Retinal Vein Pulsation Is in Phase with Intracranial Pressure and Not Intraocular Pressure

Abstract

Purpose.: As efforts to noninvasively measure intracranial pressure (ICP) increase, we thought it important to investigate the timing of retinal venous pulsation in relation to the intraocular and intracranial pressure pulses.

Methods.: Neurosurgical patients undergoing continuous direct ICP monitoring had video recordings of ICP, IOP, and retinal venous pulsation waveforms taken with constant timing relative to the cardiac cycle using pulse oximetry. Video frames of the maxima and minima of these parameters, including retinal vein diameter, were identified. The times from pulse oximetry “beep” to these parameters were measured and converted into a percentage of the respective cardiac cycle.

Results.: A total of 338 measurements from nine subjects with a mean age of 39 years, mean ICP of 4.4 mm Hg and IOP 15.1 mm Hg were taken. Vein diameter minima occurred an insignificant 0.6% of cardiac cycle before ICP minima (P = 0.6620) and a significant 3.2% after IOP minima (P = 0.0097) and significantly later than IOP (51%) and ICP maxima (74%, all P < 0.0001). Maximum vein diameter occurred an insignificant 2.0% before IOP maxima (P = 0.2267) and was significantly different from IOP and ICP minima (P < 0.00001). Mean venous pulsation pressure between the two eyes was significantly associated with ICP (r = 0.89, P = 0.0075).

Conclusions.: During pulsation, central retinal vein collapse occurs in time with IOP and ICP diastole. Venous collapse is not induced by intraocular systole. These results suggest that ICP pulse pressure dominates the timing of venous pulsation.

Introduction

Central retinal vein pulsation is a poorly understood phenomenon classically thought to be in phase with the cardiac cycle. ^1–4 This oscillation of the retinal venous wall generally occurs on the optic disc surface close to the venous exit through the lamina cribrosa. ⁵ Its presence and visibility is enhanced by low intracranial pressure (ICP) ⁶ and/or high IOP. ^7,8  

Spontaneous retinal vein pulsation is a reliable sign of normal or low ICP and cerebrospinal fluid (CSF) pressure. ⁶ Intracranial pressure is equivalent to intracranial CSF pressure and the latter appears to be equivalent to optic nerve subarachnoid space pressure when the pressure is greater than 0 mm Hg. ^9,10 There is a strong relationship between the magnitude of IOP required to induce venous pulsation (venous pulsation pressure) and the ICP. ^7,8,11 Venous pulsation is present in more than 90% of healthy subjects, but only 54% of glaucoma patients. ¹² Glaucoma subjects are thought to have increased hemi and central retinal venous resistance, which alters the distribution of the pressure gradient along the retinal vein and leads to an increased IOP required to induce venous pulsation. ¹³ The venous pulsation pressure (VPP) magnitude is closely associated with glaucoma severity and is strongly predictive of likely glaucoma progression. ^12,14 In addition to work using VPP to estimate raised ICP and in studies on glaucoma, recent work has explored its utility in estimating pressure in orbital tissue, ¹⁵ and retinal venous occlusion ^16,17 ; however, its genesis is poorly understood. Previous work by our group has demonstrated that the venous wall transmural pressure is close to 0 mm Hg at the surface of the optic disc, demonstrating that IOP is transmitted through to the intraluminal venous pressure. ¹⁸  

