Halothane and isoflurane have been shown to induce disparate effects on different brain structures in animals. In humans, various methods for measuring cerebral blood flow (CBF) have produced results compatible with a redistribution of CBF toward deep brain structures during isoflurane anesthesia in humans. This study was undertaken to examine the effects of halothane and isoflurance on the distribution of CBF.
Twenty ASA physical status patients (four groups, five in each) anesthetized with either isoflurane or halothane (1 MAC) during normo- or hypocapnia (PaCO2 5.6 or 4.2 kPa (42 or 32 mmHg)) were investigated with a two-dimensional CBF measurement (CBFxenon, intravenous 133xenon washout technique) and a three-dimensional method for measurement of the regional CBF (rCBF) distribution with single photon emission computer-aided tomography (SPECT; 99mTc-HMPAO). In the presentation of SPECT data, the mean CBF of the brain was defined as 100%, and all relative flow values are related to this value.
The mean CBFxenon level was significantly influenced by the PaCO2 as well as by the anesthetic used. At normocapnia, patients anesthetized with halothane had a mean CBFxenon of 40 +/- 3 (SE) ISI units. With isoflurane, the flow was significantly (P < 0.01, 33 +/- 3 ISI units) less than with halothane. Hypocapnia decreased mean CBFxenon (P < 0.0001) during both anesthetics (halothane 24 +/- 3, isoflurane 13 +/- 2 ISI units). The effects on CBFxenon, between the anesthetics, differed significantly (P < 0.01) also during hypocapnia. There were significant differences in rCBF distribution measured between the two anesthetics (P < 0.05). During isoflurane anesthesia, there was a relative increase in flow values in subcortical regions (thalamus and basal ganglia) to 10-15%, and in pons to 7-10% above average. Halothane, in contrast, induced the highest relative flow levels in the occipital lobes, which increased by approximately 10% above average. The rCBF level was increased approximately 10% in cerebellum with both anesthetics. Changes in PaCO2 did not alter the rCBF distribution significantly.
There is a difference in the human rCBF distribution between halothane and isoflurane with higher relative flows in subcortical regions during isoflurane anesthesia. However, despite this redistribution, isoflurane anesthesia resulted in a lower mean CBFxenon than did anesthesia with halothane.
Key words: Anesthetics, volatile: halothane; isoflurane. Brain: cerebral blood flow; regional cerebral blood flow. Carbon dioxide: hypocapnia; normocapnia. Measurement technique: single photon emission computer-aided tomography (SPECT).
ALTHOUGH isoflurane is thought to be the volatile anesthetic of choice for patients undergoing intracranial surgery, its superiority to halothane has been questioned lately. The diminished cortical cerebral blood flow (CBF) found during isoflurane anesthesia [1,2] may cause a redistribution to central brain structures as found in rabbit  and implied in rat.  However, because Young et al.  failed to reproduce these findings in the rat, this isoflurane-induced redistribution of CBF is still debated.
Using tomographic techniques, such as single photon emission computer-aided tomography (SPECT), three-dimensional studies of regional CBF (rCBF) are possible. Like a computed tomography scanner, these techniques provides slices through the brain that represent a picture of the flow. The technique is based on injection of a radioactive tracer, e.g., technetium-99m-hexamethyl-propylene amine oxime (sup 99m Technetium-HMPAO), which gives an image of the distribution of the flow at the time of injection. With these methods, not only the cortical but also subcortical regions can be studied. Intraoperative SPECT studies have been performed in patients undergoing carotid endarterectomy  during fentanyl/isoflurane anesthesia. To our knowledge, no data concerning rCBF in halothane-anesthetized patients are available.
To examine whether CBF redistribution occurs in humans, we evaluated the effect of halothane and isoflurane on the absolute level of CBF determined by intravenous133xenon  combined with measurements of the rCBF distribution with SPECT and99mTc-HMPAO.
Materials and Methods
Twenty male patients, ASA physical status 1, scheduled for inguinal herniorrhaphy participated in the study. The protocol was approved by the ethics committee for human studies and the isotope committee at the University of Lund. Written informed consent was obtained from each participant.
