CLINICAL SCIENCES. with thermotherapy or cryotherapy is an important

Transcription

1 CLINICAL SCIENCES Macular Retinoblastoma Managed With Chemoreduction Analysis of Tumor Control With or Without Adjuvant Thermotherapy in 68 Tumors Carol L. Shields, MD; Arman Mashayekhi, MD; Jacqueline Cater, PhD; Abdallah Shelil, MD; Steven Ness, MD; Anna T. Meadows, MD; Jerry A. Shields, MD Objective: To evaluate the effectiveness of chemoreduction alone and chemoreduction with thermotherapy for macular retinoblastoma. Design: Prospective, nonrandomized, single-center case series. Setting: Ocular Oncology Service at Wills Eye Hospital of Thomas Jefferson University in conjunction with the Division of Oncology at the Children s Hospital of Philadelphia (Pa). Participants: There were 68 macular retinoblastomas in 62 eyes of 49 patients managed with chemoreduction from January 1995 through January Intervention: All patients received 6 cycles of intravenous chemoreduction using vincristine, etoposide, and carboplatin. The patients were then treated according to 1 of 2 approaches: chemoreduction alone with no adjuvant focal therapy (group A) or chemoreduction combined with adjuvant foveal-sparing thermotherapy to each macular retinoblastoma (group B). Main Outcome Measure: Tumor recurrence. Results: Of the 68 tumors, 28 were in group A and 40 were in group B. A comparison of both groups revealed that the tumors were similar with regard to clinical features. The mean tumor basal dimension was 12.3 mm for group A and 12.1 mm for group B, and the mean tumor thickness was 6.8 mm for group A and 6.1 mm for group B. Tumors in group A occupied a mean of 71% of the macula, and those in group B occupied 74% of the macula. Following treatment, Kaplan-Meier estimates revealed that group A tumors showed recurrence in 25% by 1 year and 35% by 4 years whereas those in group B showed recurrence in 17% by 1 year and 17% by 4 years. All recurrences were treated with additional focal thermotherapy, cryotherapy, or plaque radiotherapy except for 1 that required external beam radiotherapy and 1 that required enucleation, both in group A. Univariate analysis revealed that predictors of tumor recurrence were intraretinal growth pattern (vs endophytic); small tumor basal dimension (less than 3 mm and occupying a smaller percentage of the macula); absence of subretinal fluid, subretinal seeds, and vitreous seeds; and chemoreduction response with less tumor calcification and tumor regression of type 0 (complete disappearance without a scar). By multivariate analysis, the most important factors predictive of tumor recurrence were smaller macular tumor size (judged by percentage of the macula occupied by the tumor), absence of subretinal or vitreous seeds, and unilateral disease. Conclusions: Treatment of macular retinoblastoma with chemoreduction plus adjuvant foveal-sparing thermotherapy provides tumor control of 83% by 4 years, and this is slightly more favorable than chemoreduction alone, which provides control of 65% by 4 years. Tumors most destined for recurrence are small tumors. Arch Ophthalmol. 2005;123: Author Affiliations: Ocular Oncology Service (Drs Shields, Mashayekhi, Shelil, Ness, Cater, and Shields), Wills Eye Hospital, Thomas Jefferson University, Philadelphia, Pa; and the Division of Oncology (Dr Meadows), The Children s Hospital of Philadelphia, Philadelphia. Financial Disclosure: None. CHEMOREDUCTION COMbined with thermotherapy or cryotherapy is an important therapeutic option in the management of retinoblastoma. 1-4 This combination therapy has permitted globe salvage in 85% of less advanced tumors, that is those classified as Reese-Ellsworth groups I to IV, and in 47% of eyes classified as group V. 5 Most prior reports on chemoreduction for retinoblastoma have focused on salvage of the eye from enucleation and external beam radiotherapy using various chemotherapy regimens and focal consolidation based on the Reese-Ellsworth stage of the globe Other authors have investigated the benefits and risks of chemotherapy alone without focal consolidation Subsequent investigators focused on specific control by chemoreduction for eyes with subretinal seeds or vitreous seeds. 14 In this article, we analyze a particularly important subset of retinoblas- 765

