Structural and vascular retinal changes and visual outcomes after treatment of compressive anterior visual pathway lesions: A prospective longitudinal study

Abstract

There is limited evidence regarding longitudinal postoperative changes in retinal vascular density in patients with compressive anterior visual pathway lesions. In this study, we aimed to investigate the changes in retinal vasculature after decompression surgery based on optical coherence tomography (OCT) and OCT angiography (OCTA) and the predictive utility of OCT and OCTA parameters for postoperative visual outcomes. This prospective study included 22 eyes (15 patients) with a compressive anterior visual pathway lesion, which were assessed preoperatively as well as at 1, 3, and 6 postoperative months. Preoperative and postoperative visual acuity (VA), mean deviation on standard automated perimetry, and OCT and OCTA parameters (e.g., retinal nerve fiber layer, macular ganglion cell complex [GCC] thickness, vessel densities of radial peripapillary capillaries, parafoveal and perifoveal superficial vessel densities, and deep vessel densities) were evaluated. At 1 month postoperatively, the average VA and mean deviation were significantly improved (P = .001, P < .001, respectively), while the average values of retinal nerve fiber layer, GCC, and vessel densities of the radial peripapillary capillaries were significantly reduced (P = .027, P < .001, P < .001, respectively), with these trends continuing until 6 postoperative months. Moreover, the average parafoveal superficial vessel densities was significantly reduced at 6 postoperative months (P = .016). Preoperative GCC thickness was a significant predictive factor for VA improvement over the follow-up period (P = .008), while an average GCC thickness of < 70 μm was associated with significantly worse visual outcomes (P = .003). Following decompressive surgery, OCT and OCTA can detect further regression of retinal structure and vasculature over 6 months as a result of retrograde degeneration of the compressed optic nerve with subsequent axonal loss and secondary superficial retinal vascular remodeling. Despite these structural changes, functional visual outcomes improved due to restored nerve transmission in surviving axons. Our findings indicate that average GCC thickness may serve as a predictor associated with visual prognosis and thus should be routinely evaluated.

1. Introduction

Compressive anterior visual pathway lesions can cause visual dysfunction, including visual acuity (VA) loss and visual field (VF) defects.^[1,2] Previous studies have shown that surgical removal can restore vision,^[3] and subsequent studies have attempted to identify prognostic factors associated with visual recovery, including age, duration of symptoms, tumor size, preoperative VA or VF, and pattern electroretinogram. However, none of these factors have consistently predicted postoperative visual outcomes.^[4,5] Optical coherence tomography (OCT) has been used to assess retinal structural damage. Tumor compression causes retrograde degeneration of retinal ganglion cells (RGC) and their axons.^[6] Eyes with chiasmal compression have shown a significant decrease in the average peripapillary retinal nerve fiber layer (RNFL) thickness and macular ganglion cell complex (GCC) thickness.^[7] Additionally, the postoperative average RNFL shows an initial decrease.^[8,9] Preoperative RNFL and GCC thickness has been associated with postoperative VF recovery.^[3,10–12]

Optical coherence tomography angiography (OCTA) is a noninvasive imaging technique that provides an adjunct method for visualizing microvascular peripapillary and macular networks.^[13] It allows detection of ganglion cell axonal damage and examination of retinal perfusion in compressive optic neuropathy.^[14,15] Decreased vascular density in the peripapillary and macular areas is correlated with the quadrants of VF defects.^[13,14,16] However, few studies have compared preoperative and postoperative retinal vascular densities.^{[8,14,17–19]} Cennamo et al reported increase in peripapillary vessel density within 48 hours and at 1 year postoperatively, whereas Lee et al observed significant decreases in parafoveal and peripapillary perfusion within 6 months postoperatively.^[8,19] Despite alterations in the retinal cellular layer and vasculature, the data remain inconclusive at variable postoperative time points. Therefore, in this longitudinal study, we aimed to evaluate postoperative vascular changes in the peripapillary and macular regions and to examine preoperative OCT and OCTA parameters as predictors of visual outcomes. These results may enhance understanding of the mechanisms underlying changes in retinal structure and vasculature morphology and potentially aid in predicting visual outcomes postoperatively.

