Intraocular Surgery After Treatment of Retinoblastoma

Objectives  To analyze the results of intraocular surgery in patients treated for retinoblastoma and to assess the ocular and systemic outcomes.

Design  Retrospective noncomparative case series.

Patients  Forty-five consecutive patients who underwent an introcular surgery after treatment for retinoblastoma.

Main Outcome Measures  (1) Recurrence of retinoblastoma, (2) need for enucleation, and (3) systemic metastasis. Overall outcome was defined as favorable in the absence of any of these measures and unfavorable in the presence of 1 or more.

Results  Thirty-four patients (76%) underwent a single procedure of cataract surgery, a scleral buckling procedure, or pars plana vitrectomy and 11 (24%) underwent a combination of 2 or more surgical procedures. In all, 16 patients(36%) achieved final visual acuity better than 20/200. Unfavorable outcomes included recurrence of retinoblastoma in 14 patients (31%), enucleation in 16 (36%), and systemic metastasis in 3 (7%). Five patients (20%) who underwent cataract surgery, 5 (63%) who underwent a scleral buckling procedure, and 9 (75%) who underwent pars plana vitrectomy manifested an unfavorable outcome. The median interval between completion of treatment for retinoblastoma and intraocular surgery was 26 months in patients with a favorable outcome vs 6 months in those with an unfavorable outcome.

Conclusions  Intraocular surgery after treatment for retinoblastoma may be justified in certain exceptional clinical situations. Cataract surgery is safe and effective in most cases. However, the need for a scleral buckling procedure and pars plana vitrectomy may be associated with a higher risk for recurrence of retinoblastoma, enucleation, and systemic metastasis, and a cautious approach is warranted.

MANAGEMENT of retinoblastoma is complex and tailored to the individual patient.^1,2 Primary enucleation is still the preferred treatment for advanced unilateral cases.^1,2 The direction in the management of less advanced cases of retinoblastoma has now shifted toward conservative measures aimed at salvaging the eye and possibly vision. External beam radiotherapy was one of the favored modalities for management until the recent emergence of chemoreduction coupled with sequential focal treatment to the eye.^2-6 Focal treatment modalities, including episcleral plaque brachytherapy, cryotherapy, laser photocoagulation, and transpupillary thermotherapy, remain important as organ-conserving treatments.^1,2,7-11

Conservative treatment methods, however, can lead to various ocular complications.^1,2,7-11 The major ocular complications of external beam radiotherapy and episcleral plaque brachytherapy include radiation-induced cataract and retinopathy.^1,7 Radiotherapy can lead to atrophic retinal breaks and consequent rhegmatogenous retinal detachment.¹² The main posterior segment complications of cryotherapy, laser photocoagulation, and transpupillary thermotherapy are retinal vascular obstruction, retinal traction, transient retinal detachment (ablatio fugax), retinal break, rhegmatogenous retinal detachment, and vitreous hemorrhage.^1,2,8-11

Opacification of ocular media, whether from cataract, rhegmatogenous retinal detachment, or vitreous hemorrhage, complicates the management of retinoblastoma by precluding visualization of the tumor and may necessitate enucleation if there is suspicion of tumor recurrence. However, surgical intervention for cataract, rhegmatogenous retinal detachment, or vitreous hemorrhage may be justified in certain exceptional clinical situations, especially if the eye in question is the only potentially seeing eye and the tumor is judged to be clinically stable and in regression.

There is limited information in the literature regarding the ideal timing and expected outcome of intraocular surgery after treatment for retinoblastoma.^13-23 It is not known whether surgery in eyes harboring regressed retinoblastoma allows for a reasonable visual outcome or could be complicated by recurrence of retinoblastoma, need for enucleation, or systemic metastasis.^14-23 Intraocular surgery in these eyes raises genuine concerns about the patient's systemic outcome because of the risks for viable tumor seeding.^24-26 In this article, we review our experience with 45 patients with treated retinoblastoma who subsequently underwent necessary intraocular surgery for cataract, rhegmatogenous retinal detachment, or vitreous hemorrhage and address the ocular and systemic outcomes.

We reviewed the computerized diagnostic records of 900 consecutive patients with retinoblastoma treated on the Oncology Service at Wills Eye Hospital, Philadelphia, Pa, between June 1974 and January 2000. The medical charts of patients who underwent intraocular surgery after treatment for retinoblastoma were selected for detailed analysis.

