Effect of Acute Intraocular Pressure Challenge on Rat Retinal and Cortical Function

Abstract

Purpose.: The global or gross response index of visual performance measured from the eye does not necessarily translate to global responses measured from the brain. A better understanding of this relationship would facilitate the monitoring of disease models that affect the visual pathway. We consider whether rod- and cone-retino-cortical-pathways are equally affected by acute IOP elevation.

Methods.: Acute, stepwise IOP elevation (10, 30, 40, 50, 60, 70 mm Hg) was induced in anesthetized dark- (N = 8) and light-adapted pigmented rats (N = 6). Electroretinogram (ERG) and visual evoked potentials (VEP) were simultaneously measured after 10 minutes at each step. Relative amplitudes (treated/baseline, %) as a function of IOP level were described with a cumulative normal function.

Results.: Our results showed decline in scotopic and photopic ERGs with IOP elevation. Photopic ERG responses were less sensitive to IOP challenge than scotopic ERG responses. Despite significant reductions of ganglion cell–mediated waveforms at 70 mm Hg, the VEP showed only subtle decreases in amplitude. Intraocular pressure elevation produced similar effects on rod- and cone-mediated VEP waveforms.

Conclusions.: We show that cone signals are less sensitive than rod ERGs to acute IOP challenge. Also, retinal signals are more sensitive than are cortical signals to IOP stress, suggesting that cortical processing may act to salvage reductions expected from attenuated retinal output.

Introduction

Intraocular pressure (IOP) elevation is the most studied physiological stress on retinal neurons. As one of the major risk factors in the development of glaucoma ¹ it is important to understand how visual function is compromised during episodes of IOP elevation. Although glaucoma is characterized by the gradual, selective dysfunction, and degeneration of retinal ganglion cell (RGC), ² recent studies indicate the presence of anatomical ^3–5 and functional alterations ^6–9 to higher visual structures. Similarly, local retinal ^10–14 and distal lateral geniculate and cortical changes ^15–22 have been reported in animal models of chronically-elevated IOP. 

Whether such central dysfunctions coincide or are secondary to ocular lesion is unclear. While studies involving acutely-raised IOP may provide insights into such mechanisms, few investigators have measured both retinal and cortical function concurrently while raising IOP. This approach provides a snapshot of sequentially activated retinal and cortical pathways. 

Several studies suggest that visual evoked potential (VEP) is more resistant to IOP elevation than the electroretinogram (ERG). ^23–25 However, these studies employ the b-wave index of retinal activity, rather than a measure indicative of ganglion cell output. To overcome this limitation, we employ, in addition to the b-wave, measurements of the scotopic threshold response (STR) ^26–30 and the photopic negative response (PhNR), ^29–33 to provide a more direct comparison of retinal output with VEP responses across a range of IOP levels. We also make recordings under dark- and light-adapted conditions to allow a more complete assessment of rod- and cone-mediated responses. ³⁴  

Methods

Animals

All experimental procedures obtained ethics clearance from our departmental Animal Ethics Committee (approval number 0911322.1) and conformed to the guidelines of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Adult Long-Evans rats (N = 14, 9–12 weeks, 250–310 g, purchased at 7–8 weeks of age; Monash University Animal Service, Clayton, VIC, Australia) were housed in a well-ventilated, temperature and humidity-controlled room (21°C, 40%–60% humidity). Light in the housing was regulated to cycles of 12 hour light/12 hour dark (on at 8 AM, maximum 50 lux), with ad libitum access to rat chow and water. 

Acute IOP Challenge

Acute IOP elevation was induced as previously described ^35,36 in two groups of rats (scotopic conditions, N = 8, photopic conditions N = 6). In brief, animals were dark-adapted overnight prior to recordings. For scotopic recording, animal preparation was conducted in a dark room with the aid of dim red light-emitting diode (LED; λ _max = 600 nm) to sustain rod sensitivity. Photopic procedures were performed after adaptation for 10 minutes to a 10 cd.m⁻² background. Rats were anesthetized by intramuscular injection of ketamine/xylazine (60:5 mg/kg; Troy Laboratories Pty. Ltd., Smithfield, NSW, Australia). A further dose of anesthesia was re-administered after 50 minutes to maintain sedation. Only a single reinjection was used per rat to minimize anesthesia-induced changes to the VEPs. ^37,38 Corneal anesthesia was achieved with a drop of proxymetacaine (0.5%; Alcain; Alcon Laboratories, Frenchs Forest, NSW, Australia), and mydriasis with a drop of tropicamide (0.5%; Mydriacyl; Alcon Laboratories) and phenylephrine (2.5% Minims phenylephrine hydrochloride; Bausch & Lomb, Rochester, NY). Body temperature was maintained with use of a circulating water heat pad (37°C). 

