Choroid plexus perfusion in sickle cell disease and moyamoya vasculopathy: Implications for glymphatic flow

Abstract

Cerebrospinal fluid (CSF) and interstitial fluid exchange have been shown to increase following pharmacologically-manipulated increases in cerebral arterial pulsatility, consistent with arterial pulsatility improving CSF circulation along perivascular glymphatic pathways. The choroid plexus (CP) complexes produce CSF, and CP activity may provide a centralized indicator of perivascular flow. We tested the primary hypothesis that elevated cortical cerebral blood volume and flow, present in sickle cell disease (SCD), is associated with fractionally-reduced CP perfusion relative to healthy adults, and the supplementary hypothesis that reduced arterial patency, present in moyamoya vasculopathy, is associated with elevated fractional CP perfusion relative to healthy adults. Participants (n = 75) provided informed consent and were scanned using a 3-Tesla arterial-spin-labeling MRI sequence for CP and cerebral gray matter (GM) perfusion quantification. ANOVA was used to calculate differences in CP-to-GM perfusion ratios between groups, and regression analyses applied to evaluate the dependence of the CP-to-GM perfusion ratio on group after co-varying for age and sex. ANOVA yielded significant (p < 0.001) group differences, with CP-to-GM perfusion ratios increasing between SCD (ratio = 0.93 ± 0.28), healthy (ratio = 1.04 ± 0.32), and moyamoya (ratio = 1.29 ± 0.32) participants, which was also consistent with regression analyses. Findings are consistent with CP perfusion being inversely associated with cortical perfusion.

Keywords

Cerebrospinal fluid choroid plexus glymphatic perfusion sickle cell disease

Introduction

The choroid plexus (CP) complexes consist of a network of modified ependymal cells and are responsible for the production of cerebrospinal fluid (CSF) in each of the four ventricles of the brain. The CP also represents the locus of exchange of blood and CSF at the blood-CSF barrier¹ and hemodynamic activity of the CP has been reported to have relevance to CSF production activity,² circulating markers of neuronal stress,³ and very recently to decrease following surgically-induced angiogenesis of the cortex in humans.⁴ As such, the CP may provide a central indicator of tissue or vascular health, a possibility that is becoming increasingly considered given CSF flow along perivascular spaces as part of the recently proposed glymphatic system.^5–7

The potential existence of the glymphatic system has expanded the description of CSF production and flow to include perivascular pathways for CSF circulation and clearance.⁸ While incompletely characterized,^9–11 central to most glymphatic hypotheses is that in addition to the bulk CSF flow pathway whereby CSF is cleared through arachnoid granulations, CSF influx additionally occurs along perivascular spaces (Figure 1).⁵ Aquaporin-4 (AQP4) channels mediate the flow of CSF from the arterial perivascular space to interstitial space. Convection currents from astrocytes allow for net fluid flow through interstitial space, and fluid exits into venous perivascular space via AQP4 channels. While the mechanism of fluid efflux from the venous perivascular space largely remains to be elucidated, net fluid flow may reach the cervical lymph nodes through lymphatic channels.¹² Central to this hypothesis is the dependence of CSF flow, along perivascular pathways, on the compliance, pulsatility, and patency of cerebrovasculature.

Figure 1.

The bulk cerebrospinal fluid (CSF) flow pathways consists of CSF production in the choroid plexus (CP) complexes; flow through the cerebral aqueduct at a peak flow velocity of 5-15 mm/s; outflow from the ventricular system via the foramina of Lushka and Magendie; and ultimately circulation along the arachnoid space overlying the cerebrum with uptake into the venous system via arachnoid granulations. Orthogonal slices of the (a) CP at the level of the atria of the lateral ventricles, (b) cerebral aqueduct, and (c) an arachnoid granulation are shown (green crosshairs depict structure). It has been proposed that CSF influx also occurs along periarterial spaces, and CSF efflux along perivenous spaces, with fluid moving between these spaces and the interstitial space through aquaporin-4 mediated channels. (d) Magnetic resonance angiography and venography of major arteries and veins which can vary in conspicuity and patency with degree of steno-occlusion and hemoglobin-induced hyperemic changes; these changes extend to smaller vessels and perivascular pathways. (e) In addition to anatomical visualization of these structures, hemodynamic activity of the CP can be evaluated non-invasively in vivo using arterial spin labeling perfusion-weighted MRI, shown here alongside anatomical T₁-weighted and T₂-weighted FLuid Attenuated Inversion Recovery (FLAIR) MRI. The CP is demarcated by orange arrows.

In support of this hypothesis, using in vivo two photon imaging, Iliff et al.¹³ reported in both ketamine/xylazine and isofluorane anesthetized mice that decreasing arterial pulsations via ligation of the internal carotid artery, and increasing arterial pulsations via administration of an adrenergic agonist, altered exchange of CSF and interstitial fluid (ISF). This provides evidence that pulsatility and compliance of the microvasculature plays a central role in CSF-ISF exchange, and directly relates to periarterial flow of CSF. However, less is known regarding the potential role of CSF-ISF exchange and arterial pulsatility in humans, largely due to safety concerns related to performing these assays in humans.

An alternative possibility to investigate analogous mechanisms in humans is to quantify CSF properties in vivo in well-characterized patients that possess different degrees of cerebral blood volume or arterial patency secondary to underlying pathophysiology. For instance, individuals with sickle cell disease (SCD) have chronic anemia, and in the 80–85% of these patients without associated vasculopathy visible on angiography,^14–17 to compensate for reduced blood oxygen content, these individuals have elevated arterial perfusion and blood volume, which is approximately 1.5–2.5 times that of healthy non-anemic adults, depending on extent of anemia.^18–21 Additionally, individuals with non-atherosclerotic moyamoya vasculopathy, which can affect individuals over a similar age span as SCD of childhood and early adulthood, have reduced arterial patency and cerebral blood volume in a manner that depends on the extent of arterial steno-occlusion and underlying collateralization.^22–26 As such, these two populations in many ways mimic the arterial pulsatility differences that have been investigated through pharmacological or ligation studies in animals, though also provide an opportunity to investigate these mechanisms in awake humans under physiological conditions. As these conditions are prominent in childhood through early adulthood, prior to the development of most common conventional stroke risk factors (i.e., atherosclerosis, type-2 diabetes, hypercholesterolemia, cardiovascular disease, etc.), fewer confounds between populations are present.

