MR Derived Volumetric Flow Rate Waveforms of Internal Carotid Artery in Patients Treated for Unruptured Intracranial Aneurysms by Flow Diversion Technique

Abstract

Little is known about the hemodynamic disturbances induced by the cerebral aneurysms in the parent artery and the effect of flow diverter stents (FDS) on these latter. A better understanding of the aneurysm-parent vessel complex relationship may aid our understanding of this disease and to optimize its treatment. The ability of volumetric flow rate (VFR) waveform to reflect the arterial compliance modifications is well known. By analyzing the VFR waveform and the pulsatility in the parent vessel, this study aimed to test the hypotheses that (1) intracranial aneurysms might disrupt the blood flow of the parent vessel and (2) the treatment by FDS might have measurable corrective effect on these changes. Ten patients followed for unruptured intracranial aneurysms treated by FDS and ten healthy volunteers as control group were included in this study. Two-dimensional quantitative phase-contrast magnetic resonance imaging (MRI) was performed on each patient on the ICA artery upstream and downstream to the aneurysm, and on each volunteer at similar locations. The aneurysms altered significantly the parent vessel pulsatility and this effect was correlated to their volume. The aneurysms treatment by FDS allowed for the restoration of a normally modulated flow and pulsatility correction in the parent vessel.

INTRODUCTION

Hemodynamic factors are considered to have an important role in progression, rupture, and endovascular treatment (EVT) efficacy of intracranial aneurysms.^1,2–4 The commonly admitted ultimate treatment goal is the vessel reconstruction and the correction of the hemodynamic disturbances in both the aneurysm sac and the parent vessel. While new generation devices acting on the parent vessel's hemodynamic are in development, such as flow diverter stents (FDS) that aim to induce the intra aneurysmal thrombosis,^5,6 little is known about the hemodynamic disturbances induced by the aneurysms in the parent arteries. Similarly, the effects of FDS on parent vessel hemodynamic changes due to the aneurysms are still poorly reported in the literature.

Previous studies on this area (i.e., aneurysmal disease) present two main limitations. (1) They are mainly based on numerical models, which require prior knowledge of reliable input functions. These latters are difficult to be measured in vivo and they are currently not readily available. Therefore, these approaches are still far from clinical practice and not routinely available for clinicians. (2) They mostly focus on the intra-aneurysmal flow modifications,^4,7 which lowlights the impact of the aneurysms on the parent vessel hemodynamic.

The ability of arterial compliance modifications, which is related to the status of the vasculature, to alter the volumetric flow rate (VFR) waveforms over the cardiac cycle is well known. ⁸ This may be of important interest for characterization of pathologic conditions. Some authors have proposed to assess the changes in VFR waveforms as a marker of diseases.^9,10 According to them, the evaluation of the VFR waveform of parent vessel (i.e., internal carotid artery (ICA)) could be of important value for the understanding of many diseases, such as intracranial aneurysms. Since phase-contrast magnetic resonance imaging (pc-MRI) is recognized to be an accurate and reliable non-invasive technique for velocity quantification in intracranial vessels,^11,12 several authors has suggested the production of archetypal VFR waveforms in ICA tree using this technique. ¹³ Using quantitative flow MRI, Gwilliam et al ⁹ proposed an original approach to assess the hemodynamic changes on carotid tree by the production of an average VFR waveform based on the averaging of identified key features of each waveform across subjects and by the computation of a pulsatility index (PI) derived from Doppler ultrasound technique. ⁹ The authors suggested that the analysis of the VFR waveforms over the cardiac cycle in the extracranial ICA might allow the monitoring or the screening of the cerebral aneurysms, similar to the one proposed for abdominal aortic aneurysms. ¹⁴ The work of Gwilliam et al established the method of waveform characterization adopted in this study. In this work using 2D phase contrast MRI, we aimed to test the hypotheses that (1) intracranial aneurysms might disrupt the blood flow of the parent vessel in patients followed for unruptured intracranial ICA aneurysms and (2) their treatment by flow diversion technique has measurable corrective effect on the parent vessel hemodynamic changes.

MATERIALS AND METHODS

Patients and Control Group

Ten patients with unruptured intracranial aneurysm selected for FDS implantation were prospectively recruited from April 2012 to December 2013. The indication for FDS implantation was posed after a multidisciplinary meeting in our institution for all patients. Of these, seven were women. The range of patient's age was 32 to 87 years resulting in a median age of 66 years with an interquartile range of 55 to 73 years. Aneurysms were located on ICA from the carotid siphon to its termination. Ten healthy volunteers were also enrolled as a control group. Of these, five were women. The range of healthy volunteers age was 22 to 55 years resulting in a median age of 29 years with an interquartile range of 25 to 39 years. Local ethics committee guidelines were followed for this study (DGRI CCTIRS MG/CP 2012.528; Comité d'Ethique du CHU de Lyon; Lyon/France). The informed consent was obtained from all patients and volunteers. Image data from those two groups were used for analysis. Five patients with no medical history were symptomatic because of aneurysm's mass effect at cavernous portion of ICA. They presented with a cavernous sinus syndrome associating headaches, ipsilateral ptosis, and ophtalmoplegia due to third, fourth, or sixth nerves palsy, without any visual acuity decreasing or pupillary abnormality. Two patients presented headaches without any obvious relationship with their aneurysms. One asymptomatic patient presented a medical history of high blood pressure, cigarette smoking, and hypercholesterolemia. One healthy volunteer presented a medical history of cigarette smoking. All other patients and control subjects did not present any symptoms or medical history, and specially, any vascular steno-occlusive lesion of the supra aortic trunks or intracranial arteries.

