Outcomes After Proton Therapy Reirradiation for Malignant Glioma: Prospective Analysis of Over 100 Patients from the Proton Collaborative Group

Purpose: The optimal salvage treatment for malignant glioma after prior radiotherapy (RT) is not well established. Proton therapy (PT) is an attractive treatment modality for reirradiation (ReRT) of recurrent disease; however, data are limited. Methods: The prospective, multi-institutional Proton Collaborative Group (PCG) registry was queried for patients with malignant glioma who previously received RT and underwent PT-ReRT between 7/2011-10/2018. Overall survival (OS) and progression-free survival (PFS) were assessed using the Kaplan-Meier method and Cox proportional hazards regression analysis. Results: 109 consecutive patients were identied. Median follow-up was 11.2 months and median time interval (TI) from prior RT (median 59.3 Gy) to PT-ReRT (median 49.6 Gy) was 45.7 months. Primary histology was glioblastoma (n=62), oligodendroglioma (n=26), and astrocytoma (n=20). Median PFS and OS was 6.1 and 11.1 months, respectively. On UVA, improved PFS was associated with oligodendroglioma (HR 0.39) and astrocytoma (HR 0.55) histologies, WHO Grade III (HR 0.36), TI (HR 0.989), surgical resection (HR 0.39), and PT-ReRT dose ≥ 50 Gy (HR 0.59) (all p<0.03), while improved OS was associated with oligodendroglioma histology (HR 0.35), WHO grade III (HR 0.33), ECOG 0 (HR 0.47), TI (HR 0.985), and PT-ReRT dose ≥ 50 Gy (HR 0.58) (all p<0.04). Late Grade 3 toxicities occurred in 5.6% of patients with no attributable acute or late Grade 4-5 events. Conclusion: In the largest series of glioma PT-ReRT reported to date, retreatment appears well tolerated with acceptable outcomes and favorable toxicities compared to photon historical controls. Dose escalation achievable with PT may improve outcomes for well-selected patients. achieving a favorable toxicity prole. Our study corroborates several known prognostic factors of recurrent malignant glioma, including tumor histology/grade, performance status, and time to recurrence. In addition, outcomes appear superior when utilizing reirradiation dose of > 50 Gy (CGE). This nding suggests a role for dose escalation in well selected


Introduction
Malignant glioma represents 25% of all primary brain tumors. [1] Unfortunately, despite standard therapies, malignant glioma inevitably recurs, most commonly in-eld and/or marginal to the original resection bed and radiotherapy (RT) treatment elds. [2,3] In the recurrent setting, outcomes and optimal salvage therapies are not well established. Treatment options include maximally safe re-resection, systemic therapies, tumor treating elds, and reirradiation. [4] A second course of RT can be considered with the goal of providing durable control of the recurrent tumor and preventing the associated morbidity of local tumor progression.
Reirradiation of recurrent glioma can be challenging due to the previous receipt of high-dose RT in close proximity to critical organs at risk (OAR), such as the brainstem and optic structures. Excess cumulative dose to OARs and normal non-target brain tissue may increase the risk of signi cant neurologic sequela and potential detriment to patient quality of life (QOL). [5][6][7][8][9] As such, administration of de nitive reirradiation doses are not typically feasible due to risks of high-grade toxicities.
Proton therapy (PT) is an auspicious treatment modality for this clinical scenario. [10] The physical properties of PT compared to photon therapy offer greater capability to cover the recurrent disease with meaningful radiation dose while minimizing excess exit-dose to OARs and normal brain tissue. PT may, therefore, facilitate safer and more effective treatment in the recurrent setting and, in a select minority of cases, offer patients a new chance of durable disease control. [11] In this context, the goal of this study was to analyze patients with recurrent glioma enrolled on the prospective multi-institutional Proton Collaborative Group (PCG) registry for treatment patterns, patient outcomes, and treatment-related toxicities.

Methods
The PCG is a research consortium of 17 proton centers in the United States. REG001-09 (NCT01255748) is a prospective registry study for which each institution obtained individual IRB approval. The study opened for accrual in 2010 and is currently enrolling patients. The registry was queried for all consecutive patients with recurrent glioma treated with PT-reirradiation from 2011-2018. Patient and tumor characteristics, prior and current radiation details, clinical outcomes, and toxicities were extracted and reviewed retrospectively.
Overall survival (OS) and progression-free survival (PFS) were de ned as time from initiation of PTreirradiation to the event of interest and censored at the time of last follow-up or death. These results were recorded individually by each institution and reviewed centrally by PCG staff. Descriptive statistics were used to report patient, tumor, and treatment characteristics. Outcomes were assessed using the Kaplan-Meier method and both univariable and multivariable Cox proportional hazards regression analysis, denoted as hazard ratios (HR) with 95% con dence intervals (CI). Demographic and treatment details included in the univariable analysis (UVA) were age, sex, race, tumor location, histology, grade, IDH mutation status, 1p19q co-deletion, MGMT methylation status, ECOG performance status, tumor size, surgical resection, extent of resection, systemic therapy use, RT dose and fractionation, and time interval from prior RT to PT-reirradiation (as a continuous variable). Each variable with a p-value < 0.1 was included in the multivariable analysis (MVA). Toxicities were scored according to CTCAE v4.0. Univariable binomial regression analysis was performed to identify clinical correlates of toxicity. To account for differences in dose fractionation, radiation dose was summed via equivalent dose in 2 Gy fractions (EQD2) using a representative α/β ratio of 10. The statistical program used was MedCalc version 19.1 (Ostend, Belgium). All statistical analyses were performed using a p-value signi cance level of < 0.05.