Unilateral IOP elevation was induced in a stepwise manner in a randomly chosen eye of each animal (10, 30, 40, 50, 60, 70 mm Hg, 12 minutes per IOP step) via a 30-G cannula placed in the vitreal chamber. The needle was connected via polyethylene tubing (inner diameter, 0.4 mm; Microtube Extrusions, North Rocks, NSW, Australia) to a pressure transducer (Transpac; Abbott Critical Care Systems, Sligo, Ireland), and a height-adjustable reservoir containing Hanks' balanced salt solution (JRH Biosciences, Lenexa, KS). Intraocular pressure was monitored continuously throughout experimentation via Chart software (Powerlab 8SP amplifier, v5.3; ADInstruments, Castle Hill, NSW, Australia). Immediately after cannulation, IOP was set to 10 mm Hg, consistent with control IOP levels reported in anesthetized rats. ³⁹ The maximal IOP of 70 mm Hg was chosen to avoid severe ischemic insult. ⁴⁰ Supplementary Figure S1 shows confirmation that the current IOP protocol allows complete recovery of waveforms by 1 hour post IOP restoration to baseline (10 mm Hg). The shorter recovery time seen with our protocol compared with a lower IOP of 50 mm Hg for half the duration (30 minutes ⁴¹ ; dysfunction evident at 1 week) further supports this notion. Given the potential influence of blood pressure on the severity of injury caused by IOP, ^42,43 systolic blood pressure was assayed at regular intervals (i.e., before, during, and after each IOP level) using tail-cuff sphygmomanometer (ML125; ADInstruments). Ten minutes of re-adaptation was allowed to ensure stable responses before IOP challenge. In both scotopic and photopic groups, the uncannulated contralateral eye served as control. 

Electrophysiological Recording

Electroretinogram and VEP (ERG/VEP) measurements were recorded simultaneously from both eyes and visual hemispheres in separate groups of rats under scotopic or photopic conditions. 

VEP Measurement.

Active electrodes were stainless steel screws (0.7-mm shaft diameter, 3 mm in length; MicroFasteners Pty. Ltd., VIC, Australia) implanted epidurally over V1 of each hemisphere (3-mm laterally to midline, 7-mm posterior to bregma ⁴⁴ ), consistent with previous studies. ^45–49 Implantation was performed under isoflurane inhalation (3%–3.5% inspired concentration for induction, 1.5%–5% once sedated: 96.5% pure O₂ gas with isoflurane, Attane, 99.9%; Bomac Pty. Ltd., Hornsby, NSW, Australia) 1 week prior to testing. The animals' head was stabilized by ear bars for surgery (Model 900; Kopf, Tujunga, CA) to allow holes to be drilled (Dremel 300 series; Bosch, VIC, Australia) for screw implantation. Holes (0.8 mm) in the skull through which the active electrodes were inserted (∼1-mm deep), were made using a dental burr (∼0.7-mm diameter, Storz Instruments, #E0824A; Bausch & Lomb, Inc., Feldkirchen, Germany). Dental amalgam (Dentsply Rapid Repair; DeguDent GmbH, Hanau, Germany) was spread over the wound to encase the electrode assembly, leaving approximately 1.5 mm of the screw exposed for recording. Intraperitoneal Carprofen (0.2 mL, Rimadyl 0.5%, 5 mg/mL; Pfizer Animal Health Group, West Ryde, NSW, Australia) and saline (1.5 mL, 0.9% sodium chloride; Baxter Healthcare, Toongabbie, NSW, Australia) were provided for analgesia and fluid replacement, respectively. Prophylactic antibiotic ointment Tobrex (0.3% Tobramycin; Alcon Laboratories) was also applied to the wound. 

On the day of recording, an alligator clip (stainless steel, 26 mm, HM3022, generic brand) soldered to platinum leads (F-E2-30; Grass Telefactor, West Warwick, RI) connected each active electrode for signal collection. The VEP reference was a silver wire hooked around lower front incisors, as in previous studies. ^34,48,50,51  

It is important to note that our binocular stimulation approach can contain contribution from the ipsilateral control eye and potential callosal routes. This former contribution was established by unilateral pharmacological inhibition of postreceptoral responses in our pilot studies (Supplementary Fig. S2, Tables S1, S2). In brief, a combination of metabotropic receptor (2-amino-4-phosphonobutyric acid; APB), ionotropic receptor (cis-2,3-piperidinedicarboxylic acid; PDA) and/or glial cell inhibition (barium chloride; BaCl₂) were used to pharmacologically transect the optic nerve with minimal trauma in one or both eyes, to determine monocular ipsilateral, contralateral, and nonvisual contributions to the rat cortical flash-VEP. A similar magnitude of interhemispheric asymmetry of VEP responses have been reported by other studies involving more (ie, enucleation) and less invasive strategies (unilateral occlusion). ^52,53 As for interemispheric transfer, enhanced contralateral bias has been noted when callosal inputs are disturbed. ⁵⁴ However, to our knowledge, the exact proportion of transcallosal contributions to the adult rat VEP has yet to be quantified. 

