Neither Hematocrit Normalization nor Exercise Training Restores Oxygen Consumption to Normal Levels in Hemodialysis Patients

Abstract

Patients treated with hemodialysis develop severely reduced functional capacity, which can be partially ameliorated by correcting anemia and through exercise training. In this study, we determined perturbations of an erythroid-stimulating agent and exercise training to examine if and where limitation to oxygen transport exists in patients on hemodialysis. Twenty-seven patients on hemodialysis completed a crossover study consisting of two exercise training phases at two hematocrit (Hct) values: 30% (anemic) and 42% (physiologic; normalized by treatment with erythroid-stimulating agent). To determine primary outcome measures of peak power and oxygen consumption (VO₂) and secondary measures related to components of oxygen transport and utilization, all patients underwent numerous tests at five time points: baseline, untrained at Hct of 30%, after training at Hct of 30%, untrained at Hct of 42%, and after training at Hct of 42%. Hct normalization, exercise training, or the combination thereof significantly improved peak power and VO₂ relative to values in the untrained anemic phase. Hct normalization increased peak arterial oxygen and arteriovenous oxygen difference, whereas exercise training improved cardiac output, citrate synthase activity, and peak tissue diffusing capacity. However, although the increase in arterial oxygen observed in the combination phase reached a value similar to that in healthy sedentary controls, the increase in peak arteriovenous oxygen difference did not. Muscle biopsy specimens showed markedly thickened endothelium and electron–dense interstitial deposits. In conclusion, exercise and Hct normalization had positive effects but failed to normalize exercise capacity in patients on hemodialysis. This effect may be caused by abnormalities identified within skeletal muscle.

Functional capacity and fitness are markedly reduced in patients with ESRD,^1–3 which increases their risk of mortality⁴ and limits their ability to perform activities of daily life.⁵ The reduction in physical fitness observed is multifactorial; however, renal anemia contributes significantly.⁶ Oxygen carrying capacity is reduced by anemia, which in turn, limits an individual’s peak oxygen uptake, ability to perform exercise, and time to exhaustion, especially in ESRD.⁷

Oxygen uptake can be thought of as a cascade of oxygen from the lung to the blood pumped by the heart to the working tissues, where it is used to perform physical work.⁸ Patients with ESRD on hemodialysis can have pathologies at most of these sites. The majority of patients on hemodialysis develop a progressive anemia, because their kidneys become unable to produce sufficient erythropoietin to maintain normal hemoglobin mass.⁹ This anemia combined with impaired cardiovascular function limit the delivery of oxygen to the tissues. As kidney function declines, patients become even less active,¹⁰ causing further deterioration of cardiovascular function, and a downward spiral ensues, ultimately leading to severe disability.

Fortunately, the anemia of renal disease can be avoided because of the advent of recombinant erythropoietin. Anemia is typically treated with erythropoiesis-stimulating agents (ESAs), which partially normalize renal anemia, improve quality of life and cognitive function,^11,12 reduce left ventricular hypertrophy,¹³ and modestly increase peak oxygen uptake relative to the increase in hematocrit.¹⁴ Likewise, intradialytic exercise training also increases peak oxygen uptake^15,16 and has additional cardioprotective benefits.¹⁷ However, similar to treatment with ESAs, exercise training in patients with kidney disease does not restore exercise capacity to normal levels.^18,19 Both interventions have been shown to produce improvements in physical fitness and peak oxygen uptake; however, they have not returned patients to the level of fitness of most sedentary people with normal renal function. It is possible in the case of raising hematocrit that, although oxygen carrying capacity of the blood is markedly improved, the other components of the oxygen cascade are unchanged and limit the potential benefit of normalizing hematocrit. Conversely, regular exercise training alone may not achieve its full benefit if oxygen transport is limited by anemia. A third scenario exists where some aspect of ESRD or dialysis²⁰ prevents normal oxygen transport that may be unaffected by either correction of anemia or regular exercise.

Evidently, the precise mechanisms underlying the modest improvements in oxygen uptake observed after exercise training or treatment of anemia in patients on hemodialysis have not been carefully elucidated. Thus, we used a crossover-designed study to test the hypothesis that patients with ESRD on hemodialysis would have the greatest improvement in peak oxygen uptake and peak power when hematocrit was increased to within a normal physiologic range (hematocrit of 42%) and combined with intradialytic exercise training (TR42) compared with exercise training alone (TR30), normalized hematocrit alone (UN42), or anemic untrained hematocrit (UN30). Secondary aims were to elucidate the possible mechanism of exercise intolerance in ESRD.