The collected patient data included age at initial presentation, race(black, Asian, white, or Hispanic), sex (male or female), laterality (unilateral or bilateral), and heredity (hereditary or sporadic). Initial visual acuity, Reese-Ellsworth group (Ia-Vb), tumor number, tumor type (endophytic, exophytic, mixed, or diffuse infiltrative), tumor dimensions, tumor proximity to the optic disc and foveola, the presence of vitreous or subretinal seeds, and the presence of subretinal fluid were recorded. Detailed information was collected regarding initial management (thermotherapy, photocoagulation, cryotherapy, episcleral plaque brachytherapy, external beam radiotherapy, and chemotherapy) and outcome of retinoblastoma (regression and recurrence). The regression of retinoblastoma was judged to be clinically stable by an experienced observer(C.L.S. or J.A.S.). Continued clinical regression for a minimum of 3 months was a prerequisite for considering intraocular surgery. Information regarding intraocular surgery included timing, indications, type (cataract surgery: intracapsular extraction, extracapsular extraction, pars plana/pars plicata/limbal lensectomy, intraocular lens implantation, primary posterior capsulotomy, anterior vitrectomy, or Nd:YAG laser posterior capsulotomy; scleral buckling procedure: drainage of subretinal fluid or anterior chamber paracentesis; or pars plana vitrectomy: number and location of ports), results of intraocular surgery (success, failure, and complications), and the status of the tumor before and after intraocular surgery (not visualized, regressed, or recurrent). If retinoblastoma recurred after intraocular surgery, the time and location of tumor recurrence and the modality of management of the recurrent tumor were recorded. Histopathologic features, including tumor involvement in the orbit, Tenon fascia, cataract surgical incision site, subretinal fluid drainage site, vitrectomy ports, episclera, anterior chamber, iris, vitreous, ciliary body, choroid, and optic nerve, were reviewed if the involved eye underwent enucleation, orbital exenteration, or an orbital biopsy.

Surgical, visual, tumor, ocular, and systemic outcomes were noted on follow-up. Cataract surgery was deemed successful if the visual axis could be cleared of lens opacity, with or without primary or secondary posterior capsulotomy. The scleral buckling procedure was termed successful if the retina could be reattached. The success of pars plana vitrectomy was assessed depending on the indication. Clearance of vitreous hemorrhage if vitrectomy was performed for vitreous hemorrhage, resolution of vitreous exudates if vitrectomy was performed for endophthalmitis, and reattachment of retina if vitrectomy was performed for retinal detachment indicated the success of the procedure. The tumor outcome (regression vs recurrence) was assessed cumulatively during the entire follow-up period, and the visual (Snellen visual acuity better than 20/200 vs 20/200 or worse), ocular (enucleation performed vs enucleation not performed), and systemic (systemic metastasis present vs systemic metastasis absent) outcomes were assessed at the final follow-up visit. Patients were categorized using the most recent intraocular surgery to assess overall outcome. Overall outcome was defined as favorable in the absence of recurrence of retinoblastoma, enucleation, and systemic metastasis and as unfavorable in the presence of 1 or more of the following: recurrence of retinoblastoma, need for enucleation, and systemic metastasis.

Of 900 consecutive patients with retinoblastoma treated on the Oncology Service over a 26-year period, 45 eyes of 45 patients (5%) underwent intraocular surgery after treatment for retinoblastoma. Mean patient age at diagnosis of retinoblastoma was 23 months (median, 12 months; range, 1 month to 31 years). There were 36 white patients (80%), 3 blacks (7%), 3 Asians (7%), and 3 Hispanics(7%); 30 patients (67%) were male and 15 (33%) were female. Retinoblastoma was unilateral in 5 patients (11%) and bilateral in 40 (89%). The mean basal diameter of the largest tumor measured 13 mm (median, 12 mm; range, 2-22 mm), and the mean thickness of the largest tumor was 7 mm (median, 7 mm; range, 2-15 mm). The mean proximity of the largest tumor to the optic disc was 1 mm (median, 0 mm; range, 0-12 mm) and to the foveola was 2 mm (median, 0 mm; range, 0-12 mm). Most eyes (27 [60%] of 45) had advanced tumor, categorized as Reese-Ellsworth group Va or Vb at initial examination.