2. Methods

2.1. Study design

This prospective cohort study was conducted at Songklanagarind Hospital, a large tertiary care center in southern Thailand. The study was approved by the Human Research Ethics Committee of the Faculty of Medicine, Prince of Songkla University (REC 63-022-2-1) and was conducted in accordance with the principles of the Declaration of Helsinki. The study was registered in the Thai Clinical Trials Registry (TCTR20200424002). Written informed consent was obtained from all participants after the nature and possible consequences of the study were explained. The recruitment period for this study was from April 9,2020 to December 31,2022.

2.2. Study population

We enrolled patients with compression of the anterior visual pathway. The inclusion criteria were as follows: VF or VA abnormality; tumor or aneurysm compression of the anterior visual pathway including the optic nerve or optic chiasm, with confirmation through neuroimaging; undergoing decompression of the optic nerve and optic chiasm via craniotomy, transsphenoidal surgery, or endovascular coiling performed by neurosurgeons; (4) complete preoperative baseline VA, VF, and OCTA measurements within 8 weeks; and willingness to adhere to the follow-up schedule at the time of enrollment. The exclusion criteria were as follows: age < 18 years; unreliable preoperative VA or VF tests; poor quality of images, including poor clarity and motion artifacts with quality score < 5/10 or segmentation errors on OCTA; evidence of coexisting ocular diseases that may cause visual abnormality or affect retinal vasculature; and intraoperative complications affecting the visual outcome (e.g., optic nerve injury).

The required sample size was calculated to achieve an 80% power of detecting a significant change in OCTA parameters over time at alpha = 0.05, with an effect size of 0.7. We defined the effect size as the change in OCTA parameters from the preoperative level to that at 6 postoperative months divided by the standard deviation of the change. Based on previous findings by Lee et al, which reported effect sizes of 0.9 for the average superficial retinal capillary plexus and 0.7 for radial peripapillary capillary (RPC) following decompression, we selected 0.7 as an evidence-based estimate for anticipating clinically meaningful changes in vascular parameters in our study.^[8] Given that eyes were the unit of analysis and may be expected to show within-patient correlation, with both eyes expected to be included in about half of the patients, the required sample size was increased based on a design effect of 1.2. Accordingly, the required sample size was 20 eyes.

2.3. Patient data collection

All participants underwent a comprehensive eye examination including VA, VF testing, and OCT/OCTA measurements, at baseline and follow-up schedule at 1, 3, and 6 postoperative months. VA was measured using the logarithm of the minimum angle of resolution (logMAR) scale with the Early Treatment Diabetic Retinopathy Study chart, followed by VF assessment. Participants underwent a VF test using the Humphrey automated perimetry program 30-2, SITA-standard strategy (Humphrey Field Analyzer; Carl Zeiss Meditec, Dublin). Patients with a compressive anterior visual pathway lesion, who could not undergo automated perimetry due to poor fixation, were considered for Goldmann perimetry (Octopus 900; Haag-Streit). Unreliable examinations, defined as false-negative errors > 30%, false-positive errors > 30%, or fixation loss > 20%, were excluded from the analysis. VF parameters, including the mean deviation (MD) and visual field index (VFI), were evaluated. Demographic data and baseline characteristics, including age, sex, tumor pathology and size, lesion location, surgical procedures, operating neurosurgeons, and surgery duration, were recorded in detail.

2.4. Optical coherence tomography and angiography measurements

OCT/OCTA were performed using a commercial spectral domain-OCT system (Optovue, Fremont, Version A2018.1.0.33) with a scanning speed of 70,000 A-scans/s and 840 nm wavelength, to investigate retinal vasculature. The system employed a split-spectrum amplitude-decorrelation angiography algorithm. Vessel density was defined as the percentage of the sampled area occupied by flowing blood vessels. Macular vessel density was measured within 6 mm × 6 mm. A rectangular scan was centered on the fovea in the AngioRetina 6.0 mm mode. Quantitative parameters of the macula included the average percentage of superficial vessel density (SVD) and deep vessel density (DVD). The superficial retinal vascular complexes were automatically segmented from the internal limiting membrane to the inner plexiform layer; further, deep retinal vascular complexes were segmented from the IPL to the outer plexiform layer. The macular area was divided into the parafovea and perifovea. Parafoveal vessel density was measured in an annular region with an inner and outer diameter of 1 mm and 3 mm, respectively, centered on the fovea. Perifoveal vessel density was measured in an annular region with an inner diameter of 3 mm and an outer diameter of 6 mm centered on the fovea. The capillary density of the radial peripapillary capillary (RPC) was measured within a 4.5 × 4.5 mm scan area centered on the optic disc in the mode of Angio-Disc 4.5 mm. The peripapillary area was automatically classified by the software as a 1.0-mm wide annulus extending from the optic disc boundary. The default vascular slab corresponds to the superficial retinal layer, extending from the inner limiting membrane to the posterior boundary of the retinal nerve fiber layer.

