Dose–response relationship in patients with newly diagnosed atypical meningioma treated with adjuvant radiotherapy

This study aimed to identify the radiation dose–response relationship in patients with newly diagnosed atypical meningioma (AM) treated with adjuvant radiotherapy (ART) using conventional fractionation. In total, 158 patients who underwent surgery and ART between 1998 and 2018 were reviewed. Among these patients, 135 with complete information on radiotherapy (RT) dose/fractionation and pathological reports were analyzed. We entered RT dose as a continuous variable into the Cox regression model using penalized spline to allow for a nonlinear relationship between RT dose and events. Local control (LC), progression-free survival (PFS), and overall survival (OS) were evaluated. The corresponding biological equivalent dose in 2 Gy fractions (EQD2) was calculated using an α/β ratio of 4 Gy. The median follow-up duration was 56.0 months. The median ART dose delivered was 61.2 Gy in 24–34 daily fractions, corresponding to a median EQD2 of 59.16 Gy. In multivariate analysis, larger size and higher mitotic count were associated with significantly reduced LC (P < 0.001 and P = 0.002, respectively), PFS (P < 0.001 and P = 0.006, respectively), and OS (P = 0.006 and P = 0.001, respectively). Meanwhile, a higher RT dose was significantly associated with improved LC, PFS, and OS. Moreover, RT showed a dose-dependent effect on LC, PFS, and OS; local failure, tumor progression, and death were reduced by 12%, 12%, and 16%, respectively, per 1 Gy increase in the dose (EQD2). The dose of ART in AM has a dose–response relationship with LC and survival outcomes.


Introduction
Meningioma is the most common type of non-glial brain tumor and accounts for more than 30% of all primary brain tumors [1]. Most meningiomas are benign and indolent, but atypical and malignant meningiomas have aggressive tumor characteristics [2,3]. Atypical meningiomas (AMs) are classified as World Health Organization grade 2 and account for 4%-15% of all meningiomas [4,5]. The recurrence rate of AM after surgery is known to be 30%-40%, regardless of the resection extent [4].
The first-line treatment for patients with AM is maximal safe surgical resection [6], but there is no clear consensus on the role of adjuvant radiation therapy (ART) in this patient population [7][8][9][10][11][12][13][14][15]. However, several previous retrospective studies have reported that ART is significantly associated with improved local control (LC), progression-free survival (PFS), and overall survival (OS) [8,10,11,15]. Additionally, in our earlier study, ART showed significant benefits in PFS and progression/recurrence rates in patients with AM than active surveillance, regardless of the extent of surgical resection [16]. Therefore, based on the results of retrospective studies, ART can be considered an important treatment option for disease control after surgical resection in patients with AM.
Several studies and current guidelines suggest that the total dose for ART in AM varies from 54 Gy to 61.2 Gy [17][18][19][20]. Previous studies of stereotactic radiosurgery or proton therapy for meningioma have reported better treatment outcomes in the high-dose group compared to the low-dose group [18,21,22]. However, few studies have reported a dose-response relationship in ART using a conventional fractionation scheme in patients with AM.
This study aimed to evaluate the predictors for treatment outcomes in patients with AM receiving ART. Furthermore, we sought to identify the dose-response relationship with treatment outcomes in ART-treated patients with AM using conventional fractionation following surgical resection.

Patient population
We previously reported an analysis comparing treatment outcomes between ART and surveillance groups following surgical resection in 518 adult patients with newly diagnosed intracranial AM between 1998 and 2018 at four institutions. The exclusion criteria for this study were as follows: ART using stereotactic radiosurgery; multiple meningiomas; neurofibromatosis type 2; optic nerve sheath meningioma; history of brain radiotherapy (RT); history of other brain tumors; and history of other malignancies within 5 years, excluding in situ tumors of the uterine cervix or breast, basal cell carcinoma, and differentiated thyroid cancer. In total, 158 patients received ART in the population of this previous study. Among these, our current cohort included 135 patients with complete information on RT dose/fractionation and pathology reports (including mitotic count).