While monocular stimulation provides a better isolation of VEPs in the pigmented rat, it does not prevent the possibility for interhemispheric interactions, nor does the monocular approach contain less nonvisual background contributions than binocular recordings. It suffers similar limitations in that some nonvisual contribution to the VEP must be acknowledged (Supplementary Figs. S2A, S2E–F, Table S1). Although post hoc removal of the ipsilateral contribution is possible, this was not undertaken on the current binocular data, as it is unlikely that a canonical ipsilateral waveform can be used for each individual animal. Subtracting an average ipsilateral waveform (established in a separate group of rats) from each animal carries the risk of distorting the VEP, if waveforms are out of phase. Concurrent bihemispheric VEP recording avoids the need for an additional control group. An internal control afforded by binocular stimulation and bihemispheric recording reduces intersubject variability associated with flash VEPs, and allows improved assessment of treatment effects. Lastly, since the major monocular VEP signal is comparable with its putative binocular form (i.e., monocular ipsilateral + contralateral VEPs in a separate group, Supplementary Figs. S2C, S2E), any underestimations of IOP-induced dysfunction from V1 (i.e., due to the ipsilateral contribution from the fellow control eye), will likely also be insignificant. This possibility is nevertheless taken in to account in the Discussion. 

Table

 

Sensitivity of ERG Parameters to IOP Elevation

	Inverse Cumulative Normal Function–Functional IOP50
	Scotopic		Photopic
	Best-Fit IOP50, mm Hg	95% CL	Best-Fit IOP50, mm Hg	95% CL
PII	60	58–64	69	65–73
pSTR	60	59–61	-	-
nSTR	53	48–58	-	-
PhNR	-	-	46	40 – 60
PhNR*	-	-	75	67 – 111

 Modeling with a cumulative normal function returns the IOP at which 50% (IOP₅₀) of ERG amplitude was lost. The 95% CL (upper–lower limits, 2.5/97.5%) for each parameter was returned from a nonparametric bootstrap.

*  PhNR identifies the conventional method for analysis that we feel is a weaker indicator of the photopic inner retinal response for the rat (see text for details).

ERG Measurement.

As detailed previously, ⁵⁵ corneal (active) and scleral (ring-shaped reference) electrodes employed for electroretinography (ERG) were customized out of chlorided silver wire braided to platinum leads (F-E2-30). While the pattern ERG is a better indicator of ganglion cell function, a full-field stimulus modality was chosen to allow simultaneous measurements of outer- and inner- retinal function as well as the VEP. Both VEP and ERG electrodes were grounded to a stainless steel needle inserted subcutaneously into the tail. After electrode positioning, a drop of carboxymethylcellulose sodium (1.0%, Celluvisc; Allergan, Gordon, NSW, Australia) was applied to maintain cornea hydration and improve signal quality. 

Data Collection.

Stimuli were white (Luxeon LEDs; Philips Lighting Co., San Jose, CA), brief (1–2 ms) flashes delivered via an Ganzfeld integrating sphere (36 cm in diameter, aperture size of 13 cm; Photometric Solutions International, Huntingdale, VIC, Australia). Luminous energy was calibrated as previously described, ⁵⁶ and controlled using a Powerlab data acquisition system (Scope Software v3.7.6; ADInstruments). From the four channels (two retinal ERG, two cortical VEP channels), ERG signals from both eyes were amplified 1000 times, recorded with band-pass filter of 0.3 to 1000 Hz (−3 dB), and digitized at 4 kHz over a 650-ms epoch. Concurrently, VEP responses from both hemispheres were amplified 10,000 times, recorded with band-pass settings of 0.1 to 100 Hz (−3 dB), with a 10 kHz acquisition rate over 250 ms. ⁵⁷ A 10-ms prestimulus baseline was collected with every signal. 

Following cannulation and electrode placement, ERG intereye variability (≤±10% between amplitudes) and VEP noise were optimized using a dim test-flash (−0.52 log cd.s.m⁻²). Eyes were then allowed to re-adapt (10 minutes) before IOP challenge. The scotopic protocol consisted of two suites of rod-dominated signals, a dim STR (average of 20 signals, interstimulus interval [ISI] 2 seconds, −4.99 log cd.s.m⁻²), and a moderate energy ERG (b-wave)/VEP sweep (20 signals, ISI 5 seconds, −0.52 log cd.s.m⁻²). This series takes two minutes to collect, with the first signal returning the scotopic ERG, whereas the average of the sweeps gave the scotopic VEP. In another group of rats, the photopic protocol consisted of 20 signals (ISI 2 seconds, 1.52 log cd.s.m⁻²), again the first signal gave the photopic ERG waveform (encompassing the PhNR) and the average of the 20 gave the photopic VEP. 

