Intensive Multimodality Therapy for Extraocular Retinoblastoma: A Children's Oncology Group Trial (ARET0321)

PURPOSE

Metastatic retinoblastoma has a poor prognosis when treated with conventional chemotherapy and radiation therapy (RT). Intensified therapy may improve the outcome.

METHODS

A prospective, international trial enrolled patients with extraocular retinoblastoma. Patients with stage II or III (locoregional) retinoblastoma received four cycles of chemotherapy, followed by involved field RT (45 Gy). Patients with stage IVa or IVb (metastatic or trilateral) retinoblastoma also received four cycles of chemotherapy and those with ≥ partial response then received one cycle of high-dose carboplatin, thiotepa, and etoposide with autologous hematopoietic stem-cell support. Patients with stage IVa or IVb with residual tumor postchemotherapy received RT. The proportion of patients who achieved event-free survival would be reported and compared with historical controls separately for each of the three groups of patients.

RESULTS

Fifty-seven eligible patients were included in the analyses. Event-free survival at 1 year was 88.1% (90% CI, 66.6 to 96.2) for stage II-III, 82.6% (90% CI, 61.0 to 92.9) for stage IVa, and 28.3% (90% CI, 12.7 to 46.2) for stage IVb/trilateral. Toxicity was significant as expected and included two therapy-related deaths.

CONCLUSION

Intensive multimodality therapy is highly effective for patients with regional extraocular retinoblastoma and stage IVa metastatic retinoblastoma. Although the study met its aim for stage IVb, more effective therapy is still required for patients with CNS involvement (ClinicalTrials.gov identifier: NCT00554788).

INTRODUCTION

Retinoblastoma is the most common primary ocular tumor of childhood. In high-income countries, the majority of patients have intraocular disease and a low risk of developing metastatic disease. Extraocular disease is common in low-income countries and is almost always fatal.^1-3

CONTEXT

• Key Objective

• To develop more effective therapy for children with extraocular retinoblastoma.

• Knowledge Generated

• The study-prescribed intensive multimodality therapy achieved the prespecified aims to reduce events in all three strata. Event-free survival results were very promising for patients with regional extraocular retinoblastoma and stage IVa metastatic retinoblastoma, but development of more effective therapy is still required for patients with CNS involvement.

• Relevance

• This treatment regimen may become the standard of care for patients with regional extraocular retinoblastoma and stage IVa metastatic retinoblastoma.

Retinoblastoma may grow through the sclera into the orbit; metastasize hematogenously to sites such as bone, bone marrow, and liver; and/or metastasize to the CNS directly through the optic nerve to the brain or via the subarachnoid space surrounding the optic nerve. The eye has limited lymphatic drainage, but regional nodal metastasis may rarely occur.

Retrospective series have suggested the benefit of intensive multimodality treatment, including high-dose chemotherapy with autologous hematopoietic stem-cell rescue, for patients with stage IVa retinoblastoma,^4-9 and some evidence of success for stage IVb or trilateral retinoblastoma.^10,11 We now report a prospective, multicenter, international study with very promising survival results for patients with stages II-III and IVa disease.

METHODS

Eligibility Criteria

Patients were age 10 years and younger with histologically or cytologically proven extraocular retinoblastoma, an Eastern Cooperative Oncology Group performance score of 0-2, and adequate organ function (Protocol, section 3.2, online only). Patients with involvement of the optic nerve subarachnoid space only were not eligible for this study. Prior therapy for intraocular retinoblastoma was permitted, but prior therapy for extraocular retinoblastoma prohibited.

All patients had brain-orbit magnetic resonance imaging (MRI) with/without gadolinium, lumbar puncture for CSF cytology, abdominal computed tomography with intravenous contrast, bone marrow aspirates and biopsies, and bone scans. If the brain MRI or CSF cytology was positive, a total spine MRI with/without gadolinium was mandated. Patients with orbital disease, optic nerve involvement at the surgical margin, or regional nodal disease were enrolled onto the stage II-III stratum. Patients with metastatic disease not involving the CNS were enrolled onto the stage IVa stratum. Patients with CNS disease (metastases or trilateral disease) were enrolled onto the stage IV stratum.

