Influence of hypercapnic vasodilation on cerebrovascular autoregulation and pial arteriolar bed resistance in piglets

Abstract

Changes in both pial arteriolar resistance (PAR) and simulated arterial-arteriolar bed resistance (SimR) of a physiologically based biomechanical model of cerebrovascular pressure transmission, the dynamic relationship between arterial blood pressure and intracranial pressure, are used to test the hypothesis that hypercapnia disrupts autoregulatory reactivity. To evaluate pressure reactivity, vasopressin-induced acute hypertension was administered to normocapnic and hypercapnic (N = 12) piglets equipped with closed cranial windows. Pial arteriolar diameters were used to compute arteriolar resistance. Percent change of PAR (%ΔPAR) and percent change of SimR (%ΔSimR) in response to vasopressin-induced acute hypertension were computed and compared. Hypercapnia decreased cerebrovascular resistance. Indicative of active autoregulatory reactivity, vasopressin-induced hypertensive challenge resulted in an increase of both %ΔPAR and %ΔSimR for all normocapnic piglets. The hypercapnic piglets formed two statistically distinct populations. One-half of the hypercapnic piglets demonstrated a measured decrease of both %ΔPAR and %ΔSimR to pressure challenge, indicative of being pressure passive, whereas the other one-half demonstrated an increase in these percentages, indicative of active autoregulation. No other differences in measured variables were detectable between regulating and pressure-passive piglets. Changes in resistance calculated from using the model mirrored those calculated from arteriolar diameter measurements. In conclusion, vasodilation induced by hypercapnia has the potential to disrupt autoregulatory reactivity. Our physiologically based biomechanical model of cerebrovascular pressure transmission accurately estimates the changes in arteriolar resistance during conditions of active and passive cerebrovascular reactivity.

hypoventilation with supplemental oxygen (“permissive hypercapnia”) is used as a therapeutic intensive care management technique to reduce ventilator-induced lung injury (4, 14, 16). Hypercapnia causes cerebrovascular vasodilation. However, the effect of hypercapnia on cerebrovascular autoregulation is poorly understood. One recent neonatal critical care study suggests that use of levels of the partial pressure of arterial blood carbon dioxide (Pco₂) above 45 Torr may abolish autoregulatory reactivity (13). Because an awareness of whether pressure regulation is intact during therapeutic hypercapnia is critical to patient management, this laboratory study was designed to examine the influence of hypercapnia on autoregulatory reactivity.

A second aim of this study was to examine the validity of a two-step modeling procedure of cerebrovascular pressure transmission to estimate changes of arteriolar resistance. With this modeling procedure, recordings of arterial blood pressure (ABP) and intracranial pressure (ICP) are used to derive a physiologically based biomechanical model of cerebrovascular pressure transmission. Because of the availability of these pressure recordings in the intensive care management of patients with brain injury, the two-step modeling procedure may have the potential to become a bedside intensive care monitoring method designed to recognize impaired cerebrovascular autoregulation. In the first step of the modeling procedure, the modal frequencies of the system identification numerical black box model of cerebrovascular pressure transmission are derived from the ABP and ICP recordings. In the second step, these modal frequencies and the ABP recording are used to derive a biomechanical parameter identification model that produces a simulated ICP recording that matches the actual ICP recording. Recently, our laboratory illustrated for a single case by subjective visual matching of the laboratory recordings to simulated ICP recordings that, during induction of acute hypercapnia, changes of arteriolar resistance and compliance derived from laboratory measures are replicated by the physiologically based biomechanical model (7). The present study builds on this previous one. We now use a two-step modeling procedure that employs objective minimization measures to obtain a match between the laboratory and simulated ICP recordings to provide objectively derived model parameters. Also, unlike the previous study, statistical group comparisons between changes of cerebrovascular resistance derived from laboratory measures and corresponding simulated resistance values are made in this study.

