Maximal safe resection is the cornerstone of treatment for low-grade and high-grade gliomas. In addition to high-resolution anatomic MRI studies that highlight tumor architecture, it is important to determine the relationship of the tumor to the eloquent cortical and subcortical areas to avoid introducing or exacerbating a neurologic deficit. The goal of this review was to highlight imaging modalities that provide functional information and can be integrated with intraoperative MRI navigation to maximize the extent of resection while preserving a patient’s neurologic function.
Key points
- •
Functional studies are key during preoperative glioma workup to appreciate the location of eloquent cortical and subcortical areas relative to the tumor.
- •
Functional cortical areas can be highlighted by using task-based or resting-state functional MRI, magnetoencephalography, or navigated transcranial magnetic stimulation.
- •
Diffusion tensor imaging–fiber-tracking algorithm allows mapping of subcortical white matter tracts, and can be used intraoperatively to determine whether shift has occurred.
- •
Additional metabolic and perfusion information can be incorporated to help delineate tumor extent, whereas intraoperative visualization of tumor tissue using 5-aminolevulinic acid can help maximize tumor resection.
- •
Combined use of several imaging modalities provides complementary information, ensuring maximal tumor resection while minimizing the risk of a new postoperative neurologic deficit.
Introduction
The goal of glioma surgery is maximal safe resection. Increasing extent of resection was shown to improve patient survival. Tumor visualization before surgery is essential to help assess feasibility of resection and for surgical planning. Preoperative MRI studies are key for understanding anatomic relationship between the tumor and the surrounding critical structures within the brain. This preoperative information is now routinely integrated into surgical workflow with the introduction of intraoperative navigation.
The major advantage of neuronavigation is that it allows the surgeon to optimize the location of the craniotomy and to minimize the size of the incision. It was not shown to improve in the extent of resection, however. Brain shift that occurs during surgery as a result of cerebrospinal fluid egress and tumor removal renders neuronavigation inaccurate at the time when it is needed the most. To mitigate the brain shift phenomenon, an intraoperative MRI (iMRI) scan can be performed to update the registration and to visualize the remaining tumor. Using iMRI as a surgical adjunct was shown to improve the extent of resection and to result in longer patient survival. A randomized controlled trial that enrolled 58 patients compared the added value of iMRI on extent of resection. This study showed that 96% of patients in the iMRI group had complete tumor resection versus 68% of patients in the regular neuronavigation group, and that this was not associated with an increased rate of neurologic deficits. As a result, iMRI is routinely used in many centers to help visualize remaining tumor and to ensure complete resection is achieved.
Maximizing the extent of glioma resection should not be pursued at the expense of neurologic function. It is important to ensure that no new neurologic deficits are introduced as a result of surgery and the functional status of the patient is preserved. A new postoperative neurologic deficit significantly reduces patient overall survival from a 2-year survival of 23% in patients without a new deficit to 8% and 0% in patients with a new postoperative motor or language deficit, respectively. Intraoperative electrical stimulation has been the gold standard for functional localization of cortical areas since introduction by Penfield and Boldrey in 1930s. It is helpful, however, to know the location of the tumor in relation to eloquent areas before taking the patient into the operating room, as it aids with preoperative surgical planning and helps to estimate the feasibility of complete resection. Many imaging modalities are now available to obtain functional information preoperatively, and many have been integrated with the iMRI navigation system. Moreover, in some instances, functional information can be updated at the time of the intraoperative scan. Here, we review these techniques, and discuss their application in conjunction with the iMRI.
Introduction
The goal of glioma surgery is maximal safe resection. Increasing extent of resection was shown to improve patient survival. Tumor visualization before surgery is essential to help assess feasibility of resection and for surgical planning. Preoperative MRI studies are key for understanding anatomic relationship between the tumor and the surrounding critical structures within the brain. This preoperative information is now routinely integrated into surgical workflow with the introduction of intraoperative navigation.
