Assessing the Structural Footprint of Minimally Invasive Brain Cannulation on Cerebral White Matter: A Cadaveric DTI Model
Background: Minimally invasive approaches to brain tumors offer the potential of decreased iatrogenic trauma related to tumor visualization compared to conventional approaches. Currently there are no validated models to examine axonal damage after minimally invasive entry into the brain.
Object: The authors present and evaluate a cadaveric model of brain cannulation using diffusion tensor imaging fractional anisotropy measurements. Two different methods of access are compared.
Methods: Freshly harvested unfixed cadaveric brains were cannulated using both direct and indirect methods. Specimens were subjected to 68-direction DTI scans and proton density imaging. FA data from a “region of interest” surrounding the entry zone was extracted from scans using imaging software and analyzed.
Results: FA values were significantly higher following indirect cannulation (less invasive method) than they were following direct cannulation. FA values for undisturbed brain were significantly higher than in either of the cannulated groups.
Conclusion: Axonal damage following brain cannulation can potentially be evaluated by FA analysis in a cadaveric model. Future studies will focus on histologic analysis and clinical validation.
Diffusion tensor imaging (DTI) is a magnetic resonance imaging methodology that can be utilized to visualize neuronal microstructure. Diffusion of water inside the human brain is largely limited by the neurolemma. Since the nerves of the brain travel in large fascicles, this diffusion process can be exploited to visualize white matter tracts using DTI. Furthermore, degree of water flow along these axonal tracts can be quantified by using a measurement called fractional anisotropy (FA). (1-3) FA is a scalar value, calculated from the eigenvalues of the diffusion tensor, which describes the independency of a diffusion process. High FA values indicate diffusion in a uniform direction whereas low FA values indicate more random motion of water. Therefore, lowering of relative FA values within specific white matter tracts has been postulated as a marker of neuronal injury in clinical studies and animal models of stroke, traumatic brain injury, radiation-induced injury, and epilepsy (4-8). Iatrogenic white matter injury from surgical trajectories into the brain can also be quantified using tractographic methods (9, 10).
Traditional open approaches to subcortical lesions are difficult to perform without causing significant trauma to the overlying normal tissue. For deep lesions, extensive retraction may be required. Minimally invasive cylindrical brain retractors have been deployed over a dilator device with success in multiple case series (5, 11-14), building on prior work demonstrating feasibility of a microsurgical cylindrical brain retractor (15-17). There are even reports of using minimally invasive approaches for hematoma evacuation18. However, the degree of brain trauma incurred from this method of brain access has yet to be radiographically assessed in human models19. In addition, potentially less invasive techniques, such as inflating a balloon within the brain to create a channel for brain surgery, have not been comparatively assessed (20-22).
This study sought to apply diffusion imaging techniques to assess iatrogenic brain injury in a cadaveric model of brain surgery. Rather than test conventional retraction methods of the brain using spatulas and brain dissection, emerging minimally invasive techniques for brain retraction were evaluated.
All specimens were procured from the hospital morgue following a protocol internally approved by the department of pathology. Within 6 hours after death, brains were harvested from patients with no known pre-existing neurologic disorders. Specimens were sectioned into two approximately 5cm by 3cm blocks to facilitate cannulation and transport. Cuts were made anteriorly at the level of the anterior genu of the corpus callosum, posteriorly at the level of the precentral sulcus, and along the midline. Lateral areas of cortex were removed to make the blocks symmetrical.
Two distinct cannulation techniques were used to access the brain, working sequentially from anterior to posterior for each block (Figure 1A). The “direct cannulation” technique was performed by passing the cannula plunger through cortex approximately 1 cm posterior to the anterior genu, and allowing the cannula to follow, similar to placement of a guide sheath for working channel neuroendoscopy. 2 cm of undisturbed cortex was maintained between cannulations. Indirect cannulations were performed using a 1.5 cc dilatable fogarty catheter. The balloon segment of the catheter initially punctured the cortex. After this, 1.5cc of air was injected to inflate the balloon. Using the balloon as a guide, the cannula was passed into cortex (Figure 1B). After cannulation, specimens were stored in sealed water filled containers and kept in refrigerated conditions until scanning.
Scans were performed within 12 hours of cannulation on a Siemens (Siemens, Erlangen, Germany) 3T Trio system. Scanning protocol included a 68-direction, 30 slice DTI scan with echo/repetition time TE/TR=99/4000ms and voxel size of 1.35 by 1.35 by 2. mm. Fractional anisotropy maps were calculated online, using the scanner software. Additionally, 75 slice proton density scans were acquired with TE/TR=9.8/3000ms and with voxel size of .45 by .45 by .8 mm. After scanning, specimens were kept in storage in 10% formalin.