The present laboratory study addresses the influence of acute hypercapnia on autoregulatory reactivity. ABP and ICP were simultaneously monitored in piglets equipped with cranial windows. Computation of pial arteriolar resistance (PAR) was determined from measurement of pial arterial diameter (PAD). Simulated resistance of the arterial-arteriolar bed (SimR) was derived from a biomechanical model of cerebrovascular pressure transmission. Here we test the hypotheses that 1) hypercapnia inhibits cerebral blood flow (CBF) regulation to increased perfusion pressure; and 2) changes in simulated arteriolar bed resistance of the biomechanical model allow assessment of cerebrovascular autoregulaton and changes of cerebrovascular resistance.

METHOD AND MATERIALS

Laboratory preparations.

Using a protocol approved by the Animal Care and Use Committees of The University of Memphis and The University of Tennessee Health Science Center, α-chloralose anesthetized piglets, ranging in weight from 2 to 4 kg, were studied. In each piglet, a cranial window was placed, and video micrometer recordings of PAD were made, as described previously (23). A fluid-filled catheter inserted in the femoral artery was used to record ABP, and ICP was recorded using a fluid-filled catheter coupled to a fluid-filled port of the cranial window. The tone of the cerebral vasculature was manipulated by ventilating the preparation with a gas mixture of 20% O₂, 10% CO₂, and 70% N₂ for 10 min, at which time vasopressin-induced acute hypertensive challenge was administered by intravenous administration of vasopressin at 0.4 μg·kg⁻¹·min⁻¹ for 5 min.

Data acquisition.

ICP recordings were obtained with an intraparenchymally inserted fiber-optic transducer (Camino Direct Pressure Monitor, Camino Laboratories, San Diego, CA). Recordings of ABP were obtained via a fluid-filled cannula in the femoral artery with a SpaceLabs model 90623A monitor (Redmond, WA). The bandwidth of ABP channel of this monitor is 40 Hz, with a ±15% variation. The fast-flush method was used to estimate the bandwidth of the entire system, including the fluid-filled femoral catheter, to be 16.9 ± 0.99 Hz (7, 8). Each recording was digitized at a rate of 250 samples/s with a system previously described (7).

PAD measurement and resistance calculation.

A video micrometer, with magnification of ×3,600, was used to videotape the diameter of three arterioles during the experiment. The video recording was then converted to still-image frames using OSS Video Decompiler software (One Stop Soft, www.onestopsoft.com) at a rate of 30 frames/s. A tracking algorithm was implemented using MATLAB software to measure the diameter of arterioles during the experiment. Each frame was converted to gray scale and contrast adjusted to enhance the arterioles from the background. A region of interest on the arteriole was then chosen for diameter measurement. A threshold intensity value was set to distinguish between background and the arteriole within the chosen region. The diameter was then computed based on the number of pixels, whose intensity values were lower than the selected threshold value iteratively for every frame. Mean value of PAD was computed on at least 7,200 frames during a period of continuous steady-state conditions. Across all piglets, the range of arteriolar diameter used in the computation of resistance was between 30 and 70 μm.

Computation of PAR and percent change arteriolar resistance.

Based on the biomechanics of steady-state laminar flow in a cylindrical tube of 1 mm length (9), the resistance of a pial arteriole can be estimated as:

(1)

With the use of this equation, PAR was computed on the assumption that the coefficient of viscosity for whole blood was 0.0027 N·s·m⁻² at 37°C. Also, percent change (%Δ) of arteriolar resistance (%ΔPAR) was calculated as:

(2)

Two-step modeling method of cerebrovascular pressure transmission.

The first step of the two-step modeling method uses system identification modeling, the details of which have been previously reported (6, 8). Briefly, system identification modeling method produces a black box model with a simulated output that matches the actual output by minimization of the least squares difference between the actual and simulated digitized output files (17). Of critical importance in the use of system identification modeling is the selection of the generalized description of the dynamic process to be modeled (17). The physiologically based biomechanical model proposed by Czosynka and colleagues (5) is used to provide the basis of the required generalized dynamic differential equation of cerebrovascular pressure transmission. The autoregressive moving average system identification technique is applied to each 2,000 paired samples of digitized pressure recordings of 8 s to obtain the modal frequencies of cerebrovascular pressure transmission. For the brief period on which each numerical model is based, perturbations of ABP and ICP are assumed to be small, and cerebrovascular pressure transmission is considered to be linear and time invariant. This strictly numerical black box model enables the computation of the model frequencies of the cerebrovascular pressure transmission, which are the roots of the polynomial equation:

