Reduced circulating endothelial progenitor cells in reversible cerebral vasoconstriction syndrome

The study protocol was approved by the Institutional Review Board of Taipei Veterans General Hospital. All participants provided written informed consent before entering the study. All clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki. The corresponding authors had full access to all of the data in the study and had final responsibility for the decision to submit the results for publication.

Patients with RCVS were recruited from the Neurology Department of Taipei Veterans General Hospital, a 2,909-bed national medical center located in Taipei City from 2011 to 2013. Age- and sex-matched volunteer participants who had neither headache history nor severe medical illness were recruited as normal controls. The inclusion criteria for the participants (both patients and controls) were the following: (1) age between 20 and 65 years, and (2) subjects could fully understand the purpose of the research and voluntarily join the study. Because many intrinsic or exogenous factors could influence the number of circulating EPCs, we applied very strict exclusion criteria and selectively enrolled matched controls to eliminate the influence of these confounders. Subjects were excluded if they met the following criteria: (1) had a smoking history, (2) had a major systemic illness such as uncontrolled hypertension (systolic blood pressure > 160 mmHg, diastolic blood pressure > 100 mmHg) at baseline, diabetes mellitus, cardiovascular or cerebrovascular disease, chronic hepatic or renal disease, or malignancies, (3) used illicit drugs; or (4) women who were pregnant or within 3 months postpartum. Subjects with a history of migraine and grade 1 hypertension (systolic blood pressure 140–159 mmHg, diastolic blood pressure 90–99 mmHg) were permitted to enroll because some patients with RCVS may have migraine and hypertension.

The diagnosis of RCVS was based on the following criteria: (1) at least two acute-onset severe headaches (thunderclap headaches), with or without focal neurological deficits; (2) vasoconstrictions demonstrated on magnetic resonance angiography (MRA); (3) reversibility of vasoconstrictions, as demonstrated by at least one follow-up MRA within 3 months; and (4) aneurysmal subarachnoid hemorrhage or other intracranial disorders were ruled out by appropriate investigations, but cortical subarachnoid hemorrhage in RCVS was allowed. The diagnostic criteria were adapted from the criteria of "benign (or reversible) angiopathy of the central nervous system" proposed by the International Classification of Headache Disorders, second edition (ICHD-2) (Code 6.7.3)[20] with the exception of the duration criterion D as well as the essential diagnostic elements of RCVS proposed by Calabrese et al.[1]. In this study, only patients with idiopathic RCVS within one month of headache onset were eligible to minimize potential confounders.

Basic demographic information and anthropometric measurements including height, body weight, and body mass index (BMI)[21] were collected. All enrolled subjects received comprehensive clinical and neurological examinations, and provided detailed medical, headache, and drug histories upon entering the study. Lipid profiles were obtained from general serological surveys. Diagnostic investigations including the first MRA, transcranial color-coded sonography and/or spinal taps were performed within the first two days after the patient was seen. Control subjects did not receive neuroimaging studies. The protocols have been reported previously[3, 7, 8]. In brief, brain MRI, MRA and MR venography (MRV) studies were performed with a 3-Tesla MR scanner. Intracranial lesions were carefully evaluated with adequate MR sequences, for example, subarachnoid hemorrhage with susceptibility weighted imaging (SWI) and intracranial aneurysms with three-dimensional time-of flight (TOF) MRA. Sequential MRAs were performed in all subjects with RCVS until the normalization of vasoconstrictions. The mean flow velocity of the major cerebral arteries was detected by transcranial color-coded sonography and recorded; the protocols were detailed elsewhere[7]. The following data were analyzed: mean flow velocity of middle cerebral artery (MCA) (V_MCA) and the Lindegaard index (LI), which is calculated by dividing the V_MCA by the mean flow velocity of the ipsilateral distal extracranial internal carotid artery. The LI was selected to evaluate the vasoconstrictions to avoid possible misreading of hyperemia as vasospasm. For computational purposes, the average V_MCA and LI derived from the bilateral MCAs in each patient were calculated. We analyzed only V_MCA and LI because MCA is the most widely studied and validated vessel for vasoconstrictions using transcranial color-coded sonography (TCCS) and there have been no validated sonographic criteria of vasoconstrictions for the other cerebral vessels with adequate sensitivity[7]. In addition, our previous studies showed a good correlation of V_MCA and LI with the severity of vasoconstrictions detected by MRA.

