Ultrasound in Med. & BioL Vol. 16, No. 8, pp. 763-772, 1990 Printed in the U.S.A.
0301-5629/90 $3.00 + .00 © 1990 Pergamon Press plc
OOriginal Contribution RELATIONSHIP BETWEEN PULSATILITY FLOW SIGNALS AND CO2-REACTIVITY CEREBRAL ARTERY IN EXTRACRANIAL
INDICES OF DOPPLER WITHIN THE MIDDLE OCCLUSIVE DISEASE
JUAN LEY-Pozo,* KLAUS WILLMES and ERICH BERND RINGELSTEIN Department of Neurology, University Hospital, Klinikum RWTH, D-5100 Aachen, Germany (Received 29 August 1989; in final form 11 April 1990) Abstract--CO2-dependent vasomotor reactivity (VMR) of the middle cerebral artery (MCA) distribution has been shown to be a reliable predictor of the patient's risk for low-flow strokes. In order to clarify whether various Pulsatility Indices (PIs) of MCA flow signals would be useful to estimate VMR, 222 subjects with various degrees of stenosing internal carotid artery lesions were studied. V M R of the MCA territory was expressed as the percentage changes in mean blood flow velocity during inhalation of CO2 and hyperventilation. Although the correlation coefficients between V M R and PIs were significantly different from zero, these relationships were not strong enough for PI values to accurately predict the individual values of VMR. Furthermore, V M R could convincingly discriminate subgroups of patients with different severities of extracranial occlusive disease, whereas PIs discriminated only weakly or not at all. PIs may give some vague idea of the vasomotor tone, but cannot predict precisely enough whether a patient is at risk for low-flow infarctions.
Key Words: Transcranial Doppler sonography, Pulsatility index, Vasomotor reserve, CO2-reactivity, Cerebrovascular disease.
ity) o f Gosling and King (1974), and more recently, the Pulsatility Transmission Index (PTI) (see below) of Lindegaard et al. (1985) have been recommended as easily obtainable parameters reflecting both the intracerebral peripheral vascular resistance and vasom o t o r responcivity. Their usefulness in clinical practice, however, has not been proven so far. The purpose of this paper was to test whether PI and PTI are precise and valid enough to replace CO2-dependent V M R measurements and to clarify to what extent these indices would be able to discriminate subgroups o f patients with various degrees of extracranial cerebrovascular occlusive disease resulting from its impact on intracranial blood flow.
INTRODUCTION In 1982, Aaslid et al. i n t r o d u c e d T r a n s c r a n i a l Doppler Sonography (TCD) into clinical practice for noninvasive investigation o f the large intracranial arteries. This m e t h o d permits the assessment of cerebral vasomotor reactivity (VMR) upon vascular stimuli, such as pCO2 d u r i n g hyper- and h y p o c a p n i a (Markwalder et al. 1984; Aaslid 1986; Widder et al. 1986; Ringelstein et al. 1988). Although the clinical usefulness of CO2 tests has been well documented (Ringelstein et al. 1988; Widder et al. 1987), they are time-consuming, and often uncomfortable for the patient. Simple n u m e r i c a l analysis o f the shape o f transcranial Doppler flow signals also gives information concerning the functional state o f the cerebral circulation (Aaslid and Lindegaard 1986). The Pulsatility Index (PI = peak-to-peak velocity/mean veloc-
M A T E R I A L S AND M E T H O D S Recruitment o f patients Between September 1987 and December 1988, 222 consecutive patients with stroke a n d / o r cerebrovascular disease were investigated. Both T C D studies o f the basal cerebral arteries and continuous-wave (CW) Doppler examinations of the extracranial brain arteries were performed. V M R values and PIs were evaluated not earlier than six weeks after a vascular
Address correspondenceto: E. Bernd Ringelstein, M.D., Associate Professorof Neurology,Department of Neurology,University Hospital, Klinikum RWTH, Pauwelsstrasse, D-5100 Aachen, Germany. t Dr. Ley-Pozois a research fellowfrom Cuba supported by a grant from the Alexander von Humboldt Foundation, Bonn/FRG (Grant No. IV 1-71147). 763
764
Ultrasound in Medicine and Biology
event. Mean age was 61.3 +_ 33 years. One hundred and twenty-three patients were males and 98 were females. In total, 392 "hemispheres," i.e., middle cerebral arteries (MCA) were considered. Measurements were not possible in 52 MCA territories due either to lack of an "ultrasound" window within the temporal bone, poor signal-to-noise ratio, or uncooperative patients.
Clinical subgroups From a clinical point o f view, the cerebral hemispheres u n d e r study were divided into four subgroups: (I) permanent neurological deficit (N = 81), (II) transient neurological deficit (N = 81), (III) asymptomatic hemispheres (N = 187), and (IV) postoperative state after carotid endarterectomy (N = 43) (all of the latter had had previous TIAs).
A ngiological subgroups On the basis of CW Doppler findings, both internal carotid arteries (ICA) were classified as either (1) " O c c l u d e d " ( " O " ) , (2) "High-grade stenosis" ( " H " ) (i.e., reduction in cross-sectional area o f more than or equal to 80%), (3) "Low-grade stenosis" ("L") (i.e., reduction in cross-sectional area o f less than 80%, but o f more than or equal to 60%), or (4) " N o stenosis" ( " N " ) (i.e., reduction in cross-sectional area o f less than 60%). The detailed methodology for CW Doppler examination o f the extracranial arteries has been described previously (Spencer and Reid 1979; Ringelstein 1989a). With respect to bilateral ICA findings, nine subgroups were considered: (1) bilaterally occluded ICA (indicated by O-O), (2) unilateral ICA occlusion with contralateral high-grade stenosis (O-H), (3) unilateral ICA occlusion with contralateral low-grade or no stenosis (O-LN), (4) bilateral ICA high-grade stenosis (H-H), (5) unilateral ICA highgrade stenosis with contralateral low-grade or no stenosis (H-LN), (6) unilateral low-grade ICA stenosis with contralateral high-grade ICA stenosis (L-H), (7) unilateral low-grade ICA stenosis with contralateral low-grade or no stenosis (L-LN), (8) ICA with no stenosis (irrespective of the situation of the contralateral side) (N-X), and (9) ICA after endarterectomy, at least two weeks after successful carotid recanalization (this group coincides with the clinical subgroup IV). It was reasonable to differentiate subgroups (3) and (5) from (6) and (8), since V M R and PI/PTI values always referred to the side mentioned first.
