Tailored Therapy in Diffuse Gliomas: Using Molecular Classifiers to Optimize Clinical Management
Tailored Therapy in Diffuse Gliomas: Using Molecular Classifiers to Optimize Clinical Management
ABSTRACT: Diffuse gliomas are the most common primary malignant brain tumors in adults and continue to be almost universally fatal. Nevertheless, a striking variability in outcome has long been observed, with a subset of patients having prolonged survival. Recent molecular discoveries have provided new insights into gliomagenesis and have enhanced clinical subclassification of gliomas. Mutations in the isocitrate dehydrogenase (IDH) genes occur frequently in low-grade astrocytomas and oligodendrogliomas (World Health Organization [WHO] grade II), and in higher-grade gliomas (WHO grades III and IV) that arise after malignant progression of low-grade tumors. IDH mutation has an established role as a favorable prognostic marker; however, the utility of IDH mutation in guiding treatment is still under investigation. A subset of IDH-mutant tumors, predominantly oligodendrogliomas, also harbor codeletion of chromosomes 1p and 19q, a feature that predicts responsiveness to chemotherapy. Here, we review the current data regarding the prognostic and predictive value of IDH mutation and 1p/19q codeletion in gliomas. We also discuss possible management algorithms using these biomarkers to tailor surgical and adjuvant therapy for specific diffuse gliomas. Ultimately, understanding the natural history of glioma subtypes and the predictive value of genetic markers may maximize survival and minimize treatment morbidity.
Diffuse gliomas continue to be difficult malignancies to manage despite the recent discovery of strong prognostic molecular markers. Gliomas account for 30% of the incidence of primary central nervous system (CNS) tumors, which is 26 per 100,000 for adults ≥ 20 years old. The World Health Organization (WHO) grades gliomas on a scale of I to IV, with higher grades corresponding to more malignant features. Pathologically, gliomas are divided into astrocytic (75%) or oligodendroglial tumors (25%), depending on the primary histology, with rare cases of mixed histology. WHO grade II diffuse gliomas (also known as low-grade gliomas) and grade III (anaplastic) gliomas are less common than grade IV gliomas (glioblastomas). Glioblastoma is the most common malignant primary brain tumor in adults and accounts for > 30% of all gliomas. Primary glioblastomas are defined as tumors that are discovered de novo, whereas secondary GBMs are tumors that arise after malignant progression via transformation of a lower-grade glioma.
In young adults (aged 20 to 34 years), gliomas account for 32% of primary brain tumors. Treatment of this population poses a particular set of challenges. It has long been observed that these patients have improved survival compared with older patients with gliomas of identical histology; however, it remains uncertain whether these patients should be treated differently. The recent discovery of recurrent somatic mutations in the isocitrate dehydrogenase 1 (IDH1) and IDH2 genes in gliomas has provided substantial insight into the improved outcomes seen in subsets of patients.[3-5] IDH mutations correlate with younger age at diagnosis, accounting for most of the prognostic effect of age. Furthermore, IDH mutations are also independently associated with improved survival.[5,7]
Although knowledge of the biology and natural history of IDH-mutant gliomas is increasing rapidly, how to incorporate this powerful prognostic marker into clinical management remains unresolved. Here, we review the epidemiology and pathophysiology of IDH-mutant gliomas, the association of IDH mutation with other molecular markers, and the value of IDH mutation as a prognostic and possibly predictive marker. Finally, we will discuss how our understanding of molecular prognostic factors could be used to stratify therapeutic decisions for patients with malignant gliomas.