Since arterial compliance has been shown to be an important driver of perivascular fluid flux in animals, it is reasonable that the CP, which is the very tissue responsible for CSF production and downstream flow, responds to changes in CSF dynamics attributable to arterial compliance, cerebral blood volume, and perivascular flow. As such, changes in CP metabolic demand and perfusion may closely relate to altered cerebral blood volume in other brain regions. Here, we apply non-invasive arterial spin labeling (ASL) MRI measures of CP and cortical perfusion in sequence with anatomical imaging in adult healthy and SCD participants to test the primary hypothesis that CP perfusion, relative to total gray matter perfusion, is reduced in SCD relatively to health control volunteers. We also explore a secondary hypothesis that CP perfusion is elevated in adults with moyamoya vasculopathy and reduced arterial patency and cortical perfusion. These hypotheses are motivated by earlier animal studies of manipulated arterial pulsatility¹³ and recent work applying ASL to measure CP blood arrival²⁷ and perfusion.^4,27 If confirmed, findings would provide additional support for glymphatic flow in humans, and specifically that CP activity is related to cortical perfusion and the compliance of perivascular pathways.

Methods

Demographics and study design

All participants (n = 75) provided informed, written consent. All components of this study were performed in compliance with the Declaration of Helsinki of 1975 (as revised in 1983), Health Insurance Portability and Accountability Act, and all protocols were approved by the Vanderbilt University Medical Center Institutional Review Board.

To test the primary study hypotheses, (i) healthy adults with no prior neurological condition or evidence of neurological pathology on anatomical MRI (n = 25) and (ii) age- and sex-matched adults with SCD without corresponding vasculopathy as confirmed by angiography (n = 25) were enrolled. As a secondary sub-study, we also evaluated an identical number (n = 25) of adults with idiopathic intracranial moyamoya vasculopathy without anemia. Moyamoya and SCD participants were recruited from the neurosurgery and hematology services, respectively, at Vanderbilt University Medical Center between 2012 and 2019. This 7-year recruitment duration allowed for enrollment of sufficient moyamoya participants meeting enrollment criteria, as moyamoya is a relatively rare disorder (estimated incidence <1/100,000²⁸) Both SCD and moyamoya vasculopathy are present in young adults, and over the first three decades of adulthood the gray matter perfusion has been measured to change minimally by 2.4 ml/100g/min (or approximately 4.1%²⁹) Additionally, it is well-known that moyamoya affects females disproportionately.^22,23 Given these additional confounds, the study focuses primarily on SCD and healthy control comparisons; the moyamoya cohort is included as a secondary analysis and should be interpreted within the context of these additional considerations of age and sex.

Pediatric participants were excluded to reduce bias from hemodynamics changing more substantially with development. Additional inclusion and exclusion criteria for each cohort were: Healthy adults. No history of neurological disease including but not limited to Parkinson disease, cerebrovascular disease, multiple sclerosis, or traumatic brain injury; homozygous hemoglobin-AA; and pre-menopausal (if female). SCD. Homozygous Hb-SS or Hb-Sβ₀ thalassemia phenotype confirmed by high performance liquid chromatography; absence of vasculopathy confirmed by magnetic resonance angiography; and compliant on hydroxyurea or blood transfusion. In transfusion participants imaging was performed late (3 weeks) in the transfusion cycle when hematocrit was near nadir. Idiopathic moyamoya disease. Clinical diagnosis of idiopathic moyamoya from catheter angiography and absence of secondary moyamoya syndrome (i.e., SCD, atherosclerosis, neurofibromatosis, or Down’s syndrome). Prior surgical revascularization was not an exclusion criterion.

Image and angiography acquisition

All participants were scanned at 3 Tesla (Philips, Best, The Netherlands) using body coil radiofrequency transmission and phased-array SENSitivity Encoding (SENSE) reception.

To characterize prior infarct and vasculopathy extent, a standard non-contrast anatomical head and neck magnetic resonance imaging (MRI) and angiography (MRA) protocol was performed including: T₁-weighted imaging (magnetization-prepared rapid gradient echo; spatial resolution = 1.0 × 1.0 × 1.0 mm³; 3 D turbo-field-echo; repetition time/echo time = 8.2/3.7 ms); T₂-weighted imaging (spatial resolution = 0.6 × 0.6 × 4.0 mm³; turbo spin echo; repetition time/echo time = 3000/80 ms); T₂-weighted fluid attenuated inversion recovery (FLAIR) (spatial resolution = 0.9 × 1.1 × 3.0 mm³; turbo inversion recovery; repetition time/inversion time/echo time = 11,000/2800/120 ms); and intracranial time-of-flight magnetic resonance angiography (spatial resolution = 0.6 ×0.6 × 1.4 mm³; 3 D gradient echo; repetition time/echo time = 23/3.5 ms).

CP and cerebral gray matter perfusion were quantified using ASL MRI. To ensure that CP perfusion was reproducible, and to quantify its dependence on ASL sequence parameters, a subset of healthy controls (n = 10) were scanned using the pseudo-continuous ASL (pCASL) sequence with common spatial resolution 3 × 3 × 7 mm for Protocol-1 (post-labeling delay = 2000 ms, label duration = 1650 ms, averages = 20, duration = 160 s) and separately for Protocol-2 (post-labeling delay = 1900 ms, label duration = 1000 ms, averages = 20, duration = 160 s), with each scan repeated once for repeatability assessment. The shorter labeling duration of 1000 ms for Protocol-2 was used here as the pCASL scan was also part of a broader multi-post-labeling delay protocol to quantify blood arrival time in SCD, and the shorter label duration was required to sensitize the sequence to blood arrival at short post-labeling delays. Note that the post-labeling delay times utilized here are much longer, as required, than the arterial transit time of blood to the CP, recently measured to be 1.24 ± 0.20 s.²⁷ This comparative analysis was performed as the pCASL protocol applied to moyamoya (Protocol-1) and SCD (Protocol-2) differed owing to known reductions in bolus arrival time in SCD compared to moyamoya and additional transit time considerations noted above.^22,30 The difference in parameter choice was also motivated by previous studies demonstrating that blood arrival times are reduced in healthy³¹ and SCD³⁰ adults relative to adults with moyamoya,^22,32 and the desire to titrate the sequence appropriately to each population. Any potential differences in quantified perfusion values between these two similar sequences were evaluated in control comparisons where both methods were repeated in healthy adults.