2 tomas, those located entirely or partly in the macular region. These tumors and their therapy can impart lifelong visual consequences to the patient In this analysis, we evaluate macular retinoblastoma control with chemoreduction alone and chemoreduction combined with foveal-sparing thermotherapy. From our point of view, the technique of chemoreduction and focal consolidation differs whether or not a tumor is in the macula. For tumors outside the macula, focal consolidation with thermotherapy or cryotherapy is provided to each regressed tumor following initial chemoreduction. 22,23 For retinoblastoma located partly or entirely in the macula, there is debate regarding the proper management strategy. Some clinicians believe that consolidation with thermotherapy should be provided to the entire tumor or at least to the extrafoveal portion (fovealsparing thermotherapy) for best tumor control. Other clinicians believe that chemoreduction alone without thermotherapy is most reasonable, to avoid foveal damage and permanent visual loss from thermotherapy. However, there is worry that the latter approach might be associated with greater tumor recurrence. In this article, we assess our results with both chemoreduction strategies for macular retinoblastoma. METHODS We identified all patients with macular retinoblastoma who were treated initially with chemoreduction (institutional review board No. 582 at the Children s Hospital of Philadelphia) on the Ocular Oncology Service, Wills Eye Hospital, Thomas Jefferson University (Philadelphia, Pa), in conjunction with the Division of Oncology at the Children s Hospital of Philadelphia. The macula was defined as a circular region in the posterior portion of the ocular fundus, measuring 3 mm in radius, centered on the foveola. All tumors involving the macula were included in this analysis. The eligibility criteria for treatment with chemoreduction were children with retinoblastoma in whom either eye would ordinarily require enucleation or external beam radiotherapy for cure of the disease based on published indications. 1,3,9 Patients whose tumors could be properly controlled with conservative methods alone (cryotherapy, laser photocoagulation, thermotherapy, plaque radiotherapy) were not eligible for inclusion in the chemoreduction protocol. Exclusion criteria for treatment with chemoreduction included biomicroscopic evidence of iris neovascularization; neovascular glaucoma; and tumor invasion into the anterior chamber, iris, optic nerve, choroid, or extraocular tissues as documented by clinical, ultrasonographic, and neuroimaging modalities. Exclusion criteria from a systemic standpoint were those patients with evidence of systemic metastasis; prior chemotherapy; or inadequate organ function of the kidney, liver, or ear. Patients who had received prior treatment for retinoblastoma were not included in this analysis. The chemotherapeutic agents included intravenous vincristine (0.05 mg/kg), etoposide (5 mg/kg), and carboplatin (18.6 mg/kg) given on day 0 and etoposide (5 mg/kg) given again on day 1. The duration of treatment was planned for 6-month cycles. The potential risks and benefits of the chemoreduction protocol were discussed with the patient s family, and informed consent was obtained. Ocular oncologic follow-up was provided at examination under anesthesia every 1 to 2 months after initiation of chemoreduction until complete control of the disease was achieved. Thereafter, examinations were provided every 2 to 4 months as needed. At each examination, the status of the individual retinal tumors, vitreous seeds, subretinal seeds, and subretinal fluid was noted and photographic documentation was performed. At cycle 1, all initial data were recorded, chemoreduction was instituted, and no patient received adjuvant therapy to macular retinoblastoma. At cycles 2 through 6, some patients were managed with chemoreduction alone and no adjuvant treatment (group A), but others received adjuvant treatment to the regressed retinoblastoma using thermotherapy (group B). The patients were not randomized to treatment. The patients selected for no adjuvant treatment generally had severe visual deficit in the opposite eye, such as macular retinoblastoma or enucleation. The patients selected for adjuvant treatment generally had a normal macula in the opposite eye with potential for good visual acuity in that eye. The adjuvant thermotherapy was provided using the indirect ophthalmoscopic diode system using 1.2-mm spot size and varied duration and power so that a light gray-white appearance of the tumor could be achieved at the end of the session. In all cases, foveal-sparing thermotherapy was directed to avoid the 1.5-mm area of the fovea and papillomacular bundle from the optic disc to the foveola. All data were collected in a prospective fashion. We evaluated each patient for age at diagnosis, race, sex, and hereditary pattern (sporadic, familial). The eye was assessed for laterality of involvement (unilateral, bilateral), total number of retinal tumors per eye, and Reese-Ellsworth classification. Each macular tumor was assessed for size in basal dimension (mm) and thickness (mm), proximity to foveola (mm), proximity to optic disc margin (mm), extent of macula involved with tumor (%), extent of optic disc overhung by tumor (%), prominent dilated feeder vessels (present, absent), presence of associated seeds (vitreous seeds, subretinal seeds, both, none), and associated subretinal fluid (present, absent). We collected the thermotherapy parameters, including