Intracranial pressure affects retinal venular complexity in idiopathic intracranial hypertension: a retrospective observational study

Abstract

Background

Increased intracranial pressure (ICP) in patients with idiopathic intracranial hypertension (IIH) affects the retinal microvasculature, which can be imaged and quantified by optical coherence tomography angiography (OCTA). We aimed to identify the mediating factor between ICP and OCTA parameters association in IIH patients.

Methods

IIH patients with active intracranial hypertension were enrolled. OCTA imaging was performed after ICP measurement. We quantified the branching complexity of the retinal arterioles and venules from the superficial vascular complex of the OCTA image. Eyes of IIH patients were stratified into eyes with papilledema (IIH-P) and eyes without papilledema (IIH-WP). All participants underwent visual acuity (VA) examination.

Results

One hundred and thirty-eight eyes from 70 IIH patients and 146 eyes from 73 controls were included. Compared to the control group, IIH patients and IIH-P had reduced arteriole complexity and increased venule complexity (p < 0.05). For IIH patients and IIH-P, increased retinal venule complexity correlated with increased ICP and reduced VA (p < 0.05); while decreased arteriole complexity only correlated with Frisen scores (p = 0.026). Papilledema mediated the effect (p < 0.001) between ICP and arteriole complexity while ICP had a direct effect (p < 0.001) on venule complexity.

Conclusion

Retinal venules imaged via OCTA may reflect ICP levels and may underpin the direct effect of increased ICP in IIH patients.

Background

Idiopathic intracranial hypertension (IIH) is a neurological condition characterized by increased intracranial pressure (ICP) [1] that predominantly affects young, obese women. Although the pathogenesis has not been fully elucidated, several clinical manifestations, such as papilledema, visual impairment, and retinal bleeding, strongly suggest that increased ICP affects the optic nerve head (ONH) and retina [1, 2]. Venous hemodynamic changes due to increased ICP cause compression of the capillaries in and around the optic nerve head by swollen axons and extracellular fluid accumulation in edema, thus increasing vascular resistance. This makes the optic disc vulnerable to ischemia, ultimately leading to papilledema [3, 4]. Since retinal vessels receive its circulation from the ONH, retinal vessels may be affects. Thus, retinal vessels may provide valuable information on ICP in IIH patients [5,6,7,8], and retinal imaging tools hold promise as an indicator for ICP.

Previous reports using fundus photography showed retinal arteriole-to-venule diameter ratio (AVR) was inversely associated with papilledema and ICP [9, 10]. Sinclair et al. [11] found that blood vessel diameter correlated with the severity of papilledema. Increased ICP causes papilledema through the enlargement of venules in ONH and retina [5, 12], highlighting the role of venules in ICP [5,6,7, 13]. Moreover, using optical coherence tomography angiography (OCTA), we previously showed that IIH patients had reduced retinal microvascular densities compared to controls [14]; we also showed reduced retinal microvascular densities in IIH correlated with their ICP levels.

Few studies have investigated the effect of ICP on retinal arterioles and venules respectively, and their associations with papilledema. The recent development of new technology to segment and quantify retina arterioles and venules in OCTA images provide a step forward in the assessment of retinal microvasculature. Therefore, in this study, we investigate the complex relationship between ICP levels, the severity of papilledema, and retinal microvasculature imaged by the OCTA by segmenting them into arterioles and venules. Through our study, we hope to reveal the role of retinal vasculature (arterioles and venules) in the mechanism of IIH and identify new biomarkers of ICP which may be promising for reducing diagnostic delays and improving the outcomes of patients with IIH. We first hypothesize that IIH patients with papilledema are likely to have worse retinal microvascular metrics. Secondly, we hypothesize that increased ICP in IIH patients may affect retinal vasculature via papilledema.

Methods

Study design and setting

This single-center retrospective, observational study was performed at the Neurology Department of West China Hospital, Sichuan University from April 2021 to June 2023. We included IIH patients and age-sex matched healthy control group. Medical histories included vascular risk factors such as hypertension, dyslipidemia, smoking status, and diabetes mellitus were recorded for all participants. ICP assessment (lumbar puncture), OCTA examination and ophthalmic examination were conduct at the same day.

The research was approved by the Ethics Committee of West China Hospital, Sichuan University, China (No. 2020 [922]) and followed the tenets of the Declaration of Helsinki. Written informed consent was obtained from all participants as a condition of their inclusion in our study.

