Hydraulic Conductivity of Fixed Retinal Tissue After Sequential Excimer Laser Ablation: Barriers Limiting Fluid Distribution and Implications for Cystoid Macular Edema

Objectives  To measure the hydraulic conductivity (HC) of human retina and to determine the presence and location of high-resistance barriers to fluid movement through the retina.

Methods  Forty-one pairs of human eyes were investigated using an HC chamber. Once baseline HC had been determined, the effect of ablating through varying thickness of retina from the vitreous or photoreceptor surface using an excimer laser (193 nm) was investigated. Tissue samples were then processed for histological investigation.

Results  The HC of fixed intact human retina was 2.54 × 10⁻¹⁰ m/s per pascal at 539 Pa (range, 0.6 × 10⁻¹⁰to 3.3 × 10⁻¹⁰ m/s per pascal; SD, 0.6 × 10⁻¹⁰ m/s per pascal [1 mm Hg equals 133 Pa]). Ablation from either surface resulted in little change in HC until a critical depth was reached, at which point there was an order of magnitude increase. The critical depth was approximately 170 µm from the inner limiting membrane when ablating from the vitreous surface and 70 µm from the inner limiting membrane when ablating from the photoreceptor surface. Histological specimens showed that these barriers were the synaptic portion of the outer plexiform layer, and the inner plexiform layer, respectively.

Conclusions  The 2 high-resistance barriers to fluid flow through the retina are the synaptic portion of the outer plexiform layer, and the inner plexiform layer.

Clinical Relevance  These observations help to explain the distribution of cystoid macular edema seen in histological studies and with optical coherence tomography.

CYSTOID MACULAR edema (CME) is a common sequel to many ocular conditions, including traumatic, vascular, inherited, and inflammatory diseases of the eye.¹ Its fundoscopic appearance may vary, but typically it may be identified as round or ovoid cysts around the fovea. The cysts are characterized by an altered light reflex with a decreased central reflex and a thin, highly reflective edge.² The retina may or may not show elevation or increased thickness. Visualization of the geometry and distribution of cysts may be enhanced by fluorescein angiography, where the cysts become hyperfluorescent over varying times.²

The recent introduction of optical coherence tomography shows the cysts as areas of low or no signal, with occasional high-signal elements bridging the retinal layers.^3-5 Histological studies show the cysts to be areas of retina in which the cells have been displaced.^2,6-9 The cysts are presumed to result from the abnormal accumulation of fluid within the retina.¹⁰ For such an accumulation to occur, abnormalities must be present in one or both elements of the vascular supply of the retina. The location and distribution of such fluid will depend on the physical constraints imposed by the structure of the adjacent retina.

The retina has 2 sources of metabolic supply, the retinal capillary system and the choriocapillaris. Both systems demonstrate barrier functions that are referred to as the inner and outer blood-retinal barrier (BRB), respectively. The inner BRB is formed by the endothelial cells lining the retinal capillaries and junctional complexes between such cells. By contrast, the retinal pigment epithelium (RPE) and similar junctional complexes between adjacent cells form the outer BRB.

To maintain retinal metabolism, there must be movement of fluid and selected metabolites across these barriers, together with removal of catabolites. Fluid movement may arise as a result of abnormal barrier leakage and may be driven by a combination of active transport, diffusion, and osmosis or by a pressure head. These systems are in equilibrium, although there is a small net movement of fluid out of the neuroretina into the choroid.^11-14 This equilibrium may be disturbed by leakage across the inner or outer BRB. It could also be postulated that fluid could accumulate in the absence of a leak but in the presence of a pump failure, particularly in the RPE. Failure of the inner BRB leading to CME has been demonstrated by means of fluorescein angiography in aphakic CME¹⁵ and of the outer BRB in retinitis pigmentosa¹⁶; however, it is likely that in many clinical situations both barriers are affected.^16-18

Hydrodynamics ensure that fluid will accumulate at points of least resistance and initially adjacent to the source of leakage. The pooling of fluid also demands that the boundaries of least resistance limit the rate of fluid loss. These boundaries can be considered resistance barriers to outflow. In the outer retina, fluid may accumulate beneath the RPE or between the RPE and the neuroretina, resulting in pigment epithelial detachments or serous detachments of the retina, respectively.^19,20 The location of accumulated fluid and the resultant histological abnormality suggest a physically weak adhesion between these adjacent layers. By contrast in the neuroretina, there are no potential horizontal cleavage planes. In this tissue, fluid may only accumulate by displacing cells and cellular components.