the number of treatment sessions, mean power (milliwatts), and total duration (minutes), for each macular tumor. Each eye was then assessed for macular tumor recurrence and its location and treatment. We calculated the interval from the initiation of chemoreduction to the macular tumor recurrence. The ocular follow-up was continued until the date the patient was last examined or until the date the eye was enucleated. STATISTICAL ANALYSIS We made a comparison of patients in group A vs those in group B to analyze differences in patient, eye, or tumor features between the 2 groups using the t test and Fisher test. We then analyzed the clinical data with regard to the single outcome of macular retinoblastoma recurrence. The effect of each individual clinical variable recorded at the time the patient came to the Ocular Oncology Service and at the end of chemoreduction on the outcome of recurrence was analyzed by a series of univariate Cox proportional hazards regressions. 24 We determined the correlation among the variables by using Pearson correlations. All variables were analyzed as discrete variables (continuous variables were analyzed by grouping them into discrete categories). We first entered variables that were significant on a univariate level (P.05) into the multivariate Cox regression analysis. For variables that showed a high degree of correlation, we entered only 1 variable at a time from the set of associated variables into subsequent multivariate models. A final multivariate model tested variables that were identified as significant predictors (P.05, Wald statistic or 95% confidence interval of the relative risk) from the initial stepwise model as well as variables deemed clinically important for the outcome of macular retinoblastoma recurrence. For validation of 766

3 Table 1. Chemoreduction for Macular Retinoblastoma. Analysis of 68 Tumors in 62 Eyes of 49 Patients: Demographic Features Group A (19 Patients, 25 Eyes, 28 Tumors) Chemoreduction Alone the final multivariable model, each predictor was analyzed 1 at a time along with the variables retained from the initial stepwise model and those deemed clinically significant. In the final model, a predictor was considered a significant risk factor if the 95% confidence interval of its relative risk did not contain a risk value of 1. The analyses accounted for the occurrence of multiple tumors for some patients by using a Cox proportional hazards model adjusting for the repeated measures via the method from Lin and Wei. 25 We determined the time to retinoblastoma recurrence using the Kaplan-Meier life table analysis. RESULTS Group B (30 Patients, 37 Eyes, 40 Tumors) Chemoreduction Plus Adjuvant Foveal-Sparing Thermotherapy Feature Mean age, mo, 8 (7, 1-29) 8 (8, 1-18) (median, range) Sex, No. (%) M 12 (63) 13 (43) F 7 (37) 17 (57) Race, No. (%) White 12 (63) 24 (80) African American 5 (26) 2 (7) Hispanic 1 (5) 2 (7) Asian 1 (5) 2 (7) Heredity, No. (%) Sporadic 16 (84) 25 (83) Familial 3 (16) 5 (17) Laterality, No. (%) Unilateral 3 (16) 10 (33) Bilateral 16 (84) 20 (67) Laterality of macular tumor, No. (%) Unilateral 13 (68) 23 (77) Bilateral 6 (32) 7 (23) Eye affected (n = per tumor), No. (%) Right 10 (36) 21 (53) Left 18 (64) 19 (47) Eye affected (n = per eye), No. (%) Right 10 (40) 21 (57) Left 15 (60) 16 (43) There were 68 macular retinoblastomas in 62 eyes of 49 patients managed with 6 cycles of this chemoreduction protocol between January 1995 and January Group A (chemoreduction alone) consisted of 28 tumors in 25 eyes of 19 patients, and group B (chemoreduction plus foveal-sparing thermotherapy) consisted of 40 tumors in 37 eyes of 30 patients. The demographic features are listed in Table 1. The mean age at treatment was 8 months for both groups, and the hereditary pattern was sporadic in 16 (84%) of 19 group A patients and 25 (83%) of 30 group B patients. The Reese-Ellsworth classification of each eye is listed in Table 2. A description of Table 2. Chemoreduction for Macular Retinoblastoma. Analysis of 68 Tumors in 62 Eyes of 49 Patients: Reese-Ellsworth Classification of Eye Reese-Ellsworth Group Group A (25 Eyes) Chemoreduction Alone, No. (%) Group B (37 Eyes) Chemoreduction Plus Adjuvant Foveal-Sparing Thermotherapy, No. (%) Ia 0 (0) 0 (0) Ib 2 (8) 5 (14) IIa 4 (16) 7 (19) IIb 4 (16) 3 (8) IIIa 5 (20) 9 (24) IIIb 1 (4) 3 (8) IVa 0 (0) 2 (5) IVb 2 (8) 3 (8) Va 4 (16) 2 (5) Vb 3 (12) 3 (8) the clinical features of the macular retinoblastomas is listed in Table 3. There were no statistical differences (t test and Fisher test) between the 2 treatment groups in any of the following features: patient age, race, and sex; tumor laterality; Reese-Ellsworth classification; tumor type (endophytic, exophytic, intraretinal); tumor size (basal dimension and thickness) and location; proximity to the foveola or optic disc; percentage of the macula occupied by tumor; percentage of the optic nerve