Inclusion and exclusion criteria

IIH patients who met the internationally accepted criteria and had a cerebrospinal fluid pressure greater than 250mmH₂O were enrolled in our study [15]. Magnetic resonance imaging (MRI) and digital subtraction angiography were performed to identify reasons for increased ICP. The included IIH patients presents with normal image in MRI and part of patients with venous sinus stenosis.

IIH patients with the following were excluded: (1) With other neurological disorders like stroke, brain tumor, trauma, hydrocephalus, and intracranial infection; (2) With other cerebral vascular diseases, like cerebral artery stenosis, arteriovenous malformation, and aneurysm; (3) With uncontrolled hypertension and diabetes mellitus. (4) With concomitant corticosteroids or immunosuppressant therapy. The control group was individuals who attended our hospital for an annual checkup without any neurological and ophthalmological disorders.

ICP assessment via lumbar puncture

Lumbar puncture was performed on all IIH patients at the 3rd – 4th lumbar intervertebral space in lateral position; initial pressure was assessed. Reading was done when fluctuations in the manometer were stabilized. Lumbar puncture was done by skilled neurologists.

Ophthalmic examination

Participants enrolled in our study underwent the following ophthalmic examination: slit-lamp biomicroscopy, fundus photography, visual acuity examination (later converted to logarithm of minimum angle resolution, LogMAR) and OCTA examination. For fundus photography, ONH and fundus images were done and assessed by two specialized ophthalmologists. Participants with ophthalmic disorder such as hemorrhage, microaneurysm, severe cataract, retinoschisis and glaucoma were excluded. Using the modified 5- point Frisen scale, the severity of papilledema was assessed [16]; 0 indicated normal while 5 represented severe papilledema. The two ophthalmologists evaluated a random sample of 20 eyes to assess inter-rate agreement (kappa 0.88, P < 0.001). Eyes from IIH patients were classified into eyes with (IIH-P) and without papilledema (IIH-WP) based on fundus imaging.

OCTA imaging

The OCTA tool (VG200S; SVision Imaging; version 2.1.016) was used for imaging the retinal microvasculature. Our previous studies [8, 17] have well-detailed the specifications of OCTA. The OCTA images covered an area of 6 × 6 mm² centered on the fovea with a raster scan protocol of 512 horizontal B-scans and each B-scan contained 512 A-scans. The en face image of the superficial vascular complex (SVC) was automatically generated by OCTA. SVC was defined as the microvessels 5 μm above the inner limiting membrane to the inner two-thirds of the ganglion cell and inner plexiform layer. Angiograms with retinal abnormalities such age macular degeneration, macular edema and poor angiogram quality (angiograms with artifacts which were caused by rapid head and eye movement during imaging) were excluded. Angiograms with signal quality less than 7 were excluded in our data analysis. OCTA data displayed in this study followed the OSCAR-IB quality criteria [18] and APOSTEL recommendation [19].

Using the OCTA-Net, we computed the complexity of retinal microvascular networks in the SVC [20]. Full details have been previously published [20, 21]. Briefly, images in PNG format were exported from the OCTA tool to the OCTA-Net. The foveal avascular zone was first detected and extracted from the angiogram using a canny edge detector algorithm and a level set al.gorithm. The image was processed to generate an image that contained large blood vessels while removing all small vessels; using global thresholding, a local gray-level change augmentation algorithm (gray-voting), Gabor filtering, and adaptive thresholding were done via Gaussian process regression. Full details on the classification of arterioles and venules in the SVC angiogram have been published [22]. From the image, retinal arterioles and venules were detected and analyzed.

Using skeletonized images, fractal dimension (FD) analysis was used to measure he geometric complexity of and branching pattern of the vasculature in the OCTA. Fractal dimension, D_box, was attained from the OCTA-Net. Values are unitless, with larger numbers reflecting greater complexity which implies denser microvasculature. Figure 1 shows the OCTA imaging and processing. All the above methods were implemented using MATLAB version R2018b (Mathworks, Inc., Natick, MA, USA). In comparison to other geometric metrics, such as the caliber of retinal vessels, fractal analysis can provide more insight into retinal vascular disease development. There is evidence that changes in the fractal dimension might be a predictor of changes in the vascular system [23, 24]; as a result, it may be possible to predict the progression and incidence of retinopathy.

