Uusitalo MS, Van Quill KR, Scott IU, Matthay KK, Murray TG, O'Brien JM. Evaluation of Chemoprophylaxis in Patients With Unilateral Retinoblastoma With High-Risk Features on Histopathologic Examination. Arch Ophthalmol. 2001;119(1):41-48. doi:10-1001/pubs.Ophthalmol.-ISSN-0003-9950-119-1-ecs90103
To identify risk factors for metastatic disease on histopathologic specimens of enucleated eyes from patients with unilateral retinoblastoma, and to evaluate the value of chemoprophylaxis in preventing disease dissemination.
Medical records from patients with unilateral retinoblastoma who underwent primary enucleation were reviewed at the University of California, San Francisco(1977-1998) and Bascom Palmer Eye Institute, University of Miami, Miami, Fla(1991-1998). All routine histopathologic specimens were reexamined. The extent of tumor invasion into the optic nerve or ocular coats and the prescribed chemoprophylactic regimen were recorded.
This retrospective study included 129 patients followed for a median of 54 months. Three patients had tumor invading the sclera. The optic nerve was involved to some extent in 82 patients, 11 of whom had tumor extension beyond the lamina cribrosa. The surgical margin of the optic nerve was involved in an additional 4 patients. The choroid was involved in 43 patients, and was considered massively affected in 12 patients. Anterior segment involvement was observed in 10 patients. Postenucleation chemoprophylaxis was administered to 4 of 4 patients who had tumor cells at the surgical margin of the optic nerve and to 7 of 11 patients with postlaminar disease, all of whom had at least 1 mm of postlaminar tumor extension. External beam radiotherapy was administered to 3/4 and 1/11 of these patients, respectively. Chemoprophylaxis was not administered to patients with choroidal or anterior chamber involvement unless the optic nerve was also involved beyond the lamina cribrosa. One patient with tumor extending to the surgical margin of the optic nerve died of metastatic disease.
Chemoprophylaxis is necessary for patients with tumor extending to the surgical margin of the optic nerve and is likely to be beneficial in preventing metastases in patients with tumor extending beyond the lamina cribrosa. We did not offer chemoprophylaxis to patients with prelaminar optic nerve disease or isolated choroidal involvement, and these patients remained free of disseminated disease.
ALTHOUGH the overall survival rate for retinoblastoma is between 84% and 95%,1- 3 metastatic retinoblastoma is almost invariably fatal. Several studies have sought to identify histopathologic features of enucleated specimens that predict those patients who are at risk for the development of metastatic disease.4 Below and in Table 1 we review all mortality rates that have been previously reported in association with each of several distinct histopathologic risk features.5- 27 The patient populations in all previous reports included both unilateral and bilateral retinoblastoma patients.
Tumor located at the surgical margin of the optic nerve suggests that residual tumor is present in the optic nerve stump following surgical resection. In the literature, this is uniformly considered a major prognostic factor for disease dissemination, with an associated mortality rate between 50% and 81% (Table 1).5- 8,10- 13,15 Invasion of the tumor posterior to the lamina cribrosa, where the meninges insert, seems to be another important risk factor for the development of metastatic retinoblastoma, with a reported mortality rate ranging from 13% to 69% (Table 1).5,6,8,10- 13,17,19,22,23,27 Other series examining postlaminar involvement, however, have not demonstrated an associated increase in mortality.2,16,20,21 Most studies suggest that tumor involvement of the optic nerve anterior to the lamina cribrosa does not significantly increase mortality,2,5,6,8,10,13,16,17,19,21,22,24 although a single study reported a 29% mortality rate associated with this circumstance (Table 1).12 The risk associated with tumor involving the choroid remains most controversial. Some series have suggested that choroidal involvement is a significant risk factor, with an associated mortality rate of 11% to 81% (Table 1),5,6,8,10,15,17,19,22,28 while other studies have failed to confirm these observations.2,11- 13,20,21,24- 26,29 The risk of metastases appears to increase when any degree of both optic nerve and choroidal involvement are present (Table 1).20,26 Tumor involvement of the anterior segment of the eye has also been reported to be a significant risk factor in some studies (Table 1),10,22 but this finding has been less frequently addressed in the literature.