Aneurysms Treatment

All patients were treated under general anesthesia by using a biplane angiographic system (ALLURA, Philips, Best, The Netherlands) after a preparation according to our institutional protocol (loading dose of 300 mg of clopidogrel administrated one day before the EVT; systemic heparinization during the endovascular procedure, stopped at the end of the treatment; double antiplatelet therapy initiated for 6 months at day 1 after the treatment, with 75 mg of acetylsalicylic acid and 75 mg of clopidogrel/day). A 3D rotational angiography of the aneurysm and parent vessel was performed before the endovascular treatment (EVT) allowing for 3D reconstructions and the treatment planning. One or more FDS were deployed in one session according to aneurysm's neck size (PIPELINE, ev3-COVIDIEN, Irvine, CA, USA).

Aneurysm's Geometries

3D aneurysms and parent vessel geometries were segmented and reconstructed from the performed 3D angiographic acquisitions (spatial resolution 0.48×0.48 mm²) by using a new active contour method dedicated to the near real-time segmentation of 3D objects based on the level-set method. ¹⁵ It allowed the calculation of the maximal diameter (mm), depth (mm), neck size (mm), volume (i.e., volume of the patent intrasaccular lumen; mm³), and aspect ratio values of all aneurysms using dedicated software (ITK-SNAP, Penn Image Computed and Science Laboratory, University of Pennsylvania, USA).

Magnetic Resonance Imaging Examinations

All patients underwent MRI examinations using a clinical system operating at 3 Tesla with 32-Channel Array Head-Neck coil before and within the 3 days after the EVT according to the MR scan availability in our institution (3 Tesla, 45 mT/m, 200 T/m/s, Skyra, SIEMENS, Erlangen, Germany). After 1 month, an additional follow-up MRI examination was performed. To visualize the circle of Willis, a 3D time-of-flight MRA (3D TOF-MRA) was performed (149 slices; repetition time/echo time 21/3.4 ms; 18° flip angle; 0.6 mm section thickness; in-plane resolution 0.27×0.26 mm²). Maximum intensity projections derived from the resultant 3D TOF-MRA data set were used to define the placement of each two-dimensional slice to be used for quantitative flow estimation. Two dimensional phase-contrast MRI (2D pc-MRI) scan slices were performed at target points in the ICA upstream (extracranial) and downstream (intracranial) to the aneurysm, respectively (one slice; repetition time/echo time 180.5/10.5 ms; 7.5° flip angle; 3.1 mm section thickness; 0.31×0.31 mm² interpolated in-plane resolution; 37.6 to 41.2 ms temporal resolution). The phase-contrast sequence consisted of a K-space segmented 2D radio frequency—spoiled gradient echo sequence with prospective ECG gating (283 phase encoding steps; 1 views per segment). It resulted in 283 cardiac cycles acquired for one phase-contrast measurement. Data acquisition resulted in a series of 2D slices representing 2D blood flow through the slice of acquisition in consecutive timeframes within the cardiac cycle. Thereby, the number of frames varied with the patient's individual heart rate. The number of measurement points over the cardiac cycle varied from 20 to 32 with a temporal resolution of 37.6 to 41.2 ms (temporal resolution was defined as following = RR interval/number of frames). Special care was taken to reduce the angulation error risk of the scan planes regarding to the axis of blood flow by ensuring that the 2D pc-MRI slices were placed orthogonal to the long axis of the vessel of interest and therefore to the direction of blood flow. Thus, upstream and downstream scan planes were placed (1) at the sub- or intra-petrous level of the ICA and at its intracranial terminal level, respectively, (2) on relatively long and straight segments, and (3) before the posterior communicating artery, as illustrated in Figure 1. The part of the blood flow for ophtalmic artery and small branches of carotid siphon was considered as negligible given their size compared with that of the ICA. Similarly, in the control group, the scan planes were placed as previously described at comparable locations on the ICA. An appropriate velocity-encoding factor was chosen by the operator based on previous experience and defined as to result in images with significant contrast between the stationary tissue and flowing blood with the lowest wrap-around artifacts. Typical settings were 80 cm/s for the upstream measurements and 110 cm/s for the downstream measurements. The same couple neuroradiologist and technologist imaged all patients and volunteers to minimize the scan set-up variability. The reproducibility of the measurements (especially the positionning of the 2D phase contrast acquisition plans) was ensured by taking anatomic criteria spotted on the 3D time of flight acquisition as reference (i.e., the intrapetrous segment of the ICA). The 2D phase-contrast MRI slices post-EVT and at follow-up were placed as close as possible to the slices plane of the pre-EVT MRI. To be bearable by the patient, the table occupancy was kept within 20 minutes, including the 3D TOF-MRA and the two 2D pc-MRI acquisitions.

Figure 1.