Treatment and follow-up
All patients underwent ART following surgical resection. Tumor size was determined by the largest diameter on preoperative enhanced T1-weighted magnetic resonance imaging (MRI). The extent of resection was determined by comprehensively evaluating the surgeon's description and postoperative MRI findings. The extent of resection was divided into gross total resection (GTR) and subtotal resection (STR); patients who underwent biopsy were excluded from this study. We defined ART as RT conducted within a year without any evidence of disease progression after surgery. The median time from surgery to ART was 1.4 (interquartile range [IQR], 1.0-2.2) months. The equivalent dose in 2-Gy fractions (EQD2) was determined using an α/β ratio of 4 Gy suggested in stereotactic RT studies of benign brain tumors [23]. The dose of ART was determined by a radiation oncologist based on the extent of resection, mitotic count, tumor location, and brain and bone invasion. The median ART dose was 61.2 (range, 45.0-66.0) Gy in 34 (range, 24-34) daily fractions, corresponding to a median EQD2 of 59.16 (range, 43.50-69.12) Gy. This study defined ART dose > EQD2 57.42 Gy (corresponding to 59.4 Gy in 33 fractions) as a high dose and ART dose ≤ EQD2 57.42 Gy as a low dose. The clinical target volume included the residual gross lesion or postsurgical tumor bed with a 1.5-2.0 cm margin to the meninges and 0.5-1.0 cm margin to the brain parenchyma. The planning target volume was expanded in all directions by 0.3-0.5 cm relative to the clinical target volume. ART was delivered using two-dimensional RT (n = 8, 5.9%), three-dimensional conformal RT (n = 33, 24.4%), or intensity-modulated RT (n = 94, 69.6%); radiation was administered with 4-15 MV photons. Follow-up brain MRI was performed every 6-12 months for 5 years after treatment. We also reviewed the presence of radiation necrosis using imaging tests and medical records to evaluate radiation toxicity. This study defined radiation necrosis as a necrotic lesion newly identified on serial brain MRI following ART. We excluded lesions that developed after re-irradiation (external beam RT or Gamma Knife surgery [GKS]) for recurrent or residual tumors.

Statistical analyses
Local failure was defined as recurrence within a 2 cm margin from the residual lesion or tumor bed. LC was defined as the time from the surgery to local failure. Intracranial progression other than local failure was defined as distant intracranial progression. PFS was defined as the time from the surgery to any disease progression, death, or the last follow-up. OS was the time from the surgery to death from any cause. LC and survival outcomes were calculated by Kaplan-Meier analysis, and groups were compared using the log-rank test.
The Chi squared test or t-test were performed to compare variables according to treatment group. The Cox proportional hazards regression univariate and multivariate analyses were performed to identify independent prognostic factors for treatment outcomes. Multivariate analysis used the stepwise method, and all covariates (age, sex, tumor location, tumor size, extent of resection, mitotic count, brain invasion, bone invasion, and RT dose [continuous variable]) were entered and analyzed. The penalized spline in the Cox model allowed a nonlinear relationship of RT dose with the logarithm of the hazard ratio (HR) for LC, PFS, or OS, estimated from the full Cox regression model adjusted for age, size, mitotic count, and brain invasion. We plotted the penalized spline using the degrees of freedom in the multivariate additive Cox models function in smoothHR to obtain the optimal number of degrees of freedom in the extended Cox-type additive multivariate analysis. We defined statistical significance as a P-value < 0.05. All statistical analyses were performed using STATA software (version 17.0; Stata Corp, College Station, TX, USA).

Dose-response relationship between ART and survival outcomes
We categorically evaluated ART dose across multiple cutoff thresholds using multivariate analysis to determine significant predictors of PFS (Supplemental Table 2). After adjustment for age, sex, tumor location, tumor size, extent of resection, mitotic count, brain invasion, and bone invasion in the entire cohort, all doses above the threshold at each cutoff value were significantly associated with improved PFS (all P < 0.05).
To quantify the dose-response relationship, we used the ART dose as a continuous variable in the Cox regression model using penalized splines in smoothHR to allow for nonlinear relationships between the dose and treatment outcomes (Fig. 2). This model indicated that the risk of local failure, disease progression, and death continuously decreased with ART dose escalation. Consequently, the maximum dose in this study (EQD2 69.12 Gy) exhibited the lowest HR for LC, PFS, and OS. These results revealed the dose-response relationship of ART on LC, PFS, and OS, implying the possibility that further dose escalation may improve the treatment outcomes.

Toxicity
Only one patient developed asymptomatic radiation necrosis 25 months after ART, which did not require medical or surgical interventions. This patient had AM in the right lateral ventricle and received EQD2 60 Gy in 30 fractions following GTR. In addition, although radiation necrosis was detected on brain MRI in three patients, all of these developed after re-irradiation (GKS) for recurrent tumors and were excluded from the toxicity analysis.

