Synopsis

Keywords: Tumors, Diffusion/other diffusion imaging techniques, Meningiomas; Diffusion-weighted MRI

An accurate assessment of the World Health Organization grade is vital in meningiomas. While many studies have investigated the usefulness of conventional DWI and DTI for noninvasive grading intracranial meningiomas, there are neither any studies comparing three advanced diffusion model including DKI, MAP and NODDI with DTI for predicting meningioma grade. Thus, it is vital to evaluate whether these advanced models derived from diffusion spectrum imaging can also be beneficial in grading meningiomas. Our results suggested that whole tumour histogram analyses of the diffusion metrics from multiple diffusion models are promising methods in grading meningiomas.

Introduction

Meningiomas are the most frequent primary intracranial tumour (38.3%) and are classified into 3 grades according to the World Health Organization Classification (WHO)^{1, 2}. The grading of meningiomas has clinical significance for determining a treatment strategy and assessing prognosis. Conventional diffusion tensor imaging (DTI) are currently the most used techniques in clinic and presume a Gaussian diffusion contribution of water molecular within tumours. More advanced diffusion models with high b values, such as Diffusional kurtosis imaging (DKI), have been shown that outperform than conventional DWI and DTI in grading meningiomas^{3, 4}. Recently, mean apparent propagator (MAP) ⁵.and neurite orientation dispersion and density imaging (NODDI) ⁶, may reflect the diffusion behavior of water molecules within brain tumours more accurately. Nevertheless, to our knowledge, there are neither any studies on the role of NODDI or MAP in meningioma grading nor any reports comparing DTI, DKI, MAP, and NODDI for predicting meningioma grade. Furthermore, these advanced diffusion models can be simultaneously derived from diffusion spectrum imaging (DSI), which is a method freely reconstructed by using multiple gradient directions and high b values of entire q-space to quantitatively estimate diffusion behaviors of water molecules. The purpose of this study was to evaluate and compare the diagnostic accuracy of DTI, DKI, MAP, and NODDI-based diffusion parameters combined with histogram analysis in grading meningiomas, and to assess the correlations between the diffusion metrics and Ki-67 index.

Methods

In this study, 122 consecutive patients with histologically proved meningiomas were enrolled. MRI examinations were performed with a 3.0 T MRI unit (Philips Ingenia, Best, Netherlands). The histogram features of multiple diffusion metrics obtained from diffusion tensor imaging (DTI), diffusion kurtosis imaging (DKI), mean apparent propagator (MAP), and neurite orientation dispersion and density imaging (NODDI) in the solid component of tumour were analyzed. All values were compared between high-grade meningiomas (HGMs) and low-grade meningiomas (LGMs) with Man-Whitney U test. Logistic regression analysis was applied to predict the grade. The correlation between diffusion metrics and Ki-67 index was analyzed.

Results

The DKI_AK maximum, DKI_AK range, MAP_RTPP maximum, MAP_RTPP range, NODDI_ICVF range, and NODDI_ICVF maximum values were lower (P<0.0001), while the DTI_MD minimum values were higher in the LGMs than those in the HGMs (P<0.001) (Table 1 and Fig 1). Among the DTI, DKI, MAP, NODDI and combined diffusion models, no significant differences were found in the areas under the receiver operating characteristic curve (AUCs) for grading meningiomas (AUCs, 0.75, 0.75, 0.80, 0.79, and 0.86, respectively; all corrected P>0.05) (Table 2 and Fig 2). Significant positive correlations were found between the Ki-67 index and the DKI, MAP and NODDI metrics (r= 0.26-0.34, all P<0.05) (Fig 3).

Discussion

An accurate and noninvasive evaluation of the WHO grade is particularly important for patients with meningiomas. In this study, with a relatively large sample size, our results demonstrated that the whole-tumour histogram features of any diffusion models (DTI, DKI, MAP, NODDI, and the combined diffusion models) were helpful in grading intracranial meningiomas (AUC, 0.75, 0.75, 0.8, 0.79, and 0.86, respectively). Our results also showed that the increased benefit of upgrading the imaging protocol for advanced diffusion model was small, meaning that DTI had similar diagnostic performance compared with the advanced diffusion methods in the prediction of meningiomas grade. In addition, weak correlations were revealed between Ki-67 index and almost diffusion metric of DKI, MAP and NODDI metrics.

Conclusions

Whole tumour histogram analyses of the multiple diffusion metrics from four diffusion models are promising methods in grading meningiomas. The DTI model has similar diagnostic performance compared with advanced diffusion models.

Acknowledgements

No acknowledgement found.