradiation.16 The aimed radiation dose was 25 Gy to the tumor margin with a surrounding isodose between 50% and 90% (eFigure). Different isodose distributions were chosen to minimize individual complication probability.
Surgery was performed as described by Damato,17 with the exception that, in endoresection, the tumor was completely shelled out through an enlarged sclerotomy. Phacoemulsification with intraocular lens implantation was carried out as standard procedure. No hypotensive anesthesia was administered. Endoresection was performed in posterior and peripheral tumors (n = 25). Combined endoresection and exoresection was chosen for tumors with ciliary body involvement (n = 4).
Tumors were categorized according to their location in the eye. These included anterior (ciliary body), anterior and peripheral (anterior to equator with ciliary body involvement), peripheral (anterior to equator), peripheral and posterior (from anterior to equator to posterior pole), and posterior (involved posterior pole near optic disc and macula).
Tumors were additionally classified according to the 7th edition of the TNM staging system.18 Tumor cell type was determined histologically according to the modified Callender classification.19,20
Variables reported as patient and tumor characteristics were age, sex, side of the tumor-bearing eye, visual acuity (VA) before treatment, tumor location, distance to the optic disc and macula, tumor thickness, largest basal diameter (LBD), TNM classification, TNM stages, and histologic cell type. Treatment variables were therapy group, surrounding tumor dose, and surrounding isodose. Primary end points were relapse-free survival and eye preservation rate. Secondary end points were final VA, loss of VA (Snellen lines), occurrence of complications, and metastases-free and total survival. Visual acuity was assessed by using the Snellen VA cards. For statistical analysis, the VA values were converted into the logMAR system. Three continuous or discrete variables were grouped in categories for analyzing subgroups: tumor thickness (0-6.0 mm, 6.1-9.0 mm, 9.1-15.0 mm, and >15.0 mm), LBD (0-12.0 mm, 12.1-18.0 mm, and >18.0 mm), and final VA (≥20/40, 20/50-20/200, and <20/200).
Commercial software (SPSS, version 17.0; SPSS, Inc) was used. Categorical data are reported as numbers and percentages, discrete data as medians with ranges, and continuous data as means (SDs). Variable balance between groups 1 and 2 was determined by univariate variance analysis (discrete and continuous data) or χ2 tests (categorical data). Survival data were calculated using Kaplan-Meier tests and described as percentages at 2, 3, and 5 years after therapy, together with 95% CIs and patients at risk, and significance of univariate subgroup analyses were tested by log-rank test (log-rank test for trends in TNM). Visual outcome was described as the median change of Snellen lines as well as the outcome with respect to the VA categories 20/40 or better, 20/50-20/200, and worse than 20/200. Correlations of distance to the optic disc or macula to loss of Snellen lines were analyzed by calculating the rank correlation coefficient due to the Kendall τ. All data were regarded as explorative. No level of significance was set in advance; thus, the P values reported are merely descriptive.
Median age was 64 years (range, 17-86 years). Mean (SD) tumor thickness was 7.55 (2.77) mm, and LBD was 13.26 (3.47) mm. Two UMs were stage IV.18 In 60 patients the primary treatment was SDRT monotherapy (group 1); in 18 patients SDRT was followed by planned resection of the tumor (group 2), with a mean time interval of 4.8 (3.3) days. The tumor-surrounding dose was 25 Gy in 77 patients and 20 Gy in 1 patient. Median follow-up of the patients was 33.7 months (range, 0.13-81.13 months). Data for each group are reported in Table 1, and a summary of the studies investigating radiotherapy is presented in Table 2.
Six clinically evident progressions in 6 patients occurred during the follow-up, all of them in group 1 (eText). The Kaplan-Meier estimate for local tumor control in the total collective was 95% (95% CI, 90%-100%) at 2 years and 88% at 3 years (95% CI, 78%-97%), with no further recurrences. Local tumor control at 2 and 3 years in group 1 was 94% and 85%, and in group 2, 100% and 100% (log-rank test, P = .22), respectively (Figure, A ).
