Elsevier

European Journal of Radiology

Volume 95, October 2017, Pages 202-211
European Journal of Radiology

Research article
Histogram analysis of diffusion kurtosis imaging estimates for in vivo assessment of 2016 WHO glioma grades: A cross-sectional observational study

https://doi.org/10.1016/j.ejrad.2017.08.008Get rights and content

Abstract

Purpose

To assess the diagnostic performance of histogram analysis of diffusion kurtosis imaging (DKI) maps for in vivo assessment of the 2016 World Health Organization Classification of Tumors of the Central Nervous System (2016 CNS WHO) integrated glioma grades.

Materials and methods

Seventy-seven patients with histopathologically-confirmed glioma who provided written informed consent were retrospectively assessed between 01/2014 and 03/2017 from a prospective trial approved by the local institutional review board. Ten histogram parameters of mean kurtosis (MK) and mean diffusivity (MD) metrics from DKI were independently assessed by two blinded physicians from a volume of interest around the entire solid tumor. One-way ANOVA was used to compare MK and MD histogram parameter values between 2016 CNS WHO-based tumor grades. Receiver operating characteristic analysis was performed on MK and MD histogram parameters for significant results.

Results

The 25th, 50th, 75th, and 90th percentiles of MK and average MK showed significant differences between IDH1/2wild-type gliomas, IDH1/2mutated gliomas, and oligodendrogliomas with chromosome 1p/19q loss of heterozygosity and IDH1/2mutation (p < 0.001). The 50th, 75th, and 90th percentiles showed a slightly higher diagnostic performance (area under the curve (AUC) range; 0.868–0.991) than average MK (AUC range; 0.855–0.988) in classifying glioma according to the integrated approach of 2016 CNS WHO.

Conclusions

Histogram analysis of DKI can stratify gliomas according to the integrated approach of 2016 CNS WHO. The 50th (median), 75th, and the 90th percentiles showed the highest diagnostic performance. However, the average MK is also robust and feasible in routine clinical practice.

Introduction

Accurately identifying and grading glioma is important to determine clinical management and the prognosis [1], [2]. Histopathologic examination is essential for definitive grading and subsequent molecular stratification of glioma [3], [4]. However, robust and reliable non-invasive tumor grading is needed for follow-up of suspected low-grade glioma, for optimal care of patients who are not eligible for surgery, for those with an increased risk of post-biopsy complications, or for those being monitored for potential tumor recurrence [5]. The recently updated World Health Organization classification (revised 4th edition) of tumors of the central nervous system (2016 CNS WHO) combines both histopathological and molecular features into an “integrated diagnosis” [1]. The molecular stratification is crucial for estimating individual prognosis [6], [7], [8]. The relevant molecular characteristics comprise isocitrate-dehydrogenase (IDH) 1/2 mutation status and chromosome 1p/19q loss of heterozygosity (LOH). They are complemented by alpha-thalassemia/mental retardation syndrome X-linked (ATRX) expression resulting in an alternative lengthening of telomeres (ALT) phenotype [9]. The ATRX status itself confers a prognostic potential in diffuse gliomas [10].

The excess diffusional kurtosis — also known as the apparent excess kurtosis coefficient (AKC) — is a dimensionless metric that quantifies the degree of deviation from Gaussian diffusion behavior [11], [12]. Diffusional kurtosis imaging (DKI) is an extension of the diffusion tensor imaging (DTI) method. It allows for simultaneous direction-dependent estimates of the apparent diffusion coefficient (ADC) and AKC [13]. DKI reflects the heterogeneity and microstructural architecture of a biological tissue such a brain tissue because diffusion barriers alter the water diffusion probability distribution [11], [14]. DKI allows for discrimination between glioma grades according to the previous 2007 CNS WHO [2], [15], [16]. In the context of the integrated diagnostic approach of 2016 CNS WHO [1], DKI has shown potential for stratifying gliomas according to their molecular profile [17]. In addition, histogram analysis can improve the diagnostic performance of ADC to grade glioma according to previous 2007 CNS WHO versus average ADC values [18], [19]. However, the role of histogram analysis of DKI parametric maps for classifying glioma according to the recent 2016 CNS WHO has not yet been investigated.

