Effect of Normalization of Hematocrit on Brain Circulation and Metabolism in Hemodialysis Patients

Correction of anemia in hemodialysis (HD) patients with recombinant human erythropoietin (rhEPO) results in reduction of the cardiovascular morbidity and mortality by decreasing cardiac output and the left ventricular mass (1). Improvement of exercise capacity, manifested by an increase in oxygen consumption and maximal work load and a decrease in the anaerobic threshold, have also been reported (2). In addition, rhEPO therapy leads to improvement in the quality of life, suggesting that the cognitive deficit resulting from anemia and uremic brain dysfunction is improved (3).

At present, many dialysis centers set the target hemoglobin levels at 10 to 12 g/dl, and numerous clinical trials have shown that achievement of this level of hemoglobin is associated with a significant relief of symptoms and a decrease in the transfusion rate (4).

There is little doubt that fear of access clotting (5), the hypertensinogenic effect of EPO that is probably related to loss of anemic vasodilation (6), and cost considerations have tempered the ambitions of most nephrologists for full correction of hemoglobin (Hb). However, failure of cardiac abnormality to completely regress with only partial correction of anemia has given reason for concern about these subnormal target Hb levels. In addition, rhEPO-related hypertension and access clotting do not exhibit a clear dose-dependent relationship, suggesting the possibility of a threshold, above which no additional excess of these side effects would be expected (7). Some nephrologists were therefore enthusiastic to normalize the Hb value to 14 g/dl. Studies examining the effect of normalization of hemoglobin for patients with end-stage renal failure are still in progress.

In this study, the effects of normalization of hematocrit (Hct) and the related change in blood rheology on the brain circulation and oxygen metabolism were investigated by the positron emission tomography (PET) technique.

Materials and Methods

Results

Hb and Hct rose significantly (P < 0.001) from 9.8 ± 1.3 to 14.2 ± 0.6 g/dl and from 29.3 ± 3.3 to 42.4 ± 2.2%, respectively. These increases were achieved over a period of 5 to 6 mo.

With normalization of the Hb level, cardiac output decreased significantly from 5.99 ± 1.21 to 4.32 ± 1.16 L/min (P <0.01) and stroke volume from 79 ± 11 to 64 ± 10 ml (P <0.01) (Table 2). MAP did not change significantly, although there was a significant increase in TPR from 2635 ± 907 to 3632 ± 1 058 dyn · s · cm^-5 · m^-2 (P < 0.05). None of the patients required adjustment in antihypertensive treatment.

Arterial blood pH, PaO₂, SaO₂, and PaCO₂ were not significantly altered by rhEPO treatment (Table 3). The arterial oxygen content increased significantly from an initial value of 5.7 ± 0.7 to a value of 8.0 ± 0.4 mmol/L after correction of anemia (P < 0.0001).

The rheologic measurements are presented in Table 4. Whole blood viscosity increased significantly after correction of anemia from 3.72 ± 0.38 to 4.57 ± 0.16 mPa · s (P < 0.05) (reference value 3.9 to 5.8 mPa · s). There was no significant change in plasma viscosity, but both the pre- and posttreatment values (1.51 ± 0.19 and 1.46 ± 0.13 mPa · s, respectively) were higher than the reference value (1.17 to 1.31 mPa · s). Erythrocyte fluidity decreased nonsignificantly from 85.8 ± 4.8 to 79.9 ± 7.4 Pa^-1 · s^-1, and both values were below the normal reference value (101 to 134 Pa^-1 · s^-1).

Table 5 presents the values for rCBF, rCBV, rOER, and rCMRO₂ in different regions of the brain, including the hemispheres, the cerebellum, the deep nuclei (thalamus and putamen), and the white matter, in HD patients and healthy control subjects. The following description of the results will be limited to the hemispherical cortices, since the behavior in the other brain areas was more or less similar to that in the cortices. The basal hemispheric rCBF was significantly greater in HD patients than in the control group, namely 65 ± 11 and 51 ± 9 ml/min per 100 ml, respectively (P < 0.05), whereas rCBV did not differ significantly between the two groups (3.5 ± 0.5 versus 3.6 ± 0.8 ml/100 ml brain tissue, respectively). However, there was a disparate response to the normalization of Hb value (Figure 1, A and B). At the same time as rCBF decreased significantly to 48 ± 12 ml/min per 100 ml (P < 0.05), rCBV increased significantly to 4.6 ± 0.6 ml/100 ml brain tissue (P < 0.05). The posttreatment value of rCBF was not significantly different from that of the healthy control subjects, whereas the posttreatment rCBV value was significantly higher (P < 0.05). The cerebral hemodynamic perfusion reserve, which is represented by the ratio of rCBF/rCBV, decreased significantly in HD patients from 18.2 ± 3.0 to 11.3 ± 2.3 (P < 0.01), whereas it was 14.5 ± 3.3 in healthy control subjects, a value not significantly different from the pre- and posttreatment values in HD patients.