Image and angiography analysis

Cervical and major intracranial vessels for each participant were assessed for vasculopathy by a board-certified neuroradiologist as previously described.³³ Each vessel was graded as normal, mild stenosis (25–50%), moderate stenosis (51–69%), severe stenosis (70–99%), or occlusion. Severity of intracranial vasculopathy was graded by the worst vessel seen, as mild, moderate, severe, or occluded. For infarcts, the scan was judged as normal or abnormal. If abnormal, the lesions were assigned as either (i) non-specific white matter hyperintensity if <3 mm in diameter;³⁴ or (ii) focal, discrete ischemic infarcts if ≥3 mm on FLAIR MRI. The above criteria were also applied to control participant datasets to confirm the absence of flow-limiting vasculopathy and infarct.

pCASL data were corrected for motion and baseline drift, normalized by the equilibrium magnetization M₀, and a model that utilized the measured hematocrit to calculate the blood T₁, as well as differing labeling efficiencies (0.72 in SCD vs. 0.85 in healthy and moyamoya disease), was applied.³⁰ Here, the recently-recommended assumption³⁵ that tissue T₁ = blood T₁ was not used, and rather we calculated the participant-specific blood T₁ using the measured hematocrit and previously published relationship between arterial (oxygenation = 92+/-7%) blood water T₁ and hematocrit: 1/T₁ = 0.52 × Hct + 0.38.³⁶ A common 3 T tissue T₁ = 1.2 s was used in all participants.³⁷ T₁-weighted images were segmented into gray matter (GM), white matter, and CSF using FSL-FAST;³⁸ CP was manually segmented from the atria of the lateral ventricles from the T₁-weighted images. While CP is present in all brain ventricles, we focused on the lateral ventricles as it is not possible to reliably segment CP in other ventricles given the spatial resolution of the pCASL method. Perfusion maps were transformed to the native T₁-weighted space using the M₀ image for co-registration and GM perfusion, and the CP perfusion values were recorded. As absolute perfusion will vary with the extent of anemia, the measurement preserved for primary hypothesis testing was the ratio of the CP-to-GM perfusion. It should be noted that in this analysis, the ASL kinetic model used becomes irrelevant as the outcome is a unitless ratio.

Statistical analysis

To understand any potential differences between the similar pCASL protocols, as well as to assess repeatability of CP perfusion, the ratios of CP-to-GM perfusion were compared using a paired Student’s t-test in healthy adults with repeated measures, and data presented as Bland-Altman and scatter plots with intraclass correlation coefficients (ICC). A Shapiro-Wilk test was applied, separately to CP-to-GM perfusion ratio values for each cohort, to evaluate normality under the null hypothesis that data were normally distributed. Standard descriptive statistics including mean, median, and inter-quartile ranges were calculated. Differences in age between groups were calculated using a Student’s t-test and differences in sex were calculated using a χ² test. To understand any potential difference in CP perfusion with (i) age or (ii) biological sex in the absence of pathology in all healthy controls (n = 25), linear regression was performed using CP-to-GM perfusion ratio as the dependent variable and age and sex as independent variables. Regression coefficients and associated p-values were calculated.

Next, to evaluate whether CP-to-GM perfusion ratios differed between groups, we performed a single one-way ANOVA between all three groups. F-statistic, Critical-F, and two-sided p-values were calculated.

As age and sex may also contribute, to test the primary hypothesis that CP-to-GM perfusion ratio is reduced in SCD relative to healthy control participants, multiple linear regression was performed on the 50 data points for these two groups according to:

\frac{{Perfusion}_{choroid plexus}}{{Perfusion}_{gray matter}} = β_{0} + β_{Age} \cdot Age + β_{Group} \cdot Group + β_{Sex} \cdot Sex

(1)

where β₀ is the model intercept, Age is the continuous age of the participant in years at the time of imaging, Group is the dichotomous group status (0 = control, 1 = SCD), and Sex is the dichotomous biological sex (0 = female; 1 = male). Finally, as a secondary analysis and to gain additional preliminary information on whether the CP-to-GM perfusion ratio differed between control and moyamoya vasculopathy participants, the above analysis was repeated for these two groups.

In all cases, two-sided p < 0.05 was required for significance (Bonferroni-corrected critical two-sided p-value for two case-control comparisons for regression analyses: p < 0.025).

Data availability

De-identified data and data processing scripts will be made available to others with Human Subjects Collaborative Institutional Training Initiative training for purposes of reproducing the study findings.

Results

Table 1 summarizes the demographic and clinical history of the healthy (n = 25; age = 26.1 ± 5.5 years; 48% male), SCD (n = 25; age = 30.2 ± 3.6 years; 60% male), and moyamoya (n = 25; age = 42.9 ± 12.4 years; sex = 16% male) participants. Consistent with adult moyamoya affecting slightly older adults and females disproportionately, age and percent-female were both greater in moyamoya participants (p < 0.001) compared to the healthy and SCD cohorts.

Table 1.

Descriptive statistics of perfusion values for the three cohorts.

	Healthy	Sickle cell disease	Moyamoya
Count	25	25	25
Age (mean±SD), years	26.1 ± 5.5	30.2 ± 3.6	42.9 ± 12.4
Sex, fraction male	0.48	0.60	0.16
CP perfusion (mean±SD), ml/100g/min	50.4 ± 14.1	75.7 ± 27.4	52.9 ± 13.7
CP perfusion (median, IQR), ml/100g/min	47.1, 20.3	73.8, 38.8	52.4, 16.1
GM perfusion (mean±SD), ml/100g/min	50.2 ± 11.0	82.1 ± 17.9	42.4 ± 13.0
GM perfusion (median, IQR), ml/100g/min	47.9, 15.9	83.9, 27.2	41.0, 15.9
CP-to-GM perfusion ratio (mean±SD)	1.04 ± 0.32	0.93 ± 0.28	1.29 ± 0.32

All healthy participants were homozygous Hb-AA, with no evidence of vascular or neurological disease. All moyamoya participants had idiopathic moyamoya with evidence of arterial steno-occlusion and moyamoya leptomeningeal collaterals on catheter angiography. Sickle cell disease participants had no evidence of vasculopathy, hematocrit = 26 +/− 6%, and homozygous Hb-SS or Hb-Sβ0. SD = standard deviation, CP = choroid plexus, IQR = interquartile range, GM = gray matter.