The institutional review board at each treating institution and the National Cancer Institute's pediatric central institutional review board approved the Protocol (ClinicalTrials.gov identifier: NCT00554788). Informed consent was obtained from a parent or legal guardian.

Treatment

Figure 1 summarizes the treatment regimen. Treatment began with four cycles of induction chemotherapy, with mesna and filgrastim supportive care. Patients with preexisting visual (neither fovea free of tumor) or hearing impairments received an alternative induction regimen omitting cisplatin (Protocol, section 4.2.2). Audiograms were required at baseline, before each cycle of induction, before consolidation, before radiation therapy (RT), and at completion of therapy.

FIG 1.

COG ARET0321 treatment schema. AUC, area under the curve; IV, intravenous; SC, subcutaneous.

Patients with stage IVa or IVb disease without bone marrow involvement underwent autologous hematopoietic stem-cell harvesting after any cycle of induction chemotherapy. Patients with bone marrow metastases had bone marrow examinations repeated after each cycle until clear and then received at least two more cycles preharvest. An adequate harvest yield was defined as ≥ 2.5 × 10⁶ CD34/kg from peripheral blood or ≥ 1 × 10⁸ total nucleated cells/kg from bone marrow.

Patients had disease evaluations after cycles 2 and 4. After cycle 4, stratum II-III patients who achieved at least a partial response proceeded to RT. Stratum 4a or 4b patients proceeded to high-dose chemotherapy with autologous hematopoietic stem-cell support assuming adequate stem-cell yield, at least a partial response, adequate organ function (Protocol, section 4.2.4), and no uncontrolled serious infection.

The high-dose carboplatin, thiotepa, and etoposide chemotherapy treatment regimen is in described in Figure 1 and Protocol section 4.2.4. Carboplatin was dosed according to the Calvert formula,¹² with each day's dose on the basis of a new daily estimate of the glomerular filtration rate, determined either by a renal technetium⁹⁹-DTPA radionuclide study or by a timed urine collection for creatinine clearance. At least 48 hours after completion of chemotherapy, cryopreserved stem cells were rapidly thawed and reinfused. Filgrastim was started on day +1 and continued until the postnadir absolute neutrophil count was ≥ 2,000/mcl.

RT

The clinical target volume for patients with stage II disease included the orbit. The clinical target volume for patients with stage III disease included the orbit (when stage II features were present) and residual tumor and regional lymph nodes after induction chemotherapy. The planning target volume included the clinical target volume surrounded by a 5-mm margin. The prescribed dose for stage II and III was 45 Gy to the planning target volume. All patients with stage II or III disease were intended to receive RT after completion of chemotherapy.

Stage IVa and IVb patients were irradiated to sites initially involved on the basis of response. Stage IVa patients were not irradiated after complete response to induction chemotherapy or to sites measuring < 5 mm after high-dose chemotherapy. RT was omitted for stage IVb patients who achieved a complete response to induction chemotherapy. Treatment volume for stage IVa patients included the pre-RT residual and 5-mm clinical target volume and 5-mm planning target volume. The prescribed dose was 36 Gy. The treatment volume (and dose) for stage IVb patients included the neuraxis clinical target volume (23.4 Gy age younger than 60 months; 36 Gy age older than 60 months) with treatment of spinal (36 Gy cumulative maximum) and intracranial (45 Gy cumulative maximum) metastatic sites measuring ≥ 5 mm at the time of planning. Pineal region tumors were treated to 50.4 Gy (cumulative maximum). Stage II and III patients were planned to begin RT within 42 days after the start of cycle 4 of induction chemotherapy. Stage IVa and IVb patients were planned to begin RT approximately day +42 postautologous stem-cell infusion. Fractionation was 1.8 Gy/day except for the treatment of whole-lung and whole-abdominal volumes (1.5 Gy/day).