(3)

where aa is coefficient, and λ is eigenvalue. The highest modal radian frequency (HMF) is defined as the eigenvalue with the greatest absolute value and is converted to modal frequency by division by 2 * π. The other two eigenvalues are the dominant modes. The dampening factor (DF) of the process defines whether dominant modes result in a process that is overdamped or underdamped. If the two eigenvalues are real, then the process is overdamped and DF is >1; whereas, if the eigenvalues are complex conjugates, then the process is underdamped and DF is <1. As DF of a process approaches zero, the process demonstrates more resonant characteristics.

In the second step of the modeling method, the mean values of the modal frequencies derived by the first-step numerical model and the corresponding 8-s interval of ABP recording are used to guide the manipulation of the parameters of the physiologically based biomechanical model, such that the simulated 8-s recording of ICP matches the actual ICP recording over the interval. Of note is that three modifications of the biomechanical model proposed by Czosynka and colleagues (5) were implemented. First, the description of the resistance of the arterial-arteriolar bed was slightly modified by placing the element representing the arterial-arteriolar bed at the midpoint as it has been modeled by others previously (1). Second, in addition to the representation of the bridging veins with variable resistance, cerebral veins within the parenchyma were also considered to vary. Third, the representation of sinus pressure was replaced by an external venous resistance component and connected to a central venous pressure generator, as previously modeled by others (20, 26, 27). Specifically, MATLAB software (The MathWorks, Natick, MA) was used to simulate using a modified state space description of the biomechanical model previously reported (6). The MATLAB model computation was constructed to simulate ICP for each ABP recording and the corresponding numerically derived modal frequencies for each modeling interval. During the computation, the resistance and compliance elements of the biomechanical model were manipulated such that the minimum square error for the 2,000-sample interval between 1) the actual and simulated ICP recordings, 2) the numerical model value of HMF and the biomechanical value of HMF, and 3) the numerical model value of DF of the lower modal frequencies and the DF of the biomechanical model were minimized. Use of DF to represent the dominant two modal frequencies simplified the computation of matching the numerically derived and simulated lower modal frequencies by eliminating the need for complex arithmetic for condition of an underdamped system. Values of each element of the biomechanical model were manipulated. Initially, a normal CBF was assumed to be 40 ml·100 g⁻¹·min⁻¹ (22, 26–27). Both compliance and resistance elements were allowed to range over values consistent with those reported in previously published modeling studies of ICP dynamics (5, 26–27, 28). Compliance of the arterial/arteriolar bed was allowed to range from 0.01 to 10 ml/mmHg, whereas venous and intracranial compliance were varied between 0.01 and 2 ml/mmHg (5, 26–27, 28). Venous resistance varied from 0.1 to 2 mmHg·ml⁻¹·min, and resistance of the arterial-arteriolar bed varied from 2 to 25 mmHg·ml⁻¹·min (15–16, 19). While the ranges of arterial and venous compliance values overlap, the mean value of arterial compliance derived by the minimization procedure was more than an order of magnitude less than the corresponding derived mean value of venous compliance for each experimental condition. A maximum mean squared error of 0.6, which results in a goodness of fit in which the mean maximum difference between the actual and simulated ICP recordings, was <0.6 mmHg for each point. Also, the minimum correlation value of 0.9 between both actual and simulated ICP recording was used to initialize the square error values. The parameters of the biomechanical model were estimated by the model computation to obtain a best fit set, which produced a minimum least squares error and maximum correlation value for each segment of the pressure recording.

Statistical methods.

All mean values are reported with ±SE. In all cases, the degree of significance between two mean values was determined by using the t-statistic. Both the Shapiro-Wilks (SW-test) and Kolmogorov-Smirnov (KS-test) were used to test distributions for uni-modality.