Figure 2 shows representative time-of-flight angiograms, highlighting the variations in arterial vascularity and macrovascular patency between cohorts. Participants with moyamoya vasculopathy have reduced arterial patency of the primarily anterior circulation, whereas SCD participants have elevated arterial cerebral blood volume and perfusion to compensate for reduced oxygen delivery from anemia and hemoglobinopathy. No included SCD or healthy control participants had evidence of vasculopathy on angiography and all moyamoya participants met criteria for flow limiting vasculopathy of either the intracranial segment of the ICA or first segment of the MCA. All moyamoya participants had evidence of prior infarcts on anatomical imaging, 9/25 (36%) of SCD participants had evidence of chronic infarcts, and no controls had evidence of infarcts.

Figure 2.

Time-of-flight angiogram from a healthy, moyamoya, and sickle cell disease participant. Participants with moyamoya vasculopathy have reduced bilateral arterial patency, whereas sickle cell disease participants have vasodilation and elevated cerebral blood volume and flow to compensate for reduced blood oxygen content from anemia. Coronal (above) and axial (below) maximum intensity projections are shown for each participant.

Figure 3 shows representative anatomical images and perfusion maps, as well as the segmentation procedure. In the preliminary protocol comparison in healthy adults, CP perfusion values between scans for both pCASL Protocol-1 and Protocol-2 were found not to be significantly different and reproducible: Protocol-1 scan-1 = 44.9 ± 8.1 ml/100g/min and scan-2 = 44.9 ± 9.9 ml/100g/min; p = 0.98; Protocol-2 scan-1 = 47.2 ± 11.9 ml/100g/min and scan-2 = 43.3 ± 11.0 ml/100g/min; p = 0.12 (Figure 4). The ICC for CP perfusion and gray matter perfusion were both 0.69, consistent with moderate-to-good agreement in both regions. Across all control volunteers (n = 25), the CP-to-GM perfusion ratio (mean ± std) was 1.04 ± 0.32 (median = 1.03; 95% confidence level = 0.13). On regression analysis, the ratio did not differ with biological sex (β = −0.06; p = 0.608), but did increase weakly but significantly with age (β = 0.03; p = 0.022). The findings from this preliminary control analysis are consistent with (i) the CP-to-GM perfusion ratios not differing significantly between the two similar ASL protocols, (ii) the measurements being repeatable within scan sessions at an ICC of 0.69, (iii) the measurements not differing with biological sex, and (iv) the measurements increasing significantly with age.

Figure 3.

(a) A representative T₁-weighted image and (b) magnification from a control obtained at the level of the atria of the lateral ventricles. (c–d) Perfusion maps from sequential scans showing reproducible signal (group ICC = 0.69) in the choroid plexus (CP) within the atria of the lateral ventricles (red arrows), consistent with CP perfusion. Example (e) manually-segmented CP and (f) automated gray matter segmentation is shown for perfusion maps in absolute units of ml/100g/min.

Figure 4.

In this study the primary hypothesis of comparing sickle cell disease and age- and sex-matched controls was performed using an identical Protocol-2, which utilized a post-labeling delay of 1900 ms. The supplementary hypothesis that evaluated perfusion values from moyamoya participants utilized Protocol-1, which had a slightly longer post-labeling delay of 2000 ms. Additional sequence details are summarized in the Methods. Choroid plexus (CP) perfusion values between scans for both protocols were found not to be significantly different and to have moderate-to-good intraclass correlation coefficient of 0.69. (a) Gray matter (GM) perfusion, (b) CP perfusion, (c) CP-to-GM perfusion ratio, and (d) Bland-Altman plots in 10 controls each scanned consecutively with each protocol. Mean values are shown in (e–f).

Figure 5 shows perfusion maps for a representative healthy, SCD, and moyamoya participant. The SCD participant has globally elevated perfusion in gray matter, consistent with known upregulation of cerebral perfusion to offset reduced blood oxygen content from anemia. The moyamoya participant has bilaterally reduced perfusion in gray matter, consistent with downstream effects of flow-limiting arterial steno-occlusion. However, in the SCD participant the CP perfusion is reduced relative to the gray matter perfusion, whereas in moyamoya the CP perfusion is elevated relative to the gray matter perfusion. In the healthy volunteer, the CP and gray matter perfusion are approximately equal.

Figure 5.

Choroid plexus (CP) perfusion for a representative (a) healthy, (b) moyamoya, and (c) sickle cell disease participant. T₁-weighted anatomical images at the level of the lateral ventricles and corresponding perfusion maps are shown above, with magnification below. Red arrows demarcate CP location on the anatomical scan and white arrows on the perfusion scan. In the healthy participant, the CP perfusion is approximately equal to the gray matter perfusion. In the moyamoya participant, the CP perfusion in the same region is markedly increased relative to the gray matter perfusion. In the sickle cell disease participant, the gray matter perfusion is increased secondary to anemia, as expected, and the CP perfusion is reduced relative to the gray matter perfusion. Group results are summarized in Figure 6.

Results of the Shapiro-Wilk test revealed no evidence to reject the null hypothesis that the CP-to-GM perfusion ratios were normally distributed for control (p = 0.808), SCD (p = 0.874), or moyamoya groups (p = 0.332). One-way ANOVA revealed a significant difference in mean CP-to-GM perfusion ratios between the three groups (p < 0.001; F-statistic = 9.22; Critical-F = 3.12), which was investigated in more detail using regression analyses. On a group level multiple regression analysis, for SCD participants a significant relationship between the CP-to-GM perfusion ratio and age (β = 0.022; p = 0.018) and group (β = −0.099; p = 0.034) were observed, but no significant relationship for biological sex (β = 0.025; p = 0.764) was observed. For moyamoya participants, no statistically significant relationship between the CP-to-GM perfusion ratio and group (β = 0.211; p = 0.088), age (β = 0.006; p = 0.227), or sex (β = 0.155; p = 0.137) was observed, however the p-value of 0.088 with group indicated a possible trend. The SCD vs. control finding (primary study hypothesis) was also significant at the Bonferroni-corrected crucial p-value of p = 0.025. Figure 6 summarizes the group-level findings and data distribution.

Figure 6.