The major advantage of neuronavigation is that it allows the surgeon to optimize the location of the craniotomy and to minimize the size of the incision. It was not shown to improve in the extent of resection, however. Brain shift that occurs during surgery as a result of cerebrospinal fluid egress and tumor removal renders neuronavigation inaccurate at the time when it is needed the most. To mitigate the brain shift phenomenon, an intraoperative MRI (iMRI) scan can be performed to update the registration and to visualize the remaining tumor. Using iMRI as a surgical adjunct was shown to improve the extent of resection and to result in longer patient survival. A randomized controlled trial that enrolled 58 patients compared the added value of iMRI on extent of resection. This study showed that 96% of patients in the iMRI group had complete tumor resection versus 68% of patients in the regular neuronavigation group, and that this was not associated with an increased rate of neurologic deficits. As a result, iMRI is routinely used in many centers to help visualize remaining tumor and to ensure complete resection is achieved.
Maximizing the extent of glioma resection should not be pursued at the expense of neurologic function. It is important to ensure that no new neurologic deficits are introduced as a result of surgery and the functional status of the patient is preserved. A new postoperative neurologic deficit significantly reduces patient overall survival from a 2-year survival of 23% in patients without a new deficit to 8% and 0% in patients with a new postoperative motor or language deficit, respectively. Intraoperative electrical stimulation has been the gold standard for functional localization of cortical areas since introduction by Penfield and Boldrey in 1930s. It is helpful, however, to know the location of the tumor in relation to eloquent areas before taking the patient into the operating room, as it aids with preoperative surgical planning and helps to estimate the feasibility of complete resection. Many imaging modalities are now available to obtain functional information preoperatively, and many have been integrated with the iMRI navigation system. Moreover, in some instances, functional information can be updated at the time of the intraoperative scan. Here, we review these techniques, and discuss their application in conjunction with the iMRI.
Functional studies that can be performed preoperatively or intraoperatively
Functional MRI
Functional MRI (fMRI) is one of the most commonly used imaging studies that allows localization of motor, sensory, visual, and language cortex. The principle of fMRI is based on the physiologic phenomenon of neurovascular coupling originally described by Roy and Sherrington in 1890 that postulates that at the time of task execution there is increased blood flow to the area of cortical activation. The amount of blood delivered to that area is in relative excess to the amount of oxygen that is consumed as a result of increased metabolic activity. Furthermore, oxyhemoglobin and deoxyhemoglobin have different magnetic properties, and it is possible to detect the excess of oxyhemoglobin in the area of cortical activation. This is the underlying principle of fMRI that measures relative increase in blood-oxygen-level–dependent (BOLD) signal while the patient is performing a particular task. To detect this relative increase in metabolic cortical activity, the MRI scans have to be acquired quickly, and spatial resolution is sacrificed as a result. The areas of activation are subsequently overlaid onto the high-resolution structural MRI scan of the brain to help with preoperative planning.
The main advantage of fMRI is that no additional equipment is needed to perform functional mapping, which led to the popularity of this method. Task-based fMRI is accurate at identifying eloquent motor areas. The results of cortical activation obtained from patients performing a motor task have demonstrated good correlation with direct cortical stimulation with the specificity and sensitivity ranging from 95% to 100%. With respect to language localization, however, fMRI studies showed much more variability in accuracy with sensitivity and specificity ranges of 37% to 91% and 64% to 83%, respectively. Therefore, it is important to appreciate that language mapping using fMRI provides only a rough estimate of location of speech areas and should be confirmed with other functional methods, such as awake cortical mapping.
The main limitation of fMRI is lack of spatial resolution and low signal-to-noise ratio resulting in false-negatives. Furthermore, fMRI results can be affected by the presence of tumor. Altered vascularization pattern of neoplastic tissue and nearby areas could result in neurovascular uncoupling, thus resulting in false-negative areas on the scan. The presence of perilesional tissue edema, on the other hand, can result in areas of false-positive activations. In summary, although fMRI is an excellent and widely accessible modality of preoperative functional mapping, its results should be used with a degree of caution and supplemented with other functional modalities to ensure preservation of neurologic function.