Results
Patient, tumor, and treatment characteristics are listed in detail in Table 1. In the database, 109 patients with recurrent glioma treated with PT across 7 treatment centers were identi ed. Of those, 50% (n = 55) were male, 80% (n = 87) were white, and median age was 49 years (range 19-85). Most patients had an ECOG performance status of 0 (n = 48) or 1 (n = 37). Primary tumor histology was glioblastoma (n = 63, including 1 gliosarcoma), oligodendroglioma (n = 26), and astrocytoma (n = 20). Of the non-glioblastoma patients, 28 had WHO grade III tumors, including 16 with anaplastic oligodendroglioma and 12 with anaplastic astrocytoma. Molecular data were not commonly available for non-glioblastoma patients. Of those with available data, IDH1 mutation and MGMT methylation was present in 20% (n = 8 of 41) and 52% of patients (n = 22 of 42), respectively. Overall, 86% and 89% of patients had surgery and chemotherapy at initial diagnosis, respectively. Most patients received temozolomide (n = 73 of 97). Most patients had a single prior course of RT (n = 102), but 6 patients had 2 prior courses and 1 patient had 3 prior courses. The median cumulative dose of all prior treatments was 59.3 Gy (range 23.4-120.9). Most prior treatment courses were conventional and/or mildly hypofractionated dose fractionation regimens; however, 5% of patients received stereotactic radiosurgery at least once prior to PT.
Of the patients with one prior course of RT, the median time from initial diagnosis to recurrence was 45.7 months (range 3.2-386.7 months). For the whole cohort, 40% of patients had surgery within 3 months of PT. The median time from the most recent prior RT course to initiation of PT-reirradiation for the entire cohort and the glioblastoma cohort was 40.7 months (range 2.9-441.0) and 14.7 months (range 3.9-89.5), respectively. Pencil beam scanning PT (n = 56) and uniform scanning PT (n = 53) were utilized. The median dose of PT-reirradiation was 49.6 Gy (CGE) (range 24.9-60.1) with a median dose/fraction size of 2.2 CGE/fraction (range 1.1-4.0). The median cumulative dose of all RT courses was 106.5 Gy (CGE) (range 78.8-167.8). Sixty-eight percent of patients had concurrent chemotherapy with PT-reirradiation. Of those who received concurrent chemotherapy, the most common agents included temozolomide (n = 36), bevacizumab (n = 12), and temozolomide/bevacizumab combination (n = 21).

Progression-free survival and overall survival
Outcomes analyses were performed for patients with only one prior course of RT (n = 102). The median follow-up time for surviving patients was 11.2 months. The median PFS was 6.1 months (95% CI 5.1-8.5), and the PFS at 3, 6, and 12 months were 74.5%, 53.8% and 23.6%, respectively. Median PFS was longer for oligodendroglioma (
There was one acute grade 4 event of thrombocytopenia in a patient receiving concurrent temozolomide. On UVA, the only variable associated with acute grade 3 toxicity was ECOG 2 (HR 2.86, 95% CI 0.58-5.14, p = 0.012).

