Adolescent and young adult glioma: systematic review of demographic, disease, and treatment influences on survival

Abstract Background Prognostic factors in adolescent and young adult (AYA) glioma are not well understood. Though clinical and molecular differences between pediatric and adult glioma have been characterized, their application to AYA populations is less clear. There is a major need to develop more robust evidence-based practices for managing AYA glioma patients. Methods A systematic review using PRISMA methodology was conducted using multiple databases with the objective of identifying demographic, clinical, molecular and treatment factors influencing AYA glioma outcomes. Results 40 Studies met inclusion criteria. Overall survival was highly variable across studies depending on glioma grade, anatomic compartment and cohort characteristics. Thirty-five studies suffered from high risk of bias in at least one domain. Several studies included older adults within their cohorts; few captured purely AYA groups. Despite study heterogeneity, identified favorable prognosticators included younger age, higher functional status at diagnosis, low-grade pathology, oligodendroglioma histology and increased extent of surgical resection. Though isocitrate dehydrogenase (IDH) mutant status was associated with favorable prognosis, validity of this finding within AYA was compromised though may studies including older adults. The prognostic influence of chemotherapy and radiotherapy on overall survival varied across studies with conflicting evidence. Conclusion Existing literature is heterogenous, at high risk of bias, and rarely focused solely on AYA patients. Many included studies did not reflect updated pathological and molecular AYA glioma classification. The optimal role of chemotherapy, radiotherapy, and targeted agents cannot be determined from existing literature and should be the focus of future studies.

Gliomas represent a diverse histologic group of central nervous system tumors (CNS) with substantial molecular heterogeneity. Taken together, gliomas represent 29-35% of central nervous system tumors within the adolescent and young adult (AYA) demographic, of which two-thirds have been categorized as low-grade or World Health Organization (WHO) grade 1 or 2 with the remainder either WHO grade 3 or 4. 1 Grade alone inadequately captures the biologic and molecular complexity of these cancers, particularly among low-grade gliomas (LGG).
Studies have demonstrated distinct clinical trajectories and underlying molecular influences in pediatric vs. adult LGG. While childhood LGG have limited propensity to undergo malignant transformation, transformation occurs in the substantial majority of adult cases. 2,3 These differing characteristics also result in important differences in treatment philosophy for children compared to adults. 3 For example, adjuvant chemoradiation has shown benefits in progression-free survival (PFS) and overall survival (OS) among LGG that occurs in patients >40 years and those <40 with subtotal resection (STR). 4 By contrast, recent combined molecular and clinical analyses have identified pediatric LGG risk-stratified subgroups that differ in the potential benefit of adjuvant therapy. 5 Furthermore, in pediatric LGG, radiation therapy has been shown to act as an independent adverse prognostic factor for OS. 6 There is less observed heterogeneity in the clinical trajectory and treatment of high-grade glioma (HGG) between pediatric and adult populations. 7,8 AYA, commonly defined as patients between 15 and 39 years of age, are a vulnerable subpopulation at the crossroads between pediatric and adult cohorts. 9,10 National brain tumor registry data from the United States suggest that AYA glioma survival is more favorable than older adults (in whom HGG is more common), though survival rates are lower when compared to pediatric patients. 11 However, AYA-specific prognostic and treatment data are rare due to overlapping inclusion in pediatric or adult cohorts combined with limited representation in clinical trials. Though it is now well accepted that glioma outcome varies by molecular alteration in both pediatric and adult cohorts, the molecular landscape of AYA glioma has not been well described, leading to a homogenous approach regardless of cancer genetics. This lack of AYA focus has consequences: mortality rates for AYA with CNS tumors have increased by 0.6% per year for males and 1% per year for females. 12 Current literature is limited in defining the ideal treatment approach for this group. Thus, AYA patients treated in pediatric centers are most often treated according to pediatric guidelines, while those treated in adult centers are often treated with adult approaches.
Given the histological and molecular heterogeneity of glioma across the age spectrum, a rigorous evaluation of the available AYA glioma literature is required to inform patient counseling, therapeutic decisions, and future research priorities. Our objective was thus to review factors associated with survival outcomes in AYA glioma.