Much effort in recent years has been devoted to developing effective chemotherapeutic regimens for the management of intraocular retinoblastoma. The use of chemoprophylaxis for patients with advanced disease has been addressed much less frequently.2,8,11,14,17,23- 26,30 Indications for chemoprophylaxis in retinoblastoma patients with high-risk features remain undetermined, and treatment centers apply different criteria for initiating such treatment. The purpose of this study was to assess previously described histopathologic risk factors for metastatic disease and to evaluate the potential value of chemoprophylaxis in preventing disease dissemination. We chose to restrict this study to patients with unilateral retinoblastoma because the risk features for metastases in these patients can be definitively determined from histopathologic examination of the enucleated specimen. This is not the case in patients with bilateral disease who have undergone unilateral enucleation, since the risk features for metastases in the remaining eye cannot be determined.
The medical records of all patients treated for unilateral retinoblastoma since the institution of postenucleation chemoprophylaxis at 2 ocular oncology centers (University of California, San Francisco [UCSF], 1977-1998, and Bascom Palmer Eye Institute, University of Miami [BPEI], 1991-1998) were reviewed. During the time periods under consideration at these institutions, 147 patients with unilateral retinoblastoma, who had not received any prior treatment, were treated with primary enucleation. Histopathologic specimens and slides were obtained in 134 cases. We selected 129 of these cases because the duration of follow-up was 12 months or longer. Eighty-one of these patients were from UCSF, and 48 patients were from BPEI.
Routine histopathologic sections of all 129 enucleated globes, including those sections treated with hematoxylin-eosin and periodic acid–Schiff stains, were reexamined by light microscopy by 1 investigator at each institution. A standardized form was used to evaluate specimens for evidence of histopathologic risk features, including tumor involvement of the orbit, sclera, optic nerve, choroid, and anterior segment. Optic nerve invasion was classified as superficial, prelaminar, postlaminar, or involving the surgical margin of the optic nerve. Choroidal involvement was characterized as massive when tumor involved more than one fourth of the choroid. The anterior segment was reported as involved when tumor cells were present in the trabecular meshwork, ciliary body, or iris. Tumor cells within the anterior chamber without evidence of invasion into anterior segment structures were not regarded as significant, as this may be observed as an artifact of processing. The following tumor characteristics were described: growth pattern, presence of necrosis and/or calcification, and degree of tumor cell differentiation. Tumor growth patterns were classified as exophytic, endophytic, mixed exophytic and endophytic, or diffuse. The tumors were classified as well differentiated when areas with well-formed Flexner-Wintersteiner rosettes were present, and poorly differentiated if there was no evidence of rosette formation.
The following clinical data were collected: patient age at diagnosis, sex, involved eye, postenucleation therapeutic regimen, follow-up time, and final outcome. Postenucleation therapies included chemoprophylaxis as well as radiation therapy. Chemotherapeutic agents, doses, and treatment schedules were recorded, as were radiation doses and treatment schedules. Any significant side effects arising from these treatments were recorded. Follow-up time was calculated from the date of enucleation to the date of last contact with the patient. Final outcome for all patients was described as disease-free, orbital recurrence, or metastatic disease. This information was entered into a standardized database at each of the 2 participating institutions.
The median age of the 129 patients at presentation was 22 months (range, 1-119 months). Seventy of the patients were male and 59 were female. The right eye was involved in 58 patients, and the left eye in 71 patients. Median follow-up time was 54 months (range, 12-165 months). One of the 129 patients succumbed to metastatic disease; 128 remained disease-free; and no patient had an orbital recurrence.
Fifty-four of 129 patients had 1 or more of the following high-risk features for which chemoprophylaxis has been recommended: postlaminar invasion of the optic nerve, scleral involvement, choroidal invasion, or anterior segment involvement. Chemoprophylaxis was administered to 11 of these patients. Specific treatment regimens administered to each of these patients are listed in Table 2. Median follow-up of these 54 patients was 50 months (range, 12-165 months).
The optic nerve was involved in 82 patients (Table 3). In 15 of these patients, tumor extended beyond the lamina cribrosa. Four of these 15 patients had tumor extending to the surgical margin of the optic nerve, and all received chemoprophylaxis (patients 1-4, Table 2). One patient in this group (patient 1) developed metastases and died 11 months postenucleation and 2 months after failing to complete a regimen of chemotherapy, which was interrupted by an episode of sepsis and by the family's noncompliance. The other 3 patients remained disease-free with follow-up times of 47, 159, and 26 months (patients 2, 3, and 4, respectively).