Initial management of retinoblastoma consisted of cryotherapy in 24 patients (53%), laser photocoagulation in 11 (24%), episcleral plaque brachytherapy in 14 (31%), chemotherapy in 19 (42%), and external beam radiotherapy in 29(64%). Fifteen patients (33%) were treated with a single modality and 30 (67%) received multiple treatment modalities. Mean duration of initial treatment of retinoblastoma was 12 months (median, 9 months; range, 1-57 months). The mean interval between completion of treatment for retinoblastoma and intraocular surgery was 21 months (median, 18 months; range, 3-57 months). The mean interval between documented regression of retinoblastoma and intraocular surgery was 18 months (median, 16 months; range, 3-54 months).

Thirty-four patients (76%) underwent a single procedure of cataract surgery, a scleral buckling procedure, or pars plana vitrectomy and 11 (24%) had a combination of 2 or more surgical procedures. Intraocular surgical procedures included cataract surgery in 34 patients (76%), a scleral buckling procedure in 11 (24%), and pars plana vitrectomy in 12 (27%).

Of 34 patients who underwent cataract surgery, 25 underwent cataract surgery alone and 9 needed a further scleral buckling procedure, pars plana vitrectomy, or both. The surgical approach to cataract included intracapsular cataract extraction in 1 patient with a subluxated cataractous lens, extracapsular cataract extraction in 28, and pars plana lensectomy in 5. A posterior chamber intraocular lens was implanted in 13 patients and primary posterior capsulotomy was performed in 8 patients who underwent extracapsular extraction. In all, 14 patients had a posterior capsule defect. Cataract surgery was successful in clearing the visual axis in 33 patients (97%).

Complications included clinically significant posterior capsule opacification necessitating Nd:YAG laser posterior capsulotomy in 4 patients, rhegmatogenous retinal detachment in 2, vitreous hemorrhage in 1, endophthalmitis in 1, and corneal decompensation in 1. Three patients had preexisting rhegmatogenous retinal detachment and 2 had preexisting vitreous hemorrhage. One patient with a dense posterior capsule opacification required pars plana membranectomy, which was performed in combination with pars plana vitrectomy for vitreous hemorrhage. In all, 4 patients underwent a scleral buckling procedure, 4 underwent pars plana vitrectomy, and 1 underwent both a scleral buckling procedure and pars plana vitrectomy after the initial cataract surgery.

Of 11 patients who underwent a scleral buckling procedure, 5 had undergone previous cataract surgery. Eight patients (18%) underwent a scleral buckling procedure alone, 2 underwent a simultaneous pars plana vitrectomy, and 1 was followed by pars plana vitrectomy. Subretinal fluid drainage was internal in 2 patients and external in 4. Of 5 patients who underwent nondrainage surgery, 1 had a dry tap and no further drainage was attempted and all had anterior chamber paracentesis. Subretinal fluid was available for cytologic examination in 3 patients, none of whom showed retinoblastoma cells. The scleral buckling procedure resulted in retinal reattachment in 8 patients (73%). Of 3 patients who had residual retinal detachment, 1 subsequently underwent pars plana vitrectomy. There was no complication related to the scleral buckling procedure.

Standard 3-port pars plana vitrectomy was performed in 12 patients (27%), 4 of whom had undergone previous cataract surgery, 1 of whom had undergone a previous scleral buckling procedure, and 1 of whom had undergone both cataract surgery and a scleral buckling procedure. Indications for pars plana vitrectomy were vitreous hemorrhage in 8 patients, rhegmatogenous retinal detachment with proliferative vitreoretinopathy in 2, endophthalmitis in 1, and vitreous hemorrhage with rhegmatogenous retinal detachment in 1. Additional procedures performed in combination with pars plana vitrectomy included a scleral buckling procedure in 2 patients, pars plana lensectomy in 2, and pars plana membranectomy in 1. Two patients with radiation retinopathy underwent endolaser panretinal photocoagulation. Findings from vitrectomy fluid cytologic examination were available in 5 patients, of whom 2 showed viable retinoblastoma cells, prompting immediate enucleation. Pars plana vitrectomy was successful (clearance of vitreous hemorrhage in 6 patients, resolution of endophthalmitis in 1, and retinal reattachment in 1) in 8 patients (67%). Neovascular glaucoma occurred in 4 patients (33%) who underwent pars plana vitrectomy.

