The aim of this study was to investigate the possible correlation between the 1H MRS mobile lipid signal, necrosis and lipid droplets in C6 rat glioma. First, the occurrence of necrosis and lipid droplets was determined during tumor development, by a histological analysis performed on 34 rats. Neither necrosis nor lipid droplets were observed before 18 days post-implantation. At later stages of development, both necrosis and lipid droplets were apparent, the lipid droplets being mainly located within the necrotic areas. Using a second group of eight rats, a temporal correlation was evidenced between mobile lipid signal detected by in vivo single-voxel one- (136 ms echo time) and two-dimensional J-resolved 1H MR spectroscopy, and the presence of necrosis and lipid droplets on the histological sections obtained from the brains of the same rats. Finally, spatial distribution of the mobile lipid signal was analyzed by chemical-shift imaging performed on a third group of eight animals, at the end of the tumor growth. The spectroscopic image corresponding to the resonance of mobile lipids had its maximum intensity in the center of the tumor where necrotic regions were observed on the histological sections. These necrotic areas contained large amounts of lipid droplets. All these results suggest that mobile lipids detected in vivo by 1H MRS (136 ms echo time) in C6 rat brain glioma arise mainly from lipid droplets located in necrosis.

In C6 rat brain glioma, we have investigated the relation between hypoxia and the presence of lipid droplets in the cytoplasm of viable cells adjacent to necrosis. For this purpose, rats were stereotaxically implanted with C6 cells. Experiments were carried out by the end of the tumour development. A multifluorescence staining protocol combined with digital image analysis was used to quantitatively study the spatial distribution of hypoxic cells (pimonidazole), blood perfusion (Hoechst 33342), total vascular bed (collagen type IV) and lipid droplets (Red Oil) in single frozen sections. All tumours (n ¼ 6) showed necrosis, pimonidazole binding and lipid droplets. Pimonidazole binding occurred at a mean distance of 114 mm from perfused vessels mainly around necrosis. Lipid droplets were principally located in the necrotic tissue. Some smaller droplets were also observed in part of the pimonidazole-binding cells surrounding necrosis. Hence, lipid droplets appeared only in hypoxic cells adjacent to necrosis, at an approximate distance of 181 mm from perfused vessels. In conclusion, our results show that severe hypoxic cells accumulated small lipid droplets. However, a 100% colocalisation of hypoxia and lipid droplets does not exist. Thus, lipid droplets cannot be considered as a surrogate marker of hypoxia, but rather of severe, prenecrotic hypoxia.

Two-dimensional J-resolved spectroscopy may be used to separate resonances which overlap in 1D NMR spectra. Coupled with spectroscopic imaging (SI), it would give unequivocal information on the distribution of such resonances. Multi-echo acquisition decreases the minimum experimental time of such 4D experiments. The water peak may be used for phase and chemical-shift reference. This study aimed to demonstrate the feasibility of J-resolved SI based on a multi-echo sequence and without water suppression, and its ability to separate the peaks for lactate and mobile lipid in a rat glioma. Experiments were performed on rat brain, without water suppression, at 7 T. The water signal was used for correcting the phase of the echoes. A FOCSY-like acquisition was used to collect the first part of the echoes at short echo times. Two different data processing methods were tested to overcome the problem of contaminations of metabolite signals by the intense water signal. Maps of N-acetylaspartate, choline, creatine, lactate and mobile lipids were obtained in vivo on a rat glioma in 70 min. The in-plane resolution was 2 mm2. The 2D spatially resolved, 2D J-resolved spectra enabled the separate mapping of lactate and mobile lipids.