Different prognostic effect of surgery, radiation and chemotherapy in pediatric medulloblastoma less and over three years old: A Surveillance, Epidemiology, and End Results (SEER) analysis

To clarify the prognostic value of extent of surgical resection, radiation and chemotherapy in pediatric medulloblastoma patients < 3 years old and ≥ 3 years old. Methods We used the Surveillance, Epidemiology, and End Results program to identify 1,495 pediatric patients diagnosed between 1973 and 2016 with medulloblastoma. Patients with incomplete or unknown clinical information were excluded. Basic characteristics between patients < 3 years old and ≥ 3 years old were compared. Then, we used Cox regression to investigate the impact of extent of surgical resection, radiation and chemotherapy on patient outcome.


Introduction
Medulloblastoma is the most common malignant brain tumors in children, comprising for nearly 20% of pediatric brain tumors and 40% of all pediatric posterior fossa tumors [1][2][3]. The main treatments for pediatric medulloblastoma are maximal safe surgical resection, craniospinal radiation (CSI) and systemic chemotherapy [3,4]. Due to most common occurrence in brainstem and cerebellum, aggressive surgical resection may be associated with increased postsurgical complications, such as neurologic morbidity and cerebellar mutism. The survival bene t of gross total resection (GTR) is controversial in the literature, which some identify GTR is an independent prognostic factor [5][6][7], but others do not [8-10].
Current clinical risk strati cation of medulloblastoma is the main reference for postsurgical therapy, and the strati cation separates children into average risk and high-risk strata. High-risk disease is de ned by age < 3 years, subtotal resection (residual tumor ≥ 1.5 cm 2 ), or metastasis at diagnosis [5,7,[11][12][13][14].
Maximal safe surgical resection followed by risk adapted CSI and systemic chemotherapy have produced the best survival bene t for pediatric patients aged between 3 and 18 years [12,13,15,16]. However, little success is achieved in infant and younger children under 3 years of age. Concerns of neurocognitive impairment limiting the application of CSI in those younger patients may account for the limited bene t.
In addition, it is not always available to received risk-adapted and "standard" therapies due to patient adherence and limited resources, especially in low and middle income countries and areas [3,17]. Thus, it is important to identify possible prognostic factors in pediatric patients under 3 years of age.
In present study, we used the Surveillance, Epidemiology, and End Results (SEER) database to identify prognostic factors in pediatric medulloblastoma patients. We rstly investigated the frequency of received treatments including surgery, radiation and chemotherapy in medulloblastoma patients with different ages and found that the rate of radiation in patients with age of 0, 1 and 2 years is signi cantly lower than their counterparts with other ages. Then, we classi ed pediatric medulloblastoma patients into < 3 years group and ≥ 3 years group, and compared the difference of basic characteristics and outcomes between these two groups. Lastly, we selected multiple parameters to identify prognostic factors by Cox regression in < 3 years group and ≥ 3 years group, respectively.

Patient cohort
We used the SEER database to identify 1,495 pediatric patients diagnosed between 2000 and 2016 with medulloblastoma, which were speci ed by four speci c ICD-O-3 codes: 9470/3-medulloblastoma, NOS, 9471/3-desmoplastic nodular medulloblastoma, 9472/3-medullomyoblastoma and 9474/3-large cell medulloblastomas, as previously reported. The age of pediatric patients de ned as younger than 18 years old, and patients age of < 1 year was de ned as 0 year old. This study was conducted in accordance with the policies of the Scienti c Ethics Committee of SEER program and Sun Yat-sen University.

Study design
As determined by SEER, race was classi ed as white, African American and others. The extent of surgical resection was de ned as: no surgery, biopsy, partial resection (PR) and gross total resection (GTR), which were created to surgical procedure codes as previously described [18]. PR and GTR were de ned as surgery performed. Patients received any type of radiation including beam radiation, radioactive implants, raioisotopes, or combinational radiotherapy were classi ed as radiation used. Patients were excluded if clinical information including race, surgery, tumor resection rate, radiation, histology, chemotherapy and survival months was incomplete or unknown, leaving 939 patients in the nal cohort. Overall survival (OS) was de ned as the time from the diagnosis until death from any cause, and progress free survival (PFS) was de ned as the time from diagnosis until death caused by the original brain cancer.

