Retinoblastoma Regression Patterns Following Chemoreduction and Adjuvant Therapy in 557 Tumors

Objective  To evaluate retinoblastoma regression patterns following chemoreduction and adjuvant therapy.

Participants  A total of 557 retinoblastomas.

Methods  A retrospective medical record review following 6 cycles of chemoreduction and tumor consolidation (thermotherapy or cryotherapy). Regression patterns included type 0 (no remnant), type 1 (calcified remnant), type 2 (noncalcified remnant), type 3 (partially calcified remnant), and type 4 (flat scar).

Main Outcome Measures  Regression pattern.

Results  Retinoblastoma regressions were type 0 (n = 10), type 1 (n = 75), type 2 (n = 28), type 3 (n = 127), and type 4 (n = 317). Tumors with an initial thickness of 3 mm or less regressed most often to type 4 (92%), those 3 to 8 mm regressed to type 3 (34%) or type 4 (40%), and those thicker than 8 mm regressed to type 1 (40%) or type 3 (49%). Factors predictive of type 1 regression included larger tumor base and closer foveolar proximity. Factors predictive of type 3 included older age, larger tumor base, macular location, closer foveolar proximity, and lack of consolidation. Factors predictive of type 4 included familial hereditary pattern, smaller tumor base, greater foveolar distance, and tumor consolidation.

Conclusions  Following chemoreduction, most small retinoblastomas result in a flat scar, intermediate tumors in a flat or partially calcified remnant, and large tumors in a more completely calcified remnant.

Chemoreduction (CRD) plus focal consolidation therapy has emerged as the leading conservative approach to the management of retinoblastoma.^1-3 This method is effective for small and large tumors, unifocal and multifocal tumors, and exophytic and endophytic tumors. Chemoreduction has largely replaced external beam radiotherapy as the treatment of choice for bilateral retinoblastoma.

Most retinoblastomas show an initial response to CRD. In an earlier analysis of 54 tumors, 2 months of CRD provided an approximately 35% reduction in tumor base and 49% reduction in tumor thickness.⁴ Focal consolidation with thermotherapy, cryotherapy, or plaque radiotherapy is provided following CRD to ensure permanent regression. Tumors posterior to the equator of the eye generally receive thermotherapy, whereas those anterior to the equator receive cryotherapy. Plaque radiotherapy is reserved for tumors that fail thermotherapy or cryotherapy consolidation.⁵

On regression, the retinoblastoma assumes a smaller size with stable margins and, frequently, some degree of calcification. Judgment of regression is challenging, as some tumors become completely calcified whereas others have minimal or no calcification. Tumor regression patterns were initially described following radiotherapy and include type 0, in which the tumor completely disappears, leaving no retinal scar; type 1, with a completely calcified mass appearing like cottage cheese; type 2, with a completely noncalcified mass; type 3, with a partially calcified mass; and type 4, with a flat atrophic scar.^6-8 In this analysis, we specifically evaluated tumor regression patterns following CRD for retinoblastoma.

All new patients with retinoblastoma who were treated with CRD (institutional review board–approved CHP [Children's Hospital of Philadelphia] 582) in the Ocular Oncology Service, Wills Eye Institute, Thomas Jefferson University, in conjunction with the Division of Oncology at The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, were identified. The eligibility criteria for treatment with CRD have been previously described.^4,9,10 Any patient whose tumor(s) could be properly controlled with focal methods alone (cryotherapy, laser photocoagulation, thermotherapy, plaque radiotherapy) was not eligible for inclusion in this study. Exceptions were bilateral cases in which the more advanced eye necessitated CRD, so the less advanced eye that might have been otherwise treated without CRD was included in the CRD protocol.

Exclusion criteria for treatment with CRD, as documented by clinical, ultrasonographic, and neuroimaging modalities, included evidence of iris neovascularization, neovascular glaucoma, extensive hyphema, extensive vitreous hemorrhage, or tumor invasion into the anterior chamber, iris, optic nerve, choroid, or extraocular tissues. Exclusion criteria from a systemic standpoint included evidence of systemic metastasis, prior chemotherapy, failure to thrive, or inadequate organ function of the kidney, liver, or ear. Patients who received prior treatment for retinoblastoma or those who received initial chemoreduction elsewhere and were then referred to our team for consultation were not included in this study. The protocol chemotherapeutic agents included intravenous vincristine, etoposide, and carboplatin.^4,9,10 Chemoreduction cycles were provided every 28 days for a total of 6 cycles. The potential risks and benefits of the CRD protocol were discussed with the patient and his or her family, at which time informed consent was signed.

