Noncontiguous Tumor Recurrence of Posterior Uveal Melanoma After Transscleral Local Resection

Objective  To describe the clinical and histopathologic features of noncontiguous tumor recurrence after transscleral local resection of posterior uveal melanoma.

Methods  Chart review was performed to identify patients with noncontiguous tumor recurrences from a series of 494 consecutive patients treated with transscleral local resection for uveal melanoma. The clinical and histopathologic features of noncontiguous tumor recurrences were studied.

Results  Nine cases were identified, for an estimated incidence of 1.8%. The average diameter of the primary tumors was 16.1 mm (range, 9-22 mm) and the average thickness was 9.8 mm (range, 6-14 mm). Noncontiguous tumor recurrences were solitary in 7 cases and multiple in 2 cases, with an average diameter of 5.8 mm (range, 3.0-9.5 mm). Eight of the 9 patients developed tumor recurrences at multiple intraocular sites. With a mean follow-up of 82.1 months (range, 22-171 months), 7 patients had a final visual acuity of counting fingers or worse, and there were 3 deaths from metastatic disease.

Conclusions  Noncontiguous tumor recurrence after local resection appears to be related to intraocular dissemination from the primary tumor, either through the natural course of the disease or secondary to surgical manipulation. Eyes with large primary uveal melanomas developing noncontiguous tumor recurrence have a high risk of developing tumor spread to multiple intraocular sites, and aggressive treatment is warranted.

TRANSSCLERAL LOCAL resection of uveal melanoma has proven to be a viable therapeutic option for the subset of patients with choroidal or ciliochoroidal tumors who are poor candidates for radiotherapy but are highly motivated to avoid enucleation.^1-5 Although a technically demanding procedure, local resection has been successful in conserving the vision of many eyes with posterior uveal melanomas, and numerous nonrandomized studies have demonstrated comparable survival rates between local resection and enucleation.^1-6 Despite recent refinements in surgical techniques that have led to improved clinical results,^3,7 12% to 32% of eyes are ultimately enucleated after local resection, most commonly for local tumor recurrence.^1,2,4,8 Recurrence of tumor at or adjacent to the margins of the surgical coloboma has been reported in 3% to 18% of cases by various authors.^1,2,4,8 Identified risk factors for local tumor recurrence after local resection include epithelioid cellularity, posterior tumor extension near the optic disc or fovea, tumors with a basal diameter greater than 16 mm, and lack of adjunctive plaque radiotherapy.⁸

To our knowledge, there have been no previous descriptions of noncontiguous tumor recurrences (NTR) after local resection surgery for posterior uveal melanomas. The only previous report of this complication is by Duker and colleagues in 1989,⁹ in which they described 4 patients who developed distant tumor recurrences after initially successful cobalt plaque radiotherapy. They estimated the incidence of this finding after brachytherapy to be only 0.68%, which is significantly lower than the 12% to 15% rate of the more typical pattern of tumor recurrence occurring at the primary tumor site.^10,11 In this study, we describe 9 patients with histologically documented uveal melanomas who were treated with transscleral local resection and subsequently developed a tumor recurrence at a site noncontiguous with the margins of the surgical coloboma. To improve our understanding of this unusual finding, the clinical and histopathologic findings for each patient were reviewed and possible mechanisms for the development of NTR were studied.

We reviewed the records of 494 consecutive patients who had been treated by local resection for choroidal and ciliochoroidal melanomas from January 1971 to December 1998, and identified 9 patients who developed tumor recurrences at a site noncontiguous with the surgical coloboma. The computerized patient database used to identify the cases was collected retrospectively from 1971 through 1984 and prospectively from 1985 through 1998. All patients in the database had signed informed consent to allow analysis of clinical data and photographs. Detailed case histories for these 9 patients were then compiled from a review of clinical notes, fundus photographs, and histopathologic specimens. For each patient, the following data from their initial examination were recorded: age, sex, eye involved, visual acuity, coronal and sagittal tumor location, and presence of ocular or systemic predisposing conditions for uveal melanoma.¹² In addition, the histopathologic findings of the primary tumor specimen were studied, including largest tumor diameter and thickness, cell type, and adequacy of surgical margins. Surgical margins were assessed to be clear if normal choroid was present beyond the tumor margin, uncertain if tumor extended close to the excision edge, and positive if tumor was present at the resection edge. Finally, we reviewed the clinical and histopathologic characteristics of the NTR, its course of treatment, and the final visual and systemic status for each patient.