Intraoperative functional MRI
The feasibility of acquisition of an intraoperative fMRI was investigated in 4 patients under general anesthetic. Passive stimulation of median and tibial nerves was accomplished with an electromagnetically shielded electrical wire. Phase-reversal potentials were used to identify central sulcus to verify the results of fMRI. Areas of activation of sensorimotor cortex obtained intraoperatively could then be used as a seed for diffusion tensor imaging (DTI) fiber-tracking (FT) protocols to improve its accuracy. This testing paradigm was successful in all 4 patients. The success of the procedure was affected, however, by the susceptibility artifacts from air/tissue boundaries, as well as from head fixation pins that interfered with data acquisition and image processing. Furthermore, acquisition of the intraoperative fMRI data is a time-consuming process and requires specialized equipment. Eloquent cortical areas can be more easily identified using intraoperative electrophysiological techniques, such as phase-reversal, or direct electrical stimulation, and subsequent shift of the cortex can be monitored directly by the surgeon. Therefore, the application of intraoperative fMRI for widespread use is limited at present.
Resting-state functional MRI
Recently, there has been an increased interest in using resting-state fMRI (rs-fMRI) for functional mapping. The physical principles of scan acquisition are the same as in task-based fMRI. The main difference is that the scans are acquired while the patient is at rest. The scan detects low-frequency (<0.1 Hz) fluctuations in BOLD signal, and computing algorithms are used to find cortical areas that are activated at the same time, indicating a functional relationship. It was noted that some of these networks involve cortical areas that are also activated during task-based fMRI suggesting that each network has a distinct function. There are several functional resting-state networks (RSNs) that were identified and include somatosensory, language, auditory, and visual networks.
Several studies used either task-based fMRI or intraoperative direct cortical stimulation to demonstrate functional correlation of RSNs. In one study of patients with low-grade gliomas undergoing surgical resection, areas of sensorimotor and language RSNs were compared with the results of intraoperative direct cortical stimulation. It was found that, on average, 96% and 84% of sensorimotor stimulation points were located within 10 and 5 mm, respectively, of the sensorimotor network map, whereas 92% and 70% of language stimulation points were found within 10 and 5 mm of the corresponding rs-fMRI language network. This further supports functional significance of rs-fMRI networks and its potential use in preoperative assessment of patients with tumor.
The main challenge with the use of rs-fMRI is computational processing of data. Identification of RSNs uses mathematical algorithms to detect areas of simultaneous activation, and the results obtained will vary based on the method used. At present, most commonly used protocols include seed-based analysis, whereby prior knowledge of functional neuroanatomy is required to select an appropriate “seed” for network tracing. Independent component analysis (ICA) is an alternative and user-independent algorithm that breaks down the 2-dimensional BOLD signal data matrix into several independent spatial components, where each component may represent a separate network map. It is not clear whether one method is superior over the other; although the seed-based method provides robust maps, the ICA method removes operator bias.
Intraoperative resting-state functional MRI
The main advantage of rs-fMRI is that it does not require patient participation, and, thus, functional mapping can be performed in children, uncooperative patients, and patients with a motor or speech deficit. Furthermore, RSNs were shown to persist in sleep and under propofol anesthesia. The latter finding suggests that rs-fMRI could be done intraoperatively using iMRI to obtain updated functional information. First experience of acquiring intraoperative rs-fMRI scan was done in 12 patients who underwent surgery for intra-axial lesions located near the sensorimotor cortex or corticospinal tract. Interestingly, patients with no or only slight postoperative deficit had significantly higher z -scores (quantification of intensity of activation) when compared with patients with severe postoperative deficits. Therefore, the results of this study suggested that intraoperative rs-fMRI results could be used to predict postoperative patient outcome.
