Oligodendroglial tumours: subventricular zone involvement and seizure history are associated with CIC mutation status

Background CIC-mutant oligodendroglial tumours linked to better prognosis. We aim to investigate associations between CIC gene mutation status, MR characteristics and clinical features. Methods Imaging and genomic data from the Cancer Genome Atlas and the Cancer Imaging Archive (TCGA/TCIA) for 59 patients with oligodendroglial tumours were used. Differences between CIC mutation and CIC wild-type were tested using Chi-square test and binary logistic regression analysis. Results In univariate analysis, the clinical variables and MR features, which consisted 3 selected features (subventricular zone[SVZ] involvement, volume and seizure history) were associated with CIC mutation status (all p < 0.05). A multivariate logistic regression analysis identified that seizure history (no vs. yes odd ratio [OR]: 28.960, 95 confidence interval [CI]:2.625–319.49, p = 0.006) and SVZ involvement (SVZ- vs. SVZ+ OR: 77.092, p = 0.003; 95% CI: 4.578–1298.334) were associated with a higher incidence of CIC mutation status. The nomogram showed good discrimination, with a C-index of 0.906 (95% CI: 0.812–1.000) and was well calibrated. SVZ- group has increased (SVZ- vs. SVZ+, hazard ratio [HR]: 4.500, p = 0.04; 95% CI: 1.069–18.945) overall survival. Conclusions Absence of seizure history and SVZ involvement (−) was associated with a higher incidence of CIC mutation. Electronic supplementary material The online version of this article (10.1186/s12883-019-1362-y) contains supplementary material, which is available to authorized users.


Background
Low-grade gliomas (LGGs) exhibiting oligodendroglial features include oligodendrogliomas and oligoastrocytomas [1]. The updated 2016 edition of the World Health Organization (WHO) Classification of tumors of the Central Nervous System (CNS) uses molecular parameters and the histology to define the main tumor categories for the first time. This represents a shift from the traditional principle of using neuropathological diagnoses, which are primarily based on the microscopic features, to using molecularly-oriented diagnoses [2,3]. Therefore, neurosurgeons increasingly depend on molecular genetic features to guide their clinical judgement and decision-making processes [4,5].
LGG samples with an isocitrate dehydrogenase (IDH) mutation and the codeletion of 1p and 19q had the most favourable outcomes for treatment [6,7].
To date, although several studies have evaluated MRI characteristics as they relate to IDH/1p19q status [18,19], no study has investigated associations between CIC mutation status and MR imaging features in oligodendroglial tumours. Radiological detection of CIC mutation status may facilitate the preoperative prediction of a patient's prognosis. Therefore, this paper reports preliminary research that can be used to determine the associations between CIC gene mutation status, MR characteristics and clinical features.

Patient population
All patient data was acquired from the published The Cancer Genome Atlas LowGrade Glioma (TCGA-LGG) project and within this publication it is stated "Specimens were obtained from patients, with appropriate consent from institutional review boards". (http:// cancergenome.nih.gov/).

Statistical analysis
Differences between CIC mutation and CIC wild-type were tested using the Chi-square test and binary logistic regression analysis (version 22.0; SPSS Company, Chicago, IL). Odd ratios (OR) and 95% confidence intervals (CI) are reported. The area under the receiver operator characteristic curve (AUC) was estimated for prediction of CIC gene mutation status. The sensitivity, specificity, positive likelihood ratio (PLR) and negative likelihood ratio (NLR) of the model in the prediction of CIC mutations were obtained. Survival analysis (SVZ-vs. SVZ+) was estimated using the Cox proportional hazards models. Hazard ratios (HR) and 95% CI are reported. The statistical significance threshold was set at a P-value of 0.05 (two-sided) to indicate statistical significance.
In univariate analysis, the clinical variables and MR features, which consisted of three selected features (SVZ involvement, volume and seizure history) were significantly associated with CIC mutation status (all P < 0.05). We demonstrated that a smaller tumour volume (OR: 9.100, P = 0.004), SVZ-(OR: 20.400 P = 0.006) and a history absent of seizures (OR: 6.462, P = 0.014) were associated with a significantly higher incidence of CIC mutations (Table 1 Fig. 2).
In multivariate logistic regression analysis, only two risk factors were significant independent predictors ( Table 2). We demonstrated that seizure history (no vs. yes OR: 28.960, 95CI:2.625-319.49, P = 0.006) and subventricular zone involvement (SVZ-vs. SVZ+ OR: 77.092, P = 0.003; 95% CI: 4.578-1298.334) were associated with a higher incidence of CIC mutation status. The nomogram displayed high discrimination, with a Cindex of 0.906 (95% CI: 0.812-1.000) and was well  Fig. 2 Smaller tumor volume, SVZ-and absent seizure history were associated with a significantly higher incidence of CIC mutation (univariate analysis P < 0.05) calibrated (Fig. 3). The sensitivity, specificity, positive likelihood ratio (PLR) and negative likelihood ratio (NLR) of this model in the prediction of CIC mutations were 0.90, 0.71, 3.09 and 0.14, respectively. Subventricular zone involvement (−) of oligodendroglial tumours in combination with absence of seizure history may therefore be used to better prognosticate CIC mutation status than the use of each variable alone (Fig. 4). Patients (follow up: 0.03-156.11 months) with SVZhad a longer median overall survival (133.6 vs.65.7)

