Pulse waveform

Our own experience indicates that pulse amplitude helps in interpretation of recordings in cases of disturbed CSF compensation, shunt blockage, or slit ventricles. It also helps in distinguishing postural changes and vasogenic ICP elevations during overnight monitoring.

We do not interpret pulse amplitude alone as a single parameter influencing management. It is interpreted in conjunction with the clinical picture, neuroimaging evidence, neuropsychological tests, etc.

Detection of pulse amplitude proves that the ICP waveform is properly transmitted to the transducer. Lack of amplitude may implicate invalid pressure recording.

Pulse amplitude of ICP is synchronized with pulse amplitude of arterial pressure and the pulse amplitude of blood flow velocity in the middle cerebral artery (Figure 14.2).

Pulse amplitude increases proportionally to mean CSF pressure during the infusion study [8]. All slow vasogenic waves of ICP produce correlated variations in amplitude. In contrast, in patients with predominant brain atrophy, the rise in AMP is very sluggish and vasogenic variations are absent.

The slope of the AMP/P line in our group analysis does not correlate with magnitude or frequency of B waves, contrary to results suggested previously [9].

Similarly, in our experience, the slope of the AMP/P line may correlate with outcome after shunting for NPH, according to recent reports [10].

Correlation of pulse amplitude with the brain elastance coefficient is insignificant.

In our own experience, pulse amplitude is not a strong predictor of outcome after shunting; however, other authors advocate such a relationship [11].

The interesting theory that an increase in pulse amplitude may reflect insufficient cerebral “Windkessel” seems to need more convincing experimental and clinical evidence. Increased pulse amplitude of ICP has been previously reported to be responsible for dilatation of the ventricles [12]. However, in idiopathic NPH there is no evidence of any association between pulse amplitude and size of the ventricles.

The cerebral “Windkessel effect” helps in damping of arterial blood pulse pressure over the cardiac cycle and assists in the maintenance of brain perfusion during diastole when cardiac ejection ceases. In brain, arterial pulsations are then progressively dissipated to render the capillary circulation smooth, almost pulseless. The Windkessel effect becomes diminished with age as the elastic arteries become less compliant. Breakdown of this phenomenon increases pulsation in the subarachnoid space and this causes an increase of pulsation in the choroidal artery and plexus, thus increasing the CSF ventricular pulsation amplitude. The new increased amplitude of ventricular CSF pressure exceeds the amplitude of pulsation of CSF in the subarachnoid space. This causes a pulsating transmantle pressure gradient, possibly explaining ventricular dilation [13]. Moreover, it is likely that the loss of the Windkessel effect results in an increased pressure in the subarachnoid capillary circulation causing narrowing or reversal of the CSF/venous gradient, responsible for CSF malabsorption. Some scientists suggest that the elevation of capillary and venous pulse pressure that occurs as a consequence of redistribution of pulsation may diminish the hydrostatic gradient necessary for CSF absorption.

In conclusion, “Windkessel theory” says that communicating hydrocephalus may be a disorder of intracranial cerebral blood flow (CBF) pulsation – but it still requires more comprehensive evidence.

On the other hand, it is also likely that CSF and brain pulsations are not only influenced by compliance of arterial walls and the brain “Windkessel” mechanism, but more importantly, by the subarachnoid space patency to compensate for the effects of brain expansion as a CSF pulsatile movement as seen using phase-coded MRI [6]. Only if CSF can move freely in basal cisterns and get out from there, can Windkessel dampening of the arterial pulse work properly.

Injection of kaolin into the basal cisterns and subarachnoid hemorrhage in the basal cisterns following aneurysmal rupture both create hydrocephalus by blocking this CSF pulse movement by clogging up basal subarachnoid spaces.

Correlation of pulse amplitude with R_out is positive and significant although weak (correlation coefficient in a range from 0.2 to 0.3). There is no difference in amplitude between males and females. Pulse amplitude increases slightly with age but does not show any correlation with duration of symptoms of NPH or their severity (measured using NPH scale). Pulse amplitude is much lower in idiopathic NPH (iNPH) patients without any evidence of coexisting cerebrovascular disease than in those patients with clear evidence of vascular problems. After shunting, pulse amplitude decreases.

While mean ICP potently reacts to a change of posture (usually becomes negative in an upright body position), the pulse amplitude usually does not follow postural ICP variations.