Mean follow-up in our series was 9 years (median, 8 years; range, 1-30 years) after the initial diagnosis of retinoblastoma. Tumor outcome was assessed cumulatively during the entire follow-up period, and visual, ocular, and systemic outcomes were assessed at the final follow-up visit. Patients were categorized using the most recent intraocular surgery to assess overall outcome. Overall outcome was categorized as favorable or unfavorable using the predefined criteria. Table 1 demonstrates the tumor characteristics and outcome in patients categorized by the most recent intraocular surgery. Table 2 provides outcome data for groups of patients segregated by the specific type of intraocular surgery.

Excluding 16 patients (36%) who underwent enucleation, final visual acuity could be assessed in 29 patients. Visual acuity better than 20/200 was achieved in 16 patients (36%) in all, including 12 (48%) who underwent cataract surgery, 3 (38%) who underwent scleral buckling procedure, and 1(8%) who underwent pars plana vitrectomy as the most recent intraocular surgery. In 13 patients (29%) overall who attained visual acuity of 20/200 or worse, the main clinically apparent cause for poor visual acuity was a regressed retinoblastoma scar involving the fovea in 6 eyes, radiation retinopathy in 3, optic atrophy in 2, and residual subretinal fluid in 2.

All patients were evaluated for tumor characteristics immediately after intraocular surgery (n = 34) or as soon as the ocular media clarity was attained(n = 11) and were monitored periodically (1- to 6-month intervals) thereafter. Retinoblastoma continued to remain regressed after intraocular surgery in 31 patients (69%). Viable tumor was detected by cytologic examination of the vitrectomy fluid in 2 patients (4%) in whom dense vitreous hemorrhage had precluded visualization of the tumor immediately before intraocular surgery; both patients underwent immediate enucleation. Twelve patients (27%) had clinical evidence of tumor recurrence a mean of 6 months (median, 4 months; range, 1-19 months) after intraocular surgery. In all, 14 patients (31%) had evidence of recurrent retinoblastoma after intraocular surgery. Recurrent retinoblastoma was detected in 5 patients (20%) who underwent cataract surgery, 5 (62%) who underwent the scleral buckling procedure, and 4 (33%) who had pars plana vitrectomy as the most recent intraocular surgical procedure (Table 1). Twelve of 14 patients with recurrent retinoblastoma underwent enucleation (6 immediately after detection of recurrent retinoblastoma and 6 after a trial of one or a combination of episcleral plaque brachytherapy, external beam radiotherapy, subconjunctival chemotherapy, and systemic chemotherapy). One patient with recurrence of retinoblastoma was treated successfully with cryotherapy and photocoagulation, and another with episcleral plaque brachytherapy.

Sixteen patients (36%) underwent enucleation after intraocular surgery. Indications for enucleation were retinoblastoma recurrence (12 eyes) and symptomatic neovascular glaucoma (4 eyes). Enucleation was performed in 5 (20%) of those who underwent cataract surgery, 4 (50%) who underwent a scleral buckling procedure, and 7 (58%) who had pars plana vitrectomy as their most recent surgical procedure(Table 1). Histopathologic examination of the enucleated eyes (n = 16) showed evidence of viable tumor in 11 eyes(69%), all of which were enucleated for clinical evidence of retinoblastoma recurrence. Five patients had histopathologic risk factors for metastasis, including 1 or more of anterior chamber infiltration (3 eyes), choroidal infiltration(3 eyes), and extrascleral extension (1 eye). Four of 5 patients with histopathologic risk factors for metastasis received standard adjuvant chemotherapy.^26,27