A possible explanation for the distribution of fluid pooling within the neuroretina seen in histological studies is as follows. Fluid accumulation can occur only within the neuroretina in the presence of resistance barriers to outflow. The morphologic appearance will be determined by the location of the high-resistance barriers and by the physical constraints imposed by the organization of the retina. These physical constraints may be identified as positions where cells are joined by junctional complexes or where their processes invaginate the surface of others or exhibit tortuousness and entwine with processes of others. By contrast, where cell bodies or cell processes are highly ordered, displacement may occur. In the outer retina, the Müller fibers display junctional complexes between themselves and the photoreceptors that collectively form the outer limiting membrane (OLM); in the inner layers, they exhibit junctional complexes between themselves and the inner limiting membrane (ILM).²¹ Thus, these cells constrain x, y-plane displacement as a result of their junctional complexes and z-plane displacement as a result of their processes. In the outer plexiform layer(OPL), the invaginated synapses of the rods and cones would theoretically limit displacement, as would the tortuousness and intertwining of the dendritic processes in the inner plexiform layer (IPL).²¹ The OPL consists of 2 zones, a large outer zone containing the inner connecting fibers of the photoreceptor cells and an inner zone of a true plexiform appearance composed of the synaptic pedicles of these fibers and the synaptic invaginations into the pedicle surface of the cells in the inner nuclear layer.²²(p31) The outer zone is further expanded in the macula, giving rise to the Henle fiber layer. Cell movement would then occur in the inner and outer nuclear layers, in the Henle fiber layer, and in the nerve fiber layers. If this concept is correct, then in the early stages of capillary leakage, fluid would be constrained between the IPL and the synaptic portion of the OPL, ie, in the inner nuclear layer. Similarly, with an outer BRB leak, fluid would pool in the Henle fiber layer and may cause some displacement of the outer nuclear layer, and, as a result, would be seen between the synaptic portion of the OPL and the OLM.

To test this hypothesis for the presence of high-resistance barriers to fluid movement through the retina and to determine their location, we have undertaken a series of studies using an excimer laser to progressively ablate retinal layers while monitoring resultant changes in the movement of fluid through the system, using hydraulic conductivity (HC) as a means of measuring fluid movement.

Forty-one pairs of human donor eyes were obtained from the United Kingdom Transplant Support Service Eye Bank at Bristol, England. Postmortem time varied from 24 to 56 hours. The median age of the donors was 69 years (range, 18-91 years). A full-thickness circumferential incision was made 5 to 7 mm behind the limbus. The vitreous, lens, and anterior segment were discarded. A 2-mm trephine was used to isolate the retina from the optic disc, and the neuroretina was gently teased away from the underlying RPE before being transferred to phosphate-buffered saline solution containing the following antibiotic and antimycotic agents: penicillin (100 000 U/L), streptomycin (100 mg/L), and amphotericin B (250 µg/L) (Sigma-Aldrich Corp, Poole, England).

Pilot experiments were undertaken using 57 trephined specimens from 22 pairs of eyes to develop a reproducible system for the measurement of retinal HC. Initial studies involved using freshly trephined retinal specimens. However, the tissue proved too friable and broke up even at very low-pressure differentials. A subsequent series of pilot studies established that HC could be measured if the retina was supported on a defined artificial membrane; however, tissue breakup still occurred in an unacceptable number of specimens. A final study was therefore undertaken using fixed retinal tissue together with a defined support membrane to ensure reproducible measurements in large numbers of specimens. They also determined that fixation resulted in no significant change in HC.

In all subsequent experiments, the isolated retina was fixed by means of immersion in 2.5% glutaraldehyde buffered with 0.1-mol/L sodium cacodylate containing calcium chloride, 10 mg/mL (final pH, 7.4), for 1 hour. With the use of a dissecting microscope, four 8-mm trephined specimens of neuroretina were isolated from the macula extending to the midperiphery. These were placed on 150-µm-thick nitrocellulose filters with 8-µm pores (Millipore Corporation, Bedford, Mass). In individual experiments, the trephined specimens were placed with the vitreous or photoreceptor surface against the filter, causing exposure of the photoreceptor or vitreous surface, respectively.