overhung by tumor; and related subretinal fluid, subretinal seeds, or vitreous seeds. The thermotherapy parameters employed for tumors in group B included mean power of 518 mw (median, 528 mw; range, mw). The mean duration of treatment was 6 minutes per tumor per session (median, 6 minutes; range, 2-11 minutes per tumor per session). The mean number of sessions per tumor was 3 (median, 3; range, 1-6 sessions per tumor). A summary of treatment results for groups A and B is provided in Table 4. Of the 3 tumors that regressed to a type 0 pattern, the mean basal tumor dimension before chemoreduction was 2.0 mm whereas the mean basal dimension of the remainder was 12.6 mm. Of the 11 tumors with 10% or less calcification following chemoreduction, the mean basal dimension was 3.7 mm before chemoreduction whereas those with less than 10% calcification following chemoreduction were 13.8 mm in mean basal dimension before chemoreduction. Table 5 lists the Kaplan-Meier analysis of time to recurrence of macular retinoblastoma and associated seeds for both groups. By the 4-year follow-up, there was 35% tumor recurrence in group A (Figure 1 and Figure 2) and 17% recurrence in group B (Figure 3 and Figure 4). The difference in recurrence rates between the 2 groups was not statistically significant (log rank P=.13). By the 4-year follow-up, 45% of macular tumors showed related subretinal or vitreous seed recurrence in group A and 46% seed recurrence in group B (log rank P=.49). Univariate and multivariate risk factors for tumor recurrence are listed in Table 6. Overall, for both groups, the factors predictive of macular tumor recurrence by multivariate analy- 767

4 Table 3. Chemoreduction for Macular Retinoblastoma. Analysis of 68 Tumors in 62 Eyes of 49 Patients: Comparison of Macular Tumors in Each Group Tumor Features Group A (28 Tumors) Chemoreduction Alone Group B (40 Tumors) Chemoreduction Plus Foveal-Sparing Thermotherapy Mean base, mm (median, range) 12.3 (9.5, ) 12.1 (13.0, ) 0.93 t test Mean thickness, mm (median, range) 6.8 (5.0, ) 6.1 (6.0, ) 0.39 t test Type, No. (%) Endophytic 11 (39) 12 (30) 0.07 Fisher Exophytic 14 (50) 28 (70) Intraretinal 3 (11) 0 (0) Mean distance to optic nerve, mm (median, range) 0.7 (0.0, ) 0.6 (0.0, ) 0.84 t test Mean distance to foveola, mm (median, range) 0.2 (0.0, ) 0.3 (0.0, ) 0.68 t test Mean macular involvement, % (median, range) 71 (90, ) 74 (93, 5-100) 0.68 t test Mean optic nerve overhung, % (median, range) 29 (0, 0-100) 20 (0, 0-100) 0.34 t test Presence of subretinal fluid, No. (%) Yes 19 (68) 30 (75) 0.59 t test No 9 (32) 10 (25) Presence of seeds, No. (%) None 17 (61) 19 (48) 0.33 Fisher Subretinal seeds 2 (7) 18 (45) Vitreous seeds 7 (25) 1 (3) Both subretinal and vitreous seeds 2 (7) 2 (5) P Value Test sis were smaller tumor size (judged by the percentage of macula occupied by tumor), absence of associated subretinal or vitreous seeds, and unilateral disease. No patients developed ototoxicity, renal toxicity, or second cancers from the chemoreduction regimen. COMMENT The treatment of macular retinoblastoma is particularly challenging because of balancing tumor control with the potential for tumor-related and treatment-related visual loss. Options for management of macular retinoblastoma include enucleation, external beam radiotherapy, plaque radiotherapy, laser photocoagulation, thermotherapy, cryotherapy, and chemoreduction with or without adjuvant thermotherapy. 1-4 Most eyes with large macular retinoblastoma, especially if associated with extensive subretinal fluid, subretinal seeds, or vitreous seeds, are managed with enucleation. Those eyes with less advanced retinoblastoma are currently managed with nonenucleation measures, most commonly involving chemoreduction, sometimes followed by focal tumor consolidation. The current chemoreduction regimen appears to lack clinically apparent damage to the surrounding normal retina and, in fact, leaves a regressed retinoblastoma remnant that is approximately 35% smaller in basal dimension than the original tumor, with the potential of improvement in visual acuity. 9 Other options for macular retinoblastoma treatment include plaque radiotherapy, laser photocoagulation, and cryotherapy. Plaque radiotherapy is reserved for solitary or recurrent macular tumor. 26 Plaque radiotherapy has the benefit of avoiding systemic toxicities, but this technique, when used for retinoblastoma of any fundus location, leads to radiation maculopathy in 25% of patients by 5-year followup. 26 Laser photocoagulation and cryotherapy are used less frequently for macular tumors as both can impart direct damage to the adjacent normal retina with profound treatment-related visual loss. Chemoreduction appears to be a favorable treatment for macular retinoblastoma from a visual standpoint. However, the question that remains is whether or not adjuvant tumor consolidation following chemoreduction is necessary for tumor control, especially when balancing the known potential for thermotherapy-related visual loss with the concern for tumor recurrence. A few reports have addressed tumor control following specific chemoreduction regimens for retinoblastoma. 