Therefore, this study sought to assess the diagnostic performance of histogram analysis of DKI maps for in vivo assessment of 2016 CNS WHO-integrated glioma grades.

Section snippets

Study design and ethics

This study is a retrospective cross-sectional analysis of prospective data acquired in a single-center, non-randomized trial that was approved by the local institutional review board of our university hospital. The trial was conducted based on the principles of the International Conference on Harmonisation Good Clinical Practice guidelines and according to the 2013 revised version of the Declaration of Helsinki. All patients provided written informed consent for the imaging surveys and the

Results

The inter-rater agreement was excellent (ICC, 0.923; 95% CI, 0.901–0.955) for VOI delineation. Table 1, Table 2 show the integrated histopathological and molecular findings according to 2016 CNS WHO as well as the grouped findings based on their clinical outcome [6], [7], [10], [25]. In addition, Table 1 shows the histogram parameters of MK values with their standard deviations of the integrated diagnoses of 2016 CNS WHO. Table 2 shows the results for grouped IDHmut AS, IDHwt GBM, and OD

Discussion

The aim of this study was to assess the diagnostic performance of histogram analysis of DKI maps for in vivo assessment of 2016 CNS WHO integrated glioma grades. Corresponding to recent neuropathological research, the MK suggests that IDHmut AS2 and AS3, IDHwt AS3 and GBM, as well as OD2 and OD3 are similar entities in regards to tumor microstructure and heterogeneity. However, we found significant differences of MK and MD between gliomas classified as IDHmut AS, IDHwt GBM, and OD1p/19q-LOH

Conclusions

Histogram analysis of AKC maps may be used to stratify gliomas according to the integrated approach of 2016 CNS WHO. For the individual histogram parameters, the 50th (median), 75th, and the 90th percentiles of MK values showed the highest diagnostic performance. However, average MK values are also robust and feasible in routine clinical practice. MK confirms recent neuropathological research by showing similar tumor microstructure and heterogeneity within the groups of gliomas classified as IDH

Conflict of interest

All authors have no conflicts of interest regarding the present paper.

Acknowledgements

We thank Robert Grimm from Siemens (Erlangen, Germany) for technical support in image post-processing. JS was supported by the Else-Übelmesser Foundation (grant no. 30.19845). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References (34)

  • M. Abedalthagafi et al.

    The alternative lengthening of telomere phenotype is significantly associated with loss of ATRX expression in high-grade pediatric and adult astrocytomas: a multi-institutional study of 214 astrocytomas

    Mod. Pathol.

    (2013)
  • D.N. Louis et al.

    The 2016 world health organization classification of tumors of the central nervous system: a summary

    Acta Neuropathol.

    (2016)
  • S. van Cauter et al.

    Gliomas: diffusion kurtosis MR imaging in grading

    Radiology

    (2012)
  • J.N. Scott et al.

    How often are nonenhancing supratentorial gliomas malignant? A population study

    Neurology

    (2002)
  • M. Watanabe et al.

    Magnetic resonance imaging and histopathology of cerebral gliomas

    Neuroradiology

    (1992)
  • A.V. Kulkarni et al.

    Incidence of silent hemorrhage and delayed deterioration after stereotactic brain biopsy

    J. Neurosurg.

    (1998)
  • D.E. Reuss et al.

    Adult IDH wild type astrocytomas biologically and clinically resolve into other tumor entities

    Acta Neuropathol.

    (2015)
  • D.E. Reuss et al.

    IDH mutant diffuse and anaplastic astrocytomas have similar age at presentation and little difference in survival: a grading problem for WHO

    Acta Neuropathol.

    (2015)
  • F. Sahm et al.

    Farewell to oligoastrocytoma: in situ molecular genetics favor classification as either oligodendroglioma or astrocytoma

    Acta Neuropathol.

    (2014)
  • M. Pekmezci et al.

