Cerebral Oxygen Extraction and Autoregulation during Extracorporeal Whole Body Hyperthermia in Humans

This study was approved by the ethics committee of the University Medical Center Utrecht, The Netherlands, and written informed consent was obtained. Patients were enrolled in the current protocol ancillary to a pilot study investigating the efficacy and safety of extracorporeal whole body hyperthermia (EWBH) for the treatment of chronic hepatitis C virus infection. This pilot study was designed after encouraging hepatitis C viral load–related observations during the experimental use of systemic hyperthermia for treatment of human immuno-deficiency virus infection. 1–3Patients aged 18–65 yr with chronic hepatitis C virus infection, genotype 1, were eligible for study inclusion. All patients had failed to show a sustained response to interferon alfa therapy and were required to have Child-Pugh classification A (no or minimal cirrhosis). 25Patients were screened for central nervous system abnormalities, cardiac disease, pulmonary disease, and other major liver or renal disease. Patients were ineligible if abnormalities were present. A history of substance abuse was not an exclusion criterion.

Thirteen patients were treated between August 2001 and April 2002 with a single session of EWBH (series A). Six additional patients and four patients from the first series were (re)treated between December 2002 and February 2003 (series B). Treatment again consisted of a single session of EWBH but was now followed after 6 weeks by the initiation of 12 months of antiviral therapy with interferon alfa and ribavirin.

All EWBH treatments were performed by the same team of anesthesiologists and perfusionists throughout the study period. General anesthesia was induced with propofol–fentanyl–rocuronium and was maintained using a target-controlled propofol infusion pump (Diprifusor; Astra-Zeneca, Ridderkerk, The Netherlands). The trachea was intubated, and the lungs were ventilated with oxygen in air. During a session of EWBH, core temperature was increased in approximately 90 min to a target of 41.8°± 0.15°C under general anesthesia, using an extracorporeal heater–cooler device with a bifemoral venovenous circuit (TEMET System 1000; First Circle Medical, Minneapolis, MN). Before the start of the extra-corporeal bypass, patients were heparinized. After 120 ± 10 min of hyperthermic plateau, patients were cooled to 39°C before the bypass was discontinued. After the procedure, patients were transported to the intensive care unit. Propofol sedation was continued until planned extubation 2–4 h later.

Anesthetic management differed with respect to two important aspects in patients treated during EWBH series A and B. First, during treatment series A, patients were ventilated in a volume-controlled manner to maintain normocapnia at a target end-tidal carbon dioxide of approximately 4.0% during normothermia, and the ventilator settings subsequently remained unchanged throughout the procedure. This resulted in a high end-tidal carbon dioxide but a normocapnic uncorrected arterial partial pressure of carbon dioxide (PaCO²), measured at 37°C, during hyperthermia (alphastat principle). During treatment series B, patients were also normocapnic during normothermia, but as body temperature was increased, the minute volume of ventilation was gradually increased to maintain an approximately normal end-tidal carbon dioxide, resulting in a hypocapnic uncorrected PaCO²during hyperthermia (pH-stat principle). 26Second, during EWBH treatment series A, changes in the propofol infusion target were made at the discretion of the attending anesthesiologist. During treatment series B, a routine gradual increase of the propofol infusion target by 1.5 mg/l during warming was implemented in the anesthesia protocol. This routine adjustment of the infusion rate was deemed necessary because blood concentrations of propofol decreased as temperature was increased.

Invasive monitoring of systemic and pulmonary hemo-dynamics was used. Temperature monitoring sites included the pulmonary artery, the nasopharynx, the esophagus, and the rectum. In addition, a right retrograde jugular bulb catheter (5.5-French Opticath Oximetrix; Abbott Critical Care Systems, Chicago, IL) was used to measure jugular venous temperature, pressure, oxy-hemoglobin saturation, and lactate. An arterial to jugular venous difference of lactate less than −0.37 mm was considered abnormal. 27The position of the catheter tip was verified by a lateral skull x-ray. The blood flow velocity in the middle cerebral artery was measured using transcranial pulsed Doppler ultrasonography (2-MHz probe, Multi-Dop T; DWL, Sipplingen/Bodensee, Germany). The position of the ultrasound transducers was fixed using a metal head frame (series A) or a special headband (series B). Mean peak flow velocities from both hemispheres were averaged. Data were stored at 0.1 Hz and at 250 Hz for off-line analysis using POLY Physiologic Analysis Package (Inspektor Research Systems, Amsterdam, The Netherlands). Laboratory parameters were measured at normothermic baseline, at fixed intervals during warming, and at hyperthermic plateau.

