10 Frontal Lobe Gliomas
The frontal lobe (defined for this chapter as the parts of the frontal lobes not part of the motor or premotor cortices or their immediate white matter connections), has long been viewed by neurosurgeons as a “safe” place to operate. It is simple to explain why: this part of the brain has a great deal of parallel processing and redundancy, making minor transgressions into this lobe for tumors like brain mets and meningiomas forgiven, meaning that the patient lacks an obvious problem. It is unquestionable that minor incursions into the anterior frontal lobe to remove a non-infiltrating brain tumor are well tolerated.
Gliomas pose a different dilemma for the frontal lobe. They cover a larger area of the frontal lobe, especially if you consider the infiltration of tumor cells into the peripheral brain, and due to their nature, often mandate removal of a large part of frontal lobe. This brings frontal lobe functional concerns front and center, and this can often vividly teach you what the frontal lobe does, especially if you talk to patients and their care givers.
Having done numerous aggressive frontal lobe glioma resections, I have begun to respect the frontal lobe in a way I never anticipated. A hemiparesis is not good, but it does not change who you are as a person, leave you intolerable to friends or family, or make you unable to meaningfully interact with others: frontal lobe injury can do all of these. Ignore the frontal lobe and its anatomy at your own peril.
We do not entirely understand how the frontal lobe is wired up and how it works. Also in many patients, frontal lobe function is extremely redundant, and can tolerate a lot of alteration before problems arise, while others less so, and we do not know how to predict this or prevent problems. This is not to say that we can avoid all frontal lobe problems and still take out the tumor to a satisfactory degree: many of these tumors are massive, and as I have stressed repeatedly, leaving a bunch of glioma growing in the brain to save this function is unlikely to achieve this goal in the long term. Those who are willing to sacrifice tumor control for function often end up with neither in the long term.
This chapter is not going to solve these mysteries in 5 pages, but it provides a framework for more thoughtful planning of frontal lobe tumors, with the goal, as always, of balancing tumor resection with functional preservation. We cannot always prevent frontal lobe dysfunction, but we should base our trade-offs on rational decisions, instead of viewing the frontal lobe as the wild west where you can do anything you want with minimal consequences.
10.2 What Does the Frontal Lobe Actually Do?
The global functions of the frontal lobe are familiar to most physicians: judgment, attention, concentration, behavioral control, inhibition, higher executive control. Frontal lobe syndromes are also known: being “frontal” is used by many as a synonym for either being abulic, or being disinhibited, careless, rude, or unable to focus. But the functional anatomy and organization of the frontal lobe is less standard in neurosurgical curricula: partly because it is not well understood, and partly because the frontal lobe is relatively forgiving, so neurosurgeons have not spent the time passionately trying to find an answer to these mysteries.
The best neuropsychological and functional imaging data suggest the following classification of frontal lobe functions:
Medial frontal lobe: Motivation and attention.
Dorsolateral prefrontal cortex: Working memory and executive function.
Orbitofrontal cortex: Emotional regulation and judgment.
It should go without saying that these functions do not easily lend themselves to simple tests in the operating room, though we are actively evaluating adaptations of some more complex tasks to make this more easily testable.
A different view of the prefrontal lobe is as an organ of control or other brain regions. For example, the orbitofrontal cortex is involved in regulating emotional centers of the basal ganglia and other structures, but may not be a principle emotional driver itself. The medial frontal cortices are not motor areas, but are critical for controlling parts of the brain that are, i.e., turning them on. The dorsolateral cortices are not essential for memory formation, but help regulate areas that are.
10.3 Global Anatomy of the Frontal Lobe
Chapters 3, 5, and 6 spend a great deal of time on various aspects of frontal lobe anatomy and this section will merely summarize the key points to tie them together.
The white matter connections leaving the prefrontal gyri generally run roughly anterior-posteriorly to some degree, especially at the frontal pole. The more superior connections have a rostro-caudal component to their general anterior-posterior direction, and the lateral connections have some medial-lateral component in addition to their general anterior-posterior angle. A simpler way to view these connections is as somewhat pyramidal with the apex pointed at the key behavioral effectors of the central core, the basal ganglia and thalamus. Of course, not all tracts neatly fit this model, but many do.