Sixteen treated eyes were enucleated, 13 in group 1 and 3 in group 2. Enucleation was performed because of tumor recurrence (n = 5), toxic tumor syndrome (n = 1), rubeotic glaucoma (n = 5), tumor disintegration (n = 1), retinopathy (n = 1), expulsive hemorrhage (n = 1), hypotonia bulbi (n = 1), and phthisis bulbi (n = 1). The Kaplan-Meier estimate for eye preservation for the whole collective was 87% (95% CI, 79%-95%) at 2 years and 79% (95% CI, 69%-89%) at 3 years; in group 1, 87% at 2 years and 77% at 3 years; and in group 2, 87% at 2 years and 87% at 3 years (log-rank test, P = .82;Figure, B), respectively. In the univariate analysis, eyes with tumors involving the ciliary body (P = .05), LBD more than 18 mm (P = .005), T4 size category tumors (P = .03), persistent exudation (P = .04), or rubeotic glaucoma (P = .03) were associated with a higher frequency of enucleation. Significantly more radiation complication (RC)–related enucleations were performed in patients with anterior tumors (P = .01), as well as in those who received the tumor-surrounding dose at the 80% isodose (P = .004), experienced persistent exudation (P = .003), or developed rubeotic glaucoma (P < .001). More surgery-related enucleations (ie, expulsive hemorrhage, persistent hypotonia, or painful phthisis bulbi) were performed in tumors with ciliary body involvement (P = .001).
Sixty-six patients had a decline in VA, 4 experienced no change, and 5 developed an improvement at the end of the follow-up period. A median of 18 Snellen lines were lost (range, –41 to +22). One of the 31 patients with a VA of 20/40 or better and 12 of the 63 patients with a VA of 20/200 or better maintained this level. In group 1, this level was retained in 1 of the 25 patients (4%) with a VA of 20/40 or better and 11 of the 48 patients (23%) with VA of 20/200 or better. In group 2, none of the patients with a VA of 20/40 or better and 1 of the 15 patients (7%)with a VA of 20/200 or better retained this level. Patients in group 1 experienced a median loss of –18 Snellen lines (range, –41 to +22); in group 2, the median loss of Snellen lines was –22 (range, –38 to 0) (Table 3).
Mean loss of Snellen lines at the end of the follow-up decreased with increasing distance to the optic disc. Mean loss was –23 in tumors with disc contact, –18 in UM of 2 disc diameters or less, and –16 with more than 2 disc diameters of distance from the optic disc (Kendall τ-b, P = .05). No association between mean loss of Snellen lines and distance to the macula was found.
Fifty-seven patients (73%) had detectable RCs, opticopathy occurred in 19 patients (24%), 33 patients (42%) had retinopathy, 10 patients (13%) developed macula edema, and exudation persisted in 4 patients (5%). In 5 cases (6%), keratopathy occurred. In 12 group 1 patients (20%), cataract was visible. Neovascular glaucoma developed in 12 patients (15%). One patient (1%) reported eyelash loss. Three patients (4%) had pronounced dry eye problems. In 9 patients (15%) in group 1, tumor disintegration occurred, and in 5 patients (8%), toxic tumor syndrome occurred.
The Kaplan-Meier analysis revealed no significant differences between groups 1 and 2 in RCs but suggested a higher frequency of retinopathy, opticopathy, and rubeotic glaucoma in group 1 (Figure, C-E).
Eighteen patients (14 in group 1 and 4 in group 2) developed metastases after a median of 22 months. Two patients had metastases at the time of therapy. In 10 patients no assessment was made.
