14.1 Introduction

This chapter is about taking on the bad boys. These are tumors which are obviously high risk, and ones which many, often good tumor surgeons deem “inoperable.” They are in diverse areas of the cerebrum, and many of the anatomic aspects of these tumors are diverse, yet I will make the following observations which will make clear why I put these tumors in a common chapter towards the end of the book.

1. 
   

   The tumor resection in these cases is almost always a lesionectomy: anatomic wedge type resection like lobectomies are rarely part of the operation as they are inadvisable in these parts of the brain.

   
   
2. 
   

   The tumors are often surrounded on all sides with critical brain. There is no “destroy” phase of the surgery as you are frequently dealing with essential functional areas on all sides. Thus, the end of the mapping is usually the end of the case.

   
   
3. 
   

   The 3-dimensional anatomy of these tumors is unfamiliar and complicated. Most of us can visualize the temporal tip because we have seen it in its entirety, but operating in the forceps major exiting the splenium and extending into the occipital white matter, or in the thalamus demands a great deal of thought to visualize what you are doing. Many of these resections require you to work based on a mental reconstructions of anatomy from conceptual knowledge, more than they are based on actual visual knowledge of the structure. Getting lost is easy to do with tragic consequences.

   
   
4. 
   

   The patients are often sicker. Many of these types of tumors are high grade gliomas, and edema in these parts of the brain have major functional consequences. The patients are harder to map, and harder to achieve results you are excited about. Worse, many times, these patients have undergone a biopsy (after being deemed “inoperable”), radiochemotherapy, and a long run of steroids and physical deterioration prior to an attempt to resect the tumor, making it even harder. Your goals need to be realistic and brave. Remember that the only thing worse than being paralyzed or aphasic from a GBM is being paralyzed or aphasic from a GBM which is still there killing you.

   
   
5. 
   

   Often, the fibers of interest are small and unpredictable. For example, it doesn’t take a major transgression of the ascending alertness fibers from the midbrain during a thalamic tumor to lead to disaster. Nor does it take a major misstep to disconnect the motor network, or to veer into the SLF at the TPO junction.

14.2 Common Pieces of Advice for Tumors Like These

1. 
   

   Go slow. The margin for error is small and the risk of disorientation is high. There is often not a great volume of tumor to remove, and the cases are rarely very long (none of these cases takes anywhere near as long as a frontal lobe resection or an insular glioma), but it is important to force yourself to slow down, stimulate more often than usual, change tasks as often as needed, and reorient yourself frequently. There is no prize for cutting 30 minutes off the resection.

   
   
2. 
   

   Do not neglect the deep boundary. If you are doing these cases in the manner described in this book, you know well that you often depend on a clear anatomic boundary to signify when you are complete. Reaching the ventricle, the falx, or a well-defined sulcus, provides you reassurance that your cut is all the way past the fibers at risk. In contrast, one common feature of these cases is you really never get this kind of event to reassure you that you are done with the cut as the deep boundary is usually a white matter tract or a critical element. Failing to spend time addressing the deep boundary ends in either transgression of the tract at the bottom of the cavity (many times this is the SLF) or the critical structure at the bottom. On the other hand, if you are not aggressive with the deep margin, then you can expect to be disappointed with your extent of resection.

In short, these cases involve a cut into the white matter on the deep surface, which is usually based on functional considerations, and loosely guided by anatomic estimations (like comparing your depth to neighboring sulci, etc). I never stop the mapping until I am certain all of the boundaries are clear, especially the deep one. Conversely, once the mapping stops, I force myself to stop addressing the deep margin. It’s tempting to keep going and trying to get more tumor; however, either functional preservation is important or it’s not, and you cannot guarantee that the tumor doesn’t possess relevant brain tissue if the patient is asleep.

3. 
   

   The visual appearance of tumor tissue counts a lot in these surgeries. In most gliomas, a good anatomic resection of an area functionally disconnected brain is the best way to ensure an oncologically satisfactory resection. In other words, there is no residual tumor in the center of a hole. This avoids the issue of fingers of infiltrative and unclear tumor mixing in making it hard to know when to stop.

