Purpose
To determine how tumor characteristics and radiation dose affect the incidence of radiation maculopathy (RM).
Design
Retrospective, consecutive case series.
Methods
A consecutive case series of 384 uveal melanomas irradiated (mean apical dose, 71.2 Gy) were followed up for a mean 47.2 months. Tumor locations included: 122 (32%) centered anterior to the equator, 27 (7%) equatorial, and 235 (61%) posterior. Tumor sizes were American Joint Committee on Cancer class T1 (n = 180), T2 (n = 150), T3 (n = 47), and T4 (n = 7).
Results
RM occurred in 8 (7%) eyes with anterior uveal melanomas. In contrast, it was found in 82 (41%) eyes with posterior tumors. Multivariate analysis revealed the risk related to posterior location was greater compared with anterior location with a hazard ratio of 6.66 (95% confidence interval [CI], 4.94 to 22.50; P = .0001). Tumor height (> 6.0 mm) also demonstrated a high risk for RM (hazard ratio, 4.5; 95% CI, 2.68 to 10.17; P = .0001). A significant dose-response relationship was found between dose to fovea and RM ( P = .0005, for trend). As compared with a dose of < 35 Gy, the risk of RM was 1.74 (95% CI, 0.98 to 3.1) for doses from 35 to 70 Gy, and the risk of RM was 2.43 (95% CI, 1.48 to 4.0) for doses of 70 Gy or more. Of interest, those anterior melanomas with RM had a mean apical height of 9.4 mm, as compared with a mean height of 3.3 mm for anterior tumors not associated with RM. Visual acuity was preserved if the fovea dose was less than 35 Gy.
Conclusions
This study suggests that tumor location, tumor thickness, and radiation dose to the fovea are risk factors for the development of RM.
Radiation oncologists strive to focus dose within the targeted zone with relative sparing of normal tissues. However, all forms of ophthalmic irradiation risk inducing cataract, radiation retinopathy, and optic neuropathy.
Ophthalmic plaque radiation therapy involves either affixing radiation seeds (radionuclides) within a gold seed carrier or purchasing the seeds as a solid metal plaque encasing the source. These plaques are sewn onto the sclera as to cover the tumors base (plus a 2- to 3-mm margin of normal-appearing tissue). Although the gold will block more than 99.75% of 125 I or 103 Pd irradiation, the episcleral aperture allows radiation to enter the eye.
Once inside the eye, radiation dose deposition is inversely related to the square of the distance from the source, photon energies, and absorption characteristics of exposed tissues. For example, although both are equally affected by the inverse square law, the lower energy photons emitted from a 103 Pd seed (versus 125 I) are absorbed more rapidly in tissue and are less likely to reach most normal ocular structures. In so far as preoperative comparative dosimetry ( 125 I, 103 Pd, and 106 Ru) can be performed to examine the dose to the tumor and normal ocular structures (fovea, optic disc, lens, subjacent sclera, and opposite eye wall), comparative dosimetry can be used to choose and thus reduce dose to selected normal ocular structures.
We examined the basic premise that decreased radiation to the fovea will diminish the incidence of radiation maculopathy (RM). Further, we evaluated tumor size, location, and radiation dose to the fovea as biomarkers (risk factors) for RM. This is important because RM is the most common irreversible cause of radiation-related loss of vision.
Methods
We report on 384 consecutive patients with uveal melanomas treated with 103 Pd. Our methods of diagnosis, informed consent, treatment, and follow-up have been described. Aspects of particular importance to this study include that tumor basal dimensions were determined by ophthalmoscopy, transillumination, fluorescein angiography, and ultrasonography.
Tumor Characteristics
The categories of tumors were grouped according to the location of the tumor apex. Tumors grouped as posterior were centered: posterior, posterior equator, and equator posterior. The categories of anterior intraocular tumors included: equator anterior, ciliary body, iridociliary, and iris. One hundred twenty-two (32%) were centered anterior, 27 (7%) were equatorial, and 235 (61%) posterior to the equator. Tumors were grouped further as either nasal, temporal, or neither nasal nor temporal. Nasal was defined as 1 to 5 o’clock clockwise for the right eye, and 7 to 11 o’clock clockwise in the left eye. Similarly, temporal was defined as 7 to 11 o’clock in the right eye and 1 to 5 o’clock clockwise in the left. Tumors located between 5 and 7 o’clock and between 11 and 1 o’clock were grouped as neither nasal nor temporal.