RESULTS

Vasopression-induced acute hypertension was administered to 12 piglets during the control normocapnic condition and the treated hypercapnic conditions. Histograms of the observed pressor responses for all piglets based on %ΔPAR and %ΔSimR are presented in Fig. 1. Based on both the SW-test (P < 0.007) and KS-test (P < 0.001), both histograms were determined not to be normally distributed, but separated into two populations. As a result, we treated the six hypercapnic piglets, which demonstrated vasopression-induced acute hypertensive responses, indicative of passive cerebrovascular reactivity, as a distinct group denoted as the pressure passive group; the other six hypercapnic piglets demonstrated responses consistent with active cerebrovascular reactivity and were considered a distinct group denoted as the intact autoregulation group.

Fig. 1.

Histograms of percent change in pial arteriolar resistance (%ΔPAR) and percent change in simulated arterial-arteriolar resistance (%ΔSimR). Pressor challenge causes an increase of arterial blood pressure and a change in PAR and SimR. Hatched bars represents hypercapnia, and solid bars represents normocapnia. A: histogram of %ΔPAR. B: histogram of %ΔSimR. By application of both the Shapiro-Wilks and Kolmogorov-Smirnov tests, both distributions were determined not to be normally distributed, with respective levels of significance of P < 0.007 and P < 0.001.

For the 12 piglets, induction of hypercapnia resulted in group mean values ± SE of Pco₂, Po₂, and pH of 78.5 ± 0.3 Torr, 99.6 ± 0.9 Torr, and 7.1 ± 0.0, respectively. Mean values ± SE of Pco₂, Po₂, and pH for the two groups and the conditions of normocapnia and hypercapnia are given in Table 1. There was no difference in the depth of hypercapnia between the intact autoregulation and pressure passive groups. Grand mean values ± SE of ABP, ICP, and cerebral perfusion pressure (CPP) for the two groups were also computed (see Table 2). Vasopressin-induced hypertensive challenge caused a significant increase of ABP and CPP for both groups. For both the intact autoregulation group during hypercapnia, ICP decreased in response to the challenge, whereas, for the pressure passive group, ICP increased significantly. For the 12 piglets, the correlation between arterial Pco₂ and CPP was found to be −0.93 ± 0.05 (n = 24).

Table 1.

Grand mean values of Pco₂, Po₂, and pH for intact autoregulation and pressure passive groups during normocapnia and hypercapnia

Groups and Condition	N	Pco2, Torr	Po2, Torr	pH
Intact autoregulation group
Normocapnic condition	6	34.70±1.20	85.60±6.70	7.40±0.05
Hypercapnic condition	6	80.40±2.61*	98.70±7.80†	7.14±0.03
Pressure passive group
Normocapnic condition	6	34.10±2.8	85.90±8.90	7.39±0.09
Hypercapnic condition	6	77.40±4.20*	113.00±10.78‡	7.09±0.02

Values are means ± SE; N, no. of animals.

^*Significant degree of difference between mean value of Pco₂ in normocapnic group and mean value of Pco₂ in hypercapnic group, P < 0.001. Significant degree of difference between mean value of Po₂ in normocapnic group and mean value of Po₂ in hypercapnic group:

^†P < 0.025,

^‡P < 0.005.

Table 2.

Grand mean values of ABP, ICP, and CPP for intact autoregulation and pressure passive groups during normocapnia, hypercapnia, and hypertensive challenge

Groups and Condition	N	ABP	ICP	CPP
Intact autoregulation group
Normocapnic condition	6	70.10±0.61	3.20±0.32	73.25±0.91
Hypertensive challenge and normocapnia	6	82.40±0.56a	2.20±0.14b	80.6±0.64a
Hypercapnic condition	6	73.8±1.84	4.6±0.91	69.5±0.44c
Hypertensive challenge and hypercapnia	6	85.60±1.76d	3.60±0.17	75.60±1.30d
Pressure passive group
Normocapnic condition	6	72.20±0.86	3.10±0.49	75.70±1.27
Hypertensive challenge and normocapnia	6	84.40±0.19a	2.10±0.09b	82.59±0.78a
Hypercapnic condition	6	75.40±1.54	6.10±1.15c	70.40±1.52c
Hypertensive challenge and hypercapnia	6	86.60±2.16e	12.80±1.40f	74.80±1.47e

Values are means ± SE in mmHg; N, no. of animals. ABP, arterial blood pressure; ICP, intracranial pressure; CPP, cerebral perfusion pressure. Significant degree of difference between baseline and challenge mean values during normocapnia:

^aP < 0.005,

^bP < 0.025.

