gms | German Medical Science

Neuronavigation is a well established computer assisted technique in neurosurgical routine and is nowadays wide spread in most departments. The conventional technique uses computer tomography (CT) or magnetic resonance imaging (MRI) as basic visualization for surgical planning and intra-operative image guidance in navigation. Different additional image modalities like functional MRI, diffusion weighted imaging or cerebral angiography were integrated into navigated surgical procedures to enhance the visualized information in different indications for surgery [1], [2], [3], [4]. Since the early eighties positron emission tomography (PET) is available to visualize metabolic activity within vital tissue. Fluoro-desoxy-glucose (FDG) PET is used frequently to visualize glucose dependent energetic metabolic state in the brain. It has been hypothesized that increased rate of glucose metabolism within brain tumors is correlating with its grade of malignancy [5], [6], [7], [8].

Gliomas are the most common brain tumors. The histopathological grading defines the prognostic interference and leads to therapeutic decisions. Low-grade gliomas tend to progress to a higher grade of malignancy over time [9]. This course of transformation occurs due to focal mutations [10]. It is well known, that gliomas may represent a heterogeneous histological pattern. In open surgery of suspected low-grade gliomas the histopathological grading depends on the area, where the tissue specimen is taken. Since only a small part of tumor tissue is used for neuropathological investigation, there remains the risk that a more malignant part of the tumor which defines the final diagnosis is missed. In this context, we ae addressing two questions: How can we visualize malignant transformation in suspected low-grade gliomas? And, how can we localize intra-operatively the area of transformation?

The correlation of glioma grading with increased metabolic rate in FDG-PET has been discussed and neuronavigation is able to localize predefined regions of interest in brain tissue during surgery. Thus, we designed the following study for integration of FDG PET imaging into neuronavigation. The correlation of MRI contrast enhancement as well as increased glucose uptake in FDG PET imaging in patients with suspected low grade gliomas by the means of PET navigated biopsies in open glioma surgery is investigated.

We present 22 patients with suspected low-grade glioma according to common MRI criteria, which are hypointense lesions in T1 and appeared hyperintense in T2 and fluid-attenuated inversion recovery (FLAIR) images, with none or marginal gadolinium enhancement (0.2 mmol gadolinium/kg BW; 1.5 T Scanner). In all patients an open tumor resection was performed.

The cohort of patients represents 10 males and 12 females with a mean age of 38 years (range 18-68). All patients presented with a Karnovsky Performance Index of 100% before operation. Three patients had a previous resection (before 43-93 month) of an astrocytoma grade II according to world health organization (WHO) calssification. None received any radiotherapy or cytotoxic agents. Eight patients visualized only slight contrast enhancement in MR imaging.

All patients underwent FDG-PET and received 370 Mbq [¹⁸F]-2-fluoro-2-deoxyglucose i.v. (Forschungszentrum Karlsruhe, Germany). Following 60 min a 2-dimensional statical PET scan was performed using the ECAT-EXACT 47 PET Scanner (Siemens, CTI, Knoxville, TN, USA). Attenuation correction was performed using ⁶⁸Ge/⁶⁸Ga rod source for 10 min. For reconstruction by filtered back projection a Hann filter had been used (cutoff 0.4). None of the patients had a seizure within 7 days previous to PET investigation. The PET data were imported in the navigation image format (Figure 1 [Fig. 1]).

For navigation an anatomical high resolution, gadolinium enhanced, three-dimensional fast field echo (3D FFE) MR sequence was performed (1.5 T Phillips Gyroscan ACS NT, Philips, Best Netherlands or 1.5 T Signa, General Electric, Milwaukee, USA) following application of six fiducial markers to the scalp. Axial reconstructed slices (1mm) were fused with the PET data set using the Brainlab iPlan V1.0 software (Figure 1 [Fig. 1]). The region of interest was defined as increased glucose uptake within the tumor compared to white matter tissue as well as areas with contrast enhancement in MR imaging. A contour segmentation of the regions was performed manually.

All patients underwent craniotomy for tumor resection. A biopsy forceps (Decker Rongeurs, Codman, Raynham, MA, USA), provided with a reference frame of reflecting markers was registered to the optical neuronavigation (Brainlab VectorVision², Heimstätten, Germany) with Cranial V1.0 software. Following dura opening and before tumor removal, navigated biopsies from the regions of interest were taken. All tissue specimens were blinded to clinical data and reviewed by CH. Grading had been performed according to WHO criteria [10].

Histopathlogical diagnosis of all patients revealed a low grade glioma in 10 patients. In this group one pilocytic astrocytoma, six astrocytomas and three oligodendrogliomas WHO grade II were diagnosed. 12 patients showed histological criteria of high grade gliomas. Thereof, nine specimens were defined as anaplastic astrocytoma, one as anaplastic oligodendroglioma (WHO III°) and two specimens revealed a glioblastoma multiforme (WHO IV°).