Spectral domain OCT was used to map and measure the macular GCC and RNFL thickness. The GCC thickness was measured within a 6 × 6 mm scan area at the macula with the center shifted 0.75 mm temporally. The peripapillary RNFL thickness was measured within a 3.4-mm diameter circular scan. All participants were examined by technicians on the same day following VA assessment. All OCT/OCTA images were obtained after pupil dilation for better image quality and were reviewed by examiners for correct segmentation. The categorizations of OCT/OCTA data were exploratory and based on sample distribution.

2.5. Statistical analysis

All comparison between pre- and postoperative data were analyzed using a mixed-effects random-intercept linear regression, in which the patients and eyes were considered as the random elements, and time was the fixed effect. The mixed model is considered to be robust to sporadic missing values without imputation. Nevertheless, a sensitivity analysis was conducted using 13 eyes (9 patients) with complete follow-up data. The predictive value of OCTA parameters associated with VA outcomes was evaluated using univariable and multivariable linear mixed-effect regression models. Preoperative OCTA values without interaction effects were included in the initial multivariable model. Next, the model was refined by successive backward elimination of variables not significantly contributing to the fit, guided by the change in Akaike information criterion (AIC). We selected the model with the lowest AIC. All statistical analyses were performed using STATA version 14 (StataCorp LP, College Station) and R Core Team version 3.6.1 (2019).^[20] P < .05 was considered statistically significant.

3. Results

3.1. Patient characteristics

Among the 19 patients initially evaluated, 3 patients were excluded because they could not undergo a VF test. There were 26 eyes (16 patients) with compressive optic neuropathy; among them, 4 eyes were excluded because of glaucoma (n = 1), diabetic retinopathy (n = 2), and intraoperative optic nerve injury (n = 1 eye). Finally, 22 eyes (15 patients) were included in the analysis (Fig. 1). During the postoperative period, 13 eyes (9 patients) had complete follow-up visits, and most patients (14 patients, 93.3%) had at least 2 follow-ups. The mean follow-up time was 7.05 ± 0.96 months. The baseline characteristics of the patients are summarized in Table 1. The mean age was 43.7 ± 10.3 years, and majority (66.7%) of the patients were female. The mean duration of visual symptoms was 6.7 ± 9 months. Twelve eyes had unreliable VF test results using Humphrey automated perimetry and were visually evaluated using Goldmann VF perimetry. Nearly half of the affected eyes had VF defects such as central scotoma and temporal hemianopia, mainly due to most tumors being located in the optic chiasm. The most common tumor was meningioma, and the average preoperative tumor volume was 27.4 ± 44.8 cm³. Most patients underwent decompressive surgery by craniotomy with tumor removal; the mean surgical time was 7.1 ± 2.4 hours. We excluded one patient with serious complications due to intracranial hemorrhage after craniectomy who underwent reoperation with intraoperative optic nerve injury.

Figure 1.

Inclusion flowchart of the patients.

Table 1.

Baseline clinicodemographic patient characteristics.

Characteristic	n (%)
Sex
Male	5 (33.3)
Female	10 (66.7)
Location of the lesions
Optic nerve	5 (33.3)
Optic chiasm	10 (66.7)
Compression type
Meningioma	9 (60)
Pituitary adenoma	4 (26)
Oligodendroglioma	1 (6.7)
ACoA aneurysm	1 (6.7)
Procedure
Craniotomy with tumor removal	11 (73.3)
TSS with tumor removal	3 (20)
Endovascular coiling	1 (6.7)

ACoA = anterior communicating artery; TSS = transsphenoidal surgery.

3.2. Comparison between pre- and postoperative visual function

The visual outcomes are summarized in Table 2. The mean preoperative VA and VF measurements, including MD and VFI, were 1.1 ± 0.9 logMAR, −15.9 ± 8.6 dB, and 54.1 ± 28.7%, respectively. The mean VA (Fig. 2A) was significantly improved at 1, 3, and 6 postoperative months (all P < .05). Furthermore, MD (Fig. 2B) and VFI were significantly recovered after decompressive surgery (all P < .05)

Table 2.