Transcranial Doppler sonography The transtemporal identification and isonation of the MCA with a T C D Doppler device (TC 2-64, EME, Ueberlingen, FRG) was based on an anterior
Volume16, Number 8, 1990 angulation of the beam, an insonation depth between 45 and 50 mm, the blood flow being directed towards the probe, and traceability of the vessel from 40-55 m m insonation depth (for details, see Aaslid 1986; Ringelstein 1989b).
Measurement and calculation of PI. PI has been defined as the ratio between th e peak-to-peak velocity and mean velocity (Gosling and King 1974). Mean flow velocity values as originally defined by Aaslid et al. (1982) were automatically calculated by the instrument over a period varying from 3-10 s depending on the sweep speed used. T C D Doppler flow signals o f both MCAs were registered with each subject resting comfortably in a supine position. Velocity spectra were recorded for 20 s during steady-state and were stored on tape. Traces with obvious artifacts were deleted. PI values were measured of the residual cardiac cycles. The mean of these PI values was calculated off-line. Calculation of PTI. T h e PTI was defined by Lindegaard et al. (1985) as PTI = PI/PI ref, where PI ref is the PI value in a reference artery without any obstruction. For this study, contralateral MCAs were used as reference arteries, but only in cases without contralateral ICA occlusive disease. This was true for subgroups O-LN, H-LN, and L-LN. A total of 93 measurements met these criteria. Measurement of the CO2-dependent VMR. The COz-reactivity test was performed as described by Ringelstein et al. (1988). Mean MCA blood flow velocity during inhalation of room air was equated to 100% and was compared with the percentage changes in mean blood flow velocity during inhalation of various CO2 concentrations (i.e., 6, 5, 4, 3, 2% CO2 in oxygen), and during various intensities of hyperventilation. No attempt was made to convert relative blood flow velocity to absolute velocity values. The relative blood flow velocity changes were plotted against the corresponding end-tidal CO2 vol% (Fig. 1). A tangent-hyperbolic function gave the best fit for these data, and the distance between the two asymptotes was considered to reflect vasomotor reactivity (VMR). The normal values were found to be 86 + 17%. Comparison of pulsatility indices and VMR The authors were aware of the fact that pulsatility indices reflect one-point measurements at rest, whereas evaluation of V M R measures a dynamic response. Therefore, it seemed invalid to compare both parameters directly. However, various authors (Halsey et al. 1986; Schneider et al. 1988) have shown
Pulsatility indices and CO2-reactivity• J. LEY-Pozo et al.
200 V%
765
V~
VMR: 86.8%
..,,......° ~...-.~ ......'
........... • ....
156.8
•
°,
100 ...," ..,.o • •
70-
0
.......................
11
• ...............
[2
i..I,.
~i "''''l
13
14
15
16
'I VOW.% CO~
Fig. 1. Vasomotor reactivity (VMR) curve of a healthy subject. Ordinate: Changes of mean flow velocity within the middle cerebral artery during various capnic stimuli. Abscissa: End-tidal Vol. % CO2 during hyper- and normoventilation, as well as during inhalation of CO2. Curve fitting results in a bi-asymptotic S-shaped curve with an upper asymptote at 156.8 V% and a lower asymptote at 70 V%. The difference between these asymptotes reflects the hemispheric VMR.
one-point evaluation o f pulsatility to be useful as a rapid, simple and reliable test to identify patients with severe reduction in perfusion pressure, suggesting that the complex and uncomfortable CO2-reactivity m e a s u r e m e n t s would not be necessary. Also, during high-altitude exposure, a significant decrease in PI paralleled a d i m i n u t i o n or loss o f CO2-reactivity, again indicating a close relationship between these parameters (Otis et al. 1989). Thus, it seemed reasonable to c o m p a r e simultaneously pulsatility indices and V M R values in patients with extracranial occlusive disease in order to test both methods in terms of a p p l i c a b i l i t y a n d clinical usefulness. T h i s p a p e r would not address the question, whether the evaluation o f changes in pulsatility indices during various capnic stimuli would be m o r e informative than onepoint measurements, or would even be equivalent to V M R values.
Statistical analysis D e s c r i p t i v e statistics a n d regression analysis were c o m p u t e d using routines from the SPSS-X statistical package. C o m p a r i s o n o f m e a n s (Student's t two-tailed tests) was p e r f o r m e d between h e m o d y n a m i c parameters in clinical and angiological subgroups.
RESULTS The V M R and PI values o f all hemispheres were plotted against each other (Fig. 2a). The linear regression coefficient r was 0.56; it was thus significantly different from zero (p < 0.001), but not very close to + 1. This value was even lower if V M R and PTI were c o m p a r e d (r = 0.48), but was still significantly different from zero (p < 0.001) (Fig. 2b). In individual cases, however, neither PI n o r P T I could predict V M R values with satisfying precision. The so-called "residuals," i.e., the actually measured V M R values m i n u s the values predicted as a function of PI, were calculated for each hemisphere. These differences were larger than 15% in 185 hemispheres, i.e., 47% of the total n u m b e r of cases, and reached even m o r e than 25% in 23% o f the measurements. W h e n V M R a n d PI values o f the clinically s y m p t o m a t i c a n d a s y m p t o m a t i c s u b g r o u p s were c o m p a r e d separately, most of the correlation coefficients (r) were significantly different from zero, but not close to +1 (Table 1). After ipsilateral carotid endarterectomy, there was no correlation between these parameters. T h e m e a n "residuals" and their standard deviations within each clinical subgroup are also listed in Table 1. In patients with p e r m a n e n t and transient deficits, a slight tendency for overestimation
766
Ultrasound in Medicine and Biology
Volume 16, Number 8, 1990
VMR 140
%
100 ,,
::
,. ,-- ~ , r • . , . ~ , . . "
_
80, •"
""
•
..2~--"=
,~..._.