In the Kaplan-Meier analysis of metastasis after SDRT, 85% (95% CI, 76%-94%) of all patients were free of metastases after 2 years, 76% (95% CI, 65%-87%) after 3 years, and 71% (95% CI, 59%-83%) after 5 years. The Kaplan-Meier estimate for metastasis-free survival was 83% (95% CI, 72%-94%) at 2 years and 73% (95% CI, 61%-86%) at 3 years in group 1. In group 2, these estimates were 92% (95% CI, 78%-107%) at 2 years and 85% (95% CI, 65%-104%) at 3 years (log-rank test, P = .33). Regarding different patient and tumor characteristics, only cell type showed an influence on the occurrence of metastases, with worse prognosis in epithelioid and mixed-cell tumors (P = .01).
Sixty-five patients (83%) were alive at the end of the follow-up period, 13 died during follow-up (6 resulting from metastases; 3, other reasons; and 4, unknown) (Table 4). The Kaplan-Meier estimate for overall survival was 93% at 2 years and 91% at 3 years in group 1 and 73% at 2 years and 67% at 3 years in group 2 (log-rank test, P = .06) (Figure, F). The only variable that influenced survival was the development of metastasis (P = .02). No association between the TNM stage and metastasis-free survival (P = .18) or overall survival (P = .40) was seen. The Kaplan-Meier estimate for melanoma-specific survival was 98% at 3 years for group 1 and 81% for group 2.
Local tumor control at 3 years was 85% for SDRT monotherapy and 100% for combined therapy. The 6 tumor recurrences in group 1 appeared within the first 3 years (mean, 21.16 months). The SDRT monotherapy tended to be less able to destroy all tumor cells. In other studies4,21- 23 reporting results after fractionated stereotactic radiotherapy, local tumor control between 95.9% and 100% was achieved (Table 2). However, in these series, the tumors being treated were much smaller than were those in our analysis. Tumor volume might be important, since larger tumors contain more radiation-resistant cells, or there may be a different control resulting from various intratumoral dose distribution. However, none of those studies4,21- 23 investigated whether specific tumor or radiation criteria influenced local tumor control. In addition, dose fractionation might influence the results. Experimental data from Logani et al14 indicating the superiority of a single dose to a fractionated dose seem to be questionable in patients. Local tumor control after radiosurgery with a gamma knife is reported13,24- 27 to be between 84% and 97%. It is striking that all studies with smaller UMs revealed better control rates.13,24,25,27 Simonová et al,26 while treating tumors with a median thickness of 8.5 mm, had a similar local control of 84%. Only Zehetmayer et al,28 using 1 to 3 fractions of gamma knife irradiation, indicated better local control of 98% in larger tumors with a mean thickness of 7.8 mm. Proton beam radiotherapy (PBRT) of UM has local tumor control rates between 94.8% and 96%.29- 33 Likewise in these studies, tumor thickness was less, thus explaining the better outcome. Conway et al12 achieved local tumor control of 67% after 2 years of PBRT of extra-large UMs (T4 size category UM with mean LBD = 18.7 mm). Our local tumor control rate of 85% with SDRT monotherapy fell between those percentages for smaller tumors29- 33 and that for extra-large tumors.12 In iodine brachytherapy studies, local tumor control was 94% and 91% at 5 years despite treatment of larger tumors.34- 36
The eye preservation rate was 77% at 3 years in group 1 and 87% in group 2. The most frequent reasons for enucleation were tumor progression and neovascular glaucoma. The fractionated stereotactic radiotherapy studies4,21- 23 reported eye preservation rates between 78.6% and 97.4%. In gamma knife therapy, eye preservation was successful in 78% to 94% of patients13,24- 28; the incidence with PBRT therapy was 54% to 92.3%12,29- 33 and, with iodine brachytherapy, 76% to 84%.34,36 These numbers are comparable with our results. The difference in enucleation frequency in different studies seems, among other things, to be affected especially by tumor size and tumor location, leading to variable percentages of neovascular glaucoma. Hirasawa et al37 have shown that radiation doses of 50 Gy equivalent or more to the iris ciliary body and irradiation of the optic disc are significant risk factors for the development of neovascular glaucoma. In the cited articles,4,12,13,21- 36 the most common reasons for enucleation were tumor recurrence and neovascular glaucoma.