5-14,23,27 Wilson and associates 11 used chemotherapy alone (vincristine and carboplatin) without tumor consolidation for 36 eyes with retinoblastoma for 8 cycles over 6 months. They found complete tumor control in only 8% of eyes, whereas 92% showed failure with progression of retinal tumor, subretinal seeds, or vitreous seeds. Our group later evaluated the specific control of retinal tumor, subretinal seeds, and vitreous seeds per treated eye. 14 We treated 158 eyes with retinoblastoma using vincristine, etoposide, and carboplatin for 6 cycles over 6 months. All retinoblastomas, subretinal seeds, and vitreous seeds showed initial regression. Tumor consolidation following chemoreduction was provided for each retinal tumor, but the vitreous and subretinal seeds were treated with chemoreduction alone without consolidation. Approximately 50% of the eyes with vitreous seeds at presentation showed at least 1 vitreous seed recurrence at 5 years, and 62% of the eyes with subretinal seeds at presentation showed at least 1 subretinal seed recurrence at 5 years. Of the 158 eyes, recurrence of at least 1 retinal tumor was found in 51% eyes by 5 years. A more recent analysis of 457 consecutive retinoblastomas by our group has focused on individual tumor control with chemoreduction with or without focal tumor consolidation. 27 Those tumors treated with chemoreduction alone showed recurrence in 45% by 7-year follow-up whereas 768

5 Table 4. Chemoreduction for Macular Retinoblastoma. Analysis of 68 Tumors in 62 Eyes of 49 Patients: Summary of Treatment Results Results Group A (28 Tumors) Chemoreduction Alone Group B (40 Tumors) Chemoreduction Plus Adjuvant Foveal-Sparing Thermotherapy Mean final tumor base, mm (median, range) 7.2 (6.0, ) 7.5 (7.5, ) Mean final tumor thickness, mm (median, range) 3.0 (2.6, ) 2.9 (3.0, ) Mean final distance to optic nerve, mm, mean (median, range) 1.7 (1.0, ) 1.6 (1.0, ) Mean final distance to foveola, mm (median, range) 1.0 (0.0, ) 0.9 (0.0, ) Mean macular involvement, % (median, range) 40 (40, 0-95) 44 (50, 3-90) Mean final optic nerve overhung, % (median, range) 4 (0, 0-100) 3 (0, 0-100) Regression type, No. (%) Type 0 3 (11) 0 (0) Type 1 8 (29) 2 (5) Type 2 3 (11) 2 (5) Type 3 14 (50) 34 (85) Type 4 0 (0) 2 (5) Tumor recurrence, No. (%) No 19 (68) 34 (85) Yes 9 (32) 6 (15) Site of tumor recurrence: foveal vs extrafoveal, No. (%) Foveal 0 (0) 1 (17) Extrafoveal 3 (33) 4 (67) Diffuse (both) 6 (67) 1 (17) Site of tumor recurrence: central vs peripheral, No. (%) Central 1 (11) 2 (34) Peripheral 2 (22) 3 (50) Diffuse (both) 6 (67) 1 (17) Mean interval between original diagnosis and recurrence of macular tumor, 8.6 (5.8, ) 8.2 (7.6, ) mo (median, range) Treatment of tumor recurrence Thermotherapy 5 0 Cryotherapy 0 1 Plaque 2 5 External beam radiotherapy 1 0 Enucleation 1* 0 Related seed recurrence, No. (%) Subretinal seeds 7 (25) 17 (43) Vitreous seeds 1 (4) 1 (3) Both vitreous and subretinal seeds 3 (11) 0 (0) Mean interval between original diagnosis and recurrence of related seeds, 10.1 (5.8, ) 5.4 (5.1, ) mo (median, range) Mean follow-up, mo (median, range) 28.2 (22.5, ) 25.5 (22.0, ) *Enucleation performed for a large recurrent macular retinoblastoma with extensive recurrent vitreous seeds. those treated with chemoreduction plus thermotherapy, cryotherapy, or both showed recurrence in 22% by 7 years. Gombos and coworkers 12 assessed primary chemotherapy alone for retinoblastoma. They evaluated 78 retinoblastomas treated with chemotherapy alone (vincristine, etoposide, and carboplatin) for 6 to 8 cycles and found complete tumor response in 72% and failure in 28%. No tumor received focal consolidation. Tumors at greatest risk for failure were small tumors under 2 mm in diameter and extramacular tumors. Macular retinoblastoma showed complete control with primary chemotherapy alone in 26 (84%) of 31 tumors over a mean follow-up of 29 months. They did not evaluate control for vitreous and subretinal seeds. There was no Kaplan- Meier analysis of time to recurrence. In this analysis, we focused only on macular retinoblastoma to determine the benefit of adjuvant focal consolidation with thermotherapy. We found that 35% of 28 macular tumors treated with chemoreduction alone (group A) showed recurrence by 4 years compared with 17% of 40 macular tumors treated with chemoreduction plus extrafoveal thermotherapy (group B). Importantly, all of the recurrences in group A were noted within the first 3 years and those in group B within the first 1 year. Of the 15 macular retinoblastomas that showed recurrence (groups A and B), the location of recurrence was the tumor periphery in 5 (33%), the tumor center in 3 (20%), and diffuse in 7 (47%). Of the 6 recurrences in group B, the recurrence occurred in the treated extrafoveal portion in 4 (67%), in the untreated foveal or papillomacular portion in 1 (17%), and diffusely in 1 (17%) (Table 4). Many factors in the univariate and multivariate analyses suggested that small retinoblastomas were most likely to show tumor recurrence. These factors included smaller tumor size (based on tumor basal dimension of 2-3 mm and also based on the percentage of macula occupied by tumor); lack of associated subretinal fluid, subretinal 769