Pilot studies showed that 10 minutes was sufficient for complete STR recovery after the 20 consecutive, moderate flashes (−0.5 log cd.s.m⁻²), whereas 4-minute recovery was found to be adequate to elicit unadapted scotopic ERGs between the series of bright flash sweeps from the same eye. This protocol represents a balance between our desire to obtain saturated STR, b-wave, and VEP waveforms, and to minimize time under anesthesia. Protocol duration was similar in both scotopic and photopic groups so that every rat had the same total duration of IOP elevation. That is, responses were measured 10 minutes after an IOP increment, allowing 2 minutes for data acquisition. Each IOP step lasted 12 minutes, and the total IOP stepwise elevation protocol per animal lasted just over 1 hour (excluding the 10 mm Hg baseline measurement). 

Electroretinogram Analysis

Signals were exported to Excel (Microsoft Office 2003, Redmond, WA) spreadsheets in digital voltage-time format for post hoc analysis. 

ERG Responses.

The photoreceptoral a-wave is small and thus not analyzed. We chose to focus on post receptoral components, which are known to be more susceptibility to IOP elevation. ^13,35,58,59 Oscillatory potentials were removed from the b-wave leading edge by band-pass filtering (Bessel filter: −3 dB at 70 and 200 Hz for scotopic, or 30 and 170 Hz for photopic ^60,61 ). The b-wave is taken as a trough-to-peak amplitude to return an accurate measure of its size when the ERG becomes electronegative at moderate IOPs. 

The STR and PhNR have been shown to reflect RGC activity in rats. ³⁰ It is conventional for STR amplitudes to be taken at fixed times of 110 (A₁₁₀) and 220 (A₂₂₀) ms for pSTR and nSTR values, respectively, ^29,30,36,62 after signal-to-noise characteristics was improved by low pass filtering (−3 dB at 100 Hz). This method was further adopted to allow comparisons with other studies. Similarly, fixed time analysis was used to obtain the PhNR amplitude, rather than taking the trough amplitude (i.e., deepest negative point post b-wave, up to 200 ms). ^31,33,63 Studies in our laboratory revealed that following inhibition of voltage-gated sodium channel using 6 μM tetrodotoxin (Tsai TI, Vinrgys AJ, Bui BV, unpublished data, 2013; and IOP elevation ³⁵ ) a larger change in PhNR amplitude was found at a fixed time of 130 ms compared with the trough minima (−66% at A₁₃₀ vs. −10% with trough amplitude). A₁₃₀ was thus adopted for PhNR quantification for a more sensitive indicator of inner retinal changes. 

VEP Responses.

The VEP is composed of several components that together reflect the integrity of the visual pathway. As per convention ^52,64,65 the visually-driven early components P1, N1, and P2 of the flash VEP were analyzed as trough-to-peak amplitudes from their preceding peak or trough, and have been expressed as P1N1 and N1P2. This scheme has been found to be less variable than are baseline-to-peak approaches for VEP signals. ^66,67  

Electroretinogram and VEP amplitudes during IOP elevation were expressed relative to their own baseline (%), measured at 10 mm Hg. Normalized amplitudes are then plotted as a function of IOP to construct an IOP-response profile. 

Statistical Analysis

All experimental data exhibited Gaussian distributions (Komogorov-Smirnov normality test, P > 0.1, Prism v5.00; GraphPad Software, Inc., San Diego, CA) and equal variance (Bartlett's statistics, P > 0.05, Prism v5.00). As such, parametric methods (t-tests and ANOVA) were employed. An α-value of 0.05 was adopted, except when adjusting for multiple comparisons (α < 0.01). 

The level of IOP-induced dysfunction (i.e., relative amplitude; y as a function of IOP; x) was modeled with a cumulative normal function, ⁶⁸ as previously described by Bui et al. ³⁵ and He et al. ⁵⁸ From this, the parameter IOP₅₀ indicates the IOP at which a 50% dysfunction is found (a measure of susceptibility, mm Hg). Sigma gives the spread at IOP₅₀ (mm Hg) or sensitivity to IOP challenge. Confidence limits (CL) of 95% for these parameters were derived from a bootstrap ^69–71 and were used to compare IOP₅₀ between the measured variables. Where the cumulative function returned a poor fit, two-way repeated measures analysis of variance was applied to compare data. When a significant interaction arose between treatment and IOP elevation, Bonferroni post hoc tests were adopted (eye × IOP level). Group averages are shown as mean (±SEM). 

