Keywords

Introduction

Much of the Chiari literature is ridden with questions, but scarcely any answers have emerged since the time of Gardner and Williams. What causes a Chiari I malformation? If it is a mesodermal problem resulting in abnormal bone development, then why do acquired Chiari malformations exist? How important are the tonsils? If the tonsils are considered essential for a diagnosis of a Chiari I malformation, then what explains the Chiari zero? Why do symptoms arise in some patients but not in others, although the MRI images seem identical? And why does syringomyelia exist in some but not in others? Which is the best surgical Chiari decompression technique? How wide should the decompression be? Should the tonsils be reduced? Why are some of the symptoms typical and others “crazy”? Why do children fare better than adults? Why are some Chiari malformations inherited and others not?

Serious multidisciplinary research efforts should be undertaken if any of these questions is to be answered. Research from the Bernard Williams days was limited to clinical observations, basic mechanical modeling, and invasive techniques of cranial and spinal pressure monitoring. The advent of MRI, computer technology, and molecular biology has radically changed both perspective and prospects. Pathophysiological theories can now be tested, not just hypothesized. Invasive procedures have been for all practical purposes replaced with noninvasive technology. Probably the most important—certainly the most used—of these technologies is the ability to track CSF flow through dynamic MR imaging, quantitate it, and analyze it using intricate software programs. In this chapter, we will not address research on the embryology or pathophysiology of the Chiari malformation itself as this is covered elsewhere in the text. Instead, we will review research efforts that ask how the Chiari malformation causes clinical problems and syringomyelia. Such efforts have been productive in large part due to collaborative efforts among neurosurgeons, radiologists, engineers, and physicists. The chapter starts with an overview of previously proposed theories of syringomyelia formation and concludes with an examination of present-day tools used to explore the pathogenesis of clinical (symptoms and signs) and imaging (syringomyelia) findings in patients with the Chiari I malformation. This will consist primarily of a review of dynamic MRI flow imaging, followed by a brief overview of other propitious research efforts aimed at understanding this enigmatic anomaly.

Proposed Theories of Syringomyelia Formation

Gardner’s Hydrodynamic/Water-Hammer Theory

In 1959, Gardner and Angel suggested that syringomyelia forms because of a persistent opening of the central canal at the obex, in the setting of closed fourth ventricular outlet foramina [1]. This hypothesis was the basis upon which plugging of the obex was suggested as part of the treat­ment for syringomyelia. Subsequently, Gardner expanded on his theory to propose that syringomyelia is a result of direct transmission of a CSF pulse through the obex in a “water-hammer” fashion [2, 3]. The theory is based on Bering’s assumption that during embryological development, pulsations from the choroid plexus contribute to expansion of the neural tube. Gardner proposed that these pulsations also help with the development of the arachnoid pathways and suggested that a balance exists between the pulsatile flow in the supratentorial and fourth ventricular choroid plexus. When this balance is disturbed, overactive supratentorial pulsations may result in tentorial migration and the development of a Chiari I malformation. In turn, compression by the posterior fossa structures leads to closure of the fourth ventricular outlet foramina, which forces CSF through the opening at the obex and into the central canal. Gardner proposed that the obstruction would first result in distension of the central canal (hydromyelia), after which the fluid would rupture into the substance of the spinal cord (syringomyelia).

Inconsistencies of Gardner’s Theory

Gardner’s theory could not explain the following observations [4]: First, if one were to assume that the pathophysiology of syringomyelia is invariable regardless of etiology, the hydrodynamic theory cannot explain cyst formation secondary to trauma, arachnoiditis, tethered cord, etc. Second, this single theory of pathogenesis at the foramen magnum does not account for the syrinx septations that are often evident on MRI. Third, West and Williams [5] showed using ventricular contrast studies that the obex is actually patent in only 10 % of patients, thus refuting Gardner’s hypothesis. Furthermore, Milhorat and colleagues have suggested that central canal ependymitis can cause an obstruction that results in dilatation of the central canal cephalad to the obstruction. This is based on the observation that CSF can be produced by the ependymal lining of the central canal and the still unproven assumption that CSF normally flows through the central canal [6, 7].