A significant correlation was found between hematocrit and rCBF (r = -0.87, P < 0.001) in HD patients, but a less significant correlation was detected between the caO₂ and rCBF (r = -0.68, P < 0.05) (Figure 2, A and B). In addition, the rOER correlated significantly with the degree of erythrocyte fluidity (r = -0.6, P < 0.05) and with the ratio of rCBF/rCBV (r = 0.72, P < 0.05) (Figure 3, A and B).

The pretreatment value for cortical rOER in HD patients (44 ± 3%) did not differ significantly from the value in healthy control subjects (43 ± 4). When the anemia was fully corrected, rOER had increased to 51 ± 6%, a value significantly higher than the pretreatment value and the value in healthy control subjects (P < 0.05 for both). On the other hand, cerebral oxygen delivery (a product of rCBF × caO₂) increased significantly from 354 ± 67 to 391 ± 84 ml/min per 100 ml (P < 0.01).

The basal cortical rCMRO₂ was 3.86 ± 0.60 ml/min per 100 ml in HD patients and was slightly increased after rhEPO therapy to 4.01 ± 0.77 ml/min per 100 ml. Both values were significantly lower (P < 0.05) than the value of 5.03 ± 0.68 ml/min per 100 ml in healthy control subjects, but there were no significant differences between corresponding values in other brain regions.

Discussion

Cardiac output is normally regulated by the interactions of venous return (preload), resistance to outflow (afterload), and myocardial contractility. The hemodynamic response to chronic anemia in our group of HD patients was similar to that reported by others (12, 13). There was an increase in the cardiac output, mainly due to an increase in stroke volume, and a decrease in blood viscosity leading to reduction in total peripheral resistance. Apart from the increase in TPR, cardiac output and stroke volume attained values within the normal range after normalization of Hb. MAP did not significantly increase and there was no need to adjust the antihypertensive medication, although TPR was significantly raised. This was probably due to the decrease in the cardiac output.

SaO₂ was within the normal range in the anemic state and did not change significantly with normalization of the Hb level, and the same was true for other Hb compounds, e.g., methemoglobin and carboxyhemoglobin. Correction of anemia led to a significant increase in caO₂, which was within the normal range after treatment. It is known that to enhance gas exchange of the flowing blood, there should be an intracellular convection within the erythrocytes related to their deformability properties and to rotation of the red cell membrane around the cell interior (14). Motthaghy et al. (15) found that rigidification of the erythrocyte membrane with diamide, which prevents elongation of the erythrocytes and the concomitant intracellular convection under shear forces, is closely paralleled by a reduction of O₂ uptake. However, the decreased deformability of erythrocytes in our patient group did not affect the Hb oxygen saturation.

At rest, the brain, which constitutes only 2% of the total body weight, receives 15% of the cardiac output of blood and consumes 25% of the total oxygen consumption. There are essentially no stores of oxygen in the brain, and all oxygen extracted from the blood is metabolized (16). In the present study, the hemispherical CMRO₂ in the uremic patients was significantly lower than that in the healthy control subjects, both before and after EPO treatment, a finding that has also been reported in another study (17). The reason might be attributable to the actions of some uremic toxins, which could decrease the respiration of mitochondria and induce mitochondrial degeneration (18, 19).

It has been reported that rCBF increases with increased cerebral function and metabolism, a process that is called coupling mechanism, and also in case of tissue hypoxia, e.g., anemic hypoxia (20). During anemia, rCBF is significantly higher than that in healthy control subjects to compensate for the reduction in the capacity of the blood to carry oxygen to the brain. The increase of rCBF is achieved through an increase in cardiac output and redistribution of blood to the more vital organs (brain and heart) at the expense of other organs, e.g., the intestine, liver, kidney, and spleen (21).