Boxplots of the choroid plexus (CP) to gray matter (GM) perfusion ratio (CP-to-GM) in each of the groups indicates that on average SCD participants have significantly reduced perfusion ratios compared to moyamoya participants and controls. All groups were significantly different on one-way ANOVA analysis (p < 0.001). On regression analyses which included age and sex as co-variates, SCD participants were different than controls and moyamoya participants, and moyamoya participants had increased perfusion ratios relative to controls, but did not meet criteria for statistical significance (p = 0.088). * two-sided p < 0.05. ** two-sided p < 0.001. P-values reflect results from regression analysis.

Discussion

We performed measurements of CP and gray matter perfusion in healthy adults and participants with non-atherosclerotic cerebrovascular disease secondary to (i) SCD without arterial steno-occlusion or (ii) moyamoya intracranial arterial steno-occlusion. ANOVA analysis yielded significantly different mean CP-to-GM perfusion ratios between the three groups. The primary hypothesis was confirmed, in that findings supported CP perfusion being reduced, relative to gray matter perfusion, in SCD participants compared to healthy adults. In moyamoya participants, multiple regression analysis which co-varied for age and sex provided evidence for an elevated CP-to-GM perfusion ratio, however this did not meet rigorous criteria for statistical significance (p-value = 0.088).

These findings expand recent animal and human work that have related arterial pulsatility and patency to CSF activity. First, Iliff et al.¹³ reported in an anesthetized murine model that decreasing arterial pulsations via ligation of the internal carotid artery, reduced exchange of CSF and ISF, whereby increasing arterial pulsatility via pharmacological manipulation with dobutamine increased exchange. These findings have the implication that reduced arterial pulsatility, ubiquitous in arterial steno-occlusive cerebrovascular disease and aging, may contribute to retention of toxic solutes within the brain. Recently, similar work has been expanded to humans, whereby it was shown that indirect surgical revascularization can lead to reductions in CP perfusion in adults with moyamoya vasculopathy, which would be consistent with a reduced need to force high levels of CSF along poorly compliant periarterial pathways.⁴ Recently, it has also been shown in a murine model that the pulsatility index of cortical vessels decreased bilaterally following craniectomy, which paralleled reduced periarterial perivascular influx.³⁹

Confounders of these prior studies are (i) how anesthesia and intracranial changes secondary to invasive procedures influence CSF production and flow in animals, and (ii) that CP perfusion has been shown to increase with increased CSF production activity, but it also may respond to independent circulating markers of glial stress, or even in a purely vascular manner if the arterial supply to the CP itself is abnormal. Findings from this study help to disambiguate these uncertainties.

First, by utilizing non-invasive approaches in humans without anesthesia, our study design obviates concerns related to possible effects of anesthesia and the influence of invasive surgical procedures on CSF flow and function. We utilized a pCASL method applied to the CP and gray matter. A similar protocol has recently been applied in healthy adults and has demonstrated that arterial arrival to the CP is 1.24 ± 0.2 s, T₁ of the CP is 2.33 ± 0.30 s, and apparent CP perfusion is 39.5 ml/100 g/min.²⁷ Our perfusion values from this study and our previous study⁴ are in good agreement with this perfusion value. It should be noted that arrival time and T₁ of the CP is longer than for cortex. In this study we used long post-labeling delays which desensitize the pCASL sequence to variations in blood arrival over this range, and also use the CP-to-GM perfusion ratio, which reduces bias from kinetic model choice. It should also be noted that as the CP is a relatively small structure, partial volume effects with CSF or periventricular tissue may also affect measurements and lead to an underestimation of true CP perfusion.²⁷ To reduce this concern we segmented the CP manually on T₁-weighted images, however, given the inherent lower spatial resolution of pCASL some partial volume contributions will persist, but are expected to be similar in all participants. The extent of atrophy, and associated size of the lateral ventricles, may influence the ability to accurately segment the CP and as such the measured CP perfusion. To reduce partial volume confounds between participants from segmentation we have chosen to only focus on the CP that is most conspicuous, in the atria of the lateral ventricles. Additionally, the participants we enrolled, young adults before the age of 50 years, have less atrophy compared to older adults where more variability between participants may exist.

Second, our findings suggest that a purely vascular explanation of the CP findings is unlikely. The CP receives its arterial supply from the anterior choroidal artery as well as the medial and lateral posterior choroidal arteries, which arise from the distal internal cerebral artery (ICA), and the first or second segments of the posterior cerebral artery (PCA), respectively.⁴⁰ As arterial steno-occlusion was an exclusion criterion for both SCD and control participants, who exhibited significantly different CP perfusion findings, and the CP has a redundant arterial supply partly arising from the posterior circulation, which is frequently preserved in moyamoya vasculopathy participants, it is unlikely that the findings could be explained by simple variations in arterial supply to the CP.

In the absence of a purely vascular explanation, or an explanation secondary to anesthetics or invasive procedures, the changes in CP perfusion could reflect a difference in CSF production or a difference in the response of the CP to circulating markers of glial stress. The CP is known to secrete various growth factors that are likely involved in maintaining the health of the brain parenchyma.³ Specifically in terms of cerebral ischemia, transforming growth factor (TGF)-β, a well-known regulatory protein involved in tissue homeostasis and immune regulation, is expressed in the CP and has been reported to be elevated during cerebral ischemia in rats,^41,42 and it has been argued that regulation of TGF-β in the CP may have relevance to ensuring calcium homeostasis and the expression of Bcl2 proto-oncogene, a regulatory protein central to cell death, even when the ischemia is spatially removed from the CP.^43,44 It is logical that CP activity could be modulated directly as a result of changes in ischemia and parenchymal need for increased or decreased growth factor expression. However, both SCD and moyamoya participants have varying levels of ischemic injury, with the mechanism of ischemia being due to arterial steno-occlusion in moyamoya versus anemia and reduced blood oxygen content in SCD. All moyamoya participants had evidence of infarcts on neuroimaging and 37% of adult SCD participants had neuroimaging evidence of infarcts. Given the older SCD cohort used here, where ischemic brain injury is more common,^45,46 it is unlikely that the different findings observed between cohorts could be explained by differences in circulating markers of ischemic stress alone, as both cohorts have varying evidence of ischemia.

