Article Text

Download PDFPDF

Endoresection in high posterior choroidal melanomas: long-term outcome
  1. J García-Arumí1,2,
  2. M A Zapata1,
  3. O Balaguer1,
  4. A Fonollosa1,
  5. A Boixadera1,
  6. V Martinez-Castillo1
  1. 1
    Hospital Vall d'Hebron, Barcelona, Spain
  2. 2
    Instituto de Microcirugía Ocular (IMO), Barcelona, Spain
  1. Dr M A Zapata, C/Aureli Campmany 2–10, ESC 3, 1, 2, 08172, Sant Cugat del Valles, Barcelona, Spain; zapatavictori{at}


Background: Eyes with high posterior choroidal melanomas are frequently enucleated because of the potential complications of radiotherapy. The aim of this study was to evaluate the safety and efficacy of endoresection at long-term follow-up.

Methods: Retrospective, non-randomised, interventional case series. Thirty-eight patients underwent endoresection. For primary procedures, inclusion criteria were tumour thickness ⩾8 mm, base <15 mm, tumours not exceeding the equatorial area. Endoresection was also undertaken as the salvage procedure in four patients. Main outcomes measured were metastatic disease, survival, local recurrences, visual acuity, enucleation rate, and surgical complications.

Results: Follow-up time ranged from 23 to 129 months (mean 70.63 months). Preoperative visual acuity ranged from “hand-movements” to 20/20 (mean, 20/60). In primary cases, mean tumour thickness was 10.1 mm and mean base diameter 9.9 mm. At the latest visit, 92.1% patients still retained the eye. Final visual acuity ranged from “no light perception” to 20/30 (mean 20/300). Two patients experienced local recurrence before 3 years of follow-up. Melanoma metastatic disease was found in two patients at 5 years of follow-up. Kaplan–Meier survival analysis for all causes was 88.2% at 5 years. Specific survival was 90.9% at 5 years.

Conclusions: At long-term follow-up, the risk of metastasis or local recurrence, and survival rates were similar to other techniques, although comparisons are difficult because of the unusual presentation of this type of melanoma. Further studies and longer follow-up are needed.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Therapeutic modalities for posterior uveal melanoma include photocoagulation, transpupillary thermotherapy, enucleation, and irradiation with plaques and charged particles.16 Peyman and Cohen7 and Lee et al8 described internal resection by pars plana using vitrectomy instruments. Later, Damato et al9 reported treating 52 melanomas by an internal approach, which they termed endoresection.

Among the ocular oncology community, endoresection remains a controversial technique. Endoresection has been described as a primary procedure for tumours that are not expected to do well after more conventional forms of treatment or as a salvage procedure for patients anticipated to respond poorly to other types of salvage treatment.9 11 The thickness of rapidly growing melanomas in the posterior pole is usually greater than the diameter of the base, and because of the morbidity associated with radiotherapy, these tumours are generally treated with removal of the eye by enucleation. This situation has been reported as an indication for endoresection.10 Local tumour recurrence after brachytherapy, transpupillary thermotherapy or photocoagulation has been described as another possible indication for endoresection, although its use in these cases is controversial.9 11 This technique has also been used in an attempt to preserve central vision in the single functional eye of some patients.11

In this report, the results of primary and salvage endoresection are presented. The safety and efficacy of the procedure are assessed, and the surgical techniques used are discussed.


Between March 1997 and February 2006, 38 patients underwent endoresection. All procedures were performed by the same surgeon (JG-A). For primary procedures, the inclusion criteria were tumour thickness 8 mm or greater, base diameter less than 15 mm and posterior tumours not exceeding the equatorial area. The single patient (case 33) with tumour thickness less than 8 mm (7.7 mm) had colon neoplasm with metastatic liver disease at the time of the diagnosis and preferred only the surgical procedure. Endoresection was also undertaken as a salvage procedure in patients with suspected or definite tumour recurrence after conventional treatment, and who were not expected to do well with conventional salvage treatment. Patients were fully informed of all relevant aspects of the procedure, including the unusual nature of the treatment. All patients signed informed consent for the surgery.

