Involvement of subventricular zone shortened the survival of pediatric glioblastoma

PURPOSE Glioblastoma involved with subventricular zone (SVZ) predicted a worse outcome in adult, but we know less about it based on children population due to the rarity of pediatric glioblastoma (pGBM). We performed this study to probe into the clinical and prognostic features of glioblastoma involved with SVZ in children. METHODS We selected thirty-one patients diagnosed with pediatric supratentorial glioblastoma at our department between January 2015 and January 2020. Clinical data and prognostic results were reviewed retrospectively.


Introduction
Glioblastoma (GBM) is a rare malignant tumor of the central nervous system in children. According to statistics, the annual incidence rate is 1.4/ million, accounting for 3% of the primary tumors of the central nervous system in children [1,2]. Several studies had demonstrated that maximum safe resection of tumor and postoperative radiotherapy (RT) and chemotherapy (CT) were effective treatment for pediatric glioblastoma (pGBM) [3][4][5][6][7]. However, the median overall survival (OS) was still between one and two years after active treatment [5,8,9]. Subventricular zone (SVZ) is a neurogenic area which harbors neural stem cells. In recent research, glioma stem cells were found in SVZ, which play a critical role in the occurrence and progression of glioma [10][11][12]. It has been proved that adult GBM involved with SVZ predict an adverse prognosis and tend to present a multifocal tumor phenotype in many studies [13][14][15][16].
Because of the rarity of pGBM, the clinical features and prognostic outcomes about pGBM involved with SVZ had never been described before. To better comprehend its clinical features and provided basis for judgment of prognosis preoperatively, we retrospectively analyzed the institutional database for 31 patients with pediatric supratentorial GBM diagnosed and treated at our hospital during 2015-2020.

Ethical statement
The implementation of this study had been approved by the ethics committee of the First A liated Hospital of Zhengzhou University. The use of data for analysis had been agreed by all patients. No personal information was disclosed in this article.

Patient selection
Thirty-one patients with histopathologically con rmed diagnosis of GBM at our hospital between January 2015 and January 2020 were selected for review. All patients were newly diagnosed with primary supratentorial GBM and were under 18 years at the time of diagnosis. In addition, cases with incomplete clinical data or not treated according to the Stupp protocol were excluded.

Data collection
Data collection included age of diagnosis, sex, preoperative KPS, tumor side, adjuvant therapy, tumor volume, edema index (EI), extent of resection (EOR), involvement of SVZ, immunohistochemistry indexes (ATRX, Ki-67, P53), genetic testing (IDH1 and IDH2), overall survival (OS) and progression-free survival (PFS). Evaluating of KPS was routinely completed by the neurosurgeon before operation. Postoperative radiotherapy was given at 2gy per day, 5 days per week, with a total dose of 50-60gy.Temozolomide was prescribed for chemotherapy according to Stupp 5/28 regime [3]. Tumor volume was measured with a preoperative T1-weighted MR images after injection of gadolinium. Peritumoral brain edema (PTBE) was evaluated on T2-weighted images. The maximum perpendicular diameters (A and B) of the tumor and the PTBE were measured on the axial images, and the diameter in the coronal direction (C) was measured on the coronal or sagittal image, the formula used to calculate tumor and PTBE volume was V = 4/3π × A/2×B/2×C/2. To reduce arti cial bias, tumor and PTBE volume were measured by three neurosurgeons and the mean value was recorded. Severity of edema was evaluated by edema index (EI), it was calculated by the formula: EI = (V tumor +V edema )/V tumor [17]. The extent of resection was veri ed by comparing preoperative MRI with that performed within 72 hours after surgery. GTR was de ned as no residual enhancement on postoperative enhanced MRI and STR was de ned while any residual enhancement was observed. Typical images of involvement with SVZ are shown in Fig. 1. The neurosurgeons who completed the collection of imaging data knew nothing about the clinical characteristics and outcomes of the patients. OS was de ned as time between the initial treatment and death or the last follow-up. PFS was de ned as time between the initial treatment and diagnosis of tumor recurrence. Follow-up was performed for all patients by making phone call once every two months or outpatient review on a timely basis.

