Cerebrospinal fluid flow in normal beagle dogs analyzed using magnetic resonance imaging

The cerebrospinal fluid (CSF) is a liquid that fills the ventricles, central canals, and subarachnoid spaces (SAS). It is produced in the choroid plexus and brain parenchyma and is absorbed from the arachnoid villi. CSF removes waste and solvents and keeps the fragile brain tissue floating in buoyancy to maintain its shape and protect against external shocks [1, 2, 3]. CSF circulation is divided into 2 types of motion. The first is bulk flow, which is based on a hydrostatic pressure gradient primarily between the choroid plexus and arachnoid granulations. It is slow enough to be difficult to recognize using magnetic resonance imaging (MRI). In contrast, pulsatile flow is more likely to be detected on MRI because it is faster than bulk flow. It is based on the Monroe-Kellie doctrine, which states that the sum of blood, CSF, and brain parenchyma volumes is constant, limited by the volume of the skull. Thus, during contraction of the heart, CSF shows backward flow. Conversely, it moves forward during diastole of the heart [4].

Phase-contrast (PC) MRI is a specific sequence used for imaging CSF flow. It generates a signal contrast between flowing and stationary nuclei by sensitizing the phase of the transverse magnetization to the velocity of motion [5, 6, 7]. A key aspect of PC-MRI is the bipolar gradient, which is placed after the radiofrequency pulse and before data collection during echo time (TE). The principle of the bipolar gradient is to develop a phase shift of spins moving with a specific velocity along the direction of the axis [4]. PC-magnetic resonance (MR) images consist of the initial magnitude images, which represent flowing CSF as a bright signal and stationary tissues as a black background, and the subsequent phase-shift-encoded images with white and black signals representing forward and backward flows, respectively. Because of their phase-dependence, these images contain velocity information, which can be quantitatively estimated. Thus, CSF flow during the cardiac cycle may be displayed as a cine loop. Currently, PC-MRI is the best-known method and the only technique for both qualitative and quantitative assessment in clinical scenarios such as differential diagnosis and postsurgical follow-up [4]. However, this technique is limited by its poor visualization of turbulent flow and the inability to measure bulk flow [8, 9, 10].

Another sequence that can complement PC-MRI is time-spatial labeling inversion pulse (Time-SLIP). This is a variant of arterial spin-labeling that produces high-quality angiograms. The application of arterial spin-labeling lends itself very well to imaging CSF movement by using CSF as its own tracer. This approach can detect not only pulsatile flow but also bulk flow in slow-flowing regions. Unlike PC-MRI, quantitative evaluation is not possible with this technique, but it offers the advantage of a high signal-to-noise ratio and provides additional information about CSF flow patterns [11]. On the basis of the results of initial baseline studies, Time-SLIP can be readily used as a screening tool to determine whether the CSF flow is typical, atypical, or absent [12].

In human medicine, there are many reports that have attempted to assess CSF flow-related pathophysiology by measuring velocity and visualizing the flow of CSF with MRI. In one study, normal-pressure hydrocephalus patients showed a significantly higher peak systolic velocity of CSF than the control group [13]. Another study revealed a shorter systolic flow period of the temporal waveform for patients with communicating hydrocephalus in comparison with healthy controls [14]. For normal volunteers, the peak systolic velocities in the SAS at the foramen magnum level were measured. In symptomatic patients with Chiari I, peak systolic velocity was significantly higher than those in control volunteers [15], and CSF flow was characterized by flow jets, regions with a preponderance of flow in one direction, and synchronous bidirectional flow [16, 17]. Additionally, one recent case report confirmed the patterns and directional changes of CSF flow in syringomyelia by using Time-SLIP [18].

Diseases associated with CSF flow were considered a rare condition in veterinary medicine, but they are now a relatively common neurologic diagnosis. Large ventricles are common findings in brachycephalic dog breeds [19]. Chiari-like malformations and syringomyelia are also often recognized in small-breed dogs. These changes can be attributed to the increased availability of MRI as well as the increased prevalence of these diseases in certain breeds, most notably the Cavalier King Charles Spaniel [20]. Although there is still much debate about the underlying mechanisms and treatment methods, syringomyelia is also thought to be a consequence of abnormal CSF dynamics in dogs as in humans [21]. Therefore, based on the pathophysiology of these diseases, abnormalities of CSF flow are expected to appear in MR examinations in dogs.

In veterinary medicine, however, only one study has been conducted using MRI to evaluate and compare CSF flows. This previous study reported the patterns and velocities of CSF flow in Cavalier King Charles Spaniels with or without syringomyelia by using PC-MRI [22], but it assessed just five control dogs and obtained measurements only at the rear of the foramen magnum.