Melanomas also were tumor–node–metastasis (TNM) classified according to the seventh edition of the American Joint Committee on Cancer staging system, which primarily classifies tumors by height and largest basal diameter. There were American Joint Committee on Cancer stage T1 (n = 180), T2 (n = 150), T3 (n = 47), and T4 (n = 7) tumors.
Exclusion Criteria
Patients were excluded from this analysis because of the inability to be assessed for RM because of less than 4 months of follow-up or lack of data (n = 25), not having a data available (n = 5), and inability to assess the macula secondary to cataract, vitreous hemorrhage, or both (n = 16). Equatorial tumors (n = 27) were excluded from analysis of anterior versus posterior tumor location, and midline tumors were not examined in comparison with nasal versus temporally located tumors.
Visual Acuity
Best-corrected visual acuities were recorded before surgery and at every post-treatment visit using Early Treatment Diabetic Retinopathy Study charts, with certified examiners of the Collaborative Ocular Melanoma Study (COMS). Using COMS visual acuity grading, if visual acuity was counting fingers, the acuity was graded as (distance able to count fingers in feet)/200; if the patient had hand movements or light perception visual acuity, the acuity was graded as 1/800; and if the patient had undergone enucleation or had no light perception visual acuity, the acuity was graded as 00/000.
Radiation Retinopathy
Radiation retinopathy was assessed by ophthalmoscopy, fundus photography, and fluorescein angiography. Each revealed evidence of retinal circulation, focal leakage, cystoid macular edema, and other signs of radiation retinopathy. RM patients were documented as having findings of macular hemorrhage, microangiopathy, neovascularization, cotton wool spots, vascular sheathing, and cystoid macular edema according to the Finger classification.
Radiation Treatment
103 Pd seeds (model 200) were affixed into gold ophthalmic plaques with a thin layer of acrylic fixative and with seed-guide inserts. Dosimetry calculations were comparable with the COMS protocol. We followed the recommendations of the American Association of Physicists in Medicine Task Groups. Seeds were calculated as point sources (no correction for anisotropy), and no attenuation was attributed to the acrylic fixative or for the (0.5 mm thick) gold plaque sidewalls. Back-scatter effects from the plaque’s posterior wall were discounted. Dosimetry was compatible with the National Cancer Institute Brachytherapy Contract Group determinations over time. A specific dose rate constant of 1.09 cGy/hour/mCi for 103 Pd was used, and our 103 Pd radial dose function was obtained from published data. Our prescription point was the tumor–apex (the farthest point of intraocular tumor extension from the inner sclera). This was consistent with The American Brachytherapy Society recommendations.
Surgical techniques for 103 Pd episcleral plaque placement have been described. In addition, intraoperative two-dimensional and three-dimensional ultrasound imaging confirmed plaque placement over the posterior choroidal melanomas. Radiation was delivered to a mean apical radiation dose of 71.2 Gy over 5 to 7 continuous days.
Statistical Analysis
We estimated the hazard ratios of rate of occurrence of RM by comparing different patient groups for each factor by a Cox proportional hazard model. To conduct this model analysis, we established a data set containing covariates of tumor location (anterior, posterior, and equator), dose to fovea (< 35.0, 35.0 to 69.9, and ≥ 70.0 Gy), tumor height (< 3.0, 3.0 to 6.0, and > 6.0 mm), tumor diameter (2.0 to 7.9, 8.0 to 11.9, and ≥ 12.0 mm), dose to optic nerve (< 8.0, 8.0 to 19.9, and ≥ 20.0 Gy), dose to opposite retina (< 2.0, 2.0 to 2.9, and ≥ 3.0 Gy), dose to tumor apex (< 70.0, 70.0 to 79.9, and ≥ 80.0 Gy), follow-up time (months), and RM status (yes = 1 and no = 0). According to the arrangement of data, we estimated corresponding univariate and multivariate-adjusted hazard ratios and 95% confidence intervals (CIs). The null hypothesis of a no-RM risk was evaluated by the Wald chi-square test. A hazard ratio larger than one with P value less than .05 indicated a significantly increased risk of RM.
Results
Follow-up time ranged from 0 (unrelated death at time of therapy) to 204.7 months, with a mean of 47.2 months. The mean time from plaque brachytherapy to diagnosis of RM was 23.2 months (range, 3 to 72 months). This was significant in that most of the patients in this series were followed up for more than 2 years. Specifically, the number of patients who were followed up for at least 12 months, 24 months, and 36 months was 296 (77%), 229 (60%), and 176 (46%), respectively. Four patients required secondary enucleation in the first year, 4 required secondary enucleation within the second year, and 3 required secondary enucleation within the third year. Five patients with poor follow-up returned for with a new diagnosis of RM after more than 1 year between visits, which may slightly lengthen the average time to this outcome.