The transscleral choroidectomy technique used in all cases is described in detail elsewhere.¹³ Briefly, a lamellar scleral flap was dissected over the tumor after the tumor margins were defined by transpupillary transillumination. The deep sclera and involved choroid were excised with 3- to 5-mm surgical margins, carefully avoiding retinal damage whenever possible. Since 1986, ocular decompression by pars plana vitrectomy was performed to decrease retinal prolapse during the procedure. Hemorrhage was minimized by lowering the systolic blood pressure to approximately 40 to 60 mm Hg during the choroidectomy portion of the procedure. Preoperative laser photocoagulation and intraoperative diathermy were not performed. Beginning in 1988, prophylactic laser photocoagulation was applied as a double row of confluent burns to the margins of the coloboma for patients who demonstrated positive surgical margins or areas of persistent pigmentation after local resection. Beginning in 1995, adjunctive treatment with a ruthenium Ru 106(¹⁰⁶Ru) radioactive plaque was used routinely to prevent local tumor recurrence at the site of the surgical coloboma. The 20-mm ¹⁰⁶Ru plaque was sutured to the episclera after closure of the lamellar scleral flap, delivering the appropriate dose of irradiation to the surgical site during the immediate postoperative period. For patients who developed retinal detachment during the perioperative or immediate postoperative period, the ¹⁰⁶Ru plaque was applied after removal of the scleral buckle. Routine follow-up for all patients consisted of clinical examinations performed at 1 month and 6 months, and then annually if no complications occurred.

The 9 identified cases represent 1.8% of the 494 consecutive local resection cases performed by our service from 1971 through 1998. Noncontiguous tumor recurrence occurred in 6 men and 3 women, and there were 7 left eyes and 2 right eyes in the study (Table 1). The average age of the patient at presentation was 43.4 years (range, 20-62 years). None of the patients had congenital ocular melanocytosis or other predisposing conditions.

Histopathologic examination of the primary tumor specimens demonstrated an average basal diameter of 16.1 mm (range, 9-22 mm) and an average thickness of 9.8 mm (range, 6-14 mm). Six of the primary melanomas were mixed cell–type tumors and 3 were spindle cell–type tumors. The surgical margins were positive in 5 cases, uncertain in 2, and clear of tumor involvement in 2. Of the 9 local resection procedures, there were 6 choroidectomies, 1 cyclochoroidectomy, and 2 iridocyclochoroidectomies. Prophylactic laser treatment to the margins of the coloboma was performed in 5 of the 9 cases.

The NTRs in the 9 patients were solitary in 7 cases and multiple in 2 (Table 2). The time interval between the local resection and the diagnosis of the NTR was 14 months or less in 6 cases, 30 months in 1 case, and 86 months in 2 cases. The average diameter of the fundus NTR before treatment was 5.8 mm (range, 3.0-9.5 mm). The average distance from the margin of the surgical coloboma to the margin of the NTR was 6.2 mm (range, 4.5-10.0 mm) (Figure 1, Figure 2, and Figure 3). No NTR lesion was directly associated with extrascleral extension, although 1 case demonstrated intrascleral invasion(case 7). Eight of the 9 patients eventually developed tumor recurrences at multiple intraocular sites (Table 2 and Figure 4, Figure 5, and Figure 6).

Eye-sparing treatment of the NTR consisted of ¹⁰⁶Ru plaque brachytherapy in 4 cases, proton beam radiotherapy in 2 cases, and laser treatment in 3 cases. Radiotherapy was successful in inducing tumor regression in all 6 of its cases, while laser treatment failed to cause tumor regression in its 3 cases. Enucleation was eventually performed in 5 cases: in 2 cases for extrascleral extension, in 1 case for an iris tumor recurrence, in 1 case for widespread tumor recurrences in both the anterior and posterior segments, and in 1 case for a noncontiguous lesion that failed to regress after laser photocoagulation.

At the end of the study period in December 1998, the mean follow-up interval for the group of 9 patients was 82.1 months (range, 22-171 months). Seven patients had a final visual acuity of counting fingers or worse in the affected eye, including 5 patients who underwent enucleation, while only 2 patients had 20/200 or better visual acuity. There were 3 deaths from metastatic disease, which occurred 13, 26, and 30 months after the diagnosis of the NTR.