Results

It was important to ensure that the effect of IOP on scotopic and photopic parameters in our rats was not influenced by differences in blood pressure (BP). ^14,43,72 Systolic BP measurements sampled throughout our protocol showed that BP was not significantly altered from beginning (IOP 10 mm Hg baseline, scotopic: 106 ± 10 mm Hg; N = 7, photopic: 102 ± 10 mm Hg; N = 8) to the end of the IOP challenge (70 mm Hg, scotopic: 93 ± 5 mm Hg, photopic: 95 ± 8 mm Hg) in either cohort (scotopic: P = 0.24, photopic: P = 0.59). Moreover, respective average BP indices with IOP stress were similar between scotopic and photopic groups (P > 0.05, two-way ANOVA). 

Effect of Acute IOP Manipulation on Scotopic Responses

Figure 1A shows average rod-dominant b-waves (−0.52 log cd.s.m⁻²), STRs (Fig. 1B; −4.99 log cd.s.m⁻²), and VEPs (Fig. 1C; −0.52 log cd.s.m⁻²) recorded simultaneously during IOP elevation. Characteristically, the moderate flash elicits an ERG with a large b-wave that peaks between 50 to 100 ms after stimulus onset. The small, negative a-wave preceding this peak is negligible at this stimulus energy. The dim-flash STR showed typical positive (pSTR) and negative deflections (nSTR) at approximately 110 and 220 ms, respectively. Finally, the accompanying VEP measured showed a large negative trough (P1N1; 20–50 ms window) followed by a larger, wider positive deflection (N1P2; 30–100 ms). To aid comparison to baseline, waveforms sampled at 10 mm Hg (thin traces) have been replotted at each IOP level (thick traces). 

Figure 1

 

Effect of IOP elevation on the scotopic ERG and VEP. Averaged (N = 8) scotopic ERG and VEP waveforms at each IOP elevation. Stimulus energies used to elicit these signals were (A) −0.52 for rod depolarizing bipolar cell b-wave, (B) −4.99 for inner retinal p- and nSTRs, and (C) −0.52 log cd.s.m⁻² for rod-mediated cortical responses. Numbers to the left (of column A) indicate the IOP (mm Hg). Intraocular pressure–treated waveforms (thick traces) are overlaid with the average baseline (10 mm Hg, thin traces) and selected (dotted) zero references lines are also shown.

Figure 1 shows that there is progressively greater loss of bright and dim waveform amplitudes as IOP increases, whereas relative changes to VEP waveforms did not show the same trend. In particular, when IOP reached 60 mm Hg, both b-wave and STR amplitudes were markedly reduced, whereas VEP components appeared largely unaffected. Even by 70 mm Hg when the b-wave and STR were abolished, only the P2 of the scotopic VEP appeared attenuated. Implicit times of all components did not vary greatly. 

In control eyes, scotopic amplitudes and implicit times did not change significantly as a function of IOP (Fig. 2, unfilled symbols, all P > 0.01). In IOP-treated eyes (filled symbols), b-wave (Fig. 2A), pSTR (Fig. 2B), and nSTR (Fig. 2C) amplitudes were reduced with IOP elevation by −88 ± 3%, −100 ± 16%, and −94 ± 3%, respectively at 70 mm Hg (all P < 0.001). Cumulative normal functions (curves in Figs. 2A–C) showed that the nSTR was more sensitive to IOP elevation (IOP₅₀, 53 mm Hg, Table) than the PII and pSTR (both 60 mm Hg, Table, P < 0.05). However, b-wave peak time was significantly slower at 70 mm Hg (P < 0.001, Fig. 2D), whereas STR peak and trough times were not significantly affected by IOP elevation (−4.99 log cd.s.m⁻² implicit time data not shown, P > 0.05). 

Figure 2

 

Scotopic ERG and VEP parameters during acute IOP challenge (mean ± SEM, N = 8) normalized to their own baseline. Left panels show (A) rod-dominant ERG b-wave, (B) pSTR, (C) nSTR amplitudes, (D) peak time, and (E) trough time. Right panels show VEP (F) P1N1, (G) N1P2 amplitudes, (H) P1, (I) N1, and (J) P2 times. Intraocular pressure–treated (solid symbols) and control (open symbols) amplitudes and implicit times are plotted as a function of IOP (mm Hg). Both data from control and treatment eyes are normalized to each group's own 10-mm Hg baseline. Dotted lines represent the 0% change from their baseline. Amplitude changes are modeled using a cumulative normal function, which provides IOP₅₀; a measure of sensitivity (arrow) to IOP. A nonparametric bootstrap estimated 95% CL (Table).