There are some caveats to be overcome to be able to use rs-fMRI data intraoperatively, however. To prevent artifact from air/fluid interface, the dura and skin had to be partially closed and resection cavity filled with fluid to optimize image data quality. Furthermore, at present it takes hours to process rs-fMRI data, and as a consequence, the results of the intraoperative scan are not immediately available during tumor resection. Advances in image and rs-fMRI data processing are required to help make these data available during surgery.
Diffusion Tensor Imaging
Although the location of eloquent cortex is important, the location of subcortical white matter tracts relative to the tumor is crucial as well. Interrupting the descending corticospinal or subcortical language pathways can result in a neurologic deficit that is just as severe as with cortical damage. DTI-FT is the only imaging modality at present that allows the surgeon to visualize the location of subcortical white matter tracts. DTI is a T2-based MRI sequence that takes advantage of the fact that movement of water molecules in brain is determined by their local environment and display anisotropy, or directionality of movement. In an axon, the diffusion of water occurs mainly along the length of the nerve fiber and is relatively restricted in the direction perpendicular to the long axis of the axon. As a result, this preferential movement of water along the fiber measured as a diffusion tensor vector is used to extrapolate the location of subcortical tracts.
Initial processing of DTI data leads to a large amount of information on all subcortical white matter fibers. It is possible to isolate individual fiber bundles by defining the regions of interest where the intended white matter tracks are expected to travel in an area that is remote from tumor. Fiber tracts that are routinely isolated in this way include corticospinal tracts, as well as pathways that are important in language processing, such as arcuate fasciculus and inferior fronto-occipital fasciculus. DTI can provide information about tract location relative to the tumor and predict whether fibers run a normal course away from tumor tissue, are displaced by white matter edema, or are going through the tumor.
The accuracy of DTI-FT in determining the location of the corticospinal tract was demonstrated in several studies and was shown to have a good correlation with the intraoperative subcortical stimulation results with the specificity and sensitivity of 93% to 95%. Knowing the exact location of the corticospinal tracts in relation to the tumor helps plan the surgical approach and estimate the degree of resectability of the tumor.
Intraoperative diffusion tensor imaging
Preoperative DTI-FT is a great tool that helps to predict the location of white matter tracts relative to the tumor. Unfortunately, intraoperative brain shift affects the deeper white matter to a great extent, making neuronavigation inaccurate, and unlike with cortical mapping, there are no visual landmarks to help predict the amount of shift ( Fig. 1 ). Several studies used iMRI to determine the amount of tract shift that occurs as a result of tumor resection. A study of 37 patients who underwent glioma surgery compared the results of preoperative DTI-FT with the results of the intraoperative scan. The tracts shifted anywhere from −8 to +15 mm, and the direction of the shift could not be predicted. Prabhu and colleagues also looked at the amount of shift of the corticospinal tract by acquiring an updated intraoperative DTI scan. The shift of the corticospinal tracts ranged from 0 to 15 mm, with an average of 5.4 mm. Another study that investigated the use of bipolar stimulation to map the corticospinal tract in 7 patients, commented on the significant degree of white matter fiber shift that occurred intraoperatively as compared with preoperative imaging. Moreover, the results of intraoperative DTI scans affect surgical decisions. A study of 30 patients who underwent resection of dominant insular gliomas in the iMRI suite showed that even though residual tumor was identified in 26 patients, in 17 of them, repeat resection was abandoned based on the results of intraoperative tractography.
To summarize, there is compelling evidence that there is significant shift of subcortical white matter tracts near the end of tumor resection (see Fig. 1 ). Ironically, tractography information is typically most useful near the end of the resection, at the time when neuronavigation based on the preoperative scan is the least reliable. Intraoperative DTI scan helps to update tractography results during surgery near eloquent subcortical white matter pathways, to alert the surgeon to the proximity of the functional fibers, and thus helps to prevent injury and minimize the risk of a new postoperative neurologic deficit.