Williams’ Modifications of Gardner’s Hydrodynamic Theory: The Suck Effect Theory

Based on manometric observations in normal subjects and Chiari I patients, Bernard Williams devised a theory that examined syringomyelia from another perspective [8]. Similar to Gardner, he postulated an obstruction at the foramen magnum. However, he theorized that the Chiari I malformation is an acquired anomaly that results from excessive molding of the head, perhaps ­during delivery through the birth canal, which may then cause hindbrain adhesions and related outlet obstruction. In support of this claim, he showed using ventricular contrast that posterior fossa arachnoiditis correlates strongly with a history of difficult birth [5]. Williams hypothesized that hindbrain adhesions can result in transient pressure differentials between the cranial and spinal compartments due to epidural venous congestion, particularly during valsalva maneuvers (coughing, sneezing, straining). This, in turn, may cause a delay of caudad CSF flow while maintaining normal craniad flow, and as a result, fluid is “sucked from the ventricle into the central canal.” Williams provided human manometric measurements demonstrating these pressure differentials and showing pressure equilibration postoperatively [8–10]. However, although this theory is more compelling than Gardner’s, it fails to provide an adequate explanation of syringomyelia from other etiologies, and as with Gardner’s theory, it assumes a patent opening between the fourth ventricle and central canal [11].

Perivascular CSF Dissection Theory

In an attempt to provide a more unified view of syringomyelia regardless of etiology, Ball and Dayan hypothesized that the impact of the tonsils on the posterior fossa structures results in distortion of the subarachnoid space, which then allows CSF to dissect into the perivascular (Virchow-Robin) spaces and subsequently into the spinal cord parenchyma [12]. Aboulker had a similar theory but thought that CSF dissection occurs via the dorsal roots with extension into the spinal cord.

Oldfield et al. expounded on the perivascular CSF dissection theory [13–15] by providing favorable observations using various imaging studies [13–15]. They showed that rostrocaudal movement of the spinal cord results in CSF dissection in the subarachnoid space and documented such movement both intraoperatively using ultrasonography, as well as on dynamic MRI studies. Unlike Williams’ theory, in which the CSF dissection is driven by valsalva maneuvers, Oldfield et al. proposed that normal CSF pulsations provides a more or less continuous reason for fluid to enter the spinal cord. They specified that the displaced cerebellar tonsils act like a piston as they are propelled caudally with systole, thus creating a pressure wave within the entrapped subarachnoid space and syrinx. One might consider as supportive evidence of the perivascular CSF dissection mechanism, and specifically Oldfield’s piston effect theory, the recent observation of a “presyrinx state,” in which spinal cord edema precedes syringomyelia [16, 17]. In addition, animal studies have provided evidence that such fluid flow between the subarachnoid space and the central canal does occur under specific experimental conditions [18, 19], and intraoperative ultrasonic studies have proven the occurrence of cardiac cycle-driven syrinx wall pulsations, which in turn decrease after dural expansion. Furthermore, more recent controversial research emerged, which hypothesizes that resonance between the subarachnoid space and syrinx fluid may be the driving force that causes fluid to enter a syrinx cavity [20]. Yet, the evidence remains incomplete, and others have presented arguments against the piston effect theory, namely, that the mechanism also relies on CSF being forced from the subarachnoid space into the spinal cord, and that the exposure of the spinal cord to an outside force would be expected to crush rather than expand a syrinx [11].

Intramedullary Pulse Pressure Theory

Based on animal experiments, Greitz’ group developed a theory that suggests that the fluid within a syrinx derives from extracellular fluid forced into the spinal cord from a high-pressure microcirculation rather than high-pressure CSF from the spinal cord subarachnoid space [11, 21, 22]. Specifically, they state that when the subarachnoid space is obstructed from any cause (Chiari I, tumor, arachnoiditis, etc.), there is significant decrease in pressure transmission to the distal CSF spaces and concomitant increased transmission of the systolic CSF pulse pressure into spinal cord parenchyma close to the obstruction. This imbalance of pressures between the spinal cord and subarachnoid space leads to distention of the spinal cord just below the blockage [11, 21, 22]. Furthermore, part of the systolic CSF pulse pressure is “reflected” into the spinal cord at the site of obstruction, also distending the spinal cord above the blockage [11, 23]. The repeated mechanical distention of the cord results in dilatation of the central canal and accumulation of extracellular fluid (of vascular origin) that ultimately coalesces into cavities [11, 21, 22].

In spite of tremendous advancements in technology and considerable effort by many researchers, no single theory has so far definitively solved the enigma of Chiari-related syringomyelia formation. However, the intramedullary pulse pressure theory seems to best explain the formation of syringomyelia independent of etiology.

CSF Flow Studies of the Foramen Magnum

Chiari Symptoms: A Functional Problem?