Since Stallard first described his transscleral choroidectomy technique in 1966,¹⁴ several other authors have published their long-term clinical results using slight modifications of this procedure.^1-4,6,8 Although distant tumor recurrence was briefly mentioned in the article by Damato et al,⁸ there have been no formal reports documenting NTR as a complication of local resection surgery for posterior uveal melanomas. An NTR appears to be an unusual finding that occurs primarily in eyes with aggressively growing, large uveal melanomas. All of the patients in our series were seen at initial examination with primary tumors that were considered too large for radiotherapy, with an average basal diameter of 16.1 mm and an average thickness of 9.8 mm. Further evidence of aggressive growth rates for the tumors in our study includes the predominance of mixed or epithelioid cell types (6 patients)¹⁵ and the propensity to develop multifocal tumor recurrences (8 patients). Similar findings were described by Duker and coworkers⁹ in their description of 4 patients who developed NTR after cobalt brachytherapy. In their study, all of the primary tumors had a basal diameter greater than 15 mm or a thickness greater than 8 mm, and all of the NTRs demonstrated rapid tumor growth. We have also treated 1 patient who developed a distant tumor recurrence after proton beam radiotherapy of an 18-mm-diameter choroidal melanoma(unpublished observation, 1996). Although our report represents the largest series of NTRs to date, this complication is not unique to local resection, and we recommend that all patients receiving eye-sparing treatments for large uveal melanomas be carefully monitored for this complication.

There are several plausible mechanisms for the formation of NTR after local resection. These include (1) multifocal melanomas, (2) uveal dissemination from the primary tumor before treatment, (3) tumor seeding secondary to surgical manipulation, and (4) tumor spread from residual melanoma cells after local resection surgery. Multiple uveal melanomas in the same eye are extremely rare, and approximately 10 documented cases have been reported in the literature.^16-24 There appears to be an association between multicentric melanocytic tumors and congenital melanosis oculi,^18,20 as well as a syndrome related to systemic malignant neoplasms.²⁵ None of our patients had either of these predisposing conditions, and the incidence of 1.8% in our series of local resection cases is higher than the expected rate for multifocal melanomas. The majority of our NTRs occurred within 6 mm of the surgical coloboma (7 patients), suggesting that local uveal spread from the primary tumor may have been the predominant mechanism. Dissemination from the primary tumor may have occurred through the natural course of the disease, and choroidal seeds may have already been growing when the patient was initially examined. Alternatively, since some choroidal manipulation is inevitable during local resection surgery, choroidal or suprachoroidal tumor dissemination may have occurred intraoperatively. It is also possible that choroidal spread occurred after surgery, from residual tumor cells left at the margins of the surgical resection.^8,26 Although we cannot exclude the possibility that these patients harbored multifocal melanomas, we believe that the majority of our cases are the result of metastatic uveal dissemination from the primary tumor, either through the natural course of the disease or somehow related to the surgical intervention.

The ability of a primary uveal melanoma to disperse tumor cells that can proliferate at a noncontiguous, intraocular site has been previously documented.^27-34 Aggressively growing ciliary body melanomas can shed viable cells through the vitreous to invade the nonadjacent retina and optic nerve, typically in the setting of secondary glaucoma, which is believed to facilitate the posterior dissemination of tumor cells.^27-31 A similar case has been reported after iridocyclectomy, in which a spindle cell melanoma of the iris and ciliary body appeared to convert into an epithelioid retinoinvasive melanoma.³² Intraocular metastasis from posterior uveal melanomas appears to be more rare, but there have been several reports of choroidal melanomas that have broken through the retina to seed the vitreous, distant retina, or optic nerve.^29,33,34 Several cases in our series provide unequivocal evidence that intraocular metastasis of posterior uveal melanomas is a real phenomenon. One of our patients was noted to have a local tumor recurrence within the surgical coloboma extending into the vitreous (case 8). Two years after this recurrence was treated with an endoresection, the patient was noted to have tumor nodules growing on the optic disc, retina, pars plana, and anterior segment. Another patient in our series underwent a retinectomy and pars plana vitrectomy because of retinal invasion by the primary tumor, and subsequently developed a distant tumor nodule in the retina (case 9). Although it is difficult to determine whether tumor spread in these 2 cases was induced by the surgical intervention, our report does provide further evidence that large posterior uveal melanomas can be associated with intraocular metastasis. Therefore, surgeons should take appropriate precautions to prevent tumor spread when performing either internal or transscleral resections of posterior uveal melanomas.