The main limitation of DTI-FT is that it is a computational model, and different mapping protocols will lead to slightly different results. This variability is independent of the tract microenvironment or presence of vasogenic edema. Standardization of protocols used to model the location of the white matter tracts would improve reliability of this method. Furthermore, the accuracy of DTI is subject to artifact at the points of crossing fibers, resulting in either false connections or premature truncation of the tract. Advanced white matter tracking techniques, such as diffusion spectrum imaging and high angular resolution diffusion imaging, may overcome some of these limitations and improve white matter tract resolution. Furthermore, using a higher field 3T iMRI while improving signal-to-noise ratio also amplifies the effects of imaging artifact generated by air/brain interface, which can interfere with data collection sufficiently to make it unusable. Careful preparation for the scan by filling up the resection cavity with fluid and partially closing the incision may help to minimize the artifact and improve image quality.
Magnetic Resonance Spectroscopy
Magnetic resonance spectroscopy (MRS) is an imaging technique that samples chemical composition of the tumor and the surrounding brain. The most common chemical markers used in clinical practice include choline-containing molecules as markers of membrane turnover, N-acetyl-aspartate (NAA) as a neuronal marker, creatinine that reflects energy status of tissue sample, lipids that mark necrosis and apoptosis, and lactate that is elevated in states of hypoperfusion and hypoxia. In tumor tissue, there is a disproportionate increase in choline relative to NAA or creatinine. Therefore, by determining the choline/NAA or choline/creatinine ratio can help better define the extent of tumor infiltration.
The results of MRS help better characterize the extent of glioma invasion of the surrounding brain. To correlate whether choline-to-NAA ratio, or index (CNI), corresponded to tumor infiltration based on 3-dimensional (3D) 1 H-MRS scans, tissue biopsies were obtained at different CNIs. It was determined that the CNI thresholds of 0.5, 1.0, and 1.5 could predict tumor infiltration with probabilities of 38%, 60%, and 79% in high-grade gliomas, and 16%, 39%, and 67% in low-grade gliomas. Moreover, metabolic margins of the tumor as determined from 3D-MRS scans extend beyond anatomic margins as visualized by conventional T1-weighted and T2-weighted MRI sequences, suggesting that it is important to incorporate metabolic information into surgical planning to maximize the extent of resection.
Transitioning MRS data in such a way that it could be easily used in surgical planning or for neuronavigation is difficult because in many centers sampling is done on individual voxels, or on a 2-dimensional static image. Attempts were made to integrate this information into neuronavigation by segmenting metabolic image of the tumor and matching it to the volumetrically obtained 3D-MRI scan. Furthermore, 3D-MRS images have a high diagnostic accuracy and thus could be used to delineate tumor margins, but require longer acquisition times and preferably an ultra-high field 3T MRI scanner for acquisition to improve signal-to-noise ratio. If obtained preoperatively, however, 3D-MRS data sets can be integrated into neuronavigation to be used continuously throughout tumor resection.
Intraoperative MRS
Unfortunately, preoperatively acquired MRS data that are integrated into standard neuronavigation are subject to the same pitfalls as with other preoperative imaging modalities and subject to brain shift. To account for shifting of tissue during resection, intraoperative MRS scans can be obtained. A recent study of 45 patients with low-grade and high-grade gliomas looked at the feasibility of intraoperative MRS and correlated presence of residual tumor using conventional MRI and histology. The investigators found good correlation between presence of residual T2-weighted signal and MRS spectra suggestive of tumor in all cases of low-grade gliomas. Interestingly, in high-grade gliomas, MRS sampling of the surrounding high-T2 signal showed a pattern highly suggestive of tumor infiltration in 8 of 12 patients, and not immediately apparent on visual inspection in 4 of those patients. Therefore, intraoperative use of MRS can help confirm presence of residual tumor and assist in decision making on whether to proceed with additional resection. Combined with recent evidence that resecting more than 53.21% of the high-T2 signal surrounding high-grade gliomas improves patient survival, this study suggests that we should be incorporating metabolic information into preoperative planning and navigation to maximize the extent of glioma resection.
