Early CSF inflammatory markers after aneurysmal subarachnoid hemorrhage and their relationship to disease severity and shunt placement

Abstract

Background:

The inflammatory response within the central nervous system is a key driver of secondary brain injury after aneurysmal subarachnoid hemorrhage (aSAH). Less is known about the impact that inflammation has on complications like persistent post-hemorrhagic hydrocephalus. To explore the association between inflammation, disease severity, and permanent shunt placement, we characterized the early cytokine profiles of the blood and cerebrospinal fluid (CSF) of patients with aSAH.

Methods:

Biological samples were collected from aSAH patients admitted to a single-center Neurosciences Intensive Care Unit between 2014 and 2024. Control CSF samples were collected from patients undergoing permanent shunt placement for normal pressure hydrocephalus. A multiplex bead-based immunoassay was used to analyze a panel of cytokines in plasma and CSF samples. Clinical variables, including demographics, disease severity, and permanent shunt placement were collected.

Results:

Plasma and/or CSF samples were collected from 83 patients (58 aSAH patients, 25 controls). In aSAH patients, C—C motif chemokine ligand-2 (CCL2), interleukin-6 (IL-6), granulocyte-colony stimulating factor (G-CSF), interleukin-8 (IL-8), and vascular endothelial growth factor (VEGF) were all elevated in CSF compared to plasma (p < 0.05 for all comparisons) and in the CSF of aSAH patients as compared to controls (p < 0.001 for all comparisons). However, only G-CSF and VEGF were associated with clinical severity at presentation when considering Hunt and Hess score as a dichotomized variable (p = 0.026 and p = 0.043, respectively). In multivariable models adjusted for age, sex, and modified Fisher Scale score, early CSF concentrations of IL-6 and IL-8 were associated with increased need for permanent shunt placement (p = 0.030 and p = 0.040, respectively).

Conclusions:

Within 72 hours of aSAH, proinflammatory cytokines can be detected at higher concentrations in CSF than in plasma, and at higher concentrations in aSAH patients compared to controls. Early concentrations of certain pro-inflammatory cytokines are associated with increased likelihood of persistent post-hemorrhagic shunt dependent hydrocephalus, independent of initial disease severity. These data support preclinical models of CNS inflammation after aSAH and suggest that early innate inflammation contributes to hydrocephalus.

Introduction

Despite significant advances in the acute management of aneurysmal subarachnoid hemorrhage (aSAH), short- and long-term outcomes remain poor. Mortality remains as high as 35 % after aSAH, and while 67 % of survivors regain functional independence, only 33 % return to work and 50 % have persistent cognitive impairment.^1–3 The inflammatory response driven by the presence of blood within the brain parenchyma, ventricles, and subarachnoid space is a key driver of secondary brain injury and poor outcome in aSAH.^4–7 In animal models of aSAH, inflammation within the ventricular system is associated with persistent post-hemorrhagic hydrocephalus,^8–10 but the relationship between cerebrospinal fluid (CSF) inflammation and post-hemorrhagic shunt dependent hydrocephalus (SDHC) in patients has not been fully elucidated.^11–14

While recent human studies on intracerebral hemorrhage have reported an initial pro-inflammatory phase involving central nervous system (CNS)-resident and circulating immune cells followed by an anti-inflammatory reparative phase,¹⁵ few studies have investigated a panel of pro- and anti-inflammatory cytokines following aSAH in the blood and CSF in a large cohort of patients.^{7, 13} In this study, we sought to more fully characterize early CNS inflammation after aSAH in patients, investigate the relationship between CSF cytokines and markers of disease severity, and determine whether early CSF cytokine levels are associated with SDHC. To achieve this goal, we analyzed a panel of cytokines in biological samples collected within the first 72 hours after symptom onset. We measured CSF and plasma levels of cytokines that were previously shown to be detectable in the CSF of aSAH patients,^{7, 13, 16–19} several of which have been shown to be produced within the CNS by astrocytes, microglia, ependymal cells, and the choroid plexus.^20–26 We then compared CSF cytokine levels between aSAH patients and controls and investigated the relationship between the cytokines, markers of disease severity, and SDHC.