The implications of these observations for rational frontal lobe surgery are significant. It is safe to say that most of prefrontal cortex function requires some anterior-posterior type connection with a distant effector structure (be it basal ganglia, thalamic, or other cortices), in order to manifest its activity in the form of an observable change in brain behavior. For example, it seems to me quite unlikely that the anterior superior frontal gyus can meaningfully change the function of other brain regions though a series of u-fibers to the MFG, then the IFG etc, after its principle apical outflow pathway has been cut (Fig. 10.1). Thus, it is possible to disconnect and inactivate a great deal of the prefrontal cortices with a simple cut in the coronal plane (Fig. 10.1). This is not to suggest that we avoid this cut, because often it is the correct cut, but rather that if you have cut across the posterior part of a prefrontal brain region, that you have usually disconnected it from the rest of the brain, and it is basically non-functional. In glioma surgery, it does not make sense to leave non-functioning tumor involved brain in the head. If it is not meaningfully connected to the rest of the brain, it basically is like a meningioma or a met.
10.3.1 The “Back Wall” of the Frontal Lobe
Many of the key large pathways at the back of the frontal lobe and other key structures such as the basal ganglia, cumulatively form a conceptual boundary in the posterior aspect of the frontal lobe, which can be thought of as the “back wall” of the frontal lobe. This is roughly a coronal plane through the frontal lobe just anterior to the head of the caudate nucleus (Fig. 10.2), but it is not perfectly planar and is warped somewhat near the frontal rami of the SLF, and near the premotor areas.
The structures of this posterior wall include the following (roughly from lateral to medial and superior to inferior):
Frontal rami of the SLF
Frontal bend of IFOF
Descending fibers from motor and premotor areas
Head of caudate nucleus
Supplementary motor areas
Cingulum/default mode network
It is not an accident that the posterior frontal cut aims to navigate in front of these structures.
10.4 The Three “Ds” Applied to Frontal Glioma Surgery
It is well appreciated that the difficulty and risk of frontal lobe surgery increases as the work moves posteriorly. This is because posterior travel in the frontal lobe brings us in contact with structures such as the SLF, the FAT, the IFOF, the basal ganglia, and the motor and attention networks, which can have a major impact on circuits such as motor or language which are essential for the meaningful function of multiple other networks.
The principle question to ask when planning these cases is which systems and tracts are in the closest proximity with the desired cortical and subcortical cuts. The goal of brain mapping is to place a cut such that the networks that we wish to save are not in the specimen, and thus leaving an essentially non-eloquent brain tumor to be removed. The answer to this question not only involves the tumor’s anatomy, but also the patient’s preoperative function (there is no reason to try to save a damaged or unsalvageable system), the presumed tumor grade, and the goals of surgery (for example some cases have supramaximal resections as an option, while many don’t).
By defining our goals, we can determine whether we need to focus our efforts on medial frontal disconnection or a lateral frontal disconnection, which in turn makes the sequence of mapping tasks, the location of the craniotomy, and the overall working angle more obvious.
Exposure and Cortical Mapping
Medial Frontal Disconnections
This is the more common disconnection in the frontal lobe and is useful for gliomas which are principally centered in the superior or middle frontal gyri. I personally prefer this angle of attack because it is less disorienting to me than the lateral disconnection, and this approach can be used to deal with most of the frontal lobe tracts.
Because the angle of attack is relatively superior to inferior, it is essential to try to get the apex of the head as high as possible in the operative field so that the remaining brain does not fall across your line of sight during the disconnection. In a sleeping patient, this is easily achieved by operating in the supine position. However, an awake patient should be on their side so they can see and interact with the examiner, so the position is a compromise achieved by tipping the apex of the head upward, and combining this with reverse Trendelenburg positioning of the bed (Fig. 10.3).
Cortical mapping of the medial frontal lobe is often negative mapping, in part because many frontal lobe intraoperative tasks are still poorly localizable, and as we know, most of the center pieces of human brain functional networks (i.e., primary motor, speech etc.) are relatively posterior in the frontal lobe. The craniotomy (Fig. 10.4) should be centered on the posterior most aspect of the planned resection as this is both the cut you are planning, and the area most likely to yield cortical findings (note that the keyhole principle predicts that the surface is the only area which cannot be easily reached by redirecting the microscope as needed, and thus, cortical surfaces which need to be mapped must be exposed). It should be planned so that you can reach the anterior pole, the midline, and the lateral extent of your cut, but these areas can be cheated and removed under the skull (Fig. 10.5).