However, resecting to anatomic boundaries is not advisable in tumors like these. In many cases, the areas we are operating in areas that if transgressed in normal patients, would have devastating consequences. In some cases, surgery is made possible by cortical reorganization. For many of these tumors, especially the deeper ones, the more likely situation is that the resectable part of the tumor is that which is densely full of cells, pushing function out of the way, and the infilitrative portions of the tumor on the boundaries are unresectable. The visual features of glioma are described in Chapter 2 and they are critical in these cases as often this is all you can safely resect (i.e., it is the dense, noninfiltrative part).

4. 
   

   Use the concepts and techniques helpful in other more straightforward gliomas as building blocks for simplifying these tumors. Many times a disconnection used in other settings can be helpful for part of these cases. The anatomy of the insula, and of major white matter tracts can also be helpful in defining boundaries for more complicated tumors, or approaches to deeper tumors. There is a reason I waited till the last section to tackle these tumors, as you need solid fundamentals before setting out to tackle these monsters.

5. 
   

   Deal with the tumor everywhere else also. It is easy to fixate on tumor in a dangerous area and to neglect a fair bit of tumor somewhere else. Doing a bunch of work in a dangerous area and not addressing the tumor elsewhere in a less dangerous area is easy to do and oncologically inadequate. There is no reason to do a risky part of the case like a butterfly glioma and neglect tumor in the frontal lobe. Again, the goal is always to leave as little disease as possible, and this is not accomplished by getting excited about the unusual part and not maximizing the mundane parts of the case.

6. 
   

   Study the T2 images closely. It is critical to have a detailed understanding of where the tumor is exactly. This is always true but here it is essential for success. The T2 can tell you the position of the internal capsule, basal ganglia, thalamus relative to the mass (they are usually not in their normal positions) which is essential for picking an entry point into the tumor. It can define the exact gyri involved, and where the vessels are. The difference between a motor strip tumor, and a tumor in the parietal lobe which has pushed its way to the area where the motor strip usually is, is enormous.

7. 
   

   Use anatomy wherever you can. You are not going to get a lot of anatomic clues during these cases as they often involve taking something amorphous out of the middle of an amorphous mass in the center of the brain. Defining sulcal and/or ventricular landmarks is ideal prior to making any irreversible moves.

8. 
   

   Check your angles often. Many of these approaches involve deep targets and some are done in a non-neutral head position. Both of these make it quite possible to lose Anterior-posterior orientation and to veer into something terrible. Image guidance is no substitute for a good understanding of anatomy, but it is peerless in reorienting in a deep hole with few landmarks.

14.3 Methods for Resecting Specific Tumors

This chapter deviates from the organization of the rest of the book as these tumors do not fit the system very well. The “division” is a “surrounded on all sides” cut, there is no “destroy” phase, and the mapping never stops until the tumor is out as the most dangerous cut is on the blind side. Instead, what follows is a run through of the steps I use to remove these tumors.

14.3.2 Planning these Cases

As the anatomy of previous chapters should make clear, the sensorimotor system is a vertically oriented system (Fig. 14.1), which is unlike the orientation of most non-thalamic projections in most networks. It is best considered as looking like a bouquet of flowers: somewhat conical and widening near the top of the branching. Most of our cuts in previous chapters addressed protecting this system by cutting parallel to these fibers with cuts in the coronal plane paralleling the precentral or postcentral sulci, which maintain the cortical motor networks, and their connections with the cerebellum, basal ganglia, thalamus and spinal cord.

It is important when planning a case for a tumor which is actually clearly in the sensorimotor system, that we consider this region as a network, and plan a way to disconnect as few parts of it as possible. This starts with a clear definition of involved gyri. For example, it is critical that if the tumor is invading the motor cortex from posteriorly, that you do not approach mistakenly from the anterior side, as you will be adding SMA syndrome or a dysfunctioning PMA system to the problem list. Similarly, a posterior approach to an anterior tumor, not only destroys the sensory and visual input to motor planning, but also probably crosses the descending motor fibers in the process (Fig. 14.2).