•
•
• .q,
, •
•
.•
60 "! ..
"iv',.-
=
-
"-
•
•
=.~
•
40 •m
20
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1.8
2.0
PI
(a) VMR 140 ~,
% 0 o
o
o~
[]
[]
[]
100 o
o
0
80E
I:1
Q
0 0 0 0n [] 0 ] Oo~
(I
o
[]
0
0[3
o 0
o O~ 0 o
[]
60
Q
~
o
o
o
B
o
o o
40
o
o
B
[] °~o
o
o
0
o o
20
o
o o
0
Lx
0
0.2
~
0.4
z~
0.6
A
/x
~x
Lx
ts
0.8
1.0
1.2
1.4
1.6
PT
I
(b) Fig. 2. Correlation between pulsatility indices (PI, PTI) and vasomotor reactivity (VMR). (a) Comparison of P! and VMR of all hemispheres (N = 392) (r = 0.56, different from zero with p < 0.001). The regression line can be described by VMR = 58.6 PI + 21.4. (b) Comparison of PTI and VMR of all hemispheres (N = 93) (r = 0.48, different from zero with p < 0.001). The regression line can be described by VMR = 57.6 PTI + 19.7.
o f V M R as a function o f PI b e c a m e apparent. T h e o p p o s i t e was true for a s y m p t o m a t i c subjects or those after e n d a r t e r e c t o m y . T h e c o r r e s p o n d i n g scatter diag r a m s o f V M R vs. PI are s h o w n in Figs. 3 a - 3 d . W h e n the s a m e analysis was d o n e for subgroups
with different degrees of extracranial carotid disease,
the relationship b e t w e e n V M R and PI was not very strong either, the m e a n "residuals" were low, and m o s t o f the c o r r e l a t i o n coefficients w e r e different f r o m zero (Table 2). T h e m e a n s and standard deviations o f V M R , PI, and PTI for each clinical subgroup are given in Table
Pulsatility indices and CO2-reactivity• J. LEY-Pozoet aL
767
Table 1. Correlation coefficients (r) between VMR and P! or PTI, and "residuals" (means and standard deviations) for clinical subgroups. V M R vs. PI
V M R vs. PTI
Residuals
Subgroup
n
r
p
n
r
p
x
SD
Stroke TIA Asym. PO
81 81 187 43
0.548 0.617 0.412 0.142
0.001 0.001 0.001 n.s.
31 28 10 24
0.405 0.751 0.721 -0.193
0.01 0.001 0.01 n.s.
- 10.5 -6.5 5.5 8.0
19.5 19.8 19.3 19.2
n: n u m b e r of MCA territories analyzed; p: level of significance of " r " different from zero; x: mean; SD: standard deviation; TIA: transient ischemic attack; Asym.: asymptomatic patients; PO: postoperative patients, i.e., after carotid endarterectomy.
3. While VMR and PI separated these subgroups equally well from each other, PTI did not show any significant difference between neighbouring subgroups. Both VMR and PI increased from subgroup to subgroup with improving clinical findings. The means and standard deviations of VMR, PI and PTI for angiological subgroups, representing different degrees of extracranial carotid disease, are shown in Table 4. The best discrimination of these subgroups was clearly obtained by VMR (Fig. 4a), whereas PI was only able to discriminate patients with occlusions or high-grade stenosis from those with low-grade lesions or without stenoses (Fig. 4b). No statistically significant differences were found between subgroups for mean PTI values. DISCUSSION Patients with hemodynamically relevant proximal arterial stenosis reveal dampened velocity waveforms downstream to the lesion. This can numerically be expressed by a reduced PI at this site (Evans et al. 1980; Gosling and King 1974; Johnston et al. 1983). At the brain arteries, these changes reflect a reduction of the cerebral perfusion pressure. Previous studies have also demonstrated a lower hemispherical VMR in such individuals (Ringelstein et al. 1988; Schneider et al. 1988). Thus, from a theoretical point of view, a close statistical relationship between VMR and PI could be expected. Our results, however, only partially support this hypothesis. Actually, the correlation coefficients between VMR and PI were significantly different from zero, and the scatter-plots showed a good fit for this data with the linear regression model. Nevertheless, this relationship was not strong enough to accurately predict the individual value of the more complex VMR parameter by exclusively measuring pulsatility indices at rest. The analysis of the "residuals" also indicated relatively large differences between the actual values of VMR and the values predicted by the regression function. If the patient is not able to perform the CO2-inhalation test,
however, measurement of the PI at rest may give only some rough information about the actual vasomotor reactivity within the hemisphere under study. Despite a considerable overlap of VMR and PI values between neighbouring clinical subgroups, there was a clear tendency to a stepwise augmentation of both parameters with an improving clinical condition, i.e., from the stroke subgroup to the asymptomatic patients. The same finding has been reported previously in patients with ICA occlusion (Ringelstein et al. 1988; Schneider et al. 1988). These findings also underline again that different degrees of functional impairment of the cerebral circulation can be found in angiologically homogeneous subgroups. The collateral vessel capacity, particularly the configuration of the circle of Willis, varies from one individual to the next (Nornes 1973; Riggs and Rupp 1963). This makes individual measurements of the VMR essential to know accurately the pathophysiological status of the patient. The identification of stroke-prone individuals based on the detection of an exhausted cerebrovascular reserve is clinically important (Norrving et al. 1982). These patients represent true cases of cerebrovascular insufficiency in the strict sense that cerebral blood flow cannot be sufficiently maintained even with recruitment of all available mechanisms for compensation. Such patients may benefit from therapeutic measures to improve large vessel flow, but no improvement may be expected from any form of anticoagulation treatment. During this study, a decrease in VMR, PI, and PTI could be found if the intracranial blood supply was impaired due to extracranial occlusive lesions. So far, our results are in line with those reported by other authors (Bullock et al. 1985; Bishop et al. 1986; Widder et al. 1986)+ VMR as a functional parameter, however, was by far superior to the static values of both pulsatility indices in separating patients with severe, moderate, mild or no obstructions of the neck arteries.