Methods
Ethics approval was not required for this systematic review.

Data Sources and Search Strategy
The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. 13 Multiple databases including OVID MEDLINE, EMBASE and EBM Reviews-Cochrane library databases from inception to July 2020 were queried in collaboration with an academic librarian at the Hospital for Sick Children. A sample search strategy can be found in supplemental materials (Supplementary Table 1). Bibliographies of relevant reviews were further queried to ensure all relevant studies were captured.
Screening and search strategy.-Study inclusion criteria included: (1) original studies that reported predictors of cancerrelated outcomes [eg, PFS, time to malignant progression (TtMP), OS]; (2) mean or median age at diagnosis within the AYA age range (15-39.

Importance of Study
Glioma is a major contributor to oncologic morbidity and mortality in the adolescent and young adult (AYA) demographic. Historically, AYA have been poorly represented in glioma research due to limited enrollment and representation in both pediatric-and adult-focused cohorts. This systematic review synthesizes available prognostic, treatment and survival data for AYA glioma patients. We demonstrate the favorable impact of younger age and higher Karnofsky Performance Status (KPS) on overall survival (OS) and event-free survival (EFS). This review identified a positive association with OS and EFS with lowgrade histology, oligodendroglial histology, isocitrate dehydrogenase (IDH) mutant molecular status and extent of surgical resection, though many included studies exhibited high bias risk and included older adults. It also highlights limited consensus on the role of adjuvant chemotherapy and radiotherapy in this population.

Results
The search strategy yielded 12 294 studies; removal of duplicates resulted in 10 336 unique studies. After abstract screening, 261 studies were identified as possibly meeting inclusion criteria and their full texts reviewed. Following full text review, 40 studies met inclusion criteria. Supplementary Figure 1 depicts the PRISMA workflow identifying included studies and reasons for exclusion. The kappa measure of agreement between reviewers for final study inclusion was 94.6% (95% CI 89.5-99.8%), or excellent.

Study Characteristics
Forty studies met criteria for inclusion in the review: 39 studies were retrospective (single center, multi-center or national database studies) and 1 study was prospective. Countries of origin included: United States (n = 19), Germany (n = 8), France (n = 4), Italy (n = 2), Japan (n = 2), Poland (n = 1), Austria (n = 1), United Kingdom (n = 1), Norway (n = 1) and Korea (n = 1). There was substantial variability in sample size among studies, ranging from 25 to 3057 patients. Together, the studies represented 12 405 patients with an age range from 3 months to 86 years. Though greater than 50% of each study cohort was required to be AYA based on inclusion criteria, older adults and children were included in many studies as illustrated in Figure 1. There were three studies that specifically included spinal cord gliomas, 1 study that included both spinal cord and intracranial glioma and the remainder included intracranial glioma. Three studies did not provide OS for the overall cohort, while another 10 did not provide EFS. All studies included OS-based univariate or multivariable analyses.

Overall Survival and Event-Free Survival
Glioma outcomes are summarized in Table 1. Two studies reported only on intracranial grade 1 glioma in which one showed an OS of 80% at 5 years and the other showed a reduced survival in the cohort undergoing external beam radiation therapy (EBRT) (< 60% 5 year OS) compared to those not undergoing adjuvant EBRT (> 75% 5 year OS). 20,21 Two studies included combined cohorts of both grade 1 and 2 glioma in which OS ranged from 75.7 to 91.0% at 5 years. 22,23 Twenty-six studies included grade 2 glioma only and reported 5-year OS ranging from 84 to 98%, with one study reporting 5-year OS of 69.2% in a subset of patients with radiographic velocity of diametric expansion over 8 mm/ year.  Among studies of grade 2 glioma, 5-year EFS ranged from 30 to 94%. Several studies included glioma subgroups across multiple pathological grades. 2 studies grouped grade 2 and 3 pleomorphic xantho-astrocytoma (PXA) with combined OS 76.3-89.5% at 5-years, 3 studies grouped grade 2 and 3 glioma together, 2 studies included grade 3 and 4 glioma, and 3 studies reported varying grades of spinal cord glioma, with 5-year OS ranging from 85.4% in grade 1 cases to 36.4% in grades 2, 3 and 4 [50][51][52][53][54][55][56][57][58][59] (Table 1).