A final possibility is that the changes in CP perfusion are attributable to changes in CSF production activity. Importantly, CP activity has been shown to be associated with CSF production levels, however not ubiquitously. In support of a direct relationship, when CSF production was evaluated in chloralose-anesthetized rabbits before and after administration of angiotensin II (100 ng/kg/min), a potent vasoconstrictor, it was observed that CSF production reduced by 24 +/– 3%, consistent with CP perfusion being related to CSF production.² A relationship between CP perfusion and CSF production was also observed in a separate study of anesthetized rabbits, in which intravenous infusion of vasopressin decreased blood flow to the CP by 50–60% for the entire 90 period of infusion and also decreased CSF production by 35 +/– 8%.⁴⁷ However, when using separate vasoactive stimuli, other studies have found an apparent uncoupling between CP perfusion and CSF production activity. In anesthetized rabbits that were administered atriopeptin, a protein hormone secreted by cardiac cells and potent vasodilator, this increased blood flow to the CP, yet no significant or only a small effect on CSF production was observed.⁴⁸ Using laser-Doppler flowmetry during ventriculocisternal perfusion with inulin-[14C]carboxylic acid, CP perfusion and CSF production were measured simultaneously in rats during intraventricular administration of vasoactive intestinal polypeptide, which resulted in a decrease in CSF production of up to 30%, yet a CP perfusion increase by 20%.⁴⁹ Lastly, in anesthetized rabbits, it was reported that administration of acetazolamide, a carbonic anhydrase inhibitor and potent vasodilator, yielded an approximately two-fold increase in CP perfusion but a large 55 +/– 5% decrease in CSF production.⁵⁰ These findings collectively highlight that changes in CP perfusion may or may not be directly related to CSF production activity and it is likely that apparent contradictions in the above animal work can be explained by differences in species, invasive methods, and pharmacological agents utilized.

In light of the above possibilities, given that CP perfusion can indicate varying levels of CSF production activity, it is reasonable that the change in CP perfusion, and therefore hemodynamic activity, between groups is related to CSF production differences and the requirement to push a higher volume of CSF along poorly compliant perivascular pathways. This explanation is also consistent with the above murine and revascularization studies in humans demonstrating changes in CSF-ISF exchange or CP perfusion for different levels of arterial pulsatility or arterial patency and cerebral blood volume, respectively. Furthermore, when interpreted in the context of the glymphatic circulation, it is logical that the CP would respond to settings of decreased cerebral blood volume and sluggish perivascular flow with increased perfusion and CSF production in order to mitigate compromised perivascular fluid flux and metabolic waste clearance.

Three aspects of the study should also be considered when interpreting the results.

First, while moyamoya vasculopathy is frequently asymmetric, it is also frequently bilateral. This complicates the use of a common reference region for gray matter perfusion estimation, as distal segments of the internal carotid arteries, and proximal middle and anterior cerebral arteries, are frequently stenosed. The posterior circulation is less commonly affected; however, we have several participants in more advanced stages that also have posterior cerebral artery involvement. As such, there is not a common, unaffected reference region that can be used in all participants. We used a global gray matter perfusion estimate here to avoid selecting individual regions with varying volumes in each participant. It was also noted that there was no significant difference in the gray matter perfusion between controls and moyamoya vasculopathy participants (p = 0.467) after controlling for age. Additionally, blood arrival times in moyamoya participants may be beyond the post-labeling delay used, as arterial circulation times can be greater than 2 s in flow territories perfused by collateral pathways in moyamoya.^22,32 This can lead to an underestimation of gray matter perfusion, or an over-estimation when high endovascular signal is present from collaterals.⁵¹ The large, total gray matter region used for gray matter perfusion quantification will reduce sensitivity to localized regions with delayed blood arrival on average. Additionally, to reduce this confound we incorporated a long labeling duration of 1.65 s and post-labeling delay of 2 s, however it is likely that in some flow territories this may still not extend beyond the blood arrival time.

Second, while the ICC was similar for both gray matter and choroid plexus perfusion, the ICC measured here was 0.69 across 10 healthy adults. Other studies in larger samples of 20 adult healthy volunteers have reported ICCs of 0.89–0.97 and which varied by flow territory,⁵² yet lower ICCs of 0.69–0.77 have been observed in other studies from 22 healthy participants.⁵³ In a similar size cohort as ours of 12 participant scanned over approximately one year, the ICC was found to be 0.63–0.74.⁵⁴ The ICC observed in this study is on the lower-end of reported ICC ranges, but is approximately within range of many other studies; differences may be attributable to number of participants included, regions analyzed (flow territories vs. total gray matter), and sequence parameters.

Third, we considered two unique cohorts: SCD and moyamoya. The primary hypothesis focused on comparing age- and sex-matched SCD and healthy control participants, and the moyamoya cohort was added as a secondary cohort but was not part of primary hypothesis testing. These three cohorts were carefully selected and we believe are appropriate given the following considerations. While we restricted our cohort to adult participants, and we include data over the range of the adult SCD and moyamoya lifespan,^28,55 the moyamoya mean age (42.9 years) was slightly higher than the SCD mean age (30.2 years). Over this age range, typical stroke risk factors (i.e., atherosclerosis, type-2 diabetes, hypercholesterolemia, and cardiovascular disease) are uncommon or have limited hemodynamic impact compared with older age, and it has been shown that brain perfusion reduces by only 0.8 ml/100g/min per decade of life over this age range.²⁹ Therefore, the large perfusion variations observed should not be attributable to age effects alone over this small range, and in all comparisons we included both age and sex as covariates. It is also unlikely that medication differences can account for these findings, as SCD participants were compliant on hydroxyurea, a myelosuppressive agent that increases hemoglobin-F with limited hemodynamic or metabolic effects beyond those related to the desired increase in hemoglobin levels, and moyamoya participants are not on major hemodynamic-modifying medications owing to the non-atherosclerotic nature of this disease.

Conclusion

CP perfusion values measured from pCASL are reproducible and similar to gray matter perfusion values, when acquired at typical ASL spatial resolutions, in healthy adults without neurodegenerative or vascular disease. In sickle cell disease participants with patent vessels and increased cerebral blood volume and flow, CP-to-GM perfusion ratios are reduced relative to ratios in age- and sex-matched healthy adults. This may indicate that there is a decreased demand for CSF production in the setting of perivascular flow associated with high arterial flow states.

