MR imaging is the method of choice for diagnosis of brain tumours but tumour delineation can be difficult and may be improved by adding metabolic information obtained from PET using O-(2-[18F]fluorethyl)-L-tyrosine (FET) ¹. The increased expression of amino acid transporters leads to an increased FET accumulation in tumour cells ². Another feature of gliomas is the increased cell turnover, which is indicated by a raised level of choline containing compounds (Cho) released from cell membranes and a reduced N-acetyl-aspartate (NAA) concentration as an indicator of neuronal loss. Such metabolites can be detected, in a 3D spatially resolved, manner by MR spectroscopic imaging (MRSI). A previous analysis of the areas of increased 18FET uptake and Cho/NAA ratio obtained from 2D spatially resolved MRSI, reported more than 75 % congruency in 15 patients ³. In the present hybrid PET MR study, the congruency of 18FET uptake and elevated Cho/NAA ratio is analysed in 3D in a series of 46 patients with brain tumours.

The study was approved by the local ethical committee. Informed consent was obtained from all patients before the measurement. Out of a series of 117 patients with brain tumours investigated by FET PET and MRSI in a hybrid PET MR scanner, 46 subjects (18 female, 28 male) of 48 ±15 years average age with suspected gliomas (WHO grade II-IV) were included in this analysis. 35 were excluded because of negative findings in FET PET and 36 were excluded because of unsuitable tumour location/bad data quality. Every patient was administered 3 Mbq/kg body weight of FET. While PET data were acquired in list mode over 50 min, a T1-weighted MPRAGE data set (with/without contrast agent), a T2+FLAIR data and a high resolution MR spectroscopic image data set with full brain coverage by means of a volumetric echo planar SI (EPSI) sequence with TE=17.6 ms and 16 min total acquisition time ⁴ were acquired simultaneously. All measurements were performed on a Siemens (Erlangen, Germany) 3T TIM TRIO equipped with the Siemens BrainPET insert inside the 3T magnet. 3D volumes representing the distribution of NAA and Cho were determined by employing the MIDAS software package ⁵. The volumes of suspected tissue were delineated based on FET uptake values greater than the 1.6 fold of the background uptake ¹ and by Cho/NAA-ratios outside the 95 % confidence interval of the distribution of normal tissue.

By means of the Dice coefficient, the overlap of the volumes of suspicious tissue depicted by the Cho/NAA-ratio and FET uptake, respectively, was evaluated at the spatial resolution of the MRSI data set (64x64x32 vxl, 5.6x5.6x10 mm³ each). The distances between the centres of gravity of the tumour volumes determined with each method were calculated.

The centres of gravity determined with both methods were located at (19±15) mm distance from each other. The overlap between the volumes defined by increased FET uptake and Cho/NAA-ratio averaged out at (30±23) % with tumour volumes of (14±15) cm³ and (39±28) cm³ in case of FET uptake and increased Cho/NAA-ratio, respectively (Fig. 1).Metabolically active tumour tissue delineated by increased FET uptake exhibits considerable differences compared with the area of elevated Cho/NAA-ratio measured by 3D spatially resolved MRSI. This finding is in contrast to previously reported results showing excellent overlap ³. Although the low extent of congruency is partially caused by the significantly different tumour volumes, to which Dice’s coefficient is sensitive, a similarity of FET uptake and Cho/NAA mapping was not found.

Although both FET uptake and Cho/NAA mapping are associated with proliferating tumour tissue, the two methods appear to reflect different metabolic properties of gliomas. The clinical relevance of these findings needs to be explored in future studies by comparing the histological and molecular parameters of the tumour tissue especially in mismatch areas.