The classical theory of venous pulsation requires the presence of a gradient down the vein between the intraocular and retrolaminar optic nerve compartments so that the intraluminal venous pressure is equivalent to IOP at its exit point on the disc surface. IOP oscillations induced by the cardiac cycle leading to an IOP peak during systole may cause a compressive force to act on the venous walls at the exit and hence for intermittent collapse to occur in time with cardiac systole. ^1,3,4 The existence of a significant 7- to 10-mm Hg pressure difference between the intraocular and subarachnoid space compartments has been well documented for the genesis of venous pulsation in dogs ^11,19 and primates. ⁷ Alternatively, Levine ¹ proposes that the IOP pulse amplitude may be greater than the ICP pulse amplitude. So, during systole this may produce a greater pressure drop across the laminar vein driving blood out of the eye and leading to collapse of the vessel. The importance of pulse amplitude is supported by observations that IOP amplitude is positively associated with spontaneous venous pulsation. ²⁰ There are several problems with both of these theories. The first is the observation that venous collapse occurs during IOP diastole in healthy subjects and glaucoma patients. ²¹ This observation is not consistent with either theory. Additionally, the pulse amplitudes of IOP and ICP are both approximately equal at 2 mm Hg. ^22,23  

Because of the known importance of both ICP and IOP in venous pulsation generation, we wished to explore the phase relationships between these parameters and venous pulsation. We studied a number of subjects having continuous monitoring of intracerebral pressure, over varying pressures using the audible (beep) of a pulse oximeter as the cardiac cycle reference frame for all three parameters. 

Methods

The study followed the tenets of the Declaration of Helsinki; informed consent was obtained from the subjects after explanation of the nature and possible consequences of the study; and with approval from the Sir Charles Gairdner Hospital Group Human Research Ethics Committee. Patients undergoing ICP monitoring in the neurosurgery department high-dependency unit with either a Codman Bactiseal external ventricular drain (EVD; Johnson and Johnson, New Brunswick, NJ) or an intraparenchymal Codman Microsensor strain gauge intracranial pressure monitor (ICPM; Johnson & Johnson, New Brunswick, NJ) were examined as part of this study. EVDs were all located in the right lateral ventricle as confirmed on computed tomography (CT) and fluid pressures were zeroed to the external auditory meatus. Both ICPMs were located in the right frontal lobe parenchyma at least 1.5 cm from the cerebral cortical surface and were zeroed to 1 atmosphere pressure intraoperatively. Intraparenchymal and intraventricular pressures are equivalent in healthy animals, ¹⁰ and fluid-coupled EVD and ICPM measures have high correlation in head-injured subjects. ²⁴ All patients were capable of giving informed consent to the protocol and were capable of sitting at a slit lamp by the bedside for examination and had clear ocular media with no known ocular pathology. Each participant was approached for enrollment in the study after the EVD or ICPM was in place and its position confirmed on postoperative CT. The experimental set up was similar to that described previously. ²¹  

One neurosurgeon (C.L.) selected the patients consecutively over a 12-month period from the Sir Charles Gairdner Hospital neurosurgery high-dependency unit. Each patient was required to be fit and cooperative enough to sit by the bed for approximately 45 minutes in order for the measurements to be taken. Patients with orbital or ocular trauma to either eye were excluded by the ophthalmologist (W.M.), who performed all measurements including IOP and VPP using previously described techniques with known reproducibility. ²⁵ All patients sat at the slit lamp for these and the ICP measurements. 

Video recording was made using a sound mixer amplifying the signal from a microphone placed near the pulse oximeter (Nellcor, Boulder, CO) and combining that signal with the video recording from the slit lamp or the camera itself. The video recording of the intracranial pressure from the monitor (Datex-Ohmeda, Helsinki, Finland) was made initially while the subject was seated at the slit lamp (Fig. 1). At least 10 cardiac cycles were recorded. The monitor also displayed the mean and pulse ICP and blood pressures. Video resolution for this and the other sequences was 720 × 526 pixels at 25 frames per second. All timing of the signals used the frames and so a precision of 0.04 seconds was possible. The pulse oximeter was placed over the right index finger. 

Figure 1. 

 

Schematic diagram showing the experimental setup and how key parameters were recorded. “Lens” refers to the ophthalmodynamometer or tonometer lens.