2D phase-contrast magnetic resonance imaging of the parent vessel. (A) 1 = internal carotid artery, 2 = aneurysmal sac, 3 = aneurysmal neck, 4 = ophtalmic artery, 5 = parent vessel blood flow direction, 6 = intrasaccular blood flow direction, 7 = upstream 2DpcMRA slice, 8 = downstream 2DpsMRA slice. (B) 2DpcMRA images of the parent vessel downstream to the aneurysm. (C) 2DpcMRA images of the parent vessel upstream to the aneurysm. (D) 3D volume rendering reconstruction of the aneurysm based on 3D angiography acquisition.

2D Phase-Contrast Magnetic Resonance Imaging Postprocessing

Data of quantitative flow MRI were analyzed with a home-developed postprocessing tool specifically developed for this study with MATLAB software (R2014a Student version 8.3.0.532, The MathWorks, Natick, MA, USA). Blood velocities were extracted by a region-of-interest analysis from the phase difference images of the flow measurements. The arterial lumen boundaries were semi-automatically segmented from the surrounding stationary tissue by using an Active Contour Algorithm allowing segmentation without applied edge on all images during the cardiac cycle. ¹⁶ In case of vessel motion through the 2D pc-MRI slice, due to superadded motion of the circle of Willis at intracranial level, a manual segmentation of the arterial lumen over the cardiac cycle was performed by an experienced neuroradiologist. The software allowed an automated extraction of mean blood velocities (cm/s), vessel area (mm²) within the analyzed slice of the vessel for each timepoint over a cardiac cycle. Volumetric flow rates (VFR, mm³/s) and its waveforms were computed by integrating over time mean blood velocities within the region of interest that enclosed the vessel lumen closely.

Volumetric Flow Rate Waveform Feature Extraction

The method of feature extraction and normalization mirrors that of Gwilliam et al adapted from that of Ford et al.^9,10 The total flow values at each location were analyzed using the MATLAB software (R2014a Student version 8.3.0.532, The MathWorks). A semiautomated algorithm was used to identify the key points on the waveform. The points M0, P1, M1, P2, M2, and P3 were noted manually. The global peak, P1, was used as the initial feature point. Working backwards from P1, the point M0 corresponded to the point at which the gradient first changed sign and was defined as time 0. Working forwards from P1, the point where the gradient first changed sign was considered as the point M1. Similarly, points P2, M2, and P3 were identified. Points H0, H1, H2, and H3 were identified using an automated algorithm that linearly interpolated from the original waveform to find the time to the associated half peak from the associated local minima and peak. Points D1 to D3 were spaced equally between P3 and D4, D4 being the final point in the waveform. Splines were then fitted to the data to produce characteristic waveforms. An average waveform for each measurements points on the ICA was produced by taking the mean of feature points, both their magnitude and time, and a spline was fitted in similar manner. Each waveform was normalized by dividing each by its mean and by dividing each timepoint by the time taken to complete a cardiac cycle for each waveform. As described by, Gwilliam et al, this feature-based approach minimizes the attenuation of more local minima and maxima, which can vary in time between subjects. ⁹ The data were presented by graphical representations of VFR waveform over a cardiac cycle.

Statistical Analysis

The Mann-Whitney-Wilcoxon's Rank test was used to compare the ages of the two analyzed groups. A Chi-Square test with a correction of Yates was used to compare both groups regarding the cardiovascular risk factors. A quantitative assessment of the waveforms was performed through the computation of a PI adapted from the index used in Doppler ultrasound for the characterization of velocity waveforms. ¹⁷

$\text{PI}=\frac{F_{\text{max}}-F_{\text{min}}}{F_{mean}}$

where F_max is the maximum flow rate in waveform, F_min is the minimum flow rate in waveform, and F_mean is the mean flow rate in waveform. Since the PI in ICA might be affected by the age or some pathologic conditions of the patients,^18,19 such as high blood pressure, the effect of the aneurysm on the parent vessel hemodynamic was assessed through the computation of the pulsatility indexes ratio (PI-ratio) value, as follows:

$\text{PI}=\text{ratio}=\frac{\text{PI downstream}}{\text{PI upstream}}$

Results are expressed as median (25% to 75% interquartile). The PI-ratio values in both patient and control groups, and within patient group before the EVT, after the EVT and at follow-up were assessed using a Student's t-test or a Mann–Whitney rank sum test when needed according to the results of the Kolmogorov–Smirnov normality test. The Spearman's Rank correlation test was used to analyze the correlation between the PI-ratio values and the aneurysm volumes and aspect ratio values, respectively. The Kruskal–Wallis rank sum test and the Mann–Whitney Rank two-sided test were used to compare the mean blood flow rates (MBFRs) in the patient group at each timepoint, and between patient and control groups. A P value of less than 0.05 was considered to indicate a significant difference. The statistical analysis was performed with SigmaPlot software (Systat Software, San Jose, CA, USA).

RESULTS

The mean ages were of 63.8 ± 16 years old in patient group and of 32.6±11 years old in control group, with a significant difference (P < 0.001). Both groups did not show any significant statistical difference regarding the cardiovascular risk factors (P = 0.57).

Aneurysms Characteristics and Treatment Details

Table 1 summarizes the morphologic characteristics of the aneurysms and the number of implanted FDS.

Table 1.