It has been suggested that the Chiari I-related symptom onset and syringomyelia formation are related, directly or indirectly, to dynamic processes in foramen magnum physiology that are not reflected on static imaging studies. More specifically, it is becoming evident that the extent of tonsillar herniation and size of the posterior fossa (both quantifiable on standard MRI) are not sufficient criteria for determining the symptomatic state of the Chiari patient. Rather, a more “functional” mechanism is at play. Possibilities include subtle chronic craniocervical instability (detailed in Chap. 14), CSF flow perturbance (central theme of this chapter), and potentially other craniocervical junction stresses that are yet elusive.

Early Work

A variety of methods have been used to study CSF flow abnormalities with no clear anatomical correlates on static MR imaging. Early on, measurements were made using invasive means. Foremost are the studies by Bernard Williams in the late 1970s [24–26], in which he calculated pressure differentials across the foramen magnum by simultaneously measuring intracranial and intraspinal pressures under a variety of clinical conditions. This allowed the study of pressure gradients in that location and showed that correction of the pressure dissociation is often associated with marked clinical improvement. Such techniques were found to support surgical indications while also providing a means of postoperative evaluation. However, with the advent of MRI, noninvasive approaches have been developed to serve the same purpose of measuring flow and pressure dynamics at the foramen magnum and elsewhere in the craniospinal axis.

Cine MRI

The idea that craniocervical hydrodynamics are altered in patients with Chiari I malformation and the possibility that partial CSF flow obstruction via tonsillar herniation and a small posterior fossa plays a role in the pathophysiology of this entity have guided most of the work in dynamic imaging. Initial applications of magnetic resonance imaging to CSF dynamic studies started in the early 1990s. The term cine MRI (Fig. 13.1) applies to this modality that evaluates dynamic processes (usually through blood and CSF) rather than the usual static structures (brain, dura, bone, etc.).

Fig. 13.1
figure 00131

Example of a typical qualitative cine MRI. Sagittal PC MRI showing severe flow restriction posterior to the spinal cord and tonsils at the foramen magnum, with craniocaudal flow present anteriorly (white). As noted in text, evaluating a severe flow restriction or near-normal flow is straightforward. The difficulties with such qualitative studies arise in patients with moderate flow restriction

Full size image

Early gated spin-echo MRI sequences were used in healthy volunteers to study the movement of the intracranial components and the pulsatile dynamics inside the cranial vault. Early evidence suggested that brain motion occurs in a funnel-shaped fashion as explained by the following: Blood influx into the cranial cavity during systole causes pulsatile propagation of ventricular compression and CSF displacement to the spinal canal, which leads to a complex interplay between the cranial contents—brain and CSF, both of which are vented through the foramen magnum [27]. When healthy volunteers were studied using cardiac-gated cine MRI, a clear relationship between the cardiac cycle and CSF flow termed “flow void sign” was detected, which consists of an area of decreased signal related to CSF flow during systole [28].

Heterogeneity of CSF Flow at the Foramen Magnum

In the early 1990s, Armonda et al. conducted one of the earliest studies, in which cine MRI results in healthy control subjects were compared to those of Chiari I patients, before and after surgery. The authors studied CSF velocity and flow direction in four particular regions in the craniocervical junction by examining the CSF velocity profile over the cardiac cycle. They found that normal subjects had a short period of CSF flow in the cranial direction followed by a sustained period of flow in the caudal direction. Conversely, subjects with tonsillar herniation had decreased velocity and obstructed CSF flow pattern with a longer craniad flow phase. In turn, postoperative changes in velocity seemed to mirror those of normal subjects. As the resistance to CSF flow decreased by elimination of the tonsillar herniation, an increase in magnitude and duration of caudad CSF velocity on MRI was observed, accompanied in some cases by syrinx resolution and symptomatic improvement [29].

Combining Invasive and Noninvasive Techniques

Accordingly, it became evident that while static MRI sequences allowed the characterization of anatomical differences between normal controls and Chiari subjects (diameter of the CSF pathways ventral and dorsal to the neural elements at the foramen magnum, syrinx dimensions, size of the lateral ventricles, any evidence of connection between the syrinx and the fourth ventricle, size of the posterior fossa, as well as cerebellar morphology including tonsillar size and displacement), dynamic assessment via phase-contrast imaging provided information that correlates the anatomy with the physiology. This was particularly true with regard to movement of fluid within a syrinx, as well as movement of CSF at the foramen magnum and in the subarachnoid space both ventral and dorsal to the spinal cord [30]. By combining these noninvasive parameters with intraoperative CSF pressure analyses, Heiss et al. were able to experimentally confirm previous hypotheses that tonsillar impaction in a smaller posterior fossa seems to cause partial intermittent occlusion of the subarachnoid space at the foramen magnum. In turn, this occlusion creates a pressure wave that propagates in the spinal subarachnoid space to compress the spinal cord and cause a syrinx to enlarge with every heartbeat. After posterior fossa decompression, craniocervical CSF flow increases while peak CSF pulse pressure decreases [31], correlating with an eventual decrease in syrinx size [30].