Because the severity of vasoconstriction of the MCA could vary between patients at the time of blood sampling, we further stratified the patients according to their LI values. Based on our previous study[7], an LI value of 2, the upper limit of our normal controls, was chosen to stratify the patients into two groups: those having either significant or insignificant vasoconstrictions of the MCA. An LI value of 3 was also used to evaluate whether patients at risk of ischemic complications[7] could have lower EPCs. Of note, the cut-off values of LI used here were based on our previous study in RCVS patients using TCCS[7] and should not be translated to the same severity of vasopasm in SAH based on studies using transcranial Doppler (TCD)[22]. In patients with SAH, the LI is disproportionately amplified because of low extracranial ICA flow velocity due to cerebral hypoperfusion caused by increased intracranial pressure (IICP)[22, 23], while patients with RCVS usually have normal ICP[1‐3]. Patients with RCVS who fulfilled the mild vasospasm criteria for SAH already had high risks of developing PRES (75%) or ischemic stroke (50%)[7].

The flow cytometric analysis of circulating EPCs was performed based on our previously established methods[24]. Peripheral blood was collected from the antecubital vein on the same day of transcranial color-coded sonography, before administration of any therapeutics. The samples were processed immediately after blood drawing. A volume of 100 μL blood was incubated for 15 minutes in the dark with monoclonal antibodies against human kinase insert domain receptor (KDR; R&D, Minneapolis, MN, USA) followed by phycoerythrin (PE)-conjugated secondary antibody, with the fluorescein isothiocyanate (FITC)-labeled monoclonal antibodies against human CD45 (Becton Dickinson, Franklin Lakes, NJ, USA), with the PE-conjugated monoclonal antibody against human CD133 (Miltenyi Biotec, Germany), and with FITC-conjugated or PE-conjugated monoclonal antibodies against human CD34 (Serotec, Raleigh, NC, USA). Isotype-identical antibodies were used as controls (Becton Dickinson, Franklin Lakes, NJ, USA). After incubation, cells were lysed, washed with phosphate-buffered saline (PBS), and fixed in 2% paraformaldehyde before analysis. Each analysis included 100,000 events. Flow cytometry was used to quantify the numbers of circulating EPCs defined as KDR⁺CD133⁺, CD34⁺CD133⁺, and CD34⁺KDR⁺ double-positive mononuclear cells (Additional file1: Figure S1). To assess the reproducibility of the EPC measurements, circulating EPCs were measured from two separate blood samples in 20 subjects (10 controls and 10 patients each). Mature circulating endothelial cells, which are generally recognized as cells expressing endothelial markers (CD 144, CD 146, vWF, VEGFR-2) in the absence of hematopoietic (CD 14, CD 45) and progenitor (CD 34, CD 133) markers (despite the lack of a clear concensus on their phenotypic definitions)[25, 26], were excluded with this method.

Descriptive statistics are presented as the mean ± standard deviation or as the number (percentage). For normally distributed variables, continuous ones were compared by the Student’s t-test, and categorical ones were compared by the Chi-square or Fisher exact tests. For variables that don’t have a normal distribution, Kruskal-Wallis test and Mann–Whitney U test were used as appropriate. For correlations between EPC counts with other variables, Spearman correlation coefficient r_s was applied because EPC counts were not normally distributed (by Shapiro-Wilk test). All calculated p-values were two-tailed. Statistical significance was defined as p < 0.05. Because there was collinearity between the number of KDR⁺CD133⁺, CD34⁺CD133⁺, and CD34⁺KDR⁺ cells (the correlation coefficients were 0.666 for CD34⁺CD133⁺ and CD34⁺KDR⁺ cells (p < 0.001), 0.735 for KDR⁺CD133⁺ and CD34⁺CD133⁺ cells (p < 0.001), and 0.570 for KDR⁺CD133⁺ and CD34⁺KDR⁺ cells (p < 0.001)), corrections for multiple comparisons were not applied in this study. All analyses were performed with the IBM SPSS Statistics software package, version 18.0.