6 Table 5. Chemoreduction for Macular Retinoblastoma. Analysis of 68 Tumors in 62 Eyes of 49 Patients: Kaplan-Meier Analysis of Time to Recurrence of Macular Retinoblastoma and Time to Recurrence of Associated Subretinal or Vitreous Seeds Kaplan-Meier Estimates Group A (n = 28 Tumors) Chemoreduction Alone, % Failed (No. Events/No. Still in Risk Set) Group B (n = 40 Tumors) Chemoreduction Plus Adjuvant Foveal-Sparing Thermotherapy, % Failed (No. Events/No. Still in Risk Set) Recurrence of macular retinoblastoma At 1-year follow-up 25 (7/20) 17 (6/26) At 3-year follow-up 35 (9/7) 17 (6/11) At 4-year follow-up 35 (9/3) 17 (6/6) At 7-year follow-up Too few Too few Recurrence of associated subretinal or vitreous seeds At 1-year follow-up 33 (9/17) 46 (18/10) At 3-year follow-up 45 (11/7) 46 (18/11) At 4-year follow-up 45 (11/1) 46 (18/4) At 7-year follow-up Too few Too few A B Figure 1. Regression of macular retinoblastoma following chemoreduction alone. A, Before treatment. B, After treatment, the tumor has remained regressed without recurrence at 14 months. A B C Figure 2. Recurrence of macular retinoblastoma following chemoreduction alone. A, Before treatment. B, Six months after treatment, the tumor was regressed. C, Twelve months after treatment, recurrence was detected. seeds, or vitreous seeds; less calcification following chemoreduction; and type 0 regression pattern (complete disappearance of tumor without scar) following chemoreduction (Table 6). In general, tumors that regress to type 0 pattern are typically small retinoblastomas, and in this series, the mean basal tumor dimension before chemoreduction of 3 such cases was 2 mm as compared with 12.6 mm for the remainder. Likewise, tumors with minimal calcification following chemoreduction tend to be small tumors, and in this series, the 11 tumors with 10% or less calcification following chemoreduction were 3.7 mm in mean basal dimension before chemoreduction, but those with less than 10% calcification were 13.8 mm in mean basal dimension before chemoreduction. These factors corroborate the results of Gombos and associates, 12 who found that small retinoblastomas ( 2 mm) were at greater risk for recurrence than larger retinoblastomas. It is possible that smaller tumors may be less responsive 770

7 A B Figure 3. Regression of macular retinoblastoma following chemoreduction and foveal-sparing transpupillary thermotherapy. A, Before treatment. B, After treatment, the tumor has remained regressed without recurrence at 40 months. A B C Figure 4. Recurrence of macular retinoblastoma following chemoreduction and foveal-sparing transpupillary thermotherapy. A, Before treatment. B, One month after treatment, the tumor was regressed. C, Six months after treatment, recurrence was detected. to chemotherapy due to better-differentiated cells or reduced chemotherapy dose because of smaller feeder vessels. This analysis was directed toward individual tumor recurrence and not salvage of the eye. The lack of subretinal fluid, subretinal seeds, or vitreous seeds was a risk for tumor recurrence. We speculated earlier that this might be related to smaller, less responsive tumors. In contrast, in other analyses from our group, the presence of subretinal fluid, subretinal seeds, or vitreous seeds were factors related to the ultimate failure of chemoreduction and loss of the eye due primarily to recurrence of seeds. 5,14 From these results, it appears that chemoreduction plus foveal-sparing thermotherapy provides better tumor control than chemoreduction alone, but it should be understood that this was not a comparative trial. The selection of therapy for macular retinoblastoma depends on many factors, such as the age of the patient; anticipation of new tumor development; size and extent of the tumor; associated features such as subretinal fluid, subretinal seeds, and vitreous seeds; status of the opposite eye; and, quite importantly, long-term visual needs. For example, a patient with a macular tumor in only 1 eye might be considered for chemoreduction plus fovealsparing thermotherapy to ensure best tumor control, but with the understanding that the long-term central vision might be compromised. On the other hand, a patient with a macular tumor in both eyes or a macular tumor in 1 eye and prior enucleation of the opposite eye might be treated with