Cardiac Gating to Improve MR Signal

Early motion-sensitive MRI techniques (primarily developed for blood flow applications) were plagued by variable signal loss within the cardiac cycle. This was rectified by the addition of cardiac gating. Since CSF flow is pulsatile and synchronous with the ­cardiac cycle, these technical improvements increased image sensitivity of CSF as well [32]. Cardiac gating was initially applied to routine spin-echo and gradient-echo MRI, which displayed CSF motion as decreased signal intensity resulting from dephasing and washout of moving spins [33].

Cardiac-Gated Phase-Contrast MRI

Phase-contrast MRI (PC MRI) is a dynamic imaging technique in which signal contrast is generated between flowing and stationary nuclei by sensitizing the phase of transverse magnetization to the velocity of motion. Two data sets with opposite sensitization are acquired. For stationary nuclei, the net phase is zero, which eliminates their signal in the final image, leaving only the residual signal from flowing CSF. The resultant signal contains information that can generate velocity data according to an intensity grayscale. Quantitative CSF velocity and qualitative flow information can be obtained by merging information from two series, usually axial and sagittal planes [34]. Further detailed result analysis using complex cardiac gating can be provided to increase sensitivity. Cardiac-gated PC MRI has become a widely used technique in CSF studies, as it allows quantification and is more sensitive to areas of slow flow [33].

CSF Flow Dynamics and Symptoms

Several cine MRI investigators have proposed that the added value of flow studies is that symptoms may be related more to the degree of CSF obstruction than the degree of tonsillar herniation, which would potentially aid in the selection of patients who are likely to benefit from surgical correction [34, 35]. Unfortunately, although several attempts have been made to determine the differences in flow parameters between symptomatic and asymptomatic Chiari I patients, flow perturbation at the foramen magnum has been unable to fully account for the significant differences in symptomatic states often observed between patients with nearly identical tonsillar anatomy. Another significant factor is that there may be considerable subjectivity in reading cine MRIs. In a 2007 study from our group, in which we asked several neuroradiologists to evaluate the same flow images in a blinded fashion, we found that the readers were more likely to agree on the presence of abnormal foramen magnum flow in symptomatic patients than in asymptomatic patients (76 % vs. 62 %) [36]. However, agreements between pairs of readers were rather low, ranging between 44 and 63 %. Of course, one would expect little disagreement when the tonsillar anatomy is very abnormal or nearly normal, with flow that is either severely restricted or near normal, respectively. In such situations, anatomical images clearly reflect the lack or presence of pathology, making flow analysis a less useful adjunct. Accordingly, most disagreements seem to occur in “gray zone” cases, i.e., in patients with moderate flow perturbation. To date, unfortunately, MRI flow analysis has been unsuccessful at separating symptomatic and asymptomatic patients under these conditions [36]. Similar investigations have been conducted to analyze CSF flow in syringomyelia. Although useful information was obtained with regard to flow velocities around the cyst cavity, these studies had equally negligible impact on understanding the pathophysiology of the anomaly or determining the need for surgery [37, 38].

Measuring Intracranial Compliance on MRI

The utility of PC MRI in the postoperative assessment of the Chiari I patient is illustrated in the work of Alperin et al., who aimed to visualize and quantify pulsatile blood and CSF flow in the craniospinal region in an effort to derive a system that determines both intracranial compliance (ICC) and intracerebral pressure (ICP) before and after posterior fossa decompression. Preliminary data suggested that intracranial compliance, as measured by these investigators, is diminished in Chiari I patients compared to healthy volunteers. Again, more work is required before this effort can result in the identification of an important diagnostic tool for guiding the treatment of patients with the Chiari I malformation [31, 39].