Comparison of pre- and postoperative visual function and optical coherence tomography angiography (OCTA) parameters using mixed-effect linear regression analysis.

Variables	Mean (95% CI)				Difference between pre-op and 1-mo post-op		Difference between pre-op and 3 mo post-op		Difference between pre-op and 6 mo post-op
	Pre-op (n = 22)	1 month post-op (n = 18)	3 months post-op (n = 20)	6 months post-op (n = 18)	Mean (SE)	P value	Mean (SE)	P value	Mean (SE)	P value
VA (logMAR)	1.1 (0.76, 1.44)	0.73 (0.38, 1.08)	0.57 (0.23, 0.91)	0.43 (0.08, 0.78)	−0.37 (0.11)	.001*	−0.53 (0.11)	<.001*	−0.67 (0.11)	<.001*
MD† (dB)	−15.95 (−19.65, −12.24)	−6.87 (−11.08, −2.65)	−5.30 (−9.30, −1.29)	−4.32 (−9.20, 0.56)	9.08 (2.21)	<.001*	10.65 (2.11)	<.001*	11.86 (2.03)	<.001*
VFI† (%)	54.1 (42.14, 66.06)	85.05 (71.33, 98.77)	86.69 (73.68, 99.69)	90.99 (78.53,103.45)	30.95 (7.46)	<.001*	32.59 (7.12)	<.001*	36.89 (6.86)	<.001*
RPC density (%)	46.68 (43.44, 49.91)	43.18 (39.8, 46.49)	41.75 (38.48, 45.01)	41.18 (37.87, 44.48)	−3.49 (0.94)	<.001*	−4.93 (0.91)	<.001*	−5.50 (0.94)	<.001*
RNFL thickness (µm)	99.18 (86.50, 111.86)	87.35 (74.10, 100.59)	76.65 (63.75, 89.56)	76.62 (63.37, 89.86)	−11.84 (5.37)	.027*	−22.53 (5.17)	<.001*	−22.58 (5.37)	<.001*
GCC thickness (µm)	79.33 (73.31, 85.36)	76.76 (70.71, 82.80)	74.30 (68.27, 80.33)	74.17 (68.13, 80.22)	−2.58 (0.71)	<.001*	−5.04 (0.67)	<.001*	−5.16 (0.70)	<.001*
Parafoveal SVD (%)	52.09 (50.04, 54.15)	50.29 (48.11, 52.47)	50.55 (48.44, 52.66)	49.54 (47.36, 51.73)	−1.80 (1.06)	.088	−1.54 (1.02)	.131	−2.55 (1.06)	.016*
Perifoveal SVD (%)	47.38 (44.60, 50.15)	46.02 (43.17, 48.88)	46.48 (43.67, 49.28)	45.83 (42.97, 46.69)	−1.35 (0.97)	.162	−0.90 (0.93)	.334	−1.54 (0.97)	.110
Parafoveal DVD (%)	55.91 (53.66, 58.16)	55.65 (53.20, 58.11)	57.58 (55.24, 59.91)	55.82 (53.36, 58.28)	−0.25 (1.47)	.863	1.67 (1.42)	0.239	−0.08 (1.47)	.955
Perifoveal DVD (%)	50.59 (47.28, 53.91)	50.40 (46.79, 54.02)	53.22 (49.78, 56.66)	51.55 (47.94, 55.16)	−0.19 (2.08)	.928	2.63 (2.01)	0.191	0.96 (2.08)	.644

CI = confidence interval; DVD = deep vessel density; GCC = ganglion cell complex; MD = mean deviation; OCTA = optical coherence tomography angiography; RPC = radial peripapillary capillary; RNFL = retinal nerve fiber layer; SE = standard error; SVD = superficial vessel density; VA = visual acuity; VFI = visual field index.

^†Pre-op (n = 10), 1 mo post-op (n = 7), 3 mo post-op (n = 8), 6 mo post-op (n = 9).

^*P < .05.

Figure 2.