In analysis of pulse waveform, peak-to-peak pulse values [8,11] or first harmonic of pulse waveform are usually taken into account. This is probably an oversimplification. Morphological analysis of pulse amplitude usually indicates several peaks: P1, P2, and often P3. Peak P1 is usually attributed to passive transmitting of the arterial pulse wave through arterial walls (percussion peak). Peak P2 is associated with transport of arterial blood volume along arteries, which produces a pressure–volume response. Peak P2 increases when compliance of the brain is reduced; therefore, the so-called P2/P1 ratio was historically interpreted as a dimensionless index of brain compliance. Automatic detection of the separate peaks is complex [14], but their continuous descriptors seem to be promising in interpretation of pulse waveform of ICP.

Finally, pulse amplitude in ICP recording may be weak and noisy. Proper computer detection is helpful in most cases.

Assessment of B waves

According to classical standards [15], when so-called “B waves” were present for more than 80% of the duration of ICP monitoring, shunting was recommended. However, using computer detection, B waves (slow waves of ICP of periods from 20 seconds to 2 minutes) are almost universally present, probably even in healthy volunteers. There are no data derived from normal subjects, as a measurement of ICP is invasive. A variable has been proposed, being an amplitude of sine wave bearing the same energy as B waves, calculated using spectral analysis. This amplitude has to remain greater than 1 mmHg for a duration longer than 15 minutes to qualify as a sign of pathological level of “B waves” [16]. However, in an original paper by Børgesen and Gjerris [15], recordings were classified for the “presence” of B waves when they reached a much larger amplitude (5–10 mmHg).

B waves are coherent with fluctuations of cerebral blood flow velocity (see Figure 14.2) and near infrared spectroscopy. The average energy of slow waves did not correlate well with R_out in those patients in whom both overnight monitoring and infusion test were performed (our own study). This result is quite disturbing: it may be explained that the survey was conducted in a limited group of “difficult” patients, for whom the results of infusion study indicated a “borderline” state of CSF compensation or the picture was dramatically divergent from the clinical or radiological findings. The utility of B waves in prognostication in hydrocephalus should be readdressed, preferably by multicenter trial.

Cerebrospinal pressure–volume compensatory reserve – RAP index

Conventionally, pressure–volume compensatory reserve is assessed using intracranial volume addition [7]. Changes in ICP in response to a known volume change allow such parameters as the pressure–volume index (PVI) to be derived from bolus volume addition or E(which characterizes the shape of the pressure–volume curve over its exponential region) using constant rate infusion. However, under certain assumptions, external volume addition is not necessary as it is possible to assess pressure–volume compensation by taking into account the change in pressure with every heart beat, where a volume of arterial blood is added to the cerebrospinal space in a pulsatile manner. Although the added volume is not known and may be variable, the pressure response is recorded continuously in the form of the pulse waveform of the ICP recording. The RAP index (correlation coefficient [R] between the pulse amplitude [A] and the mean intracranial pressure [P]) is derived by linear correlation between 30 consecutive, time-averaged datapoints of pulse amplitude of ICP (AMP) and mean ICP acquired within a 10 second-wide time-window. RAP describes the degree of correlation between AMP and mean ICP over short periods of time (5 minutes). Theoretically, the RAP coefficient indicates the relationship between ICP and changes in intracerebral volume – the “pressure–volume” curve. A RAP coefficient close to 0 indicates a good pressure–volume compensatory reserve at low ICP, i.e. the situation when the “working point” is still below the exponential region of the curve. When the pressure–volume curve starts to increase exponentially, AMP co-varies directly with mean ICP and consequently RAP rises to +1. This indicates a low compensatory reserve [17].

Cerebrovascular pressure reactivity (PRx)

Another useful ICP-derived variable is the pressure reactivity index (PRx), which is based on the concept of assessment of cerebrovascular pressure reactivity by observing the response of ICP to spontaneous fluctuations of ABP. Using computational methods similar to the calculation of the RAP index, PRx is the correlation coefficient between 30 consecutive, time-averaged datapoints of ICP and ABP. A positive PRx signifies a positive gradient of the regression line between the slow components of ABP and ICP, which has been shown to be associated with passive behavior of a nonreactive vascular bed. A negative value of PRx reflects normal reactive cerebral vessels, as ABP waves provoke inversely correlated waves in ICP. This index correlates well with indices of autoregulation based on transcranial Doppler ultrasonography (see below).

Example of clinical use of ICP monitoring: pediatric hydrocephalus

Pediatric hydrocephalus is quite different in most aspects from adult hydrocephalus, especially NPH. Most cases of pediatric hydrocephalus have a clear obstructive component regarding etiology (Chiari malformation, post-hemorrhagic, post meningitis, aqueductal stenosis, fourth ventricular outflow obstruction), even if obstruction can often only be detected using high resolution MRI techniques [18]. Despite the fact that obstruction is classically associated with pressure active hydrocephalus presenting typical signs and symptoms of raised intracranial pressure, children also exhibit pressure compensated hydrocephalus. They also might undergo shunt dysfunction or ETV failure without becoming symptomatic in the sense of the above-mentioned classical signs and symptoms of raised ICP. In these cases of compensated hydrocephalus or occult treatment dysfunction, ICP overnight monitoring and shunt infusion studies have been shown to be very helpful. Figure 14.3 shows a typical example of elevated baseline ICP, exaggerated nocturnal ICP dynamics during REM sleep, increase of RAP during these episodes indicating exhaustion of cerebrospinal reserve capacity, and elevated baseline AMP in a totally asymptomatic child with known ETV re-closure.

Figure 14.3 Example of ICP overnight monitoring during sleep of a 5-year-old boy with known Blake’s pouch (outflow obstruction of the fourth ventricle) and large ventricles, who was found to have complete shunt obstruction despite being totally asymptomatic. Because of very large ventricles he received ETV as an alternative treatment instead of shunt revision which was found to be re-closed after 6 months. He was evaluated regarding shunt-independent ventriculomegaly. His baseline ICP was elevated at 20 mmHg; during REM sleep induced vasogenic wave periods he encountered overshooting peak ICP values (minute average values) of almost 40 mmHg. RAP rose during these episodes almost to 1 indicating exhaustion of reserve capacity. Baseline AMP was slightly increased at about 1.5 mmHg and rose during vasogenic wave periods to values of about 5 mmHg. In summary highly abnormal values were recorded and a shunt was reinserted. Ventricle size thereafter decreased considerably.

R_out and baseline CSF pressure

Infusion study (constant rate [20], or any other variation of controlled but variable rate [21]) allows indices describing the state of CSF compensatory reserve to be estimated. Briefly, there are two types of infusion methods: the steady-state studies (constant pressure infusion [21] and constant flow infusion [20]), and the “dynamic study” (bolus infusion [7]). Although each examination has its advantages and drawbacks, they are all based on the same theoretical background inspired by the mathematical model developed by Marmarou et al. [7]. Traditionally, the two most important parameters are resistance to CSF outflow (R_out) and baseline CSF pressure. Elevated R_out (>13 mmHg/[ml/min] [15] or >18 mmHg/[ml/min] [22]) denotes disturbed CSF circulation. Elevated baseline pressure (>18 mmHg) may signify an uncompensated cerebrospinal volume-expanding process.

However, there is still controversy about the value of R_out for predicting improvement after shunting, which shows an interesting timeline. In 1992 Børgesen et al. [19] indicated 100% both positive and negative predictive power for an R_out threshold of 13 mmHg/(ml/min). A Dutch study in 1997 [22] showed 97% positive predictive power for R_out >18 mmHg/(ml/min) but only 36% negative predictive power. In the 2005 “NPH guidelines” Marmarou and coworkers [1] recognized sensitivity/specificity of R_out above 80% (this was done on a meta-analysis of sizable published studies). However, in 2012 the so-called “European Trial” [23] indicated no association of R_out and improvement after shunting. The trial was not very large (around 100 patients), conducted in divergent centers, and directly sponsored by one big shunt manufacturer. Nevertheless, the results were spectacularly different from historical data. Are we managing different patients than in the 1980s and 1990s? Or do we tend to estimate R_out with greater variation?

A further difficulty when trying to determine the predictability of R_out arises from the fact that different centers use different infusion methods to calculate it, and this can lead to interpreting in the same manner cut-off levels that are not necessarily the same. It has been stated that for unknown reasons [1], physiologic R_out as calculated by using the bolus infusion, results in systematically and significantly lower values (<4 mmHg/ml/min) than those determined by steady-state infusion methods. Interestingly, a recent paper from Sundström et al. [24] that aimed to compare values of R_out determined from the bolus, the constant-flow, and the constant-pressure infusion methods demonstrated that in experimental conditions, with well-known characteristics, R_out values showed good concordance among all methods. These findings confirm the robust theory of the mathematical model. Notwithstanding, when shifting to a clinical setting R_out shows clear and significantly different values between steady-state and bolus methods, as already reported in previous studies [25]. The key to understanding such differences lies in the presence of physiological pressure variations introduced when applying this model in humans (e.g. breathing, cardiac variation, and so on). The first big difference with the bolus infusion when compared to the steady-state methods is the magnitude of the created “peak pressure” that together with the volume of injected fluid permits calculation of the PVI (pressure–volume index). The bolus itself allows a wide range of increase in baseline pressure, while constant-pressure and constant-flow infusions cause much less variation. Moreover, bolus R_out values are calculated using the PVI together with the new pressure selected on the return curve at different time points. Then PVI and momentary chosen pressure on the pressure trace decaying after bolus back to baseline values determine the final R_out value. Any error in the estimation of PVI will affect the R_out estimation. At the same time, especially when resistance to outflow is high, the pressure differences on the return curve are much smaller, limiting the signal processing options for extracting R_out-related data from the curve and therefore making the method more susceptible to physiological variations. Finally, the R_out values obtained from each of these time points is averaged to yield one R_out value for each bolus. The bolus injection is then repeated at least three times and the R_out values for all three injections are averaged to yield the final R_out value for the patient. All these considerations could explain the difference in R_out between methods.

The “dynamic” response of the bolus method should be seen as an advantage, since potentially we measure a more “physiological” R_out; however, this test might require an operator with extensive experience, while the better repeatability of the steady-state methods makes them the recommended approach for estimating R_out, at least in iNPH.

A different situation, where a more dynamic physiological response might be needed, could be represented by the extremely heterogeneous population suffering from secondary forms of hydrocephalus (post-traumatic, post-hemorrhagic, post-infectious hydrocephalus, etc.). Here, determining opening pressure, brain compliance (PVI), and R_out in a relative short time might represent a true clinical advantage to distinguish between atrophy and hydrocephalus as the two possible causes of “secondary” ventriculomegaly. Here are some examples: an elevated baseline ICP (>15 mmHg) with slow return of ICP to baseline after the bolus injection indicates an elevated CSF outflow resistance (bolus R_out >5 mmHg/[ml/min]) in patients who should be promptly shunted. A normal baseline ICP with the bolus injection causing a relatively smaller increase in ICP, indicating an enhanced PVI, might reveal true brain atrophy. On the other hand, when the bolus injection causes a relatively larger increase in an initially normal baseline ICP, with rapid return to baseline, it might indicate exquisitely low brain compliance, and the need for a “low pressure” shunt. If the same peak pressure after the bolus returns to the baseline ICP at a much slower rate than normal, this suggests an elevated R_out (as occurs in iNPH) and most likely the need for an adjustable valve, since in this case there are more parameters subjected to variation after shunt that might require several changes of the valve setting.

Elastance coefficient (or elasticity)

The exact clinical significance and interpretation of the elastance coefficient (E) is not yet fully documented, though theoretically it describes brain stiffness. Tans and Poortvliet [26] showed that E weakly correlated with the resistance to CSF outflow, although no further studies exploring this concept have followed. Tisell et al. [27] demonstrated that E correlated positively with the result of third ventriculostomy: those with a stiffer brain (higher E) have a better chance of improving after surgery.

Analysis of slow waves, compensatory reserve, and pressure reactivity during infusion study

Parameters describing vascular effects and pressure–volume compensation can also be evaluated during the infusion study. The study usually starts with 10–15 minutes of baseline assessment, which can be easily extended. Consequently, the same parameters as during an overnight ICP monitoring can be calculated.

Of all the CSF compensatory parameters derived from the infusion test, the resistance to CSF outflow demonstrates significant associations with cerebrovascular reactivity (PRx): patients with lower R_out tend to have more frequently disturbed cerebrovascular reactivity.

Also the results of correlation analyses between R_out and pulse amplitude of ICP, energy of slow waves, and respiratory wave at baseline demonstrate significant associations. Increases in all vasogenic components during infusion as compared to the baseline values are commonly observed [28]. Practically these indices are used as an additional descriptor of CSF dynamics and help in understanding CSF circulatory disorders.