Of 45 patients in our series, 42 (93%) were without systemic metastasis at the final follow-up visit. Three patients (7%) (1 from the scleral buckling procedure group and 2 from the pars plana vitrectomy group) died of systemic metastasis at a mean of 44 months (median, 31 months; range, 29-71 months) after intraocular surgery. Of 3 patients who died of metastasis, 1 had bilateral retinoblastoma, with one eye enucleated immediately after the diagnosis and the other eye treated with cryotherapy and external beam radiotherapy for a Reese-Ellsworth group II disease. The patient underwent cataract surgery(extracapsular extraction with primary posterior capsulotomy) followed by a scleral buckling procedure 3 months after retinoblastoma regression. Retinoblastoma recurred 12 months after the scleral buckling procedure, and the eye was enucleated. Anterior segment seeding with retinoblastoma cells found on histopathologic examination prompted adjuvant chemotherapy. However, systemic metastasis developed 30 months after the scleral buckling procedure. Systemic metastasis developed in 2 patients who underwent pars plana vitrectomy. Both had Reese-Ellsworth group V disease treated with multiple modalities, including laser photocoagulation, cryotherapy, episcleral plaque brachytherapy, and external beam radiotherapy, for 2 years. Both had bilateral retinoblastoma, with the worse eye of each patient having undergone primary enucleation. One of these patients developed vitreous hemorrhage after initial retinoblastoma regression. Pars plana vitrectomy was performed within 4 months of regression of retinoblastoma. There was no clinically detectable recurrence of retinoblastoma, but metastasis developed 29 months after pars plana vitrectomy. The other patient developed rhegmatogenous retinal detachment and underwent the scleral buckling procedure and subsequent pars plana vitrectomy 3 months after regression of retinoblastoma. Recurrence of retinoblastoma was noted 1 month after pars plana vitrectomy, which was treated with episcleral plaque brachytherapy and finally enucleation. There was choroidal infiltration with retinoblastoma cells on histopathologic evaluation of the enucleated eye. Systemic metastasis developed 71 months after pars plana vitrectomy.

Overall outcome was favorable in 26 patients (58%) and unfavorable in 19 (42%). Most patients who underwent cataract surgery (20 [80%] of 25 patients) had a favorable outcome compared with 37% (3/8) who underwent the scleral buckling procedure and 25% (3/12) who underwent pars plana vitrectomy. Table 3 depicts differences in clinical presentation, treatment, and outcome between the group with a favorable outcome and the group with an unfavorable outcome. In the 26 patients with a favorable outcome, 12 (46%) had retinoblastoma Reese-Ellsworth group V, with median duration of treatment for retinoblastoma of 4 months; median interval between completion of treatment for retinoblastoma and intraocular surgery of 26 months; and 25 patients (96%) underwent cataract surgery, 4 (15%) underwent the scleral buckling procedure, and 3 (12%) underwent pars plana vitrectomy. Intraocular surgery was successful in all the patients in this group, and 16 (62%) attained visual acuity better than 20/200. In contrast, in the 19 patients with an unfavorable outcome, 15 (79%) had retinoblastoma Reese-Ellsworth group V, with median duration of treatment for retinoblastoma of 12 months; the median interval between completion of treatment for retinoblastoma and intraocular surgery of 6 months; and 9 (47%) underwent cataract surgery, 7 (37%) underwent the scleral buckling procedure, and 9 (47%) underwent pars plana vitrectomy. Only 2 patients (11%) in this group attained visual acuity better than 20/200. Recurrence of retinoblastoma occurred in 14 patients (74%), 6 (32%) needed enucleation, and 3 (16%) developed systemic metastasis.

Application of modern treatment techniques in the management of retinoblastoma has resulted in a decrease in the frequency of enucleation.^1,2 In some patients, preservation of visual acuity after retinoblastoma treatment can be accomplished, and this is important when it is being attempted for the patient's only potentially seeing eye. The main causes for reversible visual loss in such eyes include radiation-induced cataract, rhegmatogenous retinal detachment, and vitreous hemorrhage.^{1,2,7-11,14-23} Management of these problems not only improves visual acuity but also allows for better tumor monitoring. During the past 6 decades there have been several reports of successful and innovative surgical procedures in patients treated for retinoblastoma,^14-23 but there still remains legitimate concern regarding the safety and efficacy of performing intraocular surgery in such patients. This concern prevails because of known risks for tumor dissemination and systemic metastasis after open globe manipulation of eyes with retinoblastoma.^24-26