Methods

Study design

In this case-control study, we included 58 patients with aSAH admitted to the Yale Neurosciences Intensive Care Unit (ICU) between 2014 and 2024 with plasma and/or CSF samples collected after enrollment in the Yale Longitudinal Study of Acute Brain Injury Biorepository. CSF was collected from enrolled patients who had external ventricular drains (EVDs) placed as part of their clinical care.

This dataset represents a convenience sample of blood and CSF samples collected from eligible patients within 72 hours after symptom onset; the first sample collected from each patient was included in the analyses. Patients with plasma and CSF samples collected at the same time (n = 14) were included in our paired samples cohort. Patients with culture-confirmed ventriculitis (n = 6) were excluded from the analyses; CSF cytokine concentrations from these patients are shown in Supplemental Table 1. Samples from 25 patients undergoing first time placement of a ventriculoperitoneal shunt for treatment of normal pressure hydrocephalus, who did not have prior brain hemorrhage, were enrolled as controls. CSF was collected from the cerebral ventricles during shunt placement. This study was approved by Yale School of Medicine IRB, and consent was obtained from patients or their legally authorized representative.

Immunologic Assays

CSF was collected in no additive polyethylene tubes, and blood was collected in heparinized polyethylene tubes (BD Biosciences, Franklin Lakes, NJ). Within one hour of collection, samples were centrifuged at 500 g for 10 minutes. The supernatant was divided into aliquots, and samples were frozen at −80°C. A panel of cytokines and chemokines including interleukin-6 (IL-6), granulocyte-colony stimulating factor (G-CSF), IL-8, vascular endothelial growth factor (VEGF), C—C motif chemokine ligand-2 (CCL2), IL-1β, IL-4, macrophage inflammatory protein-1 alpha (MIP-1α), IL-10, tumor necrosis factor alpha (TNF-α), and IL-17 were analyzed using Cytometric Bead Array (BD Biosciences), a multiplex bead-based immunoassay run on an LSR Fortessa flow cytometer (BD Biosciences). VEGF was not measured in plasma samples, since prior work in our laboratory showed that VEGF was not detectable in the plasma of hemorrhagic stroke patients at these early timepoints (unpublished data). The lower limit of detection for each immune molecule was the concentration of the lowest standard on the linear curve for that assay.

Clinical Variables

Demographics, medical history, and hospitalization data were prospectively obtained from chart review and follow-up interviews with patients and/or surrogates. Clinical and radiographic scores including admission Glasgow Coma Scale (GCS) score, Hunt and Hess (HH) score, and Modified Fisher Scale (mFS) score, were extracted from patient charts by trained raters (JY, RC, SJ, LG, AA). Occurrence of delayed cerebral ischemia (DCI), defined as the occurrence of focal neurologic impairment or a decrease of at least 2 points on the GCS lasting at least 1 hour and not attributable to an alternate cause,²⁷ intraventricular hemorrhage (IVH) and permanent shunt placement were abstracted from the electronic medical record by trained raters (JY, RC, SJ, LG, AA). All scores and hospitalization data were adjudicated by a panel of board certified neurointensivists (JMB, JK, NP, LO).

Statistical analysis

Non-normally distributed concentrations of cytokines were log transformed. In our paired samples cohort, paired t-test was used to compare log CSF and plasma concentrations of the cytokines. For CSF samples, t-test was used to compare log CSF cytokine concentrations between control and aSAH samples. Linear regression was used to evaluate for significant associations between initial severity features and log CSF cytokine concentrations. Stepwise backwards variable selection linear regression was used to determine a parsimonious multivariable linear regression model to evaluate the association between log CSF cytokine concentrations and severity features. Logistic regression was used to evaluate for significant associations between log CSF cytokine concentrations and permanent shunt placement. No adjustment for multiple comparisons was done as this is an exploratory, hypothesis-generating study. There was no missing clinical or demographic data. For samples in which there was missing cytokine data, those samples were excluded from analysis only of the missing cytokine. There was no a priori sample size calculation as this was an analysis of previously collected samples. Based on the 80 total CSF samples collected, the study had 80 % power to detect a mean difference of 4 with a standard deviation of 6 or a 95 % power to detect a mean difference of 4 with a standard deviation of 4.5 in log-transformed cytokines between control (n = 25) and aSAH (n = 55) samples using a two-sample t-test with an alpha = 0.05. R software (version 4.3.1) was used for all data management and statistical analyses.

