Wide Range of Clinical Outcomes in Patients with Gliomatosis Cerebri Growth Pattern: A Clinical, Radiographic, and Histopathologic Study

Gliomatosis cerebri growth pattern (GC) is a rare presentation of a primary widely infiltrating glial tumor of the central nervous system. In this retrospective case series, the clinical, radiographic, and histopathologic findings of 34 patients with histologically proven primary, type I GC are evaluated.

Abstract

Background.

The 2016 World Health Organization Classification of Central Nervous System Tumors categorizes gliomatosis cerebri growth pattern (GC) as a subgroup of diffuse infiltrating gliomas, defined by extent of brain involvement on magnetic resonance imaging (MRI). Clinical and radiographic features in GC patients are highly heterogeneous; however, prognosis has historically been considered poor.

Subjects, Materials, and Methods.

We performed a retrospective search for patients at our institution meeting radiographic criteria of primary, type I GC (defined as diffuse tumor infiltration without associated tumor mass and contrast enhancement on MRI) and analyzed their clinical, imaging, and histopathologic features.

Results.

A total of 34 patients met radiographic criteria of primary, type I GC, and 33 had a confirmed histologic diagnosis of an infiltrating glial neoplasm. Age >47 years at diagnosis was associated with worse overall survival (OS) compared with age ≤47 years (hazard ratio [HR] 1.04, 95% confidence interval [CI] 1.01–1.07, p = .003). Patients with grade 2 tumors demonstrated a trend for improved OS compared with those with grade 3 tumors (HR 2.65, 95% CI 0.99–7.08, p = .051). Except for brainstem involvement, extent or location of radiographic involvement did not detectably affect clinical outcome. IDH mutation status identified a subgroup of GC patients with particularly long survival up to 25 years and was associated with longer time to progression (HR 4.81, 95% CI 0.99–23.47, p = .052).

Conclusion.

Patients with primary, type I GC do not uniformly carry a poor prognosis, even in the presence of widespread radiographic involvement. Consistent with other reports, IDH mutation status may identify patients with improved clinical outcome. Molecular characterization, rather than MRI features, may be most valuable for prognostication and management of GC patients.

Implications for Practice.

Patients with gliomatosis cerebri growth pattern (GC) constitute a challenge to clinicians, given their wide range of clinical, histologic, and radiographic presentation, heterogeneous outcome patterns, and the lack of consensus on a standardized treatment approach. This study highlights that radiographic extent of disease—albeit category‐defining—does not detectably influence survival and that IDH mutations may impact clinical outcome. Practicing oncologists should be aware that select GC patients may demonstrate exceptionally favorable survival times and prognosticate patients based on molecular markers, rather than imaging features alone.

Introduction

Gliomatosis cerebri is a rare presentation of a primary widely infiltrating glial tumor of the central nervous system (CNS). Nevin [1] originally described three cases with diffuse neoplastic glial cell infiltration of the brain parenchyma with relative preservation of the surrounding architecture and little mass effect. The World Health Organization (WHO) formally defines gliomatosis cerebri as involvement of more than two cerebral lobes on magnetic resonance imaging (MRI) in the presence of no or minimal contrast enhancement and, prior to 2016, identified it as a distinct nosological entity [2]. However, in the revised 2016 WHO Classification of CNS Tumors, gliomatosis cerebri was renamed “gliomatosis cerebri growth pattern” (herein abbreviated as “GC”) and categorized as a subtype of diffuse gliomas [3].

Histologically, most patients demonstrate features of diffuse infiltrating astrocytomas [2], but cases of oligodendroglial and oligoastrocytic differentiation encompassing all WHO histologic grades have been reported [4], [5], [6], [7]. Jennings [8] proposed two forms of GC: (a) “primary GC,” a diffusely infiltrating glioma without (type I) or with (type II) associated tumor mass, and (b) “secondary GC,” in which a focal glioma infiltrates the surrounding brain parenchyma over time. It is uncertain whether different forms of GC reflect a temporal and biologic spectrum of disease derived from the same cell‐of‐origin, host responses to tumor cells, or variations in cellular microenvironment. Given this uncertainty, GC, by current criteria, should exclude presentations with enhancing nodules (i.e., “primary, type II” and “secondary” cases). However, many reports of GC patients in the literature do not distinguish between the different subtypes, therefore making it difficult to interpret clinical outcome data and stratify patients according to risk groups.