Epidemiology and Diagnosis
Mutations in the IDH genes were discovered in 2008 as a result of genome-wide sequencing efforts in glioblastoma. In the initial report, recurrent mutations in the R132 position of the IDH1 gene were observed in 12% of glioblastoma cases, and these mutations were enriched in patients with secondary glioblastoma. Further investigation in diverse primary brain tumors identified recurrent mutations in IDH1 or in the analogous amino acid position R172 of the related gene IDH2 in 55% to 80% of WHO grade II and III diffuse gliomas. Conversely, IDH mutations were found in only 5% to 10% of primary glioblastomas and in < 10% of other primary CNS tumors, such as pilocytic astrocytomas, pediatric glioblastomas, ependymomas, and meningiomas.[3,4,8-11] In addition to gliomas, IDH mutations are frequently observed in acute myeloid leukemia, in intrahepatic cholangiocarcinoma, and in chondrosarcoma.
As suggested above, patients with IDH-mutated gliomas have a distinct clinical phenotype. They are significantly younger than those with IDH−wild-type tumors across all tumor grades. The incidence peaks in patients in their third decade, and mutations are rarely seen in the pediatric or elderly population. High-grade IDH-mutant gliomas also have distinct radiographic characteristics. There are predilections for frontal lobe location, larger size at diagnosis, more frequent nonenhancing tumor component, less contrast enhancement, and reduced necrotic-appearing areas, as well as for more prevalent cystic and diffuse components.[6,15] These features are similar to the classic radiographic pattern associated with lower-grade diffuse gliomas.
IDH mutations are strongly associated with other molecular markers of improved survival, including O6-methylguanine-DNA methyltransferase (MGMT) promoter DNA methylation and cytosine-phosphate-guanine (CpG) island hypermethylator phenotype (G-CIMP).[7,15,16] Furthermore, deep genomic sequencing analyses have subclassified IDH-mutant tumors based on the tightly clustered association of IDH mutation with other genetic alterations. IDH mutation is frequently observed with TP53 and alpha thalassemia/mental retardation syndrome X-linked (ATRX) mutations in astrocytic tumors, and with deletion of 1p and 19q (1p/19q codeletion) and mutations in the homologue of the Drosophila capicua gene (CIC) and the promoter of the telomerase reverse transcriptase gene (TERT) in pure oligodendrogliomas.[7,15,17-20]
Perhaps the most striking clinical feature of patients with IDH-mutated gliomas, however, is the improved survival seen across all glioma types compared with patients with IDH−wild-type gliomas. Despite the relatively recent discovery of IDH mutations in gliomas, the prognostic data have come rapidly from retrospective analyses of tissue banks from clinical trials. IDH1 and IDH2 status was determined in subsets of patients from two large prospective, randomized trials of newly diagnosed anaplastic gliomas (the German NOA-04 and the European Organisation for Research and Treatment of Cancer [EORTC] 26951 trials), and after adjusting for other prognostic factors—such as age, histology, grade, extent of resection, MGMT promoter methylation, and 1p/19q codeletion—IDH1 mutation remained a strongly significant independent prognostic factor.[5,7,21,22] In the NOA-04 analysis, IDH1 mutation was the dominant prognostic factor. Furthermore, IDH analysis of glioblastomas from the German Glioma Network and of anaplastic astrocytomas from the NOA-04 trial indicated that the median survival for IDH−wild-type anaplastic astrocytomas was worse than that of IDH-mutated glioblastoma, suggesting that IDH mutational status may be considered as important a prognostic marker as histology in grade III and IV astrocytomas.