Eleven patients had tumor posterior to the lamina cribrosa but not involving the surgical margin. Seven of these 11 patients had tumor extending at least 1 mm beyond the posterior extent of the lamina cribrosa. All of these patients received chemoprophylaxis (patients 5-11, Table 2). One patient (patient 5) also received external beam radiation therapy for gross disease beyond the posterior extent of the lamina cribrosa. Four of the 11 patients had tumor extension less than 1 mm beyond the posterior extent of the lamina cribrosa, and the families of all 4 patients refused chemoprophylaxis. None of these 11 patients developed metastatic disease with a median follow-up time of 41 months (range, 12-79 months).
Twenty-nine patients demonstrated superficial optic nerve disease, and 38 patients demonstrated optic nerve involvement anterior to the lamina cribrosa. None of these patients received chemoprophylaxis, and no metastasis occurred in these groups with median follow-up times of 76 and 50 months (ranges, 18-165 and 12-163 months), respectively.
None of the patients reviewed demonstrated orbital disease. Three patients had scleral involvement. None of these patients received chemoprophylaxis, and all were disease-free at follow-up times of 19, 35, and 85 months. Forty-three patients had choroidal involvement, which was not massive in 31 patients. Two of these 31 patients also had tumor extension to the surgical margin of the optic nerve, and they received chemoprophylaxis. The other 29 patients had prelaminar optic nerve involvement or no optic nerve involvement, and they were not treated with chemoprophylaxis. All were alive at a median follow-up time of 63 months (range, 12-165 months). Twelve patients had massive choroidal involvement. Of these 12 patients, 1 patient (patient 1) also had optic nerve invasion to the surgical margin and received an interrupted and incomplete course of chemotherapy. This patient was the only one to die in this series. Three of these 12 patients had tumor extension at least 1 mm into the postlaminar optic nerve, and they received chemoprophylaxis. Two of these 12 patients had less than 1 mm of postlaminar invasion, and 6 of these 12 patients had either prelaminar or superficial optic nerve involvement or no optic nerve involvement. None of these 8 patients received chemoprophylaxis, and all were alive at a median follow-up time of 66 months (range, 19-79 months).
Ten patients demonstrated involvement of the anterior segment. Two of these patients also had tumor extending to the surgical margin of the optic nerve, and 2 had postlaminar disease. Three of these 4 patients received chemoprophylaxis. None of the other 6 patients received chemoprophylaxis. No metastases occurred in these 10 patients at a median follow-up time of 50 months (range, 16-159 months).
Of 129 patients in this series, the tumor was exophytic in 35, endophytic in 65, mixed in 25, and diffuse in 3 patients. In 1 patient, the tumor growth pattern could not be definitively determined. Necrosis was present in 120 specimens, and calcification was observed in 87 specimens. The tumor was regarded as well differentiated in 51 patients and poorly differentiated in 78 patients.
Treatment was generally well tolerated. Acute chemotherapeutic side effects, including neutropenia, thrombocytopenia, anemia, diarrhea, and fever, were frequently observed and responded well to medical management. One patient in this series developed severe sepsis, which contributed to a 6-week delay in treatment, as well as to subsequent noncompliance with the chemotherapeutic regimen. This patient was the only patient in this study who succumbed to metastases.
Our study included 129 patients with unilateral retinoblastoma who were treated with primary enucleation. Fifty-four patients had risk factors for which chemoprophylaxis has often been suggested: postlaminar optic nerve, scleral, choroidal, or anterior segment involvement. Eleven patients received chemoprophylaxis: 4 due to involvement of the surgical margin of the optic nerve and 7 due to involvement of the postlaminar optic nerve. Chemoprophylaxis was not administered for other histopathologic risk factors. Only 1 patient with optic nerve invasion to the surgical margin and massive choroidal invasion succumbed to metastatic disease. All others remained disease-free.