Morphologic characteristics of the aneurysms in patient group

N	Aneurysm locationa	Diameter (mm)	Neck size (mm)	Depth (mm)	Volume (mm3)	Aspect ratio	Shape/partial thrombosis	Number of implanted flow diverter stents
1	Left C2 segment	12.2	6.5	12.2	984	1.88	Saccular/Regular	1
2	Left C2 segment	4.8	2.4	3.6	55	1.5	Saccular/Irregular	1
3	Left C2 segment	10.4	5.7	11.1	603	1.95	Saccular/Regular	1
4	Left C4 segment	20.9	8.8	14.8	3,648	1.68	Saccular/Regular	2
5	Right C3 segment	22	5.7	17.3	3,417	3.04	Saccular/Regular	1
6	Right C4 segment	17.1	12.9	17.6	1,817	1.36	Saccular/Regular	3
7	Left C2 segment	5.9	3.7	6.8	118	1.84	Saccular/Regular	1
8	Right C3 segment	25.4	20	23	6,735	1.15	Saccular/Regular/Part. thrombosed	0
9	Right C3 segment	30.1	6.5	20	4,888	3.08	Saccular/Regular/Part. thrombosed	1
10	Right C4-C3	22.4	13	13.7	2,319	1.05	Saccular/Regular	3
	segments
Median (25-75% IQ)		19 (10.8-22.3)	6.5 (5.7-11.9)	14,2 (11.4-17.5)	2,068 (698-3,590)	1.76 (1.39-1.93)

a

Aneurysm location according to Fisher's segmental classification of the ICA. C2: ophtalmic segment of the ICA; C3: anterior genu of the siphon; C4: intracavernous segment.

Aneurysms morphologic characteristic. Five aneurysms were located at right side. The aneurysms were located at the C2 segment (n = 4), the C3 segment (n = 3) and C4 segment (n = 3) of the carotid siphon according to the Fisher's classification of ICA segments. All aneurysms were saccular and two of them presented an intraluminal partial thrombosis. The median aneurysm's volume was 2,068 mm³ (698 to 3,590 mm³), the median maximal diameter was 19 mm (10.8 to 22.3 mm), the median neck size was 6.5 mm (5.7 to 11.9 mm), the median depth was 14.2 mm (11.4 to 17.5 mm) and the median aspect-ratio value was 1.76 (1.39 to 1.93).

Endovascular treatment. The EVT was successfully performed in 9/10 patients by deployment of one FDS (n = 6), two FDS (n = 1), and 3 FDS (n = 2) depending on the width of the neck and the patient's anatomy. Because of anatomic difficulties, patient 8 could not be treated by FDS and was scheduled for a parent artery occlusion. The patient contributed only to the pre-EVT analysis. No complication occurred during or immediately after the treatment, neither during the hospitalization period.

2D Phase-Contrast Magnetic Resonance Imaging

Before EVT, the 2D pc-MRI studies were successfully performed for all volunteers and patients (40/40; 100%). After the EVT, 2D pc-MRI was successfully performed in 17/18 acquisitions (94%) for the nine treated patients, at 6 hours (n = 4), 1 day (n =1) and 3 days (n = 4) after the endovascular procedure. The follow-up MRI examinations after 1 month were successfully performed between 1 and 12 months (median of 1 month; interquartile range of 1 to 3 months) in 17/18 (94%) of the treated patients. The downstream 2D pc-MRI was not analyzable in one patient (patient 7) after the EVT and at follow-up because the MR slice could not be appropriately placed. The FDS covered the downstream point of measurement on the ICA bifurcation resulting in a decrease of 2D pc-MRI signal due to the endovascular device. All upstream 2D pc-MRI was successfully performed and analyzable (100%). A manual correction of the segmentation was needed for two patients after the EVT and at follow-up, and for one patient at follow-up (5/58 = 8.6%), because of superadded motion of the measured arterial segment through the 2D pc-MRI slice at downstream locations. The parent vessels were patent in all treated patients at the post-EVT and at follow-up MRI examinations.

Pulsatility Indexes Ratio

The PI and PI-ratio values in control group and in patient group before the EVT, after the EVT, and at follow-up are reported in Table 2. The circle of Willis configurations, the upstream and downstream computationed volumetric blood flow rates (VBFR) before the EVT, after the EVT, and at follow-up are reported in Table 3. The measured velocities upstream and downstream to the aneurysm at each timepoint are reported in Supplementary Data (Supplementary Tables 1). The results of the statistical tests comparing the patient group and subgroups (A and B), and the control group regarding to their PI-ratio values are reported in Supplementary Data (Supplementary Table 4).

Table 2.

Pulsatility index and pulsatility indexes ratio values before EVT, after EVT, and at follow-up