The 20 patients were randomly allocated into one of the following four groups, with 5 in each group, to receive 1 MAC halothane during normo- or hypocapnia or 1 MAC isoflurane during normo- or hypocapnia.
No premedication was given. Anesthesia was induced with 0.75 mg/kg intravenous meperidine and 2–2.5 mg/kg propofol. Succinylcholine (1 mg/kg) was used to facilitate tracheal intubation. Anesthesia was maintained with halothane or isoflurane in 60–70% N2O in oxygen. Throughout anesthesia, end-tidal concentrations of halothane, isoflurane, and carbon dioxide (ETCO2) were kept as close as possible to the final experimental condition.
Mechanical ventilation was accomplished with a Servo ventilator (Siemens 900 B. Siemens Elema, Solna, Sweden). ETCO2, as well as concentrations of volatile anesthetics, were recorded on a Normocap 102–24–02 and a Normac gas monitor (Datex, Helsinki, Finland), respectively.
Nitrous oxide was discontinued 10–20 min before the end of surgery, and the patients were kept anesthetized with 1 MAC halothane or isoflurane in 30% oxygen for at least 30 min. When stable conditions were achieved, mean CBF was measured using radioactive133xenon clearance (CBFxenon).99mTc-HMPAO was injected immediately after the CBFxenonmeasurement, and anesthesia was maintained for 5 min more. No patient was studied sooner than 60 min after the induction of anesthesia. After a postanaesthetic recovery period of 1–2 h, the patients were brought to the neurophysiologic department for SPECT scanning.
During measurement of the mean CBFxenon, hemoglobin concentration and arterial blood gas samples was analyzed using an ABL 500 (Radiometer, Copenhagen, Denmark). Temperature, pulse, and arterial blood pressure were monitored.
Measurement of Mean CBF
Measurement of mean CBFxenonwas performed with injection of 0.5 Gbq (10 mCi)133xenon in a cubital vein, followed by a rapid injection of 20 ml of isotonic saline. The uptake and clearance of the tracer substance was recorded with a scintillation detector with a wide collimation (about 90 degrees view) placed on the right side of the head over the parietotemporal region, including most of the brain in its field. Clearance through the lungs was recorded from the expired air. A Novo Cerebrograph 10a (Simonsen Medical A/S, Randers, Denmark) was used for data collection with a sampling time of 11 min and subsequent flow calculation. The CBFxenonwas expressed as the initial slope index (ISI),  because it represents the blood flow of all tissue recorded but is highly dominated by the gray matter blood flow and very little influenced by extracerebral components. .
Measurement of Three-dimensional CBF Distribution
Measurement of the regional distribution of CBF was performed through injection of a tracer substance, 0.5 Gbq (10 mCi)99mTc-HMPAO (Ceretec, Amersham, England) in a cubital vein, followed by a rapid injection of 20 ml isotonic saline. This tracer substance is lipid-soluble at the time of injection and is distributed in proportion to the blood flow. In the brain cells, the carrier molecule HMPAO is, within a few minutes, converted to a water-soluble form that cannot cross the cell membrane. Hence, the amount of99mTc (half-life 6 h) that remains trapped in the brain cells is proportional to the rCBF distribution. The cerebral99mTc distribution was recorded three-dimensionally with a SPECT scanner (Tomomatic 564, Medimatic A/S, Denmark), giving a picture of the rCBF distribution at the time of the sup 99m Tc-HMPAO injection. [9,10] The three-dimensional distribution of99mTc-HMPAO in the brain was recorded in ten contiguous, about 1-cm thick slices, parallel to the orbitomeatal (OM) line, with the center of the lowest slice located 1 cm below the OM line. The head position was controlled with light beams on the external auditory meatus and the nasion. The slices were recorded in two interlacing sets of five slices each, with a recording time of 10 min for each set, which gave about 106counts/slice and an intraslice resolution of about 1 cm. The regions of interest were automatically positioned within each slice, with adjustment to the subject's brain size.