MR imaging visualized eight lesions with slight contrast enhancement (Table 1 [Tab. 1]). In these patients the histological diagnosis was defined of a higher grade of malignancy (WHO III° & IV°) in four patients (50%). No contrast enhancement shown in 14 patients correlated in six specimen (42.6%) with the diagnosis of low grade gliomas (WHO I° & II°).

FDG-PET showed an increased glucose metabolism in nine patients. In this group of patients the navigated biopsy leads to a histological diagnosis of anaplastic gliomas in 5 patients (55.6%). Low uptake of glucose in PET imaging could be visualized in 13 patients, which correlated with lower grade of malignancy in 6 patients (46.2%, Table 2 [Tab. 2]). However, specimens with raised glucose did show higher cellular density in histological sections independent of grade.

Correlating MR with FDG PET imaging, in 14 patients with no contrast enhancing lesion in the preoperative MRI scan five had a focally increased glucose uptake in FDG-PET. No contrast enhancement in MRI and raised glucose uptake did lead in only three patients to the diagnosis of an anaplastic glioma (Table 3 [Tab. 3]).

In our study, 22 patients with no or slight contrast enhancing lesions in MRI with suspicion of low grade glioma [11], [12] underwent PET-navigated tumor resection. In our study grade of malignancy in potential low grade gliomas could be predicted by contrast enhancement in MRI in 50% of enhancing lesions. PET imaging sowed a slightly higher chance of correct prediction, Here, glucose uptake was coherent with high grade gliomas in 55.6%. In three of five patients with no contrast enhancement and increased glucose uptake an anaplastic glioma was proven. Compared to other studies we had a relatively high percentage (57.1%) of anaplastic gliomas without contrast enhancement in MRI [13], [14], [15]. However, the population investigated in our study is limited, but the distribution is representative concerning age, symptoms, histology, localization and sex distribution [9], [13], [14], [15].

Image fusion of FDG-PET and MRI within neuronavigation was save and feasible in open tumor resection of gliomas according to the authors opinion. Possible limitations of accuracy within the procedure might result especially from the fusion procedure. The software does not enable to give an exact value of fusion accuracy. Due to low anatomical resolution of PET imaging the fusion could not be performed automatically by the provided software. Anatomical landmarks need to be carefully verified for correct image fusion. However, manual fusion could be achieved with good results using parallel visualization of the fusing data sets in all three basic spatial orientations (transverse, sagittal, coronal). Intra-operatively, further attention is needed to check for brain-shift related inaccuracy of the biopsy. Thus, tissues samples of the related regions of interest were extracted immediately after dura opening, when brain shift is rather irrelevant.

Previously, FDG-PET was integrated into navigation to provide functional data for locating speech activation by subtracting the rate of glucose utilization in alternating speech-developing and silent periods. In this investigation the baseline FDG-PET was not introduced for tumor analysis although gliomas of WHO grade II and III had been investigated [16]. Moreover, image fusion of [¹¹C]-methionine-PET with MRI was used with neuronavigation to achieve a more complete tumor resection in a case of anaplastic oligodendroglioma [17]. All studies were in line with our opinion that technically no significant problems were noted in the fusion and integration to navigation process of PET image modality.

Target setting for biopsy using FDG-PET had been evaluated previously in stereotactic procedures [6], [18], [19], [20], [21]. Here, in FDG-PET guided stereotactic series of non contrast enhancing lesions the rate of anaplastic tumors was described to be only 27% [19]. This might be due to the fact of very small specimens obtained by stereotactic procedures, taking into concern that the neuropathologist underestimated the grade, if only a needle biopsy is available [22], [23]. In our series we used the navigated biopsy forceps obtaining tissue samples of about 2x6 mm. In consistency to our investigation stereotactic studies show a significant correlation of glucose and cell density [6], [18]. However, this is limited in terms of anaplasia [6].

Francaville and coworkers published 12 patients with focal hypermetabolic changes as a hint for malignant transformation [24]. FDG-PET was accomplished if patients with previously diagnosed low-grade gliomas had signs of tumor progression. Six of twelve tumors showed histological progression in grading compared to the time point of biopsy. However, 50 percent of the lesions did not receive a second biopsy and progression was assumed on MRI and PET scan changes.

In our opinion FDG-PET may be slightly superior to predict histological grading. However we are not able to prove significant better sensitivity of PET imaging with our study. Nevertheless, pathological evaluation remains by far the gold standard and surgical biopsies will not be replaced by imaging modalities so far. Further investigations are needed to quantify accuracy of image fusion and evaluate other imaging modalities like fluoro-ethyl-thyrosin (FET)-PET and MR spectroscopy.

PET integration into neuronavigation is a helpful tool to localize metabolic active regions of interest to obtain representative histological specimens. Compared to MRI criteria increased glucose uptake in PET imaging did not significantly increase the rate of prediction for histological grading. Neuronavigation is an excellent tool to prove modern image modalities for their clinical significance.