Preoperative work-up included determination of best-corrected Snellen visual acuity, a complete ophthalmoscopic examination, and A- and B-scans with measurement of the maximum thickness and diameter of the tumour. Fundus photography and fluorescein angiography were performed in all patients; orbit imaging was carried out when orbital dissemination was suspected. Screening for metastatic disease was done in all cases and included serum biochemistry with liver-function determinations, ultrasound study of the liver, chest radiography and bone scanning.

The surgical technique varied depending on the degree of retinal involvement. If the tumour had not invaded the retina, 20-gauge vitrectomy was followed by posterior hyaloid dissection, 120° anterior retinotomy, and 810 nm diode laser endophotocoagulation (800–100 mW) 2 mm beyond the tumour margins. Melanoma was removed with the vitrectomy probe using a bimanual technique. Intraocular pressure (IOP) was increased to 100 mm Hg for 5 min to inhibit choroidal vessel bleeding. The patient’s blood pressure remained within normal values. Excision began at the apex of the tumour and continued to the scleral bed. Cellular remnants at the scleral bed were photocoagulated with high doses (800–1000 mW) of endodiode laser. The retina was reattached with liquid perfluorocarbon and air. Laser retinopexy endophotocoagulation was performed at the limits of the retinotomy and was followed by fluid–air exchange and silicone oil–air exchange. If the tumour had invaded the retina, the diode laser was applied through the retina, and the tumour and retina were removed together.

Silicone oil was removed at 3 months. Since January 2003, adjunctive brachytherapy treatment has been performed following surgery in primary cases to prevent local recurrence, with the exception of one patient (case 33) who refused radiotherapy. A ruthenium 106 plaque was sutured to the sclera to cover as much of the coloboma as possible. The plaque was removed after a dose of approximately 80 Gy had been delivered to a depth of 3 mm.

Patients were instructed to maintain face-down position for 1 week and were examined at 1 week, 1 month, 3 months and every 6 months thereafter. Ophthalmoscopic examination and screening for metastasis were performed every 6 months. Cytological and histological examinations were carried out in the vitrectomy specimens.

Patient data were analysed with SPSS 12.0 for Windows (SPSS). Kaplan–Meier survival curves were calculated and statistically compared by logrank analysis.


Patient characteristics

The 38 patients (28 men and 10 women) had a mean age of 50.3 years (range 21–77, SD 13.5) (table 1). The follow-up time ranged from 23 to 129 months, with a mean of 70.63 months (SD 30.32). Concurrent disease included diabetes mellitus (one case), hypertension (3), schizophrenia (1), colon carcinoma with metastatic disease (1) and treated breast carcinoma (1). In 89.5% of cases (34 patients), endoresection was performed as primary treatment, and 10.5% (four patients) as salvage treatment. Preoperative best-corrected visual acuity ranged from “hand movements” to 20/20 (mean, 20/60). In primary cases, mean tumour thickness was 10.1 mm (range 7.7–13.5 mm, SD 1.74) and mean base diameter was 9.9 mm (range 5–15 mm, SD 2.7). The tumour was in a juxtapapillary position in five patients, temporal to the optic disc in 21 patients (fig 1), nasally located in eight patients, above the optic disc in two cases (fig 2), and in an inferior location in two patients. All patients presented exudative retinal detachment, and in five cases detachment was bullous. One melanoma was amelanotic. Histological data were available for all 38 tumours, and the evidence of malignancy was unequivocal in all of them. No patients were lost during the follow-up time.