DTI tractography is critical. As stated many times in discussions of thinking in terms of networks, there is no benefit to sparing the motor cortex if you cut the fibers. These can be pushed or infiltrated, and it is not always possible to resolve them in some cases when they are densely involved with the tumor. Equally critical is the fact that the SLF is running underneath the sensorimotor system, and this is the deep boundary of this resection. It is essential to note that you can add neglect or aphasia to a hemiplegic patient by neglecting the SLF in these cases (or by not removing the tumor and letting it destroy these fibers).

Finally, goals matter here and they should drive the surgical decision making. Sometimes patients can regain some function with careful resection, but this is not always the case. While we do not wish to be nihilistic, we should also be realistic, and try to put everyone in the best situation we can, even if that situation is still bad.

14.3.3 Executing the Case

Lower Sensorimotor Gliomas

It is at first, important to note that sensorimotor cases of the inferior (opercular) third are substantially simpler than removing tumors from the other parts of the motor and sensory cortices. Not only are the face and mouth bilaterally innervated and compensable, but they do not directly overlie the descending corticospinal fibers the way the hand motor cortex does. Further, the insular cortex gives you a key depth gauge (Fig. 14.3), as it tells you the approximate depth of the SLF, and of the stem of the descending motor fibers (they are just medial to the SLF). Cortical mapping can define the key players in the network, but I tend to be aggressive in these cases, on the assumption that the mouth/face motor and sensory function can be taken unilaterally if absolutely needed, and that speech will be preserved in the long term if the SLF remains intact in the depth ( Fig. 14.4 ).

Lower sensorimotor tumor cases begin with cortical mapping. It is usually best to look for spontaneous movement and sensation, as this is easy to find and useful for interpreting premotor sites. It is important to also look for SLF functions, such as speech or neglect, as they can have contributory sites throughout the area worth noting. Motor planning tasks can also be used to define the network before making a cut.

Once we have made an overall plan, resection of these tumors starts by subpial resection of the involved gyrus. The first goal is to find and preserve all en passage arteries exiting the Sylvian fissure (they are very likely going somewhere important). By resecting the brain around them, you provide slack and allow them to move out of the way. The goal eventually is to remove the opercular tissues to find the Sylvian cleft and follow it until the insula is visible, as this is the key landmark in these cases.

Once the insula and its orientation is clear, I direct my attention to the superior cut, which is the key step in the operation. This is made across the gyrus in the sagittal plane as you are disconnecting the tumor from the rest of the sensory and/or motor strip. It is important to make this cut aggressive and square with the initial axial trajectory (in other words, to not try to shortcut towards the insula too quickly); however, you must realize that this is the dangerous cut which risks injury to the SLF and motor fibers, and thus functional considerations will often push you off your geometrically ideal path.

At this point, I skeletonize the anterior and posterior pial boundaries of the involved gyri (it is rarely more than one or two gyri in this region), and join all of the cuts along the deep boundary of the tumor just lateral to the sagittal plane of the insula. By removing this piece, I can get a clear view of the deep margin without opercular brain in the way. I then slowly advance the deep boundary in all directions as far as motor and/or SLF functions will allow me to do. It is important that if the tasks don’t overlap (i.e., a hand task to test neglect on the right overlaps both major tracts whereas speech and hand function are less overlapping) that you rotate tasks often so you are not missing something. The insula serves as a critical landmark to remain aware of the SLF and its position.

14.4 Case Examples

Fig. 14.5 demonstrates a simple lower motor cortex low grade glioma, and this was removed without deficit as the mapping demonstrated reorganization of the face motor function superiorly in the motor strip.

1. 
   