768
Ultrasound in Medicine and Biology
Volume 16, Number 8, 1990
VMR 140~ 0 0
%
0
(ID
0
O0 0
0
0 0
0
°°°o°~o°oO
100
0
o
oo
~
o Ooo°_....---"~o o
0
O0
80
0
, ~ ~ O o 0
60
0 0 " ~ 0 ~
o o o ~ % °0
o
o
0
0
o~
0 ~o ~
Ooo
o
0
o 01~ 0
0 °°
o
0 0
° 0
0
0
00
-oo-
o
0
0
0
0
40~
0 0 0
0
0
0
20
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1.2
1.4
1.6
1.8
2.0
PI
(a)
VMR 140 %
100
80
60
...J.
40
-'..-. • ".-
.
20
0
~
0
0.2
0.4
0.6
0.8
1.0 PI
(b) Fig. 3. Correlation between pulsatility index (PI) and v a s o m o t o r reactivity ( V M R ) in clinical subgroups. (a) A s y m p t o m a t i c patients ( N = 187 hemispheres) (r = 0.41, different from zero with p < 0.001). The regression line can be described by V M R = 38.9 PI + 45.2. (b) Patients with transient ischemic attacks ( N = 81 hemispheres) (r = 0.62, different from zero with p < 0.001). The regression line can be described by V M R = 61.5 PI + 12.4. (c) Patients with p e r m a n e n t deficits ( N = 81 hemispheres) (r = 0.55, different from zero with p < 0.001). The regression line m a y be described by V M R = 59.4 PI + 10.3. (d) Patients after carotid endarterectomy ( N = 43 patients/hemispheres) (r = 0.14, not different from zero). T h e regression line m a y be described by V M R = 12.2 PI + 73.8.
Pulsatility indices and CO:-reactivity • J. LEY-POzO et al.
769
VMR 140 %
100
.'.
80'
60
,,
•
•
•
40
•
• •
•
•
~• i ' ~ , "
20
•
•
,,,,
•
mo
nn
•• •
" "
•
"
I
nun
0
z~
0
0.2
z~
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1.4
1.6
1.8
2.0
PI (c)
VMR 140t % o
[]
o o o°o
100
o
o
o
o
o
oq6o
80
o
o
[]
Oo o
[3o []
o
60
D
o
o
o
40
20
A
0
0.2
0.4
0.6
0.8
1.0
1.2
PI
(d)
Fig. 3. (Contd)
With respect to clinical subgroups, both V M R and PI were able to discriminate stroke, TIA and asymptomatic patients. By contrast, PTI turned out to be useless in this context. Probably, because of high standard deviations, the changes in PTI were not statistically significant between neighbouring groups.
This may in part be explained by the vast variability in the configuration of the circle of Willis. Furthermore, because most of the patients have bilateral lesions, it is sometimes impossible to find a normal "reference" artery, and therefore, this parameter is not always measurable in clinical practice.
770
Ultrasound in Medicine and Biology
Volume 16, N u m b e r 8, 1990
T a b l e 2. C o r r e l a t i o n coefficients (r) b e t w e e n V M R a n d PI o r P T I , a n d " r e s i d u a l s " ( m e a n s a n d s t a n d a r d d e v i a t i o n s ) for angiological s u b g r o u p s . VMR vs. PI
VMR vs. PTI
Residuals
Subgroup
n
r
p
n
r
p
x
SD
O-O O-H O-LN H-H H-LN L-H L-LN N-X
44 17 57 25 44 23 27 112
0.506 0.411 0.529 0.506 0.654 0.053 0.331 0.290
0.001 0.05 0.001 0.01 0.001 n.s. 0.05 0.01
--31 -28 -10 --
* * 0.427 * 0.591 * 0.203 *
--0.01 -0.01 -n.s. --
- 17.1 - 10.1 -7.5 -9.2 -3.6 3.7 15.5 9.5
17.5 24.0 17.5 17.6 16.3 23.0 14.9 17.2
n: n u m b e r of MCA territories analyzed; p: level of significance of " r " different from zero; x: mean; SD: standard deviation; *: not calculated (for explanation, see text).
T a b l e 3. M e a n s (x) a n d s t a n d a r d d e v i a t i o n s ( S D ) o f V M R (in %), PI, a n d P T I for e a c h clinical g r o u p (for n-values, see T a b l e 1). VMR
Subgroup
x
p
Stroke
52.9
PI
SD
x
p
23.6
0.72
0.005 TIA
64.6
25.2
81.4
0.85
20.7
85.4
x
p
0.22
0.82
0.25
0.93
0.17
0.82
0.16 n.s.
0.22
0.94
0.32
n.s,
17.0
0.96
SD
n.s.
0.005
n.s.
PO
SD
0.005
0.005 Asym.
PTI
R.S.
0.20
0.96
0.19
p: level of significance in the comparison between means of neighbouring subgroups; TIA: transient ischemic attack; Asym.: asymptomatic patients; PO: patients after carotid endarterectomy.
T a b l e 4. V M R (in %), PI, a n d P T I w i t h r e f e r e n c e to angiological s u b g r o u p s . VMR
Subgroup O-O
x (SD)
PI
p
43.6 (20.2)
x (SD)
55.2 (26.5)
H-H
58.1 (19.6)
H-LN
67.2 (21.6)
L-H
80.0 (20.0)
L-LN
95.2 (12.6)
0.79 (0.25)
86.7 (16.4)
0.78 (0.15) n.s. n.s. n.s.
0.84 (0.24) 0.025 0.005 0.005
1.06 (0.22)
*
* 1.07 (0.22)
n.s. 0.95 (0.20)
n.s.
0.81 (0.17) 0.005 0.005 n.s.
0.99 (0.22) 0.02
N-X
* n.s.
0.76 (0.22) n.s. 0.05 0.05
p
*
0.75 (0.16)
58.8 (22.2)
x (SD)
n.s.
n.s. O-LN
p
0.67 (0.20) 0.05
O-H
PTI
*
p: level of significance in the comparison between means of neighbouring subgroups; *: not calculated.
n.s.
Pulsatility indices and CO2-reactivity• J. LEY-POZOet al.