^cSignificant degree of difference between normocapnia and hypercapnia mean values, P < 0.05. Significant degree of difference between baseline and challenge mean values during hypercapnia:

^dP < 0.025,

^eP < 0.005,

^fP < 0.001.

Experimental grand mean values of PAD and the corresponding PAR and HMF and the corresponding SimR of cerebrovascular pressure transmission for the two groups were also calculated for each condition (see Table 3). Dilation induced by hypercapnia resulted in an increase of PAD and a decrease of both PAR and SimR. Consistent with active arteriolar pressure regulation for the intact autoregulation group during normocapnia and hypercapnia, hypertensive challenge decreased PAD and increased PAR. During active cerebrovascular reactivity, HMF decreased with increasing CPP. Changes in the parameters of the biomechanical model of cerebrovascular pressure transmission were also consistent with active pressure regulation of CBF in that SimR increased with increasing CPP. In contrast, for the pressure passive group during hypercapnia, vasopression-induced hypertension 1) increased PAD and decreased PAR; 2) increased HMF; and 3) SimR decreased.

Table 3.

Grand mean values of PAD, PAR, HMF, and SimR for the intact autoregulation and pressure passive groups

Groups and Condition	N	PAD, μm	PAR, mmHg·ml−1·min	HMF, Hz	SimR, mmHg·ml−1·min
Intact autoregulation group
Normocapnia condition	6	37.50±1.60	2729±100.4	35.80±1.2	4.68±0.05
Hypertensive challenge and normocapnia	6	29.60±0.54a	7032.50±68.7a	30.20±0.42a	6.83±0.20a
Hypercapnia condition	6	75.10±1.1c	170.62±3.03c	45.70±1.27d	3.20±0.14c
Hypertensive challenge and hypercapnia	6	61.60±2.78e	374.90±8.70e	22.70±2.28e	6.25±0.20e
Passive group
Normocapnia condition	6	36.80±2.13	2943.10±26.2	31.60±0.86	6.08±0.34
Hypertensive challenge and normocapnia	6	30.10±0.34a	6576.70±59.3a	25.20±0.94a	7.98±0.14a
Hypercapnia condition	6	62.20±3.99c	360.70±25.2c	39.30±2.25d	5.03±0.12d
Hypertensive challenge and hypercapnia	6	97.80±5.56f	59.10±1.47e	49.60±2.60f	3.23±0.20f

Values are means ± SE; N, no. of animals. PAD, pial arteriolar diameter; PAR, pial arteriolar resistance; HMF, highest modal radian frequency; SimR, simulated arterial-arteriolar bed resistance. Degree of significant difference between baseline and challenge mean values during normocapnia:

^aP < 0.0001,

^bP < 0.005. Degree of significant difference between normocapnia and hypercapnia mean values:

^cP < 0.0001,

^dP < 0.005. Degree of significant difference between baseline and challenge mean values during hypercapnia:

^eP < 0.0001,

^fP < 0.005.

As noted previously, the computation of PAR is based on PAD and a length of 1 mm. As a result, the orders of magnitude in the difference between PAR and SimR represent the difference between high-arteriolar resistance and that of the global resistance of the arterial-arteriolar bed of the entire brain. Changes in these values of resistance, %ΔPAR, %ΔSimR, and %ΔCPP, during vasopressin-induced hypertensive challenge for the conditions of normocapnia and hypercapnia were computed and plotted (see Fig. 2). Values of percent change in resistance to hypertensive challenge for all preparations during normocapnia (see Fig. 2A) were positive. Similarly, during hypercapnia and hypertensive challenge, the intact autoregulation group demonstrated a positive increase in both %ΔPAR and %ΔSimR (see Fig. 2B). In contrast, for the passive pressure group, the changes in both the experimental arteriolar resistance and simulated bed resistance decreased with increased %ΔCPP during hypertensive challenge and were significantly (P < 0.001) different than those determined for both the group of all preparations and the intact autoregulation group (see Fig. 2C). Intragroup comparisons of mean %ΔPAR and %ΔSimR for each condition were not significantly different.