Calculations and Statistical Methods
The mean CBFxenonvalue recorded from our detector corresponds to a weighted average of different brain regions with the weight of each region in the average mainly determined by its133xenon content and the distance from the detector.  Consequently, the CBFxenonvalue from the parietal detector can be seen as an average between the CBF in the closest cortical tissue (parietal cortex) and the mean CBF, with dominance for the parietal cortex.
In the SPECT99mTc-HMPAO measurements the average CBF level, calculated as the average number of counts through all regions, gray as well as white matter, was defined as 100% for the presentation of relative flow values. Regions of interest were outlined in the brain slices by a program constructed from anatomic templates in a computed tomography brain atlas.  Three-dimensional cerebral regions of interest were calculated from adjoining regions of interest in different brain slices representing the same structure. The program gave the mean value in each region of interest as well as the size of the region of interest area.
Because the parietal and the average CBF had virtually the same value ([nearly equal] 100%) in the SPECT99mTc-HMPAO measurements, it was a good approximation to estimate the recorded CBF sub xenon value to the parietal cortical CBF. Regional CBF values were calculated from the CBFxenonmean flow values and the99mTc-HMPAO rCBF distribution, the133xenon CBFxenonvalues being equated with the parietal values in the regional distribution. Carbon dioxide response was calculated as Delta CBFxenondivided by the corresponding Delta PaCO2. The conversion factor from kPa to mmHg was 0.13–33.
All values are given as mean plus/minus SE. Analysis of variance (ANOVA) and conservative post hoc testing with Student's unpaired t test was used for statistical comparison of the groups. In the ANOVA test of the SPECT99mTc-HMPAO data, the different regions of interest were within group factors and PaCO2and isoflurane/halothane were between groups factors. Subsequent to the ANOVA on the SPECT99mTc-HMPAO measurements, a post hoc testing with t test was performed to identify which regions participated most to the regional differences, t tests were done only when the ANOVA indicated a significant difference in the regional distribution. The P values for the ANOVA interaction terms were corrected for departure from sphericity  making the evaluation more conservative. P less or equal to 0.05 was considered statistically significant.
Because the effects of the anesthetic agents are expected to be similar on the right and left hemispheres, the corresponding regions on each side cannot be seen as independent. We, therefore, have used a design that employs the mean values of the symmetrically located regions in our ANOVA design. In corroboration, an ANOVA test for hemisphere effects shows no significant side effects or interactions from the anesthetics and/or ventilation.
Physiologic values of the four groups are presented in Table 1. Aside from the expected pH and PaCO2differences between the normo- and hypocapnic groups, there were no statistically significant differences. Temperature was measured in only two of the five patients in the hypocapnic groups.
Effects of Halothane and Isoflurane on Mean CBF sub xenon Level Using sup 133 Xenon
Patients with no surgical stress anesthetized with halothane at 1 MAC (0.75%) had a mean CBFxenonof 40 plus/minus 3 ISI units during normocapnia. CBFxenonduring isoflurane anesthesia at 1 MAC (1.2%) was lower (33 plus/minus 3 ISI units) than that during administration of halothane (F = 9.585, P < 0.01, ANOVA). Hypocapnia decreased CBFxenonsignificantly (F = 40.353, P < 0.0001, ANOVA) during both anesthetics. In halothane-anesthetized patients, hypocapnia decreased CBFxenonto 24 plus/minus 3 ISI units (P < 0.001), and in isoflurane-anesthetized patients to 13 plus/minus 2 ISI units (P < 0.0001). The carbon dioxide responses during halothane and isoflurane anesthesia were calculated to 1.4 and 1.5 ISI units mmHg1carbon dioxide, respectively, with no significant difference. The mean CBFxenonvalues in all four groups are presented in Figure 1.