All measurements were performed following pupil dilation using 1% tropicamide. At the slit lamp, the video recording (Leica, Wetzlar, Germany) was made of the IOP Goldman applanation tonometry mires and the central retinal vein pulsation. IOP was measured taking the midpoint of the mire oscillation as the mean IOP. The upper bound and the lower bound of the mire oscillation were observed to estimate the IOP amplitude also. 

Central retinal vein pulsation was examined initially using a 60-diopter noncontact lens to see whether pulsation was spontaneously present or not. An ophthalmodynamometer (Meditron, Völklingen, Germany) using a contact lens surrounded by a ring force transducer was applied to the eye using contact gel after local anesthetic instillation. Gentle graded force was applied until the onset of venous pulsation in either of the hemiveins or the central retinal vein was first visualized. The force measured from the ophthalmodynometer was then recorded. The IOP induced was calculated using our previously described formula; VPP = baseline IOP + 0.89 × ODF, where VPP is the calculated venous pulsation pressure and ODF is the force measured by the ophthalmodynometer. ²⁵ The venous pulsation was then recorded by video for at least 10 cardiac cycles. Where data sets were complete from each eye, data from both eyes were collected and analyzed. 

Each parameter was measured five times for each eye where possible. The pulse oximeter beep was used as the reference point for all measurements. The “beep” to IOP minima, ICP minima, and vein diameter minima (collapse) were calculated (T1, Fig. 2). The “beep” to IOP maxima, ICP maxima, and vein dilation were calculated (T2). The beep to next beep, maxima to maxima (ICP, IOP, and vein diameter), and, similarly, minima to minima were measured and calculated as T3. 

Figure 2. 

 

Schematic diagram illustrating the key measurements in relation to the parameter pulsation. “Beep” refers to the onset of the pulse oximeter beep. Parameters include IOP, CSF pressure, and retinal vein diameter.

We assumed the T3 values for beep and the maxima and minima of IOP, ICP, and vein pulsation should be equal and that tests comparing these parameters should produce an estimate of our parameter measurement precision. The values of T1 and T2 were converted into a percentage of the cardiac cycle by dividing them into the beep-to-beep T3 (and multiplying by 100). The T2 values tended to cluster around 0%, with some maxima values occurring just before or just after the beep. This meant that values just before the beep were close to 100% with values just after the beep close to 0%. To eliminate the creation of an artifactual bimodal distribution, if these values were greater than 50% then we subtracted 100% and hence produced a negative value for some T2 parameters. 

All data are expressed as mean with SD. A one-way ANOVA comparing the timing of the various parameters as a percentage of the cardiac cycle was performed using a nonlinear mixed model incorporating two random factors. The prime random factor was the variation of multiple recordings from each subject and the other was the variation between right and left eyes. The latter (right or left) was nested within the patient random factor using the nonlinear mixed effect model package within R. ^26,27 Each parameter was compared with five others, so a P value of less than 0.01 was considered significant and similarly 99% confidence intervals (CIs) were calculated where appropriate. 

Results

Nine patients were examined with a mean age of 39 years (range 21–58, Table 1) with measurements taken from 44 ICP recordings, 61 IOP recordings, and 64 vein pulsation recordings. Their mean cardiac cycle interval was 0.73 (± 0.11) second with a mean pulse rate of 81.8 per minute. Mean IOP was 15.1 ± 8.9 with a mean amplitude estimate of 1.1 mm Hg. Mean ICP was 4.4 ± 7.1 mm Hg with a mean amplitude of 3.7 ± 1.9 mm Hg. Figure 3 shows the shape of the ICP pulse curves from seven of the subjects. The ICP pulse curve shape from subjects H and I were similar to that of subject C. One subject (A) had an IOP of 38 mm Hg in both eyes and venous pulsation was spontaneous, so this value was not used in the regression analysis with ICP. Patient I had ICP −9 mm Hg, below the level where ICP variation alters vein pulsation, so this participant's data were also not included in the regression analysis. Patient A had no antecedent ocular history and was reviewed in the eye clinic 3 days later with a normalization of his IOP at that time and no pathology found. The mean of left and right venous pulsation pressure was calculated from each subject and analyzed in relation to ICP using linear regression analysis. A significant linear relationship was found (r = 0.89, slope = 1.16, P = 0.0075, y intercept = 13.2 mm Hg). 