chemoreduction alone and cautious follow-up for recurrence in an effort to spare vision in the only remaining eye. There are limitations in this analysis that should be realized. First, the patients, eyes, and tumors were not randomized to treatment, and there may have been inherent bias regarding the favored treatment strategy. As mentioned previously, patients with a normal opposite macula were more likely treated with thermotherapy consolidation whereas those without normal opposite macula were more likely treated with chemoreduction alone. Second, the extent of the tumor treated with extrafoveal thermotherapy varied and depended on the amount of tumor situated directly under the foveola and in the papillomacular bundle. For example, temporal macular retinoblastoma received more complete thermotherapy whereas nasal macular retinoblastoma received less complete thermotherapy to avoid treating the papillomacular bundle. Third, although we attempted to have a uniform gray-white end point of thermotherapy, the end points varied slightly depending on the size of tumor, degree of pigmentation in the uvea, and amount of tumor calcification following chemoreduction. Fourth, there is a possibility that some of the presumed recurrences were 771

8 Table 6. Chemoreduction for Macular Retinoblastoma. Analysis of 68 Tumors in 62 Eyes of 49 Patients: Risk Factors Predictive of Tumor Recurrence Risk Factors Group A: Chemoreduction Alone (28 Tumors) No Recurrence (15 Patients, 18 Eyes, 19 Tumors) Recurrence (4 Patients, 7 Eyes, 9 Tumors) P Value Risk (95% Confidence Intervals) Univariate analysis Heredity, sporadic vs familial* 13 (87%) 3 (75%) ( ) Laterality, unilateral vs bilateral* 1 (7%) 2 (50%) ( ) Mean proximity to optic disc, mm ( ) Tumor basal dimension, 2 mmvs 1 (5) 2 (22) ( ) 2mm(%)* Macula with rb, % per 10% decrease ( ) Associated subretinal fluid, absent vs 3 (16) 6 (67) ( ) present (%)* Calcification of regressed rb following ( ) chemoreduction, % per 10% decrease Multivariate analysis Macula with rb, % per 10% decrease NA NA ( ) Laterality, unilateral vs bilateral* NA NA ( ) Group B: Chemoreduction Plus Adjuvant Foveal-Sparing Thermotherapy (40 Tumors) (27 Patients, 31 Eyes, 34 Tumors) (3 Patients, 6 Eyes, 6 Tumors) Univariate analysis Tumor basal dimension, 3 mmvs 2 (6) 2 (33) ( ) 3mm(%)* Associated subretinal fluid, absent vs 6 (18) 4 (67) ( ) present (%)* Associated subretinal or vitreous seeds, 13 (38) 6 (100).03 Zero cell nonestimable absent vs present (%)* Multivariate analysis Associated subretinal fluid, absent vs NA NA ( ) present* Groups A and B: All Tumors in Both Groups (68 Tumors) (42 Patients, 49 Eyes, 53 Tumors) (7 Patients, 13 Eyes, 15 Tumors) Univariate analysis Tumor growth pattern, intraretinal vs 0 (0) 3 (20) ( ) endophytic (%)* Tumor basal dimension, 3 mmvs 4 (8) 5 (33) ( ) 3mm(%)* Macula with rb, % per 10% decrease ( ) Associated subretinal or vitreous seeds, 23 (43) 13 (87) ( ) absent vs present (%)* Associated subretinal fluid, absent vs 9 (17) 10 (67) ( ) present (%)* Regression pattern of tumor following 0 (0) 3 (20) ( ) chemoreduction, type 0 vs types 1, 2, 3, or 4 (%)* Calcification of regressed rb following ( ) chemoreduction, % per 10% decrease Multivariate analysis Associated subretinal or vitreous seeds, NA NA ( ) absent vs present* Macula with rb, % per 10% decrease NA NA ( ) Laterality, unilateral vs bilateral* NA NA ( ) Abbreviation: NA, not applicable; rb, retinoblastoma. *Reference variable. Overall model modified score for group A and B tumor growth patterns, P.001. Time-dependent covariate. actually new tumors or superimposed seed recurrence and not true recurrence of the previously regressed macular tumor. 28 However, it should be realized that most new tumors tend to occur in the retinal periphery, outside the macular region. 28,29 Finally, the visual function in these young patients, many of whom were infants, will only be realized at a later date. The singular purpose of this report was to assess tumor control. In summary, management of macular retinoblastoma with chemoreduction alone provides control in 65% of tumors by 4-year follow-up whereas chemoreduction plus foveal-sparing thermotherapy provides 83% control by 4-year follow-up. The risks and benefits of these strategies should be considered for all children with macular retinoblastoma. This is particularly important for children who have reduced visual acuity in the opposite eye when preservation of vision in the treated eye is desired. The choice of therapy depends on many factors, and the goal of treatment should be tumor control while preserving as much useful vision as possible. 772