The HC chamber used in the present series of experiments was identical to that previously described,²³ except that the tissue-mounting cassette was constructed in transparent plastic with a central lumen of 4 mm. The retina and nitrocellulose samples were mounted in the cassette as previously described,²³ and the cassette was inserted into the chamber such that the nitrocellulose surface was always distal to the direction of flow. Procedures for introducing fluid into the cassette were as previously described,²³ except that the phosphate-buffered saline solution was degassed with the use of a vacuum. The assembled chamber was maintained at 37°C. Tissue was exposed to pressures varying from 343 to 1715 Pa by altering the height of the fluid reservoir (1 mm Hg equals 133 Pa). For each pressure at which flow was measured, a 30-minute equilibration was first allowed, during which the reservoir height slowly changed. The fluid height was then returned to this pressure and the position of the meniscus was noted at 3-minute intervals for 20 minutes.

An excimer laser (Apex Plus; Summit Technology, Boston, Mass) was used in the phototherapeutic keratectomy mode with a radiant emission of 180 mJ/cm,² but with a beam diameter of 3 mm. From each pair of eyes, up to 8 trephined specimens were obtained and mounted in the transparent plastic cassettes. One sample was then mounted in the HC chamber with the vitreous surface exposed and the baseline flow determined. Measurements were then repeated on a second sample with the photoreceptor surface exposed. Subsequent samples were placed under the operating microscope of the excimer laser. The surface of the neuroretina was then carefully blotted using eye sponges (Visispear; Visitec, Sarasota, Fla). The phototherapeutic keratectomy software was programmed into the laser control system, the helium-neon aiming beams of the excimer system were focused on the surface of the retina, and the ablation sequence was initiated. Three specimens were ablated on the vitreous surface and 3 on the photoreceptor surface. In early experiments, specimens received 10, 20, or 30 pulses, whereas in the later experiments values were steadily increased up to 150, 200, and 250 pulses. After ablation, the cassettes were immersed in phosphate-buffered saline solution before being mounted in the chamber, and measurements of flow were undertaken. A total of 99 trephined specimens from 22 pairs of eyes were examined. Fifty-nine trephined specimens were mounted with the vitreous surface outward, of which 38 were ablated with excimer laser, and 40 trephined specimens were mounted with the photoreceptor surface outward, of which 32 were ablated with excimer laser. Finally, all specimens were removed from the cassette and processed for light microscopy (LM) and scanning electron microscopy (SEM). Thirteen additional samples were not included in the results because of holes detected during flow measurement or subsequent morphologic examination.

Samples were rinsed briefly in 0.1-mol/L sodium cacodylate containing 7.5% sucrose before they were fixed for 1 hour in 2% osmium tetraoxide in 0.2-mol/L sodium cacodylate. Samples were then hemisected. Half were processed for LM, and the other half for SEM. The LM samples were dehydrated in ethanol and embedded in epoxy resin. Semithin (1 µm) sections were cut on an ultramicrotome (Huxley; Leica, Milton Keynes, England) and stained with toluidine blue. The SEM sections were dehydrated in acetone, dried in a critical point drier, sputter coated with gold, and examined in an SEM (510S; Hitachi, Wokingham, England).

Flow, defined as the rate of volume change per unit of time per unit of surface area, was calculated from F = (x × C)/(t × A), where F indicates flow; x, distance moved by capillary column (in meters); C, manometer calibration constant; t, time (in seconds); and A, exposed membrane area (1.26 × 10⁻⁵ m² for the 4-mm-diameter tissue cassette).

The manometer calibration constant was determined by introducing a known weight of mercury into the capillary tube of the reservoir, lowering the tube into a horizontal position, and measuring the length of the mercury thread with the traveling microscope.²³ From the density of mercury (13 000 kg/m²), the constant was calculated to be 1.170 × 10⁻⁶ m³/m.

The HC was calculated from HC = F/P, where P indicates pressure (in pascals). The units of HC are cubic meters per second × square meters per second per pascal, which simplifies to meters per second per pascal. As no empirical studies of the pressure reduction across the retina have been published, it was assumed that pressure reduction would be similar to that across the Bruch membrane and choroid (estimated at 535 Pa); therefore, the results in the present report will be expressed at a similar pressure.²⁴