Patient Factors
Several patient factors were associated with superior OS and EFS across glioma grade following adjusted multivariable analysis (Tables 2-4). Increased age was often associated with worse OS when age was evaluated as a continuous variable, 20,23,34,38,58,59 including cohorts of pilocytic astrocytoma alone, combined grade 1 and 2 gliomas, combined grade 2 and 3 gliomas, and of peri-ventricular HGG. Within the AYA group, the following younger age clusters were associated with improved OS: age <18 years, 51 age <30 years, 53 and age <40 years. 22,42 Only one study showed a negative impact of age younger than 40 on OS. 46 Several studies in contrast did not find a significant association between age and OS in multivariable analysis. [30][31][32][33]36,43,46,52,54 Three studies demonstrated that younger age was associated with improved EFS. 45,47,53 The relationship between sex and OS and EFS was conflicting with no clear prognostic effect. 22,29,32,38,46,54 Three studies showed no effect of patient sex on OS. 30,33,39,53 Other patient-related factors associated with favorable OS included private health insurance in a United States cohort, 30 median annual income greater than $38 000, 20 Charles-Deyo Comorbidity Index score of 0 vs. 2, 20 and Karnofsky Performance Score (KPS) greater than 80. 32,33,42 KPS over 80 was associated with favorable EFS in 1 study following multivariable analysis, 42 and though KPS was significantly associated with EFS in univariate analysis in three additional studies, it lost significance when adjusted for other factors. 32,33,47 Disease and Treatment-Related Factors Grade 1 glioma.-Several disease and treatment-related factors were significantly associated with OS and EFS among patients with grade 1 glioma or studies combining grade 1 and 2 gliomas (Table 2). Pre-operative lesion size over 19 mm 20 and grade 2 compared to grade 1 histology 22,23 were associated with inferior OS, while location of tumor in the supratentorial compartment was favorable compared to spinal cord or infratentorial locations following univariate analysis, though non-significant after multivariable analysis (though brainstem lesion inclusion in the infratentorial category may have biased this finding). 20 Symptom duration in spinal cord glioma was not significantly associated with OS after multivariable analysis. 52 Treatment-related factors positively influencing OS included gross-total resection (GTR) in spinal cord glioma cases. 51 Three studies found adjuvant radiation to be associated with inferior OS even after adjustment for other factors. 20,23 The first study by Lee et al. examined a national cohort of patients with pilocytic astrocytoma and adjusted for age, median income, tumor volume and comorbidity scores. They found adjuvant external beam radiotherapy (EBRT) was associated with a significantly worsened OS compared to no radiotherapy (patients undergoing EBRT 5-year OS < 60% compared to ≥ 75% 5-year OS in patients receiving other therapies). 20 The same study showed a trend towards inferior OS, though non-significant, when stereotactic radiotherapy was compared to no radiotherapy. 20 The authors nonetheless attributed their finding to confounding by other important factors including eloquent location and tumor resectability. The second study examined the effect of pregnancy on LGG survival. 23 They showed that post-operative radiation therapy was associated with significantly inferior OS in combined grade 1 and 2 gliomas as well as grade 2 gliomas alone following multivariable adjustment, though the authors did not provide a list of what variables were adjusted for. The third study, examining     Bolded fields indicate statistical significance of the variable in cited study at an alpha of 0.05.    Table 3. Imaging-related factors negatively associated with OS following multivariable analysis included: eloquent location, 25 tumor volume over 100 cm 3,29,44 larger tumor size as a continuous variable, 38 velocity of diametric expansion over 8 mm/year, 29,44 size greater than 5 cm 35,49 and size greater than 6 cm. 30 Factors initially significantly associated with OS in univariate analyses but which lost association in multivariable analyses included contrast enhancement on MRI 29,40 and corpus callosum involvement. 29 There was significant negative influence of eloquent location, 25 Table 5. Among patients with astrocytomas, grade 2 histology conferred significantly worse OS than grade 1 histology. 22 Diffuse astrocytoma histology was associated with inferior OS compared to oligoastrocytoma or oligodendroglioma histology following multivariable analysis. [34][35][36]41,42,49 Oligodendroglioma was variably defined either histologically or molecularly across articles. Oligodendroglioma showed significantly favorable OS compared to IDH mutant and IDH wildtype astrocytoma. 38,46 IDH mutant status 29,33,37,42 and 1p19q co-deletion 32,33 were positively associated with longer EFS. In one cohort of diffuse supratentorial low-grade gliomas, 1p19q co-deletion status was non-significant after adjusted multivariable analysis. 29 In multivariable analysis, EFS was significantly inferior among those with diffuse astrocytoma histology, 34,49 adjusted for IDH mutational status. 46 IDH mutant status, 29 1p19q co-deletion 32,33 and O6-methylguanine-DNA methyl-transferase (MGMT) methylation 33 were favorably associated with prolonged EFS when compared to IDH wild type gliomas. Diffuse astrocytic histology 43,47 and p53 over-expression 47 were significantly negatively associated with EFS in univariate analysis but after adjustment in multivariable analysis were no longer significant. Notably, the studies that described IDH mutational status and influence on prognosis all comprised of cohorts that despite meeting our inclusion criteria, included substantial numbers of older adults (Figure 1). For example, of the 26 studies that included AYA patients with grade 2 glioma, 24 had a mean or median age above 30.