Statistical analysis
Age, tumor volume, EI, Ki-67, PFS and OS were analyzed as continuous variables, whereas gender, tumor side, immunohistochemistry indexes (positive or negative), genetic testing (mutant or wild-type), KPS (≥ 80 or 80), EOR (GTR or STR), adjuvant therapy (Radiotherapy combined with chemotherapy; Other: incomplete adjuvant therapies including chemotherapy only, radiotherapy only and none) and involvement with SVZ were analyzed as a categorical variables. Mann Whitney U test was used to evaluate the distribution of continuous variables and Fisher's exact test was employed to evaluate the distribution of categorical variables. Kaplan-Meier method was used to estimate progression-free survival (PFS) and overall survival (OS). Variables analyzed in univariate Cox regression model analysis included the variables that may affect the survival including age, sex, tumor volume, EI, preoperative KPS, EOR, adjuvant therapy, Ki67 and SVZ involvement. Log-rank test was employed to compare survival distribution. Multivariate Cox proportional hazard model analysis included all variables with P < 0.05 in univariate analysis. P < 0.05 was de ned as statistically signi cant. All statistical analyses were performed using IBM SPSS 22.0 software. Box plots and survival curves were completed by Graph Pad Prism 9.

Results
Involvement of SVZ relates to large tumor volume, mild edema, passive treatment and poor prognosis.
The distributions of continuous variables and categorical variables with statistical analysis results were summarized in Table 1 and Table 2 respectively. Box graphs showing the distribution of continuous variables were presented in Fig. 2. No gender or side predominance was found for tumor involved with SVZ. Age, Ki-67, EI, EOR and preoperative KPS didn't differ in patients who had a tumor involved with SVZ. However, patients who had a tumor involve with SVZ were more likely to receive passive adjuvant therapy (p = 0.029). There were no differences in the results of ATRX and p53 immunohistochemistry between patients with SVZ involvement and patients without SVZ involvement. The tumor volume was signi cantly larger when SVZ was involved (p = 0.007), but it tended to have a mild PTBE (p = 0.010). We observed a decline on PFS (p = 0.001) and OS (p < 0.001) in subset of tumor involve with SVZ.
Involvement of SVZ is an independent prognostic factor for pGBM.

