The Effect of Diabetes on the Interrelationship Between Jugular Venous Oxygen Saturation Responsiveness to Phenylephrine Infusion and Cerebrovascular Carbon Dioxide Reactivity

Methods

This study was approved by the ethics committee of our institution, and written informed consent was obtained from all patients. Forty diabetic patients consecutively scheduled for elective coronary artery bypass graft surgery were studied. Patients were defined as having diabetes mellitus if their medical records showed a diagnosis of type 2 diabetes and current medical treatment with antidiabetic therapy (diet, glibenclamide, or insulin therapy). Duration of disease was defined as time elapsed since commencement of medical treatment. Patients with a history of cerebrovascular disease, psychiatric illness, renal disease (creatinine concentration >2.0 mg/dL), or active liver disease (glutamine oxaloacetate transaminase or glutamine pyruvate transaminase >50 U/dL) were excluded. Patients with moderate-to-severe atherosclerotic lesions in the ascending aorta or carotid artery stenosis, confirmed by preoperative ultrasonography and magnetic resonance imaging, were also excluded. Forty age-matched nondiabetic patients consecutively scheduled for elective coronary artery bypass graft surgery served as controls.

Because glycosylated hemoglobin (HbA1c) (normal value 4.5%–5.8%) is an indicator of blood glucose control, preoperative HbA1c levels were determined in all diabetic patients.

All patients received 10 mg of diazepam (orally) 1 h before anesthesia. Anesthesia was induced with IV midazolam (0.2 mg/kg), fentanyl (20 μg/kg), and vecuronium (0.2 mg/kg), and the trachea was intubated. After anesthesia induction, 4–6 mg · kg⁻¹ · h⁻¹ of propofol was infused IV and was continued until the patients arrived in the intensive care unit.

Intermittent administration of fentanyl was performed for additional analgesia. The maximum fentanyl dosage was 50 μg/kg. Muscular relaxation was obtained from vecuronium. No volatile anesthetics were given. After anesthesia induction, a pulmonary arterial catheter (Vigilance®, Swan-Ganz CCO Thermodilution Catheter, Baxter, Irvine, CA) was inserted through the right internal jugular vein. For continuous monitoring of SjvO₂, a 4F fiberoptic oximetry oxygen saturation catheter (Dual-lumen oximetry catheter®, Baxter) was inserted into the right jugular bulb using a modified Seldinger technique. This catheter was connected to an analysis system (Explorer™ system, Baxter) and was calibrated in vivo by drawing a blood sample from the catheter. The position of the jugular bulb catheter tip was verified by radiograph. The correctly positioned catheter tip should lie cranial to a line extending from the atlanto-occipital joint space and caudal to the lower margin of the orbit. SjvO₂ data were collected and processed with a monitor-computer interface and were displayed and stored every 5 s in an Apple Macintosh computer (Apple Macintosh Computer Company Ltd, Cupertino, CA). The catheter was calibrated every 20–30 min.

The partial pressures of arterial and jugular venous blood gases were analyzed using a Stat Profile Ultmita® (NOVA Biomedical Co, Boston, MA) and a CO-oximeter (OSM3, Hemoximeter®, Radiometer Co, Copenhagen, Denmark). The lungs were ventilated with 50% oxygen and 50% nitrogen. End-tidal CO₂ (Petco₂) was monitored (Ultima®, Datex, Helsinki, Finland) and maintained between 35 and 40 mm Hg. Tympanic membrane temperature was continuously monitored by Mon-a-Therm® (Mallinckrodt Co, St Louis, MO).

CPB was primed with a crystalloid, nonglucose containing solution, and a nonpulsatile pump flow rate of 2.2–2.5 L · min⁻¹ · m⁻² was maintained. A membrane oxygenator and a 40-μm arterial line filter were used, and Paco₂, uncorrected for temperature, was maintained between 35 and 40 mm Hg by varying fresh gas flow to the membrane oxygenator (α-stat regulation).

Target tympanic membrane temperature was 34.5°C–36.0°C, whereas the limit on maximal inflow temperature was set at 37.0°C. Hematocrit was maintained at >0.20 during CPB, with the addition of blood as required.