A substantial decline in VA occurred in both treatment groups, especially in group 2. The preoperative VA of all patients in group 2 was better than 20/200. Only 1 patient maintained that level postoperatively; all other patients had a VA worse than 20/200. This loss of vision was also reported in other studies analyzing alternative forms of UM radiotherapy.4,12,13,22,23,25,28,29,32,34- 36
Kaplan-Meier plots suggest that more patients in group 1 developed RC, especially retinopathy, opticopathy, and rubeotic glaucoma. Thus, we assume that tumor resection decreases the risk of RC by removing RC-promoting cells in the tumor mass; in this way, it is protective against the development of RC. In the fractionated stereotactic radiation studies of Dieckmann et al21 and Emara et al,22 fewer patients developed retinopathy and more patients developed opticopathy. The rate of neovascular glaucoma in the case series of Emara et al22 was similar to ours, probably because of the exclusive juxtapapillary location of the UM in the Emara et al study. With regard to gamma knife radiotherapy, a high proportion of patients had retinopathy (84%) and neovascular glaucoma (47%) in the study by Haas et al.13 Langmann et al24 reported neovascular glaucoma in 35% of their patients. The higher radiation doses used could be responsible for the increased complication rates despite smaller tumors in comparison with those in our treatment population. Even with PBRT, high percentages of radiation retinopathy,29,33 opticopathy,29,33 and neovascular glaucoma developed12,29 despite smaller tumors treated in the studies of Dendale et al29 and Höcht et al.33 Iodine brachytherapy caused retinopathy more frequently than in our study, and opticopathy and neovascular glaucoma rates were similar or even higher than in other investigations.35,36
The Kaplan-Meier estimate for metastasis-free survival at 3 years was 73% in group 1 and 85% in group 2. Metastasis-free survival achieved with differing interventions was 84.6% to 97% in fractionated stereotactic radiation studies,4,21- 23 65% to 98% in gamma knife studies,13,24,25,27,28 80.6% to 90% in studies using PBRT,12,29 and 70% in a study of iodine brachytherapy.36 On the other hand, a reliable comparison of the different studies with regard to the effect of the treatment on the development of metastasis has limitations because none of the cited studies provided information about monosomy 3 or class 2 expression profile of the tumors. In addition, we know that the development of metastasis in the first years after the treatment of the primary tumor is the result of pretreatment dissemination.38 The Kaplan-Meier estimate for metastasis-free survival of 65% at 5 years in the study by Fakiris et al27 is striking because tumor sizes were small. However, a high percentage of monosomy 3 tumors in the treated population may be responsible for the high metastasis rate at 5 years. In group 1, 91% of the patients were alive at 3 years; the incidence in group 2 was 67% . This difference was caused by the treatment protocol–related deaths shortly after surgery in group 2.
In conclusion, the combined treatment approach using SDRT and tumor resection achieved better local tumor control and fewer RCs than SDRT alone, but we must consider the treatment protocol–related deaths resulting in a higher survival rate in group 1. The rates of local control, eye preservation, and RCs in our study are similar to those achieved in studies of comparable UMs.
Correspondence: Daniela Suesskind, MD, Center for Ophthalmology, Schleichstrasse 12-16, Eberhard Karls University of Tuebingen, 72076 Tuebingen, Germany (firstname.lastname@example.org).
Submitted for Publication: April 26, 2012; final revision received August 23, 2012; accepted August 24, 2012.
Published Online: March 14, 2013. doi:10.1001/jamaophthalmol.2013.697
Author Contributions: Drs Grisanti and Paulsen contributed equally to the work.
Conflict of Interest Disclosures: None reported.