771
VMR 120,
T
T
L-H
L-IN
%
T
80
60
40
20
0-0
O-H
V p
0301-5629/90 $3.00 + .00 © 1990 Pergamon Press plc
OOriginal Contribution RELATIONSHIP BETWEEN PULSATILITY FLOW SIGNALS AND CO2-REACTIVITY CEREBRAL ARTERY IN EXTRACRANIAL
INDICES OF DOPPLER WITHIN THE MIDDLE OCCLUSIVE DISEASE
JUAN LEY-Pozo,* KLAUS WILLMES and ERICH BERND RINGELSTEIN Department of Neurology, University Hospital, Klinikum RWTH, D-5100 Aachen, Germany (Received 29 August 1989; in final form 11 April 1990) Abstract--CO2-dependent vasomotor reactivity (VMR) of the middle cerebral artery (MCA) distribution has been shown to be a reliable predictor of the patient's risk for low-flow strokes. In order to clarify whether various Pulsatility Indices (PIs) of MCA flow signals would be useful to estimate VMR, 222 subjects with various degrees of stenosing internal carotid artery lesions were studied. V M R of the MCA territory was expressed as the percentage changes in mean blood flow velocity during inhalation of CO2 and hyperventilation. Although the correlation coefficients between V M R and PIs were significantly different from zero, these relationships were not strong enough for PI values to accurately predict the individual values of VMR. Furthermore, V M R could convincingly discriminate subgroups of patients with different severities of extracranial occlusive disease, whereas PIs discriminated only weakly or not at all. PIs may give some vague idea of the vasomotor tone, but cannot predict precisely enough whether a patient is at risk for low-flow infarctions.
Key Words: Transcranial Doppler sonography, Pulsatility index, Vasomotor reserve, CO2-reactivity, Cerebrovascular disease.
ity) o f Gosling and King (1974), and more recently, the Pulsatility Transmission Index (PTI) (see below) of Lindegaard et al. (1985) have been recommended as easily obtainable parameters reflecting both the intracerebral peripheral vascular resistance and vasom o t o r responcivity. Their usefulness in clinical practice, however, has not been proven so far. The purpose of this paper was to test whether PI and PTI are precise and valid enough to replace CO2-dependent V M R measurements and to clarify to what extent these indices would be able to discriminate subgroups o f patients with various degrees of extracranial cerebrovascular occlusive disease resulting from its impact on intracranial blood flow.
INTRODUCTION In 1982, Aaslid et al. i n t r o d u c e d T r a n s c r a n i a l Doppler Sonography (TCD) into clinical practice for noninvasive investigation o f the large intracranial arteries. This m e t h o d permits the assessment of cerebral vasomotor reactivity (VMR) upon vascular stimuli, such as pCO2 d u r i n g hyper- and h y p o c a p n i a (Markwalder et al. 1984; Aaslid 1986; Widder et al. 1986; Ringelstein et al. 1988). Although the clinical usefulness of CO2 tests has been well documented (Ringelstein et al. 1988; Widder et al. 1987), they are time-consuming, and often uncomfortable for the patient. Simple n u m e r i c a l analysis o f the shape o f transcranial Doppler flow signals also gives information concerning the functional state o f the cerebral circulation (Aaslid and Lindegaard 1986). The Pulsatility Index (PI = peak-to-peak velocity/mean veloc-
M A T E R I A L S AND M E T H O D S Recruitment o f patients Between September 1987 and December 1988, 222 consecutive patients with stroke a n d / o r cerebrovascular disease were investigated. Both T C D studies o f the basal cerebral arteries and continuous-wave (CW) Doppler examinations of the extracranial brain arteries were performed. V M R values and PIs were evaluated not earlier than six weeks after a vascular
Address correspondenceto: E. Bernd Ringelstein, M.D., Associate Professorof Neurology,Department of Neurology,University Hospital, Klinikum RWTH, Pauwelsstrasse, D-5100 Aachen, Germany. t Dr. Ley-Pozois a research fellowfrom Cuba supported by a grant from the Alexander von Humboldt Foundation, Bonn/FRG (Grant No. IV 1-71147). 763
764
Ultrasound in Medicine and Biology
event. Mean age was 61.3 +_ 33 years. One hundred and twenty-three patients were males and 98 were females. In total, 392 "hemispheres," i.e., middle cerebral arteries (MCA) were considered. Measurements were not possible in 52 MCA territories due either to lack of an "ultrasound" window within the temporal bone, poor signal-to-noise ratio, or uncooperative patients.
Clinical subgroups From a clinical point o f view, the cerebral hemispheres u n d e r study were divided into four subgroups: (I) permanent neurological deficit (N = 81), (II) transient neurological deficit (N = 81), (III) asymptomatic hemispheres (N = 187), and (IV) postoperative state after carotid endarterectomy (N = 43) (all of the latter had had previous TIAs).
A ngiological subgroups On the basis of CW Doppler findings, both internal carotid arteries (ICA) were classified as either (1) " O c c l u d e d " ( " O " ) , (2) "High-grade stenosis" ( " H " ) (i.e., reduction in cross-sectional area o f more than or equal to 80%), (3) "Low-grade stenosis" ("L") (i.e., reduction in cross-sectional area o f less than 80%, but o f more than or equal to 60%), or (4) " N o stenosis" ( " N " ) (i.e., reduction in cross-sectional area o f less than 60%). The detailed methodology for CW Doppler examination o f the extracranial arteries has been described previously (Spencer and Reid 1979; Ringelstein 1989a). With respect to bilateral ICA findings, nine subgroups were considered: (1) bilaterally occluded ICA (indicated by O-O), (2) unilateral ICA occlusion with contralateral high-grade stenosis (O-H), (3) unilateral ICA occlusion with contralateral low-grade or no stenosis (O-LN), (4) bilateral ICA high-grade stenosis (H-H), (5) unilateral ICA highgrade stenosis with contralateral low-grade or no stenosis (H-LN), (6) unilateral low-grade ICA stenosis with contralateral high-grade ICA stenosis (L-H), (7) unilateral low-grade ICA stenosis with contralateral low-grade or no stenosis (L-LN), (8) ICA with no stenosis (irrespective of the situation of the contralateral side) (N-X), and (9) ICA after endarterectomy, at least two weeks after successful carotid recanalization (this group coincides with the clinical subgroup IV). It was reasonable to differentiate subgroups (3) and (5) from (6) and (8), since V M R and PI/PTI values always referred to the side mentioned first.