In 1939, Reese first reported successful intracapsular cataract extraction after radiotherapy for retinoblastoma.¹³ Since then, extracapsular cataract extraction, intraocular lens implantation, and numerous other innovations in pediatric cataract surgery have evolved.¹⁹ In 1990, our group contributed to a multicenter collaborative study¹⁶ of 38 patients with radiation-induced cataract who underwent surgery a mean of 29 months after external beam radiotherapy. Eleven patients included in this earlier series are part of the present series. There was retinoblastoma recurrence in 3 eyes (8%), necessitating enucleation of 2 eyes. In 1 case, orbital exenteration was performed for subconjunctival retinoblastoma recurrence that developed at the site of cataract incision. Retinoblastoma recurrence was confined primarily to eyes with persistent vitreous haze or vitreous hemorrhage at the time of surgery. There was no systemic metastasis in this series. Use of a limbal approach and avoidance of primary posterior capsulotomy and scleral incision was recommended. Portellos and Buckley¹⁹ later demonstrated the safety of extracapsular cataract extraction and posterior chamber intraocular lens implantation in combination with pars plana posterior capsulotomy and anterior vitrectomy in a series of 8 patients (11 eyes) with radiation-induced cataract after retinoblastoma treatment. In their series, surgery was performed a mean of 54 months after external beam radiotherapy, and mean follow-up after cataract surgery was 20 months. They did not report retinoblastoma recurrence or systemic metastasis in their patients.

In the present study, 25 patients underwent cataract surgery alone, 13 of those with intraocular lens implantation, a mean of 26 months after the final treatment for retinoblastoma. Recurrence of retinoblastoma was observed in 5 patients (20%), all of whom underwent subsequent enucleation. The approach to cataract surgery, intraocular lens implantation, and posterior capsule status seemed to have minimal influence on the ocular or systemic outcome. None of the patients who underwent cataract surgery developed metastasis. Thus, modern techniques for cataract surgery are successful and appropriate for radiation-induced cataracts after complete regression of retinoblastoma. However, when realizing a 7% to 20% risk for tumor recurrence, we advise a cautious approach, including clear corneal incision, extracapsular cataract extraction, and preservation of the posterior capsule if possible. The clear corneal incision may reduce the risk of inadvertent conjunctival implantation of viable tumor cells and may allow for direct inspection of the incision site for tumor recurrence (unlike the limbal or scleral incision, which may be obscured by the overlying conjunctival flap). Presence of a posterior capsule defect theoretically increases the risk of dissemination of viable retinoblastoma cells to the anterior chamber and extraocular extension through the incision site. If retinoblastoma regression has been deemed stable for at least 6 to 12 months after cataract surgery, Nd:YAG laser posterior capsulotomy may be cautiously performed where required. Implantation of an intraocular lens by itself may not increase the risk of recurrence of retinoblastoma, enucleation, or systemic metastasis and could be considered for providing optimal visual rehabilitation after cataract surgery.

Since the study by Pruett¹⁴ in 1975, there have been several publications of a successful scleral buckling procedure after treatment for retinoblastoma.^15,18,20-23 Our review of the literature found 18 cases of rhegmatogenous retinal detachment managed with a scleral buckling procedure 1 week to 15 years after treatment for retinoblastoma.^{14,15,18,20-23} External drainage of subretinal fluid was performed in 14 eyes (78%) and nondrainage surgery was performed in 4 (22%). The surgery was successful in reattaching the retina in 14 eyes (78%) and unsuccessful in 4 (22%). Four eyes (22%) developed recurrence of retinoblastoma, all of which underwent enucleation. There was no study of systemic metastasis. In our series of 8 patients who underwent a scleral buckling procedure at a mean of 14 months after completion of treatment for retinoblastoma, the retina was reattached in 6 (75%), comparable to published results, but 5 (62%) developed retinoblastoma recurrence, 4 (50%) needed enucleation, and 1 (12%) developed systemic metastasis. It has been stated that failure to reattach the retina may indicate active retinoblastoma.²⁵ However, we found complete reattachment of the retina in 3 of 5 patients in whom retinoblastoma later recurred. It should be realized that more than 60% of eyes requiring a scleral buckling procedure had Reese-Ellsworth group V disease with vitreous tumor seeds at initial examination. These advanced eyes carry a poor ocular outcome. Considering the high risk of retinoblastoma recurrence in eyes with retinal detachment, it may be prudent to restrict the scleral buckling procedure to only exceptional situations when complete tumor control has been achieved for 6 to 12 months and attempt nondrainage surgery when possible.