Measurement of cerebrovascular CO₂ reactivity was performed after anesthesia induction, before the start of surgery, and during the stable hemodynamic period (approximately 20–30 min after the induction of anesthesia). A 2.5-MHz pulsed transcranial Doppler (TCD) probe was attached to the patient’s head at the right temporal window, and the mean blood flow velocity of the middle cerebral artery (Vmca) was measured continuously (Hewlett Packard SONOS 5500®, 2.5-MHz transducer, Andover, MA). After the signals were identified at a depth of 45–60 mm, the probe was fixed using a probe folder so as not to change the insonation angle. Vmca value was recorded at end-expiration.

After measuring the baseline Vmca, arterial blood gases, and cardiovascular hemodynamic values, Petco₂ was increased by reducing the ventilatory frequency by 2–5 breaths/min. This resulted in an increase in Paco₂ of approximately 6–9 mm Hg within several minutes. Measurements were repeated when Petco₂ increased and remained stable for 10–15 min.

The cerebral vasodilatory response to hypercapnia in each patient was calculated as both the absolute change in Vmca (cm · s⁻¹ · mm Hg⁻¹) and the percentage change in Vmca (percentage of baseline Vmca/mm Hg) per millimeter of mercury change in Paco₂ using the following formula (9):

and

where ΔVmca is the difference between flow velocity after Paco₂ increase and baseline flow velocity, and ΔPaco₂ is the difference between final Paco₂ and baseline Paco₂.

The SjvO₂ response to phenylephrine was measured during the stable CPB period (approximately 30 min after initiation of CPB). After measuring the baseline partial pressure of the arterial and jugular venous blood gases and cardiovascular hemodynamic values, MAP was increased by repeated administration of a 10-μg bolus of phenylephrine until reaching 100% of baseline values, with an allowed maximum of 100 mm Hg (4). The study end-point was defined as a MAP of 100% of baseline values that was maintained at least 60 s.

Pre- and posttreatment SjvO₂ were analyzed in sampled blood. Obtained data were later analyzed by an individual who was blinded as to patient groupings. Cerebral perfusion pressure (CPP) was defined as CPP = MAP − internal jugular venous pressure, which was measured by monitoring pressure at the distal end of the SjvO₂ catheter.

All data were expressed as mean ± sd. Data were examined by the Bartlett test to determine whether the variance was normally distributed among groups. Unpaired t-test or Mann-Whitney U-test was used for comparison of demographic data. Paired t-test was used for comparison between prevariables and postvariables. SjvO₂ values in the diet therapy, glibenclamide, insulin, and control groups were analyzed using one-factorial repeated-measures analysis of variance. For multiple comparisons of the four groups, the Scheffé test was used. Simple linear regression was used for associations between absolute CO₂ reactivity and the mean slope of SjvO₂ versus CPP for increasing CPP, HbA1c and absolute CO₂ reactivity, and HbA1c and the mean slope of SjvO₂ versus CPP for increasing CPP. Statistical significance was set at P < 0.05. All calculations were performed on a Macintosh computer with SPSS (SPSS Inc, Chicago, IL) and StatView 5.0 software packages (Abacus Concepts Inc, Berkeley, CA).

Results

Table 1 shows the demographic data from the two groups. There were no significant differences between them. There was no significant difference in phenylephrine dosage required to increase MAP to 100% of baseline values (control group, 65 ± 28 μg; diabetic group, 72 ± 28 μg; P = 0.21). There was no significant difference in time from the induction to MAP reaching 100% of baseline values after the administration of phenylephrine (control group, 7.9 ± 2.5 min; diabetic group, 8.2 ± 2.5 min; P = 0.69).

There were no significant differences in internal jugular venous pressure, tympanic membrane temperature, SjvO₂, Paco₂, Hb concentration, or pump flow between the two groups at either pre- or posttreatment time points (Table 2).