Figures 2F and 2G show that despite variations in the IOP-treated VEP waveforms over the course of the IOP protocol seen in Figure 1C, their corresponding group averaged parameters were not significantly removed from baseline. It was not until 70 mm Hg that the N1P2 amplitude became significantly smaller than baseline (−35 ± 13%, P < 0.01, Fig. 2G). In contrast, acute IOP challenge had little effect on the scotopic VEP P1N1 amplitude (70 mm Hg only −7 ± 14%, P = 0.21). As for implicit times of VEP features, the P1 was faster at 70 mm Hg (by 8 ± 2 ms, P < 0.01, Fig. 2H), whereas N1 was slower (by 9 ± 3 ms, P < 0.001, Fig. 2I) and P2 remained similar (Fig. 2J, P = 0.27). 

Effect of Acute IOP Manipulation on Photopic Responses

Figure 3 compares photopic ERG (Fig. 3A) and VEP waveforms (Fig. 3B). Similar to the scotopic signals, attenuation and delays of both ERG and VEPs were noted at 70 mm Hg, however these appear less severe in comparison to Figure 1. 

Figure 3

 

Effect of IOP elevation on the photopic ERG and VEP. Averaged (N = 6) cone-mediated (A) ERG and (B) VEP waveforms at each IOP elevation. These were elicited by a 1.52 log cd.s.m⁻² flash on a light-adapting background (10 cd.s.m⁻²) for cone-mediated responses. Numbers to the left (of A) indicate IOP (mm Hg). Intraocular pressure–treated waveforms (thick traces) were overlaid on the same average 10-mm Hg baseline for IOP-treated eyes (thin traces). Also shown are average waveforms from fellow control eyes (dashed gray traces) at 70 mm Hg (∼60 minutes into protocol), and selected reference (dotted) lines.

Both the cone b-wave (Fig. 4A) and PhNR (Fig. 4B) were significantly attenuated at 60 mm Hg. They were reduced by −52 ± 6% and −100 ± 44% of their baseline amplitudes, respectively, at 70 mm Hg IOP (both P < 0.001). However, the b-wave showed a lower sensitivity to IOP in comparison to the PhNR (IOP₅₀ of 69 vs. 46 mm Hg, respectively, Table, P < 0.05), as the inner retinal response was affected to a greater extent. Maximal reduction of the photopic b-wave was also up to 40% less severe than the scotopic equivalent, while the PhNR and STR responses were similarly affected. It should be noted that if however, the PhNR amplitude was assessed at its minima (Fig. 4B, gray symbols) as is usually the case (see Methods), it would be seen to be less sensitive to IOP stress (−36 ± 11% at 70 mm Hg, with IOP₅₀ of 75; Table, PhNR*). Associated photopic peak/trough times were delayed at 70 mm Hg (by 6 ± 2 and 9 ± 8 ms, Figs. 4C, 4D, P < 0.01). 

Figure 4

 

Photopic ERG and VEP during acute IOP challenge (mean ± SEM, N = 6) normalized to their own baseline. Left panels: (A) cone-dominant ERG b-wave, (B) PhNR amplitudes, (C) peak time, and (D) trough time. Right panels: VEP (E) P1N1, (F) N1P2 amplitudes, (G) P1, (H) N1, and (I) P2 time. Intraocular pressure–treated (solid symbols) and control (open symbols) amplitudes and implicit times are plotted as a function of IOP (mm Hg). Both control and treatment datasets are normalized to each group's own 10 mm Hg baseline. Dotted lines: 0% change from respective baselines. A cumulative normal function modeled amplitude changes in (A) and (B), which provided a measure of sensitivity (arrow) to IOP (IOP₅₀ ). A nonparametric bootstrap estimated 95% CL (Table). Panel (B) also includes the PhNR analyzed at its maximum trough (solid gray symbols); the general convention, for comparison with the fixed time method used in this paper (see text for details).

Figure 4 also shows that the photopic VEP is resistant to IOP challenge. Even at 70 mm Hg, P1N1 (Fig. 4E) and N1P2 (Fig. 4F) were not significantly reduced (92 ± 13%; P = 0.17 and 80 ± 19%, P = 1 or baseline). Likewise, VEP implicit times were within baseline limits (Figs. 4G–I, all P > 0.05). It should be noted that the b-wave and PhNR amplitude in control eyes decreased as the protocol progressed. These changes occurred at approximately 60 minutes into the protocol (i.e., at 70 mm Hg, b-wave: −16 ± 4%, PhNR: −23 ± 3%, both P ≤ 0.01, Fig. 3A, gray dashed trace). All other photopic ERG and VEP parameters assessed here remained stable throughout (maxiumum change ±10%, all P > 0.05). This reduction may reflect a combination of prolonged anesthesia, corneal drying associated with our photopic protocol, or a reduction in pupillary mydriasis. 

Discussion

Acute IOP Challenge on Retinocortical Pathways

We find that transient IOP elevation produces attenuation of ERG signals derived from inner retinal sources more than it does the VEP. In addition, we found that IOP elevation affects scotopic responses more than photopic responses. 