N	Before the EVT			After the EVT			At follow-up > 1 month
	Upstream PI	Downstream PI	PI-ratio	TE	Upstream PI	Downstream PI	PI-ratio	TE	Upstream PI	Downstream PI	PI-ratio
Patient 1	0.616	0.611	0.992	D3	0.896	0.881	0.983	M1	1.037	0.899	0.867
Patient 2	0.942	0.973	1.033	D0	1.162	1.141	0.982	M12	0.935	1.062	1.136
Patient 3	0.572	0.616	1.077	D1	0.904	0.829	0.917	M11	0.855	0.775	0.906
Patient 4	0.813	0.946	1.164	D3	1.029	N/A	N/A	M3	0.788	N/A	N/A
Patient 5	0.826	0.580	0.702	D0	0.814	0.779	0.957	M1	0.920	0.872	0.948
Patient 6	1.199	0.860	0.717	D3	0.781	0.823	1.054	M1	1.184	1.111	0.938
Patient 7	1.230	0.761	0.619	D0	1.079	0.966	0.895	M1	0.788	0.714	0.906
Patient 8	0.920	0.346	0.376	N/A	N/A	N/A	N/A	N/A	N/A	N/A	N/A
Patient 9	1.008	0.625	0.620	D0	1.159	1.037	0.895	M1	0.860	0.849	0.987
Patient 10	0.877	0.898	1.024	D3	0.909	0.851	0.936	M3	0.728	0.805	1.106
Control 1	0.951	0.990	1.041
Control 2	0.980	0.878	0.896
Control 3	0.750	0.824	1.099
Control 4	0.736	0.760	1.033
Control 5	0.674	0.678	1.006
Control 6	0.665	0.558	0.839
Control 7	1.079	0.976	0.905
Control 8	1.245	1.278	1.027
Control 9	0.810	0.910	1.123
Control 10	0.891	1.010	1.134

D, day; EVT, endovascular treatment; M, month; N/A, non available data; PI, pulsatility index; PI-ratio, pulsatility indexes ratio; TE, time of MRI examination.

Table 3.

Configuration of circle of Willis and mean blood flow rates

N	Complete circle of Willis a	Anterior communicating artery	Ipsilateral A1 segment b	Ipsilateral posterior communicating artery c	MBFR BEVT (mL/min)		MBFR AEVT (mL/min)		MBFR FU (mL/min)		P-value
					Up-S	Down-S	Up-S	Down-S	Up-S	Down-S
Patient group
1	−	+	I < C	−	181	173	264	243	246	228
2	−	+	I = C	−	222	211	210	198	243	226
3	+	+	I = C	+f	277	258	299	278	328	317
4	−	+	I = C	−	338	286	351	na	296	na
5	−	+	I = C	+d	194	234	202	223	262	239
6	−	+	I = C	+f	253	227	264	244	227	212
7	−	+	I > C	−	310	269	296	252	232	218
8	+	+	I < C	+s	138	133	na	na	na	na
9	+	+	I < C	+s	156	156	205	185	179	167
10	−	+	I = C	−	267	232	241	348	316	272
Mean ± s.d.					234 ± 66	218 ± 50	259±50	246±51	259 ± 47	235 ± 44	0.44d
Control group
1	−	+	I = C	−	272	260	272	260	272	260
2	+	+	I = C	+s	368	350	368	350	368	350
3	−	+	I = C	−	266	255	266	255	266	255
4	+	+	I = C	+s	274	259	274	259	274	259
5	−	+	I = C	+f	316	359	316	359	316	359
6	+	+	I = C	+d	317	294	317	294	317	294
7	+	+	I = C	+s	316	300	316	300	316	300
8	+	+	I < C	+d	325	306	325	306	325	306
9	+	+	I = C	+d	341	314	341	314	341	314
10	+	+	I = C	+f	265	241	265	241	265	241
Mean ± s.d.					306±35	294 ± 40	306±35	294 ± 40	306±35	294 ± 40	0.24e
P-value					0.02e	< 0.01e	0.02e	0.03e	0.02e	0.01e

AEVT, after endovascular treatment; BEVT, before the endovascular treatment; Down-S, downstream; FU, at follow-up; MBFR, mean blood flow rate; Up-S, upstream. ^aNo complete circle of Willis (-) versus Complete circle of Willis (+). ^bI < C: Dominant contralateral A1 segment and small ipsilateral A1 segment; I = C: co-dominant A1 segments; I > C: dominant ipsilateral A1 segment. ^c(-): No ipsilateral posterior communicating artery; (+d): developed ipsilateral posterior communicating artery; (+f): fetal type ipsilateral posterior communicating artery; (+s): small ipsilateral posterior communicating artery. ^dKruskal–Wallis rank sum test for the comparison of MBFR in patient group. ^eMann–Whitney Rank two-sided test for the comparison of MBFR between patient and control groups.

The distribution of the PI-ratio values fulfilled the normality test (Kolmogorov–Smirnov normality test). Therefore, the PI-ratio values are expressed as mean±s.d. In control group, the mean PI-ratio value was 1.01 ±0.10. In patient group, the mean PI-ratio values were 0.83 ±0.26 before the EVT, 0.95 ±0.05 after the EVT, and 0.97±0.10 at follow-up without any significant difference (P = 0.79). These PI-ratio values before the EVT, after the EVT, and at follow-up were also not statistically different to the value in the control group (P = 0.13, P = 0.16, and P = 0.45, respectively). In patient group, the PI-ratio values before the EVT were significantly inversely correlated to the aneurysm's volumes (R = − 0.82, P = 0.001) as shown in Figure 2. They were not correlated to the aneurysm's volumes after the EVT (R = − 0.12, P = 0.75) or at follow-up (R = 0.14, P = 0.70). There was not any significant correlation between the aneurysm's aspect ratio and the PI-ratio values before the EVT (= − 0.05; P = 0.86), after the EVT (R = − 0.26; P = 0.50), or at follow-up (R = 0.14; P = 0.70).

Figure 2.