   By tracing out the right angle (superior frontal sulcus and precentral sulcus) and using the “hand knob:” as land mark, we can follow this tumor down to see that this tumor is mainly located in the face/tongue motor cortex. The critical thing to note here is on the coronal images which show the tumor following U-fibers into the superior insula at the circular sulcus. We need to take the cut into the motor strip as far superiorly as possible based on mapping to address the top of these U-fibers, and to see the circular sulcus clearly. The insula will serve as the depth gauge as well, so we need to know where it is.

   
   
2. 
   

   None.

   
   
3. 
   

   None.

   
   
4. 
   

   The face motor was at the top of this cut (best seen on coronal images). This cut has drifted down inferiorly and underestimates its height. We did not feel we needed to sacrifice ipsilateral face motor in this case to get a good resection

Fig. 14.6 demonstrates a lower sensory strip GBM.

1. 
   

   At a glance, this may seem like a motor area tumor, but careful evaluation of the T2 suggests that it is in the sensory strip, and is pushing the motor strip slightly anteriorly. This enhancing part of this tumor was able to be completely removed.

   
   
2. 
   

   The pre- and post-resection images demonstrate the depth of this resection, and how distorted sulci can be shortly after opening. A low lying biopsy site can be visualized and this was made after motor mapping, under the microscope to ensure that the opercular arteries were not injured.

   
   
3. 
   

   As is typical for these tumors, the SLF and motor fibers are running along the deep border of these tumors. Looking at the coronal image, will reinforce the need to use the insula as a depth gauge in lateral sensorimotor tumors, the SLF and motor fibers really are never lateral to the insular surface.

   
   
4. 
   

   There is some T2 change left in the more superior sensorimotor cortex, but in a case like this, this is a satisfying result, especially given that the patient was intact afterwards.

Fig. 14.7 is the remarkable case of an extensive low grade glioma filling the lower third of the motor and sensory strips, and the posterior IFG. Surprisingly, the entire cortical surface had no function in it, and we were able to resect the tumor until encountering the SLF running deep to the cortical surface which we left tumor in. The patient had no neurologic problems with this surgery despite an aggressive resection of the tumor.

1. 
   

   This is a very extensive tumor in the motor and sensory regions as well as threatening the speech networks. The greatest risk is the depth of this tumor, which extends nearly to the ventricle. Note that this tumor seems to have a denser area (which is reasonable to expect might be noninfiltrative), and a less dense area (more likely to be infiltrative). We should be very concerned about the lateral part given its proximity to critical tracts.

   
   
2. 
   

   None.

   
   
3. 
   

   Not surprisingly, the SLF and motor are plastered on the deep surface of this tumor. Simply put, they have nowhere else to go. What is encouraging is that the frontal, parietal, and temporal SLF ramus, and the FAT do not seem to run through it, but rather around it. Remember that we cannot assume that there are no areas in this tumor which feed to it, DTIs are not that accurate presently and you shouldn’t follow it off a cliff. Tractography shows you bus routes, not all the stops. But it is promising, and should be studied with the possibility that this tumor can be aggressively removed.

   
   
4. 
   

   The amount of reorganization in this case was remarkable, and an aggressive resection was possible with no neurologic sequelae. In this case, we left a rind of tumor in the SLF but otherwise the involved brain was nonfunctional.

   

   

   

   

   

14.5 Case Examples

Fig. 14.11 shows a prototypic anterior approach to a medial premotor glioma. The degree of enhancement underestimates this avastin treated tumor. Note that the cut proceeds vertical and parallels the direction of the sensorimotor system as allowed by the mapping.

1. 
   

   The preoperative image, as usual in these cases, is complex, and defining goals is critical in a case like this. The enhancement begins in the SMA, and appears to follow the descending motor fibers towards the basal ganglia. This is unfortunately not the only time I have seen this ominous pattern of spread. The T2 changes are extensive and fill much of the premotor and SMA regions. The goal was to make a posterior frontal type disconnection as best possible given the functional anatomy, and to perform a large anatomic resection of as much of the T2 change as possible. Of note, the patient had significant preoperative weakness from this tumor, but was not hemiplegic, being weaker in the leg than the arm. Our goal is to try to save whatever function we can in this case, recognizing that without an aggressive turn of events, his motor prognosis is poor.