Effect of IOP Challenge on Scotopic and Photopic Retinal Processes

In general agreement with past studies in rodents, our results show that both scotopic (Figs. 1A, 1B, 2A–C) and photopic ERGs (Figs. 3A, 4A, 4B) declined in amplitude with IOP elevation. ¹⁴ Also, that the IOP needed to attenuate responses by 50% was lowest for the nSTR and PhNR (IOP₅₀ = 53 and 46 mm Hg, respectively, P > 0.05). The pSTR and scotopic b-wave were less sensitive (IOP₅₀ = 60 mm Hg). Finally, the most robust component was the photopic b-wave (IOP₅₀ = 69 mm Hg). As touched upon in the Methods, ERG and VEP components amplitudes returned to with baseline values 1 hour after IOP was restored to normal level (Supplementary Fig. S1). This is consistent with a previous study, ⁴⁰ indicating that the current acute protocol did not induce severe retinal ischemia. 

That significant reduction of ERG parameters occurred when the IOP exceeds 60 mm Hg is well documented. ^25,35,73,74 This appears to involve vascular compromise, ^75,76 as evidenced by the relationship between IOP and blood flow. ^43,72,77 As the choroidal blood supply is more robust to IOP elevation than is the inner-retinal supply, ^42,75,78 RGCs would be more sensitive if the latter was compromised. A vascular mechanism may explain why the nSTR and PhNR are more affected by IOP elevation than their corresponding b-waves. However, the observation that the pSTR was more sensitive than the scotopic b-wave was perplexing. He et al. ⁷⁴ has noted using a similar stepwise IOP protocol in rats, that changes to the nSTR were more pronounced than the pSTR. ^58,74  

It is unclear why the nSTR was more sensitive to stepwise acute IOP elevation than is the pSTR. One might have expected the pSTR to be equally, if not more sensitive, given that it is dominated by RGC responses in the rat. ^29–31 Studies have found that the nSTR and PhNR contain additional contributions from amacrine or glial cell activity. ^30,32,63,79 We postulate the higher nSTR and PhNR susceptibility may result from the widespread effect of acute IOP elevation on Müller cells ^40,80,81 or amacrine cells. ⁸² Also, underestimation of the full pSTR deficit will occur in the presence of a simultaneous loss of the opposing nSTR. 

An interesting finding regarding rat PhNR is the apparent lower sensitivity of the parameter to IOP-induced dysfunction if it is analyzed at its trough (Fig. 4B, gray plot, and Table, PhNR*), than if it is analyzed prior to the deepest negative point post b-wave, at 130 ms after stimulus onset (Fig. 4B, black plot, and Table, PhNR). As similarly evidenced in previous studies with tetrodotoxin (see Methods), IOP-induced changes to the PhNR component were also more prominent near the descending edge of the b-wave (i.e., 70 mm Hg IOP effect at 130 vs. 160 ms; the mean PhNR implicit time of our control data, versus maximum trough was ∼ −100% vs. −54% vs. −36%, respectively). This suggests that the conventional approach of taking PhNR amplitude at the maximal trough may not be optimal to expose the full extent of inner retinal changes. Rather, the inner retinal component of the rat photopic ERG appears to occur earlier on the waveform. That a recent study in mice reported similar IOP-induced PhNR amplitude changes from both mean implicit time-based fixed criterion (A₁₄₅) and maximum trough analyses, ⁴¹ may signify differences between rat and mice PhNR origins. Perhaps, in terms of their unique contributions from amacrine and glia. 

The more notable finding here is that the photopic ERG b-wave was less sensitive to IOP elevation than was its scotopic counterpart. This outcome is consistent with data from Uenoyama et al. ²⁵ who found rod-mediated responses in cats to be more affected by IOP elevation than those of the cone system. This also agrees with Kothe and Lovasik's ²⁴ findings in humans showing the photopic ERG to be less susceptible to changes in retinal vascular perfusion pressure, and with Bui et al., ⁸³ where the photopic ERG persisted for a longer time after postmortem ischemia in rats. However, a previous study in Brown Norway rats failed to find a difference between scotopic and photopic b-wave attenuation during IOP elevation. ³⁵ In the Bui et al. ³⁵ study, separate groups of animals were used for each IOP level, whereas the current study and that of Uenoyama et al. ²⁵ employed a stepwise IOP elevation in the same eye. The within animal design with a stepwise approach may reduce noise and help expose small differences in IOP susceptibility. 

Difference in the sensitivity of photopic and scotopic responses is consistent with the marked differences in metabolic demand under dark- and light-adapted conditions. In particular, the energetic cost of maintaining the dark current is highly ATP demanding. ⁸⁴ Lower oxygen ^84–87 and glucose consumption ^88–92 has been reported under constant light compared with darkness. Not surprisingly, cone-driven function in cats was less vulnerable to mild reductions in glucose concentration compared with rod responses. ⁹³ In the case of IOP elevation, there is evidence for increased glucose consumption in the primate retina, consistent with a shift to more anaerobic glycolysis secondary to IOP-related ischemia. ^89,94 This would act to deplete glucose reserves and have a greater impact on rods compared with cones. 