Figure 1 (A) Case 14: Funduscopy findings in a 42-year-old patient with inferotemporal choroidal melanoma, in an amblyopic eye. Base 10 mm, thickness 9.8 mm. (B) Early postoperative funduscopy image showing bleeding at the scleral bed with a submacular haemorrhage. (C) One year after surgery, primary endoresection without brachytherapy.
Figure 2 Case no 27. (A) Funduscopy images of a 43-year-old patient with superior choroidal melanoma, base 11 mm, thickness 11 mm. (B) One year after surgery, primary endoresection with brachytherapy.
Table 1 Patient demographic data

Anatomical and functional outcomes

At the latest visit, 35/38 (92.1%) patients still retained the eye, and three patients had undergone enucleation. Final visual acuity ranged from “no light perception” to 20/30 (mean 20/300) as follows: less than 20/400 in 27 patients, from 20/400 to 20/100 in eight patients, and better than 20/100 in three patients.

Surgical complications

The main postoperative complication was retinal detachment after silicone oil removal, which was done in 36 patients; among them, 10 had retinal detachment (26% of the series). Postoperative retinal detachment consisted of a break at one of the margins of the surgical coloboma in all patients except one, who had a tear in a peripheral lattice. Four patients additionally developed vitreoretinal proliferation. In all these patients, retinal detachment surgery was performed with scleral buckling and pars plana vitrectomy.

Other postoperative complications included bleeding at the scleral bed (100%), early postoperative ocular hypertension in 12 patients (31.5%), epiretinal proliferation in four (11%), postradiation retinopathy in two, severe subretinal fibrosis in one, and two cases of subretinal neovascularisation 2 years after endoresection.

Over the follow-up period, cataract surgery was performed in 32 patients: in 11 of them, with silicone oil removal 3–6 months after endoresection, and in 21, in a third procedure at 3 months to 6 years (mean 20 months). Two patients developed phthisis bulbi.

Local recurrence, metastasis and mortalities

Local recurrence was documented in 2/34 patients before the third year of follow-up (5.8%); enucleation was performed in both cases. No other local recurrence was detected. There were no conjunctival or orbital recurrences. At the start of the study, one patient presented liver metastasis from a colon carcinoma (case 33). At 3 years, there was no new metastatic disease. At 5 years of follow-up, three patients had metastasis to the liver (three of 23, 13%); all of them died at 39, 60 and 63 months after endoresection. Two of these patients had a prior carcinoma, and one died due to liver metastasis of a colon carcinoma (case 33). The Kaplan–Meier survival analysis for all causes was 88.2% at 5 years (SD 0.065). Specific survival for melanoma was 90.9% at 5 years (SD 0.061) (figs 3, 4). There were no statistical differences in survival between patients who received postoperative brachytherapy and those that did not (p>0.1). Twenty-three patients completed at least 5 years of follow-up; all-cause survival in this group was 87%, and melanoma-specific survival was 91.3%.

Figure 3 Kaplan–Meier visual survival curve: all causes of death.
Figure 4 Kaplan–Meier visual survival curve: choroidal melanoma-specific survival.


The surgical technique used in our patients is similar to the techniques reported in other studies.9 10 12 We apply certain surgical measures to minimise the risk of local recurrence and metastasis, as was described in 2001,10 and attribute our relatively low rates of local recurrence and metastasis to these measures. Diode laser applied 2 mm beyond the tumour margins before resection closes the tumour’s feeding vessels and decreases the risk of haematogenous dissemination and bleeding during removal. Bleeding and haematogenous dissemination are additionally averted by increasing intraocular pressure during resection. Some authors reduce ocular blood flow by lowering the systolic blood pressure to 50–60 mm Hg.9 We believe that the safest way to close choroidal circulation is by increasing intraocular pressure, which does not affect cerebral circulation and avoids life-threatening risks. Experimental study with animal models has shown that irreversible retinal damage begins at 97 min after central retinal artery occlusion;13 in our patients, intraocular pressure is increased for only about 5 min. The other measures we used to minimise local, conjunctival and orbital recurrence are similar to those reported in other publications:9 10 diode laser applied to the coloboma bed, fluid–air exchange after removal to eliminate residual cells in the vitreous cavity, lowering intraocular pressure when removing the instruments, and conjunctiva and sclera examination and washing after the procedure. We did not consider adjuvant cryotherapy for the vitrectomy port.