Each patient was evaluated for age at diagnosis, race, hereditary pattern of retinoblastoma (familial, sporadic), tumor laterality (unilateral, bilateral), and affected eye (right, left, both). Each eye was classified according to the Reese-Ellsworth classification⁷ and the International Classification of Retinoblastoma.^11-13 Each tumor was evaluated for basal dimension, thickness, proximity to the optic disc, and foveola (all in millimeters), related subretinal fluid, subretinal seeds, and vitreous seeds. Ocular oncology follow-up was provided with examination under anesthesia every 1 to 2 months after initiation of CRD, and tumor consolidation was provided using thermotherapy or cryotherapy until tumor control was achieved. After completion of the CRD, the tumor regression pattern was listed as type 0 (no visible remnant), type 1 (completely calcified remnant), type 2 (completely noncalcified remnant), type 3 (partially calcified remnant), and type 4 (atrophic chorioretinal flat scar) (Figures 1, 2, 3, 4 and 5). Thereafter, examinations were performed every 2 to 4 months, as needed. Tumor recurrences were treated with thermotherapy, cryotherapy, or plaque radiotherapy if minor and external beam radiotherapy or enucleation if major.

The clinical data were then analyzed regarding the single outcome of retinoblastoma regression pattern. The effect of each individual clinical variable on this outcome was analyzed by logistic regression analysis. All variables were analyzed as discrete variables except for patient age, tumor base, tumor thickness, proximity to optic disc, and proximity to foveola, which were analyzed as continuous variables. The factors found significant on a univariable level were entered stepwise into a model to determine the significance on a multivariable level. Statistical significance was assigned at P < .05.

Between July 1994 and March 2007, there were 157 patients, with 239 affected eyes and 557 retinoblastomas treated with CRD and adjuvant methods. The mean patient age at presentation was 9 months (median, 7 months; range, <1-48 months). There were 93 boys (59%) and 64 girls (41%). The patient races were white (n = 115; 73%), African American (n = 23; 15%), Hispanic (n = 12; 8%), Asian (n = 5; 3%), and Middle Eastern (n = 2; 1%). The hereditary pattern was sporadic (n = 129; 82%) or familial (n = 38; 18%).

The initial tumor features are listed in Table 1. The mean basal diameter for all retinoblastomas was 7 mm (median, 4 mm; range, <1-25 mm) and thickness was 4 mm (median, 3 mm; range, <1-24 mm). The mean basal diameter for retinoblastomas in the macula was 11 mm; macula to equator, 5 mm; equator to ora serrata, 3 mm; and diffuse tumors, 18 mm. The mean tumor thickness for retinoblastomas in the macula was 6 mm; macula to equator, 3 mm; equator to ora serrata, 2 mm; and diffuse tumors, 9 mm.

Tumor regression patterns per clinical features are listed in Table 2 (Figures 1, 2, 3, 4 and 5). Overall, most retinoblastomas regressed to type 3 (n = 127; 23%) or type 4 (n = 317; 57%). Small retinoblastomas (0-3 mm thickness) showed predominantly type 4 regression (n = 220/240; 92%), medium retinoblastomas (3-8 mm thickness) showed predominantly type 3 (n = 78/232; 34%) or type 4 (n = 92/232; 40%) regression, and large retinoblastomas (>8 mm thickness) showed predominantly type 3 (n = 42/85; 49%) or type 1 (n = 34/85; 40%) regression. Peripheral retinoblastomas between the equator and ora serrata showed predominantly type 4 regression (n = 138/156; 89%), whereas macular retinoblastomas showed predominantly type 3 regression (n = 64/161; 40%). Tumors consolidated with thermotherapy showed type 4 (57%) or type 3 (24%) most often, whereas those consolidated with cryotherapy showed type 4 (82%) or type 3 (8%).

Univariate and multivariate analyses for clinical factors predictive of each regression pattern are listed in Table 3 and Table 4. Based on multivariate analysis, the factors predictive of regression pattern type 1 included larger tumorbase and location nearer to the foveola. The factors predictive of regression pattern type 3 included older age, larger tumor base, location nearer to the foveola, and lack of consolidation with thermotherapy or cryotherapy. The factors predictive of type 4 regression included familial hereditary pattern, smaller tumor base, greater distance from the foveola, and consolidation with thermotherapy or cryotherapy. There were no features predictive of type 0 or type 2 regression patterns in the multivariate model.