The diffuse nature of GC limits the effectiveness of surgery and shaped‐field radiotherapy (RT). Chemotherapy, such as procarbazine, lomustine, vincristine (PCV) and temozolomide (TMZ), is associated with variable response rates [7], [9]. Response to treatment is difficult to assess as imaging endpoints have not been clearly defined. The clinical course of GC patients is variable: Although prognosis is generally considered poor with a median overall survival (mOS) ranging from 9.5 to 16 months [4], [5], [6], rare patients with OS from 3 to 7 years have been described [5], [6]. Recent reports suggest that molecular markers found in gliomas, such as mutations in the isocitrate dehydrogenase genes (IDH1/2) [10], [11] and methylation of the O⁶‐methylguanine DNA methyltransferase (MGMT) promoter [12], are associated with improved OS in GC patients.

In this retrospective case series, we evaluated the clinical, radiographic, and histopathologic findings of 34 patients with primary, type I GC treated at the Massachusetts General Hospital (MGH) between 1999 and 2016. Our findings highlight the heterogeneity of GC in terms of histopathological, molecular, and imaging features and provide examples of long‐term survivors even in the presence of extensive radiographic disease involvement.

Subjects, Materials, and Methods

Patient Selection

Following MGH Institutional Review Board approval, we searched the neuroradiology and neuro‐oncology database for patients with GC and diffuse infiltrating gliomas between 1999 and 2016. We defined GC as (a) diffusely infiltrative tumor involving more than two cerebral lobes or regions on T2‐weighted or fluid‐attenuated inversion recovery (FLAIR) MR imaging in the absence of a solid tumor mass; and (b) lack of enhancement on T1‐weighted imaging after gadolinium‐based contrast administration. We excluded patients whose initial MRI demonstrated significant enhancement after contrast administration as well as “secondary GC” patients whose focal glioma infiltrated adjacent brain parenchyma over time.

Radiologic features were assessed on T2‐weighted FLAIR and T1‐weighted sequences performed before and after contrast administration. In addition to the anatomic division of the brain into frontal, parietal, temporal, and occipital lobes, we defined involvement of the pineal region, brainstem, cerebellum, and deep gray matter structures (e.g., basal ganglia and thalamus) as separate anatomic locations [11], [12]. Response to treatment was assessed by MRI before and at least 4 weeks after completion of therapy, and classified as complete response, partial response (PR), minor response (MR), stable disease, or progressive disease (PD) according to the Response Assessment in Neuro‐Oncology (RANO) criteria for low‐grade gliomas [13].

“Overall survival” (OS) was defined as the time between radiographic diagnosis and death or time of last follow‐up. “Time to progression” (TTP) was defined as the time between the start of first‐line treatment and the day of radiographic progression.

Molecular Testing for IDH Mutations and MGMT Promoter Methylation

To test for the most common IDH mutation (IDH1 variant R132H), immunohistochemistry (IHC) on formalin‐fixed paraffin‐embedded (FFPE) tissue was performed with the mIDH1R132H antibody [14]. Tumor genotyping was performed using the SNaPshot methodology [15], version 3, a multiplexed allele‐specific assay to detect somatic mutations in tumor DNA extracted from FFPE samples, which also detects noncanonical IDH1 mutations. Testing of MGMT promoter status was performed using methylation‐specific polymerase chain reaction based on a standardized clinically validated protocol at the MGH Department of Pathology.

Statistical Analysis

For each potential prognostic factor or covariate of interest, the Cox proportional model was used to estimate hazard ratios (HRs) and their 95% confidence intervals (CIs). Two separate Cox proportional models were fitted for OS and TTP. Group differences were compared using the log‐rank test. Because there were missing values across some variables (such as IDH status), a tree‐based classification model instead of a multivariate analysis was performed to screen for additional variables associated with OS/TTP. The tree‐based model first carries out a global null hypothesis of independence between covariates and OS/TTP and then splits the data into subgroups with respect to the variable that has the strongest association with OS/TTP. The model repeats this splitting algorithm until the global null hypothesis of independence cannot be further rejected. The covariates included in this model were age, tumor grade, number of involved brain regions, and tumor location. To detect a potential age cutoff that predicts OS, a maximally selected log‐rank procedure was applied. For each cutpoint, the log‐rank statistic was computed for the OS greater than the cutpoint versus OS equal to or lower than the cutpoint. In general, a large log‐rank statistic indicates that the distributions of the two groups are different. The maximum of these log‐rank statistics corresponds to the best cutpoint that separates the two distributions. The significance of the cutpoint was assessed with a permutation approach, with 10,000 permutation size. All statistical tests were two‐sided, and a p value <.05 was considered statistically significant. Analyses were performed with the use of the R software, version 3.3.1.

Results

Cohort Definition (Fig. 1)

Figure 1.

Search criteria.