During the two-year study period, we prospectively recruited 31 patients with RCVS. After excluding one patient with a previous history of intracerebral hemorrhage, one patient with previous cerebral infarction, two patients who used psueodephedrine, one patient who had a history of marijuana use, one patient who was postpartum, and one heavy smoker, 24 patients were eligible for study. Another 24 age- and sex-matched controls were recruited during the study period. There was no difference in the demographics, BMIs, medical illnesses, and serum cholesterol levels between the two groups (Table 1). Most of the patients (91.7%) had identifiable triggers for their thunderclap headaches, with defecation (58.3%) and bathing (29.2%) being the most common. The cerebrospinal fluid was studied in 12 (50%) patients, and their results were all normal.

None of the patients had posterior reversible encephalopathy syndromes, ischemic stroke, intracerebral hemorrhage or intracranial arterial dissection on the initial or follow-up MR studies, but one patient had cortical subarachnoid hemorrhage confined to right anterior high frontal cortex without identifiable aneurysm. None of the patients had aneurysmal subarachnoid hemorrhage in consecutive MR studies.

The blood samples were taken at a mean of 5.4 ± 3.7 days (range 2–13) after headache onset. Of the 20 subjects who had two separate blood samples, there was a strong correlation between the two measurements suggesting high reproducibility (CD34⁺CD133⁺: rs = 0.973, p < 0.001; CD34⁺KDR⁺: rs = 0.972, p < 0.001; and KDR⁺CD133⁺: rs = 0.976, p < 0.001). There were no correlations between the EPC counts and age, gender, BMI, menopausal status, migraine, hypertension, or serum cholesterol level.

In comparison to the controls, patients with RCVS had reduced numbers of CD34⁺KDR⁺ cells (0.009 ± 0.006% vs. 0.014 ± 0.010%, p = 0.031), but they had no reduction in KDR⁺CD133⁺ cells (0.031 ± 0.015% vs. 0.039 ± 0.020%, p = 0.176) or CD34⁺CD133⁺ EPCs (0.045 ± 0.028% vs. 0.060 ± 0.041%, p = 0.140). The EPC counts in patients with comorbid migraine (n = 4) did not differ from that in patients without comorbid migraine (n = 20) (CD34⁺KDR⁺ cells (0.007 ± 0.006% vs. 0.009 ± 0.006%, p = 0.751), KDR⁺CD133⁺ cells (0.027 ± 0.010% vs. 0.032 ± 0.015%, p = 0.642), CD34⁺CD133⁺ EPCs (0.042 ± 0.007% vs. 0.045 ± 0.030%, p = 0.781)).

In comparison with the controls, patients with an LI > 2 (n = 13) had fewer CD34⁺KDR⁺ cells (0.007 ± 0.005% vs. 0.014 ± 0.010%, p = 0.010), but there was no difference in the number of KDR⁺CD133⁺ cells (0.032 ± 0.014% vs. 0.039 ± 0.020%, p = 0.186) or CD34⁺CD133⁺ EPCs (0.042 ± 0.023% vs. 0.060 ± 0.041%, p = 0.087).The EPC counts of patients with LI ≤ 2 did not differ from those of the controls (CD34⁺KDR⁺ cells: 0.012 ± 0.006% vs. 0.014 ± 0.010%, p = 0.392; KDR⁺CD133⁺ cells: 0.036 ± 0.016% vs. 0.039 ± 0.020%, p = 0.404; CD34⁺CD133⁺ cells: 0.052 ± 0.032% vs. 0.060 ± 0.041%, p = 0.534.) (Figure 1)

In patients with LI > 3 (n = 3), the trend was similar: Their CD34⁺KDR⁺ cell counts were lower than the controls (0.006 ± 0.002% vs. 0.014 ± 0.010%, p = 0.015), but there was no difference in the numbers of KDR⁺CD133⁺ cells (0.023 ± 0.008% vs. 0.039 ± 0.020%, p = 0.070) or CD34⁺CD133⁺ EPCs (0.032 ± 0.012% vs. 0.060 ± 0.041%, p = 0.054).

The number of CD34⁺KDR⁺ cells was negatively correlated with the LI (rs = -0.418, p = 0.047) (Figure 2), whereas there was no significant correlation of LI with the number of KDR⁺CD133⁺ (rs = -0.265, p = 0.221) or CD34⁺CD133⁺ cells (rs = -0.207, p = 0.344). None of the other clinical variables such as triggers (including defecation, bathing, exertion, cough, rage, or sex) or the number and duration of thunderclap headaches before blood drawing had significant correlations with the EPCs counts (data not shown).