Neuro-Oncology Advances
Treatment-related variables are summarized in Table 6. The impact of adjuvant chemoradiotherapy on OS and EFS was mixed. Combined adjuvant chemotherapy and radiotherapy positively impacted OS and EFS among grade 2 glioma patients in one study compared to adjuvant radiotherapy alone following multivariable analysis. 42 Within this study the effect of adjuvant chemoradiotherapy was most pronounced in cases of IDH 1/2 mutant cases. By contrast Pal'a et al 43  Bolded fields indicate statistical significance of the variable in cited study at an alpha of 0.05 Table 3. Continued    Bolded fields indicate statistical significance of the variable in cited study at an alpha of 0.05. Table 4. Continued  LGG that combined chemoradiotherapy (temozolomide) was superior in EFS compared to chemotherapy alone in a multivariable model with covariates gender, tumor size, molecular characteristics and adjuvant therapy regimen. 46 Several studies did not specify the adjuvant therapy regimen used, though showed chemoradiotherapy was associated with an unfavorable effect on OS following multivariable analysis. 28,37 Gousias et al 28 showed a negative association between adjuvant therapy and OS, but did not conduct multivariable analyses for this outcome; only 5% of their cohort underwent either chemotherapy and or radiotherapy. In their multivariable analyses conducted for EFS however, including eloquent location as a covariate, adjuvant therapy had a favorable impact on EFS.