Because of the heterogeneous nature of the diabetes mellitus population in this study, we subdivided the diabetes group into three subgroups (Table 3). SjvO₂ values in the insulin group did not change after administration of phenylephrine. In contrast, SjvO₂ values in the other three groups increased after phenylephrine infusion (Table 3). There was a significant difference in SjvO₂ values between the insulin group and the other three groups after phenylephrine infusion. In addition, there were significant differences in the mean slopes of SjvO₂ versus CPP for increasing CPP, absolute and relative CO₂ reactivity, and HbA1c between the insulin group and the other three groups (P < 0.01). There was a significant difference in absolute CO₂ reactivity between the diabetic and control groups (control, 2.8 ± 0.7 cm · s⁻¹ · mm Hg⁻¹; diabetic, 2.2 ± 1.1 cm · s⁻¹ · mm Hg⁻¹; P = 0.02). Among the diabetic subgroups, absolute CO₂ reactivity in the insulin group was less than that in the other two groups.

There was a good correlation between absolute CO₂ reactivity and the mean slopes of SjvO₂ versus CPP for increasing CPP (y = 7.9x + 1.66; r = 0.54; P < 0.0001; Fig. 1) and an inverse correlation between absolute CO₂ reactivity and HbA1c (y = −0.8x + 8.1; r = 0.63; P < 0.0001; Fig. 2) or the mean slope of SjvO₂ versus CPP for increasing CPP (y = −8.0x + 6.9; r = 0.422; P < 0.0001; Fig. 3).

Cerebrovascular CO₂ reactivity was only seen in the insulin group, and thus, we examined these relationships in this group (Table 3). There was a good correlation between absolute CO₂ reactivity and the mean slope of SjvO₂ versus CPP for increasing CPP (y = 8.2x + 0.79; r = 0.54; P = 0.043) and an inverse correlation between absolute CO₂ reactivity and HbA1c (y = −1.6x + 9.3; r = 0.81; P = 0.0004) or the mean slope of SjvO₂ versus CPP for increasing CPP (y = −17.4x + 8.3; r = 0.595; P = 0.024).

Discussion

We found that cerebrovascular CO₂ reactivity in diabetic patients was associated with the response of SjvO₂ to phenylephrine infusion and that decreased cerebrovascular CO₂ reactivity and the mean slope of SjvO₂ versus CPP for increasing CPP were associated with HbA1c.

A previous study showed that increasing MAP had no effect on SjvO₂ in insulin-dependent patients (4). In contrast, SjvO₂ values in control patients or diabetics treated with diet or glibenclamide were increased after phenylephrine infusion during tepid CPB. To clarify the mechanisms for this lack of effect of increased MAP on SjvO₂ in insulin-dependent patients, we examined the relationship between cerebrovascular CO₂ reactivity and the response of SjvO₂ to phenylephrine infusion in diabetic patients.

There have been several reports studying cerebrovascular CO₂ reactivity in diabetic patients (6–8,10). Kawata et al. (10) examined the effects of diabetes mellitus on CO₂ reactivity using TCD and found that the relative values of CO₂ reactivity in diabetic patients were equivalent to those in control patients during isoflurane anesthesia. In contrast, Dandona et al. (7) reported that under awake conditions, there were significant variations in CBF after administra-tion of 5% CO₂ in insulin-dependent diabetics when compared with normal subjects and thus concluded that diabetics had diminished cerebrovascular reserve and were unable to compensate with increased CBF when required. Rodriguez et al. (8) reported that under awake conditions, when compared with control subjects, acetazolamide administration significantly impaired the percentage of global CBF increment in four insulin-dependent diabetic patients and gave a borderline response in two insulin-dependent diabetic patients. This discrepancy might have been partly attributable to differences in demographic data or anesthetic method. Kawata et al. (10) examined CO₂ reactivity using TCD during isoflurane anesthesia. In contrast, we examined CO₂ reactivity under propofol-fentanyl anesthesia in the present study. Nishiyama et al. (11) investigated the effects of different anesthetic drugs on cerebrovascular CO₂ reactivity and found that different anesthetic drugs exert different effects.