Results

Subject Characteristics

Fifty patients with ESRD requiring hemodialysis were recruited to participate in this study. Three patients were deemed ineligible after randomization. Of the 47 remaining patients, 27 patients completed the entire study (study design is given in Figure 1). In total, 20 patients withdrew from the study, predominantly during the first and final 12-week periods (n=10 and n=7, respectively). Reasons for withdrawal included personal reasons (n=8), renal transplant (n=5), unable to achieve target hematocrit of 42% (n=2) (despite epoetin α doses >2000 U/kg per week), failure to return to hematocrit of 30% (n=1), poor hemodialysis compliance (n=3), and cancer (n=1). The characteristics of patients who were randomized and those who completed the study are presented in Table 1. The majority of patients were black (51%) or Hispanic (32%). Hypertension was the predominant cause of renal failure followed by GN and diabetes.

Figure 1.

Schematic diagram of the study protocol. Subjects were required to have a stable hematocrit (Hct) for a minimum of 4 weeks during the maintenance (increasing or decreasing) phase before any testing. Similarly, patients were required to complete ≥33 good exercise sessions (92%) before completing post-testing for each training phase. EXT, exercise training.

Table 1.

Baseline characteristics of the enrolled cohort and completed participants

Characteristics	Randomized, n=47	Completed, n=27	P Value
Age, yr [range]	44±13 [20–71]	45±11	0.36
Sex, n (% men)	36 (77)	22 (81)	0.62
Height, cm	—	172±9
Weight, kg	—	79±17
Body fat, %	—	20±7
Ethnicity, n (%)
White	6 (13)	3 (11)	0.83
Black	24 (51)	15 (56)	0.71
Hispanics	15 (32)	8 (30)	0.84
Asian Indian	2 (4)	1 (4)	0.91
Cause of kidney disease, n (%)
Hypertension	23 (49)	16 (58)	0.39
GN	10 (21)	5 (19)	0.77
Diabetes	8 (17)	5 (19)	0.87
Other	6 (13)	1 (4)	0.20
Duration of hemodialysis, yr [range]	3±3 [4 mo to 17 yr]	3±3	0.99
Hematocrit, %	30.7±1.8	30.8±1.9	0.73
ESA dose, U/kg per wk	142±112	97±89	0.07

Values are presented as mean±SD or n (%) for categorical variables.

The dosage of ESA required to normalize hematocrit levels was approximately four times that of the baseline dose of ESA (epoetin α: 415±196 versus 97±89 U/kg per week). The higher dose of ESA significantly increased hematocrit, arterial O₂ content, hemoglobin, total blood volume and red cell mass and decreased plasma volume (Table 2). Normalization of hematocrit also increased resting diastolic BP and mean arterial pressure (MAP); however, systolic BP was unchanged (Table 2).

Table 2.

Response of resting measures of metabolic and cardiovascular function to exercise training and hematocrit normalization

Parameter	UN30	TR30	UN42	TR42	Main Effect Hct	Main Effect Training	Interaction
Hct, %	31.4±1.7	30.9±3.9	41.5±2.1	42.1±2.4	<0.0001	0.92	0.28
Hemoglobin, g/dl	10.7±0.8	10.5±1.3	14.4±0.9	14.6±1.0	<0.001	0.99	0.31
RBC volume, ml/kg	17.7±2.9	17.6±3.3	24.4±3.9	25.3±4.9	<0.0001	0.59	0.50
Plasma volume, ml/kg	46.3±8.5	46.8±8.1	42.3±6.2	43.3±8.7	0.016	0.624	0.87
Blood volume, ml/kg	64.0±11.2	64.3±10.4	66.6±9.7	68.6±13.3	0.113	0.596	0.69
VO2, ml/kg per min	3.0±0.6	2.8±0.5	3.1±0.6	3.0±0.6	0.179	0.179	0.65
Cardiac output, L/min	7.4±1.8	7.3±2.0	7.5±2.1	7.3±1.9	0.89	0.69	0.89
Heart rate, bpm	73±9	77±12	72±12	73±12	0.25	0.25	0.49
a-vO2 difference, vol%	3.2±0.6	3.2±0.7	3.4±0.9	3.4±1.0	0.20	0.99	0.99
aO2, vol%	14.3±1.2	14.2±1.9	18.6±1.6	18.8±1.9	<0.0001	0.88	0.64
Resting systolic BP, mmHg	145±21	142±22	146±23	140±21	0.90	0.28	0.72
Resting diastolic BP, mmHg	84±13	83±12	90±15	87±11	0.045	0.42	0.686
Resting MAP, mmHg	104±12	103±13	109±15	104±11	0.23	0.23	0.42

Data are presented as means±SDs for a given variable at a given time point. Hct, hematocrit; RBC, red blood cell; a-vO₂ difference, arteriovenous oxygen difference; aO₂, arterial oxygen content.