Distribution of the pulsatility indexes ratio (PI-ratio) values in patient group according to the aneurysm's aspect-ratio values and volumes before the endovascular treatment (EVT), after the EVT and at the follow-up. The PI-ratio values before the EVT were significantly inversely correlated to the aneurysm's volumes (R = −0.82, P = 0.001). They were not correlated to the aneurysm's volumes after the EVT (R = −0.12, P = 0.75) or at follow-up (R = 0.14, P = 0.70). There was not any significant correlation between the aneurysm's aspect ratio and the PI-ratio values before the EVT (= −0.05; P = 0.86), after the EVT (R = −0.26; P = 0.50), or at follow-up (R = 0.14; P = 0.70).

By analyzing retrospectively the PI-ratio values for each patient before EVT, two subgroups of patients could be identified according to whether their PI-ratio values before the EVT was significantly decreased or not compared with those of the control group.

Five patients (subgroup A; mean age of 64.4±11.4 years old) presented a mean PI-ratio value of 1.06 ±0.07 before the EVT, which was not statistically different from that in the control group (P = 0.36). None of these patients were symptomatic because of their aneurysms. In this subgroup, the mean PI-ratio values were 0.95 ±0.03 after the EVT and 1.00 ±0.14 at follow-up, without any significant difference (P =0.17), nor when compared with that in control group (P =0.31 after the EVT and P = 0.92 at follow-up).

A second group of five patients (subgroup B; mean age of 63.2 ±21 years old) presented a mean PI-ratio value of 0.61 ±0.14 before the EVT, which was significantly decreased compared with that in the control group (P < 0.001). All these patients were symptomatic because of their aneurysms and presented a cavernous sinus syndrome. In this subgroup, the PI-ratio values increased significantly after the treatment with mean PI-ratio values of 0.95 ±0.07 after the EVT (P < 0.001) and of 0.94±0.03 at follow-up (P < 0.001). The latter PI-ratio values were not statistically different (P = 0.90), nor when compared with that in the control group (P =0.31 after the EVT and P = 0.24 at follow-up).

The two subgroups presented PI-ratio values significantly different before the EVT (P < 0.001), but not after the treatment (P = 0.92 after the EVT and P = 0.89 at follow-up). The aneurysm's volumes in subgroups A and B were significantly different (P = 0.01), with a median aneurysm's volume of 603 mm³ (118 to 984 mm³) in subgroup A and 3648 mm³ (3,417 to 4,888 mm³) in subgroup B. The ages in these two subgroups were not statistically different (P =1). But, they were significantly different from those in the control group (P < 0.01 and P = 0.01 for subgroups A and B, respectively).

Mean Blood Flow Rates

In patient group, the MBFRs at upstream and downstream locations were 244±66 mL/min and 218±50 mL/min before the EVT, 259±50 mL/min and 246±51 mL/min after the EVT, and 259±47 mL/min and 235±44 mL/min at follow-up. No significant differences were observed between the upstream and downstream MBFR at each timepoint (P = 0.44), except in two patients (patients 4 and 10). They presented an important increase of the MBFR at downstream locations compared with the upstream locations (+20% and +44%, respectively) because of heart rate increases during downstream flow measurements (76 bpm versus 106 bpm for patient 4 and 53 bpm versus 75 bpm for patient 10). A discomfort felt by these two patients within the MR scan explained the heart rate variations. In control group, MBFR was 306±35 mL/min at upstream location and 294 ± 40 mL/min at downstream location, without any significant difference (P = 0.24). The MBFR was higher in the control group compared with those in patient group at both upstream and downstream locations before the EVT (P = 0.02 and P < 0.01, respectively). This difference was also significant after the EVT and at follow-up (P = 0.02 and P = 0.03 after the EVT, and P = 0.02 and P = 0.01 at follow-up).

Volumetric Flow Rate Waveforms

The normalized VFR waveforms in control group and in patient group before EVT are shown in Figure 3. The normalized VFR waveforms for patients in subgroup A and subgroup B before EVT, after EVT, and at follow-up are shown in Figure 4. The analysis of the VFR waveforms showed moderate modifications of the flow modulation at downstream measurement point in patient group before the EVT compared with the normal modulation in the control group. These modifications did not seem to affect the upstream segment of ICA. They were mainly characterized by a moderate increase of diastolic velocities (+14% of mean flow from P3 to D4 plate) associated with a slight decrease in systolic peak (-5% at P1) compared with the upstream VFR waveform. These modifications were more pronounced in the average VFR waveform of subgroup B before the EVT, with a clear increase in the diastolic velocities (+25%), a decreased (-10%) and moderately delayed (+6% of proportion of cardiac cycle) systolic peak with attenuation of the M1 and M2 minima. The average upstream and downstream VFR waveforms after the EVT and at follow-up were comparable to those of the control group. The subgroup A presented modulated VFR waveforms before the EVT, after the EVT, and at follow-up similar to those of the control group.

Figure 3.

Normalized volumetric flow rate (VFR) waveforms in control group and patient group before the endovascular treatment (EVT). Mean normalized VFR waveforms at upstream (i), downstream (ii) and both upstream and downstream (iii) segments of the internal carotid artery in control group (A) and in patient group (B). Features of the individual waveforms are also shown (x) for the upstream and downstream VFR waveforms (Ai, Aii, Bi, and Bii).