Transcranial Doppler sonography The transtemporal identification and isonation of the MCA with a T C D Doppler device (TC 2-64, EME, Ueberlingen, FRG) was based on an anterior
Volume16, Number 8, 1990 angulation of the beam, an insonation depth between 45 and 50 mm, the blood flow being directed towards the probe, and traceability of the vessel from 40-55 m m insonation depth (for details, see Aaslid 1986; Ringelstein 1989b).
Measurement and calculation of PI. PI has been defined as the ratio between th e peak-to-peak velocity and mean velocity (Gosling and King 1974). Mean flow velocity values as originally defined by Aaslid et al. (1982) were automatically calculated by the instrument over a period varying from 3-10 s depending on the sweep speed used. T C D Doppler flow signals o f both MCAs were registered with each subject resting comfortably in a supine position. Velocity spectra were recorded for 20 s during steady-state and were stored on tape. Traces with obvious artifacts were deleted. PI values were measured of the residual cardiac cycles. The mean of these PI values was calculated off-line. Calculation of PTI. T h e PTI was defined by Lindegaard et al. (1985) as PTI = PI/PI ref, where PI ref is the PI value in a reference artery without any obstruction. For this study, contralateral MCAs were used as reference arteries, but only in cases without contralateral ICA occlusive disease. This was true for subgroups O-LN, H-LN, and L-LN. A total of 93 measurements met these criteria. Measurement of the CO2-dependent VMR. The COz-reactivity test was performed as described by Ringelstein et al. (1988). Mean MCA blood flow velocity during inhalation of room air was equated to 100% and was compared with the percentage changes in mean blood flow velocity during inhalation of various CO2 concentrations (i.e., 6, 5, 4, 3, 2% CO2 in oxygen), and during various intensities of hyperventilation. No attempt was made to convert relative blood flow velocity to absolute velocity values. The relative blood flow velocity changes were plotted against the corresponding end-tidal CO2 vol% (Fig. 1). A tangent-hyperbolic function gave the best fit for these data, and the distance between the two asymptotes was considered to reflect vasomotor reactivity (VMR). The normal values were found to be 86 + 17%. Comparison of pulsatility indices and VMR The authors were aware of the fact that pulsatility indices reflect one-point measurements at rest, whereas evaluation of V M R measures a dynamic response. Therefore, it seemed invalid to compare both parameters directly. However, various authors (Halsey et al. 1986; Schneider et al. 1988) have shown
Pulsatility indices and CO2-reactivity• J. LEY-Pozo et al.
200 V%
765
V~
VMR: 86.8%
..,,......° ~...-.~ ......'
........... • ....
156.8
•
°,
100 ...," ..,.o • •
70-
0
.......................
11
• ...............
[2
i..I,.
~i "''''l
13
14
15
16
'I VOW.% CO~
Fig. 1. Vasomotor reactivity (VMR) curve of a healthy subject. Ordinate: Changes of mean flow velocity within the middle cerebral artery during various capnic stimuli. Abscissa: End-tidal Vol. % CO2 during hyper- and normoventilation, as well as during inhalation of CO2. Curve fitting results in a bi-asymptotic S-shaped curve with an upper asymptote at 156.8 V% and a lower asymptote at 70 V%. The difference between these asymptotes reflects the hemispheric VMR.
one-point evaluation o f pulsatility to be useful as a rapid, simple and reliable test to identify patients with severe reduction in perfusion pressure, suggesting that the complex and uncomfortable CO2-reactivity m e a s u r e m e n t s would not be necessary. Also, during high-altitude exposure, a significant decrease in PI paralleled a d i m i n u t i o n or loss o f CO2-reactivity, again indicating a close relationship between these parameters (Otis et al. 1989). Thus, it seemed reasonable to c o m p a r e simultaneously pulsatility indices and V M R values in patients with extracranial occlusive disease in order to test both methods in terms of a p p l i c a b i l i t y a n d clinical usefulness. T h i s p a p e r would not address the question, whether the evaluation o f changes in pulsatility indices during various capnic stimuli would be m o r e informative than onepoint measurements, or would even be equivalent to V M R values.
Statistical analysis D e s c r i p t i v e statistics a n d regression analysis were c o m p u t e d using routines from the SPSS-X statistical package. C o m p a r i s o n o f m e a n s (Student's t two-tailed tests) was p e r f o r m e d between h e m o d y n a m i c parameters in clinical and angiological subgroups.
RESULTS The V M R and PI values o f all hemispheres were plotted against each other (Fig. 2a). The linear regression coefficient r was 0.56; it was thus significantly different from zero (p < 0.001), but not very close to + 1. This value was even lower if V M R and PTI were c o m p a r e d (r = 0.48), but was still significantly different from zero (p < 0.001) (Fig. 2b). In individual cases, however, neither PI n o r P T I could predict V M R values with satisfying precision. The so-called "residuals," i.e., the actually measured V M R values m i n u s the values predicted as a function of PI, were calculated for each hemisphere. These differences were larger than 15% in 185 hemispheres, i.e., 47% of the total n u m b e r of cases, and reached even m o r e than 25% in 23% o f the measurements. W h e n V M R a n d PI values o f the clinically s y m p t o m a t i c a n d a s y m p t o m a t i c s u b g r o u p s were c o m p a r e d separately, most of the correlation coefficients (r) were significantly different from zero, but not close to +1 (Table 1). After ipsilateral carotid endarterectomy, there was no correlation between these parameters. T h e m e a n "residuals" and their standard deviations within each clinical subgroup are also listed in Table 1. In patients with p e r m a n e n t and transient deficits, a slight tendency for overestimation
766
Ultrasound in Medicine and Biology
Volume 16, Number 8, 1990
VMR 140
%
100 ,,
::
,. ,-- ~ , r • . , . ~ , . . "
_
80, •"
""
•
..2~--"=
,~..._.