Using the aforementioned search terms, 401 patients were identified. Of these, 367 did not meet inclusion criteria, resulting in 34 patients with primary, type I GC that were included in the analysis (6 from the neuro‐oncology and 28 from the neuroradiology database).

Patient Clinical Characteristics

There was a slight male predominance (18 males vs. 16 females). Median age at time of diagnosis was 57 years in males and 49 years in females (range 2–83). Presenting symptoms are summarized in Tables 1 and 2. Case 2 was asymptomatic, and abnormal T2/FLAIR hyperintensity was detected during routine MRI performed as part of a clinical study for a nonneurologic condition. Karnofsky Performance Score (KPS) was ≥90 in all patients except Case 9, 32 (both KPS = 50), and 27 (KPS = 30).

Table 1. Summary of patient characteristics at time of diagnosis.

^a Included personality and behavioral changes, memory difficulties, confusion, and disorientation. Other/nonspecific symptoms included lightheadedness, nausea, and syncope (not attributed to seizures).

^b IDH‐wt: IDH‐wild‐type; IDH‐mut: IDH‐mutant; NOS: molecular testing for IDH status and/or 1p19q co‐deletion could not be performed or was inconclusive. Cases of “diffuse astrocytoma,” “oligodendroglioma,” and “oligoastrocytomas” correspond histologically to WHO grade 2. Cases of “anaplastic astrocytoma” correspond histologically to WHO grade 3.

^c Histopathology read as “infiltrating glioma” without further specification of histologic subtype.

^d No specification of histologic grade.

Abbreviations: CHT, chemotherapy; CRT, chemoradiation; PCV, procarbazine, lomustine, vincristine; RT, radiotherapy; TMZ, temozolomide; WHO, World Health Organization.

Table 2. Individual patient characteristics.

^a At time of MRI diagnosis.

^b DA, IDH‐wt: diffuse astrocytoma, IDH‐wild‐type, WHO grade 2; DA, IDH‐mut: diffuse astrocytoma, IDH‐mutant, WHO grade 2; DA, NOS: diffuse astrocytoma, IDH status has not been fully assessed, WHO grade 2; AA, IDH‐wt: anaplastic astrocytoma, IDH‐wild‐type, WHO grade 3; AA, IDH‐mut: anaplastic astrocytoma, IDH‐mutant, WHO grade 3; AA, NOS: anaplastic astrocytoma, IDH status has not been fully assessed, WHO grade 3; OD, NOS: oligodendroglioma, molecular testing for IDH status and 1p19q co‐deletion could not be completed or was inconclusive, WHO grade 2; OA, NOS: oligoastrocytomas, molecular testing for IDH status and 1p19q co‐deletion could not be completed or was inconclusive, WHO grade 2.

^c Pathologic enhancement after gadolinium‐based contrast administration was absent in all cases at the time of diagnosis.

^d Deceased.

Abbreviations: adj, adjuvant; AMS, altered mental status; B/l, bilateral; BCNU, vincristine; Bx, biopsy; Carbo, carboplatin; CCNU, lomustine; CRT, chemoradiation (concurrent RT + TMZ); F, female; FLAIR, fluid‐attenuated inversion recovery; GTR, gross total resection; HA, headache; L, left; M, male; MRI, magnetic resonance imaging; N/A, not available; OS, overall survival; PCV, procarbazine, lomustine, vincristine; R, right; SRS, stereotactic radiosurgery; STR, subtotal resection; Sz, seizure; TAM, tamoxifen; TMZ, temozolomide; VA, visual acuity; VF, visual field; WBRT, whole‐brain radiotherapy.

Diagnosis of GC

Histopathology.

Thirty‐three patients (97%) had tissue available for pathologic evaluation. Analysis from 32 patients revealed a neoplasm of astrocytic (n = 26), oligodendroglial (n = 4), or mixed oligoastrocytic origin (n = 2). A left temporal biopsy in Case 27 was equivocal, showing infiltrating atypical cells and reactive astrocytes in the absence of high‐grade features. One individual (Case 12) did not undergo brain biopsy due to clinical and radiographic stability but was diagnosed with GC based on characteristic imaging features.

WHO grade 3 tumors were reported in approximately half the cases (n = 18; 53%) and were all classified as astrocytomas. Fourteen patients (41%) had a WHO grade 2 tumor. No WHO grade 4 tumors were identified. In one patient, tumor grade was not specified. In 11 patients, biopsies were specifically referred to as “diffuse” or “infiltrating,” reflecting areas of neoplastic cells infiltrating cortex and hypercellular white matter.