Fig. 2.

%ΔPAR and %ΔSimR induced by vasopressin-induced acute hypertensive challenge vs. percent change of cerebral perfusion pressure (%ΔCPP). A: vasopressin-induced hypertension challenge during normocapnia for all preparations. Hatched bar graph represents mean %ΔPAR (±SE), and solid bar represents %ΔSimR (±SE). For all preparations (N = 12), hypertensive challenge during normocapnia caused pial arteriolar vasoconstriction and increased simulated cerebrovascular resistance. B: vasopressin-induced hypertension challenge during hypercapnia for the intact autoregulation group (N = 6). For this group, hypertensive challenge caused pial arteriolar vasoconstriction and increased simulated cerebrovascular resistance. C: vasopressin-induced hypertension challenge during hypercapnia for the pressure passive group (N = 6). For this group, hypertensive challenge caused pial arteriolar vasodilation and decreased cerebrovascular resistance (N = 6). Intragroup comparisons of mean of %ΔPAR and %ΔSimR were not different, whereas intergroup comparisons of %ΔPAR and %ΔSimR were significantly different.

DISCUSSION

The new findings of this study are that 1) induced hypercapnia has the potential to disrupt autoregulatory reactivity; and 2) changes in PAR are matched by equivalent changes of SimR of the physiologically based biomechanical model of cerebrovascular pressure transmission.

Percent changes of the values of both PAR and SimR for each group were the same, even though a difference of several orders of magnitude in the values of resistance was determined. During active autoregulatory reactivity, both resistances, PAR and SimR, increased, and, during the passive pressure condition, both decreased in response to hypertensive challenge. In this study, measured values of PAD ranged from 30 to 70 μm, which corresponds to a range of resistance of 6,665 to 225 mmHg·ml⁻¹·min. Because the arterioles play a major role in the control of CBF, such a range of high-resistance value is to be expected. In contrast, SimR values of the physiologically based biomechanical model of cerebrovascular pressure transmission derived from ABP and ICP recordings during baseline ranged from 3.0 to 3.5 mmHg·ml⁻¹·min. Because the SimR value represents the resistance of the cerebral arterial vascular tree and capillaries, which is incredibly rich in parallel pathways, the overall value of equivalent resistance should be several orders of magnitude less than that of a single arteriole. Hypercapnia seems to induce a relatively uniform vasodilation for the entire arterial-arteriolar bed. As a result, both PAR and the simulated equivalent resistance of the cerebral arterial-arteriolar tree decreased similarly. Furthermore, the nature of the hypertensive challenge is that it also uniformly changes the resistance of pial arteriole: during active vasoreactivity resistance increases and during passive vasoreactivity resistance decreases. Here again, changes in the resistance of the entire cerebral arterial-arteriolar tree should follow the corresponding changes in arteriolar resistance. Thus our findings of no intragroup differences, but significant intergroup changes in the mean values of percent change of PAR and SimR of the biomechanical model, are reasonable.

Previous studies on mature dog of the influence of arterial carbon dioxide on autoregulation involving direct measures of CBF report that a change of arterial Pco₂ from 40 to 80 Torr results in an 80–100% increase of CBF (12). Furthermore, because a consistent loss of autoregulation during severe hypercapnia in four preparations was observed, it was suggested that, during near-maximal dilation of the microvasculature, autoregulation is reduced or absent (11). Perhaps because of the genetic and maturity differences between laboratory preparations, our findings are not in complete agreement with these previous studies.