Fig. 1

Optical coherence tomography angiography imaging for an idiopathic intracranial hypertension patient with papilledema (red arrow) and arteriole/venule classification with fractal complexity. OCTA: optical coherence tomography angiography; SVC: superficial vascular complex

Sample size

Prior sample size calculation was not performed due to limited previous reports. The post-hoc power analysis (PASS 2021) was performed to estimate the statistical power; the power for two-group comparison of main results (arteriole and venule complexity) was 0.72 and 0.61 respectively.

Statistical analysis

The normality of our data was test by the Shapiro–Wilk test. Mean ± standard deviation (SD) or median and interquartile ranges (IQR) were used to describe continuous variables while frequencies and percentages were used for categorical variables. The t-test or Mann-Whitney test was used for the comparison of continuous variables between IIH patients and the control group and the Fisher exact test was used for categorical variables.

A generalized estimating equation (GEE) was used when comparing the OCTA parameters between IIH patients and the control group. Multivariable linear regression was used when exploring the correlation between the OCTA parameters and clinical features (ICP, Frisen scores, and VA) or when comparing OCTA parameters among IIH-P, IIH-WP, and control eyes. ICP and Frisen scores were independent variable and OCTA parameters were outcomes; while when investigate VA, OCTA parameters were independent variable and VA was the outcome. When stratified into IIH-P and IIH-WP, subgroup analysis to explore the association between the OCTA parameters and clinical features, was also performed in linear regression.

Mediation analysis was performed to investigate whether the association between ICP and OCTA parameters was mediated by the severity of papilledema (as measured by Frisen scores). Nonparametric bootstrapping (B = 1000) was used to compute 95% confidence intervals (95% CI) for the total effect, mediation effects, and direct effects.

The covariates for GEE, linear regression, and mediation analysis were set as age, gender, vascular risk factors (hypertension, diabetes, dyslipidemia, smoking, and drinking), and inter-eye dependencies. The added variable plot was used to demonstrate the partial correlation between clinic features and the arteriole/venule FD values. The coordinate axis of added variable plots were the residuals of the independent variable and the dependent variable when these variables were regressed on the covariates. Statistical analysis and plotting were conducted in R version 4.2.3 (gee package for GEE, mediation package for mediation analysis), and p < 0.05 were considered significant.

Results

84 IIH patients and 74 controls were initially enrolled;14 IIH patients were excluded because of coexisted neurological or ophthalmic diseases as shown in Supplementary Fig. S1. Final analysis included 138 eyes from 70 IIH patients (mean age: 34.71 ± 11.77 years; 45.71% males) and 146 eyes from 73 controls (mean age: 40.10 ± 11.93; 34.25% males). Of the 138 eyes in IIH patients, 126 eyes had papilledema (IIH-P) while 12 eyes did not have papilledema (IIH-WP). Importantly, IIH patients had reduced VA (p < 0.001) compared to controls. Compared to the control group, IIH patients had reduced arteriole (p = 0.014) and increased venule (p = 0.015) complexity after adjusting for risk factors. Table 1 shows the demographics and clinical information of our study participants.

Table 1 Demographics, clinical and OCTA data of study participants

Supplementary Fig. S2 shows the comparison of OCTA parameters among IIH-P eyes and IIH-WP eyes compared to the control group. Compared to the control group, IIH-P eyes showed reduced arteriole complexity (p = 0.008) and increased venule complexity (p = 0.012). No significant differences (p > 0.05) were seen when the control group was compared to IIH-WP eyes; correspondingly, no significant differences (p > 0.05) were seen when IIH-P eyes were compared to IIH-WP eyes.

Figure 2 displays the association between OCTA parameters and clinical features in IIH patients. Increased venule complexity was significantly correlated with increased ICP levels (p < 0.001) and reduced VA (p = 0.003), but not with the severity of papilledema (p = 0.131). Regarding retinal arterioles, reduced arteriole complexity was only significantly correlated with the severity of papilledema (Friesen scores, p = 0.026).