The rapid diagnostic assessment of IDH status has been greatly facilitated by the development of an antibody directed against the most prevalent IDH-mutant isoform (IDH1 R132H substitution—which accounts for approximately 85% of all IDH mutations in gliomas[4,6]), as this antibody has become widely available and has been adopted in many clinical diagnostic laboratories. Compared with sequencing, the sensitivity and specificity of the IDH1 R132H antibody to detect all variants of IDH1 by immunohistochemistry (IHC) have been reported to be 94% and 100%, respectively. There may be higher inter-laboratory agreement for IHC than for sequencing. One study demonstrated 100% concordance for IDH1 R132H IHC results across laboratories in a test set of gliomas, whereas several laboratories had inconsistent results using sequencing. In addition to facilitating the genotyping of IDH in diffuse gliomas, IHC for IDH1 status has improved the accuracy of histopathological diagnosis in cases with limited sample available, thereby aiding in distinguishing between diffuse gliomas and pilocytic astrocytomas, ependymomas, and nonneoplastic reactive gliosis.[24-26]
The Pathophysiology of IDH Mutations
Mutations in IDH are nearly all heterozygous missense somatic mutations at codon 132 (IDH1) or 172 (IDH2), which are important structural residues for enzymatic function and ligand binding. Upwards of 90% of IDH1 mutations are single-base transition substitutions from G to A, resulting in an arginine to histidine substitution (R132H), with other mutations occurring less frequently. IDH2 mutations are much less common (~3%) and are associated with oligodendroglial histology.[4,6,10]
IDH is a catalytic enzyme with at least three isoforms—IDH1, IDH2, and IDH3. IDH3 is the canonical nicotinamide adenine dinucleotide (NAD+)-dependent enzyme involved in the Krebs cycle, and mutations in IDH3 have not been observed in gliomas. The normal function of IDH1, located in the cytoplasm, is to catalyze the oxidative decarboxylation of isocitrate into alpha-ketoglutarate (alpha-KG) and nicotinamide adenine dinucleotide phosphate (NADPH). These metabolites are then thought to be involved in cholesterol and lipid synthesis. The identical reaction is carried out by IDH2 in the mitochondria. The mutated IDH enzyme loses the ability to generate alpha-KG and gains the ability to convert alpha-KG to 2-hydroxyglutarate (2-HG), which then accumulates to significant levels in gliomas.
Due to the exceedingly high levels detected in glioma tissue—up to one to two orders of magnitude higher than normal—2-HG has been proposed as a novel oncometabolite. However, its exact mechanistic role in gliomagenesis is still under active investigation. Emerging data suggest that 2-HG exerts its oncogenic effect by competitively inhibiting 2-oxoglutarate (2-OG)-dependent enzymes, such as hypoxia-inducible factor prolyl hydroxylases, methyl-cytosine hydroxylases (which regulate DNA methylation), and histone demethylases,[30-32] which result in global histone and DNA hypermethylation and in induction of G-CIMP.[15,32,33] Some of these effects may also result in accumulation of hypoxia-inducible factor (HIF)-1α and increased angiogenesis.[34,35]
As mentioned above, other genetic alterations have recently been found to cluster tightly with IDH mutations in gliomas, raising the possibility that the mechanistic basis of prognostic and predictive impact may, in fact, be due to these additional alterations. These alterations therefore present the opportunity for further stratification. Patients with concurrent IDH, p53, and ATRX mutations have an “intermediate” median overall survival of approximately 5 years. Mutations in the CIC gene on chromosome 19q and in the far upstream binding protein 1 (FUBP1) gene on chromosome 1p are associated with classical histologic oligodendrogliomas. As would be expected, CIC mutations strongly correlate with 1p/19q codeletion and, when present with IDH mutations, are associated with a median overall survival of 8 or more years. If none of these alterations are present, the median overall survival is approximately 1 year, regardless of tumor grade, essentially mirroring the natural history of classical glioblastoma.
Prognostic and Predictive Implications
Recently mature data from the large, prospective, randomized Radiation Therapy Oncology Group (RTOG) and EORTC trials investigating the role of chemotherapy (procarbazine, lomustine [CeeNU], and vincristine [PCV regimen]) in addition to radiation in newly diagnosed anaplastic oligodendroglial tumors provide intriguing insight into prognostic and predictive markers for anaplastic gliomas. An estimated 25% to 50% of anaplastic oligodendrogliomas and oligoastrocytomas harbor 1p/19q codeletion and associated IDH mutations. Although these trials opened to accrual in the 1990s, prior to the discovery of IDH1 mutation or the prognostic and potentially predictive importance of 1p/19q codeletion, 1p/19q status was determined in 91% of patients in the RTOG study and in 86% in the EORTC study.