During the mean follow-up of 66 months (median, 64 months; range, 6-154 months), recurrence was found in 100 tumors (18%) at a mean interval of 11 months (median, 9 months; range, 2-48 months). Treatment of recurrence included thermotherapy (n = 21; 21%), cryotherapy (n = 31; 31%), plaque radiotherapy (n = 36; 36%), external beam radiotherapy (n = 10; 10%), and enucleation (n = 2; 2%). Tumor recurrence was found following regression pattern type 0 in 2 of 10 cases (20%), type 1 in 35 of 167 cases (21%), type 2 in 3 of 22 cases (14%), type 3 in 44 of 183 cases (24%), and type 4 in 16 of 175 cases (9%).

Retinoblastoma demonstrates a recognizable feature of intralesional calcification at presentation, and the degree of calcification often increases following therapy. Calcification within retinoblastoma is dystrophic and occurs within regions of necrosis. Intralesional calcification is found with other cancers including malignancies of the breast, prostate, and lung. In children, both neuroblastoma and Wilms tumor can display intralesional calcification. It is estimated that approximately 85% of neuroblastomas show calcification on computed tomography, typically in a speckled linear or curvilinear (rim calcification) pattern.¹⁴ Wilms tumors display calcification in only 10% of cases.¹⁴

Retinoblastoma shows a variety of regression patterns following therapy. In 1993, Singh and colleagues¹⁵ reported a retrospective analysis of 180 retinoblastomas treated with external beam radiotherapy at St Bartholomew's Hospital in London, England, between 1970 and 1989 and found type 0 (n = 33; 18%), type 1 (n = 90; 50%), type 2 (n = 31; 17%), and type 3 regression (n = 26; 14%).¹⁵ There was no comment on type 4 regression in that analysis. The only statistically significant factor predictive of regression pattern was the initial tumor size. Abramson and associates¹⁶ evaluated the long-term stability of regression patterns in 89 tumors treated only with external beam radiotherapy and followed for a minimum of 7 years. They noted that regression patterns often slowly changed over time and that types 0, 1, and 4 patterns increased each by approximately 10%, whereas types 2 and 3 decreased by 19% and 8%, respectively. Overall, they observed that smaller tumors were most likely to become type 0, whereas larger tumors were more likely to become type 1 pattern.

From a histopathologic standpoint, Ts’o and colleagues¹⁷ reported that the residual noncalcified tumor following irradiation (type 2 or the noncalcified portion of type 3 regression patterns) generally proved to be well-differentiated retinoblastoma, often with features of retinocytoma with fleurettes and rosettes. This finding was most notable in those tumors that were deemed radioresistant and showed little response to external beam radiotherapy.

In 2004, Shields and colleagues¹⁰ evaluated the more current therapy of CRD and its risks for recurrence. Following CRD and focal consolidation, tumor recurrence was found in 18% of tumors at 7 years, and the most important factor predictive of recurrence was increasing tumor thickness. In that analysis of 457 tumors, regression patterns were identified as type 0 in 14 tumors (3%), type 1 in 45 tumors (10%), type 2 in 15 tumors (3%), type 3 in 149 tumors (33%), and type 4 in 234 tumors (51%).¹⁰ Recurrence was found in 18% of type 0, 31% of type 1, 0% of type 2, 36% of type 3, and 8% of type 4. In that study, tumor regression pattern was not a factor predictive of recurrence using multivariate analysis.¹⁰ Tumor size and location were most important in predicting recurrence. In this analysis, we found that tumor size and location were most important in predicting tumor regression pattern. Of the 557 tumors, recurrence was found in 20% of type 0, 21% of type 1, 14% of type 2, 24% of type 3, and 9% of type 4. Most recurrences (88%) were treatable with thermotherapy, cryotherapy, or plaque radiotherapy, and only 12% required external beam radiotherapy or enucleation.

In our analysis, type 1 regression represented only 10% of tumors following CRD, whereas it was the most common pattern following radiotherapy, accounting for 50% of tumors.¹⁵ Furthermore, the type 4 regression pattern had emerged as the most common pattern following CRD, representing 51%, whereas it was not mentioned following radiotherapy.¹⁵ Some of these differences are due to the post-CRD consolidation with thermotherapy or cryotherapy that often leads to a flat atrophic scar, whereas consolidation treatments were not generally necessary or available after standard radiotherapy.