Discussion
Molecular genetic studies demonstrated distinct glioma entities with specific epigenetic and genetic profiles [23]. Some oligoastrocytomas and most oligodendrogliomas are characterized by a typical and unique unbalanced translocation, der (1,19), resulting in a 1p/19q codeletion (codeletion of 1p and 19q). Candidate tumour suppressor genes (TSGs) targeted by these losses, including FUBP1 on 1p31.1 and CIC on 19q13.2, were only recently discovered [10]. CIC-mutant oligodendroglial tumours are also linked to better prognoses [8,9]. There are a number of studies regarding the relationship between imaging features and gene mutations. Rios Velazquez E et al. [24] confirmed that quantitative features related to intratumour heterogeneity that were able to successfully discriminate (AUC = 0.69) between EGFR-and EGFR+ lesions. Brendle C et al. [25] showed that the differentiation of high-grade gliomas and lowgrade gliomas (sensitivity: 100% specificity: 80%) is made possible by the dynamic contrast-enhanced MR perfusion parameter Ve (P = 0.024), while arterial spin labelling perfusion shows the potential for the discrimination of the ATRX and IDH mutation statuses (sensitivity: 75% specificity: 88%, P = 0.014). Dagher J [26] implied that wild-type von Hippel-Lindau (VHL) renal cell carcinomas were associated with lymph nodal metastases. Rizzo S [27] found that a pleural effusion related to ALK mutations while nodules located in non-tumour lobes or round lesion shapes were related to a KRAS mutation in subgroups of non-small cell lung cancer patients. So far, no study has investigated associations between CIC mutation status and MR imaging features in oligodendroglial tumours. This study suggests that SVZ involvement and seizure history can be conveniently used to facilitate the prediction of CIC mutation status.
The subventricular zone has been associated with the origination and development, as well as the biological behaviour of LGGs [22]. Recently, limited studies have reported associations between SVZ involvement and patient prognosis. Nakagawa Y et al. [28] implied that the loss of 19q and lack of SVZ+ might be prognostic for longer survival. Liu S et al. [22] reported that multivariate analysis showed that a shorter distance between the tumor centroid and the SVZ (p = 0.039) was significantly associated with poor overall survival in SVZinvolved patients (low-grade astrocytoma). Liu S et al. [22] also reported that a longer distance between the SVZ and the tumour centroid was significantly related to better overall survival in SVZ-involved LGG patients. Adeberg S et al. [8] confirmed that the tumour location with regard to the subventricular zone is related to a patient's prognosis (p < 0.05).
Nomograms are user-friendly tools that give relative contexts and probabilities of cancer prognoses [11]. In Fig. 4 ROC curve analysis showed that subventricular zone involvement of oligodendroglial tumors in combination with seizure history may therefore be used to prognosticate CIC mutation status better than use of variables alone the present study, we constructed three models for the preoperative prediction of CIC mutation statuses in oligodendroglial tumours. ROC curve analysis showed that use of seizure history combined with SVZ involvement (AUC = 0.906) was superior to simply the use of seizure history (AUC = 0.725) or SVZ involvement alone (AUC = 0.804). The discriminatory power of the nomogram which combined SVZ involvement and seizure history was also very strong and well calibrated.
The limitations of our study included the retrospective nature of our data collection, the relatively limited number of cases (n = 59) and the automatic imaging feature extraction not being implemented. In addition, we do not know whether re-classification by an expert neuropathologist has been performed, however, similar to other papers (2017-2019 more than 250 articles) published using this dataset, we therefore believe our conclusions are not largely influence by such factors. Further study should be conducted by using a larger pool of oligodendroglial tumours patients with the use quantitative image analysis tools being required [29]. The newly emerged field of radiogenomics allows specific MR imaging phenotypes to be linked with gene expression profiles. Further work is needed to better define the relationships identified in our study and to explore additional relationships (SVZ involvement and CD133, SVZ involvement and CD44, SVZ involvement and MMPs).

Conclusions
In conclusion, this study presents that SVZ involvement (−) and absence of seizure history may therefore be used to facilitate the prediction of CIC mutation status. Patients with SVZ-had a longer median overall survival months than patients with SVZ+.This work represents a practical application of imaging findings for personalized medicine. External validation of this model in other cohorts of patients is needed.