Fig. 2

The association between optical coherence tomography angiography parameters and clinical features (ICP for A and VA for B) in idiopathic intracranial hypertension patients. The X and Y axis for the added variable plot were the residuals of X and Y when they were regressed on the covariates. ICP: intracranial pressure; VA: visual acuity; FD: fractal dimension; LogMAR: logarithm of the minimum angle of resolution

Figure 3 illustrates the subgroup analysis for the association between OCTA parameters of IIH patients and their clinical features. Increased venule complexity in IIH-P eyes correlated with ICP (p < 0.001) and VA (p = 0.005) while reduced arteriole complexity in IIH-P eyes correlated (p = 0.045) with Frisen scores. In IIH-WP eyes, increased venule complexity correlated with ICP (p = 0.031); no other significant correlations (p > 0.05) were found between OCTA parameters and their clinical features.

Fig. 3

Subgroup analysis for the association between optical coherence tomography angiography parameters (arteriole complexity for A and venule complexity for B) of idiopathic intracranial hypertension patients and their clinical features. The X and Y axis for the added variable plot were the residuals of X and Y when they were regressed on the covariates. ICP: intracranial pressure; VA: visual acuity; FD: fractal dimension; LogMAR: logarithm of the minimum angle of resolution; IIH-P: eyes with papilledema in idiopathic intracranial hypertension patients; IIH-WP: eyes without papilledema in idiopathic intracranial hypertension patients

Figure 4 displays the mediation analysis of whether the association between ICP and retinal microvasculature was mediated via the severity of papilledema (Frisen scores). ICP had a significant direct effect (p < 0.001), and a total effect (p < 0.001) on increased venule branching complexity in IIH patients. The association between increased ICP levels and lower arteriole branching complexity was mediated (p < 0.001) by the severity of papilledema (measured by Frisen scores).

Fig. 4

A mediation analysis of the association between ICP and optical coherence tomography angiography parameters (arteriole complexity for A and venule complexity for B) via higher Frisen scores. Total (red), indirect (blue), and direct effect (green) were displayed. ICP: intracranial pressure

Discussion

Our current study characterized retinal arterioles and venules in IIH patients and found that IIH patients have reduced arteriolar and increased venular branching complexity compared to control group. We also found that retinal vascular branching complexity correlates with ICP levels and papilledema. In addition, the effect between ICP and arteriole branching complexity was mediated by papilledema.

IIH frequently has retinal manifestations due to increased pressure of the cerebrospinal fluid and impacts retinal vessels [3, 14]. It is suggested that changes in the retinal microvasculature are an indicator of retinopathy onset and progression in IIH. Using the OCTA, our previous report [14] showed IIH patients had reduced retinal microvascular density when compared to controls. Our current study extended our previous findings and investigated the retinal arterioles and venules in IIH patients compared to controls. We showed IIH patients had reduced arteriolar and increased venular branching complexity when compared to controls, which is in line with previous fundus photograph reports [6] that showed narrower retinal arterioles and wider retinal venules. There is increasing evidence that during IIH, vessels in and around the ONH and retina are affected [2, 25, 26]. Fundus imaging reports on IIH patients showed venous congestion and compression of capillaries due to edema (resulting from papilledema) [3, 6, 27]. This results in reduced blood flow in the eye ultimately leading to hypoxia and/or ischemia. Here we suggest that lower arteriolar branching complexity may be due to the hypoxia and/or ischemia while increased venular branching complexity may be due to venous congestion as a result of inflammation in IIH patients. Taken together, retinal arteriolar and venular branching complexity may be due to the disease cascade of IIH.

We further showed that IIH-P eyes had reduced retinal arteriolar and increased venular branching complexity compared to the control group. Papilledema is the clinical hallmark of increased ICP [28] where increased cerebrospinal fluid (CSF) pressure in the subarachnoid space compresses the optic nerve fibers within the ONH when passing through the retrobulbar space of the orbital cavity [5, 27]. Furthermore, it is generally believed that increased CSF pressure in the ONH leads to significant changes in the retinal vasculature [29]. These retinal vascular changes may be due to the fact these vessels originate from the ONH thus liable to changes in the ONH with subsequent vessel compression and congestion (i.e., retinal arteriolar narrowing and venular engorgement).