Therefore, in this experimental study, CSF flow velocity data was obtained from the cerebral aqueduct and SAS at the foramen magnum level using PC and Time-SLIP MRI in healthy beagle dogs. The purpose of this study was to provide baseline data of CSF flow velocity in dogs.

PC-MR and Time-SLIP imaging were successfully performed in all dogs over a full study period lasting approximately 40 min/dog. CSF flow was detected in all ROIs, and velocity measurements were obtained.

Time-SLIP

The signal changes indicating CSF movement in the SAS and cerebral aqueduct were displayed as a cine loop, which was converted into color to obtain better performance with the postprocessing technique on the MR system (Fig. 2). In the conversion image, red and green represent the bright regions in the primary image, which is interpreted as tagged CSF. In this experiment, the range of red areas was shown to be increased in the dorsal SAS and cerebral aqueduct. In the ventral SAS, tagged CSF was observed extending the caudal part with a decrease and increase in green areas.

Fig. 2
Sagittal Time-SLIP MR images after a transverse labeling pulse show the contrast difference between tagged and untagged CSF at various TIs. The image is converted to color to maximize the expression of light and shade. Scale bar of primary images (A) and converted images (B). Labeled CSF in the cervical arachnoid space (C) and cerebral aqueduct (D) enters the caudal SAS and 4th ventricle. The red and green areas, indicating tagged CSF, increase in the dorsal SAS and cerebral aqueduct (arrows). In addition, in the ventral SAS, the green areas are seen extending backwards (arrow heads).
Time-SLIP, time-spatial labeling inversion pulse; MR, magnetic resonance; CSF, cerebrospinal fluid; TI, inversion time; SAS, subarachnoid spaces.

• Click for larger image
• Download as PowerPoint slide

PC-MRI

The mean waveforms of 6 dogs' CSF flow studies were plotted with velocities in centimeters per second on the y-axis and fractions of the cardiac cycle on the x-axis (Fig. 3). The positive deflections represent the craniocaudal flow, and the negative deflections represent caudocranial flow.

Fig. 3
CSF flow waveforms of 6 dogs in three ROIs. The positive deflections represent craniocaudal flow and the negative deflections represent caudocranial flow.
FM, foramen magnum; CSF, cerebrospinal fluid; ROIs, regions of interest.

• Click for larger image
• Download as PowerPoint slide

In the ventral part of the cervical SAS and cerebral aqueduct, the CSF was characterized by bidirectional flow throughout the cardiac cycle. This can be seen by the time point of zero velocity, which indicates a change in the direction of CSF flow and was noted twice and three times, respectively. The point at which positive values begin after the R wave indicates the start of the systolic period. This time tended to be faster in the ventral SAS than in the cerebral aqueduct. In both the ventral SAS and cerebral aqueduct, CSF flow waveforms were characterized by relatively narrow systolic peaks than the diastolic troughs. Therefore, the diastolic slopes were less steep than the systolic slopes. The systole duration was shorter than that of the diastole. The diastole duration could be measured at the point where the third zero point in the CSF waveform at the cerebral aqueduct begins with the second contraction of the heart. However, the waveform in the ventral SAS was assumed to be greater than that since the third zero point did not appear within 576 ms. On the other hand, the waveform in the dorsal SAS revealed no greater to-and-fro movement than anywhere else. Moreover, it showed a tendency to flow steadily at a similar flow rate in the present study.

The mean ± SD peak velocity was 1.39 ± 0.13 cm/s through the ventral aspect of the SAS, 0.32 ± 0.12 cm/s through the dorsal aspect of the SAS, and 0.76 ± 0.43 cm/s through the cerebral aqueduct. The peak velocities at each region differed significantly from each other. The values for each dog are summarized with their respective 95% confidence intervals in Table 1.

Table 1
Peak CSF flow velocity of 6 Beagle dogs at three ROIs on PC-MRI.

• Click for larger image
• Click for full table
• Download as Excel file

The results were analyzed on the basis of ICCs (Table 2). The assessments confirmed that the repeated measurements were reliable. ROIs at all locations also showed high agreement in the analysis of the inter-observer agreement.

Table 2
Intra- and inter-observer agreement evaluation of peak velocity at 3 ROIs

• Click for larger image
• Click for full table
• Download as Excel file

In this study, it was possible to assess CSF flow in all dogs by using Time-SLIP and PC-MR imaging, allowing noninvasive quantitative evaluation of velocity and qualitative evaluation of the flow pattern. First, by using the time-SLIP technique, we investigated whether it was possible to identify communication between the CSF spaces. We also could get a velocity values measured using PC-MRI. In addition, it was confirmed that the flow showed a normal reflux pattern, not a turbulent flow that can be seen in an abnormal state on PC-MR images.