It is well known that local tumor recurrence after various forms of radiotherapy portends a poor prognosis for patient survival.^10,11,35-37 In our series, 3 of 9 patients with NTR developed systemic metastatic disease. Duker and colleagues⁹ reported that 3 of their 4 patients died of metastatic disease within 7 months after the diagnosis of their NTR. However, the average basal diameter of the distant tumor recurrences in their study was 13.75 mm,⁹ whereas the average diameter of the recurrent lesions in our study was only 5.8 mm. Because of these confounding variables and the small number of patients in both series, there is insufficient evidence to make conclusions regarding the relationship between NTR and systemic prognosis. Tentative data from Damato and coworkers' local resection series⁶ suggest that enucleation for small local tumor recurrence does not increase the risk of death from metastases, while enucleation for large recurrent tumor (>10 mm) may double the rate of metastatic disease. Small tumor recurrences, including most noncontiguous lesions, probably do not serve as a new source of metastatic cells as long as appropriate and prompt secondary treatment is administered

As evidenced in our study, NTR after local resection was associated with a poor prognosis for vision. Seven of our 9 patients had a final visual acuity of counting fingers or worse, with 5 patients ultimately undergoing enucleation. The poor vision was not usually attributable to the NTR, which was typically small and peripheral. Instead, these eyes tended to develop other tumor recurrences that eventually necessitated enucleation, either locally at the coloboma, at multiple intraocular sites, or extending extraocularly. Another concerning feature of these eyes was that they frequently harbored areas of recurrent tumor that were evident only on histopathologic examination. Therefore, we recommend enucleation for eyes developing NTR with nonambulatory levels of vision, as well as for patients who are not willing to accept the high risk of future tumor recurrence. If the patient is highly motivated to keep the eye, further conservative treatment can be considered if there is a single lesion and the eye possesses long-term potential for good vision. An isolated NTR can usually be treated successfully with radiotherapy; the 6 patients who received either ¹⁰⁶Ru plaque or proton beam radiotherapy in our series all experienced successful tumor regression. Although laser photocoagulation was previously thought to be acceptable for small and flat tumor recurrences,^8,38 laser treatment was not found to be an effective mode of therapy or prophylaxis in our study. Because of the possibility of scleral invasion (case 3), we also do not recommend transpupillary thermotherapy as a primary treatment modality for noncontiguous tumor foci.³⁹ An NTR after local resection appears to be a clinical sign of aggressive tumor growth, necessitating close long-term follow-up and definitive treatment of detected tumor recurrences.

Local resection is a demanding procedure for both the patient and the surgeon, and it should be reserved as an alternative to radiotherapy or enucleation for selected patients with large uveal melanomas. The primary mechanism for the formation of NTR after local resection appears to be intraocular dissemination from the primary tumor, either through the natural course of the disease or secondary to iatrogenic factors. Because NTR can occur after noninvasive treatment modalities such as radiotherapy, it is possible that intraocular tumor spread is primarily related to the biological growth potential of large uveal melanomas, rather than the surgical intervention. The significance of this finding for clinicians is that it indicates aggressively growing tumors with a poor prognosis for vision and a high risk of developing tumor recurrences at multiple intraocular sites. Whether this risk for local spread translates into an increased susceptibility for systemic metastasis remains to be elucidated. Although NTR is an infrequent complication after transscleral local resection, it reemphasizes the importance of a cautious approach in the management of large uveal melanomas if eye-sparing surgical resection is undertaken.

Submitted for publication April 19, 2001; final revision received July 22, 2002; accepted August 16, 2002.

This study was supported in part by the Pacific Vision Foundation, San Francisco, Calif.

Corresponding author: Bertil E. Damato, PhD, FRCS, FRCOphth, Royal Liverpool University Hospital, Prescot Street, Liverpool L7 8XP, England (e-mail: bertil.damato@btinternet.com or jwkim333@yahoo.com).