   
   
2. 
   

   The incision follows a previous open biopsy incision and parallels the cuts. Note in the post-resection images that this cut is quite irregular and complex, and this is due to the need to define appropriate safe paths to cut parallel to the motor planning system, which can be quite complex in the coronal plane. Slow detailed subcortical dissection is critical in these cases.

   
   
3. 
   

   The DTIs quickly reveal the nature of his motor problems. The FAT no longer enters the SMA and this has likely been disrupted by the tumor or the biopsy, or at least it is hard to visualize on the tractography. The motor system is splayed outward around the tumor and contibutions to the corticospinal tract are pushed quite far anteriorly. There are almost certainly connections that we do not see in this tumor due to edema.

   
   

   An additional noteworthy feature is the fact that at first glance it seems to be crossing the midline, but in fact this is simply an enlargement of the midline frontal lobe herniating under the falx. The coronal images demonstrate that the callosum is largely uninvolved but is merely pushed by the tumor bulk.

This patient on presentation has a kind of zoned-out affect, which I have seen many later staged GBM patients look like (this man came to us for a second opinion 6 months after completing radiotherapy elsewhere, when he was quite a bit worse than initial presentation). By this I mean he is responsive, conversant, able to follow commands, etc., but he is not driving the conversation mainly, he is passive. I think this happens for a lot of reasons, including fatigue, but in his case I suspect much of it stems from compromise of the DMN and cingulate seen in this image.

1. 
   

   The resection is aggressive, and we did not make him any worse; however, he had a marginal recurrence in about 3 months and by this point his poor functional status made all of us decide not to treat him further. One thing this case highlights the need to get these tumors under control if at all possible, as early as possible. It is much easier to manage a patient through subsequent treatments when they haven’t been accumulating edema and steroid toxicity for many months prior to starting an aggressive strategy. I try to aggressively treat all with gliomas that seek my help and understand the risks, but it is beyond question that it is much more fulfilling when we get to them early and treat them aggressively from the start.

Fig. 14.12 demonstrates a case of a patient who presented hemiplegic from this GBM straddling the motor strip and SMA. Despite his inability to participate with motor tasks, we used direct stimulation to identify and preserve motor cortices and descending motor fibers, while resecting the enhancing portion of the tumor. His function improved to 4/5 the following day, but we could only preserve motor function in him for about 6 months before the tumor destroyed it.

1. 
   

   As usual, the enhancing part of the tumor underestimates the T2 changes, and much of the T2 change is in the motor strip and not worth going after. The degree of T2 change is less than the previous case though, and given the more posterior location, we decided that a lesionectomy was all we should try, as an anterior resection would be in the SMA and not really be in the T2 areas.

   
   
2. 
   

   The incision looks similar to the previous case as it had been similarly resected. The tumor was clearly visible upon opening the dura, and the markers demonstrate areas where we were able to stimulate motor responses with a cortical stimulator. We used this technique to define boundaries and primarily resected abnormally appearing areas.

   
   
3. 
   

   The DTI shows an interesting pattern: the tumor lies between the SMA and FAT complex, and the motor strip. The fact that he regained his function following a lesionectomy argues that the key communications between the motor strip and SMA are not through the U-fibers.

   
   
4. 
   

   The resection is limited to the enhancing region but with that as the goal this scan looks good.

Fig. 14.13 demonstrates the case of a premotor glioblastoma operated at another facility who suffered hemiplegia immediately after the resection which only improved to 0/5 in the arm and 3/5 in the leg. Examination of the DTIs demonstrated that although the entry point selected was anterior, the resection drifted in the subcortical white matter entering the descending motor tracts. We were able to preserve his remaining function at the repeat surgery with our standard techniques.

1. 
   