The HC of fixed intact human retina was measured to be 2.54 × 10⁻¹⁰ m/s per pascal (range, 0.6 × 10⁻¹⁰ to 3.3 × 10⁻¹⁰ m/s per pascal; SD, 0.6 × 10⁻¹⁰ m/s per pascal) at 539 Pa. The average values for the HC measured in 4 paired samples of fixed and unfixed retina were 2.75 × 10⁻¹⁰ m/s and 2.97 × 10⁻¹⁰ m/s per pascal, respectively (ranges, 2.47 × 10⁻¹⁰ to 3.1 × 10⁻¹⁰ and 2.85 × 10⁻¹⁰ to 3.1 × 10⁻¹⁰ m/s per pascal, respectively; SDs, 0.3 × 10⁻¹⁰ and 0.1 × 10⁻¹⁰, respectively; P = .30 by t test) at 539 Pa. For a given eye, there was no difference in HC measured when pressure was applied to the vitreous surface or the photoreceptor surface (15 eyes; P = .64 by t test). There was no correlation between HC and time from death to fixation or time from enucleation to fixation (R² = 0.013 [P = .66] and R² = 0.117 [P = .13], respectively). There was no correlation between HC and age (R² = 0.055 [P = .20]). There was a logarithmic reduction in HC with increased pressure(R² = 0.98 [P<.001]; Figure 1).

Irrespective of the surface of the retina receiving ablation pulses, there was a linear relationship between the depth of tissue removed and the number of excimer pulses applied (R² =0.91 [P<.001]). The average ablation depth per pulse was 0.5 µm.

The results of measuring HC after ablation are shown in Figure 2. Ablation from either surface resulted in little change in HC until a critical depth was reached, at which point there was an order-of-magnitude increase. The critical depth was approximately 170 µm from the ILM when ablating from the vitreous surface and 70 µm from the ILM when ablating from the photoreceptor surface.

Results of histological examination showed that these barriers were the OPL and the IPL, respectively (Figure 3). Sections showed that the synaptic portion of the OPL was always totally ablated such that the inner connecting fibers of the photoreceptor cells (Henle fiber layer) were exposed. Thus, the synaptic portion of the OPL presented the outer barrier.

The measurements of HC reported in this study support the hypothesis that the plexiform layers of the retina are regions of high resistance to the movement of fluid under pressure. These findings identify 2 of the 3 elements in the horizontal structural framework that would be required to produce the distribution of cysts seen in histological specimens.^2,6-9 The remaining element is at the location of the OLM.^25,26 It should be remembered that HC did not change significantly until considerable amounts of the IPL had been ablated. This observation probably reflects a lack of sensitivity of the measuring technique used, given an inability to detect change as this inner barrier was traversed. This concept is further supported by the histological findings that, in ablations from the vitreous surface, the spatially confined synaptic portion of the OPL alone was always completely ablated before barrier loss was detected. This layer is typically 10 µm thick. Given the 0.06-µm thickness of the OLM and the apparent insensitivity of the present method, it is perhaps not surprising that its demonstrable barrier properties were not elucidated by our method using a 0.5-µm ablation rate per pulse.^25,26

Under normal physiological conditions, fluid must move across the entire retina and through the high-resistance barriers; however, there is evidence that diffusion limits for metabolic supply in the system are approximately 150 µm. This would explain the thickness of the retina within the capillary-free zone where it rarely exceeds this value, and it would also explain cell loss in relation to vascular closure in either the retinal or choroidal supply. In large areas of capillary closure, all the layers internal to the OPL are lost and replaced by gliosis,²⁷ whereas choroidal infarction may lead to loss of RPE and photoreceptor cells but to preservation of all internal layers.^28-30 It could be argued, therefore, that fluid leaking from a given vascular system easily may diffuse up to 150 µm. If such fluid were leaking at an abnormal rate and pooling within this diffusion limit, it could rapidly change the microanatomical configuration of the tissue components and its boundaries. Increasing pressure within the pool would lead to increasing compactness in the surrounding tissues. Thus, in conditions leading to macular edema, fluid leaking from retinal capillaries would result in displacement of nuclei in the INL and, ultimately, in compression of fibers within both plexiform layers. This concept is supported by observations in the HC studies, where resistance increased significantly with pressure. If pressure-related barrier changes in the OPL were slower or of less magnitude than those in the IPL, then fluid could pass through this layer into the region of photoreceptor inner segments and nuclei before compressing junctions in the OLM and enhancing its barrier properties. In conditions where the outer BRB leaked, fluid would rapidly reach the photoreceptor inner segments and outer aspect of the OLM. It could pass through this barrier before becoming further impeded by the synaptic portion of the OPL. Again as fluid progressively accumulated, resistance to fluid movement through the OLM and the synaptic portion of the OPL would ultimately govern the size of any given cyst.