Effects of Halothane and Isoflurane on the Three-dimensional CBF Distribution
The carbon dioxide level did not significantly alter the regional distribution of CBF. Halothane and isoflurane anesthesia resulted in different patterns of CBF distribution (F = 6.520, P < 0.0001, ANOVA). During isoflurane anesthesia, the regions with the highest CBF were the thalamus and basal ganglia (10–15% greater than average), followed by pons (7–10% greater than average). On the other hand, during halothane anesthesia, the highest rCBF levels were in the occipital lobes (about 10% greater than average). The rCBF distribution in percent of mean was higher for isoflurane in the pons (P < 0.01) and the thalamic (P < 0.05) regions and higher for halothane in the occipital (P < 0.005) region. Regardless of the anesthetic used, the CBF level in the cerebellum was approximately 10% greater than average. The CBF distributions are presented in Figure 2and representative SPECT scans in Figure 3, rCBF values, calculated from the two- and three-dimensional blood flow, are presented in Figure 4.
Volatile anesthetics have long been known to affect the cerebral circulation, but it has been difficult to determine how blood flow through various parts of the human brain is influenced by these agents. CBF may be determined via different methods, each with its own inherent anatomic sampling bias. The technique by Eintrei et al.,  using washout of133xenon employed directly to the brain surface, measures flow mainly in the superficial layer of the cortex. Intracarotid or intravenous injection of133xenon and extracranial recording to the clearance curves with scintillation detectors also reflect mainly cortical blood flow, whereas Kety and Schmidt's method  measures the mean CBF of the entire brain.
These methods have been used to investigate the effects of halothane and isoflurane on human CBF revealing different results. Using the Kety and Schmidt method, no real difference could be detected between the anesthetics. [15,16] On the other hand, cortical blood flow measured with extracranial recording of133xenon wash-out was higher during halothane anesthesia compared to isoflurane.  These differences were even greater when cortical blood flow was measured with the technique of Eintrei et al.  If, indeed, there is no difference in the effects on total CBF in humans between the volatile agents, the decrease in cortical CBF during isoflurane anesthesia would imply a redistribution of CBF with an increased flow in subcortical gray matter regions of the brain. Such a redistribution has been found in studies comparing the effects of halothane and isoflurane on rCBF in rabbit  and rat. .
Our measurements of the rCBF distribution during normocapnia showed that isoflurane caused a relative flow distribution with higher CBF levels in pons and the subcortical structures (thalamus and basal ganglia) as compared to cortex. We also found a relative increase in flow to the occipital cortex and cerebellum during halothane anesthesia. Thus, our results support the theory that there is a redistribution of the CBF to subcortical structures during isoflurane anesthesia as implied by Hansen et al.  This redistribution of the flow is minor and, if one compares the flows in absolute values, the redistribution is hardly visible (Figure 4).
The reason for the selective difference in relative flow through different regions during isoflurane and halothane anesthesia is unknown. One theory is a direct influence on the cerebral arteries,  which differs between the vessels within the cerebrovascular tree as found for oxygen in the rabbit  and proposed by Hansen et al.  Another possible reason is based on the fact that the normal brain has a tight coupling between brain metabolism and flow.  The tight connection between CBF and brain function seems to be mediated mainly by chemical factors, particularly carbon dioxide, released by the tissue acting locally on the neighboring cerebral vascular smooth muscle.  During exposure to halothane and isoflurane, changes in perivascular carbon dioxide modulate the tone of human pial arteries,  and in vivo a flow metabolism relationship is evident during both isoflurane and halothane anesthesia in the rat.  Assuming that volatile anesthetics have different impacts on metabolism in various brain regions,  differences would result also in the relative flow distribution.
In the current study, we found that mean CBFxenonduring anesthesia with halothane was higher than when isoflurane was administered. This observation is in accordance with former human studies mainly reflecting cortical blood flow [2,21] as well as studies in the cat  and the rabbit  but in contrast to global CBF findings in the rat, [4,5] where no difference between the two anesthetics was found. The cortical CBFxenonvalues found in our study were less than those previously reported, probably because our patients were completely unstressed. [21,23] Anesthesia was induced with propofol at least 1 h before the CBF measurement. Propofol is rapidly redistributed, and the cerebral effect of propofol should be negligible at the CBF measuring time. However, because propofol decreases human CBF  with a maintained carbon dioxide response,  the possibility that propofol might have contributed to the low CBF values cannot be ruled out, even if such an influence seems unlikely.