Figure 4.

Normalized volumetric flow rate waveforms in patient group, subgroup A and subgroup B before the endovascular treatment (EVT), after the EVT, and at follow-up. Mean normalized waveforms at upstream and downstream segments of the internal carotid artery in patient group (A), in patient subgroup A (B) and in patient subgroup B (C) before the EVT (Ai, Bi, Ci), after the EVT (Aii, Bii, Cii), and at follow-up (Aiii, Biii, Ciii).

DISCUSSION

The parent artery geometry and the position of the aneurysm with respect to it are known to impact on the intra aneurysmal flow and the FDS efficiency.^1,20 In this study, we highlighted two main findings: (1) the intracranial aneurysms induce some hemodynamic disturbances in parent vessels. This effect of the aneurysms was strongly correlated to their volumes. (2) The FDS have a measurable corrective effect on these flow modifications in the parent vessel for large aneurysms. To our knowledge, this is the first work quantifying the hemodynamic impact of the intracranial aneurysms on the parent vessel and its correction by FDS. Our results are in good agreement with previous data of Gwilliam et al. ⁹ The average VFR waveforms presented similar shape and modulation over a cardiac cycle as described previously, with no significant change in pulsatility (i.e., PI) between upstream and downstream measurement points on ICA (represented by PI-ratio values close to 1) in both control group and patient group after the treatment.

As hypothesized by Gwilliam et al the geometric changes in artery caused by an aneurysm induce some hemodynamic modifications in the parent vessel. It was responsible for a flow disruption downstream to the aneurysm which was characterized by the decrease in the systolic peak and the relative increase in the diastolic flow, which was reflected by a decreased PI-ratio value. This flow disruption was strongly and inversely correlated to the aneurysm's volume and not to its aspect-ratio. These modifications were visible only for large aneurysms (subgroup B, median volume: 3,648 mm³), whereas the smaller ones (subgroup A, median volume: 603 mm³) did not seem to impact significantly the parent artery hemodynamic. The mechanisms involved in this dampening effect of the aneurysms remain unclear. One can presume that during systole a part of blood flow might be retained in the aneurysm first, explaining the relative decrease in systolic peak velocity downstream, and might be « regurgitated » into the parent vessel during diastole in a second time, explaining the relative increase in diastolic velocities. This pattern of hemodynamic changes in the parent vessel visible with the large aneurysms leads us to make two hypotheses: (1) the parent vessel segment in-between the upstream and downstream measurement locations, including the diseased aneurysmal segment, presents an increased compliance (or distensibility) and/or (2) the aneurysm geometry induces significant flow modifications that affect the measurement in the downstream segment of the parent vessel. The first hypothesis is in good agreement with some biomechanical properties of aneurysm wall reported by some authors,^21–23 and especially a higher distensibility at the aneurysm's dome. ²² Indeed, we can reasonably assume that in large aneurysms the dome surface should be large and therefore induce significant impact on flow. The second hypothesis cannot be excluded since the downstream flow measurements were performed close to the aneurysms because of the anatomic constraints (i.e., the need of a straight and sufficiently long segment to have a reliable and reproducible measurements with 2D pc-MRI technique). Interestingly, the modifications of the VFR waveforms were depicted only within the segment of the ICA downstream to the aneurysm and not at upstream level (extracranial), as suggested by Gwilliam et al. This result weakens the possibility to implement a screening of the patients with intracranial aneurysms at cervical level of the carotid tree by a simple method such as Doppler ultrasounds technique.

The FDS for the treatment of unruptured intracranial has already shown a great promise by allowing the ‘morphologic reconstruction’ of the diseased arterial segment.^5,24 We observed and quantified a ‘hemodynamic reconstruction’ of the parent vessel treated by FDS. This ‘hemodynamic reconstruction’ was observable only for large aneurysms (subgroup B) immediately after the EVT and was characterized by an increase in the PI-ratio values reflecting the increase in PI values downstream to the aneurysm. It showed the restoration of a normally modulated flow pattern distal to the aneurysm, which remained unchanged at follow-up examination. Regarding the MBFR in upstream and downstream segments of the parent artery, they were comparable and seemed not modified by the presence of the aneurysm, which is consistent with the law of conservation of mass. It suggests that large aneurysms affect the PIs (i.e., the flow pattern) rather than the MBFR in the downstream segment of parent vessel, as shown by the graphical representation of the volumetric flow waveforms. This result suggests that the changes in the downstream vessels after the EVT, and thereby more distally in the vascular bed within the brain parenchyma, are more related to the acute PI correction than to the MBFR modifications, which should not be significant according to the law of conservation of mass.

Interestingly, all the symptomatic patients had the lower PI-ratio values before the EVT, which were associated with the larger volumes. Whereas all the asymptomatic patients had PI-ratio values comparable to those in control group. This analysis is biased by the aneurym's volume, which can be considered as a confounding factor responsible for both the symptoms and the PI-ratio values decrease.