Medial frontal disconnections usually involve disconnection from more than one tract, and thus it is important to map motor and motor planning (which is posterior), but also speech, spatiotemporal function, and other tasks. For example, primary language planning areas are unlikely to reside in the medial frontal lobe, but speech arrest can be useful for finding motor planning areas like the SMA.
Lateral Frontal Disconnections
This disconnection is used when the principle goal is to separate the tumor from the frontal rami of the SLF, and when the medial frontal involvement is of lesser concern. In these cases, the head is not laterally flexed as in the medial frontal disconnection (Fig. 10.6), and the craniotomy is centered over the surface involvement of the tumor. The DTIs are quite useful at predicting where the speech network is going, and planning the craniotomy accordingly.
There are two options that make sense to me: expected positive sites are either definitely in the field or definitely out of it (Fig. 10.7). By this I mean, that if after looking at the tractography, I expect speech tracts to be densely involved, I want to be certain I know where they are, and this means extending the craniotomy to make sure I get a positive map. On the other hand, if they are not involved, I would prefer not to put them at risk by exposing them, and try to plan the craniotomy so they are not in the field.
I start by mapping the motor strip first, as it is the easiest to find and it helps to rapidly define the rest of the cortical map. Speech arrest is typically found just anterior to the face/tongue motor cortex (it is the motor planning area for speech). I also check naming here to make sure I know the network, and also to get the patient ready to double task. I then look for motor arrest to ensure I know where motor planning areas are. Finally, we have found neglect centers in this area, especially on the right (this is probably being an SLF mediated task), so I look for this as well
10.4.2 “Divide” Phase
Medial Frontal Disconnection (Fig. 10.8)
This is an “L” shaped cut that separates the tumor from the motor system and SMA/FAT network posteriorly, the SLF and parts of IFOF laterally, eventually ending in the frontal horn of the lateral ventricle.
The lateral limb of this cut is always done first, typically with an arm/naming double task. This is because the posterior cut can lead to abulia or mutism, which can recover but will prevent you from meaningfully mapping speech functions to protect the SLF and IFOF. It is critical to get the angle correct, or you will greatly short change your resection in the depth. It is important to note that your natural angle of attack with the patient in the lateral position for brain mapping is angled medially towards the falx, and not on the superior to inferior as your brain generally thinks of it. Therefore, I start by subpially exposing the lateral sulcus of the most lateral gyrus I plan to resect (SFG or MFG), and keep this pia as a guide to the correct angle, which typically parallels the floor. I use a combination of the DTI and feedback from mapping to define my lateral borders. Initially, the goal is to separate the tumor from SLF, but inferiorly you can encounter IFOF, and realization of this caused me to begin to keep the patient awake until the lateral cut is down to the ventricle or orbital roof. Before switching to the posterior cut, I make sure I have an idea where the caudate head is, and that the anterior to posterior length of the frontal horn has been opened.
The posterior cut often involves a switch to motor based tasks, especially ones which require concentration as these can be used to protect the SMA/FAT and the medial networks such as the DMN. If not possible the double task is perfectly acceptable. The goal of this cut is to widely open the width of the posterior frontal lobe from the surface down to the medial-lateral aspect of the frontal horn from the midline until it joins the lateral cut. After subpializing the gyri on the surface, down to the bottom of the sulci, I start by finding the midline and continue the cut down to the falx. This allows me to find the cingulate sulcus, which is where I expect to find the attention networks. Depending on the surgical goals, I either remain lateral to the sulcus depths to stay out of the networks, or attempt to resect the cingulate cortices using repeated stimulations. Either way I try to find this early, so I am not encountering it in an uncontrolled fashion in a deep hole. The rest of the cut follows the direction of the sulci as superior to inferior as possible, the goal being not to drift into the FAT or the descending communications with the basal ganglia/thalamus in the sub cortex. Once the L-shaped cut is down to the ventricle on all sides, the cut is complete.