Figure 3. 

 

The CSF pressure traces from seven of the nine subjects (A to G) are shown to illustrate the variation in pulse pressure waveform. They all had a rapid rise in pressure from minimum values. The inset contains schematic ICP pulse, arterial pressure pulse, and IOP pulse curves across one cardiac cycle superimposed, with amplitude expressed as percent to allow comparison.

Table 1. 

 

Patient Features and Mean Pressures (mm Hg)

Patient	Age (y)	Sex	Diagnosis	ICP Monitor Type	ICP mm Hg	ICP Amp mm Hg	IOP R mm Hg	IOP Amp R mm Hg	IOP L mm Hg	IOP Amp L mm Hg	VPP R mm Hg	VPP L mm Hg
A	21	M	Pineal tumor	EVD	2	1	38	1	38	1	Spont	Spont
B	34	M	Subarachnoid hemorrhage and hydrocephalus	EVD	14	2	14	2	14	1	28.2	30.5
C	21	F	Hydrocephalus	ICPM	8	5	11	1	12	1	22.6	26.2
D	20	M	Colloid cyst of third ventricle	EVD	0	3	10	1	10	1	10.0	11.8
E	57	M	Subarachnoid hemorrhage	EVD	0	7	12	1	10	1	13.9	16.2
F	46	M	Hippocampal tumor	EVD	4	2	10	1	10	1	16.8
G	53	F	Cerebellar tumor	EVD	11	5	14	1	14	1	18.5	22.6
H	53	F	Cerebellar tumor	EVD	10	5	17	2	18	2	40	18.7
I	44	F	Idiopathic intracranial hypertension with working shunt	ICPM	−9	3	10	0.5	11	0.5	19	18
Mean	39.1			4.4	4.4	3.7	15.1	1.2	15.2	1.1	21.1	20.6
SD	15.2			7.1	7.1	1.9	8.9	0.5	8.9	0.4	9.4	6.3

 ICP, IOP, and VPP recordings are in the sitting position at the slit lamp. Amp, pulse pressure amplitude, which is the difference between maximum and minimum; F, female; L, left eye; M, male; R, right eye; Spont, spontaneous venous pulsation.

Box plots of the various parameters timing in relation to the “beep” as percent of cardiac cycle are shown in Figure 4. The timing of the various parameters was compared using the nonlinear mixed model, finding that ICP maxima (−38% ± 15%), IOP maxima (−15% ± 14%) and vein diameter maxima (−17% ± 17%) were all significantly different from ICP minima (37% ± 11%), IOP minima (33% ± 12%), and vein diameter minima (36% ± 8%, all P < 0.0001). ICP maxima were found to be significantly different from all other parameters (P < 0.0001). Vein diameter minima timing was not significantly different from ICP minima (P = 0.6620), but significantly different from IOP minima (P = 0.0097). Vein diameter maxima were not significantly different from IOP maxima (P = 0.2267). ICP minima occurred an insignificant 3.5% (CI −0.6% to 7.7%) of cardiac cycle after IOP minima (P = 0.0263). Vein diameter minima occurred a mean 3.24% (CI 0.02% to 6.46%) of the cardiac cycle after IOP minima, and a mean 0.6% (CI −2.9% to 4.1%) of cardiac cycle before ICP minima. Vein diameter maxima occurred a mean 2.0% (CI −2.4% to 6.4%) of cardiac cycle before IOP maxima and a mean 20.8% (CI 15.8% to 25.9%) after ICP maxima. 

Figure 4. 