Exercise Training Compliance

The number of completed training sessions and mean training impulse (TRIMP) scores, which provides an overall measure of training load, were similar between the anemic (TR30) and normalized hematocrit (TR42) conditions (Table 3). However, total work performed was slightly higher under the TR42 condition, meaning a greater amount of energy was expended during these exercise sessions.

Table 3.

Comparison of training compliance and load at the anemic and normalized hematocrit

Parameter	TR30	TR42	P Value
Completed sessions	38±4	36±2	—
Mean TRIMP per session	54±35	50±29	0.65
Work performed per session, kJ	136±55	152±61	0.32
Kilojoules per TRIMP score per session	3.0±1.5	3.7±1.7	0.12

Data are presented as means±SDs. kJ, kilojoules.

Effects of Hematocrit Normalization and Intradialytic Training on Cardiovascular and Metabolic Function

The effects of hematocrit normalization, exercise training, and the combination are presented in Table 4. Peak power and oxygen consumption (VO₂) were both significantly improved by normalization of hematocrit and exercise training. However, the effect of the combination of the two perturbations, although incrementally greater than either intervention alone, was modest. The normalization of hematocrit increased peak arterial oxygen content and caused a widening of the arteriovenous oxygen difference, although exercise training alone tended to increase peak arteriovenous oxygen difference and arterial lactate concentration. The effects of exercise training and higher hematocrit were additive for arteriovenous oxygen difference. However, the combination did not normalize these values, which remained at only two thirds of expected values in healthy sedentary controls (Table 4). Finally, tissue diffusing capacity, a measure of oxygen transport by diffusion from the microcirculation to the mitochondria, was improved by exercise training at peak exercise by 11%.

Table 4.

Peak metabolic and cardiovascular response to exercise training and hematocrit normalization

Parameter	UN30	TR30	UN42	TR42	Healthy Controls	Main Effect Hct	Main Effect Training	Interaction
Peak power, W	101±35	124±45	110±37	134±48	—	0.238	0.004	0.95
aO2, vol%	15.7±1.1	15.5±2.1	20.3±1.3	20.5±1.7	18.6±1.6	<0.0001	0.99	0.50
VO2, ml/kg per min	17.9±3.9	19.3±4.4	19.6±4.2	21.5±5.1	30.0±5.0	0.024	0.055	0.77
a-vO2 difference, vol%	7.9±1.3	8.3±1.4	8.5±1.8	9.0±1.8	15.6±2.1	0.036	0.14	0.87
Cardiac output, L/min	19.0±5.7	19.3±5.1	18.7±6.0	20.0±6.1	14.9±3.0	0.83	0.40	0.60
RER	1.22±0.12	1.20±0.10	1.15±0.10	1.16±0.09	1.26±0.08	0.007	0.80	0.45
Heart rate, bpm	144±24	145±22	143±22	144±24	177±8	0.82	0.82	0.99
Stroke volume, ml	131±40	135±38	129±35	135±35	84±17	0.89	0.48	0.89
MAP, mmHg	127±17	123±11	129±16	130±15	123±12	0.12	0.60	0.39
SVR, dynes⋅s−1 per cm5	7.8±3.2	7.0±2.6	7.8±2.5	7.1±1.8	—	0.92	0.13	0.92
Tissue diffusing capacity, ml/m per torr	28.3±8.8	31.5±8.0	27.6±7.3	30.8±9.1	—	0.66	0.048	0.99
Arterial lactate, mmol/L	5.5±1.8	6.3±2.0	4.8±1.3	5.6±1.5	10.4±1.7	0.032	0.014	0.99
Arterial pH	7.33±0.06	7.32±0.06	7.29±0.03	7.29±0.05	—	0.001	0.615	0.62

Data are presented as means±SDs for a given variable at a given time point. For comparison, we have included a subset of historical healthy sedentary age–matched (n=20; men =75%; age =48±6 years old; Hct=41.3±3.2) control data from our laboratory (work by Carrick-Ransom and colleagues⁴²) to highlight the disturbances to the capacity of the oxygen transport system in the patients on hemodialysis. Despite exercise training and hematocrit normalization, peak VO₂ and a-vO₂ difference remain severely impaired compared with those in normal sedentary individuals. Hct, hematocrit; —, no data; aO₂, arterial oxygen content; a-vO₂ difference, arteriovenous oxygen difference; RER, respiratory exchange ratio; SVR, systemic vascular resistance.