The use of chemoprophylaxis has been recommended for selected groups of patients at high risk for disseminated disease, but no guidelines for administration of chemoprophylaxis have received universal acceptance. Chemoprophylaxis has been recommended for patients with tumor extending to the surgical margin of the optic nerve,2,8,14,24 and for those with postlaminar disease,17,30 but not, in general, for patients with prelaminar disease.2 Recommendations for chemoprophylaxis in cases of choroidal involvement,2,8,11,17,23- 26,30 simultaneous choroidal and optic nerve involvement,26 or tumor cells invading the anterior chamber,17 are less consistent in the literature.
In contrast to previous studies, this series was selected to include patients who had uniform indications for chemoprophylaxis. This is the first study in the literature that has been restricted to patients with unilateral retinoblastoma. We consider this an important distinction, because the sole indication for chemoprophylaxis among patients in this series was the presence of high-risk features on histopathologic examination. Indications for chemotherapy among patients with bilateral retinoblastoma frequently include vision and globe salvage. These patients receive chemotherapeutic regimens which are frequently not comparable to those administered for chemoprophylaxis. Patients with bilateral retinoblastoma were also excluded from this series because differences in histopathology between the 2 eyes could confound assessment of risk factors for disease dissemination. Patients with bilateral disease frequently retain 1 eye, and the possibility that this eye contains significant histopathologic risk factors cannot be excluded. The 2 institutions, UCSF and BPEI, and the time periods studied were chosen because the indications for postenucleation chemoprophylaxis were identical at these institutions during these time periods. Chemoprophylaxis is always offered to patients at these institutions when tumor cells involve the optic nerve beyond the posterior extent of the lamina cribrosa. Adjuvant external beam radiation therapy in combination with chemoprophylaxis is offered to patients with tumor invasion to the surgical margin of the optic nerve. For several years now, systemic administration of vincristine, carboplatin, and etoposide has been the standard chemoprophylactic regimen for these patients. Chemoprophylaxis has not been recommended at these institutions to patients with isolated prelaminar disease, choroidal involvement, or anterior segment involvement.
Previously reported mortality rates for patients with tumor invasion to the surgical margin of the optic nerve have ranged from 50% to 81%.5- 8,10- 13,15 Recently, however, aggressive chemotherapy has been administered to these patients, sometimes with hematopoietic stem cell rescue. Such treatments have successfully reduced mortality rates in these patients to between 20% and 33%.2,14,18 Our study reported 4 patients in this category, and the 25% mortality rate for these patients falls within the previously cited range. In the current series, the single patient who developed metastatic disease did not receive the chemotherapeutic regimen as prescribed. Convincing evidence exists that patients with tumor involvement to the surgical margin of the optic nerve should always be treated with chemoprophylaxis.
Mortality rates between 13% and 69% have been reported for patients with postlaminar disease.5,10,12,13,19 In 1 study in which chemoprophylaxis was provided to 11 patients with postlaminar disease, the mortality rate was 9.1%.2 In our study, we believe chemoprophylaxis contributed to the 0% mortality rate observed for 7 patients with postlaminar disease extending 1 mm or more beyond the posterior extent of the lamina cribrosa; these patients were all treated with chemoprophylaxis. It should be noted, however, that the mortality rate for 4 patients with postlaminar disease extending less than 1 mm beyond the posterior extent of the lamina cribrosa who did not accept chemoprophylaxis was also 0%. Although our study cannot unequivocally confirm the efficacy of chemoprophylaxis in patients with postlaminar disease, based on the excellent survival of patients in this series compared with historical rates of survival, we continue to recommend chemoprophylaxis for patients with tumor extending posterior to the lamina cribrosa.
Chemoprophylaxis has not been suggested for prelaminar optic nerve invasion,2 but it has been recommended when tumor cells invade the anterior segment.17 Among our patients with tumor extension to the superficial or prelaminar optic nerve, or to the anterior segment without concurrent optic nerve disease beyond the lamina cribrosa, none of the 29, 38, or 6 patients, respectively, received chemoprophylaxis, and no metastases occurred. These results support the idea that patients with these histological features do well without chemoprophylaxis, and we do not recommend chemoprophylaxis for patients in these categories.
Choroidal involvement was present in 33% of patients; this falls within the range reported by other investigators.21 A previous study reported that choroidal involvement could be observed in 22% of retinoblastoma cases using routine random sections,25 whereas examination of serial sections demonstrated choroidal involvement in 62% of cases.31 In this study, we evaluated routine random sections only, as this is the standard method used to generate histopathologic reports on which clinical decisions are based.