Tumor Location Affects Radiation Maculopathy
RM developed in only 8 (7%) patients with anterior uveal melanomas. In contrast, 82 (41%) of patients with posterior uveal melanoma were affected. Examining only tumor location (anterior vs posterior) as a possible risk factor, the Fisher exact test revealed that the risk of developing RM was significantly associated ( P < .0001) with tumor or plaque location ( Table 1 ). Of interest, the mean tumor height for anterior versus posterior tumors was 3.7 mm (range, 1.0 to 11.9 mm) versus a relatively equivalent 3.6 mm (range, 1.6 to 11.0 mm), respectively ( Table 2 ).
| Tumor Location | Patients with RM/Patients in Whom RM Could Be Determined |
|---|---|
| Anterior (n = 122) | 8/112 (7.0%) |
| Posterior (n = 235) | 82/199 (41.2%) |
| Hazard ratio (adjusted Cox model) | 6.7 (95% CI, 4.9 to 22.5) |
| P value a | .0001 |
| Mean Foveal Dose (Gy) | |
| Nasal (n = 112) | 44.8 ± 8.1 (SD) |
| Temporal (n = 162) | 128.7 ± 14.2 (SD) |
| P value b | <.0001 |
a Fisher exact test (2 tailed).
| Tumor Location | No. | Mean Tumor Thickness (mm) | SD | P Value b |
|---|---|---|---|---|
| Anterior (n = 122) | 116 | 3.7 | 2.3 | |
| Posterior (n = 235) | 214 | 3.6 | 1.7 | .537 |
| Nasal | 112 | 3.8 | 2.2 | |
| Temporal | 162 | 3.7 | 2.0 | .697 |
a Both equatorial and midline tumors have been excluded from this table.
Tumor location also was found to be significant in that nasal tumor location shifts the source away from the fovea. In this subset, 329 fovea doses were available for review. Location was grouped as follows: nasal (n = 112), temporal (n = 162), and neither nasal nor temporal (n = 55). Mean fovea dose for nasal tumors was 44.8 Gy (standard deviation [SD], 8.1 Gy; range, 1.4 to 221.2 Gy), and mean fovea dose for temporal tumors was 128.7 Gy (SD, 14.2; range, 1.6 to 545.2 Gy). Nasal tumor location was associated with an 83.9-Gy mean decrease in dose to the fovea. Using a t test analysis, the relative risk for tumors in the temporal location was statistically significant ( P < .0001). Similarly, we evaluated the size of the tumors in this subset and found that the mean tumor thickness of nasal tumors was 3.8 mm (SD, 2.2 mm) versus a mean temporal tumor thickness of 3.7 mm (SD, 2.2 mm; Table 2 ). Therefore, as was anterior plaque location, nasal tumor and plaque locations were both generally farther from the fovea, associated with a lower dose to the fovea and less likely to be associated with RM.
Anterior Tumors Associated with Radiation Maculopathy
Despite their anterior location, RM developed in 8 patients with anterior tumors. We examined this subset of patients and found that these anterior tumors had a mean apical height of 9.4 mm (as compared with 3.3 mm for those anterior tumors with which RM did not develop; n = 110). All 8 tumors were in the (relatively posterior) equator anterior location. It is important to note that with none (0%) of the iris, iridociliary, or ciliary body tumors did either RM or optic neuropathy develop.
Visual Acuity is Related to Fovea Dose
We examined visual acuity outcomes in relation to fovea dose ( Table 3 ). Of 304 patients whose initial visual acuity was 20/200 or better, 193 patients received a fovea dose of less than 35 Gy, 44 received a fovea dose between 35 and 69.9 Gy, and 67 received a fovea dose of 70 Gy or more. For those with a fovea dose of less than 35 Gy, the median initial visual acuity was 20/25 (range, 20/16 to 20/200), and the median end visual acuity was unchanged at 20/25 (range, 20/16 to 0/0). For those with a fovea dose between 35 and 69 Gy, the median initial visual acuity was 20/22.5 (range, 20/16 to 20/200), and the median end acuity was 1-line decreased at 20/32 (range, 20/16 to 1/800). For the highest fovea dose (≥ 70 Gy), the initial median visual acuity was 20/32 (range, 20/20 to 20/160), and the median end visual acuity was 20/80 (range, 20/20 to 00/00), representing a doubling of the visual angle.