Examination of cyst size in uveitic patients using optical coherence tomography revealed a maximum z-plane dimension of approximately 460 µm(R.J.A., unpublished data, November 1999). Beyond this size, fluid spread laterally or accumulated in the subretinal compartment underneath the OLM. The concept of multiple discrete cysts has developed as a result of examining histological sections and, more recently, optical coherence tomography scans. Both modalities give an erroneous picture in that they sample a single confined plane through the retina. In 3-dimensional examination of cysts such as those derived from SEM, it is clear that the cysts are a single compartment spanned by the trunks of the Müller fibers.³¹ Because these trunks are some 10 to 30 µm in diameter, they are artificially identified as apparent compartmental barriers in sectioning techniques. The limitation in cyst size must mean that equilibrium exists between the pressures increasing the volume of any given cyst and the forces restraining the size of the cyst. The forces restraining the size of the cyst are the Müller fibers themselves and the compression of the high-resistance barriers as the cysts increase in size.

Tso⁶ showed in histological studies of CME that vascular diseases were more likely to lead to cystoid spaces accumulating in the inner retina, specifically the inner nuclear layer. By contrast, RPE disease led to cysts accumulating in the outer retina in the Henle fiber layer.⁶ Tso and Shih³² also showed with horseradish peroxidase that leakage through the RPE in monkeys was constrained by the OPL. These observations are consistent with the evidence of the plexiform layers being high-resistance barriers to flow, as presented herein.

Although HC increased as the retina was ablated, there was no incremental increase on passing through the first of the 2 plexiform layers. Ablating through significant thickness of the retina would lead to the pressure being dissipated across a thinner section of retina. This would be equivalent to increasing the pressure across intact retina, which causes tissue compression. This compression of the remaining plexiform layer would enhance its barrier properties and thus exaggerate the increase in flow on ablating through it. In addition, the current technique may not be sufficiently sensitive to detect a difference in the barrier properties between the 2 plexiform layers. Hence, if each layer confers a similar rate-limiting barrier to fluid movement, ablation of both layers would be required before an increase in flow is detected.

There was no significant difference in HC measured at various postmortem times up to 50 hours and no significant differences generated by time between enucleation and death. In addition, no measurable difference was apparent between fixed and unfixed specimens. Previous work has suggested that vacuolation and cytoplasmic swelling occur after 10 hours post mortem but that, at least up to 24 hours post mortem, these changes are reversible.³³ The effect at a cellular level of fixation in 2.5% glutaraldehyde with its attendant protein cross-linking is uncertain, but such fixation would arrest any further postmortem changes. Fisher³⁴ compared the HC of rat lens capsule before and after fixation with glutaraldehyde. He found the HC at low-pressure levels of fixed lens capsule to be about 0.6 times that of unfixed tissue, but he found that fixation stabilized the HC of the lens capsule, probably by making the capsule less deformable. Similar findings have been shown in kidney glomeruli.³⁵ Although caution must be expressed in applying the actual measurements presented in this article to the clinical situation, the concepts and locations of the barriers should hold.

To our knowledge, studies of the HC of human retina have not previously been published. Fatt and Shantinath³⁶ investigated rabbit retina, but it was not possible to transform their data to the units used in this study. Pederson³⁷ described unpublished data showing the results of a study using dog retina with an HC of 0.38 × 10⁻¹⁰ m/s per pascal (converted from 0.03 µL/min per millimeters of mercury per square centimeter), approximately 7 times less than the HC in this study, which would be within species and technique variation but might also reflect the freshness of their tissue. The HC of intact human retina showed no change with age. This compares with the exponential decrease of the HC with age in human Bruch membrane (Figure 4).^38,39 In the young, the HC of the Bruch membrane is much greater than that of retina, and as a consequence, this would promote retinal apposition in the presence of a flow of fluid from vitreous to choroid. Although the pressure difference, and thus the flow, across the retina from vitreous to choroid is likely to be small, Fatt and Shantinath³⁶ estimated that a pressure difference of as little as 0.52 × 10⁻³ mm Hg would be sufficient to keep the retina firmly attached.

In conclusion, we have demonstrated the HC of intact human retina and have shown that 2 of the major barriers to fluid flow through the retina are the IPL and the synaptic portion of the OPL. These observations help to provide an explanation for the distribution of CME seen in histological studies.

Accepted for publication Octover 18, 2000.

This study was supported by the Allerton Fund, London, England; the Iris Fund for Prevention of Blindness, London; and the TFC Frost Fund, London.

Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Fla, May 14, 1999.

We thank Ann Patmore and Anne Weston for their help throughout the paper and Timothy J. ffytche, FRCS, FRCOphth, for his advice.

Corresponding author and reprints: Richard J. Antcliff, FRCOphth, GKT Department of Ophthalmology, The Rayne Institute, Saint Thomas' Hospital, Lambeth Palace Road, London SE1 7EH, England.