Neuro-Oncology Advances
Conflicting results related to the role of adjuvant chemotherapy were observed; one group showed a positive association with both adjuvant chemotherapy and radiotherapy with increased EFS in multivariable analysis that included covariates age, histology, presenting symptoms, size and extent of resection. 49 Another study showed increased EFS but no significant change in OS with adjuvant chemotherapy following LGG resection after multivariable analysis with covariates age, tumor diameter, pathology and adjuvant therapy. 41 Few studies analyzed the role of adjuvant radiotherapy alone upon OS, though one included study demonstrated a significant negative impact on OS after multivariable analysis including age at diagnosis, molecular class, eloquent location, and post-operative residual volume. 38 Adjuvant radiotherapy significantly improved EFS in two studies, 38,49 and the effect was suggested to be greater with immediate as opposed to delayed radiotherapy following univariate analysis alone in two other reports. 32,33 Non-significant prognostic variables are shown in Supplementary Table 1. Following multivariable analysis, several studies found a non-significant association between OS for LGG and adjuvant chemother apy, 23,28,38,41,49,52,58 adjuvant radiotherapy 22,39,41,49,52 and combined adjuvant chemoradiotherapy. 54 Several studies looked at the impact of surgeryrelated factors. Increased extent of resection compared to biopsy alone was associated with both OS and EFS in multivariable adjusted models. 29,30,46 Extent of resection measured as either a continuous variable 27,34,45 or lower magnitude of post-operative volumetric tumor residual 34,38 correlated with prolonged OS and/or EFS. Several studies showed in adjusted multivariable analysis that GTR resulted in superior OS or EFS benefit compared to other resection categories, 26,32,36,49,53 though one study showed negative effect on EFS in IDH mutant astrocytoma. 43 One study found that first line surgical therapy compared to observation did not significantly influence OS though it favorably impacted EFS. 29 Factors associated with positive impact on OS following univariate analysis (in absence of *Inverse hazard ratios were reported to compile into common categories. Bolded fields indicate statistical significance of the variable in cited study at an alpha of 0.05. Table 5. Continued   Table 6. Continued   Table 6. Continued adjusted multivariable analysis) included: decreasing postoperative T2-weighted MRI signal volume, 34 greater extent of resection across histological types, 26,28,33,34,37,39,41,44,48 and smaller post-operative tumor volume. 38,39,45 Grade 3 and 4 glioma.-Groupings of Grade 3 and 4 glioma in included studies may not have reflected current classification schemes that include IDH mutational status. In addition, Grade 3 glioma may or may not be included in the definition of high-grade glioma. However, grouping Grade 3 and 4 glioma best reflected the categorization used by the papers identified in this systematic review. Table 4 summarizes disease and treatment-related factors influencing EFS and OS in HGG. Among high-grade spinal cord glioma, there was no significant influence on localized vs. regional or invasive location on OS. 56 Oligodendroglioma histology showed superior influence on OS compared to astrocytic histology in pooled grade 2 and 3 cases following univariate analysis (no multivariable analysis reported). 57 Grade 3 and 4 spinal cord glioma were negative influences on OS when compared to grade 2 histology. 52 1p19q co-deletion, IDH mutant status, low nestin level, and mitotic index less than 4% all positively impacted OS in combined grade 2 and 3 glioma cases. 54,57,58 No EFS analysis was conducted using these variables.

Neuro-Oncology Advances
Some studies included in this review showed adjuvant radiotherapy demonstrated favorable impact on OS in pooled grade 2 and 3 glioma, 58 pooled grade 3 and 4 spinal cord glioma, 56 and peri-ventricular HGG. 59 STR or biopsy-only resulted in worse OS than GTR or near-total resection (NTR) in two studies. 40,54 Though in peri-ventricular HGG STR and GTR were favorably associated with OS in univariate analysis compared to no surgery, they lost significance following adjusted multivariable analysis. Adjuvant chemoradiation positively impacted EFS in grade 2 and 3 glioma, though chemotherapy alone was not significant. 57 Grade 2 and 3 adjuvant radiotherapy also favorably influenced EFS. 57 One combined cohort of grade 2 and 3 glioma showed a non-significant influence of adjuvant chemoradiotherapy on OS following multivariable analysis. 54 Excluding spinal pilocytic astrocytoma, Fakrehddine et al 52 showed adjuvant chemotherapy significantly improved EFS in infiltrative spinal cord glioma (grades 2, 3 and 4) after adjusting for treatment modality, age at diagnosis, grade, number of spinal levels, neurological deficits and symptom duration. In the same analysis, adjuvant radiotherapy did not significantly impact EFS nor did either chemotherapy or radiation contribute to OS benefit after multivariable analysis. 52

Quality Assessment
Given the absence of methodological limitation reporting across studies, the QUIPS assessment tool was utilized the provide a standardized risk of bias assessment (Supplementary Table 2). Most studies (35/40)