Static autoregulation tests were performed during normothermic flow over the extracorporeal bypass circuit (T1) and during hyperthermic plateau (T2), using a ramped phenylephrine infusion. The static rate of auto-regulation was calculated as the ratio of the percent change in cerebrovascular resistance by the percent change in mean arterial pressure, with cerebrovascular resistance estimated as the ratio of mean arterial pressure by cerebral blood flow velocity (CBFV). 28–30A static rate of autoregulation of 1 indicates perfect adaptation of cerebral resistance vessels to changes in mean arterial pressure, whereas a value close to 0 indicates complete absence of cerebrovascular reactivity. In addition to these formal tests of static pressure–flow auto-regulation, transient hyperemic response tests were performed at fixed stages during the EWBH procedure. Transient hyperemic response testing involves measurement of changes in middle cerebral artery flow velocity during and after the release of a 10-s compression of the ipsilateral common carotid artery. If cerebral autoregulation is intact, the decrease in perfusion pressure in the middle cerebral artery at the onset of compression triggers vasodilatation in the distal vascular bed, resulting in a transient increase in the flow velocity after the release of compression. The transient hyperemic response ratio (THRR) was calculated as the ratio of the mean peak flow velocity of the Doppler waveform immediately after the release of compression by that of the waveform immediately preceding the compression. A THRR value close to 1 indicates absence of a vascular response, whereas higher values indicate increasing levels of vasoreactivity. The THRR reflects changes in the static rate of autoregulation and provides a valid measure for assessing graded impairments of cerebral pressure–flow autoregulation. 30,31 

Systemic and cerebral physiologic parameters recorded during normothermia and hyperthermia were compared using paired-samples t  tests. In addition, changes in these parameters from T1 to T2 were tested for differences between treatment series A and B using independent-samples t  tests. With respect to cerebrovascular reactivity testing, the THRR was considered the primary measure of pressure–flow autoregulation in this study because transient hyperemic response tests could be repeatedly performed during various stages of the EWBH procedure. Therefore, the THRR was expected to give a more precise estimate of cerebral vasomotor responses than the static rate of autoregulation. Because changes in PaCO², mean arterial pressure, and propofol blood concentration during EWBH treatment might confound the relation between temperature and pressureflow autoregulation, 29,31a mixed-effects regression analysis was used to model possible confounding by these parameters, taking into account the nesting of repeated observations within an individual. To allow for differences in vasoreactivity between individual subjects, an intercept was included in the model as a random effect. The crude and adjusted effect estimates of the association of temperature with the THRR are presented with 95% confidence intervals. Data in the text are presented as mean ± SD.

Thirteen sessions of EWBH treatment were performed during series A, and 10 sessions were performed during series B, in a total of 19 patients (age 44 ± 9 yr, 14 [74%] male). Table 1shows various anesthetic management–related parameters at T1 (normothermia) and T2 (hyperthermia). End-tidal carbon dioxide and uncorrected PaCO²values show the effect of the changes in ventilation management that were introduced between treatment series A and B. Likewise, the differences in propofol infusion targets show the effect of protocol changes between series A and B. The calculated target blood concentration of propofol grossly overestimated the true blood concentration at T2.

Table 2shows systemic parameters. During hyperthermia, systemic vascular resistance decreased considerably, despite the use of increased infusion rates of phenylephrine. As a consequence, cardiac output and heart rate increased, and mean arterial pressure decreased. Simultaneously, systemic oxygen consumption increased by 47% and 23% for treatment series A and B, respectively. The hemodynamic and metabolic changes were more outspoken during treatment series A (high–carbon dioxide group) than during series B (low–carbon dioxide group), although these differences did not reach statistical significance.

