INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE VALSALVA MANEUVER (VM) is a widely used procedure for measuring the integrity of the autonomic nervous system and the noninvasive assessment of baroreceptor sensitivity (13, 18, 22). More recently, with the advent of transcranial Doppler ultrasonography (TCD) (1), which has a good temporal resolution compared with the more traditional techniques for assessing cerebral blood flow (CBF), the changes in cerebral hemodynamics with this maneuver have been examined, with the possibility of refining this technique for the noninvasive assessment of cerebrovascular reactivity and cerebral autoregulation (CA) (12, 23).

The VM is usually achieved by raising intrathoracic pressure to 40 mmHg for ~20 s, as described by Palmero et al. (18) and Smith et al. (22). There are four recognized phases to the maneuver: phase I is associated with the beginning of the strain and a transient rise in mean arterial blood pressure (MABP) as the increase in intrathoracic pressure is transmitted to the arterial tree; during phase IIa, the atrial filling pressure falls so MABP decreases; in phase IIb there is increased sympathetic activation, causing a rise in peripheral vascular resistance, which leads to a small increase in MABP and heart rate; phase III is associated with release of the strain and a sudden fall in MABP due to the release of the intrathoracic pressure influence on the arterial tree; and, finally, phase IV sees an immediate "overshoot" in MABP because of the persistence of increased sympathetic tone and systemic vascular resistance. A reflex bradycardia then results due to stimulation of arterial baroreceptors, and both MABP and heart rate return to baseline values. If CA is the process regulating CBF during phase IV of the maneuver, then the rapid rise in MABP should be followed by a rapid rise in CBF, which is quickly returned to baseline values by dynamic CA, before MABP returns to normal.

Greenfield et al. (12) examined nine male subjects (demographics unknown) 10-20 days after neurosurgery for resection of malignant cerebral tumors. Cerebrospinal fluid pressure (CSFP), internal carotid artery blood flow (_ICA), and intra-arterial MABP were measured during the VM (20 mmHg for 15 s). The changes detected were as follows: during phase I, MABP increased by 17 mmHg, CSFP by 7 mmHg, and _ICA by 23 cm³/min; in phase II, MABP fell 28 mmHg, CSFP rose a further 3.5 mmHg, and _ICA fell 21% below baseline; during phase III MABP fell more, CSFP rose to baseline values, and _ICA increased to 16% below baseline; and, finally, during phase IV MABP increased to 18% above baseline, with CSFP at 21% and _ICA at 22% above baseline. Cerebrovascular resistance (CVR) was calculated to assess CA from the equation As expected, resistance fell during phases II and III, then increased in phase IV, implying that the autoregulatory mechanism was intact, but this change was considered to be both slow, compared with the MABP change, and also not of sufficient magnitude to return CBF to normal during phase II, accounting for the fall in cerebral perfusion pressure.

Tiecks et al. (23), studying normal volunteers (mean age 30 ± 9 yr), measured middle cerebral artery (MCA) cerebral blood flow volume (CBFV) by using TCD, CBFV being regarded as a valid surrogate marker of CBF (2). Similar results were found with a fall in MABP of 10.2% and in CBF of 35% in phase II and a rise of 29.6 and 56.5%, respectively, during phase IV. CA during phase IV was calculated by using the formula Again, it was thought that CA was occurring and that the larger overshoot in CBFV occurred because of the time delay in the onset of CA.

Because of this disparity in the magnitude of changes between mean cerebral blood flow volume (MCBFV) and MABP, and the temporal pattern of events seen during the VM, our goal was to find a more physiological explanation for these results by examining the temporal evolution of instantaneous pressure-velocity relationships throughout the VM. We studied a group of volunteers with a wide age range because this technique may find a clinical application in subjects with, e.g., cerebrovascular disease.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Subjects and study design. We examined 53 healthy volunteers (29 men), mean age 66 yr (range 31-80 yr), mean blood pressure 134/79 mmHg (range 108-164/63-108 mmHg), and mean body mass index 26.2 kg/m² (range 20-35).

Data analysis. Off-line recordings of CBFV and ABP signals were continuously transferred to a microcomputer at a rate of 200 samples · s¹ · channel¹. These signals were low-pass filtered with a cutoff frequency of 20 Hz, and the beginning and end of each cardiac cycle were detected from the ABP waveform.

(1)

View larger version (31K): [in this window] [in a new window]   Fig. 1.   A: representative recording of arterial blood pressure (ABP) and cerebral blood flow volume (CBFV) waveforms during Valsalva maneuver (VM). B: pressure-velocity (P-V) relationships for a single cardiac cycle during phases II and IV. C: corresponding temporal changes in critical closing pressure (CCP) throughout VM estimated by beat-to-beat extrapolation of linear regression of pressure-velocity curves.