Pars plana vitrectomy is rarely performed after treatment for retinoblastoma.^16,17,20-22 Monge and associates¹⁷ described 2 patients who underwent pars plana vitrectomy for complications of external beam radiotherapy. Our review of the literature revealed only 8 patients who underwent pars plana vitrectomy after retinoblastoma treatment for reasons of retinal detachment in 6, vitreous hemorrhage in 1, and persistent vitreous haze in 1. In these 8 patients, retinoblastoma recurred in 1 (13%) and systemic metastasis occurred in none at 1- to 3-year follow-up.^16,17,20-22 Surgery in these patients was performed 3 weeks to 13 years after treatment for retinoblastoma. In our series, pars plana vitrectomy was performed in 12 patients a mean of 15 months after completion of treatment for retinoblastoma. In all cases, the surgery was performed in the patient's only potentially seeing eye in a desperate attempt to provide hope for visual rehabilitation. All eyes had tumor control for a mean of 21 months after completion of treatment of retinoblastoma and had remained so when the ocular fundus was last visualized, no more than 3 months before pars plana vitrectomy. Recurrence of retinoblastoma was detected in 4 patients (33%), 2 of whom had viable retinoblastoma cells on cytologic evaluation of vitrectomy fluid, prompting immediate enucleation. Two patients (17%) developed systemic metastasis. Seven patients (58%) needed enucleation, 3 for retinoblastoma recurrence and 4 for symptomatic neovascular glaucoma. It should be realized that 75% of these eyes treated with pars plana vitrectomy had Reese-Ellsworth group V disease and most had previous chemotherapy and radiotherapy by the time the patient was referred to us or manifested vitreous media opacity. These advanced eyes carry a known poor ocular prognosis. Our results indicate that treated cases of retinoblastoma undergoing pars plana vitrectomy may have a high risk for subsequent enucleation, retinoblastoma recurrence, and systemic metastasis. Thus, this procedure is useful only in exceptional clinical situations.

The optimal interval between completion of treatment of retinoblastoma and intraocular surgery is not clearly established. Based on our results, we believe that 6 to 12 months of cautious observation after complete retinoblastoma regression is necessary before attempting intraocular surgery for visual rehabilitation.²⁰ We realize that proper monitoring of the neoplasm may be especially difficult in an eye with opaque media. In addition, observation for a prolonged interval may not be practical in certain situations, such as retinal detachment occurring in a child's only seeing eye. Tumor status and risks of surgery must be assessed individually in such cases and discussed in detail with the family before arriving at a decision.

Our study involves patients with retinoblastoma managed during a period of 26 years. Several changes have occurred in the techniques of cataract surgery and pars plana vitrectomy during this period, which might affect the outcome. Because of the small number of patients in our series, we are unable to analyze this issue. Moreover, most patients (82%) in our series who underwent intraocular surgery were treated in the past 15 years, and with surgical techniques comparable to the current standard in the given clinical situation. The success of intraocular surgery in clearing lens opacity from the papillary axis, reattaching the detached retina, clearing vitreous hemorrhage, and resolving endophthalmitis, was reasonably good (87%) in our series. Visual recovery after intraocular surgery, however, depends largely on factors such as the location of the scar of regressed retinoblastoma, the presence of radiation retinopathy, or the presence of optic atrophy. Excluding those who underwent enucleation, 13 patients in our series (29%) achieved final visual acuity of only 20/200 or worse. On the other hand, satisfactory visual results (visual acuity better than 20/200) were achieved in 16 patients (36%), in most of whom it was their only potentially seeing eye.

Recurrence of retinoblastoma after intraocular surgery is a potentially serious problem. Tumor recurrence has been reported to range from 0% to 45% after various intraocular procedures.^14-23 Recurrence of retinoblastoma was observed in 14 patients (31%) in our series at a mean of 6 months after intraocular surgery. Most recurrences occurred within the first year, with the longest interval being 19 months in our series. Importantly, patients needing a scleral buckling procedure or pars plana vitrectomy seemed to be at greater risk for retinoblastoma recurrence compared with those needing cataract surgery. We do not imply that the particular intraocular surgical procedure precipitated the recurrence of retinoblastoma. Conversely, eyes that need a scleral buckling procedure or pars plana vitrectomy may have advanced retinoblastoma, with higher risk for recurrence. Close monitoring of ocular and systemic status is warranted in these children.

The anatomic and functional outcomes of intraocular surgery in eyes treated with retinoblastoma should be carefully considered before planning any surgery. Enucleation for retinoblastoma recurrence or failed surgery is ultimately required in 20% to 44% of these patients.^14-23 In our series, 16 patients (36%) needed enucleation, 12 for recurrent retinoblastoma and 4 for painful blind eye. Advanced tumor at presentation (Reese-Ellsworth group Va or Vb) and the need for a scleral buckling procedure or pars plana vitrectomy seem to increase the risk for enucleation.