Results

Plasma and/or CSF samples from 83 patients (n = 58 aSAH, n = 25 control) were analyzed. Patients had a median age of 62 years (interquartile range [IQR] 18-90), 42.1 % were male, 84.2 % were non-Hispanic, and 76.3 % were Caucasian (Table 1). The median time from symptom onset to sample collection was 2 [1-3] days for CSF and 1 [1-3] days for plasma. The measured levels of each cytokine and lower limits of detection are shown in Supplemental Table 2 for plasma samples and Supplemental Table 3 for CSF samples. Demographic features including age, sex, race, and ethnicity were not associated with CSF cytokine concentrations (Supplemental Table 4).

Table 1.

Baseline Demographics (All patients).

	Control (N = 25)	aSAH (N = 58)	Overall (N = 83)
Sample type †
Plasma	NA	18	18
CSF	25	55	80
Age
Mean (SD)	76.9 (4.30)	57.8 (14.1)	63.5 (14.9)
Median [Min, Max]	77.0 [67.0, 88.0]	59.0 [18.0, 90.0]	63.0 [18.0, 90.0]
Sex
Female	7 (28.0 %)	39 (67.2 %)	46 (55.4 %)
Male	18 (72.0 %)	19 (32.8 %)	37 (44.6 %)
Ethnicity
Hispanic	0 (0 %)	12 (20.7 %)	12 (14.5 %)
Non-Hispanic	25 (100 %)	46 (79.3 %)	71 (85.5 %)
Race
White or Caucasian	24 (96.0 %)	40 (69.0 %)	64 (77.1 %)
Black or African American	1 (4.0 %)	5 (8.6 %)	6 (7.2 %)
Asian	0 (0 %)	3 (5.2 %)	3 (3.6 %)
Other	0 (0 %)	6 (10.3 %)	6 (7.2 %)
Unknown	0 (0 %)	4 (6.9 %)	4 (4.8 %)

aSAH, aneurysmal subarachnoid hemorrhage; CSF, cerebrospinal fluid; ICH, intracerebral hemorrhage; SD, standard deviation.

^†14 aSAH patients had both CSF and plasma samples collected.

Baseline demographics of control and aneurysmal subarachnoid hemorrhage patients from whom CSF and plasma samples were collected.

Cytokine concentrations are elevated in the CSF compared to plasma

CCL2, IL-6, G-CSF, IL-8, and VEGF were detectable in the CSF of patients with aSAH within 72 hours after symptom onset. These proinflammatory cytokines were present in higher concentrations in the CSF compared to the plasma (Figure 1A, p < 0.05 for all comparisons).

Fig 1.

Comparison of cytokine concentrations between CSF and plasma samples.

In CSF and plasma samples collected between day 0 and 3 after symptom onset, log concentrations of VEGF, IL-8, IL-6, G-CSF, and CCL2 were all significantly higher in CSF collected from aSAH patients (n = 58, n = 55 for VEGF, n = 57 for IL-8) compared to control CSF samples (n = 25) and plasma collected from aSAH patients (n = 18, VEGF not measured), as shown in heat map (A). (B) In paired CSF and plasma samples collected from the same patients with aSAH at the same time, log concentrations of CCL2, IL-6, G-CSF, and IL-8 were all significantly higher in CSF compared to plasma samples (n = 14).

aSAH, aneurysmal subarachnoid hemorrhage; CCL2, C—C motif chemokine ligand-2; CSF, cerebrospinal fluid; G-CSF, granulocyte colony stimulating factor; IL-6, Interleukin-6; IL-8, Interleukin-8; VEGF, vascular endothelial growth factor.