All data were analyzed using OsiriX Imaging Software (http://www.osirix-viewer.com/). FA maps were co-registered to the proton density scans for anatomical accuracy. Two Regions of Interest (ROIs) were drawn as concentric circles centered on the cannula (Figure 1C). The larger circle had a constant radius of .5cm radii around the cannulae. The smaller circle was constructed to cover the area of the cannula. “Average FA” was recorded for both direct and indirect cannulation for a given slice and both ROIs. The FA of the cannula ROI was subtracted from the FA of the .5cm ROI to obtain the “Adjusted FA” of the area immediately around the cannula. (Equation 1) A third ROI, constructed to match the radius of the cannula, was placed between the two cannula ROIs in an equidistant fashion. This ROI served as a control, measuring the FA of undisturbed brain. This technique was repeated on a slice-by-slice basis over a depth of 5 slices and the resulting data was averaged to determine the total FA for a given cannulation.
This data was subjected to a paired T-Test using Microsoft Excel. The p value of .05 was prospectively determined to be statistically significant.
Six specimens were analyzed for FA changes after cannulation using both direct and indirect methods. For each specimen 13 slices of MR data were used for analysis. The mean average FA for direct and indirect cannulation was, respectively, 0.1893 and 0.2956. Average Adjusted FA for direct cannulation varied from 0.0267 to 0.105 with a mean value of 0.0645 while indirect varied from 0.0596 to 0.208 with a mean of 0.137. (Table 1) Analysis of all 78 slices revealed an Adjusted FA range of .0211 to .241 for direct cannulation and .0545 to .611 for indirect. Control FA varied from 0.0648 to 0.2152, with a mean of 0.1576. Not a single slice exhibited direct cannulation FA greater than indirect.
Paired 2 tail t-tests were conducted between direct and indirect datasets using both average and adjusted values. All analyses were statistically significant. (Table 2).
Minimally invasive brain retraction systems offer the potential to decrease trauma to normal brain during surgical removal of deep-seated intraparenchymal masses. However, an assessment of the utility of such systems requires an appropriate experimental model that can quantify brain injury incurred via various access routes through the cortex. The investigators have developed such a model, via MRI assessment of cadaveric specimens.
Previous research in an animal model has demonstrated that a dilatable balloon retraction system causes less damage to white matter as compared to blunt microsurgical dissection (23). In this particular study, both histological and clinical evidence demonstrated that balloon dilation resulted in less brain injury as compared to standard techniques. In addition, recent research has demonstrated that iatrogenic white matter injury can be quantified following brain surgery using fiber tracking methods (9, 11). The investigators submit that the above imaging method is a relatively simple technique for assessing iatrogenic white matter injury following brain dissection, which can then be correlated with functional outcomes in live patients.
The concept of FA correlation to axonal integrity is not new. Many investigators across multiple different disciplines in brain research have used FA as a metric of neuronal injury. (6-8, 24) FA is highest when there is uniformity in a diffusion process such as in intact white matter tracts. It decreased progressively to a theoretical value of zero as diffusion becomes more dimensional. In the case of traumatized axons, damage increases the degrees of freedom of diffusion and thus decreases FA. (25)
The use of fresh cadaveric specimens with diffusion tensor imaging has also been previously reported and validated26, 27. Using unpreserved specimens ensures the retention of brain tissue fluidity. Thus, retraction in these specimens directly mimics operative conditions. In addition, cadaveric brains can be imaged easily and accurately, without motion artifact. Many authors have previously used diffusion imaging in postmortem brain tissue to study neuropathologic features of the brain28-30. In fact, Seehaus et al (31) published a histological validation of postmortem DTI. Perhaps most convincing, Budde et al32 published a recent study evaluating blast traumatic brain injury in rats using DTI. In their study, rats were subjected to controlled shockwave exposure. After running post-trauma functional tests the authors sacrificed the rats and performed tractography on the formalin fixed post-mortem brain specimens. The authors found significant differences in FA in multiple cortical regions that furthermore correlated to poor functional outcome. For example, a significant decrease in FA was discovered in the hippocampus, which correlated to poor performance on the Morris Water Maze task.
In light of this evidence, the authors believe that this model is a reasonable estimation of iatrogenic brain injury following tissue manipulation in cadaveric specimens. However, live imaging in vivo would clearly be superior to a cadaveric model, and is planned for future studies.
The data clearly illustrates a general increased fractional anisotropy using the indirect cannulation technique in comparison to the direct cannulation method. Fractional anisotropy is a scalar measurement of the independency of a diffusion process. Higher values indicate more uniform diffusion. In the case of neural tissue, diffusion is largely limited by the cell membrane. Thus, lower FA values indicate compromised neuronal integrity. We believe the statistically significant difference in FA between indirect and direct cannulation is evidence for increased preservation of neural tissue using indirect, balloon-guided cannulation. Generally increased FA values in the control specimens is further evidence of this hypothesis.
Brain cannulation for minimally invasive access to deep-seated brain tumors is an area of growing interest in neurosurgery. To assess the collateral damage to surrounding neural tissue, validation studies need to be conducted to assess white matter integrity following manipulation. Diffusion imaging provides unique visual and quantitative measurements that can be utilized to answer these critical questions. The application of DTI to assess cannulation techniques is a simple but novel idea. The authors hope that this work facilitates analysis of brain cannulation and retraction technology with the goal of maximum preservation of cerebral white matter.
We would like to thank Dr. Julia Koeffler (UPMC Neuropathology) and Denise Davis (UPMC Radiology) for their assistance on this project.