•
•
• .q,
, •
•
.•
60 "! ..
"iv',.-
=
-
"-
•
•
=.~
•
40 •m
20
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1.8
2.0
PI
(a) VMR 140 ~,
% 0 o
o
o~
[]
[]
[]
100 o
o
0
80E
I:1
Q
0 0 0 0n [] 0 ] Oo~
(I
o
[]
0
0[3
o 0
o O~ 0 o
[]
60
Q
~
o
o
o
B
o
o o
40
o
o
B
[] °~o
o
o
0
o o
20
o
o o
0
Lx
0
0.2
~
0.4
z~
0.6
A
/x
~x
Lx
ts
0.8
1.0
1.2
1.4
1.6
PT
I
(b) Fig. 2. Correlation between pulsatility indices (PI, PTI) and vasomotor reactivity (VMR). (a) Comparison of P! and VMR of all hemispheres (N = 392) (r = 0.56, different from zero with p < 0.001). The regression line can be described by VMR = 58.6 PI + 21.4. (b) Comparison of PTI and VMR of all hemispheres (N = 93) (r = 0.48, different from zero with p < 0.001). The regression line can be described by VMR = 57.6 PTI + 19.7.
o f V M R as a function o f PI b e c a m e apparent. T h e o p p o s i t e was true for a s y m p t o m a t i c subjects or those after e n d a r t e r e c t o m y . T h e c o r r e s p o n d i n g scatter diag r a m s o f V M R vs. PI are s h o w n in Figs. 3 a - 3 d . W h e n the s a m e analysis was d o n e for subgroups
with different degrees of extracranial carotid disease,
the relationship b e t w e e n V M R and PI was not very strong either, the m e a n "residuals" were low, and m o s t o f the c o r r e l a t i o n coefficients w e r e different f r o m zero (Table 2). T h e m e a n s and standard deviations o f V M R , PI, and PTI for each clinical subgroup are given in Table
Pulsatility indices and CO2-reactivity• J. LEY-Pozoet aL
767
Table 1. Correlation coefficients (r) between VMR and P! or PTI, and "residuals" (means and standard deviations) for clinical subgroups. V M R vs. PI
V M R vs. PTI
Residuals
Subgroup
n
r
p
n
r
p
x
SD
Stroke TIA Asym. PO
81 81 187 43
0.548 0.617 0.412 0.142
0.001 0.001 0.001 n.s.
31 28 10 24
0.405 0.751 0.721 -0.193
0.01 0.001 0.01 n.s.
- 10.5 -6.5 5.5 8.0
19.5 19.8 19.3 19.2
n: n u m b e r of MCA territories analyzed; p: level of significance of " r " different from zero; x: mean; SD: standard deviation; TIA: transient ischemic attack; Asym.: asymptomatic patients; PO: postoperative patients, i.e., after carotid endarterectomy.
3. While VMR and PI separated these subgroups equally well from each other, PTI did not show any significant difference between neighbouring subgroups. Both VMR and PI increased from subgroup to subgroup with improving clinical findings. The means and standard deviations of VMR, PI and PTI for angiological subgroups, representing different degrees of extracranial carotid disease, are shown in Table 4. The best discrimination of these subgroups was clearly obtained by VMR (Fig. 4a), whereas PI was only able to discriminate patients with occlusions or high-grade stenosis from those with low-grade lesions or without stenoses (Fig. 4b). No statistically significant differences were found between subgroups for mean PTI values. DISCUSSION Patients with hemodynamically relevant proximal arterial stenosis reveal dampened velocity waveforms downstream to the lesion. This can numerically be expressed by a reduced PI at this site (Evans et al. 1980; Gosling and King 1974; Johnston et al. 1983). At the brain arteries, these changes reflect a reduction of the cerebral perfusion pressure. Previous studies have also demonstrated a lower hemispherical VMR in such individuals (Ringelstein et al. 1988; Schneider et al. 1988). Thus, from a theoretical point of view, a close statistical relationship between VMR and PI could be expected. Our results, however, only partially support this hypothesis. Actually, the correlation coefficients between VMR and PI were significantly different from zero, and the scatter-plots showed a good fit for this data with the linear regression model. Nevertheless, this relationship was not strong enough to accurately predict the individual value of the more complex VMR parameter by exclusively measuring pulsatility indices at rest. The analysis of the "residuals" also indicated relatively large differences between the actual values of VMR and the values predicted by the regression function. If the patient is not able to perform the CO2-inhalation test,
however, measurement of the PI at rest may give only some rough information about the actual vasomotor reactivity within the hemisphere under study. Despite a considerable overlap of VMR and PI values between neighbouring clinical subgroups, there was a clear tendency to a stepwise augmentation of both parameters with an improving clinical condition, i.e., from the stroke subgroup to the asymptomatic patients. The same finding has been reported previously in patients with ICA occlusion (Ringelstein et al. 1988; Schneider et al. 1988). These findings also underline again that different degrees of functional impairment of the cerebral circulation can be found in angiologically homogeneous subgroups. The collateral vessel capacity, particularly the configuration of the circle of Willis, varies from one individual to the next (Nornes 1973; Riggs and Rupp 1963). This makes individual measurements of the VMR essential to know accurately the pathophysiological status of the patient. The identification of stroke-prone individuals based on the detection of an exhausted cerebrovascular reserve is clinically important (Norrving et al. 1982). These patients represent true cases of cerebrovascular insufficiency in the strict sense that cerebral blood flow cannot be sufficiently maintained even with recruitment of all available mechanisms for compensation. Such patients may benefit from therapeutic measures to improve large vessel flow, but no improvement may be expected from any form of anticoagulation treatment. During this study, a decrease in VMR, PI, and PTI could be found if the intracranial blood supply was impaired due to extracranial occlusive lesions. So far, our results are in line with those reported by other authors (Bullock et al. 1985; Bishop et al. 1986; Widder et al. 1986)+ VMR as a functional parameter, however, was by far superior to the static values of both pulsatility indices in separating patients with severe, moderate, mild or no obstructions of the neck arteries.