| Fovea Dose (Gy) | No. | Mean x | SD | Median | P Value a |
|---|---|---|---|---|---|
| < 35.0 Gy | |||||
| Initial visual acuity | 193 | 32.1 | 28.3 | 25.0 | .184 |
| End visual acuity | 188 | 211.0 | 1263.1 | 25.0 | |
| 35.0 to 69.9 Gy | |||||
| Initial visual acuity | 44 | 30.2 | 17.4 | 22.5 | .0707 |
| End visual acuity | 44 | 110.9 | 307.4 | 32.0 | |
| ≥ 70.0 Gy | |||||
| Initial visual acuity | 67 | 45.9 | 42.4 | 32.0 | .0001 |
| End visual acuity | 67 | 749.5 | 2160.6 | 80.0 |
Time to Radiation Maculopathy
We analyzed the time to RM development (months) after plaque brachytherapy. Three hundred fifty-five patient records were available to evaluate the length of time to retinopathy. Patients were excluded if time to retinopathy could not be assessed because of vitreous hemorrhage or cataract (n = 17). The mean time to retinopathy for those eyes in which this complication developed was 22.8 months (SD, 20.2 months; range, 2 to 153 months). Patients who were found to not develop RM were followed up for a mean 45.8 months (SD, 41.3 months; range, 1 to 172 months).
We used the product limit (Kaplan-Meier) method to draw the cumulative RM incidence for the 3 tumor apex locations of anterior, posterior, and equator. Anterior, posterior, and equator categories previously were defined in the Methods section. As shown in the Figure , the cumulative RM incidence curves were significantly different among the 3 tumor locations (log-rank test: chi-square = 47.5, degrees of freedom = 2, P < .0001). The incidence of RM increased with time, reached a peak 5 years after treatment, and then plateaued for posterior tumors. The cumulative risk of incidence of RM for patients with posterior tumors was 56.8% at 5 years after brachytherapy. Anterior tumors had a much flatter curve because of the low incidence of RM. The cumulative risk of RM for patients with anterior tumors was 8.3% at 5 years. Equatorial tumors had a 5-year cumulative incidence of RM between posterior and anterior tumors (25.5%).
Cox Model Analysis
Table 4 lists the distributions of proportion of patients according to tumor and radiation characteristics. Except for the factor of dose to tumor apex, tumor location, height, largest diameter, dose to fovea, dose to optic nerve, and opposite retina dose showed statistically significant differences in the distribution of proportion of patients. These findings indicated the necessity for multivariate adjustment by Cox model.
| Characteristic/Dose | Total No. of Patients | No. of RM | % | P Value |
|---|---|---|---|---|
| Tumor location | ||||
| Anterior | 112 | 8 | 7.1 | <.0001 |
| Posterior | 199 | 82 | 41.2 | |
| Equator | 24 | 3 | 12.5 | |
| Tumor height (mm) | ||||
| < 3.0 | 167 | 33 | 19.8 | .0008 |
| 3.0 to 6.0 | 128 | 40 | 31.3 | |
| > 6.0 | 39 | 19 | 48.7 | |
| Tumor diameter (mm) | ||||
| 2.0 to 7.9 | 70 | 9 | 12.9 | .0034 |
| 8.0 to 11.9 | 156 | 47 | 30.1 | |
| ≥ 12.0 | 108 | 37 | 34.3 | |
| Dose to fovea (Gy) | ||||
| < 35.0 | 191 | 26 | 13.6 | <.0001 |
| 35.0 to 69.9 | 44 | 19 | 43.2 | |
| ≥ 70.0 | 61 | 36 | 59.0 | |
| Dose to optic nerve (Gy) | ||||
| < 8.0 | 85 | 3 | 3.5 | <.0001 |
| 8.0 to 19.9 | 52 | 8 | 15.4 | |
| ≥ 20.0 | 159 | 70 | 44.0 | |
| Dose to opposite retina (Gy) | ||||
| < 2.0 | 101 | 17 | 16.8 | .0001 |
| 2.0 to 2.9 | 92 | 21 | 22.8 | |
| ≥ 3.0 | 95 | 41 | 43.2 | |
| Dose to tumor apex (Gy) | ||||
| < 70.0 | 15 | 6 | 40.0 | .3240 |
| 70.0 to 79.9 | 139 | 42 | 30.2 | |
| ≥ 80.0 | 179 | 45 | 25.1 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