Statistical Analysis
Baseline patient characteristics were presented as percentages and compared with Chi-Squared test. Multivariable analyses with the Cox proportional-hazards model were used to assess the impact of multiple covariates on patients' OS and PFS. And the selected covariates were age, sex, surgical resection, radiation and chemotherapy. To further clarify the role of radiation and chemotherapy in pediatric medulloblastoma, adjuvant treatment strategy was rede ned as none, chemotherapy, radiation, and radiation plus chemotherapy, and then was put into Cox regression analysis. Statistical analysis was carried out using SPSS 22.0 (Chicago, IL, USA). A p value < 0.05 was considered statistically signi cant.

Result
A total of 939 patients were included after exclusion of 556 patients. As shown in Fig. 1, most of the patients aged between 0 to 10 years old. Of these cohort of patients, the number of patients with medulloblastoma, NOS (9470/3), desmoplastic nodular medulloblastoma (9471/3), medullomyoblastoma (9472/3) and large cell medulloblastomas (9474/3) were 784, 101, 3 and 51, respectively ( Fig. 1 and Table 1). Due to the concerns of adverse effects, each treatment was differentially applied in every age strata. Only 13.2%, 27.8% and 35.2% of patients in age of 0, 1 and 2 years received radiation, which were signi cantly lower than that in other age strata, ranging from 74.8-100% (Fig. 2). However, patients in each age strata received comparable chemotherapy and surgical resection (Fig. 2). Hence, we classi ed the cohort of patients into < 3 years group and ≥ 3 years group and compared the difference of basic characteristics. Consistent to the previous report that the incidence of medulloblastoma is higher in male, and even higher in ≥ 3 years group (55.1% vs 64.5%, p = 0.016) ( vs 34.5% (p = 0.127) between those two groups, respectively (Table 1).
To identify prognostic risk factors, parameters including age, sex, histology, surgical resection, radiation and chemotherapy were selected for Cox regression analysis for both groups. We found that patients age, chemotherapy independently bene t patient OS (p < 0.001) and PFS(p = 0.008), however, radiation and extent of resection did not in < 3 years group (Table 2). For those patients ≥ 3 years patients, the more frequently applied radiation signi cantly prolonged patient OS and PFS (all p < 0.001) ( Table 3). Unlike < 3 years group, surgery performed could signi cantly prolong patient OS (p < 0.05) ( Table 3). However, chemotherapy did not bene t patient outcome (p > 0.05) ( Table 3). In both groups, desmoplastic nodular medulloblastoma negatively correlated to patient outcome compared to medulloblastoma, NOS (p < 0.05) ( Tables 2 and 3).  Since less concerns of adverse effect of adjuvant radiation and chemotherapy in patients over 3 years old, we further clari ed the survival bene t of radiation and chemotherapy in those patients who underwent biopsy, PR or GTR. Parameters including age, sex, histology and adjuvant therapy were selected for Cox regression. Radiation alone did improve patient outcome in those received PR or GTR, but not in those received biopsy (Tables 4 and 5). Surprisingly, chemotherapy alone could not bene t patient OS (p > 0.05) and PFS (p > 0.05) neither in those underwent biopsy, PR or GTR (Tables 4 and 5). And radiation could only signi cantly prolong patient OS (p = 0.011; OR 0.230; 95% CI, 0.074 to 0.716) and PFS (p = 0.017; OR 0.206, 95% CI, 0.056 to 0.754) in those received GTR. Noteworthy, radiation plus chemotherapy signi cantly improved patient outcome even more (Tables 4 and 5).

