Intraoperative dynamic susceptibility contrast weighted magnetic resonance imaging (iDSC-MRI) — Technical considerations and feasibility

Abstract

DSC-MRI was applied intraoperatively during brain tumor removal. Immediately after presumed complete tumor resection an MRI including a dynamic susceptibility contrast T2⁎-weighted EPI sequence was performed in 6 patients while the skull was still open using a flexible two-channel coil system at an intraoperative 1.5-Tesla MR scanner. After an initial baseline period of this iDSC-MRI sequence a bolus of contrast agent was administered intravenously. Maps of relative regional blood flow (rCBF), blood volume (rCBV) and the mean transit time (MTT) were calculated. These maps were compared to preoperatively acquired DSC-MRI data. The extent of the resection was compared with the postoperative MRI performed 24 h after the operation. In five patients complete tumor removal was already achieved at the time of iDSC-MRI and no areas of elevated perfusion values adjacent to the resection cavity were found. Complete removal was again documented on the postoperatively performed MRI. In one case there was residual tumor that showed both contrast enhancement and identical perfusion ratios as in the preoperatively acquired data. Removal of the remaining tumor was performed. iDSC-MRI is technically feasible as there are no significant susceptibility artifacts. DSC-MRI has been used to distinguish different tumor entities preoperatively and recurrent disease from radiation necrosis. Despite brain shift and thus invalidated preoperative image data or contrast leakage caused by intraoperative manipulation, iDCS-MRI furthermore reliably detects residual tumor intraoperatively at a timepoint where further resection is still possible and thus enables the neurosurgeon to complete the resection during the same procedure.

Introduction

In surgical neuro-oncology conventional MRI is the gold standard for preoperative imaging. Using this technique a lesion can be can nicely delineated and, according to specific imaging characteristics, the tumor entity diagnosed. However, physiologic information is lacking. PET provides these data (Subramanyam et al., 1978, Phelps et al., 1979, Hanson et al., 1991) but is limited in terms of availability. More recently, it has become possible to acquire physiologic data on a wider scale by using either dynamic contrast-enhanced T1-weighted perfusion MRI studies (DCE-MRI; Brix et al., 1991, Bullock et al., 1991, Worthington et al., 1991, Donahue et al., 1995, Parker et al., 1997, Li et al., 2000, Ludemann et al., 2001) or dynamic susceptibility contrast-weighted perfusion studies (DSC-MRI; Villringer et al., 1988, Belliveau et al., 1990, Rosen et al., 1990, Edelman et al., 1990, Sugahara et al., 1998, Sorensen et al., 1999a, Sorensen et al., 1999b, Provenzale et al., 2002, Conturo et al., 2005, Ostergaard, 2005). It has previously been shown that to a certain extent these modalities are interchangeable (Haroon et al., 2007, Ludemann et al., in press). DSC-MRI enables calculation of regional maps for relative blood volume and flow by administering conventional MR contrast agents while T2⁎-weighted images are being acquired. In these areas of blood-brain barrier breakdown DSC-MRI can also define distinct zones with increased cerebral blood flow and volume corresponding to neovascularization and active metabolism within the tumor. This helps to distinguish high-grade from low-grade tumors as well as to create characteristic maps for lesions such as tumefactive lesions and lymphoma (Knopp et al., 1999, Law et al., 2002, Provenzale et al., 2002, Hartmann et al., 2003, Lev et al., 2004, Boxerman et al., 2006). This sequence can easily be integrated into a routine protocol since it only takes 1 min 20 s. As of now this technique has only been used preoperatively for differential diagnosis and both postoperatively and during follow-up to distinguish recurrent disease from changes caused by radiation (Sugahara et al., 2000, Covarrubias et al., 2004). Furthermore, these data can be integrated into intraoperative computer-assisted guidance systems (neuronavigation).

However, in the presence of intraoperative brain deformation, so-called brain shift, these preoperative acquired data are no longer valid and the amount of brain shift is unpredictable (Nabavi et al., 2001). To overcome this, intraoperative imaging was introduced to detect residual tumor during the resection while the patient is still under anesthesia and the skull is still open. This enables the surgeon to complete the resection during the same procedure. The scanner employed intraoperatively in this study is a diagnostic standard clinical 1.5-Tesla scanner. Thus, the quality of the pre-, post-, and intraoperative images is comparable. However; intraoperative imaging poses a higher challenge with regards to possible image distortion caused by susceptibility of intracranial air or blood in the resection cavity, restricted choice of MR head coils, and time constraints. These factors preclude the application of highly susceptible or time-consuming sequences. Total resection of the enhancing tumor in high-grade gliomas has been demonstrated to correlate significantly with long-term survival (Stummer et al., 2008); therefore, adding physiologic information as well as precise definition of tumor location is crucial in surgical decision making.

We therefore assessed the feasibility of iDSC-MRI in an intraoperative setting in terms of possible technical drawbacks caused by the susceptibility of air or an air-water level in the resection cavity and the reliability of the results as compared to preoperatively acquired information.