There have also been reports regarding the relationship between cerebrovascular CO₂ reactivity and duration of diabetes, serum glucose concentration, and HbA1c (7). Rodriguez et al. (7) reported that duration of diabetes, serum glucose concentration, and HbA1c did not correlate with the percentage of postacetazolamide global CBF changes. Their findings are inconsistent with our results. This discrepancy might be partly attributable to differences in demographic data. Rodriguez et al. (8) used subjects with a mean age of 30 years, whereas in the present study mean subject age was 61 years. Age is one of the factors associated with cerebral vasodilatory response to hypercapnia (12). Hartl and Furst (12) reported that absolute and relative mean CO₂ reactivities in elderly subjects were markedly less than those in younger subjects. However, Yamamoto et al. (13) examined the effects of aging on cerebral vasodilatory response to hypercapnia and found that average CBF in elderly patients was approximately 10%–20% less than in younger volunteers, whereas the vasodilatory response to hypercapnia in elderly patients without risk factors was similar to that in younger volunteers.

We confirmed a relationship between cerebrovascular CO₂ reactivity and the response of SjvO₂ to phenylephrine infusion. White and Markus (14) showed that there was a correlation between autoregulatory index and CO₂ reactivity. Pallas and Larson (6) provided a possible mechanism for impaired CO₂ reactivity or poor response to CBF. They suggested that the cerebral vasculature of the diabetic has both morphological and functional changes when compared to the nondiabetic. These may include endothelial hyperplasia, thickening of the basement membrane, and loss of normal autoregulatory function in response to changes in perfusion pressure in addition to metabolic factors. The endothelial cellular-dependent functions that seem to be the most affected by the diabetic state were assumed to be linked to the nitric oxide pathway.

HbA1c in diabetic patients indicated that blood sugar was controlled for one to two months before the study. Aroca et al. (15) reported that HbA1c predicts microangiopathy in diabetic patients. In a previous study, we found that cerebrovascular CO₂ reactivity was inversely related to HbA1c. This indicates that the primary cause of microvascular disease is hyperglycemia itself and that poorly controlled blood sugar induces disturbed cerebrovascular microcirculation, rather than the severity of diabetes mellitus. This assumption was confirmed by Stratton et al. (16) who demonstrated in patients with type II diabetes that the risk of diabetic complications, such as macrovascular and microvascular disease, was strongly associated with previous hyperglycemia. We speculated that because insulin-dependent patients have endothelial cellular-dependent dysfunction because of previous hyperglycemia, administration of phenylephrine had no vasoconstrictive effect on cerebral vasculature.

The usefulness of measuring SjvO₂ in the perioperative period remains controversial (17,18). However, a report by Bell et al. (19) showed an increase in interstitial brain adenosine and xanthine during jugular venous desaturation in humans after traumatic brain injury and demonstrated that SjvO₂ might be useful for detecting brain ischemia.

In hypertensive patients, the flow-pressure autoregulatory curve is shifted rightward, and higher levels of CPP are required to maintain adequate CBF. Because most patients in this study had hypertension, this might have affected our results.

TCD facilitates the measurement of CBF velocity in a noninvasive and continuous manner. However, TCD is not a direct measurement of CBF but rather a measurement of flow velocity. Using blood flow velocity to estimate CBF makes the assumption that middle cerebral artery (MCA) diameter does not change with Paco₂. Eng et al. (20) reported that because MCA is a conductance, and not a resistance, vessel, changes in cerebral vascular resistance occur primarily through dilation of arterioles and not the arteries of the Circle of Willis. Consequently, the MCA is unlikely to be affected by cerebral vasoactive drugs.

During CPB, it remains controversial as to whether pressure-flow autoregulation is intact. Newman et al. (21) reported that increases in MAP had a small effect on CBF during normothermic CPB. In contrast, Sadahiro et al. (22) reported that in an animal model, CBF was constant when MAP was more than 50 mm Hg. If the cerebral metabolic rate for oxygen was constant after phenylephrine infusion, increases in SjvO₂ might indicate increases in CBF. If autoregulation is impaired in insulin-dependent patients, increases in CPP would result in increased SjvO₂. SjvO₂ is an index of CBF/metabolism, not CBF, and thus, increases in SjvO₂ do not directly indicate increases in CBF. Further study is required to confirm whether increasing MAP has a beneficial effect on reduced SjvO₂ during CPB in diabetic patients.

In conclusion, cerebrovascular CO₂ reactivity in diabetic patients correlated with the response of SjvO₂ to phenylephrine infusion. In addition, we confirmed the interrelationship between SjvO₂ responsiveness to phenylephrine infusion and cerebrovascular CO₂ reactivity.