Discussion
In present study, we rstly found that there was more desmoplastic nodular medulloblastoma cases (20.9%) in patients < 3 years ( Fig. 1 and Table 1), however, which is lower than 44% in the previous reports [14]. Secondly, radiation was less frequently applied in patients < 3 years, which is consistent to the clinical practice of reduced or delayed application of radiation in those patients due to concerns of neurocognitive impairment [19,20]. Thirdly, we identi ed that patients < 3 years was really in high risk compared to patients ≥ 3 years with lower 6-month, 1-year, 3-year and 5-year survival rate. Differential epidemiological characteristics, such as age and histology, and application of radiation as shown in Table 1 might account for survival difference between patients < 3 years old and ≥ 3 years old. Indeed, before use of CSI, medulloblastoma remains incurable in older children even with GTR [4,21].
In the literature, the bene t of GTR remains controversial. Albright et al [6] and Zeltzer et al [7] reported survival bene t of GTR over PR or biopsy in patients > 3 years old, however, no signi cance was found in Akyuz's study with patients age of 0 to 18 years old [9] and in Evans's study with patients age of 2 to 16 years old [10]. The above studies implied that age might affect signi cance of GTR in survival. In a large retrospective study, Thompson et al found signi cant interaction between age and extent of resection, however, they did report overall and progress-free survival bene t of GTR compared to near total resection and subtotal resection in patients < 3 years old [5]. Notably, the prognostic bene t of increased extent of resection was attenuated when molecular subgroup a liation is taken into account [5]. Similarly, the present study demonstrated that GTR did not confer OS and PFS bene t in patients < 3 years old (Table 2). However, GTR showed OS bene t but no PFS bene t in ≥ 3 years old group (Table 3).
It is well established that radiation alone could largely improve patient perspective and cure the disease [4,21]. Early studies elicited that chemotherapy alone could prolong tumor free survival in younger children less than 18 months of age [22]. For those patients aged < 5 years old, Grill et al proved that conventional chemotherapy alone was su cient to cure patients who had GTR, but not those with incompletely resection or metastatic medulloblastoma [20]. The present study identi ed that chemotherapy but not radiation signi cantly improved OS and PFS in younger patients < 3 years old (Table 2), however, the OS and PFS bene t of chemotherapy and radiation was inverted in those patients ≥ 3 years old (Table 3). Different sensitivity of histological subtypes to chemotherapy might account for the difference between patients < 3 years old and ≥ 3 years old [23,24], since there are more desmoplastic nodular medulloblastoma cases in < 3 years old patients. Furthermore, in those patients ≥ 3 years old and underwent surgical procedures, the bene t of radiation alone largely depends on the extent of resection (Tables 4 and 5). It is worth noting that adjuvant radiation together with chemotherapy largely improves patient outcome even in those underwent partial resection (Tables 4 and 5).
With the introduction of molecular diagnosis by the 2016 WHO classi cation of central nervous system tumor [25], different molecular subtype and origin of medulloblastoma determine the sensitivity to chemo-, radio-or targeting therapy [4,5,26,27]. Hence, pathological diagnosis from surgical procedures is highly recommended in the treatment of pediatric medulloblastoma. Due to apparent reasons, the epidemiology of molecular subtype and its impact on the outcome of pediatric medulloblastoma is out of the scope of the study.
Due to limited cases, we were unable to further investigate the survival bene t of radiation and chemotherapy in medulloblastoma patients less than 3 years old who received biopsy, PR or GTR. Also, the study is subject to inherent limitations of retrospective study. The treatment patterns might vary from each patient and doctor, and the "standard" treatments including radio-and chemo-therapy differ from each year or each version of guideline. Moreover, the further details of radiation or chemotherapy are largely unknown and might impact the results of present studies and cause bias.
Notwithstanding these limitations, data from present study demonstrates that the extent of surgical resection confer overall survival bene t in those ≥ 3 years old patients but not in those < 3 years old. Chemotherapy shows OS and PFS bene t in those ≥ 3 years old patients but not in those < 3 years old, however, radiation shows inverted effect between those two populations. Moreover, adjuvant radiation plus chemotherapy largely is the most effective adjuvant treatment for those underwent tumor resection.
In conclusion, the extent of resection should be differentially considered and applied between pediatric medulloblastoma patients < 3 years old and ≥ 3 years old, so are the adjuvant treatments.

Declarations
The authors declare that the work described is original work and is not under review by any other journal. The data access is approved by the SEER program and the usage is in accordance to the policies of SEER program.

Funding
Not applicable.

Con icts of interest
There is no con ict of interest exits in the submission of this manuscript.

Availability of data and material
In the present study, all data were selected from the Surveillance, Epidemiology, and End Results (SEER) database. Researchers can request and get access to the data at https://seer.cancer.gov.

Ethics approval
The study is conducted in accordance with the policies of the Scienti c Ethics Committee of Sun Yat-sen University.
Consent to participate Line chart shows the percent of medulloblastoma patients received surgery, chemotherapy and radiation in every age strata.