Diagnosis of glioma recurrence using multiparametric dynamic 18F-fluoroethyl-tyrosine PET-MRI

Highlights

• The study introduces combined dynamic 18F-FET-PET/MRI for the diagnosis of recurrence in gliomas.

• We demonstrate the feasibility of the approach and evaluate the accuracy of the different modalities in the hybrid imaging protocol.

• Multiparametric analysis is shown to be particularly useful at confirming foci of recurrence.

Abstract

Objectives

To investigate the value of combined 18F-fluorethyltyrosine-(FET)-PET/MRI for differentiation between recurrence and treatment-related changes in glioma patients.

Methods

63 lesions suggestive of recurrence in 47 glioma patients were retrospectively identified. All patients had a dynamic FET scan, as well as morphologic MRI, PWI and DWI on a hybrid PET/MRI scanner. Lesions suggestive of recurrence were marked. ROC analysis was performed univariately and on parameter combination.

Results

50 lesions were classified as recurrence, 13 as radiation necrosis. Diagnosis was based on histology in 23 and follow-up imaging in 40 cases. Sensitivities and specificities for static PET were 80 and 85%, 66% and 77% for PWI, 62 and 77% for DWI and 64 and 79% for PET time-to-peak. AUC was 0.86 (p < 0.001) for static PET, 0.73 (p = 0.013) for PWI, 0.70 (p = 0.030) for DWI and 0.73 (p < 0.001) for dynamic PET. Multiparametric analysis resulted in an AUC of 0.89, notably yielding sensitivity of 76% vs. 56% for PET alone at 100% specificity.

Conclusion

Simultaneous dynamic FET-PET/MRI was reliably feasible for imaging of recurrent glioma. While all modalities were able to discriminate between recurrence and treatment-related changes, multiparametric analysis added value especially when high specificity was demanded.

Introduction

Maximum safe resection followed by external beam radiation combined with concomitant and adjuvant radiochemotherapy is considered the first-line therapy for high-grade gliomas [1]. According to the RANO response criteria, follow-up MRI is scheduled 4–6 weeks after irradiation is completed [2]. Many patients show signs of progression (contrast enhancement and/or T2 hyperintensities) at this early point in time, and the number increases during the following months [3]. However, these imaging findings do not prove tumour progression, as therapy-related changes such as necrosis and inflammation may be indistinguishable on standard MRI, a phenomenon called pseudoprogression. While these early reactive changes normally subside spontaneously, a more severe form which usually occurs at a later time point, namely radiation necrosis, is more reluctant to regress and more likely to evoke neurological symptoms. Since pseudoprogression and radiation necrosis show considerable overlap and share common pathophysiological features, we will refer to both phenomena as therapy- or treatment-related changes.

Differentiation between therapy-related changes and true progression is of high clinical importance. Early recognition of progression offers the possibility for therapeutic intervention, such as re-resection or re-irradiation. Consequently, several imaging methods have been investigated for this purpose, and are now finding their way into clinical practice. Different techniques aim at the determination of tissue perfusion, i. e. perfusion-weighted MRI (PWI). Commonly used, dynamic susceptibility contrast (DSC) determines blood supply and vascularization by contrast-induced T2/T2* signal loss, and several studies have shown that vital tumour tissue exhibits increased perfusion in DSC compared to treatment-related changes [4]. Dynamic contrast-enhanced T1-weighted imaging is another possibility of obtaining perfusion-related measures, but is more difficult to quantify and, according to current guidelines, is regarded more as a supplement to DSC [5].

Diffusion-weighted imaging (DWI) is a modality used for estimating water mobility, with higher values of the apparent diffusion coefficient (ADC) favouring post-treatment changes [6,7]. When multiple diffusion gradients are acquired, measures such as fractional anisotropy (FA) might be able to detect alterations of the fibre structure caused by tumour infiltration [8]. MR spectroscopy has been employed to detect abnormal concentrations of certain metabolites in the case of recurrence, although reliable clinical implementation is challenging [9,10]. On the other hand, amino acid PET with 11C-methionine and 18F-fluroethyltyrosine (FET) has gained wide acceptance especially in Europe and several Asian countries as an add-on technique for cases that cannot be adequately solved by MR imaging alone [11,12]. Additional information can be obtained by dynamic FET-PET imaging, with active tumour showing ‘faster’ tracer uptake with subsequent wash-out, further improving the already remarkable diagnostic accuracy of static PET [13,14].

With the advent of hybrid PET/MR imaging, both morphological and functional imaging can be performed in one examination, increasing comfort for the often severely ill patients, and optimizing spatial and temporal co-registration of the different modalities [15]. Although there is an increasing number of studies on multimodal advanced MR imaging [16], little is known about the possible benefit from combining PET and MRI in terms of increased sensitivity and specificity. Sogani et al. have examined a cohort of 32 low and high grade glioma patients with a hybrid PET/MRI protocol and demonstrated added value of the modalities, but did not use dynamic FET imaging [17]. In contrast, Jena et al. combined FDG and PET/MRI and could likewise show a benefit from multimodal analysis [18].

In this work, we investigate a cohort of glioma patients who had received a state-of-the-art combined dynamic FET-PET/MRI protocol and seek to determine the benefit of static and dynamic FET imaging, diffusion (DWI) and perfusion weighted MRI (PWI) and their multiparametric combination in gliomas.