768
Ultrasound in Medicine and Biology
Volume 16, Number 8, 1990
VMR 140~ 0 0
%
0
(ID
0
O0 0
0
0 0
0
°°°o°~o°oO
100
0
o
oo
~
o Ooo°_....---"~o o
0
O0
80
0
, ~ ~ O o 0
60
0 0 " ~ 0 ~
o o o ~ % °0
o
o
0
0
o~
0 ~o ~
Ooo
o
0
o 01~ 0
0 °°
o
0 0
° 0
0
0
00
-oo-
o
0
0
0
0
40~
0 0 0
0
0
0
20
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1.2
1.4
1.6
1.8
2.0
PI
(a)
VMR 140 %
100
80
60
...J.
40
-'..-. • ".-
.
20
0
~
0
0.2
0.4
0.6
0.8
1.0 PI
(b) Fig. 3. Correlation between pulsatility index (PI) and v a s o m o t o r reactivity ( V M R ) in clinical subgroups. (a) A s y m p t o m a t i c patients ( N = 187 hemispheres) (r = 0.41, different from zero with p < 0.001). The regression line can be described by V M R = 38.9 PI + 45.2. (b) Patients with transient ischemic attacks ( N = 81 hemispheres) (r = 0.62, different from zero with p < 0.001). The regression line can be described by V M R = 61.5 PI + 12.4. (c) Patients with p e r m a n e n t deficits ( N = 81 hemispheres) (r = 0.55, different from zero with p < 0.001). The regression line m a y be described by V M R = 59.4 PI + 10.3. (d) Patients after carotid endarterectomy ( N = 43 patients/hemispheres) (r = 0.14, not different from zero). T h e regression line m a y be described by V M R = 12.2 PI + 73.8.
Pulsatility indices and CO:-reactivity • J. LEY-POzO et al.
769
VMR 140 %
100
.'.
80'
60
,,
•
•
•
40
•
• •
•
•
~• i ' ~ , "
20
•
•
,,,,
•
mo
nn
•• •
" "
•
"
I
nun
0
z~
0
0.2
z~
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
1.4
1.6
1.8
2.0
PI (c)
VMR 140t % o
[]
o o o°o
100
o
o
o
o
o
oq6o
80
o
o
[]
Oo o
[3o []
o
60
D
o
o
o
40
20
A
0
0.2
0.4
0.6
0.8
1.0
1.2
PI
(d)
Fig. 3. (Contd)
With respect to clinical subgroups, both V M R and PI were able to discriminate stroke, TIA and asymptomatic patients. By contrast, PTI turned out to be useless in this context. Probably, because of high standard deviations, the changes in PTI were not statistically significant between neighbouring groups.
This may in part be explained by the vast variability in the configuration of the circle of Willis. Furthermore, because most of the patients have bilateral lesions, it is sometimes impossible to find a normal "reference" artery, and therefore, this parameter is not always measurable in clinical practice.
770
Ultrasound in Medicine and Biology
Volume 16, N u m b e r 8, 1990
T a b l e 2. C o r r e l a t i o n coefficients (r) b e t w e e n V M R a n d PI o r P T I , a n d " r e s i d u a l s " ( m e a n s a n d s t a n d a r d d e v i a t i o n s ) for angiological s u b g r o u p s . VMR vs. PI
VMR vs. PTI
Residuals
Subgroup
n
r
p
n
r
p
x
SD
O-O O-H O-LN H-H H-LN L-H L-LN N-X
44 17 57 25 44 23 27 112
0.506 0.411 0.529 0.506 0.654 0.053 0.331 0.290
0.001 0.05 0.001 0.01 0.001 n.s. 0.05 0.01
--31 -28 -10 --
* * 0.427 * 0.591 * 0.203 *
--0.01 -0.01 -n.s. --
- 17.1 - 10.1 -7.5 -9.2 -3.6 3.7 15.5 9.5
17.5 24.0 17.5 17.6 16.3 23.0 14.9 17.2
n: n u m b e r of MCA territories analyzed; p: level of significance of " r " different from zero; x: mean; SD: standard deviation; *: not calculated (for explanation, see text).
T a b l e 3. M e a n s (x) a n d s t a n d a r d d e v i a t i o n s ( S D ) o f V M R (in %), PI, a n d P T I for e a c h clinical g r o u p (for n-values, see T a b l e 1). VMR
Subgroup
x
p
Stroke
52.9
PI
SD
x
p
23.6
0.72
0.005 TIA
64.6
25.2
81.4
0.85
20.7
85.4
x
p
0.22
0.82
0.25
0.93
0.17
0.82
0.16 n.s.
0.22
0.94
0.32
n.s,
17.0
0.96
SD
n.s.
0.005
n.s.
PO
SD
0.005
0.005 Asym.
PTI
R.S.
0.20
0.96
0.19
p: level of significance in the comparison between means of neighbouring subgroups; TIA: transient ischemic attack; Asym.: asymptomatic patients; PO: patients after carotid endarterectomy.
T a b l e 4. V M R (in %), PI, a n d P T I w i t h r e f e r e n c e to angiological s u b g r o u p s . VMR
Subgroup O-O
x (SD)
PI
p
43.6 (20.2)
x (SD)
55.2 (26.5)
H-H
58.1 (19.6)
H-LN
67.2 (21.6)
L-H
80.0 (20.0)
L-LN
95.2 (12.6)
0.79 (0.25)
86.7 (16.4)
0.78 (0.15) n.s. n.s. n.s.
0.84 (0.24) 0.025 0.005 0.005
1.06 (0.22)
*
* 1.07 (0.22)
n.s. 0.95 (0.20)
n.s.
0.81 (0.17) 0.005 0.005 n.s.
0.99 (0.22) 0.02
N-X
* n.s.
0.76 (0.22) n.s. 0.05 0.05
p
*
0.75 (0.16)
58.8 (22.2)
x (SD)
n.s.
n.s. O-LN
p
0.67 (0.20) 0.05
O-H
PTI
*
p: level of significance in the comparison between means of neighbouring subgroups; *: not calculated.
n.s.
Pulsatility indices and CO2-reactivity• J. LEY-POZOet al.
771
VMR 120,
T
T
L-H
L-IN
%
T
80
60
40
20
0-0
O-H
V p