Discussion
Involvement of SVZ had been proved to be a prognostic factor for adult GBM in many researches [14][15][16][18][19][20][21]. The clinical signi cance of this nding was to help us judge the prognosis of patients before surgery. On the one hand, several studies have reported the discovery of glioma stem cell in SVZ, where provides a protective environment and increase resistance to irradiation and chemotherapy [22][23][24][25][26][27][28][29]. On the other hand, tumor cells could shed into the lateral ventricle and spread with cerebrospinal uid, resulting in poor progression free survival and overall survival. However, due to the rarity of pediatric glioblastoma, nearly all of the reports about SVZ were based on adult population in the past. The only report containing some cases of high-grade glioma in children found a poor prognosis for high-grade glioma patients with involvement of SVZ, suggesting that the prognosis of high-grade glioma patients with involvement of SVZ in children may be as poor as that of adults [15]. We are the rst to research on the impact of tumor involved with SVZ on survival in children population. Our results indicated that involvement of SVZ is a prognostic factor for PFS and OS in pGBM independent of EOR and adjuvant therapy.
Researchers held different opinions on the relationship between SVZ involvement and total resection rate in the past reports [14,21]. In order to avoid the risk of communicating hydrocephalus and the spread of tumor with cerebrospinal uid caused by ventriculotomy, many neurosurgeons tend not to remove the entire tumor which involved lateral ventricle [30]. However, it had also been reported that surgical incision of lateral ventricle to achieve complete resection of supratentorial GBM has a better prognosis [31]. Our results con rmed that there was no relationship between GTR and involvement of SVZ, however, since GTR could prolong the survival of pGBM in our study, we advise to realize GTR with the application of microsurgical technology and intraoperative neuronavigation as far as possible even facing tumor involved with SVZ.
Adult GBM with no SVZ involvement were more likely to receive active adjuvant therapy in the literature [14], they attributed this difference to raised gumption for treatment given by mild symptoms (higher KPS) of patients, we also found active treatment were given to patients with no SVZ in pGBM cases, but our patients didn't represent a serious condition when SVZ was involved.
Harat M et al. showed SVZ in ltration of GBM by O-(2-[18F] uoroethyl)-L-tyrosine (FET) PET scan, they found SVZ in ltration was correlated to larger tumor volumes [32]. Our results also showed that supratentorial GBM with SVZ in ltration were larger in children. To some extent, tumor itself may not originated from SVZ, but invaded SVZ with its growth. This view was supported by the theory that tumor stem cells migrate along the CXCL12 / CXCR axis or pleiotrophin-driven axis to SVZ [33,34]. According to our results, we could speculate that the same mechanism of glioma stem cell migration also exists in children.
Although it has been reported that the degree of PTBE is not related to tumor volume [35], however, surprisingly, we found that pGBM exposed to SVZ had lower edema index (EI), which may be related to the expression of aquaporins and tumor microenvironment, we hope the speci c mechanism can be studied in future. Meanwhile, mild PTBE suggests that patients with SVZ involvement may needn't to use steroids or mannitol before surgery to reduce tumor edema.
Status of IDH gene has been tested routinely in our center. Because of the low incidence of IDH gene mutation, relationship between IDH gene mutation and SVZ involvement cannot be analyzed properly. In consideration of it had been reported that IDH gene mutation is not associated with SVZ in ltration in adults GBM [25,36], we need a larger sample size to con rm the result in pGBM. In addition, we found that there was no difference in the expression of ATRX and p53 between SVZ contact subset and no SVZ contact subset, which indicated that the poor prognosis caused by SVZ exposure was not related to these pathological prognostic indicators.
At present, no particular treatment was applied on GBM with SVZ involvement. In the retrospective analysis, whether patients can bene t from radiotherapy for SVZ is still controversial [37,38]. A prospective study found that patients with radiation necrosis in SVZ have a longer survival time [39]. Several drug treatments are being studied, it had been reported that CXCL12 inhibitors can improve radiosensitivity and reduce tumor cell proliferation in animal models [33]. Besides, administration of drugs, vectors or cells in the lateral ventricles, which can bypass the blood-brain barrier, have been considered as a more effective treatment in theory, and obtained positive results in animal models [40][41][42][43]. Our study shows that RT + CT is still an effective therapy of pGBM, but the treatment of pGBM with SVZ in ltration needs further study in the future.

Conclusion
We have proved that SVZ involvement is an independent prognostic factor in this study, and we found it also associate with large tumor volume, mild PTBE and passive adjuvant therapy. Even though there weren't any available speci c treatment for GBM with SVZ involvement, some possible therapies have been proved to be effective in animal models. We hope more research could be developed to optimize the treatment strategy of GBM involved with SVZ. We thank all the patients who trusted us and all the physicians and staff who helped this study and authors whose work was referred or quoted in this study.

Funding
Medical Science and Technology Project of Henan Province (SBGJ202003032)

Con icts of interest
Neither I nor my spouse/partner has a commercial interest, nancial interest, and/or other relationship with manufacturers of pharmaceuticals, laboratory supplies, and/or medical devices or with commercial providers of medically related services.

Availability of data and material
The datasets used during the current study are available from the corresponding author on reasonable request. The implementation of this study has been approved by the ethics committee of the First A liated Hospital of Zhengzhou University. The use of data for analysis has been agreed by all patients. No personal information will be disclosed in this article.

Consent to participate
Consent to participate in our study had been obtained from parents of every patient.

Consent for publication
Written informed consent for publication was obtained from all participants. Table 1 Comparisons of categorical variables and statistical results based on SVZ involvement. Unavailable EOR, extent of resection; GTR, gross total resection; STR, sub-total resection; RT, radiotherapy; CT, chemotherapy.

Figure 2
Box graphs showing the distribution of continuous variables were presented in Fig2.