Re-Irradiation for Recurrent Brain Tumors: A Retrospective Study from a Tertiary Hospital in Saudi Arabia

Objective To analyze the post-re-RT progression-free survival (PFS) and incidence of radio-necrosis (BRN) in patients with recurrent primary brain tumors and to explore the associated factors. Method A retrospective cohort study that included 15 pediatric and adult patients with primary brain tumors who were treated with re-RT between 2011 and 2020. The study endpoints included the post-re-RT PFS, which were analyzed using Kaplan-Meier survival analysis, and the incidence of radio-necrosis. Baseline demographic and clinical data, primary radiation therapy (RT1) parameters and outcomes, and re-RT parameters and outcomes, were analyzed as factors for the two outcomes.


Introduction
Primary brain tumors constitute a heterogenous group of benign and malignant tumors that develop in the brain structures including parenchyma and annex tissues. They are characterized by a high mortality and poor functional outcome, resulting into a substantial disease burden, and high propensity for recurrence. [1][2][3] The worldwide incidence of primary brain cancers was estimated as 10.82 per 100 000 person-years, with a signi cant age-disparity depending on the tumor type and histology. 4 Recent progress in molecular biology, imaging, surgery and radiation therapy has enabled advanced understanding and management of primary brain tumors, especially gliomas and primary central nervous system lymphomas that represent the most common pathological types. Nevertheless, the standard treatment remains relatively aggressive, including surgical resection followed by adjuvant radiotherapy (RT), using either conformal intensity-modulated radiation therapy (IMRT) or stereotactic radiotherapy (SRS), and in some cases adjuvant chemotherapy (CT). 3,5−8 Furthermore, in refractory or relapsing cases a re-irradiation (re-RT) and or re-resection can be proposed, entailing concerns regarding the risk of radionecrosis that may occur several months to several years following re-RT. [9][10][11][12][13] Currently, there is no standard of care, notably dose regimens, regarding re-RT of brain tumors and the prospective data addressing this approach are scarce 9 . The safety of re-RT approach is limited by the capacity of brain recovery after radiation, which depends on the initial biologically effective dose (BED) as well as on the time interval between the primary radiation (RT1) and re-RT (RT1-re-RT interval). 11 Consequently, a low RT1 dose and an increased RT1-re-RT interval are supposedly favorable for a re-RT indication 11 ; however, some data have shown no correlation between RT1-re-RT interval and brain tolerance to re-RT. 9, 10 On the other hand, recent advances in RT technique, such as IMRT and SRS, allowed reduction of the treatment volume, thereby sparing normal tissue and reducing the proportional risk of radiation toxicity. 9,10,14 In our center, which is a referral center in radiation oncology, a selected number of patients with recurrent primary brain tumor have bene tted from re-RT over the last years. This study aims at analyzing the safety and e cacy of re-RT among this cohort of patients, by estimating the incidence risk of brain radionecrosis (BRN) and the progression-free survival (PFS), post re-RT, and analyzing the associated treatment parameters, notably the cumulative radiation doses for brain and organs at risk. Such data would provide a more nuanced insight into the clinical bene t of re-RT for brain tumors and enable determine eventual irradiation dose thresholds and toxicity pro les, which may give direction for future randomized trials. Additionally, data from this study would supply further systematic reviews and metaanalyses.

Design & Setting
A retrospective cohort study was conducted at the Radiotherapy Unit of XXX in XXX, XXX, between 2011 and 2020. The study received ethical approval of the institutional review board of XXX.

Participants
Fifteen patients were identi ed with primary brain tumors who were treated with re-RT during the study period, either with 3-D conformal radiotherapy, IMRT or SRS. Both pediatric and adult patients were included.

Data collection
A pre-formatted Excel sheet was used to collect the following data: 1) baseline demographic and clinical data including age, gender, and pathology type, grade, and side; 2) RT1 parameters including radiation technique, total radiation dose, dose per fraction, number of fractions, duration, and maximal dose for optic nerves, optic chiasm and brainstem; 3) RT1 outcomes including time from RT1 to disease progression and re-resection; 4) re-RT parameters including RT1-re-RT interval, technique, total radiation dose, dose per fraction, number of fractions, duration, and maximal dose for optic nerves, optic chiasm and brainstem, in addition to the cumulative radiation doses for the same structures; 5) re-RT outcomes including progression and time from end of re-RT to progression, and BRN. Alpha beta ratio of 3 (for late responding organs) was used for re-RT composite dose calculations. Patients were assessed clinically for neuropathy in follow-up clinic.