Changes in visual function and OCTA parameters during the postoperative (PO) period. (A) Visual acuity (VA); (B) Mean deviation (MD); (C) Average radial peripapillary capillary (RPC) density; (D) Average retinal nerve fiber layer (RNFL) thickness; (E) Average parafoveal superficial vessel density (SVD); (F) Average ganglion cell complex (GCC) thickness. Data points indicate mean, and error bars indicate 95% CI.

3.3. Comparison of pre- and postoperative OCT and OCTA parameters

The pre- and postoperative OCT/OCTA parameters are summarized in Table 2. Comparison between pre- and postoperative values showed that vessel densities of RPC (Fig. 2C), peripapillary RNFL (Fig. 2D), and macular GCC thickness (Fig. 2F) were significantly lower at 1 postoperative month than during the preoperative period, with a continued significant decrease until 3 and 6 postoperative months (all P < .05). Furthermore, the postoperative parafoveal SVD (Fig. 2E) was significantly decreased at 6 postoperative months (P = .005). However, parafoveal and perifoveal DVD and perifoveal SVD showed no significant changes throughout the follow-up period. Figure 3 illustrates a representative case of chiasmal compression at 6 months following decompressive surgery.

Figure 3.

A representative case diagnosed with pituitary adenoma compressed optic chiasm and underwent transsphenoidal surgery. Gray scale of computerized visual field defect (CTVF 30-2) and enface OCTA images of radial peripapillary capillary (RPC) density, retinal nerve fiber layer (RNFL) thickness, macular ganglion cell complex thickness (GCC), perifoveal and parafoveal superficial vessel density (SVD), and perifoveal and parafoveal deep vessel density (DVD) in a representative case.

Nevertheless, a sensitivity analysis using only 13 eyes (9 patients) with complete follow-up data revealed change patterns in VA and OCTA parameters closely similar to those using 22 eyes (15 patients). The only difference was that the subset model showed slightly poorer VA and OCTA parameters, indicating that patients lacking complete follow-up had better visual function on average.

3.4. Predictive factors for visual outcomes among preoperative OCT and OCTA measurements

To identify predictors of visual outcome over time, we categorize the predictors as structural OCT parameters or vascular OCTA parameters and performed mixed-effect linear regression analysis. The average GCC thickness was classified as < 70 mm, 70 to 79 mm, 80 to 89 mm, and ≥ 90 mm. The average parafoveal SVD and DVD were categorized into 2 groups as < 50% and ≥ 50%. The initial variables included in the multivariable model were macular GCC thickness, vessel density of RPC, parafoveal SVD, and parafoveal DVD. Peripapillary RNFL, perifoveal SVD, and perifoveal DVD were not included owing to high correlations with vessel density of RPC, parafoveal SVD, and parafoveal DVD, respectively. The mean preoperative and postoperative logMAR values for VA with respect to preoperative OCTA parameters in the final multivarible mixed-effect linear regression model are summarized in Table 3. Table S1, Supplemental Digital Content, https://links.lww.com/MD/R409 presents the full model output, including regression coefficients (ß), standard errors, 95% confidence intervals (CIs), and P-values for all main effects and interaction terms. Among the preoperative OCTA parameters in the final model, the average GCC thickness was a significant predictor for changes in visual outcome over time (P = .008). At the preoperative period, eyes with GCC thickness < 80 mm had a poorer initial VA than eyes with GCC thickness ≥ 80 mm (P < .001). At the 6-month follow-up, there was pronounced VA improvement in eyes with GCC thickness of 70 to 79 mm. Nevertheless, eyes with GCC thickness < 70 mm had significantly worse postoperative VA than those in other thickness categories (P = .003). Parafoveal DVD was retained in the final model because it showed marginally significant.

Table 3.

Marginal effects of preoperative OCTA measurements adjusting for all other variables and mean logMAR VA in the multivariate mixed-effect linear regression model.