The CSF velocity values were compared with those in a previous veterinary study [22] conducted on 59 animals with neurological symptoms of the Cavalier King Charles Spaniel breed, which is predisposed to Chiari malformations that are responsible for syringomyelia. For comparison, four beagles and one mixed dog were included in the control group. The measurement locations for the analysis were foramen magnum level, with the mean of peak velocities for the ventral and dorsal aspects of SAS, based on normal dogs included in the control group, being 0.75 ± 0.24 cm/s and 0.59 ± 0.13 cm/s, respectively. Their results were not consistent with ours. This discrepancy may be attributable to multiple factors including scan parameters, the positional difference of the neck, anesthetic agents, and location and size of ROIs. Infants are known to have higher CSF flow velocities than adults in humans. This is thought to be due to arterial pulse wave variations with age [23]. Since both the previous and present studies used adult dogs, age would not have affected the results. Studies assessing the effects of breed and/or body weight on CSF flow have not yet been performed, but since dogs of similar breed and body weight were used, these factors also cannot explain the differences in results. In addition, we controlled the heart rate to ensure CSF flow within a range of 60–120 bpm, as in the previous study.

A CSF flow pattern of to-and-fro movement resulting from heart contraction and relaxation was observed in the ventral SAS and cerebral aqueduct. Humans have been reported show pulsatile waves in both the dorsal and ventral SAS, but some normal individuals may also show loss of this movement. In this study, there was no clear contrast change in the dorsal SAS at the time of drawing ROIs for measurement in PC-MR images, and no obvious to-and-fro movement was observed during the measurement as a result. This may be due to the head angle. In a previous study, the head angle was set in the range of 100° to 138° in a similar standing position. In this experiment, however, the dogs were positioned in basic ventral recumbency without flexion of the neck. This may have made the dorsal SAS smaller than the neck flexed position [22]. In fact, Sofia et al. also claimed that when scanned with the neck in extension and in flexion, the dorsal aspect of the SAS was particularly large in the flexion state.

Although the effects of changes in CSF flow from drugs have not yet been shown, anesthetics are thought to change physiological conditions directly related to CSF flow, such as heart rate and blood pressure. In a previous study [22], each dog was premedicated with fentanyl, and anesthesia was induced with propofol and maintained using inhaled sevoflurane. Since these drugs do not match the anesthetic drugs in the current study, it is assumed that the results may change. Alfaxalone, which was used for anesthesia in this experiment, is known to produce dose-dependent changes in the cardiovascular system [24]. According to previous studies, transient tachycardia and hypotensive effects have been noted immediately after drug administration when larger doses are administered [25]. Although this effect is minimal at a dose of less than 6 mg/kg [24], it is impossible to say that there was no effect in this experiment. In this respect, the subjective measurement of blood pressure in this experiment is a limitation.

It is important to consider that, depending on the situation, variations can be introduced by the measurement method. Plane selection and manual ROI placement in relation to the anatomic region are also known to affect the overall obtained values. In particular, for very narrow aqueducts, this error may be even higher because noise and poor contrast make the placement of the ROI difficult [26]. The flow in the center is higher than at the periphery, so a small ROI placed in the center of the aqueduct will yield higher readings of mean velocity, and a larger ROI will underestimate it. However, the peak velocity representing the fastest flow in any pixel in the given ROI is not affected by the ROI size [13].

One limitation of PC-MRI is that proper setting of the recording time is required through adjustment of TR and phase number. Since the heart rate of dogs is approximately twice that of humans, TR and recording time were set at half values. However, in the ventral SAS, it was difficult to measure the exact duration of the diastole period because the second onset of systole was not recorded. Therefore, the findings are thought to be more accurate if the recording time is set such that it can be increased. Also, this study was performed only in same purpose-bred dogs with similar size.

A uniform relative evaluation is difficult because each individual has a different heart rate, measured ROI, etc., and there is no method to analyze the amount of CSF produced, but it is thought to be meaningful in understanding the direction of the overall circulation. The results of this study confirmed that normal net flow rate/cycle would, in most cases, flow positive from the third to the fourth ventricle.

In conclusion, studies of CSF circulation using MRI with visible images of CSF circulation are a useful method for understanding CSF flow dynamics. We obtained a reference data set of CSF flow peak velocities through the cervical SAS and cerebral aqueduct in healthy dogs. This approach is thought to be especially important considering the possibility of various errors occurring in these assessments. Therefore, specific guidelines for CSF MR examination are crucial to obtain comparable and reliable data across different studies. In that sense, the results of this study provide valuable data for future research on CSF flow disorders such as hydrocephalus, Chiari-like malformation, and syringohydromyelia. A large-scale future study is also needed to establish a standard technique for normal CSF flow in the future.