In the 14 aSAH patients with paired CSF and plasma samples, the mean log concentrations (in pg/mL) of all detectable cytokines were significantly higher in the CSF of aSAH patients compared to paired plasma samples, including CCL2 (CSF 8.13 vs plasma 4.79, p < 0.001), IL-6 (CSF 6.80 vs plasma 2.99, p < 0.001), G-CSF (CSF 2.54 vs plasma 1.65, p = 0.004), and IL-8 (CSF 6.34 vs plasma 2.48, p < 0.001; Figure 1B).

Differences in CSF cytokines between patients with aSAH and controls

In univariate analyses, compared to CSF collected from controls, patients with aSAH had significantly higher CSF log concentrations (in pg/mL) of CCL2 (controls 6.50 vs aSAH 8.25, p < 0.001), IL-6 (controls 1.32 vs aSAH 7.41, p < 0.001), G-CSF (controls 2.04 vs aSAH 3.25, p < 0.001), IL-8 (controls 2.23 vs aSAH 6.58, p < 0.001), and VEGF (controls 2.25 vs aSAH 3.94, p < 0.001) (Figure 2A). In multivariable linear regression analyses, adjusted for age and sex, log concentrations of all detectable cytokines remain significantly higher in aSAH patients compared to controls: CCL2 (β=1.59, 95 % CI 1.12-2.05, p < 0.001), IL-6 (β=6.24, 95 % CI 5.32-7.16, p < 0.001), G-CSF (β=1.04, 95 % CI 0.12-1.97, p = 0.028), IL-8 (β=4.42, 95 % CI 3.58-5.27, p < 0.001), and VEGF (β=2.20, 95 % CI 1.29-3.11, p < 0.001).

Fig 2.

Comparison of CSF cytokine concentrations

(A) In CSF samples collected from patients with aSAH and controls who were undergoing permanent shunt placement for normal pressure hydrocephalus, log concentrations of CCL2, IL-6, G-CSF, IL-8, and VEGF were all significantly higher in aSAH patient samples (n = 58, n = 55 for VEGF, n = 57 for IL-8) compared to control CSF samples (n = 25). (B) In univariate analyses, higher early log concentration of CSF IL-6 in patients with aSAH was associated with shunt dependent hydrocephalus (n = 58).

aSAH, aneurysmal subarachnoid hemorrhage; CCL2, C—C motif chemokine ligand-2; CSF, cerebrospinal fluid; G-CSF, granulocyte colony stimulating factor; IL-6, Interleukin-6; IL-8, Interleukin-8; VEGF, vascular endothelial growth factor.

Association of cytokines and features of severity

Patients included in our study demonstrated features associated with severe neurological injury, including low admission GCS (median 10 [IQR 3-15]), high HH score (3 [1-5]), and high rate of DCI (43.1 %; Table 2). On initial imaging, patients had radiographically severe disease with high mFS scores (median 4 [IQR 1-4]) and very high rates of IVH (81.8 %) and hydrocephalus (78.2 %) (Table 2). Three patients without EVDs, from whom only plasma samples were collected, were included in our dataset. These patients had significantly higher admission GCS scores and lower HH scores compared to patients with EVDs from whom CSF and/or plasma samples were collected (Table 2).

Table 2.

aSAH Patient and Sample Characteristics.