Footnotes

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by NIH/NIA (1R01 AG062574), NIH/NINS (5R01 NS097763), NIH/NINDS (5R01 NS096127), American Heart Association.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Manus J Donahue receives research related support as Principal Investigator from Philips Healthcare, the Lipedema Foundation, the American Heart Association, and the National Institutes of Health (NIA, NINDS, NINR, and NCCIH); is a paid consultant for Global Blood Therapeutics and Pfizer, Inc; and is a paid member of the advisory board for bluebird bio and Novartis. He is the CEO of biosight, LLC which provides healthcare technology consulting services to academic and industry clients. These agreements have been approved by Vanderbilt University Medical Center in accordance with its conflict of interest policy. Lori C Jordan receives research funding from the National Institutes of Health (NINDS, NIDDK and NHLBI) and has served as a consultant to bluebird bio. Rohan Chitale receives research grants from Medtronic, Microvention, and Cerenovus. He serves as a consultant for Medtronic.

Authors’ contributions

SEJ designed experiments, acquired data, performed data analysis, and drafted the manuscript. CDM designed experiment, graded imaging findings, and critically reviewed the manuscript. LCJ acquired data, interpreted imaging and neurological findings, and critically reviewed the manuscript. DOC designed study aspects, evaluated neurological findings, and critically reviewed the manuscript. SW acquired data, analyzed data, and critically reviewed the manuscript. CL recruited participants, analyzed data, and critically reviewed the manuscript. MG recruited participants, analyzed data, and critically reviewed the manuscript. NJP recruited participants, analyzed data, and critically reviewed the manuscript. LTD graded imaging findings, analyzed data, and critically reviewed the manuscript. SP graded imaging findings, analyzed data, and critically reviewed the manuscript. PT analyzed data and critically reviewed the manuscript. RC assisted with moyamoya participant recruitment and critically reviewed the manuscript. MF assisted with moyamoya participant recruitment and critically reviewed the manuscript. MJD designed experiments, acquired data, performed data analysis, and drafted the manuscript.

References

Sakka

Coll

Chazal

Anatomy and physiology of cerebrospinal fluid. Eur Ann Otorhinolaryngol Head Neck Dis 2011; 128: 309–316.

Maktabi

Stachovic

Faraci

FM.

Angiotensin II decreases the rate of production of cerebrospinal fluid. Brain Res 1993; 606: 44–49.

Emerich

Vasconcellos

Elliott

, et al. The choroid plexus: function, pathology and therapeutic potential of its transplantation. Expert Opin Biol Ther 2004; 4: 1191–1201.

Johnson

McKnight

Lants

, et al. Choroid plexus perfusion and intracranial cerebrospinal fluid changes after angiogenesis. J Cereb Blood Flow Metab 2020; 40: 1658–1671.

Iliff

Wang

Liao

, et al. A paravascular pathway facilitates CSF flow through the brain parenchyma and the clearance of interstitial solutes, including amyloid beta. Sci Transl Med 2012; 4: 147ra111.

Kylkilahti

Berends

Ramos

, et al. Achieving brain clearance and preventing neurodegenerative diseases – a glymphatic perspective. J Cereb Blood Flow Metab. Epub ahead of print 18 January 2021. doi:10.1177/0271678X20982388

McKnight

Rouleau

Donahue

, et al. The regulation of cerebral spinal fluid flow and its relevance to the glymphatic system. Curr Neurol Neurosci Rep 2020; 20: 58.

Jessen

Munk

Lundgaard

, et al. The glymphatic system: a beginner's guide. Neurochem Res 2015; 40: 2583–2599.

Bakker

Bacskai

Arbel-Ornath

, et al. Lymphatic clearance of the brain: perivascular, paravascular and significance for neurodegenerative diseases. Cell Mol Neurobiol 2016; 36: 181–194.

10.

Hladky

Barrand

MA.

Mechanisms of fluid movement into, through and out of the brain: evaluation of the evidence. Fluids Barriers CNS 2014; 11: 26.

11.

Spector

Robert Snodgrass

Johanson

CE.

A balanced view of the cerebrospinal fluid composition and functions: focus on adult humans. Exp Neurol 2015; 273: 57–68.

12.

Ahn

Cho

Kim

, et al. Meningeal lymphatic vessels at the skull base drain cerebrospinal fluid. Nature 2019; 572: 62–66.

13.

Iliff

Wang

Zeppenfeld

, et al. Cerebral arterial pulsation drives paravascular CSF-interstitial fluid exchange in the murine brain. J Neurosci 2013; 33: 18190–18199.

14.

Choudhury

DeBaun

Ponisio

, et al. Intracranial vasculopathy and infarct recurrence in children with sickle cell anaemia, silent cerebral infarcts and normal transcranial Doppler velocities. Br J Haematol 2018; 183: 324–326.

15.

Farrell

Panepinto

Carroll

, et al. End points for sickle cell disease clinical trials: patient-reported outcomes, pain, and the brain. Blood Adv 2019; 3: 3982–4001.

16.

Ikwuanusi

Jordan

Lee

, et al. Cerebral hemodynamics and metabolism are similar in sickle cell disease patients with hemoglobin SS and sbeta(0) thalassemia phenotypes. Am J Hematol 2020; 95: E66–E68.

17.

Quinn

McKinstry

Dowling

, et al. Acute silent cerebral ischemic events in children with sickle cell anemia. JAMA Neurol 2013; 70: 58–65.

18.

DeBaun

Gordon

McKinstry

, et al. Controlled trial of transfusions for silent cerebral infarcts in sickle cell anemia. N Engl J Med 2014; 371: 699–710.

19.

Guilliams

Fields

Dowling

MM.

Advances in understanding ischemic stroke physiology and the impact of vasculopathy in children with sickle cell disease. Stroke 2019; 50: 266–273.

20.

Juttukonda

Donahue

MJ.

Neuroimaging of vascular reserve in patients with cerebrovascular diseases. Neuroimage 2019; 187: 192–208.

21.

Guilliams

Fields

Ragan

, et al. Red cell exchange transfusions lower cerebral blood flow and oxygen extraction fraction in pediatric sickle cell anemia. Blood 2018; 131: 1012–1021.

22.

Donahue

Ayad

Moore

, et al. Relationships between hypercarbic reactivity, cerebral blood flow, and arterial circulation times in patients with moyamoya disease. J Magn Reson Imaging 2013; 38: 1129–1139.

23.

Kuroda

Houkin

Moyamoya disease: current concepts and future perspectives. Lancet Neurol 2008; 7: 1056–1066.

24.