Response Criteria

A complete response was defined as resolution of the metastatic disease on bone marrow examinations and relevant imaging studies (those that revealed an abnormality before enrollment on this study). A very good partial response was defined as a > 90% decrease in the sum of the two-dimensional measurements of the tumor(s) on relevant (computed tomography and/or MRI) imaging studies, no new lesions, bone marrow free of tumor, and improved bone scan. A partial response was defined as a 50%-90% decrease in the sum of the two-dimensional measurements of the tumor(s) on relevant (computed tomography and/or MRI) imaging studies, no new lesions, and bone marrow free of tumor. Stable disease was defined as a < 50% decrease and < 25% increase in the sum of the two-dimensional measurements of the tumor(s) on relevant (computed tomography and/or MRI) imaging studies and no new lesions. Progressive disease was defined as at least a 25% increase in the sum of the two-dimensional measurements of the tumor(s) on relevant (computed tomography and/or MRI) imaging studies and/or a new lesion that unequivocally represented retinoblastoma.

Statistical Considerations

The primary aim of the study was to compare the event-free survival (EFS) between the current strata-specific strategies and historical experiences separately for each stratum. The 1-year historical EFS rate for patients with orbital disease was 40% with very few failures after 1 year.¹³ The 1-year historical EFS rate for distant metastatic patients treated with less intensive strategies was at most 20% for CNS-negative patients and approximately 5% for CNS-positive patients.¹⁴ The hypothesis was that the proposed treatment strategies could improve the 1-year EFS to 70%, 70%, and 20% for strata II-III, IVa, and IVb, respectively.

The study was planned for 6 years of enrollment and 1 year of follow-up, after which the primary analysis was to take place. It was anticipated that there would be a total of 60 patients enrolled and split about equally in the three strata. We assumed that the distribution of EFSs can be adequately modeled using the exponential cure model. For the primary analysis, we planned to compare the observed survival experience to the expected distribution using a method adapted from Woolson's test.¹⁵ For 20 stage II-III patients, assuming that the EFS at 1 year from historical experience was 40%, the power to detect a true increase in the EFS rate of 30% (70%-40%) was 89%. For 20 stage IVa patients, assuming that the EFS rate for the current therapy was 20% at 1 year, we would have > 95% power to detect a true increase in the EFS of 50% (70%-20%). For 20 stage IVb patients, assuming that the EFS rate for the current therapy was 5% at 1 year, we would have 80% power to detect a true increase in the EFS of 15% (20%-5%). A one-sided alpha level of .05 was chosen in the primary analyses.

Outcome data for the stage II-III stratum were formally reviewed for excessive events using the Lan-DeMets implementation of the O'Brien and Fleming group sequential boundaries, after 50% and 100% of the expected information is observed. An alpha spending function of (alpha × t) was used to specify the monitoring boundaries as it is of interest to detect early indications of increased harm relative to historical experience.¹⁶ For the stage IVa and IVb strata, we did not monitor for futility because the historical outcome for these patients was so poor, and we wanted to avoid acting on early failures when there might be a delayed effect.

An event was defined as relapse, second malignancy, or death from any cause. The primary outcome measure EFS was defined as the time from study enrollment to first occurrence of any experienced event or last contact. A patient who did not experience an event was considered censored at last contact; in all other cases, an EFS event was considered to have occurred. Overall survival (OS) was defined as the time from enrollment to death or last contact. Patients who were alive at last contact were considered censored for survival; in all other cases, an OS event was considered to have occurred. EFS and OS, as a function of time since enrollment, were calculated by the method of Kaplan and Meier. The distribution of the complementary log-log transformation of the Kaplan-Meier estimate was used for the construction of CIs.

RESULTS

Patients

Sixty patients (including 21 to the stage II-III stratum, 18 to the stage IVa stratum and 21 to the stage IVb stratum) were enrolled from 2008 to 2014. Two patients from the stage II-III stratum and one patient from the stage IVb stratum were declared ineligible because of the timing of enrollment, having started treatment before consent, and/or registration on study (Fig 2). The Children's Oncology Group data safety monitoring committee elected not to permit replacement of the ineligible patients because that was felt unlikely to influence the study results, and the data presented reflect the 57 eligible patients. Forty-two eligible patients were enrolled at Children's Oncology Group institutions, and 15 were enrolled by international collaborators in Argentina (n = 8), Egypt (n = 6), and Brazil (n = 1). The data reported are current as of December 31, 2019.