   There are a number of points to be made with these images. First, note that the enhancement is mainly in the premotor areas with extension into the deep white matter. I speculated that the tumor may be following the premotor communication with the basal ganglia (see (Fig. 6.4) for an example of this). The main epicenter is in the inferior frontal gyrus and operculum. Second, there is the common issue with defining what is tumor and what is treatment effect in these cases. As always, I try to work a margin around the tumor as much as I can under the assumption that there almost always are tumor cells in this T2 change around the tumor, and by taking it out, if functionally possible, I am reducing tumor burden to its lowest reasonable amount.

Finally, the cortical entry path for the previous resection starts relatively premotor, and in a traditional view of the brain where cortex comes first, this should be good enough to avoid motor problems. But careful examination clearly demonstrates that the resection continues medial and inferior into the motor network and even into the basal ganglia, which were plausibly mistaken for tumor (the tissue looks so similar that if you are not relatively certain where you are and that you might be near it, your mind will not connect the fact that you are in the basal ganglia). This is easy to do if the patient is asleep for the subcortical portion, or if your subcortical angles are off. The surprising part wasn’t that there were motor problems in a case with a “safe” anterior entry point, it was that it improved at all.

1. 
   

   In the left panel, the cortical mapping has been completed. In this picture, the right Sylvian fissure is at 12 o’clock in the image and is a bit under the bone flap. The gyri are clearly enlarged and distorted. The two small anterior sites indicate premotor sites, where we found negative motor events (i.e., freezing) which in this setting represent probable motor planning areas. The more posterior marker indicates positive motor sites for the face found in the anatomic location of the motor strip. Following this, we performed a complex version of a lateral frontal disconnection. This had to account for the fact that the FAT is running on the medial (i.e., deep to us) surface of the tumor, and may or may not be participating in this patient’s motor function. It has to parallel these sites in the coronal plane to respect the coronal orientation of the motor system. It has to extend deep to the insular level, but should preserve the IFOF. Regardless, the steps are ultimately the same: subpialize the operculum and find the insula for orientation, protect the arteries, make the cut with the premotor areas ensuring that the cut remains oriented roughly parallel with the sulcal orientation, make the superior cut and then disconnect the deep boundary to free up and remove the tumor.

   
   
2. 
   

   The DTIs make clear that the SLF and motor fibers have been entered at the previous surgery, especially on the axial images. The FAT has also likely been transected partially as well, as only a small number of fibers connect the SMA to the premotor areas and the coronal images suggest that the previous resection tract enters them as well. It is always difficult for me to interpret networks of previously injured brains, as it is unclear how they recover, but my working assumption in this case was that the recovery this patient saw resulted from recovering from an SMA type syndrome, and the permanent deficit resulted from the partial transection of the descending motor fibers. I think in this case, working on that assumption, that it is probably wise to try to save whatever FAT and SLF is still remaining, as they may contain some crossed FAT fibers or other mechanism for rewiring this area to allow him to recover spontaneous motor function. I think this may account for the more prominent premotor mapping in this case: in the setting of this injury, more of it may be essential then normal.

   
   
3. 
   

   The resection is aggressive, but it is important to compare the postoperative coronal to the DTI to note how the deep cut tries to conform to the residual FAT. This patient did not lose any additional motor function as a result of this resection.

Fig. 14.14 demonstrates a left-sided premotor tumor. Tumors like this add the additional problem of protecting speech, as well as the FAT/crossed FAT, needed to initiate speech.

1. 
   

   This tumor has undergone a lesionectomy at another facility as seen, and has recurred within 3 months of completing chemoradiation. We are returning to complete the resection and obtain a better margin including the T2 changes in hopes of improving his chance of responding to a new agent.

   
   
2. 
   

   None.

   
   
3. 
   

   The DTIs highlight the complexity of a case like this. First, like previous cases, the tumor splits the SMA/FAT complex and the motor strip, putting both aspects of this system at risk. The SLF is at risk on the deep and lateral parts of this resection. Even the DMN is potentially in play in this case. This will require a lot of work for a small amount of tumor removal.

   
   
4. 
   

   This postoperative scan demonstrates the aggressive resection we achieved. He has some issues initiating speech, but otherwise is high functioning, working full time, and recurrence free at 2 years after this surgery.