Table 3shows cerebral parameters. For series A, cerebral oxygen extraction (arterial to jugular venous difference of oxygen content) decreased 2.2-fold, compared with a simultaneous 2.0-fold increase in CBFV. For series B, cerebral oxygen extraction decreased 1.6-fold, and CBFV increased 1.5-fold. For all 23 treatment sessions of series A and B combined, jugular venous oxygen saturations below 50% were observed in two instances at T1 and never at T2. Signs of cerebral lactate production (arterial to jugular venous difference of lactate < −0.37 mM) were observed in three instances at T1 and in five instances at T2. The simultaneous occurrence of jugular venous oxygen desaturation and abnormally low arterial to jugular venous difference of lactate was not observed. Figure 1shows the cerebral and systemic arterial to venous oxygen extraction at different temperatures during the EWBH procedures. For both treatment series A and B, the systemic oxygen extraction decreased linearly with increasing temperature. In contrast, cerebral oxygen extraction decreased in a nonlinear manner, with the greatest decrease occurring at temperatures above approximately 40°C.

Table 4shows measures of cerebrovascular reactivity. Cerebral pressure–flow autoregulation was assessed by 37 static autoregulation tests during 17 EWBH procedures and by 220 transient hyperemic response tests during 14 procedures. The static rate of autoregulation and the THRR significantly decreased from T1 to T2 in both series A and B. Figure 2shows the THRR at fixed temperatures during the EWBH procedures (series A and B combined). The THRR did not decrease until the jugular venous blood temperature reached approximately 40°C. Subsequently, the crude mixed-effects regression model (table 5) showed a decrease in the THRR of 0.07 (95% confidence interval, 0.05–0.09) per degree temperature increase (P = 0.000). After adjustment of the crude association for variations in PaCO², mean arterial pressure, and propofol blood concentration during the EWBH procedure, the association remained (P = 0.000).

Experimental treatment using EWBH in patients with chronic hepatitis C virus infection under propofol anesthesia resulted in a decreased cerebral oxygen extraction that was accompanied by an increase in CBFV of a similar magnitude. These effects were more pronounced when end-tidal carbon dioxide was allowed to increase during warming (series A), compared with when end-tidal carbon dioxide was more rigidly maintained (series B). Second, heating resulted in impaired cerebrovascular responses to both a vasoconstrictor stimulus (static autoregulation testing) and a vasodilator stimulus (transient hyperemic response testing), indicating impaired pressure–flow autoregulation. This effect remained after adjustment for changes in PaCO², mean arterial pressure, and propofol blood concentration. Together with systemic findings, these data indicate that a transient partial vasoparalysis develops during profound hyperthermia.

There are several methodologic issues that must be discussed. First, we measured arterial to jugular venous difference of oxygen content and found an apparent decrease in cerebral oxygen extraction. However, this could be a spurious finding if cerebrovenous blood would be grossly contaminated with extracerebral venous blood, known to have a higher oxygen saturation, during hyperthermia. In the current study, of 164 fiber-optic jugular venous oxygen saturation readings, 106 were verified by cooximetry using a laboratory-based blood gas analyzer. Cooximetry values were on average 3.5 ± 7.4% higher than fiberoptic values, and these differences were independent of temperature. These slightly higher values may indeed be explained by admixture of extracerebral blood when a sample is actively drawn from the jugular bulb catheter. 32Because this effect was still apparent and independent of temperature, we consider it unlikely that gross contamination occurred during hyperthermia. Second, the reduction of cerebral oxygen extraction was accompanied by an increase in CBFV by a factor of almost equal magnitude. One interpretation of these observations is that cerebral oxygen consumption did not change appreciably during hyperthermia. However, the increase in peak flow velocity may have underestimated the true increase in erythrocyte flux during hyperthermia. This could have occurred if the insonated segment of the middle cerebral artery had increased in diameter over time. Studies that have specifically looked at the cross-sectional area of the middle cerebral artery have concluded that the diameter of this large vessel is grossly unaffected by changes in temperature, PaCO², and blood pressure. 33–37Also, changes in hematocrit and blood viscosity in theory may confound the relation between peak flow velocity and erythrocyte flux. An in vitro  study examining the effect of varying temperature (in a range of 19°–37°C) and hematocrit (in a range of 0.05–0.54) showed that the linear relation between flow and velocity was not affected. 38Another study in anesthetized patients reported on the effect of a sudden decrease in hematocrit from 0.38 to 0.30 and concluded that CBFV well reflects true cerebral blood flow changes after hemodilution. 39In the current study, the effect of hemodilution by the circuit priming had stabilized before the first (normothermic) measurements. Nonetheless, we cannot rule out that small increases in middle cerebral artery diameter occurred during either EWBH series A or EWBH series B, which would result in a significant underestimation of true erythrocyte flux.