 

Box plots showing distributions of all maxima and minima from the parameters measured. The data are time as a percentage of cardiac cycle in relation to the pulse oximeter “beep.”

Thirty-one T3 measures were taken from the subjects for each parameter and were compared. The largest difference was between ICP minima T3 and beep T3 (0.036, CI −0.009 to 0.081 seconds). All other T3 parameters were less than 0.025 seconds and none were significantly different from 0 seconds. 

Discussion

There was a strong relationship between mean venous pulsation pressure and ICP (r = 0.89, P = 0.0075), with a slope close to 1 and the VPP tending to be 13 mm Hg above ICP. This is consistent with previous reports in humans, ⁸ and a little higher than the 8-mm Hg difference found in dogs ¹⁹ and monkeys. ⁷ Previously, we have found that a negative venous transmural pressure occurs close to the point of venous pulsation on the pig optic disc. ¹⁸ This suggests that the venous pressure gradient within the optic disc extends into the intraocular compartment and can be accentuated by raising IOP or lowering ICP to favor a negative transmural pressure within the eye leading to intermittent venous collapse. It should be noted that all ICP measurements were performed in the sitting posture with transducers zeroed at the external auditory meatus, which is in the same horizontal plane as the eye and occipital prominence. Magnaes ²⁸ has shown that in this posture and position, normal ICP ranges from −10 mm Hg (−14 cm H₂O) to 0 mm Hg and is approximately 13 mm Hg below that in the lateral decubitus position. 

The vein pulsation minima (collapse) occurs in time with ICP minima and close to IOP minima. Hence, venous collapse is occurring in time with diastole, not systole. This observation is consistent with our earlier timing experiments relating IOP with vein pulsation pulse only. ²¹ This occurrence in phase with ICP suggests enhanced outflow into, or suction by, the retro-laminar portion of the central retinal vein under the influence of ICP is tending to dominate the timing of the vein pulsation cycle. 

These results were derived from neurosurgical patients and so, many of their ICPs were somewhat higher than the normal range of 0 to 10 mm Hg when sitting. ²⁸ However, the IOP and venous pulsation timing results are identical to our previous work in healthy subjects and glaucoma patients ²¹ so these timing relationships are probably occurring in healthy and glaucoma patients also. Cerebrospinal fluid pressure maxima occurred significantly before vein diameter maxima (by 21% of cardiac cycle); however, IOP maxima occurred at the same time as vein diameter maxima. The typical ICP pulse, ²⁹ arterial pressure pulse, ³⁰ and representative IOP pulse curves ³¹ are superimposed in Figure 3 (inset). The shape of the curves leads to ICP increasing more rapidly than IOP in the phase segment, leading to systole with a likely tendency for intraocular blood accumulation, with dilation in systole, which is enhanced by compression of the central retinal vein within the optic nerve due to the rise in ICP tending to impede downstream flow. ³⁰ Furthermore, the tendency for ICP to drop before the fall in IOP in the phase segment leading to diastole may favor more rapid drainage of blood from the intraocular compartment toward the CSF space and hence retinal venous collapse in diastole. 

This difference in IOP and ICP curves will cause dynamic fluctuation in the translaminar pressure gradient with an expanded gradient (larger pressure drop) occurring in the phase leading to diastole with a more negative transmural pressure in the retinal veins. ¹⁸ When fluid drains from a collapsible vessel within a pressurized chamber to the exterior, this leads to oscillation of the collapsible vessel walls within the chamber near the exit region. ^2,3 In the body, the pressurized chamber is analogous to the eye and IOP, with the outside being analogous to CSF and ICP and the exit region analogous to lamina cribrosa. Modeling experiments have shown that oscillations in the pressurized chamber, ^2,32 or the downstream, outside region, ³² will force the collapsible vessel oscillations near the exit region to be in time. The collapse part of the cycle will occur in time with the most negative transmural pressure induced in the vessels near the exit region. ³³ The difference in IOP and ICP wave forms may be small, but the difference may be large enough to create these effects, which can be enhanced through a form of mode-locked resonance. ³⁴  