Statistical methods
Statistical analysis was performed with the Statistical Package for Social Sciences version 21.0 for Windows (SPSS Inc., Chicago, IL, USA). Categorical variables are presented as frequency and percentage, while numerical variables are presented as mean ± standard deviation (SD) or median (range) as applicable. Kaplan-Meier survival function was used to plot the survival curve for the progression-free time after re-RT, and to estimate the mean and median PFS. Further, factors associated with PFS were analyzed using Kaplan-Meier survival with calculation of the Log-rank; aLog-rank<0.05 was considered to reject the null hypothesis.

Demographics and rst radiation therapy parameters
Fifteen patients with brain re-irradiation were eligible and were included in the study, age range was 11 -75 years and 8 of them were male. Pathology showed 7 cases of glioblastoma and 5 cases of anaplastic ependymoma, and 13 patients were grade III or IV. The rst RT used VMAT in 12 patients, the mean duration was 42 days, and the median number of fractions was 30. The median maximal equivalent dose in 2 Gy (EQD2) received to brainstem was 37.05 Gy. Disease progression occurred after a mean time of 29 months and 7/15 underwent re-resection (Table 1).

Re-irradiation parameters
The mean interval from rst RT to re-RT was 24 months (range=2 -60). The mean total cumulative dose after re-RT as per EQD2 was 101.97 Gy (max 135.6 Gy). The total mean (max) cumulative doses for organs at risk as per EQD2 after re-RT were 54.05 (92.93) Gy for brain stem, 41.19 (87.94) Gy for optic chiasma, and 28.79 (77.18) Gy and 28.6 (88.71) Gy for left and right optic nerves respectively. The median total re-radiation dose was 35 Gy delivered over a median 10 fractions and over a mean period of 18.7 days. The brainstem received a median maximal dose EQD2 of 17 Gy, resulting in a median cumulative dose of 54.7 Gy (Table 2). No evidence of neuropathy was noted clinically. Outcomes -Progression and radiation necrosis Subsequent to re-RT, disease progression occurred in 10/15 patients after a median follow up time of 4.5 months (range = 0 -13). The mean and median PFS were 5.13 months (95%CI=2.51 -7.76) and 4 months (95%CI=0 -9.1) respectively (Table 2, Figure 1).
Radiation necrosis occurred in 2 patients, giving an incidence of 13.3% (95%CI = 1.7% -40.5%) ( Table 2). It was con rmed on MRI spectroscopy. The characteristics of the two patients who developed brain radiation necrosis are presented in Table 3. Both were female, aged 11 and 46 years old. Intervals between rst RT and re-RT were 20 and 12 months. Both patients had disease progression. Factors associated with progression-free survival The Kaplan-Meier survival analysis showed no statistically signi cant factor associated with PFS (Log-rank>0.05); however, some observations are worth noting. PFS was relatively longer among patients with grade II disease (9.0 vs <6 months). Regarding treatment parameters, a longer PFS was associated with a longer RT1 -re-RT interval, a longer re-RT duration. The two patients who developed brain necrosis -one had glioblastoma and the other anaplastic ependymoma-had a relatively shorter PFS (mean=3.0 vs 5.6 months, p=0.342) compared to their counterparts respectively (Table 4). Kaplan-Meier survival analysis; event= progression or death after re-radiation RT1: First radiotherapy; re-RT: Re-irradiation.

Summary of ndings
Over the 10-year period of the study, only 15 patients underwent re-RT for primary brain tumor in our institution. With respect of the size limitation of this cohort, ndings support that recurrent or treatmentresistant, high-grade glioblastomas and anaplastic ependymomas represent the most common indications of re-RT in primary brain tumors. These tumors are characterized with early progression after the rst treatment, leading to shortened intervals between the rst RT and re-RT in majority patients. Regarding safety in our institute, ndings suggest that re-RT is associated with an incidence of brain radio-necrosis of 13.3%, depending on radiation dose and pathology.