	CSF and/or plasma (N = 55)	Plasma only (N = 3)	p-value	Overall (N = 58)
Admission Glasgow Coma Scale score
Mean (SD)	9.64 (4.58)	15.0 (0)	<0.001	9.91 (4.61)
Median [Min, Max]	9.00 [3.00, 15.0]	15.0 [15.0, 15.0]		10.0 [3.00, 15.0]
Hunt Hess Score
Mean (SD)	3.24 (1.15)	2.00 (0)	<0.001	3.17 (1.16)
Median [Min, Max]	3.00 [1.00, 5.00]	2.00 [2.00, 2.00]		3.00 [1.00, 5.00]
modified Fisher Scale Score
Mean (SD)	3.62 (0.623)	2.00 (1.00)	0.104	3.53 (0.731)
Median [Min, Max]	4.00 [2.00, 4.00]	2.00 [1.00, 3.00]		4.00 [1.00, 4.00]
Intraventricular hemorrhage on initial imaging
Yes	45 (81.8 %)	0 (0 %)	0.009	45 (77.6 %)
No	10 (18.2 %)	3 (100 %)		13 (22.4 %)
Hydrocephalus on initial imaging
Yes	43 (78.2 %)	0 (0 %)	0.020	43 (74.1 %)
No	12 (21.8 %)	3 (100 %)		15 (25.9 %)
Delayed cerebral ischemia
Yes	25 (45.5 %)	0 (0 %)	0.342	25 (43.1 %)
No	30 (54.5 %)	3 (100 %)		33 (56.9 %)
Sample collection time (post-bleed day)
0	7 (12.7 %)	0 (0 %)	0.77	7 (12.1 %)
1	15 (27.3 %)	1 (33.3 %)		16 (27.6 %)
2	8 (14.5 %)	0 (0 %)		8 (13.8 %)
3	25 (45.5 %)	2 (66.7 %)		27 (46.6 %)
Shunt dependent hydrocephalus
Did not survive to shunt decision	12 (21.8 %)	0 (0 %)	0.435	12 (20.7 %)
Yes	8 (14.5 %)	0 (0 %)		8 (13.8 %)
No	35 (63.6 %)	3 (100 %)		38 (65.5 %)

aSAH, aneurysmal subarachnoid hemorrhage; CSF, cerebrospinal fluid; SD, standard deviation.

Clinical and imaging characteristics of aneurysmal subarachnoid hemorrhage patients from whom CSF and plasma samples were collected.

When investigating the association between early CSF cytokine concentrations and disease severity features including admission GCS, HH score, mFS score, presence of IVH on initial imaging, and hydrocephalus on initial imaging, we found very few significant associations. Higher log concentrations of G-CSF were associated with a higher HH score at admission (p = 0.042), and there was a trend towards an association between a higher log concentration of VEGF and higher HH score (p = 0.060). In linear regression analyses in which HH score at admission was dichotomized (low grade defined as HH score 1-3, high grade defined as HH score 4-5), there was a significant association between a higher log concentration of G-CSF and a higher HH score at admission (β coefficient −0.89, 95 % CI −1.67 to −0.11, p = 0.026). Additionally, in these analyses, there was a significant association between a higher log concentration of VEGF and a higher HH score at admission (β coefficient −0.81, 95 % CI −1.59 to −0.03, p = 0.043). There were no associations between early CSF cytokine concentrations and admission GCS or mFS score in univariate analyses (Table 3). There was also no association between early CSF cytokine concentrations and presence of IVH or early hydrocephalus on initial imaging (p > 0.05 for all cytokines).

Table 3.

Association of CSF cytokines with severity factors (univariate analysis).

CSF samples (n = 55)†	β Coefficient	95 % CI	R2	p-value
log[CCL2]
Admission GCS	−0.03	−0.07 to 0.01	0.04	0.161
Hunt and Hess score	0.11	−0.06 to 0.27	0.03	0.195
modified Fisher Scale score	0.08	−0.24 to 0.39	0.004	0.628
Intraventricular hemorrhage	0.32	−0.17 to 0.81	0.03	0.199
Early hydrocephalus	0.23	−0.23 to 0.69	0.02	0.323
log[IL-6]
Admission GCS	−0.02	−0.11 to 0.06	0.01	0.580
Hunt and Hess score	0.11	−0.23 to 0.45	0.01	0.517
modified Fisher Scale score	−0.09	−0.73 to 0.54	0.002	0.769
Intraventricular hemorrhage	0.73	−0.26 to 1.73	0.04	0.147
Early hydrocephalus	0.19	−0.76 to 1.14	0.003	0.687
log[G-CSF]
Admission GCS	−0.05	−0.14 to 0.04	0.02	0.250
Hunt and Hess score	0.35	0.01 to 0.70	0.08	0.042*
modified Fisher Scale score	0.11	−0.54 to 0.77	0.002	0.730
Intraventricular hemorrhage	0.23	−0.82 to 1.28	0.004	0.664
Early hydrocephalus	−0.53	−1.51 to 0.44	0.02	0.276
log[IL-8]
Admission GCS	−0.003	−0.08 to 0.08	0.00	0.941
Hunt and Hess score	0.06	−0.26 to 0.37	0.002	0.711
modified Fisher Scale score	0.24	−0.35 to 0.82	0.01	0.418
Intraventricular hemorrhage	0.33	−0.60 to 1.26	0.01	0.482
Early hydrocephalus	0.24	−0.63 to 1.11	0.01	0.586
log[VEGF]
Admission GCS	−0.09	−0.17 to 0.002	0.07	0.051
Hunt and Hess score	0.33	−0.01 to 0.67	0.07	0.060
modified Fisher Scale score	−0.20	−0.84 to 0.44	0.01	0.529
Intraventricular hemorrhage	0.08	−0.94 to 1.10	0.001	0.873
Early hydrocephalus	−0.35	−1.30 to 0.60	0.01	0.463