There is concern that surgical violation of an eye with viable or regressed retinoblastoma may predispose to extraocular tumor spread and metastasis.^24-26 Retinoblastoma invasion through the cataract surgery incision has been observed.¹⁶ We found extraocular extension in only 1 enucleated eye in our series, which had previous pars plana vitrectomy. It may be difficult to judge microscopic extraocular extension unless an eye is enucleated or the orbit undergoes biopsy. Cytologic examination of subretinal and vitrectomy fluid is a reasonable precaution in anticipating extraocular spread.²⁰ Prompt enucleation and adjuvant chemotherapy are warranted if there is cytologic evidence of viable tumor cells.^26,27 Of 8 patients who had cytologic examination of subretinal or vitrectomy fluid, results were positive in 2 (25%) for viable retinoblastoma cells, prompting immediate enucleation.

Three patients (7%) in our series had systemic metastasis. It is believed that the presence of anterior chamber, choroidal, postlaminar optic nerve, or extrascleral extension of retinoblastoma on histopathologic examination of the enucleated eye is a risk factor for systemic metastasis.²⁷ We suggest that patients with histopathologic risk factors for metastasis be considered for prophylactic chemotherapy to prevent systemic metastasis.²⁷

Overall outcome was favorable in 26 patients (58%) and unfavorable in 19 (42%). Most patients who underwent cataract surgery (20 [80%] of 25 patients) had a favorable outcome compared with 37% (3/8) who underwent the scleral buckling procedure and 25% (3/12) who underwent pars plana vitrectomy. The 19 patients with an unfavorable outcome included a higher proportion of patients(79%) with retinoblastoma Reese-Ellsworth group V at initial examination, with a longer median duration of treatment for retinoblastoma (12 months) and a shorter median interval between completion of treatment for retinoblastoma and intraocular surgery (6 months) (Table 3). Other factors, including initial tumor characteristics, modality of initial treatment for retinoblastoma, and the need for multiple intraocular surgical procedures, seemed no different in the 2 groups.

In conclusion, in our series of 45 patients who underwent cataract surgery, a scleral buckling procedure, or pars plana vitrectomy at a median of 18 months after completion of treatment for retinoblastoma, intraocular surgery was successful in 87%, and 36% achieved final visual acuity better than 20/200. Retinoblastoma, however, recurred in 31% of patients, enucleation was needed in 36%, and systemic metastasis occurred in 7%. Although surgery for radiation-induced cataract can be gratifying, the need for a scleral buckling procedure and pars plana vitrectomy may be associated with a high risk for retinoblastoma recurrence, enucleation, and systemic metastasis. Despite the modest success in visual outcome and concern for tumor recurrence and ultimate enucleation, intraocular surgery may still be warranted in selected clinical situations, especially if it is the patient's only potentially seeing eye. The clinician should weigh the risk of tumor recurrence, enucleation, and metastasis against the expected benefit of visual rehabilitation and discuss the facts with the family before making a decision. Intraocular surgery should be withheld if the tumor is viable or if there is uncertainty about its activity. Even in patients with documented tumor regression, it may be worthwhile to allow observation of at least 6 months before attempting intraocular surgery. Cataract surgery may be planned with clear corneal incision and posterior capsule salvage, with or without intraocular lens implantation. When indicated, a scleral buckling procedure can be performed, without external drainage of subretinal fluid if possible. Cytologic examination of subretinal and vitrectomy fluid samples can provide direct intraoperative evidence of viable retinoblastoma. Prompt enucleation and adjuvant chemotherapy with or without orbital radiotherapy may be considered in such situations. Close observation is warranted after intraocular surgery for several years to detect possible tumor recurrence and systemic metastasis.

Accepted for publication March 30, 2001.

This study was supported by the Hyderabad Eye Research Foundation, Hyderabad, India (Dr Honavar); Orbis International, New York, NY (Dr Honavar); the Macula Foundation, New York (Dr C. L. Shields); the Eye Tumor Research Foundation, Philadelphia, Pa (Drs C. L. Shields and J. A. Shields); and the Paul Kayser International Award of Merit in Retina Research, Houston, Tex (Dr J. A. Shields).

Corresponding author and reprints: Carol L. Shields, MD, Oncology Service, Wills Eye Hospital, 900 Walnut St, Philadelphia, PA 19107