(2)

(3)

(4)

	RESULTS

Top Abstract Introduction Methods Results Discussion References

A representative recording of ABP and CBFV changes during the VM is given in Fig. 1A, and instantaneous pressure-velocity relationships for a single cardiac cycle in phases II and IV are represented in Fig. 1B. The pressure-axis intercept of the linear regression that is used to estimate the CCP shows a clear shift between the two phases; the complete evolution of the CCP during the VM is depicted in Fig. 1C.

The increased CCP during phases I and II, and its sudden fall after phase IIb, as observed in Fig. 1C, could be observed in most subjects. In a small number of cases the linear regression extrapolation becomes less reliable as the result of noise in the pressure-velocity relationship (Fig. 1B). Nevertheless, this temporal pattern of CCP during the VM was confirmed for the whole group of subjects studied by the average results displayed in Fig. 2, A-D, for the right MCA. Similar results were obtained for the left MCA (Table 1), although the CCP during phase IV tends to be lower than for the right MCA.

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Population mean values of ABP (A), CBFV (B), CCP (C), and resistance-area product (RAP; D) throughout VM and recovery period. D: solid line, RAP that considers CCP (RAPC); dotted line, alternative estimate that neglects contribution of CCP (RAPZ). t = 0, Start of phase IV of VM; n = 53 subjects.

The changes in MABP and MCBFV that we detected were not unlike those found previously (12, 23), i.e., a 15 and 20% fall from baseline in MABP and MCBFV, respectively, during phase II, and a 5 and 20% overshoot from baseline in phase IV. However, the rise in MCBFV precedes the increase in MABP, as does the change in RAP_Z (Fig. 2D, dotted line). It would appear that CCP mirrors the changes in intrathoracic pressure and that the fall in CCP leads to the unexpected sudden drop in RAP_Z and the associated rise in MCBFV. As the MABP then increases, the RAP_Z increases as the autoregulation mechanism begins to take effect, and, as expected, all parameters return to baseline within 10-15 s.

However, calculating RAP_C (see METHODS), which takes into account the non-zero-pressure axis intercept (Fig. 1B), provides a more physiological response (Fig. 2D, solid line). This parameter falls gradually from phase I to phase III. After the MABP overshoot, it rises with a 3- to 4-s delay, thus providing a characteristic autoregulatory response. These changes are independent of the CCP temporal pattern.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

We have confirmed the characteristic changes in CBF and MABP seen during phase IV of the VM described in previous studies (12, 23). However, by taking into account CCP, and calculating a measurement of cerebrovascular resistance (RAP_C) that includes this parameter, we have provided a more physiological explanation for these changes than have previous reports, which implies an inability of the CA to cope with changes in MABP of the magnitude recorded.

The early studies of Burton (3) and Dewey et al. (6) suggest that CCP probably reflects the interplay between active arteriolar tone and extravascular pressure. This needs to be interpreted with some caution in the present study. We have defined the CCP as the pressure-axis intercept of instantaneous pressure-velocity relationships after compensating for the time lag between these two waveforms. The main determinant of the time lag (70-100 ms) is the additional time required for the pressure wave to reach the finger, compared with the time it takes for it to reach the MCA. We have found that this parameter is not critical; without lag compensation, the pressure-velocity relationships represented in Fig. 1B tend to form a loop rather than a straight line, but the linear regression still provides CCP values that are significantly >0. Our conclusions only apply, however, to the model adopted in this study (Eq. 1), corresponding to a vascular resistance (RAP) in series with a back pressure (CCP). If more complex models were considered, e.g., including compliance and inductance parameters, different results could have been obtained. First, it is important to compare the CCP model with the more simplistic representation of the cerebral circulation by a single vascular resistance parameter (RAP_Z). This model assumes that pressure-velocity relationships intercept the pressure (and velocity) axis at the origin. From Fig. 1B, which can be generalized to all our subjects, this approach is obviously in error and leads to temporal changes in resistance that are not only very distinct from RAP_C but also more difficult to interpret physiologically, as indicated by the sudden reduction in RAP_Z during phase III (Fig. 2D). It could be argued that pressure-velocity relationships, as depicted in Fig. 1B, can become nonlinear for smaller values of ABP and therefore pass through the origin. This explanation is more physiologically acceptable (3) and should be pursued in the future in the search for more realistic models. The incorporation of cerebrovascular compliance as a model parameter would represent the next step forward in the attempt to develop more realistic models. Giller and Aaslid (9) have obtained evidence suggesting that cerebral vessels are less compliant than those of the systemic circulation. This observation is supported by the relatively flat impedance ratios obtained by Giller et al. (11), using carotid artery pressure waveforms. However, when ABP measurements are obtained with the Finapres device, the contribution of the peripheral circulation needs to be taken into account. More complex models are likely to reflect the compliance of peripheral arteries and the inertia of blood between the aorta and the finger, rather than the input impedance of the MCA. In addition, the incorporation of more parameters frequently leads to instability and unacceptable variability in parameter values. Using intra-aortic ABP measurements, Panerai et al. (19) studied pressure-velocity relationships in neonates, examining the significance of different mathematical models. The most robust model obtained was the RAP+CCP model adopted in the present study. Higher-order dynamic models, which can represent the contribution of compliance and inductance, or quadratic static models, frequently fail to produce statistically significant parameter values, which tend to have very large coefficients of variation (19). These problems stem from the fact that these models depend on the calculation of first- and second-order derivatives of the pressure and velocity waveforms, which are frequently corrupted by noise. Nevertheless, a parameter equivalent to the CCP can also be extracted from these models, albeit with a much higher coefficent of variation (19). The robustness of the two-parameter model adopted is well expressed by the correlation coefficient of the linear regressions, exemplified in Fig. 1B, which is typically in the range 0.85-0.95. The consistency of the RAP+CCP model is reflected in the population values presented in Table 1.