Repeat biopsy was performed in nine patients (Case 6, 10, 15, 16, 17, 20, 22, 28, and 33) after appearance of new contrast‐enhancing lesion(s) or worsening extent of T2/FLAIR hyperintensity. In five individuals (Case 6, 10, 17, 20, 33), progression from either a grade 2 or 3 tumor to a glioblastoma (GBM) occurred. One patient (Case 22) had a recurrent anaplastic astrocytoma after treatment of an original grade 3 tumor with TMZ. Another individual (Case 28) had recurrent infiltrating glioma and reactive gliosis after RT and TMZ. Biopsy revealed radiation effect without tumor progression in two patients (Case 15 and 16) following chemoradiation and adjuvant monthly TMZ.

IDH status results were available for 16 patients (47%). An IDH1 mutation was found in four individuals through both IHC and genetic sequencing (Case 4) or genetic sequencing alone (Case 10, 14, and 31). Of these, two patients (Case 4 and 31) had a grade 3 anaplastic astrocytoma and one each had a grade 2 astrocytoma (Case 10) and grade 2 oligodendroglioma (Case 14; Table 3). Table 3 summarizes additional genetic alterations found in IDH‐mutant patients. Twelve patients were classified as IDH1‐wild‐type (wt) based on IHC (n = 10) or genetic sequencing (n = 2). In 18 patients (53%), IDH status could not be obtained, either due to insufficient tissue or lack of tissue access for additional testing.

Table 3. Characteristics of patients with IDH mutations.

^a AA, IDH‐mut: anaplastic astrocytoma, IDH‐mutant, WHO grade 3; DA, IDH‐mut: diffuse astrocytoma, IDH‐mutant, WHO grade 2; OD, IDH‐mut, 1p19q co‐deleted: oligodendroglioma, IDH‐mutant, 1p19q co‐deleted, WHO grade 2.

Abbreviations: EGFR, epidermal growth factor receptor; IHC, immunohistochemistry; L, left; MGMT, O⁶‐methylguanine DNA methyltransferase; MRI, magnetic resonance imaging; R, right; TTP, time to progression; WHO, World Health Organization.

1p19q testing with fluorescent in‐situ hybridization (FISH) was performed in seven patients, including two of four patients with histologically suspected oligodendrogliomas (Case 3 and 14). Case 14 was found to be 1p19q co‐deleted and IDH‐mutant, and was classified as an oligodendroglioma. In Case 3, FISH failed due to technical difficulties and 1p19q status could not be determined, but the patient was classified as an oligodendroglioma based on the presence of typical cellular morphology, including rounded nuclei and perinuclear halos. In the other five patients (Case 10, 20, 21, 31, and 33), 1p19q was either maintained or partially deleted. These cases were all classified as astrocytomas. The integrated diagnoses encompassing histopathologic and molecular classification criteria based on the 2016 WHO Classification of CNS Tumors are summarized in Tables 1 and 2.

Testing for MGMT promoter methylation status was performed in four patients. Two patients had an unmethylated (Case 10 and 22), one a partially methylated (Case 33), and one a methylated (Case 17) MGMT promoter.

MRI Findings.

Both cerebral hemispheres were affected with similar frequency (n = 24; 71% for right hemisphere vs. n = 25; 74% for left hemisphere). Bi‐hemispheric T2/FLAIR hyperintensity was seen in 16 (47%) subjects. The median number of affected regions was 4 (range 3–9). Twenty patients (59%) had T2/FLAIR hyperintensity of >3 cerebral regions. Significant concomitant contrast enhancement (≥1 cm in diameter) on T1‐weighted imaging was absent in all patients at the time of diagnosis. Figure 2 shows representative cases.

Figure 2.

Magnetic resonance imaging (MRI) of gliomatosis cerebri growth pattern. (A1, A2): Case 27 presented with a 10‐day history of confusion, behavioral changes, visual difficulties, and left‐sided localizing signs. (A1): T2‐weighted imaging revealed T2/ fluid‐attenuated inversion recovery (FLAIR) abnormality of the bilateral parietal, occipital, temporal, and frontal lobes, with extension into the right internal, external, and extreme capsules. (A2): No concomitant enhancement was noted on T1‐weighted imaging. The patient succumbed 20 days after supportive treatment. (B1, B2): T2/FLAIR imaging in Case 23, performed after 2 months of memory loss and confusion, showed a hyperintense infiltrative lesion of the bilateral frontal and temporal lobes, thalami, basal ganglia, and brainstem (B1), which did not enhance (B2). Chemoradiation (CRT) and adjuvant temozolomide (TMZ) stabilized disease for 17 months. (C1, C2): Case 22 had episodes of headaches. MRI revealed a T2/FLAIR hyperintense infiltrating (C1) but nonenhancing (C2) lesion of the right cerebral hemisphere extending into the midbrain and across the posterior body and splenium of the corpus callosum. Biopsy revealed a grade 3 IDH‐wild‐type astrocytoma. The patient died 82 months after initial diagnosis. (D1, D2): Case 14 had a 4‐year history of seizures before diagnosis of an infiltrative nonenhancing T2/FLAIR hyperintense mass within the bilateral frontal lobes and left anterior temporal lobe was made. Biopsy revealed an IDH‐mutant grade 2 oligodendroglioma. She has remained stable for >10 years after subtotal resection, 24 cycles of TMZ, and radiotherapy. (E1, E2): Case 12 presented with seizures and personality changes. MRI showed T2/FLAIR signal abnormality of the left temporal and occipital lobe, insular cortex, and splenium of the corpus callosum (E1), in the absence of concomitant enhancement (E2). Surgery was not performed and treatment not administered, given clinical and radiographic stability for >7 years.