Variables		Mean logMAR VA (95%CI) at each time point				P value*
		Pre-op	1 mo post-op	3 mo post-op	6 mo post-op
GCC thickness (μm)	<70	1.91(1.41, 2.41)a	1.66 (1.16, 2.16)a	1.49 (0.99, 1.99)a	1.40 (0.89, 1.91)a	.008
	70–79	1.67 (1.14, 2.19)a	1.24 (0.68, 1.80)a	0.73 (0.21, 1.25)a,b	0.10 (−0.51, 0.72)b
	80–89	0.82 (0.47, 1.18)b	0.28 (−0.09, 0.64)b	0.22 (−0.14, 0.59)b	0.10 (−0.26, 0.47)b
	≥90	0.32 (−0.15, 0.79)b	0.12 (−0.40, 0.65)b	0.08 (−0.41, 0.58)b	0.02 (−0.46, 0.49)b
P value†		<.001	<.001	.002	.003
Parafoveal DVD (%)	<50	1.48 (0.97, 1.99)	0.69 (0.18, 1.21)	0.60 (0.09, 1.12)	0.32 (−0.21, 0.85)	.059
	≥50	0.99 (0.74, 1.26)	0.76 (0.49, 1.03)	0.58 (0.32, 0.85)	0.42 (0.16, 0.69)
P value‡		.120	.835	.951	.749

CI = confidence interval; DVD = superficial vessel density; GCC = ganglion cell complex; OCTA = optical coherence tomography angiography; logMAR = logarithm of the minimum angle of resolution; VA = visual acuity.

Values within columns not having a superscript in common differ significantly (P Wald test <.05).

^†P value for difference in mean logMAR VA across GCC thickness at each follow-up time.

^‡P value for difference in mean logMAR VA across parafoveal DVD at each follow-up time.

^*P value for difference in change over time across GCC thickness or perifoveal DVD category.

4. Discussion

A critical finding of this longitudinal study is the divergence between functional recovery and structural integrity. While decompression led to significant visual improvement, it did not halt the progressive decline of retinal structure and microvasculature over the 6-month follow-up. We included patients with compression in the anterior visual pathway caused by various types of tumors (e.g., meningioma, pituitary adenoma) or aneurysms. Most patients manifested severe visual impairment and thus required decompression surgery for treatment. Regarding compressive optic neuropathy following surgery, OCTA is used to visualize anatomical changes around the optic nerve head and macular capillary plexus.^{[8,13,16,21,22]} An advantage of the present study is its prospective design, allowing comparison of dynamic changes in peripapillary and macular vessel density, GCC, and RNFL thickness between preoperative and postoperative periods. Furthermore, the tests were serially conducted at 1, 3, and 6 months postoperatively, enabling a better understanding of the temporal changes, whereas previous studies only performed a single postoperative test.^{[13,18,22,23]}

Previous studies have shown that tumor-induced chiasmal compression can cause neuronal degeneration and decrease vessel density.^[22–24] In the absence of decompression, axonal injury worsens, leading to damage of retinal nerve fibers and ganglion cells. In case of decreased metabolic activity in neurons and axons, the oxygen demand is reduced, resulting in superficial vessel regression. The deep retinal vessel may be secondarily affected by anastomosis with the superficial vessels.^[25] Lee et al found that parafoveal and peripapillary perfusion significantly decreased after decompression surgery for chiasmal compression, with the superficial and deep retinal capillary densities showing a relatively high decrease.^[8] In our study, postoperative RPC and parafoveal SVD were also significantly decreased at 1 and 6 postoperative months, respectively. These results might be explained by the chronology of retrograde degeneration. However, parafoveal and perifoveal DVD showed no significant changes throughout the postoperative period. GCC damage might result from delayed retrograde degeneration, leading to secondary changes in the superficial retinal capillary structures before the deep layer. This may be explained by the higher density of retinal ganglion cells in the parafoveal compared to the perifoveal regions.

The decrease in vessel density may be secondary to neural degeneration of the retina, characterized by reduced RNFL and GCC thickness.^[15,25] This was observed in cases of postoperative chiasmal compression in the present study. Anatomically, the retina and optic nerve are closely interconnected.^[25,26] Compression-induced damage to the optic nerve particularly affects the RNFL. Axonal injury that affects retinal ganglion cells and causes axonal death consequently results in RNFL thinning.^[27] In our study, there was significantly reduced thickness of the RNFL and GCC at 1, 3, and 6 postoperative months, which is consistent with previous findings.^[6,11,28]