Increases in CCP, as estimated during phases I and II (Fig. 2C), are equivalent to a reduction in perfusion pressure (i.e., MABP  CCP). This reduction in the effective perfusion pressure can be caused by different factors, the most obvious being an increase in intracranial pressure (ICP); Fry et al. (8) demonstrated a significant reduction in perfusion pressure at the Vein of Galen (i.e., the collapsible interface between the intracerebral venous system and the "rigid" venous sinuses linking with the extracerebral venous circulation) in studies using baboons in situations of raised ICP. However, Greenfield et al. (12) observed relatively small changes in ICP during the VM, compared with the large changes observed in CCP (see Fig. 2C). Also important is whether the peripheral MABP we are measuring with the Finapres reflects the true arterial pressure in the MCA. With the increase in intrathoracic pressure and subsequent reduction in stroke volume during phase II, it is conceivable that MCA pressure will fall more than aortic pressure. However, Greenfield et al. (12), who measured MABP invasively via a catheter in the common carotid artery (i.e., centrally), found changes of similar magnitude to the peripheral values obtained in this study. It is also conceivable that cerebral venous pressure will consequently rise above ICP, further contributing to a fall in perfusion pressure; both of these factors will lead to a substantial increase in the value of CCP that is calculated.

Furthermore, it is conceivable that posture might modify the responses we have observed. Because of the TCD measurement, we placed our subjects in the supine position with the head elevated. However, if the VM is performed in the upright position, the corresponding changes in venous return, transmission of the intrathoracic pressure to the cerebral circulation, and autonomic nervous system responses might lead to significant differences in the time course of CBFV, CCP, and RAP_C. Future work will be necessary to estimate the relative contribution of these different factors; it is clear, however, that the contribution of the effective perfusion pressure, and hence CCP, in the hemodynamic changes seen in the cerebral circulation secondary to the VM cannot be ignored.

Other possible confounding factors include the influence of changes in PCO₂ during the VM. We monitored PCO₂ by using the TINA (5, 14) (Radiometer), and this did not detect any change, but its temporal resolution may not be sufficient to detect changes over 1-2 min in keeping with the blood pressure findings during the VM; however, Myer et al. (15) previously found that, in the time window of 15-20 s, there is a 2-mmHg change in PCO₂, which might be expected to trigger vasodilatation and could explain the gradual reduction in RAP_C during phases I-III (Fig. 2D). We have also not considered whether the diameter of the MCA is constant during the VM, i.e., whether using changes in MCA volume to reflect changes in CBF is a valid methodological technique in this situation. During validation studies of TCD, no change was found in MCA diameter during CA, and therefore changes in velocity are extrapolated to represent changes in CBF (10). However, to our knowledge, the effect of the VM in particular has not been assessed; because we used noninvasive methods, we are unable to comment on this.

The physiological changes we are detecting may also be altered by advancing age (the majority of our subjects being elderly); however, Tiecks et al. (23) studied younger subjects and found the same MABP and CBF changes, although CCP per se was not assessed.

In conclusion, CCP of the cerebral circulation has been previously studied in animals (6) and humans (16, 19-21) because it provides a more accurate model for instantaneous pressure-velocity relationships that tend to intercept the pressure axis at values significantly >0 (3, 16). We have demonstrated, in a large group of subjects, that this parameter provides a coherent explanation for the dynamics of CBFV in response to the VM and that, unless this parameter is taken into account, assessment of autoregulation on the basis of the VM will lack a sound physiological basis and might produce misleading results. In addition, estimates of cerebrovascular resistance (or RAP) (RAP_C) do not show the aberrant sudden changes displayed by classical estimates (RAP_Z), which ignore this parameter. Because of its close relationship to physiological changes of the cerebral circulation, the RAP_C parameter has potential to grade the performance of CA during the response to the MABP overshoot in phase IV of the VM (21).