Effect of IOP Challenge on Scotopic and Photopic Cortical Processes

As might be expected from the upstream changes in the ERG, the scotopic VEP (Figs. 1C, 2F–J) was more sensitive to acute IOP elevation than was the photopic VEP (Figs. 3B, 4E–I; i.e., for N1P2: P = 0.04). Accordingly, the scotopic, but not the photopic, N1P2 parameter was significantly reduced at 70 mm Hg (∼19% difference) and the associated N1 wave delayed. Uenoyama and colleagues ²⁵ qualitatively compared the effect of IOP on scotopic and photopic VEPs, suggesting that photopic VEPs were less sensitive to IOP in cats. These authors report more severe deficits at 70 mm Hg than with our findings. That they held the IOP at each elevation for twice the duration than we did in our procedure (20 vs. 10 minutes) may account for the more severe outcome of their study. In regards to the delay in waveform, Uenoyama et al. ²⁵ did not consider this issue. However, one study found that full body inversion of human subjects (IOP elevation by ∼10 mm Hg) increased the latency (∼7%) of the N1 component of the pattern-onset VEP (e.g., equivalent to our N1 wave ⁹⁵ ). 

Effect of IOP Challenge on Retinal and Cortical Full-Field Responses

Both scotopic and photopic VEPs were resistant to IOP elevation, consistent with other reports. ^23–25,49 Jehle's study ⁴⁹ found that flicker ERG was more affected than flicker VEP during IOP elevation to ischemic levels in rats. Similarly, a differential vulnerability of ERG and VEP to acute alterations in IOP was reported for pattern-evoked responses in humans ^23,24 and nonhuman primates. ⁹⁶ This situation is also evident in our current data, with full-field flash ERG components being more severely attenuated at any given IOP elevation than the simultaneously recorded full-field flash VEP. As has been previously speculated, ²³ cortical dysfunction during acute IOP manipulation may be only an indirect and possibly distorted reflection of retinal dysfunction. Despite reduced retinal output, the VEP remains relatively normal. 

However, it should be noted that the fellow eye is still viable and provides input to the midbrain and cortex contralateral to the IOP-treated eye. ^{52,53,97–99} This input alone elicits a remnant signal that can be up to 40%, of which nonvisual contributions also make up 20% to 30% (Supplementary Fig. S2F, Table S1). Contributions of control signals via callosal routes may also exist. ^54,100,101 Thus, the current approach leads to an overestimation of VEP amplitude from the eye of interest. Nevertheless, even taking into account the possible ipsilateral contribution (i.e., 20%), the VEP was still much less affected by IOP elevation than the ERG (i.e., in rod-driven more so than in the cone-driven pathway: nSTR −94 ± 3% versus scotopic N1P2 −55 ± 13%, P = 0.01; PhNR −100 ± 44% versus photopic N1P2 −40 ± 14%, P = 0.16). Moreover, our observations for a relative preservation of the VEP compared with the ERG during IOP elevation, are consistent with studies employing a monocular occlusion method during recordings. ^{23–25,49,96} Thus, we feel that the present results are consistent with the VEP being less affected by the ERG during IOP elevation. It would be interesting to further investigate how and under what other circumstances this adaptive mechanism can be activated, as well as when it ceases to be adequate. Like the cellular gain control for photoreceptor light adaptation, ^102,103 the onset and degree of adaptation may represent a balance of the benefits and the drawbacks associated with signal amplification processes. 

Conclusions

In summary, this study demonstrated that in rats, rod-mediated retinal function was more sensitive to acute IOP elevation than were cone-mediated retinal signals. Also, similar to past simultaneous ERG/VEP studies, the ERG was more sensitive to transient IOP challenge than is the VEP. This points to the possibility of a post retinal functional compensatory mechanism(s) that can be activated or upregulated when the retina is under stress (i.e., with acute IOP challenge) to maintain the VEP. Moreover, this apparent ERG to VEP response disparity and in particular VEP signal preservation despite stark ERG dysfunction is both relevant for interpreting the results of functional assessment of the visual pathway in animal models and humans. Specifically, in the case of VEP preservation, failure to acknowledge a fast-acting functional compensation would lead to underestimation of retinal injury. 

Supplementary Materials

docx S1 

Acknowledgments

The authors thank Vickie Wong, Zheng He, and Christine Nguyen for their assistance throughout this project. 

Supported by the National Health and Medical Research Council Project (566570 and APP1046203). 

Disclosure: T.I. Tsai, None; B.V. Bui, None; A.J. Vingrys, None