These findings provide new insights on the complex relationship between aneurysms and the parent vessel hemodynamic. We believe that the hemodynamic changes in both the parent vessel and the aneurysmal sac are indissociable for a better comprehension of the intracranial aneurysmal disease. From a clinical point of view, it allows us to make assumptions about some complications reported after treatment of large aneurysms. One can reasonably assume that the sudden increase in the PI in a probably long-standing downstream dampened blood flow (i.e., with a decreased PI) within the ipsilateral hemisphere could have a role in intraparenchymal hemorrhage or in hyperperfusion syndromes that have been previously reported after the treatment of large aneurysms either by clipping or by endovascular techniques.^6,25,26 The knowledge of this eventual effect for large aneurysms could be taken into account for the treatment decision and the postoperative management of the patients. We did not observe such complications in our series of patients.

Different methods have been described to measure flow boundary conditions in aneurysmal disease. Levitt et al measured patient-specific inflow and outflow boundary conditions by using an intravascular dual-sensor pressure and Doppler velocity guidewire, during the EVT before and after the FDS placement. ²⁷ This approach is invasive and provides boundary conditions for only retrospective computational flow dynamic studies, which limits its use for a prospective purpose. It has been also shown that position of the dual-sensor guidewire and its catheter in the vessel lumen induces flow modifications and lead to the underestimation of the measured velocities. ²⁸ Thus, it weakens the reliability of the method. Furthermore, the location and angle of the wire in the vessel lumen may also influence the measurements and limit its use in tortuous arteries. ²⁹ With the aim of setting an easily usable and reproducible method for physicians, we used 2D pc-MRI technique for flow measurements. This technique has been shown to be an accurate and reliable method for velocity quantification in cervical and intracranial vessels in various cerebrovascular conditions,^{11,12,30–33} such as aneurysmal disease. ³⁴ It is highly correlated to Laser Doppler Velocimetry as reference method and endovascular Doppler techniques for flow quantifications, especially in large arteries as in ICA.^35–37 This noninvasive and nonirradiating technique is widely available in clinical practice. In this work, a time-resolved 2D pc-MRI technique was used rather than a time resolved 3D pcMRI technique. This latter requires long total scanning times that are prohibitive in clinical practice, ³⁸ presents lower both in-plane spatial and temporal resolutions at the same magnetic field strength, and might lead to an underestimation of peak velocities due to low-pass filtering effect. ³⁹ Thus, 2D pc-MRI appeared to be more suited to this work, since it combines reduced acquisition times and high both (in-plane) spatial and temporal resolutions. Since, the table occupancy was voluntarily limited (20 minutes) the method allowed for the measurements of only 20 to 31 points over the cardiac cycle for each phase-contrast acquisition. This was clearly lower than the 40 measurements points in the work of Gwilliam et al. However, it did not seem to have any impact on the reproducibility of the method and the VFR waveform analysis. Moreover, we used a 2D pc-MRI acquisition with higher spatial resolution (0.31×0.31 mm² in-plane resolution versus 1.72×1.60 mm² in previous work), which allowed for more accurate signal with a short image times. The 2D pc-MRI was successfully performed by most of the patients and healthy volunteers.

Limitations

This study had several limitations. Since only aneurysms of the intracranial ICA treated by FDS were included in this feasibility study, the number of patients was limited. Only ICA aneurysms were included to simplify the analysis and to limit the velocity measurements on relatively large and straight arterial segments for which no underestimation in flow and velocity measurements is expected with 2D pc-MRA. ¹⁸ Only two partially thrombosed aneurysms were analyzed. Therefore, the effect of the intrasaccular thrombosis on the flow waveform could not be assessed. Thus, the significance of PI-ratio values regarding intrasaccular thrombosis was not evaluated. Larger prospective studies could be considered to investigate this point. But, the large range of aneurysm's volumes (from 55 to 6,735 mm³) enabled the demonstration of a significant correlation between the aneurysm's volume and its impact on the parent vessel hemodynamic. Control and patient groups were not matched for age because of the limited size of our series. The mean age of control group was significantly lower than the mean age of patient group. The effect of age on the PI of large arteries, such as the ICA, has already been reported in the literature.^18,19 Since we analyzed the PI-ratio values between the upstream and downstream segments of the parent vessel relative to the aneurysm's location instead of the PI alone, our methodology should overcome the effect of age. Furthermore, the PI-ratio values in subgroup A at each timepoint, the PI-ratio values in subgroup B after the EVT and at follow-up, and those in the control group, as well, were comparable and not significantly different. These latter results highlight the robustness of the PI-ratio value analysis in time regarding the potential effects of age. The effect of arterial normal stiffening associated with the age affecting the arterial compliance was not observed in our limited cohort, either in patient or in control groups. However, the significant age difference between patient and control group had an impact on the computationed MBFR. The MBFR was higher in the control group compared with the patient group before the EVT, after the EVT, and at follow-up. This result is in good agreement with previous studies that had shown an inverse correlation between the age and the MBFR.^18,19

CONCLUSION

The intracranial aneurysms induce some flow modifications in the parent vessel which are measurable by 2D pc-MRI technique, through a pulsatility indexes ratio computation and VFR waveform analysis. This impact of aneurysms on parent hemodynamic is strongly correlated to their volumes and corrected by FDS for large aneurysms. This approach could help to better understand the complexity of aneurysm-parent vessel hemodynamic interactions. Further studies on a larger number of patients with different aneurysmal localizations and geometries will be needed to confirm our results and to clarify the significance of the computed PI-ratio values regarding the interaction between the aneurysm and parent vessel hemodynamics, and also the long-term efficiency of the FDS.