Morphologic patterns of noncontrast-enhancing tumor in glioblastoma correlate with IDH1 mutation status and patient survival

Highlights

• Glioblastomas can be classified by the pattern of nCET.

• IDH1^mut glioblastomas often demonstrate mass-like nCET.

• Using the morphology of nCET can thus improve the prediction of IDH1 mutations.

• The pattern of nCET may also influence patient survival.

Abstract

Glioblastomas with a substantial proportion of noncontrast-enhancing tumour (nCET) have a variety of imaging appearances. We aimed to determine whether glioblastomas demonstrating a substantial proportion (>33%) of nCET can be sub-classified by different morphologic pattern of nCET. We then assessed whether this improves the ability of MRI to predict isocitrate dehydrogenase-1 (IDH1) mutation status and whether this has prognostic significance independent of IDH1 mutation status. Pre-operative MRIs of patients with a new diagnosis of glioblastoma were reviewed. Tumours with >33% nCET were sub-classified by the dominant morphologic pattern of nCET: mass-like expansion, white matter dissemination, grey matter dissemination or a combination. IDH1 mutation status (by immunohistochemistry) and survival were compared for each pattern. 153 patients met the inclusion criteria, of whom 34 patients demonstrated >33% nCET. 10 patients had a significant mass-like component, either as the dominant pattern (n = 4) or as part of a mixed pattern (n = 6). The 10 patients with a significant mass-like component had longer survival than those without (median 387 days, compared to 241 days), though this was not statistically significant (p = 0.242). Three patients had R132H-IDH1 mutations and >33% nCET, and all three had a mass-like component. Using the presence of a mass-like component of nCET for predicting IDH1 mutation status improved the positive predictive value, specificity and overall accuracy of MRI. Classification of nCET by morphologic pattern improves the ability of MRI to predict IDH1 mutations and may provide useful prognostic information.

Introduction

Tumours with the same histological diagnosis of glioblastoma and an equivalent proportion of noncontrast-enhancing tumour (nCET) according to the current VASARI criteria [1] often, however, have vastly different imaging appearances. We sought to determine if variation in appearance has clinical or prognostic implications for the patient. Isocitrate dehydrogenase mutation testing has become a crucial part of glioma assessment and is recommended as routine in the recently-updated World Health Organisation guidelines due to its prognostic importance [2]. Carrillo et al. have previously described an association between isocitrate dehydrogenase-1 (IDH1) mutations in glioblastoma patients with the presence and a greater proportion of nCET [3]. In our own patient cohort we also found that the presence of >33% nCET improved the prediction of IDH1 mutation status, however this association was only modest [4]. Due to the low incidence of IDH1 mutations in glioblastoma, reported as being <5% [5], the majority of patients with >33% nCET nonetheless have IDH1-wildtype (IDH1^wt) tumours [4].

Illustrative images were provided for four patients with IDH1-mutated (IDH1^mut) glioblastomas in the study by Carrillo et al. [3], and closer inspection suggests that all patients had a dominant, partially-enhancing mass, rather than a more widely-disseminated pattern of nCET. In fact, the presence of nCET distant to the dominant glioblastoma has been shown to be a feature of IDH1^wt tumours [6]. Similarly, Seiz et al. found significant differences in the incidence of IDH1 mutations in patients with gliomatosis cerebri depending on the MRI appearances. IDH1 mutations were demonstrated in 10 of 24 patients who met the criteria for gliomatosis cerebri but also had a dominant mass, while no mutations were detected in 11 patients with widespread tumour but no dominant mass [7]. All patients in the above study can be assumed to have had extensive nCET given the diagnosis of gliomatosis cerebri, suggesting that it is not just the extent of nCET that has a role in determining IDH1 mutation status, but also the morphologic pattern.

We hypothesise that the current definition of nCET is too broad, encompassing a variety of molecular phenotypes, and thus limiting the predictive value of MRI. The purpose of this study was to determine whether glioblastomas demonstrating a substantial proportion (>33%) of nCET can be sub-classified by different morphologic pattern of nCET, and whether this classification can both improve the ability of MRI to predict IDH1 mutation status and provide prognostic information independent of IDH1 status.