Discussion
The current study describes the largest series of patients with recurrent malignant glioma treated to date with PT-reirradiation. This multi-institutional prospectively-collected dataset demonstrates comparable e cacy to photon reirradiation series while achieving a favorable toxicity pro le. Our study corroborates several known prognostic factors of recurrent malignant glioma, including tumor histology/grade, performance status, and time to recurrence. In addition, outcomes appear superior when utilizing reirradiation dose of > 50 Gy (CGE). This nding suggests a role for dose escalation in well selected patients with favorable clinical features and good performance status, a treatment paradigm which may be particularly suited for PT.
Recurrent glioma is a challenging clinical scenario. Most patients have already received high-dose chemoradiotherapy, commonly to 50.4-60 Gy, and tumors are often in close proximity to critical neurologic OARs that already received radiation doses at or close to tolerance. In addition, as noted in our study, patients commonly already suffer from neurological sequalae related to prior therapies as well as ongoing tumor-related de cits. Subsequent treatment decisions can, therefore, be exceptionally di cult to make.
There is currently a robust collection of studies examining photon reirradiation for recurrent glioma, particularly for recurrent WHO Grade IV glioblastoma. [12,13]  developed a late grade 3 event. [15] Survival outcomes in these studies were comparable to our results, which demonstrated a median OS of 9.5 months, 16.3 months, and 22.9 months for patients with glioblastoma, astrocytoma, and oligodendroglioma, respectively. RTOG 1205, a randomized phase-II trial comparing bevacizumab alone or bevacizumab with hypofractionated reirradiation to 35 Gy/10 fractions for recurrent glioblastoma, was recently presented in abstract form, demonstrating a clear PFS bene t (7.1 months vs. 3.8 months) but no difference in OS (10.1 months vs. 9.7 months) with combinedmodality therapy. Despite this, toxicity was quite low after reirradiation, with acute and late grade 3 + events for the whole cohort of 5% and 0%, respectively. [21] Another randomized trial was recently conducted by Henry Ford Hospital comparing bevacizumab alone to bevacizumab with fractionated radiosurgery to 32 Gy/4 fractions with similar clinical and toxicity outcomes. [22] These studies appear to con rm the safety pro le and potential to delay disease progression for modestly-dosed hypofractionated reirradiation regimens. This strategy may be particularly useful for high-grade recurrences. Despite a tumor control bene t, survival outcomes remain disappointing. With low reported rates of treatment-related toxicity, dose intensi cation may be a reasonable strategy to enhance reirradiation e cacy within the therapeutic window.
To date, reports using de nitive or near-de nitive reirradiation dose regimens are much more limited, likely due to concerns for treatment-related toxicities. Historical studies found reirradiation doses > 40 Gy using conventional 3D-conformal RT to be associated with radiation damage.
[6] Similar dosimetric variables have been associated with treatment-related necrosis following repeat stereotactic radiosurgery for brain metastases. [23] Dose escalation in the recurrent setting should, therefore, be performed thoughtfully.
Several contemporary studies utilizing higher-dose reirradiation regimens have identi ed dose-response relationships in regard to treatment outcomes. The Italian Association of Radiation Oncology described 300 recurrent glioma patients retreated with a median biological effective dose (BED) of 43 Gy. They identi ed reirradiation dose > 43 Gy was associated with improved OS compared to < 43 Gy (12.2 months vs 6.7 months). [ . The median PFS and OS were encouraging at 13.9 and 14.2 months, respectively. They also reported a modest incidence of late grade 3 toxicities at 8.3%. [29] The overall incidence of grade 3 toxicity in our study was similar (5.6%) and there were notably no grade 4-5 PT-related toxicities. These outcomes are particularly encouraging given the higher reirradiation doses used in this series. Based on these results, PT-reirradiation appears similarly effective but potentially favorable in terms of toxicity compared with photon reirradiation and may facilitate more aggressive salvage therapy.
Malignant gliomas are highly heterogeneous tumors with varying prognoses. Therefore, patient selection and use of molecular biomarkers are critical to identify patients who may bene t most from aggressive salvage therapies and advanced radiation modalities. In our study, survival curves for the ≥ 50 Gy (CGE) cohort prominently separates around 9-12 months, suggesting patients with > 1 year life expectancy would derive the greatest bene t from this treatment paradigm. In addition, future trials should also focus on QOL and neurocognitive preservation, domains that were not available for analysis in the present dataset. Prior studies have demonstrated more optimal QOL preservation for patients with recurrent glioblastoma treated with PT-reirradiation. [30] For patients with newly diagnosed low-grade glioma, PT is reported to preserve neurocognitive test scores, and even result in improved scores for left-sided tumors. [31,32] Given the extended survival of many patients, particularly those with non-glioblastoma, the risk of neurocognitive deterioration should be considered when deciding on RT modality, and pencil-beam scanning PT may further preserve QOL and/or reduce toxicities compared with passive-scattering PT.
Lastly, there remains no clear standard for dose and fractionation after recurrence; however, fractionated RT may be better suited for dose escalation to minimize the late effects on normal brain tissue given its inherent low α/β ratio. [33] Despite the use of a prospectively-collected multi-institutional dataset and large sample size, several limitations to the current work are notable. There were limited molecular data, particularly for nonglioblastoma patients, which could more accurately stratify patient analysis. In addition, the database was unable to collect patterns of recurrence and detailed dosimetric parameters, including extent of eld overlap and OAR doses. These speci c factors are critical to optimal patient selection and more detailed outcomes analyses. More robust prospective evaluations of these factors will be needed on future prospective institutional trials. Lastly, PCG toxicity data relies on institutional evaluation and the database does not provide primary radiographic data. Therefore, rates of radiation necrosis and estimates of PFS has a degree of uncertainty given the di culty in distinguishing pseudoprogerssion versus disease progression. Some patients did not continue regular follow-up at their proton center of treatment, potentially underestimating certain events.
In conclusion, our series reports 109 patients with recurrent malignant glioma who received PTreirradiation. This study demonstrates comparable e cacy and favorable tolerability relative to photon series. Dose escalation > 50 Gy may be more safety achievable with PT and further investigation is warranted for well-selected patients. Future studies are needed to quantify potential neurocognitive and QOL advantages of PT in the reirradiation setting and correlate cumulative OAR doses with risk of highgrade toxicities.

Declarations
Compliance with Ethical Standards Funding: No funds, grants, or other support was received for this manuscript.
Con icts of interest: CBS reports a Varian Medical Systems honorarium outside of the submitted work.
MVM reports a Varian Medical Systems honorarium outside of the submitted work. The remaining authors have no relevant nancial or non-nancial interests to disclose.
Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964