Tumor involvement varied depending on the location in the brain (Table 1), with the temporal lobe being most frequently involved. Of the four IDH‐mutant patients, three had disease limited to three regions (Table 3).

Treatment of GC (Tables 1, 2)

Surgery was performed in 33 of 34 patients, including stereotactic biopsy in 22 of 34 (65%), subtotal resection of the T2/FLAIR hyperintense tumor in 10 of 34 (29%), and gross total resection in 1 of 34 patients (3%).

Postoperative treatment varied between subjects. Thirteen patients (38%) received chemotherapy alone, including single‐agent TMZ (n = 8), PCV (n = 2), carboplatin (n = 1), or a combination regimen (n = 2). Seventeen patients (50%) received some form of RT, either alone (n = 3), with concurrent TMZ (n = 9), or sequentially with adjuvant TMZ (n = 3) or PCV (n = 2). Six patients were treated with concurrent RT and TMZ, followed by adjuvant TMZ, which is the current standard of care for GBM [16]. Most patients received focal RT, and only two patients underwent whole‐brain RT (WBRT). Three patients did not receive any tumor‐directed treatment after surgery: Case 9 and 17 were initially observed after diagnosis until MRI progression was noted after 8 and 4 months, respectively; Case 27 was managed supportively due to severe functional impairment and died 20 days after diagnosis of a diffuse infiltrating glioma.

In the recurrent setting, six (18%) patients underwent repeat surgery. RT alone was provided to 10 patients (29%) whereas 4 patients (12%) were treated with concurrent TMZ, including 1 (Case 29) who had received chemoradiation as first‐line therapy. Fifteen (44%) patients received chemotherapy alone, which included antiangiogenic, targeted, and investigational agents. TMZ was the most frequently used chemotherapeutic agent (n = 14; 41%), followed by bevacizumab with or without irinotecan (n = 12; 35%). Other types of chemotherapy included alkylating agents (cyclophosphamide, carboplatin), tyrosine kinase inhibitors (vandetanib, erlotinib, sorafenib, dasatinib), and antiproliferative and other antiangiogenic agents (cilengitide, etoposide, sirolimus, thalidomide, tamoxifen, tipifarnib, cabozantinib, aflibercept).

Treatment Response and Outcome (Table 4)

Table 4. Univariate Cox regression analysis of clinical, histopathologic, and radiographic factors and their association with OS and TTP.

^a Bolded p values are statistically significant.

^b Includes RT alone, concurrent CRT, RT+ adjuvant CHT, and concurrent CRT + adjuvant CHT.

Abbreviations: CHT, chemotherapy; CI, confidence interval; CRT, chemoradiation; HR, hazard ratio; MRI, magnetic resonance imaging; mut, mutation; OD/OA, oligodendroglioma/oligoastrocytoma; OS, overall survival; RT, radiotherapy; TTP, time to progression; wt, wild‐type.

Median OS for all patients was 41.9 months (range 20 days to 301.4 months) from the time of radiologic diagnosis, with a median follow‐up time of 123.6 months (10.3 years; 95% CI >67.6 months). Median TTP (mTTP) was 19.5 months (range 9 days to 74.2 months) and median follow‐up time 94.8 months (7.9 years; 95% CI >57 months). Eleven patients developed PD by RANO criteria while on first‐line therapy, and 17 individuals remained stable until last follow‐up or before disease progression occurred. Four patients had an MR to treatment, and one individual (Case 24) had a PR after receiving sequential RT and TMZ. Twenty‐two patients (65%) have died at the time of last follow‐up.