Discussion
This study assessed treatment outcomes in patients with AM treated with ART and investigated dose-dependent effects on LC and survival outcomes. In the multivariate analysis, tumor size, mitotic count, and brain invasion were significantly associated with treatment outcomes. Importantly, our results indicated that ART exerted a dose-response relationship, and high-dose ART was associated with improved treatment outcomes.
Current guidelines generally recommend ART in patients with AM; however, the benefits of ART after GTR remain controversial [19,20]. Several retrospective studies have assessed the effects of ART compared to surveillance after GTR, but with inconsistent results due to their small sample  [25]. Furthermore, according to a recent meta-analysis, ART significantly increased PFS and OS, regardless of whether GTR or STR [26]. In addition, our previous study also demonstrated that ART resulted in significantly better PFS and progression/ recurrence rate than the active surveillance group, even in patients who underwent GTR [16]. Therefore, although randomized controlled trials to compare ART and active surveillance in AM are ongoing [27,28], several retrospective studies and meta-analyses suggest that ART should be considered following surgery in patients with AM, regardless of the extent of resection. The present study identified that larger tumor size, higher mitotic count, and brain invasion were significant risk factors for survival outcomes in patients treated with ART. These findings were consistent with those of previous studies [29][30][31]. Notably, our results indicated that survival outcomes according to the extent of surgical resection were not significantly different in patients treated with ART. Similarly, an analysis of Surveillance, Epidemiology, and End Results data revealed no significant OS difference between the GTR and STR groups. However, patients who underwent GTR had significantly better PFS than those with STR [32]. Based on these results, the authors suggest that maximal safe resection is reasonable in routine meningioma management to prevent disease progression, but caution is required since radical surgical strategy may cause severe complications after surgery. Therefore, we suggest that ART following adequate surgical resection can be considered, instead of excessive resection, in patients with expected postoperative sequelae.
The retrospective studies analyzing the effects of stereotactic radiosurgery in meningioma found that a higher radiation dose corresponds with better LC and survival outcomes [21,33]. For instance, Sethi et al. analyzed local recurrence in 101 patients with benign, atypical, or malignant meningioma who underwent GKS. Their results showed a relative reduction of 42% in local recurrence for each 1-Gy dose escalation [21]. Moreover, some studies have reported that proton therapy in meningioma also showed a dose-dependent effect [18,22]. The French group reported 24 patients with non-benign meningiomas who received postoperative proton/photon therapy; survival was significantly associated with higher radiation dose, and the cutoff value was 60 Gy (Relative Biologic Effectiveness [RBE]) [18]. Additionally, McDonald et al. reported the treatment outcomes after proton therapy in 22 patients with AM and showed significantly improved LC at radiation dose > 60 Gy (RBE) [22].
To the best of our knowledge, this is the first study to investigate the dose-response relationship in patients with AM treated with ART provided as a conventional fractionation dose scheme. Wang et al. analyzed the association between ART and OS benefit in 2515 patients with AM using the National Cancer Database. They reported that the radiation dose was not related to OS in patients who received STR [15]. However, this previous study compared only the OS rates between the high-dose (≥ 54 Gy in 30 fractions) and low-dose (< 54 Gy in 30 fractions) groups by subgroup  [34]. Meanwhile, our study observed a dose-response relationship between RT dose and treatment outcomes, and high-dose ART was significantly associated with better survival. Moreover, because asymptomatic radiation necrosis was observed in only one patient in our study, dose escalation within the ART dose range used in this study can be considered a safe and effective treatment strategy. Given that ART is widely used in AM, detailed consensus guidelines regarding ART dose prescription should be established. Currently, the standard dose for AM recommended by the guidelines is 54-60 Gy [19,20]. However, several studies are underway to attempt radiation dose escalation by additional boost using proton or particle therapy in patients with non-benign meningioma [35][36][37]. The PANAMA trial will apply a total dose of up to 68 Gy using proton therapy to patients with AM who do not undergo complete resection. Thus, the efficacy of a dose-intensified ART will be evaluated [36]. Additionally, MARCIE-Study has been developed to assess the survival outcomes of carbon ion boost after photon therapy in patients with AM who underwent incomplete resection or biopsy [37]. These ongoing trials will provide answers to important clinical questions about the need for dose escalation of ART in patients with AM. Our results lead us to expect promising results from these trials and may help determine the optimal dose until complete data are developed. Furthermore, the advancement in molecular stratification of meningiomas should be noted [32,38,39]. With these advancements, dose escalation of ART in meningioma may be applied to more suitable patients.
Our study has some limitations. First, this study had a retrospective nature, and inherent selection bias could not be completely avoided. Second, the sample size was insufficient for subgroup analyses under various conditions. However, our study selectively enrolled patients with AM who underwent ART, and the cohort size of this study was larger than that of other retrospective series. Third, we lacked detailed data on treatment toxicity, such as neurological dysfunction or quality of life, which plays a significant role in determining optimal radiation dose cutoff values. Fourth, multiple modalities, including two-dimensional RT, threedimensional RT, and intensity-modulated RT, were used for ART in this study. Thus, the heterogeneity of these RT techniques may affect treatment outcomes. Fifth, we evaluated LC, PFS, and OS as endpoints in this study. Among these, PFS included distant tumor control outside the irradiation field, which is not directly attributed to ART. Also, since all-cause mortality was included in OS, thus bias may occur in patients with local tumor control. Finally, because we used the ART dose recommended in the guidelines for most patients, we could not fully validate the dose-response relationship in wider dose distribution. Therefore, further prospective studies fully verifying ART dose escalation for patients with AM are warranted.
In conclusion, we found that larger tumor size, higher mitotic count, and brain invasion were significantly associated with worse treatment outcomes in patients with AM who received ART. In contrast, the extent of surgical resection did not significantly influence the prognosis of patients after ART. Furthermore, our results demonstrate that ART reveals a dose-response relationship and that the high-dose group shows better treatment outcomes compared to the low-dose group in patients with AM. Therefore, promising results are expected from ongoing dose escalation studies with additional boosts.