The presence of choroidal involvement has been controversial as a risk factor for tumor dissemination and mortality. Several large series have reported choroidal involvement to be a significant risk factor, with an associated mortality rate between 11% and 81%.5,6,8,10,15,17,19,22,28 Other large series have failed to confirm these findings.2,11- 13,20,21,24- 26,29
Rootman et al,8 Khelfaoui et al,17 and Howarth et al24 recommend chemoprophylaxis when massive choroidal involvement is observed. Massive choroidal involvement and invasion beyond the optic nerve head are highly related. In the present study, 6 (50%) of 12 patients with massive choroidal involvement also demonstrated involvement of the optic nerve beyond the lamina cribrosa, which was an independent indication for chemoprophylaxis. This was the case in only 2 (6%) of 31 patients with nonmassive choroidal involvement. Our series included 29 patients with nonmassive choroidal involvement, either with associated prelaminar optic nerve involvement or without optic nerve involvement. Six of our patients demonstrated massive choroidal involvement, either with associated prelaminar optic nerve involvement or without optic nerve involvement. None of these patients received chemoprophylaxis, and all are alive without disease dissemination. The only patient of 43 patients with choroidal involvement who developed metastases had massive choroidal involvement and tumor extension to the surgical margin of the optic nerve, and was noncompliant with therapy. With so few patients in this category, we cannot definitively evaluate the utility of chemoprophylaxis for isolated massive choroidal involvement.
In an ideal world, the prognostic significance of every histopathologic risk factor should be studied independently and in combination with other risk factors. Even at the busiest ocular oncology centers, however, numbers of patients within each subclassification are small. This difficulty has prompted some investigators to group patients with different histopathologic risk factors together to obtain statistical significance, making a clear-cut examination of individual risk factors difficult. Furthermore, the indications and regimens for chemoprophylaxis have varied over time and among institutions, making accurate comparisons among studies impossible.
Any chemotherapeutic regimen has short-term side effects, and these must be carefully discussed with parents prior to the institution of therapy. Another consideration in choosing chemoprophylaxis is the potential for chemotherapy to increase second tumor risk in patients with underlying germ line mutations at the retinoblastoma locus. In this study, patients with unilateral retinoblastoma were selected. These patients are less likely to carry germ line mutations which predispose them to second tumor development. We do apply the same guidelines to our patients with bilateral disease, however. Metastatic retinoblastoma remains a fatal disease in the majority of cases. Consequently, we support the use of chemoprophylaxis in all cases in which risk for disease dissemination is significant.
Overall survival of patients in this study was unexpectedly high, at 99%. These 2 centers see large numbers of patients from outside the United States who frequently present with advanced disease. Although small numbers of patients in some groups do not allow us to draw definite conclusions, the data suggest that our criteria for chemoprophylaxis are reasonable. For patients who have tumor invasion to the surgical margin of the optic nerve, we also routinely offer external beam radiation therapy.
Additional information is needed to define the value of postenucleation chemoprophylaxis in preventing metastases in retinoblastoma patients with advanced disease at presentation. An international, multi-institutional prospective study, with access to the largest series of patients with retinoblastoma ever compiled, could resolve current uncertainties regarding the indications for chemoprophylaxis, the most appropriate agents and doses, the ideal duration of therapy, and the effect of this treatment on second tumor risk in children with retinoblastoma. We recommend that institutions participating in the proposed international retinoblastoma study collect information toward the development of guidelines for chemoprophylaxis in patients afflicted with advanced forms of this disease.
Accepted for publication April 28, 2000.
Supported by the Mary and Georg C. Ehrnrooth Foundation, Helsinki, Finland; the Finnish Eye Foundation, Helsinki, Finland; The Academy of Finland, That Man May See Foundation, San Francisco, Calif; Research to Prevent Blindness, Inc, New York, NY; and UCSF core grant EY02162 from the National Eye Institute, Bethesda, Md.
Corresponding author: Joan O'Brien, MD, Division of Ocular Oncology, Department of Ophthalmology, University of California, San Francisco, 10 Kirkham St, Box 0730, San Francisco, CA 94143-0730 (e-mail: firstname.lastname@example.org).