Indications of brain re-irradiation
The rst published data regarding re-RT of the cranium and CNS date back to nearly one century, with the works of Beclere and Levy reporting cases of single or twice re-RT. 15,16 In a historical series of 16 patients with brain tumors, mainly gliomas, published in 1926, 2 of the cohort patients were reirradiated after 6 and 12 months of the rst RT respectively. 17 Relapsed brain tumors constitute the most common indication of fractioned re-RT, which remains the only treatment option in majority of these patients.
Several pathological types of primary brain tumors have been reported in re-RT cohorts, notably glioblastomas, anaplastic gliomas, medulloblastomas, ependymomas and meningiomas. 9,18 In the pediatric population, recurrent medulloblastoma/primitive neuroectodermal tumors, germinoma and nongerminomatous germ cell tumors, and all-grade gliomas were reported as indications for re-RT, which is often adjunctive to re-resection. [19][20][21] Both in adults and children, recurrence in primary brain tumors is often observed in high-grade tumors or atypical ones. [21][22][23] This is consistent with the present study's cohort including 40.0% and 46.7% of grade III and IV tumors, respectively. Consequently, the primary therapeutic plan for such patients with high recurrence pro le should be adapted at two levels: rst, to reduce the risk of recurrence or delay its occurrence, by performing a more aggressive surgery to minimize the residual disease; The indication of re-RT is not limited to primary brain tumors, brain metastasis was early considered among the eventual indications of brain re-RT. 24 However, unlike primary brain tumors, the metastatic indications often require whole brain radiotherapy due to the frequent presence of multiple, diffused lesions to be treated, which may result in a higher risk of brain atrophy or failure. 25,26 Nonetheless, this indication was out of the scope of the present study.
E cacy and safety balance of brain re-irradiation It is common knowledge that the growing use of re-RT in both primary and metastatic brain tumors is associated with improved overall survival especially with the advent of high-precision RT techniques.
However, the expected clinical e cacy should be assessed cautiously and weighed against the risk of radiotoxicity on the healthy structures notably the brain parenchyma and organs at risk such as the optic chiasm and optic nerves. Such caution balance combined with the scarcity of clinical data result in lack of evidence-based consensus and persisting controversies regarding the dosimetry and treatment regimens of such patients. 27,28 Further, the growing number of single-center reports is inconsistent with respect of the e cacy and safety pro les.
Findings from the present study suggest a low-e cacy pro le for re-RT given the high incidence of disease progression (66.7%) occurring early, i.e., with a median time of 4.5 months from the end of re-RT. On the other hand, the safety pro le of re-RT in the present cohort was relatively high, as only 2 patients developed brain radio-necrosis representing an incidence risk of 13.3%; however, the severity of the necrotic lesions in the two concerned patients and their respective functional outcomes were not reported.
Further, given the small size of the present cohort, the incidence of radio-necrosis can statistically be inferred with a large con dence interval of 1.7% -40.5%, which is conclusive regarding the safety pro le.
By comparison, a retrospective cohort by Stiefel et al. analyzed the outcomes of 76 patients with recurrent brain tumors, of whom 34 (44.7%) were primary and the others metastatic tumors. Outcomes in patients with primary tumors showed a median overall survival of approximately 14 months (range=21 -42 months) after re-RT, while the median time to local recurrence was not reached due to high proportion of censoring and was less than 6 months in majority of non-censored patients. Regarding radio-toxicity, authors did not carry out separate analysis for the primary tumors group; instead, they reported an overall incidence of radio-necrosis of ~12%, including 5.3% in the acute phase (<12 weeks post-re-RT).
Furthermore, low-grade toxicity events such as edema, headache and fatigue, were observed in 74% of the patients. 23 A comparable incidence of radio-necrosis was in our present study; however, we did not report the low-grade toxicity events.
A clinical trial by Møller et al. randomized 31 candidates for brain re-RT for recurrent high-grade glioma, of whom 81% had glioblastoma, into 4 treatment groups, each a speci c sequential regimen. Of the 4 groups, 3 had a planning target volume (PTV)<100cm 3 including Group 1 (3.5 Gyx10); Group 2 (3.5 Gyx10+7Gy boost); and Group 3 (5.9 Gyx5); whereas the 4th group (39 Gy x 10) had a PTV 100-300cm 3 . The median PFS in the total study population was 2.8 months and the median overall survival was 7 months. However, due to high censoring and early mortality, only 7 patients reached a PFS of 2.5 months, 5 of them received the rst regimen (3.5 Gyx10, PTV<100 cm 3 ) and 2 received the second one (3.5Gy x 10+7Gy boost, PTV<100 cm 3 ). Beside these low e cacy pro les, a low-safety pro le was reported including high rate of minor early toxicity and serious late events in 3 out of 7 patients with PFS>2.5 months, including radio-necrosis and irreversible white matter changes with neurofunctional sequelae. 