We propose that the same phenomenon is occurring in the eye and that the ICP waveform dominates the generation of pulsation in the central retinal vein. This may be due to differences in the pressure waveforms as described previously (Fig. 3 inset). Another explanation could be that the ICP pulse pressure amplitude is greater than IOP pulse amplitude and this would lead to an increased pressure drop between the two compartments during diastole with a possible tendency toward increased outflow of blood and intraocular venous collapse. However, we cannot find published evidence to support a clear increase in ICP pulse pressure above intraocular pulse pressure. Our intraocular pulse pressure measurements appear to be less than ICP pulse pressure (Table 1), but we are not convinced that our intraocular pulse pressure measurements are accurate and suspect that they are somewhat damped. Additionally, ICP and IOP pulse amplitudes both rise with mean ICP and IOP respectively. ^22,23 Our patients had somewhat higher than normal ICP, so their ICP pulse amplitudes could be expected to be greater than IOP pulse amplitude. This is unlikely to be the case in healthy subjects with the same IOP and VPP phase relationships reported previously. ²¹  

There are several other less likely explanations for this phenomenon. It has been postulated that pulsation of the central retinal artery in proximity to the central retinal vein may induce venous pulsation. ¹ However, central retinal vein pulsation is seen in animals without similar proximity, such as dogs, which have separate cilioretinal arteries and veins, yet still have prominent retinal venous pulsation that is similarly affected by ICP. ^11,19 Another possible influential factor is movement of the lamina cribrosa during the cardiac cycle. The IOP amplitude is, however, only 2 mm Hg and this lamina movement is determined not just by the IOP alone but rather by the difference between IOP and ICP, ³⁵ which probably remains approximately constant apart from slight variation due to the shape of the pulsation curves. There have been no reports of significant anterior optic disc or lamina movement with IOP changes less than 5 to 12 mm Hg, so it is unlikely that lamina cribrosa movement is a significant factor. ^35,36  

The ICP and IOP parameters were not measured concurrently but sequentially, with only several seconds between video sequences. All measurements were concurrent with a constant cardiac cycle marker and so the phase relationship found is likely to be robust. All measurement-to-measurement (T3) durations were insignificantly different from the pulse oximeter beep-to-beep time and the mean differences were all less than one video frame interval (0.04 seconds). So, the average measurement precision was within 0.04 seconds, equivalent to 5% of the cardiac cycle length. 

Our results may suggest that there is a low resistance connection between the intracranial CSF and optic nerve subarachnoid spaces; however, these results do not depend on such a fluidic connection occurring because just as there is no low resistance pathway between the vitreous cavity and anterior chamber of the eye, the pressures in those compartments are identical due to the easily deformable iris and zonular barriers. Ultrastructural studies demonstrate that optic nerve subarachnoid space septae exist and may form high resistance to CSF flow. ³⁷ Our results do suggest pressure transference from the intracranial to optic nerve subarachnoid space in our patients; however, such pressure transference may not occur in all individuals. 

There was a strong relationship between mean venous pulsation pressure and ICP (r = 0.89, P = 0.0075), consistent with previous reports. ^7,8 So, not only is venous pulsation pressure dependent on ICP, but its phase timing is also dependent on ICP. The central retinal vein lies within the retrolaminar optic nerve for some 10 mm before exiting into the orbit. Hence, the influence of ICP is likely to be modified by retrolaminar tissue pressure characteristics. Further clarification of these properties may allow noninvasive estimation of retrolaminar and ICPs in human subjects. 

Footnotes

 Disclosure: W.H. Morgan, None; C.R.P. Lind, None; S. Kain, None; N. Fatehee, None; A. Bala, None; D.-Y. Yu, None