FIG 2.

CONSORT diagram.

Patient characteristics including age and sex are summarized in Table 1. Treatment of the intraocular disease before enrollment on this study included surgery (n = 41, 71.9%), chemotherapy (n = 19, 33%), and external beam RT (n = 5, 8.8%).

TABLE 1.

Baseline Characteristics for Eligible Patients

The median interval from diagnosis of intraocular disease to diagnosis of extraocular disease was 27 days, with a range from 0 to 2,287 days. Sites of distant metastatic disease in the stage IVa patients (available for 19 patients) were bone marrow (n = 14), bone (n = 11), liver (n = 2), and other soft-tissue sites (n = 4). Of the patients enrolled on stratum IVb, 10 had disease metastatic to the CNS, two had trilateral disease, and the data regarding the type of CNS retinoblastoma were missing for eight. Six patients had localized CNS disease, six had leptomeningeal dissemination, and data regarding leptomeningeal metastasis were missing for eight.

Table 2 describes the institutional assessments of response to induction therapy. Twenty-two patients underwent consolidation high-dose chemotherapy with autologous stem-cell rescue; 21 received peripheral blood stem cells, and one received both peripheral blood stem cells and bone marrow. Twenty-six patients received RT: 15 with stage II-III, seven with stage IVa, and four with stage IVb disease. Data are unavailable regarding the number of patients treated with the standard versus the alternative induction regimen.

TABLE 2.

Institutional Assessment of Response to the Induction Phase of the Regimen

Outcomes

The median follow-up was 7.3 years. For the stage II-III patients, the 1-year EFS was 88.1% (90% CI, 66.6 to 96.2), and the one-sided P value from the final analysis was .003, indicating that the ARET0321 therapy significantly reduced the risk of events. The 1-year OS was 88.1% (90% CI, 66.6 to 96.2). Interim analyses to assess the difference between observed and expected failures were not performed as the required number of failures in the first look was not obtained when the accrual was completed. There was no excessive number of failures concern. The 3-year EFS and OS were the same as the 1-year survival.

For the stage IVa patients, the 1-year EFS was 82.6% (90% CI, 61.0 to 92.9), and the one-sided P value from the final analysis was < .001, indicating that the ARET0321 therapy significantly reduced the risk of events. The 3-year EFS was 76.7% (90% CI, 54.6 to 89.1). The 1-year and 3-year OSs were 88.5% (90% CI, 67.7 to 96.3) and 76.7% (90% CI, 54.6 to 89.1), respectively.

For the stage IVb patients, the 1-year EFS was 28.3% (90% CI, 12.7 to 46.2), and the one-sided P value for the final analysis was .002, indicating that the ARET0321 therapy significantly reduced the risk of events. The 3-year EFS was 14.2% (90% CI, 3.7 to 31.5). The 1-year and 3-year OSs were 42.1% (90% CI, 23.6 to 59.5) and 12.3% (90% CI, 3.0 to 28.4), respectively.

The Kaplan-Meier plots of EFS and OS by stratum are shown in Figure 3. Patient-level outcome data are included in Appendix Table A1 (online only).

FIG 3.

Kaplan-Meier curves and at-risk table of EFS and OS for (A) stage II-III patients, (B) stage IVa patients, and (C) stage IVb patients. EFS, event-free survival; OS, overall survival.

Toxicity

Toxicity was significant as expected and included two therapy-related deaths due to septicemia during induction therapy. Appendix Table A2 (online only) describes the grade 3-5 nonhematologic adverse events encountered during the induction and consolidation phases.

Recurrences

Three patients in the stage II-III stratum suffered recurrences, two in the CNS at 5 and 10 months, and one in bone at 6 years. Two patients in the stage IVa stratum suffered recurrences, both in the CNS at 5 and 9 months. Ten patients in the stage IVb stratum suffered recurrences, all in the CNS at a median of 6 months (range, 1-18 months).

Second Malignancies

Six patients have developed secondary malignancies. These include one case each (and time since enrollment) of osteosarcoma (8 years), papillary carcinoma of the thyroid (9 years), glioblastoma multiforme (8 years), large B-cell EBV-associated lymphoproliferative disorder (8 months), acute myeloid leukemia (2 years), and alveolar soft part sarcoma of the bladder (5 years). Five patients had unilateral retinoblastoma and two received RT (one each with unilateral and bilateral retinoblastoma).