Given the small number of patients with oligodendroglioma and oligoastrocytoma, these histologic subtypes were grouped together in the survival analysis. 1p19q analysis was not available for the patients with oligoastrocytomas. Tumor histology did not affect OS (HR 1.71, p = .34) or TTP (HR 2.24, p = .12). Notably, three of the four patients with oligodendrogliomas (Case 5, 14, and 19) had survival times ranging from 123.6 to 301.4 months (10.3–25.1 years) and were still alive at the time of last follow‐up. There was a trend toward an association between tumor grade and OS (Fig. 3A), with a survival advantage in patients with grade 2 (mOS 76.7 months) versus grade 3 tumors (mOS 43.7 months; HR 2.65, p = .051).

Figure 3.

Overall survival. Kaplan‐Meier curves stratified by age (A), histologic grade (B), and extent of radiographic involvement (C).

Abbreviation: WHO, World Health Organization.

There was also a trend toward shorter TTP in IDH‐wt patients (HR 4.81, p = .052). Median OS has not been reached in the IDH‐mutant group, compared with an mOS of 43.7 months in the IDH‐wt group. Of the four IDH1‐mutant patients, two have remained stable after first‐line treatment (Case 4 and 31) and two progressed after 41.1 (Case 10) and 67.9 (Case 14) months. Three of these patients (Case 4, 14, 31) are still alive, whereas one (Case 10) died 201.1 months (18 years) after initial diagnosis (Table 2). By contrast, mTTP in IDH1‐wt patients was only 12 months. Notably, some cases without confirmed IDH mutation demonstrated an exceptionally favorable clinical course: Case 5 was diagnosed with a grade 2 oligodendroglioma at age 9 and has remained in remission for 25 years after treatment with RT and PCV. IDH status was not tested at the time of original surgery in 1991 and was negative on subsequent IHC. However, there was insufficient tissue for genetic sequencing to evaluate for a noncanonical IDH mutation. In addition, three patients (Case 7, 13, 25) with astrocytomas demonstrated survival times >100 months (8.3 years). Case 7 and 25 had a grade 3 astrocytoma and lived 150.7 and 103.8 months (12.6 and 8.7 years) after their initial diagnosis, respectively. Case 13 was diagnosed with a grade 2 astrocytoma at age 2 and remains alive 160.6 months (13.4 years) later.

The number of involved regions (stratified by 3 regions vs. >3 regions) was not associated with worse OS (HR 0.73, p = .47; Fig. 3C). For instance, Case 7 had involvement of 7 regions but survived 150.7 months. By contrast, Case 27 was found to have extensive T2/FLAIR hyperintensity of both hemispheres affecting >8 regions and died 20 days after diagnosis. Tumor location also did not affect outcome, except for brainstem involvement, which was associated with shorter mTTP (24.0 months in those without and 10.8 months in those with brainstem involvement; HR 2.27, p = .049).

Type of first‐line treatment (stratified by provision vs. no provision of RT of any type) did not affect TTP or OS (HR 1.15, p = .75). A correlation between type of RT (WBRT vs. focal RT) and clinical outcome was not possible given that only two patients received WBRT. Age appeared to have a significant effect on OS, with a cutoff point at 47 years dividing patients into low‐ and high‐risk groups (HR 1.04, p = .003; Fig. 3A). Median OS for patients ≤47 years was 103.8 months, compared with 30 months in those >47 years.

Using the tree‐based classification model, age was the covariate with the strongest association with OS and TTP (p = .004 for OS, p = .019 for TTP). None of the other covariates (tumor grade, number of involved brain regions, and tumor location) demonstrated a significant association with OS and TTP.

Discussion

GC is an infiltrative manifestation of a glial brain tumor, which can be primarily of astrocytic, oligodendroglial, or mixed oligoastrocytic phenotype. It represents a challenging disease entity for clinicians and pathologists alike, given its rarity, association with heterogeneous outcome patterns, and the fact that its diagnosis is primarily based on radiographic criteria. The goal of this retrospective study was to describe the clinical and imaging features of a large single‐institution patient cohort and to identify factors that may help stratify patients into distinct risk groups. Our findings revealed that (a) subgroups of patients exist with significantly prolonged survival up to 25 years after diagnosis, (b) extent of radiographic involvement did not affect OS, while age >47 was associated with worse OS, even when controlling for additional covariates, and (c) there was a trend toward longer TTP in patients with IDH‐mutant tumors and improved OS in those with grade 2 tumors.

Age is a well‐established prognostic factor in high‐grade glioma patients [17]. In addition, a study of patients with primary GC found a trend toward decreased OS in patients >46 years [18]. Our data closely mirror these findings and confirm that more advanced age is associated with shorter OS. This finding holds true even in a model that included additional covariates, including tumor grade, number of involved brain regions, and tumor location. We were unable to control for IDH status due to the lack of available molecular studies in more than half (53%) of patients. It is possible, however, that the impact of age on OS is a reflection of underlying IDH status because IDH mutations are predominantly found in patients <50 years [19].