The most interesting findings of this study are the survival, metastatic disease and local recurrence rates for the size of the tumours, as compared with other studies. In the COMS large-tumour trial, the estimated all-cause mortality was nearly 40% at 5 years, and the estimated tumour-specific mortality was 27%.14 15 Considering the mean tumour thickness of 10 mm in our series, patients could be included in the large melanoma group. Nevertheless, mortality in our patients was lower, with 5-year all-cause mortality at 12% and melanoma-specific mortality at 9.1%. When only patients with at least 5 years of follow-up were analysed, the differences were similar (13% all-cause mortality, 8.7% melanoma-specific mortality).

In fact, our series is difficult to compare with the populations of other studies because of the unusual presentation of this type of melanoma. Patients enrolled in the COMS large tumour trial had a mean age of 60 years (50 years in our study), and the mean tumour base diameter was greater than in our study (COMS: 17.2 mm, present study: 9.9). Nevertheless, the mean tumour thickness was smaller (COMS: 9.5 mm, present study: 10.1 mm). Strictly speaking, 18 of our primary cases could be considered as having large tumours by the COMS, and in this group the melanoma-specific mortality at 5 years was 0%. Sixteen primary cases in our series would be considered medium-size melanomas by the COMS, and specific mortality in this group was 14.3% (one death). Four patients underwent endoresection as a salvage procedure, and their specific mortality at 5 years was 33% (one death).

As other authors have mentioned,16 tumour size must be differentiated in two ways. A larger tumour base diameter has been linked to patient mortality, and in most studies tumour thickness has been associated only with increased ocular morbidity.17 Attending to this theory, we can also compare our survival rate with the COMS cases of medium-size tumours, which had a mean base diameter of 11.4 mm (9.9 mm in the present study). The estimated 5-year all-cause mortality for the COMS tumours was 20%, with 10% disease-specific mortality, rates that are comparable with the 12% all-cause and 9.1% disease-specific mortality found in the present study. These data suggest that endoresection is a safe procedure, despite the theoretical risk of dissemination. The medium-sized tumours in the COMS trial had a mean thickness of 4.8 mm, whereas the tumour thickness in our patients was 10.1 mm; hence, once again, comparisons are difficult.

As several authors have mentioned, there are a number of concerns regarding the endoresection approach. The surgeon cannot be sure that the tumour has been completely resected, and there is a theoretical risk of dissemination of malignant cells; in fact, all other tumours are removed en bloc.6 However, if this theory were true, the number of local recurrences and metastases should be higher with this technique. Five-year local recurrence has been reported at 4.2% with iodine-125 and 5.2% with proton beam radiation,18 and 3-year recurrence is described at 42% with transcleral local resection.19 The COMS medium-sized tumour study had a 5-year risk of treatment failure of 10.3% with radiotherapy.20 Our local recurrence rate at 3 and 5 years was 5.8%.

The present study has the limitation of a non-randomised design; hence, we cannot conclude that endoresection has a similar or lower risk of local recurrence as compared with other techniques, but it seems that local recurrence was not comparatively higher with this procedure. In an attempt to prevent local recurrence after surgery, adjunctive treatment with brachytherapy has been performed since January 2003. Brachytherapy seems to inhibit local recurrence by two main mechanisms, first by killing remnant cells in the scleral bed, and second by preventing scleral diffusion of malignant cells.21 In the present study, no statistical differences in 5-year survival were found between patients receiving brachytherapy and those who did not. Nonetheless, this comparison may not be reliable because of the difference in the number of patients studied and the follow-up time. In other studies investigating endoresection, Damato reported two cases of recurrent disease in 41 primary endoresections, and one after salvage endoresection. These authors used laser photocoagulation in areas of increased pigmentation in seven patients because they could not differentiate between reactive pigment epithelial hyperplasia and tumour.9 Kertes et al reported one local recurrence in 32 patients with a mean follow-up of 3.5 years.22 Karkhaneh described one recurrence at the margins of the surgical coloboma and one new tumour focus among 20 patients with a mean follow-up of 89.5 months.12