9 Submitted for Publication: October 13, 2003; final revision received October 14, 2004; accepted October 14, Correspondence: Carol L. Shields, MD, Ocular Oncology Service, Wills Eye Hospital, 840 Walnut St, Philadelphia, PA Funding/Support: Support was provided by the Eye Tumor Research Foundation, Philadelphia, Pa (Dr Carol Shields); the Macula Foundation, New York, NY (Dr Carol Shields); the Rosenthal Award of the Macula Society, Barcelona, Spain (Dr Carol Shields); and the Paul Kayser International Award of Merit in Retina Research, Houston, Tex (Dr Jerry Shields). Previous Presentations: This study was presented in part at the LuEsther Mertz Lecture; April 20, 2004; New York, NY. It was also presented as a paper at the International Congress of Ocular Oncology; January 25, 2004; Hyderabad, India; and presented as a poster at the American Association of Pediatric Ophthalmology and Strabismus; March 27-31, 2003; Washington, DC. REFERENCES 1. Shields CL, Shields JA. Recent developments in the management of retinoblastoma. J Pediatr Ophthalmol Strabismus. 1999;36: Epstein J, Shields CL, Shields JA. Trends in the management of retinoblastoma: evaluation of 1,196 consecutive eyes during J Pediatr Ophthalmol Strabismus. 2003;40: Shields JA, Shields CL. Management and prognosis of retinoblastoma. In: Shields JA, Shields CL. Intraocular Tumors: A Text and Atlas. Philadelphia, Pa: WB Saunders; 1992: Shields JA, Shields CL. Retinoblastoma. In: Shields JA, Shields CL. AtlasofIntraocular Tumors. Philadelphia, Pa: Lippincott Williams & Wilkins; 1999: Shields CL, Honavar SG, Meadows AT, et al. Chemoreduction plus focal therapy for retinoblastoma: factors predictive of need for treatment with external beam radiotherapy or enucleation. Am J Ophthalmol. 2002;133: Kingston JE, Hungerford JL, Madreperla SA, Plowman PN. Results of combined chemotherapy and radiotherapy for advanced intraocular retinoblastoma. Arch Ophthalmol. 1996;114: Gallie BL, Budning A, DeBoer G, et al. Chemotherapy with focal therapy can cure intraocular retinoblastoma without radiotherapy. Arch Ophthalmol. 1996;114: Murphree AL, Villablanca JG, Deegan WF, et al. Chemotherapy plus local treatment in the management of intraocular retinoblastoma. Arch Ophthalmol. 1996; 114: Shields CL, DePotter P, Himmelstein B, et al. Chemoreduction in the initial management of intraocular retinoblastoma. Arch Ophthalmol. 1996;114: Gunduz K, Shields CL, Shields JA, et al. The outcome of chemoreduction treatment in patients with Reese-Ellsworth Group V retinoblastoma. Arch Ophthalmol. 1998;116: Wilson MW, Galindo CR, Haik BG, et al. Multiagent chemotherapy as neoadjuvant treatment for multifocal intraocular retinoblastoma. Ophthalmology. 2001; 108: Gombos DS, Kelly A, Coen PG, et al. Retinoblastoma treated with primary chemotherapy alone: the significance of tumor size, location and age. Br J Ophthalmol. 2002;86: Levy C, Doz F, Quintana E, et al. Role of chemotherapy alone or in combination with hyperthermia in the primary treatment of intraocular retinoblastoma: preliminary results. Br J Ophthalmol. 1998;82: Shields CL, Honavar SG, Shields JA, et al. Factors predictive of recurrence of retinal tumor, vitreous seeds, and subretinal seeds following chemoreduction for retinoblastoma. Arch Ophthalmol. 2002;120: Lam BL, Judisch GF, Sobol WM, Blodi CF. Visual prognosis in macular retinoblastoma. Am J Ophthalmol. 1990;110: Weiss AH, Karr DJ, Kalina RE, et al. Visual outcomes of macular retinoblastoma after external beam radiation therapy. Ophthalmology. 1994;101: Watts P, Westall C, Colpa L, et al. Visual results in children treated for macular retinoblastoma. Eye. 2002;16: Lueder GT, Goyal R. Visual function after laser hyperthermia and chemotherapy for macular retinoblastoma. Am J Ophthalmol. 1996;121: Desjardins L, Chefchaouni MC, Lumbroso L, et al. Functional results after treatment of retinoblastoma. J AAPOS. 2002;6: Hall LS, Ceisler E, Abramson DH. Visual outcomes in children with bilateral retinoblastoma. J AAPOS. 1999;3: Dorey SE, Adam GW, Lee JP, Sloper JJ. Intensive occlusion therapy for amblyopia. Br J Ophthalmol. 2001;85: Shields CL, Santos C, Diniz W, et al. Thermotherapy for retinoblastoma. Arch Ophthalmol. 1999;117: Shields CL, Shields JA, Needle M, et al. Combined chemoreduction and adjuvant treatment for intraocular retinoblastoma. Ophthalmology. 1997;104: Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. New York, NY: Springer-Verlag Inc; Lin DY, Wei LJ. Robust inference for the proportional hazards model. JAmStat Assoc. 1989;84: Shields CL, Shields JA, Cater J, et al. Plaque radiotherapy for retinoblastoma: long term tumor control and treatment complications in 208 tumors. Ophthalmology. 2001;108: Shields CL, Mashayekhi A, Shelil A, et al. Chemoreduction for retinoblastoma: analysis of tumor control and risks for recurrence in 457 tumors. Am J Ophthalmol. 2004;138: Shields CL, Shelil A, Cater J, Meadows AT, Shields JA. Development of new retinoblastomas after 6 cycles of chemoreduction for retinoblastoma in 162 eyes of 106 consecutive patients. Arch Ophthalmol. 2003;121: Abramson DH, Gombos DS. The topography of bilateral retinoblastoma lesions. Retina. 1996;16: