Non-invasive Estimation of Pulsatile and Static Intracranial Pressure by Optical Coherence Tomography

This study examined whether optical coherence tomography (OCT) can be used to estimate the pulsatile and static intracranial pressure (ICP) in patients with idiopathic intracranial hypertension (IIH). Continuous overnight monitoring of the pulsatile and static ICP and a single OCT examination were performed in 20 patients with IIH and in 12 reference subjects without cerebrospinal uid disturbances. The ICP measurements were obtained using a parenchymal sensor. The pulsatile ICP was determined as the mean ICP wave amplitude (MWA) and the static ICP was determined as the mean ICP. The IIH group had increased overnight MWA and mean ICP scores, and OCT demonstrated anterior deection of the peripapillary Bruch’s membrane angle (pBA) and increased optic nerve head height (ONHH). There was a signicant positive correlation between the pBA and ONHH and the overnight MWA scores, and between mean ICP and ONHH. The pBA and the ONHH also predicted abnormal overnight MWA, but the predictive power of mean ICP was non-signicant after adjusting for age and body mass index. This study suggest that the OCT parameters pBA and ONHH can predict abnormal MWA scores.


Introduction
Continuous monitoring of intracranial pressure (ICP) is a cornerstone in the diagnosis and management of a wide range of intracranial disorders, such as traumatic brain injury, intracranial hemorrhage and cerebrospinal uid (CSF) disturbances. 1 Still, ICP monitoring is invasive and requires insertion of a probe or catheter into the intracranial compartment, thereby carrying a risk of complications. 2,3 A wide range of non-invasive ways of measuring the static and pulsatile components of ICP have been explored, 1,4−11 but so far no methods have been found clinically suitable.
The eye is a potential window for indirect assessment of the ICP due to the direct communication between the intracranial and intraorbital subarachnoid spaces, by which the ICP directly affects the optic nerve and eye globe. 12 Papilledema is a widely recognized indirect sign of intracranial hypertension. Recently, optical coherence tomography (OCT) has gained increasing interest as a potential tool for noninvasive ICP (nICP) assessment. [13][14][15] The OCT methodology measures the re ection of back-scattered light and allows for detection and quanti cation of even subtle morphological changes in the retina and the optic nerve. 16 In patients with idiopathic intracranial hypertension (IIH), 17 the mean ICP, measured using a telemetric parenchymal ICP probe, correlated with OCT measures of the optic nerve head.
Moreover, in a pediatric study 18 OCT measurements of the retina and the optic nerve head detected elevated mean ICP, measured with a subdural probe. However, with the exception of these studies, the published reports on estimating ICP from OCT have merely studied the relationship between OCT ndings and the lumbar CSF opening pressure, 13-15,19−22 which may differ from the ICP measured dynamically with an intracranial probe. 1,23 In addition, the previous studies only examined the static ICP, even though ICP consists of both static and pulsatile ICP. The latter results from the pressure uctuations created by the cardiac contractions. 1 The pulsatile ICP may be expressed as the mean wave amplitude (MWA) 24 and has been reported to better re ect the pressure-volume reserve capacity (or intracranial compliance) than the static ICP. 1,25 This study explored the value of OCT as a potential non-invasive tool for estimating the pulsatile ICP, determined as the MWA, and the mean ICP measured overnight in patients with IIH.

Results
Patients. The study cohort consisted of 20 patients ful lling the diagnostic criteria of IIH 26   Regarding the OCT parameters, the peripapillary Bruch's angle (pBA) was signi cantly smaller in the IIH than the REF group, demonstrating a more anterior de ected peripapillary Bruch's membrane. The optic nerve head height (ONHH) was signi cantly increased in the IIH group ( Table 1). The other OCT parameters did not differ signi cantly between IIH and REF (Table 1).
Association between ICP and OCT parameters. There was a positive correlation between the OCT parameters pBA and ONHH and the ICP scores. In particular, the pBA became increasingly smaller (anteriorly de ected) with increasing pulsatile ICP, expressed as overnight average MWA (Fig. 1a) or percentage proportion of MWA ≥ 5 mmHg (Fig. 1b). This was not signi cant for static ICP (mean ICP; Figs. 1c-d). Furthermore, there was a signi cant positive correlation between ONHH and overnight average MWA (Fig. 2a), proportion of MWA ≥ 5 mmHg (Fig. 2b), and overnight mean ICP (Fig. 2c). The proportion of mean ICP ≥ 15 mmHg did not correlate with ONHH ( Fig. 2d).
Ability of OCT to predict abnormal pulsatile and static ICP. The ability of the OCT parameters pBA and ONHH to predict abnormal pulsatile and static ICP is presented in Table 2. Both pBA and ONHH predicted abnormal MWA, with optimal cut-points of -0.65° and 405 µm. Their predictive ability remained signi cant after adjusting for age and BMI ( Table 3). The ONHH predicted mean ICP with optimal cut point of 517 µm ( Table 2), but this was non-signi cant after adjusting for age and BMI (Table 3).
Accordingly, both the pBA and ONHH most accurately predicted the MWA scores. Figure 3 displays the receiver operating curves (ROCs).

Discussion
This study investigated whether OCT may be used to non-invasively estimate the pulsatile and static ICP.
The main observations were that the OCT parameters pBA and the ONHH correlated with the overnight pulsatile ICP (MWA) scores, while the other OCT parameters investigated did not differ between IIH and REF. The optimal cut-off values for the pBA could predict elevated MWA with a sensitivity and a speci city of 75% and 92%. Similarly, the ONHH could predict elevated MWA with a sensitivity of 88% and speci city of 69%. The OCT parameters did, however, not reliably predict the static ICP (mean ICP).
This study included both a patient group with IIH diagnosed according to current criteria 26 and a reference group without CSF disturbances. The groups were comparable with respect to age and BMI. Both the pulsatile and static ICP scores were increased in the IIH group. An abnormal MWA in the majority of the IIH patients (95%) and mean ICP being elevated in less than half of the subjects (40%) compares to a previous study of IIH patients with failed conservative and medical treatment. 28 The MWA shows a better correlation with the intracranial compliance (i.e., intracranial pressure volume reserve capacity) than mean ICP, 25 and intracranial compliance can be impaired despite normal mean ICP. 29 Accordingly, IIH patients can have impaired intracranial compliance even though mean ICP is within normal thresholds. Another possible reason for the abnormal pulsatile ICP may be alterations at the capillary level, which may underlie the impaired glymphatic function (i.e. impaired paravascular molecular transport) that was recently demonstrated in IIH patients. 30 The ICP scores were derived from continuous overnight measurements. Previous studies have explored various types of continuous non-invasive source signals to estimate ICP. [4][5][6][7] Continuous monitoring of the optic nerve with OCT is, to our knowledge, not available. Yet, a single time point examination may be used to estimate the presence of normal or abnormal ICP. Thus, OCT could serve as a method to differentiate elevated from non-elevated pulsatile ICP. A de nite diagnosis of abnormal pulsatile and static ICP still requires an invasive procedure.
The pBA was signi cantly smaller in the IIH group than the REF group. A smaller angle implies a more anterior position of the peripapillary Bruch's membrane. The position of this anatomical area re ects the balance between the pressure in the perioptic subarachnoid space and the intraocular pressure. 31 Elevated ICP relative to the intraocular pressure causes an anterior de ection. Lowering ICP has been shown to reverse the anterior de ection. 19,31,32 Illustrating the highly dynamic properties of this parameter, one study observed reversing of the position of the peripapillary Bruch's membrane within one hour after lowering the ICP. 20 Previous studies have also shown anterior de ection of the Bruch´s membrane in patients with elevated lumbar CSF opening pressure. 19,21,31,33 However, contradictory to previous ndings 20,22,34 , we did not nd any correlation between the anterior de ection and the mean ICP. There are several potential reasons for this discrepancy. Both the lumbar CSF opening pressure and the mean ICP refer to the static ICP component; these pressures are not identical, as has been previously shown by simultaneous measurements. 23 Moreover, a lumbar CSF pressure is measured over a very short period and may be extensively affected by stress or body position. The ICP scores presented in this study, however, were retrieved from multi-hour overnight measurements. The patients were resting in their beds and allowed to sleep.
Our observation of a signi cant correlation between pBA and MWA indicates that the abnormal pulsatile pressure uctuations during the cardiac beat are more susceptible to cause anterior deviation of the peripapillary Bruch's membrane than a static pressure gradient. Accordingly, repetitive small cardiacrelated pressure uctuations (in the IIH group the average overnight MWA was 6.7 ± 2.0 mmHg) seem to exert a different effect on the soft tissue than a static difference in pressure (the overnight mean ICP in IIH subjects was 13.4 ± 5.6 mmHg).
We did not nd between-group differences in the RNFL. Automatic algorithm for measuring the RNFL thickness is available in all OCT devices and widely used in the diagnosis and management of papilledema. 13,14,35 Elevated ICP compresses the optic nerve and leads to axoplasmatic stasis and swelling of the retinal nerve bers. 12 However, since dead axons don't swell, a major drawback of measuring the RNFL is the potentially false impression of a normal or even thin RNFL in patients with increased ICP, if optic nerve atrophy has developed. Notably, the pBA is not affected by optic atrophy. Accordingly, it may be a more reliable biomarker for nICP estimation if atrophy has developed. Intriguing in this regard is a previous case report. 19 This suggested the anterior de ection of the peripapillary Bruch's membrane to better indicate elevated ICP than the RNFL per se in a patient with optic nerve atrophy treated with CSF diversion (shunt) surgery.
The ONHH was higher in the IIH than the REF group. Elevated ONHH has been shown to correlate with the optic nerve head volume and likely re ects the swelling and protrusion of the optic nerve head caused by the increased ICP. 14 The increased ONHH in the IIH group may re ect papilledema, though papilledema would expectedly be accompanied by increased thickness of the RNFL. 14,36 Thus, the elevated ONHH in the absence of signi cantly elevated RNFL thickness observed in the IIH group may also be due to chronic changes like overlying gliosis. 37 Previous studies of ONHH in relationship to ICP vary. Two studies 17,18 demonstrated a correlation between the height of the optic nerve head and invasively measured mean ICP. Similarly, a second study found a trend towards a decrease in ONHH after lowering the CSF opening pressure. 38 Other studies, however, found no 14 or even a negative 39 association between the ONHH and the lumbar CSF opening pressure.
Some limitations of this study should be noted. First, due to the lack of an automatic algorithm in the NIDEK software for pBA and ONHH determination, these may be examiner-dependent. To mitigate this potential bias, a consensus between two examiners was made. Second, for logistic reasons, we were not able to perform ICP and OCT measurements simultaneously. There was a time lag of weeks to months between the two examinations. However, we consider it less likely that the ICP or OCT scores would change signi cantly during this period. At least, the difference in time cannot explain the positive correlations between the independent variables. Third, the low number of study subjects is a limitation. Of note in this regard is the observed difference in the mean RNFL, which may be due to lack of statistical power.

Conclusions
The peripapillary Bruch's membrane angle and optic nerve head height correlated with the pulsatile ICP and differentiated between elevated and non-elevated ICP levels. Accordingly, our study provides proof of concept for a non-invasive method of detecting abnormal pulsatile ICP.

Methods
Approvals. The ICP scores are stored as part of a Neurovascular-Hydrocephalus quality registry approved by the Institutional Review Board of Oslo University Hospital (2011/6692). The OCT assessment was done as part of a study approved by the The Regional Committee for Medical and Health Research Ethics of Health Region South-East, Norway (2018/1074). The study was conducted according to the ethical standards of the Helsinki Declaration from 1975 and as revised in 1983. Informed consent was given by the participants prior to inclusion.
Design. A prospective observational study design was used. In order to examine how OCT parameters may be used to estimate pulsatile and static ICP, we 1) compared OCT and ICP scores between two patient cohorts, and 2) correlated the OCT and ICP scores of individual patients, and 3) determined the predictive ability of OCT regarding ICP.
Patients. The rst group included patients with a diagnosis of IIH, 26 who underwent ICP monitoring as work up for tentative shunt surgery due to failed conservative-medical treatment. The second group included an age-matched patient group that underwent ICP monitoring as pre-operative work-up for tentative CSF disturbance, but in whom no CSF disturbance was identi ed and no surgical intervention recommended. Both groups underwent overnight ICP monitoring at the Department of Neurosurgery, Oslo University Hospital -Rikshospitalet, Norway.
Overnight ICP monitoring. Monitoring of the ICP was performed overnight as part of routine clinical work up in order to diagnose and select patients for CSF diversion surgery. 28 Continuous ICP waveforms were recorded at 100-200 Hz using an intracranial pressure sensor (Codman ICP MicroSensor, Codman, Johnson & Johnson, Raynham, Massachusetts) placed 1-2 cm into the parenchyma through a burr hole in the frontal bone under local anesthetics. 24 For the MWA, single heart beat induced pressure waves are identi ed by an automatic algorithm, and the pulse amplitude is calculated as the pressure difference between the diastolic and systolic pressures over 6-second time windows. 24 The mean ICP is the sum of all pressure measurements during the 6-second time windows divided by the number of measurements.
With regard to our previously reported upper normal ICP scores, 27,28 abnormal pulsatile ICP was de ned by average overnight MWA scores ≥ 4 mmHg and/or ≥ 5 mmHg in ≥ 10 % of the recording time, and abnormal static ICP was de ned by overnight average mean ICP ≥ 15 mmHg and/or ≥ 10 % of measurements ≥ 15 mmHg.
Ophthalmological examination and OCT imaging protocol All ophthalmic examinations were performed at the Department of Ophthalmology, Oslo University Hospital -Ullevål, Norway by an experienced ophthalmic nurse and HHJ. To avoid dependent data, only the "worst eye" de ned by mean defect on threshold perimetry was included. Exclusion criteria were refraction error with spherical equivalent outside the range of -6.0 to + 3.0 diopters, concurrent eye disorders affecting the retina or optic nerve, and OCT scans with artifacts or poor quality (signal strength index < 7 of 10).
Standard white-on-white threshold perimetry of the central 30° was performed using an Octopus 900 perimeter (Haag-Streit AG, Koeniz, Switzerland) and the G-TOP program. The perimetry was performed by a trained ophthalmic nurse. Presence of fundoscopic papilledema was evaluated by HHJ and graded as present or not present.
Spectral-domain OCT scans were acquired in mydriasis using a RS-3000 OCT Advance (NIDEK Co. LTD, Singapore) and its Navis-EX software (NIDEK Co.).
The pBA (Fig. 4) and the maximum ONHH ( For the ONHH measurement we used the tools in the Navis-EX software. A reference line was drawn across the scleral opening, connecting the nasal and temporal ends of the retinal pigment epithelium/Bruch's membrane layers. The ONHH was measured by a straight line to the highest point of the optic nerve head (Fig. 2). The OCT parameters mean RNFL, the mean peripapillary total retinal thickness (TRT), and the macular ganglion cell complex (mGCC) were also measured and compared between the groups. The mean RNFL thickness and the mean TRT were automatically measured by the Navis-EX software using a ring scan centered at the optic disc with a diameter of 3.45 mm. The mean TRT was measured from the inner limiting membrane to the retinal pigment epithelium/ Bruch's membrane. The mean mGCC was measured automatically by the Navis-Ex software using the 9x9mm Macular Map program.

Statistical analyses
An independent t-test was used for the comparison of continuous data between the groups. Correlations between independent variables were determined by the Pearson correlation coe cient. Logistic regression models calculated the odds ratio for continuous exposure variables.

Data availability
The data by which this paper is based is available upon reasonable request. Figure 1 Association between the peripapillary Bruch's membrane angle (pBA) and pulsatile and static ICP scores.

Figures
There was a signi cant positive correlation between (a) the peripapillary Bruch's membrane angle (pBA) and the average overnight mean ICP wave amplitude (MWA) and (b) between the pBA and the overnight percentage of MWA ≥5 mmHg. No signi cant correlation was found between pBA and (c) average of overnight mean intracranial pressure (ICP) or (d) overnight percentage mean ICP ≥15 mmHg. Each plot shows the t line and Pearson correlation coe cient with signi cance levels (ns=non-signi cant).

Figure 2
Association between the optic nerve head height (ONHH) and pulsatile and static ICP scores. There was a signi cant positive correlation between (a) the optic nerve head height (ONHH) and the average overnight mean ICP wave amplitude (MWA), and (b) between ONHH and the overnight percentage of MWA≥5 mm.
There also was signi cant positive correlation between (c) ONHH and average of overnight mean intracranial pressure (ICP), but no correlation between (d) ONHH and overnight percentage mean ICP ≥15 mmHg. Each plot shows the t line and Pearson correlation coe cient with signi cance levels (ns=nonsigni cant).  The peripapillary Bruch's membrane angle (pBA). The Bruch's membrane/ retinal pigment epithelium is marked by a red line. The peripapillary Bruch's membrane angle (pBA, yellow) is the angle which is formed between the deviated peripapillary Bruch's membrane (red line) and the unaltered part (blue line).
A mean of nasal and temporal angles was calculated. The optic nerve head height (ONHH). A line connecting both sides (nasal-temporal) of the scleral canal opening (red line) was used as a reference line. The height was measured from the highest point of the optic nerve head to the reference line (yellow line).