After normalization of Hb, the rCBF decreased significantly in the HD patients, and this was also reported in another study after partial correction of hematocrit (22). In this study, there was a highly significant correlation between rCBF and hematocrit and a lesser degree of significance with caO₂. Two studies reported that caO₂ appears to be of fundamental importance in the regulation of CBF in humans (23, 24), since it is considered a critical determinant of the total amount of oxygen released into the tissues and the tissue pO₂. This contradiction to our study may indicate that the increase in hematocrit may not be associated with a correspondent increase of the tissue pO₂.

The rCBV is of interest in the evaluation of cerebrovascular abnormality because it reflects the vasodilation in response to a decrease in perfusion pressure (25). In healthy control subjects, rCBV is positively related to rCBF, indicating that the vascular space is larger where tissue perfusion is high (26). The rCBF depends on the cerebral perfusion pressure (CPP), which is the difference between the systemic arterial pressure and the venous pressure at the exit from the subarachnoid space (27). In healthy humans, within the autoregulatory range (MAP between 60 and 160 mmHg), as CPP falls there is, within seconds, vasodilation and an increase in rCBV to keep rCBF constant (28). When the vasodilation has almost reached its maximum and rCBV can no longer increase, and if CPP continues to fall due to a fall in MAP, rCBF starts to decline; in other words, the cerebral perfusion reserve becomes exhausted and the metabolic activity is maintained by increasing rOER (28). It was considered that rCBF/rCBV is a measure of cerebral hemodynamic perfusion reserve. When the ratio is below about 6, vasodilation and rCBV are maximal and the reserve is exhausted, even if rCBF is still normal (29).

In the present study, normalization of the Hb value was accompanied by an increase rCBV, reflecting in the occurrence of cerebral vasodilation. The rCBF/rCBV ratio, representing the cerebral hemodynamic perfusion reserve, decreased with normalization of Hb but without reaching the critical value, i.e., 6, which could not explain the increase in rOER. Schäbitz (30) found that an increased blood and plasma viscosity, a decreased deformability of erythrocytes, and increased tendency to aggregation are pathogenetically important for a disturbance of the microcirculation. Of special importance among these factors is the decreased erythrocyte deformability, when the erythrocytes have difficulty in traversing the capillaries resulting in a low perfusion rate in these capillaries, which will make the transfer of oxygen more difficult (31). Therefore, we assume that normalization of the hematocrit in our group of patients with hemorheologic abnormalities, i.e., decreased erythrocyte fluidity and high plasma viscosity, would have lowered the velocity of erythrocyte transfer inside the brain capillaries and may explain the high oxygen extraction by the brain tissue. In addition, rigidity of the vascular wall due to atherosclerosis, a common finding in uremic patients (32), would make the viscosity of the blood essentially the main determinant of blood flow in accord with Poiseuille's law.

Another important factor is that most of the patients in our study had hypercholesterolemia, which increases the risk of atherosclerosis, and hyperfibrinogenemia, which negatively affects the whole blood and plasma viscosity as well as the deformability and aggregability of red blood cells, causing a reduction in the microcirculatory flow (33). Thus, reduction of plasma fibrinogen and blood cholesterol may help to improve the microcirculatory blood flow and allow the patients to gain benefit from raising the hemoglobin value.

In summary, Eschbach et al. (7) demonstrated an improvement in the quality of life when Hct was raised to 42%. Bárány et al. (34) reported that the exercise capacity increased with normalization of Hct, but still remained subnormal compared with EPO-treated nonuremic subjects. From another study (35), it has been suggested that raising Hct to 42% from the commonly achieved target of 30% significantly improves the brain cognitive function. In our study, normalization of Hct greatly improved the general condition of the HD patients, but on the other hand the fraction of oxygen extraction by the brain tissue increased. The reason for this is not clear, but we assume that it may be attributed to a decrease in erythrocyte velocity inside the brain capillaries due to the coexisting problems of decreased erythrocyte fluidity and increased plasma viscosity in the these patients. In our view, an adequate hemoglobin concentration that fulfills the oxygen demand is an individual characteristic of every patient. Although Hino et al. (36) reported that a normal hematocrit provides a better brain oxygenation than any level lower than normal, their study was conducted on normal subjects with healthy erythrocytes, but the condition needs to be evaluated in patients with poorly deformable erythrocytes. However, our study comprised only a small number of patients, and more confirmative studies with a larger number of patients need to be undertaken.