    Adult infiltrating gliomas with WHO integrated diagnosis: additional prognostic roles of ATRX and TERT

    Acta Neuropathol.

    (2017)
  • J.H. Jensen et al.

    MRI quantification of non-Gaussian water diffusion by kurtosis analysis

    NMR Biomed.

    (2010)
  • J.H. Jensen et al.

    Diffusional kurtosis imaging: the quantification of non-gaussian water diffusion by means of magnetic resonance imaging

    Magn. Reson. Med.

    (2005)
  • D.H.J. Poot et al.

    Optimal experimental design for diffusion kurtosis imaging

    IEEE Trans. Med. Imaging

    (2010)
  • H. Lu et al.

    Three-dimensional characterization of non-gaussian water diffusion in humans using diffusion kurtosis imaging

    NMR Biomed.

    (2006)
  • P. Raab et al.

    Cerebral gliomas: diffusional kurtosis imaging analysis of microstructural differences

    Radiology

    (2010)
  • R. Jiang et al.

    Diffusion kurtosis imaging can efficiently assess the glioma grade and cellular proliferation

    Oncotarget

    (2015)
  • J.-M. Hempel et al.

    In vivo molecular profiling of human glioma using diffusion kurtosis imaging

    J. Neurooncol.

    (2017)
  • Cited by (27)

    • Texture analysis based on ADC maps and T2-FLAIR images for the assessment of the severity and prognosis of ischaemic stroke

      2020, Clinical Imaging
      Citation Excerpt :

      Kurtosis is the measure of the thickness or heaviness of the tails of a distribution, and quantiles represented how many data points of a given dataset fall within a certain percentage range. Previous studies have shown that these histogram parameters qualitatively and quantitatively assess tumour heterogeneity by analysing the distribution of voxel-level grey levels in images [23,24]. Only a few studies have used histogram parameters to evaluate stroke.

    • Performance of Diffusion Kurtosis Imaging Versus Diffusion Tensor Imaging in Discriminating Between Benign Tissue, Low and High Gleason Grade Prostate Cancer

      2019, Academic Radiology
      Citation Excerpt :

      Since cancer tissue could include normal tissue, calculating the mean of diffusion parameters in ROIs may not be the best quantification of diffusion metrics in PCa. For this reason, we performed histogram analysis (37), by measuring, in addition to the mean (m) and the standard deviation (SD), the 10th (p10), 25th (p25), and 90th (p90) percentile of the distribution, ie, the histogram, of diffusion parameters measured in volumetric ROIs. The percentile of a distribution indicates the value below which a given percentage of observations falls: in our case, the 10th, 25th, and 90th percentiles represent the values below which the 10%, 25%, and 90% of the measures lie, as indicated in Figure 2.

    • Diffusion kurtosis imaging histogram parameter metrics predicting survival in integrated molecular subtypes of diffuse glioma: An observational cohort study

      2019, European Journal of Radiology
      Citation Excerpt :

      Regarding the different DKI parameters we found that MK performed better than MD in predicting survival on preoperative images. These findings so far correspond to previous reports [26], where MK histogram parameters performed better than those of MD in non-invasively predicting the integrated molecular glioma profile. We did not find corresponding values in the literature.

    • Quantitative analysis of neurite orientation dispersion and density imaging in grading gliomas and detecting IDH-1 gene mutation status

      2018, NeuroImage: Clinical
      Citation Excerpt :

      Several studies (White et al., n.d.; Kothari et al., 2013; McDonald et al., 2016) have focused on tumour response monitoring and HGG margin delineation, but RSI has rarely been applied for grading gliomas. Furthermore, recent studies on the identification of IDH-1 mutation status by DTI (Xiong et al., 2016) and DKI (Hempel et al., 2017b) have shown that IDH-1-mutated gliomas have lower maximum FA and mean MK values. The authors hypothesized that lower FA and MK values represent more homogeneous cellular architecture and lower cell density (Beppu et al., 2005; Kinoshita et al., 2008; Popov et al., 2013).

    View all citing articles on Scopus
    View full text