We found that brainstem involvement was associated with an increased risk for disease progression; however, the number of involved brain regions, tumor location, and presence of bilateral hemispheric involvement did not predict OS. Earlier reports of primary GC patients have suggested that bilateral brain involvement may be associated with shorter survival [11], but a recent study [12] did not find any association between OS and tumor location (supra‐ vs. infratentorial), extent of disease (<6 regions vs. ≥6 regions), bi‐hemispheric disease, or presence of contrast enhancement. Of note, these studies included primary, type II GC patients, that is, those with focal contrast enhancement on MRI, which were excluded from our cohort. Taken together, however, these observations suggest that radiographic features—albeit disease‐defining—may not carry prognostic value in GC patients.

We observed a trend for shorter OS in patients with grade 3 versus grade 2 tumors, which is consistent with previous reports [6], [18]. However, we did not find an association between tumor histology and clinical outcome, contrasting earlier studies that have shown that oligodendroglial histology in GC was a favorable prognostic marker compared with oligoastrocytic and astrocytic tumors [6]. The lack of such an association in our cohort may partly be related to the relatively small number of patients with oligodendrogliomas (n = 4) as, in fact, we did observe remarkable prolonged survival times in three individuals with oligodendrogliomas (beyond 25 years in one patient). Another explanation for this lack of association may be due to potential misclassification of cases as astrocytomas or oligodendrogliomas, given that IDH mutation and 1p19q co‐deletion status were available in only 16 and 7 patients, respectively. Of the four patients with oligodendrogliomas, only one was confirmed to be IDH‐mutated and 1p19q co‐deleted. Other tumors that can histologically mimic oligodendrogliomas include clear cell ependymoma, neurocytoma, dysembyroplastic neuroepithelial tumors (DNETs), and pilocytic astrocytomas [3]. However, of the three patients with oligodendroglial morphology and lack of molecular markers in our cohort, none demonstrated features suggestive of these histologic mimics such as perivascular pseudorosettes (suggestive of clear cell ependymoma), positive immunostaining for NeuN and synaptophysin (suggestive of neurocytoma), glioneuronal features (suggestive of DNET), or presence of a mural nodule and midline location on MRI (suggestive of pilocytic astrocytoma).

We did not identify patients with GBMs in our cohort, which may have been due to surgical sampling error in those who underwent a biopsy. The absence of GBMs may also be explained by selection bias, given that we focused on nonenhancing tumors and the low prevalence (≤4%) of nonenhancing GBMs in general [20], [21], [22].

In low‐ and high‐grade gliomas, the presence of a mutation in the IDH gene is considered a favorable prognostic marker across all histologic grades and subtypes [23], [24], [25], [26]. Similarly, Desestret et al. found that IDH mutations were associated with improved survival in primary, type I GC, with an mOS of 73.9 months in IDH‐mutant compared with 23.6 months in IDH‐wt patients [10]. These findings were in line with results from a study by Seiz et al., which included primary, type I and II GC patients [27]. Notably, mOS in this cohort was shorter than in the study by Desestret et al. (36.3 months in IDH‐mutant and 8 months in IDH‐wt patients). One explanation could be the inclusion of primary, type II GC patients, that is, those with radiographic evidence of a solid tumor mass. The relationship between primary, type I and type II GC remains unclear, but some studies have described the development of solid tumor foci in patients who initially met radiographic criteria for primary, type I GC [28], [29]. Thus, the development of primary, type II GC may indicate a more advanced stage in disease development and more aggressive tumor behavior, and account for the worse survival times observed in the study by Seiz et al. In our cohort, we found a positive trend toward improved TTP in IDH‐mutant patients but no association with OS, most likely due to the small number of patients with a confirmed IDH mutation. We also cannot exclude the possibility of unconfirmed and noncanonical IDH mutations in some individuals, such as three patients (Case 7, 13, and 25) with astrocytomas whose clinical history and survival times (8.7–13.4 years) were, in fact, consistent with the survival data reported in IDH‐mutant grade 2 (10.9 years) and 3 (9.3 years) astrocytomas [30].

In addition to IDH status, MGMT promoter methylation [11], [12] and 1p19q co‐deletion [31] may carry prognostic significance in GC patients. This was corroborated by other studies using alpha‐internexin (INA), a surrogate marker of 1p19q co‐deletion, which demonstrated statistically significant improved progression‐free survival (PFS) and a trend for longer OS in INA‐positive patients on univariate, but not multivariate, analysis [10], [11]. Coexpression of IDH1 and INA has been linked to a threefold increase in OS and PFS compared with double‐negative patients [10]. Based on the 2016 WHO classification, these double‐positive patients likely represent cases of oligodendrogliomas [3], which carry the most favorable prognosis of all glioma subtypes [32], [33].