CCL2, C—C motif chemokine ligand-2; CI, confidence interval; CSF, cerebrospinal fluid; IL-6, Interleukin-6; IL-8, Interleukin-8; GCS, Glasgow Coma score; G-CSF, granulocyte colony stimulating factor; VEGF, vascular endothelial growth factor.

^*p < 0.05.

^†n = 57 for IL-8, n = 55 for VEGF.

Univariate analysis of the relationship between clinical and imaging severity factors and CSF cytokine concentrations. Higher early log CSF concentration of G-CSF was associated with higher presenting Hunt and Hess score. There were no additional significant relationships between early CSF cytokine concentrations and severity factors including admission GCS, modified Fisher Scale score, intraventricular hemorrhage, and early hydrocephalus.

Using stepwise backwards linear regression modeling, we found that higher log concentrations of VEGF were associated with a higher HH score at admission when corrected for mFS score and age (β=0.39, 95 % CI 0.05-0.73, p = 0.026). There were no additional associations found between early CSF cytokine concentrations and severity features in multivariable models using this methodology.

Association of early CSF cytokines and shunt dependent hydrocephalus

Of the 55 aSAH patients with CSF samples collected within 72 hours of ictus, 43 patients (78 %) survived to the time of decision making regarding EVD removal or permanent shunt placement and 8 (18.6 %) of these patients developed SDHC. In univariate analyses, higher early log concentration of IL-6 within the CSF of aSAH patients was associated with increased likelihood of SDHC (p = 0.021, Figure 2B). Additionally, there were trends toward an association between higher early CSF log [IL-8] (p = 0.079) and log[G-CSF] (p = 0.063) and SDHC, but neither reached statistical significance. Older age was also significantly associated with SDHC (p = 0.028), but other demographic and clinical variables including sex, HH score, and mFS score were not associated with SDHC (all p > 0.05). In multivariable models, when adjusted for age, sex, and mFS score, early CSF log[IL-6] (OR 7.91, 95 % CI 1.86-88.2, p = 0.030) and log[IL-8] (OR 4.18, 95 % CI 1.33-23.1, p = 0.040) were significantly associated with SDHC (Supplemental Table 5).

Discussion

By using techniques to study inflammation in the CNS after hemorrhagic stroke, we were able to detect cytokines in the CSF of patients with aSAH. In this study, we have shown that the innate proinflammatory cytokines CCL2, IL-6, IL-8, VEGF, and G-CSF have higher concentrations within the CSF as compared to the plasma of aSAH patients in the first 72 hours after hemorrhage. This is consistent with previous literature investigating CSF cytokines after aSAH.^{7, 17, 19, 28} Additionally, in the current study, these proinflammatory cytokines were higher in CSF collected from aSAH patients compared to older patient controls with ventricular sampling of CSF, an important comparator often lacking in previous studies.