Our results demonstrated that VA and VFI were significantly improved at 1 postoperative month, with this trend continuing until 6 postoperative months. The duration of visual recovery has varied across previous reports. Lee et al reported a significant improvement in VA at the 4th month after decompression surgery for a pituitary tumor,^[8] while Moon et al reported no significant improvement in chiasmal decompression at the 3rd postoperative month.^[1] Our findings demonstrated that visual function is recovered after decompression surgery, indicating restoration of the nerve transmission by the residual neurons. Visual disturbances may precede ganglion cell loss during compression.^[29] A previous meta-analysis showed that there are multiple predictive factors for postoperative VF improvement in patients with pituitary adenoma, including age, degree of preoperative VF defects, duration of symptoms, and RNFL thickness.^[30] Recent studies have also indicated that RNFL thickness, GCC thicknesses, and RPC density are crucial predictive factors for postoperative visual outcomes.^[12,30,31] In our study, the average GCC was associated with VA outcome. Eyes with preoperative GCC thickness ≥ 80 mm had good VA at both pre- and postoperative periods; additionally, eyes with an average GCC thickness ≥ 70 mm had significantly greater visual improvement at 6 postoperative months than those with GCC thickness < 70 mm. This could be attributed to the profound GCC loss resulting from retrograde degeneration of axonal injury, which may cause delays in visual function recovery. Taken together, average GCC thickness is among the predictive factors for visual recovery in patients with compressive optic neuropathy.

This study has several limitations. First, this study had a small sample size as well as loss to follow-up due to the COVID-19 pandemic. Because patients with incomplete follow-up tended to have better visual function, our findings regarding long-term structural deficits might be overestimated, and functional recovery results should be interpreted with this potential bias in mind. The small sample size impeded analysis of various risk factors that can predict visual outcomes. Our findings regarding average GCC thickness as a predictor should be interpreted as exploratory due to risk of overfitting, lack of adjustment for other confounders (age, symptoms duration, baseline VF defect), and potential bias from categorizing continuous varibales. These predictors require validation in larger cohorts. Second, the postoperative follow-up period of 6 months was relatively short, and further studies are warranted to evaluate the postoperative long-term outcomes. Finally, we included patients with a variety of types of lesions, and we did not evaluate any medical assessment to exclude systemic vascular diseases that may affect the retinal and choroidal vasculature. Despite these limitations, our study remains valuable in that it describes postoperative changes in retinal structure and vasculature. The findings might be helpful in predicting visual recovery in patients with compressive optic neuropathy.

In conclusion, visual function improves following decompression surgery, indicating restoration of nerve transmission in the surviving neurons. However, postoperative decreases in retinal ganglion cell density and perfusion of the peripapillary and superficial parafoveal areas may occur, likely due to retrograde degeneration from prior axonal injury with secondary vascular changes. The average GCC thickness may serve as a predictor of postoperative visual recovery and require validation in adequately powered studies. Furthermore, more severe optic nerve injury and visual impairment may result in delayed recovery of retinal structure and visual function.

Acknowledgments

The authors gratefully thank Dr Alan Geater and Ms. Parichat Damthongsuk from the Faculty of Medicine, Prince of Songkla University, for their assistance with statistical analysis. We would also like to thank Dr Thara Tuntanatip and Dr Chin Taweesomboonyat from the Neurosurgery Unit, Department of Surgery, Faculty of Medicine, Prince of Songkla University, for their care and support of the study patients. Additionally, the authors thank Editage (www.editage.com) for English language editing services.

Author contributions

Conceptualization: Juthamat Witthayaweerasak.

Data curation: Juthamat Witthayaweerasak, Panalee Larpudomphun, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Formal analysis: Juthamat Witthayaweerasak, Panalee Larpudomphun.

Funding acquisition: Juthamat Witthayaweerasak.

Methodology: Juthamat Witthayaweerasak, Panalee Larpudomphun, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Project administration: Juthamat Witthayaweerasak, Panalee Larpudomphun, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Software: Juthamat Witthayaweerasak, Panalee Larpudomphun, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Supervision: Juthamat Witthayaweerasak, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Validation: Juthamat Witthayaweerasak, Panalee Larpudomphun, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Visualization: Juthamat Witthayaweerasak, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Writing – original draft: Juthamat Witthayaweerasak, Panalee Larpudomphun.

Writing – review & editing: Juthamat Witthayaweerasak, Panalee Larpudomphun, Anukoon Kaewborisutsakul, Thakul Oearsakul.

Investigation: Juthamat Witthayaweerasak, Panalee Larpudomphun.

Resources: Juthamat Witthayaweerasak, Panalee Larpudomphun.

Supplementary Material

medi-105-e47769-s001.docx ^{(19.2KB, docx)}