The current data indicate that the cerebrovascular bed during hyperthermia is still responsive to a dilatory stimulus, such as a brief carotid compression, at temperatures up to approximately 40°C (fig. 2). At higher temperatures, however, we found that cerebral pressureflow autoregulation becomes progressively impaired. There are few studies assessing cerebral vasomotor responses during whole body hyperthermia, and the ones that are available have shown conflicting results. A study in dogs concluded that the main cerebral autoregulation system is paralyzed during whole body hyperthermia at 41.5°C. 21Similarly, a recent study in humans has shown an increased pressure–flow dependency during an inflow of hyperthermic blood into the brain during rewarming from hypothermic cardiopulmonary bypass. 24These reports of hyperthermia-induced cerebral vasoparalysis contrast with an animal experiment that indicated that the dilated cerebrovascular bed during hyperthermia is still responsive to indomethacin as a constrictor stimulus. 20This finding, as well as the results of the current study, suggests that hyperthermia-induced vasoparalysis is partial rather than complete. Furthermore, the association between temperature and loss of cerebrovascular reactivity may be nonlinear, as suggested by figure 2and by a report of improved dynamic autoregulation during a 0.4°C temperature increase in awake human volunteers submersed in hyperthermic baths. 23The results of the adjusted mixed-effects regression analysis suggest that the THRR becomes also increasingly depressed as PaCO²increases, mean arterial pressure decreases, and propofol blood concentration increases, although none of these effects reached statistical significance (table 5). The current findings may have clinical implications, as transient cerebral vasoparalysis during periods of induced hyperthermia or fever could potentially predispose to neurologic injury. Although the blood–brain barrier is most likely grossly preserved at temperatures below 42–43°C, 21,22it has been suggested that brain edema may occur easily when arterial blood pressure fluctuates excessively and cerebral autoregulation is absent.

In conclusion, whole body hyperthermia is associated with decreased cerebral oxygen extraction and increased cerebral blood flow velocities in patients under propofol anesthesia. These findings are more marked when end-tidal carbon dioxide is allowed to increase compared with when it is more rigidly controlled. In addition, hyperthermia is associated with impaired cerebrovascular responses to both blood pressure increase and carotid occlusion, indicating that transient partial cerebral vasoparalysis develops when temperature exceeds approximately 40°C.

Special Note: The research presented in this article was performed as an ancillary protocol to pilot trials on the safety and efficacy of extracorporeal whole body hyperthermia for treatment of chronic hepatitis C virus infection. This project has been halted by the Utrecht University Medical Center after an internal investigation concluded that the principal investigator of the therapeutic trials had failed to report serious side effects, in particular peripheral neuropathy, to the hospital’s medical ethics committee and that funds had not been properly declared from personal agreements with the commercial study sponsor. The authors related to this manuscript have not been implicated with respect to these issues.

The authors thank John C. Drummond, M.D., F.R.C.P.C. (Professor and Chair, Department of Anesthesiology, University of California, San Diego, California), and Gert W. van Dijk, M.D., Ph.D. (Staff Neurologist, Department of Neurology, University Medical Center, Utrecht, The Netherlands), for valuable suggestions and comments while reviewing the manuscript.