29 In pediatric patients, a 24-year retrospective study by Bouffet et al. reported a higher e cacy pro le in a series 18 children who were selected for full re-RT, with or without re-resection, among a total 47 with recurrent ependymoma. Re-irradiation was associated with a 3-year survival rate of 81%, compared with only 7% among non-re-irradiated children. Additionally, re-irradiation was associated with a signi cantly delayed disease progression/relapse (3-year PFS=61% vs 25%) compared to the initial relapse after the rst RT respectively, p=0.003. However, this high-e cacy and survival pro le was achieved at the price of a signi cant decline in the intellectual quotient in re-irradiated children from pre-to post-re-RT times. 30 Improving e cacy and controlling radiotoxicity in reirradiation Although not statistically signi cant, due to small sample size, ndings from the present study suggest the presence of factors that may improve the post -re-RT prognosis, notably by prolonging the PFS.
Tumor characteristics that showed association trend with PFS included lower tumor grade having longer PFS than high-grade. PFS showed likelihood of positive association with RT1-re-RT interval and re-RT duration, and was likely to be increased in patients who received adjuvant chemotherapy.
Observations from animal and clinical studies support the relevance of these factors as determinant of both the e cacy and brain tolerance to re-RT. 9 In an interesting cohort including 233 patients with recurrent grade II (40%) and III (22%) gliomas and glioblastoma (38%), Combs et al. proposed a prognostic score to determine potential candidates for re-RT based on the prognostic value of the baseline clinical and pathological characteristics. Findings demonstrated that the tumor histological type, patient age at diagnosis, and RT1-re-RT interval were the strongest predictors of survival post re-RT. 31 The use of this score should be encouraged and further validated in different settings and cohorts, as it has a direct clinical impact on decision-making regarding the re-RT indication and could result into a consensual approach.
Further efforts are made to improve the prognosis of patients with recurrent brain tumors, while improving their quality of life. Some studies have suggested that the risk of radiotoxicity associated with re-RT can be mitigated by using proton beam therapy (PBT). A study by Mizumoto et al. (2013) analyzed the e cacy and toxicity of re-RT using conventional PBT versus conventional RT and stereotactic radiotherapy, in a cohort of 26 pediatric and adult patients with recurrent malignant brain tumors, including 15 (57.7%) glioblastomas multiform and 6 (23.1%) grade 3 gliomas. Toxicity outcomes showed 2 (7.7%) cases of radio-necrosis, which were well controlled in the second year following re-RT; beside minor acute toxicity events. E cacy outcomes showed one-and 2-year overall survival rates (55.4% and 45.1%) and one-and 2-year local control rates (43.0% and 18.4%) respectively in the total population; and the one-year overall survival rate was higher in patients treated with PBT (75.0%). 32 More recently, Scartoni et al. assessed the health-related quality of life among 33 patients with recurrent glioblastoma who underwent re-RT using PBT. Findings showed a clinically and statistically signi cant improvement in global health at the early post-re-RT phase, followed by progressive improvement in social functioning and motor dysfunction dimensions. However, a decline in the cognitive and emotional functioning was observed among these patients and was deemed as being non-signi cant. Further, the median PFS and overall survival were 5.9 and 8.7 months respectively. 33

Limitations
The present study is limited by the retrospective design and the small sample size which impact the external validity of the ndings and conclusions. This highlights the urgent need for randomized trials and prospective cohorts to provide a more accurate evaluation of the clinical bene t of re-RT in the local population.

Conclusion
Fractioned re-RT is likely to be a safe therapeutic option for recurrent primary brain tumors, with relatively low incidence of radio-necrosis. However, its e cacy may be dependent on the pathology parameters and treatment regimen, and could be confounded with factors impacting radiotoxicity notably the maximal and cumulative radiation doses on the brain parenchyma and organs at risk. Such observations highlight the relevance for an anticipative approach to determine candidates for re-RT based on predicted safetye cacy pro les by weighing the bene ts in overall and progression-free survival with the adverse effects and other relevant parameters of quality of life. There is unmet need for conducting further randomized trials and prospective cohorts to evaluate more accurately the clinical bene t of re-RT in the local population.

Declarations
Con ict of Interest: The Authors declare that they have no competing interest.

Funding Statement
There was no funding for this work.