Analyzing Flow Velocity Voxel by Voxel: Bidirectional Flow and Velocity Jets

Analytical techniques using computer algorithms based on cardiac-gated PC MRI images can be used to generate spatial and temporal velocity plots that illustrate the qualitative and quantitative characteristics of particular regions. In a 2004 study, our group analyzed voxel by voxel the velocity of CSF throughout the cardiac cycle in both systole and diastole, and corresponding surface contour and time-course color plots were displayed. This work showed that, in Chiari I patients, there was a preponderance of regional flow jets with significant elevations in flow velocity (Fig. 13.2). These jets, which occurred primarily anterior to the spinal cord, comprised only a small percentage of the voxels, which meant that the average flow velocity across the foramen magnum was normal. In addition, select regions within the foramen magnum of Chiari I patients exhibited synchronous bidirectional flow, i.e., CSF that travels simultaneously in the cranial and caudal directions. Such bidirectional flow was obviously absent in volunteer subjects [40, 41]. An excess of 50 qualitative and quantitative parameters was designed to assess the temporal and spatial heterogeneity within the foramen, of which four were found to be particularly useful in separating Chiari patients from control subjects. Still, our studies have not so far been able to identify parameters that specifically distinguished between symptomatic and asymptomatic states within the Chiari I population.

Fig. 13.2
figure 00132

Example of CSF flow velocity plots. Color plot of CSF flow velocity in the foramen magnum of a child with Chiari I malformation. The velocities are color-coded and displayed in consecutive images representing 14 time points during the cardiac cycle (last plot represents the throughput). Note that there is a predominance of cephalad flow in seven of the images and caudad flow in the other seven images. This color plot of velocities displays the cephalad velocities in green, yellow, and red, with green being slowest and red being fastest; caudad flow is ­displayed with light blue, deep blue, and violet/black, with light blue being slowest and violet/black fastest. In children, Chiari I malformation jets (single arrow) of elevated velocities occur in the anterior quadrants of the foramen magnum (note the red color for velocities nearing 10 cm/s). Finally, note the bidirectional flow evident in some of the images (double arrows), in which cephalad and caudad velocities coexist at one time point. Such bidirectionality of flow was not present in healthy (control) subjects

Full size image

Computational Fluid Dynamics

Over the past decade, a number of physicists and engineers in the fields of biomechanics and fluid dynamics showed interest in developing other modern noninvasive methodologies to study Chiari I and syringomyelia. This recent surge in enthusiasm among non-clinicians seems to have resulted largely from a serious effort by the Chiari and syringomyelia societies (the American Syringomyelia and Chiari Alliance Project, the Chiari and Syringomyelia Foundation, Conquer Chiari, and others) to enlarge the scope of research to specialists outside of neurosurgery by providing grant money and forums for discussion. This culminated in several multidisciplinary research conferences, including the first CSF Hydrodynamics Symposium (Zurich, 2011) organized and attended almost exclusively by engineers and physicists. A major part of this effort was spent on applying principles of computational fluid dynamics (CFD) to the Chiari and syringomyelia pathologies (as well as hydrocephalus). This consists of hydrodynamic modeling of the anatomic region of interest (e.g., foramen magnum) and the prediction of physical interactions between its components (Fig. 13.3). CFD allows characterization of CSF dynamics at a more advanced level as it derives its model from the anatomical and flow data obtained from PC MRI. By applying knowledge and equations from fluid dynamics to simulate CSF flow under normal and pathologic conditions, the goal is to provide greater temporal and spatial resolution [43, 44]. Such modeling technology can be individualized to the patient’s specific parameters and could potentially greatly advance the noninvasive armamentarium aimed at improving treatment and surgical planning for these and other conditions.

Fig. 13.3
figure 00133

Example of computational fluid dynamics. The spatial distribution of velocity magnitude is shown in the spinal canal below the foramen magnum for a healthy volunteer and a Chiari I patient with significant tonsillar herniation. The velocity distribution is shown at two different time points in the cardiac cycle: peak systolic flow and peak diastolic flow. These distributions of velocity were calculated using the Navier-Stokes equations under the assumption of rigid walls. The geometry and CSF flow waveform were obtained using magnetic resonance imaging data. Note that the velocity is much larger in the Chiari patient due to a reduction in the cross-sectional area of the spinal canal (Courtesy of Frank Loth, PhD [42])

Full size image

Conclusions

It is becoming evident that the simple descent of cerebellar tonsils to the foramen magnum may not be sufficient in creating a pathophysiological disturbance that fully explains Chiari I symptoms and syrinx formation. In fact, syringomyelia can develop as a result of foramen magnum abnormalities without obvious tonsillar herniation [45]. In addition, mere tonsillar descent can create flow alterations at the foramen magnum without the onset of either symptoms or syringomyelia. Meaningful advances in research aimed at improving patient care require close correlation between symptom states and imaging advancements. This requires close collaboration between radiologists, medical physicists/engineers, and neurosurgeons. Novel imaging and simulation tools aimed to improve diagnosis and treatment will derive from these collaborative efforts.