Federau

Christensen

Zun

, et al. Cerebral blood flow, transit time, and apparent diffusion coefficient in moyamoya disease before and after acetazolamide. Neuroradiology 2017; 59: 5–12.

25.

Lee

Zaharchuk

Guzman

, et al. Quantitative hemodynamic studies in moyamoya disease: a review. Neurosurg Focus 2009; 26: E5.

26.

Zaharchuk

Marks

, et al. Arterial spin-labeling MRI can identify the presence and intensity of collateral perfusion in patients with moyamoya disease. Stroke 2011; 42: 2485–2491.

27.

Zhao

Taso

Dai

, et al. Non-invasive measurement of choroid plexus apparent blood flow with arterial spin labeling. Fluids Barriers CNS 2020; 17: 58.

28.

Scott

Smith

ER.

Moyamoya disease and moyamoya syndrome. N Engl J Med 2009; 360: 1226–1237.

29.

Rodrigue

, et al. Alterations in cerebral metabolic rate and blood supply across the adult lifespan. Cereb Cortex 2011; 21: 1426–1434.

30.

Juttukonda

Jordan

Gindville

, et al. Cerebral hemodynamics and pseudo-continuous arterial spin labeling considerations in adults with sickle cell anemia. NMR Biomed 2017; 30.

31.

Donahue

Faraco

Strother

, et al. Bolus arrival time and cerebral blood flow responses to hypercarbia. J Cereb Blood Flow Metab 2014; 34: 1243–1252.

32.

Fan

Guo

Khalighi

, et al. Long-delay arterial spin labeling provides more accurate cerebral blood flow measurements in moyamoya patients: a simultaneous positron emission tomography/MRI study. Stroke 2017; 48: 2441–2449.

33.

Hulbert

McKinstry

Lacey

, et al. Silent cerebral infarcts occur despite regular blood transfusion therapy after first strokes in children with sickle cell disease. Blood 2011; 117: 772–779.

34.

Casella

King

Barton

, et al. Design of the silent cerebral infarct transfusion (SIT) trial. Pediatr Hematol Oncol 2010; 27: 69–89.

35.

Alsop

Detre

Golay

, et al. Recommended implementation of arterial spin-labeled perfusion MRI for clinical applications: a consensus of the ISMRM perfusion study group and the European Consortium for ASL in dementia. Magn Reson Med 2015; 73: 102–116.

36.

Clingman

Golay

, et al. Determining the longitudinal relaxation time (T1) of blood at 3.0 tesla. Magn Reson Med 2004; 52: 679–682.

37.

Donahue

Jones

, et al. Theoretical and experimental investigation of the VASO contrast mechanism. Magn Reson Med 2006; 56: 1261–1273.

38.

Jenkinson

Beckmann

Behrens

, et al. Fsl. Neuroimage 2012; 62: 782–790.

39.

Plog

Lou

Pierre

, et al. When the air hits your brain: decreased arterial pulsatility after craniectomy leading to impaired glymphatic flow. J Neurosurg 2020; 133: 210–214.

40.

Wiesmann

Yousry

Seelos

, et al. Identification and anatomic description of the anterior choroidal artery by use of 3D-TOF source and 3D-CISS MR imaging. AJNR Am J Neuroradiol 2001; 22: 305–310.

41.

Klempt

Sirimanne

Gunn

, et al. Hypoxia-ischemia induces transforming growth factor beta 1 mRNA in the infant rat brain. Brain Res Mol Brain Res 1992; 13: 93–101.

42.

Knuckey

Finch

Palm

, et al. Differential neuronal and astrocytic expression of transforming growth factor beta isoforms in rat hippocampus following transient forebrain ischemia. Brain Res Mol Brain Res 1996; 40: 1–14.

43.

Chodobski

Szmydynger-Chodobska

Choroid plexus: target for polypeptides and site of their synthesis. Microsc Res Tech 2001; 52: 65–82.

44.

Prehn

Bindokas

Marcuccilli

, et al. Regulation of neuronal Bcl2 protein expression and calcium homeostasis by transforming growth factor type beta confers wide-ranging protection on rat hippocampal neurons. Proc Natl Acad Sci U S A 1994; 91: 12599–12603.

45.

Jordan

Kassim

Donahue

, et al. Silent infarct is a risk factor for infarct recurrence in adults with sickle cell anemia. Neurology 2018; 91: e781–e784.

46.

Kassim

Pruthi

Day

, et al. Silent cerebral infarcts and cerebral aneurysms are prevalent in adults with sickle cell anemia. Blood 2016; 127: 2038–2040.

47.

Faraci

Mayhan

Heistad

DD.

Effect of vasopressin on production of cerebrospinal fluid: possible role of vasopressin (V1)-receptors. Am J Physiol 1990; 258: R94–98.

48.

Schalk

Faraci

Williams

, et al. Effect of atriopeptin on production of cerebrospinal fluid. J Cereb Blood Flow Metab 1992; 12: 691–696.

49.

Nilsson

Lindvall-Axelsson

Owman

Simultaneous and continuous measurement of choroid plexus blood flow and cerebrospinal fluid production: effects of vasoactive intestinal polypeptide. J Cereb Blood Flow Metab 1991; 11: 861–867.

50.

Faraci

Mayhan

Heistad

DD.

Vascular effects of acetazolamide on the choroid plexus. J Pharmacol Exp Ther 1990; 254: 23–27.

51.

Waddle

Juttukonda

Lants

, et al. Classifying intracranial stenosis disease severity from functional MRI data using machine learning. J Cereb Blood Flow Metab 2020; 40: 705–719.

52.

Lin

Zuo

, et al. Test-retest reliability and reproducibility of long-label pseudo-continuous arterial spin labeling. Magn Reson Imaging 2020; 73: 111–117.

53.

Steketee

Mutsaerts

Bron

, et al. Quantitative functional arterial spin labeling (fASL) MRI–sensitivity and reproducibility of regional CBF changes using Pseudo-Continuous ASL product sequences. PLoS One 2015; 10: e0132929.

54.

Jiang

Kim

Chodkowski

, et al. Reliability and reproducibility of perfusion MRI in cognitively normal subjects. Magn Reson Imaging 2010; 28: 1283–1289.

55.

DeBaun

Ghafuri

Rodeghier

, et al. Decreased median survival of adults with sickle cell disease after adjusting for left truncation bias: a pooled analysis. Blood 2019; 133: 615–617.