The main limitation of MRS relates first of all to the ability to acquire the spectrum and delineate metabolite peaks above the background noise. Higher field magnet dramatically improves the signal-to-noise ratio, thus resulting in better spatial resolution. In centers already equipped with the iMRI suite, it would require an upgrade of the existing MR scanner, however, which is a major expense for a clinical center. Furthermore, metabolite ratios can be affected by peritumoral edema, by presence of inflammatory cell infiltrates around the tumor, or inflammation. Thus, additional imaging characteristics of the tissue, as well as clinical impression of the surgeon during resection, need to be taken together to make the decision on whether to proceed with further resection.
Perfusion Imaging
Contrast-enhanced MRI is fundamental to diagnosis of high-grade gliomas and helps to delineate tumor margin. During resection, however, the blood-brain barrier is disrupted at the site of surgery, resulting in contrast enhancement at the edge of the resection cavity. It is difficult to discern at this point whether the enhancement is due to the remaining tumor or secondary to surgical trauma.
Intraoperative MRI perfusion studies take advantage of the differences in perfusion dynamics of the tumor relative to surrounding uninvolved brain to help visualize the remainder of the neoplastic tissue. Dynamic susceptibility contrast (DSC) MRI is a T2*-weighted technique that allows calculation of the regional cerebral blood flow and volume. It was shown to reliably distinguish between residual tumor and changes introduced as a result of surgical resection.
Another perfusion technique that is used intraoperatively is dynamic contrast-enhanced T1-weighted perfusion. It looks at the rate of contrast uptake that is faster in neoplastic tissue when compared with other areas of blood-brain barrier disruption, such as changes introduced by surgery. Using perfusion imaging intraoperatively can help discriminate between tumor residual and tissue trauma as a result of surgery, and help ensure complete tumor resection.
The main limitation of this technique is that its application is limited to situations in which there is only robust contrast enhancement (ie, high-grade gliomas), and not applicable to low-grade gliomas. In addition, it is subject to artifact at the air/brain interface, which can significantly degrade the quality of imaging.
5-Aminolevulinic Acid
Direct visualization of tumor tissue during surgery is helpful in guiding resection to ensure complete removal of neoplastic tissue. Over the past decade, there has been increasing use of fluorescent compounds as an adjunct to glioma surgery. Among these, 5-aminolevulinic acid (5-ALA) is the most widely studied. A derivative of hemoglobin breakdown, it is administered exogenously before operation and accumulates in tumor tissues where it undergoes further metabolism to protoporphyrin IX. The latter is a red-pink fluorescent compound under blue light. Intraoperative use of 5-ALA allows the surgeon to identify neoplastic tissue with more than 90% sensitivity and specificity. The use of 5-ALA in malignant glioma surgery results in a higher rate of complete tumor resection (65% in 5-ALA group vs 36% in control group) and a higher rate of progression-free survival (41% vs 21%) that was demonstrated in a randomized multicenter prospective trial.
There is evolving evidence that the combined use of 5-ALA and iMRI results in a greater extent of resection. A recent study evaluated the extent of glioblastoma resection in patients in whom both 5-ALA and iMRI were used and compared it with patients in whom only stand-alone iMRI was used. The investigators found a significantly higher rate of gross total resection in patients in whom both adjuncts were used (100% vs 82%). Combined use of both modalities also can be useful in cases of recurrent glioblastoma to ensure tumor tissue is visualized and fully resected. On the other hand, iMRI may help identify areas of tumor infiltration that do not display fluorescence. A recent study showed that 39.3% of biopsy samples that were negative for fluorescence and had contrast enhancement on iMRI contained tumor tissue. Thus, combined use of both modalities is complementary to enable the surgeon to fully visualize neoplastic tissue and to ensure maximal surgical resection.
Related posts:
Intraoperative Imaging
iMRI During Transsphenoidal Surgery
A Novel Use of the Intraoperative MRI for Metastatic Spine Tumors
Magnetic Resonance Thermometry and Laser Interstitial Thermal Therapy for Brain Tumors
Intraoperative 3D Computed Tomography
Intraoperative Ultrasound for Peripheral Nerve Applications
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
