Prediction of Response to Chemotherapy in Anaplastic Glioma: How Many Markers Does It Take?

Abstract

Journal of Clinical Oncology recently published the results of the long-term follow-up of patients enrolled onto the Radiation Therapy Oncology Group (RTOG) 9402 trial and the European Organization for Research and Treatment of Cancer (EORTC) 26951 trial. These studies demonstrate that first-line procarbazine, lomustine, and vincristine (PCV) chemotherapy and radiation dramatically improve overall survival in patients with anaplastic oligodendroglioma or anaplastic mixed oligoastrocytoma whose tumors contain allelic loss of chromosomes 1p and 19q, compared with radiation alone. In the article that accompanies this editorial, Erdem-Eraslan et al attempt to explain the survival benefits by publishing detailed molecular analyses of tumor tissue from a subset of patients who were treated on the EORTC trial. This article demonstrates the prognostic and predictive value of intrinsic glioma subtype in patients with anaplastic glioma and suggests that narrow classification using one or two molecular markers may not adequately identify all patients who will benefit from radiation and PCV chemotherapy. This finding is significant because it represents an important advance in prognostication; that is, it demonstrates not just the need to move beyond the histologic information that is used to enroll and categorize patients in clinical trials but may also indicate the need to extend enrollment beyond a single molecular marker, chief among them loss of heterozygosity (LOH) of 1p19q. It is well known that histologic subtyping (WHO criteria) is insufficient to accurately classify glioma subtypes. Histology alone is subjective and prone to interobserver variability. These deficiencies may reflect the genetic heterogeneity that is intrinsic to poorly differentiated tumors, including malignant gliomas, and underscores the need to develop molecular signatures that better predict tumor behavior. Since the discovery that 1p19q LOH is associated with up to 70% of anaplastic oligodendrogliomas and response to PCV chemotherapy, it is widely used as a molecular marker of oligodendroglioma. The EORTC 26951 trial was initiated before the identification of such markers that are now used to enrich patients for anaplastic oligodendroglioma. One limitation of the trial itself was that patients were enrolled and classified using only histology, although this now allows the opportunity for unbiased analysis of the molecular data. Up to 30% of oligodendroglioma lineage tumors do not have 1p19q LOH, yet may still respond to chemotherapy. Thus, additional methods to identify patients who might benefit from chemotherapy are necessary. Several groups have attempted to improve on histology using unsupervised gene expression profiling to identify subgroups of patients whose tumors are molecularly and histologically similar but who nevertheless have different prognoses. The authors previously identified seven molecular glioma clusters (so-called intrinsic glioma subtypes, or IGS) that were characterized by specific genetic changes; the IGSs were shown to be better predictors of survival than histology. These data were validated using several large external data sets, but intrinsic subtypes have not been validated in a large randomized clinical trial. In their study, Erdem-Eraslan et al used a subset of tumors from patients enrolled onto EORTC 26951 to further validate their subtype grouping in a randomized trial. In their analysis, IGS was highly prognostic for overall survival for patients who were treated with radiation and chemotherapy, independent of the patients’ clinical, molecular (IDH1 mutation and 1p19q LOH), and histologic profiles. Although statistically significant, the contribution of IGS to prognostication seems to have been lower than the contributions of other markers; for example, the hazard ratio for intrinsic subtype was 1.06 (95% CI, 1.03 to 1.12), whereas that for 1p19q LOH was 0.32 (95% CI, 0.17 to 0.59). These data suggest that although intrinsic subtype can separate patients into biologically distinct groups with different outcomes, molecular markers such as IDH1 mutation and 1p19q LOH may be more predictive of behavior than either histology or IGS. Because EORTC 26951 only enrolled patients with anaplastic oligodendroglioma or anaplastic mixed oligoastrocytoma, the true predictive power of IGS is not known: IGS may be a strong independent prognostic factor when applied to all infiltrative gliomas. This possibility should be tested going forward. Perhaps the most exciting information from the article by Erdem-Eraslan et al was their identification of one glioma subtype that significantly benefited from chemotherapy. Specifically, they found that patients whose tumors were within IGS-9 had significantly longer progression-free and overall survival when treated with adjuvant PCV. In one sense, this is not surprising, given that the IGS-9 group likely was enriched for anaplastic oligodendrogliomas. However, the IGS-9 subtype includes other histologies, as well as tumors with intact or only partially deleted 1p and/or 19q. The fact that Erdem-Eraslan et al found that only 63% of IGS-9 tumors had 1p19q LOH and only 70% had IDH1 mutation suggests that intrinsic subtyping JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 31 NUMBER 3 JANUARY 2