DISCUSSION

Patients with intraocular retinoblastoma have an excellent prognosis for survival in higher-income countries, but patients who develop metastatic disease have historically had a very poor prognosis when treated with conventional therapy. This current report, which we believe, to our knowledge, to be the largest prospective series of children with extraocular retinoblastoma, indicates that intensive multimodality therapy is curative in a majority of patients with stage IVa disease and suggests that the inclusion of high-dose thiotepa-based chemotherapy may be a crucial component of therapy for patients with metastatic disease. The patients treated for stage II-III disease also fared very well, similar to other recent experience,¹⁷ suggesting that high-dose chemotherapy with autologous hematopoietic stem-cell support is not required for patients with locoregional extraocular retinoblastoma.

Some patients apparently cured of their extraocular retinoblastoma developed second malignancies that affected EFS and OS. The study treatment likely caused some of these, but others may be due to the patients' RB1 germline mutations, although data regarding the subjects' germline status were not collected.

Strengths of this trial were its prospective nature and the inclusion of patients from middle-income countries, particularly since patients with extraocular retinoblastoma disproportionately live in low-income countries. However, this treatment is resource intensive and not currently possible in most centers in low-income countries. A major challenge is to determine how to offer curative therapy for extraocular retinoblastoma in lower-income countries. Limitations include some missing data (such as the treatment previously administered for the intraocular retinoblastoma and sites of RT administered to the stage IV patients), including whether leptomeningeal disease was present for some of the stage IVb patients and which patients received the alternative induction regimen and whether that had any impact on outcome. It would be valuable to know the proportion of survivors afflicted with clinically significant ototoxicity, but long-term data regarding this issue were not collected.

Given the limited number of potential patients, it is unclear whether future trials will be able to demonstrate superiority or noninferiority to these stage II-III and IVa results. Future trials should attempt to improve the prognosis of patients with stage IVb disease. Interventions might include the use of other active agents such as melphalan and/or topotecan, tandem transplants rather than a single high-dose chemotherapy cycle,¹⁸ intrathecal chemotherapy and/or radioimmunotherapy,¹⁹ and/or RT in older patients.

In conclusion, intensive multimodality therapy is highly effective for patients with stage II-III and stage IVa retinoblastoma. More effective therapy is required for patients with CNS retinoblastoma.

APPENDIX

TABLE A1.

Eligible Patient Data

TABLE A2.

Grade 3+ Nonhematologic AEs After Induction or After Consolidation Cycle 1

AUTHOR CONTRIBUTIONS

Conception and design: Ira J. Dunkel, Guillermo L. Chantada, Sherif Abouelnaga, Jeffrey C. Buchsbaum, Thomas E. Merchant, Meaghan M. Granger, David H. Abramson, Carlos Rodriguez-Galindo, Murali M. Chintagumpala

Administrative support: Meaghan M. Granger, Murali M. Chintagumpala

Provision of study materials or patients: Ira J. Dunkel, Guillermo L. Chantada, Anuradha Banerjee, Meaghan M. Granger, Rima F. Jubran, David H. Abramson

Collection and assembly of data: Ira J. Dunkel, Guillermo L. Chantada, Anuradha Banerjee, Sherif Abouelnaga, Thomas E. Merchant, Meaghan M. Granger, Rima F. Jubran, Mark D. Krailo, Murali M. Chintagumpala

Data analysis and interpretation: Ira J. Dunkel, Jin Piao, Guillermo L. Chantada, Anuradha Banerjee, Sherif Abouelnaga, Jeffrey C. Buchsbaum, Thomas E. Merchant, Meaghan M. Granger, Rima F. Jubran, Joanna L. Weinstein, Lauren Saguilig, Mark D. Krailo, Carlos Rodriguez-Galindo, Murali M. Chintagumpala

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Intensive Multimodality Therapy for Extraocular Retinoblastoma: A Children's Oncology Group Trial (ARET0321)

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).