Although it was previously thought that the immunologically privileged CNS had no contact with the peripheral immune system due to the presence of the blood-brain barrier, it has now become clear that interactions between the CNS and the immune system are crucial both in states of homeostasis and during times of injury.²⁹ CNS production of cytokines occurs within the CNS by astrocytes, microglia, ependymal cells, and cells within the choroid plexus, and these cell types can play key roles in the central-peripheral immune interface.^20–26 Thus, while it is possible that the measured cytokines are entering the CNS from the periphery when blood spills into the subarachnoid space during aneurysm rupture, the significantly lower concentrations of these cytokines in the plasma despite the acute inflammatory state of critical illness, as well as the known central production of these molecules, indicates that the majority of the CSF cytokines are likely produced within the CNS in response to the initial insult.

Additionally, these cytokines likely play a role in the recruitment of circulating immune cells to the CNS, therefore making it less likely that cells have translocated from the periphery into the CNS within the time of sample collection. To further illustrate this point, CCL2, a monocyte chemoattractant, has previously been shown to be higher in the CSF compared to the plasma of aSAH patients within the first 48 hours after hemorrhage, and has also been reported to be associated with elevated IL-6 and correlated with higher risk of vasospasm.^{7, 19, 30} IL-6, a multifunctional proinflammatory cytokine that stimulates production of acute phase proteins, has previously been shown to be elevated in the CSF compared to plasma in aSAH patients, and in several studies has been shown to be associated with worse outcome, vasospasm, DCI, and increased risk of SDHC.^{14, 16, 18, 31–38} IL-8, a neutrophil chemoattractant, has been less well studied after aSAH, but it has been shown to be elevated in the CSF of aSAH patients in the first week after hemorrhage.^{7, 17} There is limited prior investigation of VEGF and G-CSF in aSAH patients. G-CSF, is primarily produced by monocytes in the periphery but can also be produced by astrocytes within the CNS.²² While G-CSF is classically involved in the proliferation and differentiation of granulocytes, it has also been shown to be neuroprotective and play a role in plasticity in ischemic stroke.^22,39 To our knowledge, G-CSF has not been previously measured in the CSF of patients with aSAH, but in a rat model of intraparenchymal hemorrhage it was found to be neuroprotective and associated with increased VEGF expression after hemorrhage.⁴⁰ VEGF appears to play a dual role in the CNS, as it is involved in blood brain barrier permeability but it also has an anti-inflammatory effect and is involved in vessel repair.^{41, 42} In one prior study which measured VEGF in the plasma and CSF of aSAH patients, the cytokine was found to be elevated in CSF from aSAH patients compared to control CSF and in aSAH patient CSF compared to plasma, but no differences were found between aSAH patient and control plasma.⁷

Consistent with prior literature, we found very high early levels of CCL2 in the CSF of patients with aSAH, but we did not find an association between CCL2 at this early time point and DCI or vasospasm (data not shown). IL-6, the cytokine most widely studied within the CSF of aSAH patients, is also significantly elevated in aSAH patient CSF compared to plasma and compared to control CSF in our study. Although we did not find an association between CSF IL-6 and disease severity, we found a significantly increased likelihood of SDHC in aSAH patients with higher early levels of IL-6 in their CSF in both univariate and multivariable analyses. While it will be important to further investigate persistent elevations in this proinflammatory cytokine in aSAH patients, high early concentrations of IL-6 in the CSF of aSAH patients suggests that some patients may have a more robust inflammatory response to aneurysm rupture and the presence of subarachnoid and intraventricular blood. Additionally, we found high levels of IL-8 early after hemorrhage in CSF collected from aSAH patients compared to plasma levels and compared to control CSF. Although we found no association between CSF IL-8 and disease severity, in a multivariable model adjusted for age, sex, and mFS score we found that higher early CSF IL-8 was associated with increased likelihood of SDHC, a novel finding. As it has been shown in animal models that inhibition of IL-6 can mitigate cerebral vasospasm,⁴³ it is possible that in the future IL-6- or IL-8-targeted immunotherapy may similarly be beneficial in preventing unfavorable outcomes like the development of SDHC in a select population of patients.