Regardless of treatment assignment, in both studies patients with 1p/19q codeletion survived much longer than patients with non-codeleted tumors (eg, overall survival of 123 vs 23 months in the EORTC study), confirming the strong prognostic effect of 1p/19q codeletion.[37,38] In addition, these studies report data suggesting that 1p/19q codeletion predicts sensitivity to PCV chemotherapy. In the RTOG study, patients with codeletion who received PCV in addition to radiation at diagnosis lived much longer than those treated with radiation alone (median overall survival of 14.7 vs 7.3 years, P = .03). Conversely, there was no difference in the survival of patients with non-codeleted tumors in the different treatment arms. In the EORTC study, the median survival of patients with 1p/19q codeleted tumors who received radiation and PCV had not yet been reached, compared with 112 months for patients with codeleted tumors treated with radiation alone. Of note, this difference was a trend for extended survival in the chemotherapy group (P = .059). As with the RTOG study, in the non-codeleted subset, there was no difference in survival between treatment arms.[7,37]
The prognostic impact of IDH mutation was also confirmed in both studies. The EORTC study retrospectively evaluated IDH1 status in 179 of 368 enrolled patients, and IDH mutation was found to be an independent prognostic indicator, with median overall survivals of 8.4 years for patients with IDH-mutated tumors vs 1.4 years for patients with IDH−wild-type tumors. IDH mutation was confirmed to be an important prognostic factor in the RTOG study as well. However, the question of whether IDH mutation itself predicts sensitivity to chemotherapy remains unresolved. In the EORTC study, IDH-mutant tumors seemed to derive more benefit from PCV chemotherapy; however, tests for interactions with chemotherapy were not statistically significant.[7,37]
In NOA-04, a randomized phase III trial of radiation vs chemotherapy (either PCV or the currently more widely used alkylating agent temozolomide [Temodar]) in anaplastic gliomas, IDH1 mutation was retrospectively reviewed in a subset of patients and was found to be associated with longer time to treatment failure but was not predictive of response to chemotherapy. A follow-up study demonstrated that IDH1 mutation was an independent prognostic factor for overall survival, and was in fact the most prominent prognostic factor for survival.
As a result of these retrospective analyses, the National Comprehensive Cancer Network (NCCN) guidelines now recommend the use of chemotherapy with radiation for treatment of anaplastic gliomas harboring a 1p/19q codeletion.
In low-grade gliomas, the predictive value of IDH mutations with regard to radiation or chemotherapy has yet to be determined. One retrospective study of 84 patients with low-grade gliomas who were treated with temozolomide upfront after surgical resection reported that IDH-mutant tumors had higher rates of radiographic responses; the response rate was highest in patients with both IDH1 mutation and 1p/19q codeletion. However, other retrospective studies of low-grade astrocytomas failed to correlate IDH1 mutation with temozolomide response.[41,42]
Given the mounting evidence that 1p/19q codeletion conveys sensitivity to chemotherapy, two large, multicenter randomized trials will address the issues of temozolomide efficacy and chemosensitivity in 1p/19q codeleted and non-codeleted anaplastic tumors. The CODEL trial (NCT00887146), a phase III study currently under review by the RTOG, is considering randomization of patients with newly diagnosed anaplastic gliomas with 1p/19q codeletion to either PCV, temozolomide, or combined chemotherapy and radiation. The specific regimens in each arm remain to be determined. For patients with newly diagnosed anaplastic gliomas without 1p/19q codeletion, the counterpart trial, CATNON (NCT00626990), is currently randomly assigning patients to one of four arms: radiation alone; radiation with concurrent daily temozolomide without further adjuvant therapy; radiation alone, followed by 12 cycles of adjuvant temozolomide; and finally, concurrent daily chemoradiation followed by 12 cycles of temozolomide. In this study, patients will be stratified on the basis of MGMT status.