Regression patterns following CRD have been evaluated pathologically following enucleation. Bechrakis and colleagues¹⁸ examined 5 eyes with advanced retinoblastoma that were treated with 2 to 6 cycles of chemoreduction and found that type 1 regression correlated with complete tumor necrosis. Types 2 or 3 regression, found in 4 cases, correlated with either complete tumor necrosis (n = 1) or residual active tumor (n = 4). Enucleation of these eyes was indicated for suspected viable intraocular tumor and only 1 of the 5 eyes received the full 6 cycles of CRD. Dithmar and colleagues¹⁹ evaluated 2 eyes with type 3 regression following 2 cycles or less of CRD and noted that this regression type represents a regressed gliotic mass with calcification in one eye and a partially necrotic tumor adjacent to a viable component in another eye. Demirci et al²⁰ evaluated 10 eyes that were treated with CRD and no consolidation therapy for 1 to 6 cycles and were enucleated for suspected recurrence or opaque media. All 10 eyes showed evidence of retinoblastoma regression. In 8 eyes the main tumor was completely regressed without viable tumor, but 6 had a basal, residual, inactive, well-differentiated component consistent with retinocytoma. Two remaining eyes had viable retinoblastoma. Of the 2 eyes classified as regression pattern type 1, both were completely regressed. Of the 8 eyes classified as regression pattern type 3, 6 were completely regressed but had histopathologic features of a basal retinocytoma component, whereas 2 had a viable tumor. These findings indicate the challenge in judging viability of retinoblastoma. From a clinical perspective, the nonviable tumor should remain regressed for months, whereas the viable tumor would show progressive regrowth generally in 1 to 4 months.

In the current analysis, we intended to further explore regression patterns after CRD and identify significant factors that determine each pattern. We found that small retinoblastomas of 3 mm or less (n = 240) regressed most often to type 4 (n = 220; 92%), medium tumors of 3 to 8 mm (n = 232) regressed most often to type 3 (n = 78; 34%) or type 4 (n = 92; 40%), and large tumors of more than 8 mm (n = 85) regressed most often to type 1 (n = 34; 40%) or type 3 (n = 42; 49%). The consolidation technique affected regression patterns, as tumors treated with post-CRD thermotherapy showed type 4 regression in 57% (136/238 tumors) and those treated with post-CRD cryotherapy showed type 4 regression in 82% (129/156 tumors). By multivariate analysis, the main factors affecting the regression pattern were tumor size and location. Larger tumors and those nearer the foveola showed type 1 or type 3 regression patterns, whereas smaller tumors or those located more peripherally showed type 4 regression pattern. Older age was more predictive of type 3 pattern, and this could reflect larger tumor size in older children, whereas familial hereditary pattern predicted type 4 pattern, and this could reflect smaller tumor size in neonates.

The management of retinoblastoma continues to be challenging, with critical judgment regarding tumor regression following CRD.¹ The physician managing the child with retinoblastoma using CRD should be familiar with tumor regression patterns and be able to differentiate them from incomplete response and tumor recurrence. In this analysis we found that one might anticipate type 1 regression with large retinoblastomas, types 3 or 4 regression with medium tumors, and type 4 regression with small tumors.

Correspondence: Carol L. Shields, MD, Ocular Oncology Service, Wills Eye Institute, 840 Walnut St, Ste 1440, Philadelphia, PA 19107 (carol.shields@shieldsoncology.com).

Submitted for Publication: February 9, 2008; final revision received July 3, 2008; accepted July 4, 2008.

Author Contributions: Dr C. L. Shields has had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the Retina Research Foundation (Charles L. Schepens Lecture) of the Retina Society in Cape Town, South Africa (Dr C. L. Shields); the Paul Kayser International Award of Merit in Retina Research, Houston, Texas (Dr J. A. Shields); a donation from Michael, Bruce, and Ellen Ratner, New York, New York (Drs C. L. and J. A. Shields); Mellon Charitable Giving from the Martha W. Rogers Charitable Trust, Philadelphia, Pennsylvania (Dr C. L. Shields); and the Eye Tumor Research Foundation, Philadelphia, Pennsylvania (Drs C. L. and J. A. Shields).

Role of the Sponsors: The sponsors had no role in the design and conduct of the study, in the collection, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.

Additional Contributions: Statistical analysis was provided by Rishita Nutheti, PhD, International Centre for Advancement of Rural Eye Care, L.V. Prasad Institute, Hyderabad, India.