Metastatic disease occurred in three of our patients at 5 years of follow-up, but one of them had liver metastasis from colon carcinoma and no melanoma metastasis. Among 23 patients who completed 5 years of follow-up (22 survivors, one death), two had melanoma metastasis to the liver (9.1%), a rate comparable with the 5-year rate in the COMS medium-size melanoma study.23 Metastasis at 10 years in the COMS large tumour trial was over 40%.24 Again, we cannot conclude that metastatic disease in endoresection is similar to that of other procedures, but it seems that despite the theoretical risk, endoresection is not associated with more frequent metastatic disease. In the other related studies, Damato reported one death due to metastatic disease at 41 months after surgery, Kertes reported three deaths due to metastatic disease, and Karkhaneh reported one death caused by liver metastasis.

Outcomes for local recurrence and metastasis in our study are similar to those of other reports, but it should be remembered that our patients had larger tumours (in diameter and thickness), and the mean follow-up time was 70.63 months (5.8 years), whereas Damato reported a mean follow-up of 20 months and Kertes, 40.1 months. Karkhaneh had a longer follow-up time in 20 patients with tumours smaller than those in our series. Haddden, Hiscot and Damato found five unknown local recurrences in a series of 12 enucleations.25 In agreement with these authors, we also believe opaque media may contribute to underdiagnose local recurrence.

In an effort to prevent intraoperative dissemination of tumour cells and distant recurrences, some authors have proposed preoperative coadjuvant treatment, Bechrakis et al26 performed proton beam irradiation in 58 patients before endoresection. The mean follow-up time was 21.7 months, and there were no local recurrences, but three patients developed metastatic disease. Shilling et al27 performed stereotactic gamma-knife radiosurgery in 46 cases. Seven patients developed liver metastasis, and six patients died over a short follow-up time of just 410 days. Therefore, at this time, coadjuvant preoperative treatment does not seem to avoid tumour cell dissemination.

The primary goal of endoresection is saving the patient’s life, a secondary goal is globe salvage, and the third goal is preservation of visual acuity. Our functional results are similar to those reported in other endoresection series: in over 15% of patients, a visual acuity greater than 20/200 was preserved.9 22 In another study in which conservative treatments were applied to large melanomas, only 5.6% of patients had visual acuity over 20/200 in the brachytherapy group, but this rate improved to 61.1% in patients treated with transcleral resection.16 Tumour location is also an important factor for functional outcome. In the present study, all tumours were posterior to the equator, and in 55% of cases, the tumour was temporal to the optic disc. Surgery for temporal tumours and its associated complications more frequently affect the macular area and, consequently, visual function than the techniques generally used for tumours in a more anterior location, such as transcleral resection.

The most serious complication was retinal detachment (26%) after silicon oil removal. The incidence of this complication in the related endoresection studies ranges from 9.4% to 32.6%,9 18 and the rates of other commonly described complications, such as cataracts, ocular hypertension and epiretinal proliferation9 10 12 22 are similar to our results. Additional procedures are often needed following endoresection, such as silicon oil removal, or cataract surgery. Patients should be informed of this possibility before giving their consent for endoresection.

In conclusion, endoresection remains a controversial technique, but in large uveal melanomas it may preserve the eye and vision, whereas other forms of conservative treatment are likely to cause severe ocular complications, with enucleation the only alternative treatment. Long-term follow-up of these patients did not show a higher risk of metastasis or local recurrence, and survival rates were similar to other techniques, although comparisons are difficult because of the unusual presentation of this type of melanoma. Further studies and longer follow-up are needed to establish the safety of this procedure, and to consider endoresection a standard of care and not merely an investigational technique.



  • See Editorials, p 1013 and 1015

  • Competing interests: None.

  • Ethics approval: Ethics approval was obtained from Comité ético Hospital Vall d'Hebron, Barcelona, Spain.

  • Patient consent: Obtained.

Linked Articles