Discussion
This systematic review identified 40 studies that reported on demographic, disease and treatment predictors of EFS and OS among AYA glioma patients in high income countries. Despite stringent definitions utilized to capture an adequately sized AYA cohort, several included studies captured a proportion of older adults (Figure 1). This points to a severe limitation in the existing AYA glioma literature, with all interpretation limited by the potential impact of older adult glioma biology in these cohorts. In contrast, only two studies included pediatric patients. 52,53 Furthermore, many papers scored in the high-risk bias category in at least one domain. Despite this, several patient epidemiological, disease and treatment factors with prognostic impact on EFS and OS were identified. chemotherapy and radiation therapy is usually considered. 68 A major challenge is the lack of studies in this review including details about the presence of pediatric-type alterations in AYA glioma, [69][70][71] thus limiting any meaningful molecularly informed conclusions about adjuvant chemoradiotherapy. Whether there is a role for adjuvant therapy among AYA with LGG either totally resected or with residual disease is a crucial question that should be prioritized.
Though HGG in pediatric and adult patients may share similarities in overall prognosis, there are important differences that exist between treatment regimens and biological considerations. At a molecular level, the profile of HGG is different with distinct copy number aberrations and driver mutations in pediatric HGG compared to adults. 72,73 Furthermore, cancer predisposition syndromes are more common in pediatric populations compared to adults. The extent to which these pediatric-type alterations and predispositions exist in AYA demographics is not well known and was not clarified through this review, thus highlighting a major gap in understanding. Stupp et al showed that adults with HGG had improved OS with adjuvant temozolomide in combination with fractionated radiotherapy compared to radiotherapy alone. 74 Radiotherapy typically begins 3-5 weeks following surgical resection and is typically administered at 50-60 Gy in 1.8-2 Gy fractions with limited evidence suggesting any added benefit at higher doses. 75,76 For patients with MGMT methylated promoter glioblastoma, recurrent or progressive HGG, second line alkylating chemotherapeutics may be considered. 76,77 By contrast, the benefit of adjuvant temozolomide in the treatment of pediatric HGG is debatable. This is highlighted by contrasting two prospective trials. Cohen et al. showed temozolomide administration during and after adjuvant radiotherapy in pediatric HGG did not improve outcomes. 78 In contrast, Jakacki et al 79 demonstrated that children with maximally resected nonmetastatic HGG treated with radiotherapy and concomitant temozolomide followed by lomustine and temozolomide adjuvant chemotherapy experienced significantly improved outcomes. Despite the complexity in decision making surrounding HGG adjuvant therapy, our review highlights that AYA-specific data to guide clinicians is lacking.
Limitations stem from the predominance of retrospective studies included in this systematic review as well as the inclusion of older adults in many study cohorts. Despite intentions to identify and assess prognostic factors in AYA glioma, the inclusion of older adults skews the results and limits generalizability. However, stricter age-based inclusion criteria would have resulted in the exclusion of nearly all studies. Pediatric glioma mutational markers were rarely examined, precluding assessment of their prognostic value in AYA populations. Our review included all CNS gliomas, including spinal gliomas, though the latter may require different treatment approaches owing to differing biology anatomical considerations. Finally, the majority of studies were classified as at high risk of bias in at least one domain.

Conclusion
Although this study reveals some traditional factors that appear prognostically important in AYA glioma, most, including tumor grade, pathological subtype and genetic mutations such as IDH1/2, need to be considered with care given bias from the inclusion of older adults in many studies. Interestingly, the role of cytoreductive surgery remains an important prognostic factor in AYA gliomas and may not change until effective adjuvant medical therapies emerge. As such, the current literature does not provide clinicians with an evidence-based approach to treating AYA with gliomas, particularly regarding the role of adjuvant chemotherapy and radiotherapy. Available evidence is heterogenous, of mixed quality, at high risk for confounding, and predominantly derived from older adult cohorts. Prospective studies of histopathological and molecularlydefined gliomas exposed to uniform treatment including both short-and long-term outcomes will allow the identification of optimal AYA-specific glioma management strategies.

Supplementary material
Supplementary material is available online at Neuro-Oncology Advances online.

Funding
No financial disclosures.