It is suggested that changes in ICP levels are directly correlated with retinal vascular dynamics [30]. The retinal venules are associated with variations in cerebral venous pressures downstream [6]. Secondly, as the central retinal vein leaves the eye, elevated ICP in the subarachnoid space inside the optic nerve sheath exerts external pressure on it, increasing vascular resistance and resulting in venous congestion [31, 32]. Previous reports [5, 6] using fundus photography showed enlarged retinal venules correlates with increased ICP. Moreover, previous reports [9, 10] showed increased arteriovenous ratio correlated with increased optic disc elevation, which is suggested to be linked with ICP. In our current study, we showed increased venular branching complexity in IIH patients and eyes with IIH-P correlated with ICP. Importantly, we showed that increased ICP had a direct effect on retinal venules in IIH patients. It is suggested that papilledema occurs as a result of increased ICP, which exerts external pressure on the retinal vein, ultimately resulting in venous congestion [32]. Since elevated ICP may cause papilledema, which may be linked it venular changes [5, 33], it is plausible to suggest that increased venular branching complexity is linked with ICP levels in IIH. Importantly, we showed increased venular branching complexity in IIH-WP eyes correlated with ICP. Notably, papilledema did not develop in these eyes, reinforcing that elevated ICP can induce retinal venular changes in the absence of papilledema. Taken together, our data provide convincing results to support the potential of retinal venules as a biomarker to aid in diagnosing and monitoring ICP states in IIH patients.

The main clinical method for evaluating papilledema in IIH patients is the Frisen scale [11]. Although the Frisen scale evaluates the optic nerve changes, we showed reduced arteriolar branching complexity correlated with Frisen scores in IIH patients. This may suggest that quantitative measures of retinal arterioles may be useful in evaluating the severity of papilledema in IIH patients. Even though the causation cannot be proven in this cross-sectional, non-interventional study, we showed that increased ICP in IIH patients alters retinal arterioles through papilledema. ICP has several deleterious effects on the retinal microvasculature that can be quantified. Based on previous reports [11, 28], we identified that papilledema is the most likely mediator of retinal arteriolar changes. We found papilledema may be involved in retinal arteriolar changes by increased ICP in IIH patients, suggesting that severe papilledema may be the main driver of arteriolar changes in IIH patients.

To our knowledge, this is the first study of a comprehensive assessment of quantitatively measured retinal arterioles and venules in IIH patients using OCT angiograms. We used the fractal dimension which is a sensitive indicator of early retinal vascular changes. Unlike retinal vascular caliber assessment such as AVR, which represent a single parameter of the retinal vessels, retinal fractal dimension is a global measure of the complexity of vascular structure measure because it reflects relatively static blood distribution optimality and efficiency [23, 24]; similarly, the retinal vessel density used in our previous studies also reflects the blood flow in the vessels. The differences in the retinal microvasculature in IIH patients compared to controls show how effectively vascular patterns span the retina in determining susceptibility to IIH and how fractal dimensions metrics could answer potentially significant clinical questions. Here we showed that IIH patients had lower arteriolar branching complexity (lower fractal dimension due to hypoxia) and higher venular branching complexity (higher fractal dimension due to inflammation) when compared to controls. We also showed that arteriolar branching complexity decrease with the progression of IIH and may be a pointer of papilledema. Importantly, we found that increased retinal venular branching complexity correlated with ICP levels in IIH. Increased retinal venular branching complexity (which may be due to the venous stasis [inflammation]) correlated with ICP levels in IIH. As lumbar puncture is invasive, significant correlation between increased retinal venular branching complexity and ICP levels suggest that quantitative measurement of the retinal venules in IIH patients may be a promising imaging indicator of ICP levels because of its non-invasiveness. Further validation would be of immense interest.

Several limitations need to be addressed. First, the observational, non-interventional design of our present study did not allow us to determine the temporal sequence of the associations observed. Further, longitudinal, and interventional study designs are required to warrant our findings. Moreover, the sample size of the IIH-WP was relatively small, which may influence statistical power. In our current cohort, over 40% of IIH patients were male. This may be due to the exclusion of patients with ophthalmic disorders such as retinoschisis or extremely poor visual acuity; most of the participants excluded were female. More IIH patients are proposed to be recruited and more rational analyses are required to conduct. The results were obtained from a single center, which limits the generalizability of the data to the general Chinese race.

In conclusion, we found that IIH patients have reduced arteriolar and increased venular branching complexity. We also found that ICP levels directly affect retinal venular branching complexity while indirectly affect arteriolar branching complexity via papilledema. Future studies with intervention and follow-up will be needed to test whether restoring the mediator improves the retinal vasculature or prevents dysfunction. This will enable us to prove its causal role in ICP-related retinal vascular changes.