Hypocapnia reduced CBFxenonduring both halothane and isoflurane anesthesia. Hypocapnia directly constricts human pial arteries in vitro  and, if this is a general feature on the arteriolar level, it also should decrease CBF as found by Kety and Schmidt. [14,27] They found a carbon dioxide reactivity for the whole brain of 1 ml *symbol* 100 g1*symbol* min1*symbol* mmHg1in spontaneously breathing volunteers. In contrast, Messeter et al.,  using a method reflecting mainly cortical areas, reported a carbon dioxide reactivity of 2.2 ISI units mmHG , which, assuming a cortical lambda value of 0.8, would correspond to about 1.8 ml *symbol* 100 g1*symbol* mmHg1carbon dioxide change. The direct vasodilation of halothane was less pronounced in vitro during hypocapnia as compared to normocarpia.  This is in accordance with previous in vivo investigations, in which the carbon dioxide response was intact during halothane anesthesia. [29,30] Our study shows a preserved carbon dioxide reactivity, because a lowering in Pa sub CO2of 1.4 kappa Pa (1.4 (11 mmHg) diminished the cortical CBF sub xenon by 16 ISI units, which equals a carbon dioxide response of 1.4 ISI units per 1 mmHg change in PaCO2.
The carbon dioxide response has been reported to be intact in rabbit during isoflurane anesthesia.  In isolated human pial arteries, isoflurane had a smaller relaxant effect than halothane.  At hypocapnia, isoflurane was a vasoconstrictor,  an effect also found in vivo in rabbits,  suggesting a greater carbon dioxide response than with halothane. We found that the carbon dioxide response was preserved with 1.5 ISI units per 1 mmHg change in PaCOsub 2 during administration of isoflurane, which was in accordance with previous findings in humans.  It did not significantly differ from the effects of halothane, in contrast to the findings in rabbits.  in which the carbon dioxide response was greater during isoflurane compared to halothane.
Isoflurane profoundly decreased CBFxenonduring hypocapnia and, consequently, may reduce the cerebral blood volume, thereby reducing an increased intracranial pressure. [32,33] Such a decrease is especially welcome during neurosurgical procedures, because a reduction in brain volume improves the working condition for the surgeon. Furthermore, a reduction of the cerebral blood volume may be vital in patients with reduced intracranial compliance. The lowest mean CBFxenonvalues were seen during hyperventilation and isoflurane anesthesia, 8 ISI units, a value that may cause speculations concerning whether it could cause ischemia. However, the ISI unit may give an underestimation of the CBFxenonin the low-flow situation because of its inherent assumption of a mean xenon solubility in brain tissue equal to that in blood (lambda = 1). In the low-flow situation, the flow component in white brain matter (lambda = 1.5) increasingly influences the CBF measurement. Thus, it can be shown that 8 ISI units may correspond to a cortical CBF of about 10–12 ml *symbol* 100 *symbol* g *symbol* min . It also should be noted that measurements of low CBFxenonvalues have a higher measurement error because of methodologic reasons.  This is because there may be an increased delay between the recording of the lung and the brain clearance curves due to the lower CBF. Such an increased delay will decrease the calculated CBFxenonwith about 2–3% per second of delay if the ISI value is about 50.  It can be shown that the size of this kind of error is inversely proportional to the true ISI value of the CBFxenonmeasurement.  Thus, for an ISI value of about 10, the effect of 1 s of increased delay would result in a 10–15% underestimation of the true ISI value. However, this delay was not possible to measure with our equipment, even though it certainly exists. Because isoflurane, in addition, lowers the metabolic demand in the brain,  the low CBFxenonvalues found in the current study does not necessarily imply an ischemic condition. .
In conclusion, we found a difference in the human cerebral flow distribution with higher flows in subcortical regions during isoflurane than during halothane anesthesia at 1 MAC. However, despite this redistribution, isoflurane anesthesia resulted in a lower mean CBF sub xenon value than did anesthesia with halothane.
The authors thank the staff at the outpatient operating department. University Hospital of Lund, for their support, and Majvi Persson, for assistance with single photon emission computer-aided tomography (SPECT) scans and isotope delivery.