Some limitations exist in our present study, including its retrospective nature and the heterogeneity of treatment modalities, which preclude us from performing correlative analyses between specific therapies and clinical outcome parameters. In addition, we were unable to obtain a complete set of molecular markers on tissue specimens, given limited access to original tissue specimens from referring hospitals and limited tissue availability after brain biopsy. Notably, the prognostic significance of MGMT promoter methylation was not established until 2005 [34]. Given that 15 of 34 patients in our cohort were diagnosed before 2005, this partly explains the lack of available MGMT testing in these patients. Furthermore, the past standard practice was to perform MGMT testing only in patients with GBMs (none of whom were present in this cohort), which is in contrast to today's practice at most academic centers, which includes routine screening for MGMT promoter methylation in WHO grade 3 gliomas. Lastly, we only included primary, type I GC patients in our analysis, and our findings may thus not be applicable to primary, type II and secondary GC cases.

The strengths of our study include its fairly large cohort size and a well‐defined patient population (i.e., only primary, type I GC patients). The latter is particularly important to avoid confounding effects on outcome analysis imparted by patients with potentially more aggressive and advanced disease as may be seen in those with a pre‐existing tumor mass or contrast enhancement on MRI at the time of diagnosis.

To date, no standardized treatment approach exists for GC patients. The benefit of RT is contested [6], [18], partly due to selection bias on which patients are considered for treatment, and provision of RT is often limited by the widespread area of involvement. Use of certain chemotherapy regimens, such as PCV and TMZ, are extrapolated from the glioma literature [6], [7], [9], [18], but the optimal drug and timing of chemotherapy in relation to RT has not been established. The only prospective trial in GC to date is the German NOA‐05 trial, a single‐arm phase II study evaluating procarbazine and lomustine [11]. Median PFS was 14 months and mOS was 30 months, which compared favorably with the survival times of 11–24 months from retrospective data of patients treated with primary RT [35], [36]. A similar proportion of patients in our cohort received either chemotherapy alone or a combination of RT and chemotherapy, reflecting the lack of consensus on how to best treat GC patients. Future studies should therefore focus on the systematic prospective evaluation of therapeutic strategies in these patients, stratified by subtype of GC and molecular features.

In summary, our findings suggest that there are subsets of patients with GC who demonstrate exceptionally prolonged survival up to 25 years and that extent of radiographic involvement does not carry prognostic significance. Rather, molecular markers such as IDH mutations appear to influence tumor behavior and disease outcome, consistent with published reports. Management of GC patients should thus include comprehensive molecular evaluation, including testing for canonical and noncanonical IDH mutations, MGMT promoter methylation, and 1p19q co‐deletion, to aid in patient risk stratification and prognostication.

Conclusion

Patients with primary, type I GC present with heterogeneous clinical, histopathologic, and imaging findings and highly variable survival times. Importantly, although disease‐defining, extent of radiographic involvement does not detectably influence survival. Rather, IDH mutation status appears to be a strong marker of prolonged TTP. Prospective studies of GC patients stratified by GC subtype and molecular features are needed to better define treatment strategies.

Author Contributions

Conception/design: K. Ina Ly, Jorg Dietrich

Provision of study materials or patients: K. Ina Ly, Derek H. Oakley, Matthew P. Frosch, Stuart R. Pomerantz, Fred H. Hochberg, Tracy T. Batchelor, Jorg Dietrich

Collection and/or assembly of data: K. Ina Ly, Derek H. Oakley

Data analysis and interpretation: K. Ina Ly, Derek H. Oakley, Alexander B. Pine, Sy Han Chiou, Rebecca A. Betensky, Jorg Dietrich

Manuscript writing: K. Ina Ly, Derek H. Oakley, Alexander B. Pine, Matthew P. Frosch, Sy Han Chiou, Rebecca A. Betensky, Stuart R. Pomerantz, Fred H. Hochberg, Tracy T. Batchelor, Daniel P. Cahill, Jorg Dietrich

Final approval of manuscript: K. Ina Ly, Derek H. Oakley, Alexander B. Pine, Matthew P. Frosch, Sy Han Chiou, Rebecca A. Betensky, Stuart R. Pomerantz, Fred H. Hochberg, Tracy T. Batchelor, Daniel P. Cahill, Jorg Dietrich

Disclosures

Tracy T. Batchelor: Merck & Co., Inc., NXDC, Amgen, UpToDate, Inc., Jiahui Health (C/A, SAB, H). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board