VEGF and G-CSF, perhaps the least well understood cytokines we measured and those least well studied in this population, were also both significantly elevated in early CSF samples collected from aSAH patients compared to aSAH patient plasma and control CSF. Interestingly, higher concentrations of early CSF G-CSF and VEGF were significantly associated with higher HH score at admission in univariate and multivariable models, while none of the other measured cytokines were associated with disease severity factors. It is intriguing that early CSF levels of VEGF and G-CSF are associated with disease severity in our study, as these molecules have been previously shown to have proinflammatory, as well as neuroprotective and anti-inflammatory effects. This is perhaps less surprising for VEGF which appears to have a time-dependent role in the CNS after injury.^{20, 22} It has also been shown that VEGF and G-CSF are produced within the CNS by ependymal cells and astrocytes, respectively, in response to acute injury and during the chronic phase of repair after brain injury.^{20, 24} As the role of these cytokines has not previously been explored in aSAH, their acute production within the CNS may represent an early marker of severe disease and a potential risk factor for poor outcome. We are continuing to explore the role of these cytokines in the CSF of patients with aSAH in ongoing studies, including investigating temporal dynamics and associations with neurological outcome. Additionally, the fact that early CSF concentrations of IL-6 and IL-8 are associated with SDHC but not clinical or radiographic severity features, highlights that this relationship is independent of disease severity.

Our study has several limitations and is meant to be hypothesis generating. Firstly, we only analyzed samples collected within the first 3 days after ictus, limiting our investigation to the acute inflammatory response. Future work with longitudinal CSF samples from patients with aSAH will be needed to further understand the temporal changes in these and other inflammatory molecules throughout the disease course. Secondly, we are only able to collect CSF samples from patients with EVDs placed as part of their clinical care, and as such we only have CSF data on more severely affected aSAH patients. Additionally, it should be noted that patients from whom only plasma samples were collected had less severe presentations (higher admission GCS, lower HH score, and less hydrocephalus on initial imaging) than those with EVDs placed (Table 2), which limits comparison between these groups. While this means that our analysis is limited to a subset of aSAH patients, these severely affected patients are at highest risk for poor outcomes so a better understanding their disease course and developing therapeutics for this population is likely to have a substantial impact on their outcome. Our small sample size limits our statistical power but has generated several hypotheses for ongoing investigation into the role of CNS inflammation after hemorrhagic stroke. Thirdly, patients with normal pressure hydrocephalus from whom we collected control ventricular CSF samples were significantly older than the aSAH patients included in this study. Prior studies have shown that aging is typically associated with a pro-inflammatory state with higher levels of pro-inflammatory cytokines and chemokines produced both peripherally and by microglia within the CNS.^44–46 Additionally, the few studies that have investigated cytokines within the CSF of patients with normal pressure hydrocephalus have shown elevated levels of pro-inflammatory cytokines including IL-6 and IL-8 when compared to age-matched controls.^{47, 48} Thus, the presence of higher concentrations of all of the measured cytokines within the CSF of a slightly younger population of aSAH patients indicates that the acute inflammatory state post-hemorrhage is able to overcome even these age-related and hydrocephalus-related differences in CNS inflammation. Finally, our analysis was limited to 11 cytokines and chemokines and our methodology has a relatively high lower limit of detection and thus was not a comprehensive study of all factors that may be associated with brain injury after aSAH and outcomes.

Measurement of inflammatory molecules within the CNS is critically important to characterize the inflammatory cascade that occurs after aSAH and its relationship with the development of SDHC. Identification of CSF cytokines that are highly associated with SDHC can prompt the clinical team to discuss an earlier transition from EVD to permanent shunt in the appropriate clinical context, thus decreasing the infectious risk that occurs with prolonged EVD use. In the future, immunomodulatory medications targeting cytokines associated with SDHC and poor outcome may help mitigate these effects. In addition, since inflammation is involved in both initial injury and resolution and tissue repair, incorporation of the kinetics of inflammation in the CNS following hemorrhage is critical. In this study we have begun to explore the relationship between early inflammatory molecules, disease severity, and SDHC. Future studies on the relationship with clinical outcomes and later time points will be important to further define potential therapeutic strategies.

Supplementary Material

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jstrokecerebrovasdis.2025.108395.