Studies on healthy human subjects and patients have employed a noninvasive method in which the relationship between middle cerebral artery blood flow velocity (MCAFV) and ABP have assessed autoregulation using static and dynamic indexes of autoregulation. In one such study, estimated cerebrovascular resistance was derived as the ratio of ABP to MCAFV (9). Static rate of autoregulation (SROR) index was defined as the ratio of percent change in estimated cerebrovascular resistance to percent change in ABP. The SROR index was reported to reflect percentage of autoregulatory capacity, with high values being indicative of high autoregulatory capacity and low values indicative of low capacity (9). Instead of using MCAFV, we used simulated cerebral blood flow (sCBF) to compute a modified index. For the normocapnic group and hypercapnic group with intact cerebrovascular regulation, these indexes were 99.7 ± 3.7 and 92.4 ± 5.7. These high values of our modified index were consistent with the reported interpretation of the SROR index, which would indicate high autoregulatory capacity during intact cerebrovascular regulation. For the hypercapnic group with impaired autoregulation, the modified computed index was 22.6 ± 8.4, indicative of low autoregulatory capacity. Another study of dynamic regulation of healthy adult subjects concluded that acute hypercapnia almost completely destroys the autoregulatory mechanism (3). However, another study of healthy adult subjects exposed to both acute and long-term elevation of environmental carbon dioxide found that static autoregulation was impaired initially, but, after 3 days of exposure, autoregulation was unaffected (24). Clinical studies of neonates, which used the static autoregulatory index, have concluded that hypercapnia impairs autoregulatory reactivity in these patients (13, 18). Finally, laboratory studies of acute hypercapnia in the newborn piglet and rat and the adult cat also have reported that hypercapnia impairs autoregulation (19, 21, 25). Our findings that induction of acute hypercapnia can impair autoregulation are consistent with these previously reported studies.

Caution should be taken when relating our findings to the clinical use of hypercapnia. First, our experiments represent the consequences of the induction of acute hypercapnia; whereas therapeutic hypercapnia is maintained by permissive hypercapnia techniques over a much longer time period, in which decreases in arterial pH are mitigated by infusion of bicarbonate (4, 14, 16). Second, the mechanisms that underlie our observations remain uncertain and will require considerably more experimentation. We can speculate that it seems likely that the induction of acute hypercapnia placed the vessels on the border between active and passive so that some lost responses, but others did not. Beyond that, it should be emphasized, there were no noticeable differences between the piglets that subsequently segregated the two groups before CO₂ challenge. Finally, we do not know whether vasopressin might be uniquely related to our observations. Vasopressin was just a tool to increase CPP, and we cannot envision an acute mechanism where the responses to equivalent hypertension by another means may be different. However, our findings and those reported by others (11–12) do suggest the use of caution with induction of hypercapnia, particularly during the use of permissive hypercapnia in neonatal critical care. During severe hypercapnia at maximal or near-maximal vasodilation, autoregulatory reactivity may become passive. As a result, in neonatal patients without active vasoconstriction, increased CPP is not attenuated by the arteriolar bed. Thus the small arterioles, capillaries, and venules are unprotected from transient increases in pressure and can be injured. In addition, loss of autoregulation and high CPP can impair capillary fluid balance and increase cerebral edema.

In summary, vasodilation induced by hypercapnia has the potential to disrupt autoregulatory reactivity. The described two-step modeling procedure of the physiologically based biomechanical model of cerebrovascular pressure transmission accurately estimated the changes in arteriolar resistance during conditions of active and passive cerebrovascular reactivity. The potential clinical value of the proposed methodology is the ability to continuously assess changes of the microvasculature, a capability that other technologies, including imaging techniques, lack. Note that this ability is important because, as shown here, some individuals will become pressure passive, but others maintain active flow regulation in response to increasing intravascular pressure. Differentiating between the groups a priori does not appear to be possible with available information in the clinic, or so previously, even in the laboratory.

GRANTS

This research was supported in part by grants from the National Heart, Lung, and Blood Institute and National Institute of Neurological Disorders and Stroke of the National Institutes of Health, the Southeast Affiliate of the American Heart Association, a Graduate Fellowship to N. Narayanan from the Herff College of Engineering Trust, and a University of Memphis Faculty Research Grant.