Figure 2 compares healthy normal subjects (from the work by Hartley et al.²¹) with the patients on hemodialysis after hematocrit normalization before and after exercise training for 12 weeks. Figure 2 highlights the blunted effects of exercise training and hematocrit normalization in patients on hemodialysis as well as the severely impaired capacity of patients at baseline. The normal training response is characterized by an increase in peak VO₂ because of increased cardiac output and tissue diffusing capacity. In patients on hemodialysis, the increase in peak VO₂ was one half of the observed change in the normal subjects. The exercise training had little effect on the arteriovenous oxygen difference, which remained low compared with that in the normal subjects, along with peak tissue diffusing capacity.

Figure 2.

Peak exercise response to endurance training. Comparison of maximal oxygen uptake and components in the subjects in the work by Hartley et al.²¹ with these data in patients on hemodialysis with a normal hematocrit. The first column in each category is normal subjects’ pretraining data, which are assigned 100%. The second column is the change from initial values in normal subjects post-training. The third column is the initial data from the patients on hemodialysis as percentages of those in the work by Hartley et al.²¹ The fourth column is the data from patients on hemodialysis post-training. aO₂, arterial oxygen content; a-vO₂, arteriovenous oxygen difference; Qc, cardiac output; TDC, tissue diffusing capacity.

Quadriceps Histology and Biochemistry

The effects of the perturbations on muscle histology and biochemistry are summarized in Table 5. Muscle histology was not significantly altered by any intervention (Table 5), although exercise training significantly increases citrate synthase activity by 43% and acyl-CoA dehydrogenase (HAD) by 16%. This indicates that exercise training increased the concentration of oxidative enzymes. However, phosphofructokinase and myoglobin were not altered by any intervention. Change in citrate synthase activity did not correlate with change in peak VO₂, tissue diffusing capacity, or the work performed during exercise training.

Table 5.

Effects of intradialytic training and hematocrit normalization on skeletal muscle strength, structure, and function

Parameter	UN30	TR30	UN42	TR42	Main Effect Hct	Main Effect Training	Interaction
Type 1 fiber, %	33±10	32±9	31±9	29±9	0.16	0.40	0.78
Area type 1, μm2	5366±1955	5226±1757	5171±1575	5140±1665	0.68	0.80	0.87
Area type 2, μm2	3528±1498	3364±1106	3393±1168	3260±1062	0.61	0.53	0.95
Capillary density, μm2	271±85	280±103	268±145	270±142	0.78	0.81	0.88
Capillary-to-fiber ratio	1.1±0.5	1.2±0.6	1.1±0.7	1.0±0.7	0.99	0.99	0.99
Citrate synthase, μm/min per dry gram	12.6±4.8	18.0±7.2	12.2±4.7	17.9±7.5	0.83	<0.0001	0.90
HAD, μm/min per dry gram	25.8±8.2	30.0±14.7	22.8±7.5	30.6±12.9	0.58	0.0066	0.41
Phosphofructokinase, μm/min per dry gram	87.2±50.2	74.1±34.0	86.4±50.3	97.7±64.5	0.25	0.93	0.22
Buffer capacity, μm H+ per pH per dry gram	356±89	387±85	359±71	361±79	0.46	0.29	0.36
Myoglobin, ng/wet g	2.3±0.7	2.0±0.7	2.4±0.9	2.3±0.7	0.17	0.17	0.49

Data are presented as means±SDs for a given variable at a given time point. Hct, hematocrit; H⁺, hydrogen ions.

Adverse Events

One serious adverse event occurred during the study; a patient suffered a seizure during dialysis at the normal hematocrit level. In addition, 10 access thromboses required treatment: four occurred while patients were at the higher hematocrit, three occurred while patients were at the lower hematocrit, and three occurred while either increasing or decreasing hematocrit.

Comparison of Healthy Aged–Matched Controls with Patients on Hemodialysis: Quadriceps Electron Micrographs

We compared a subset of patients on hemodialysis (n=18) at baseline with a group of controls with normal renal function (n=7). A typical sample from a healthy control is shown in Figure 3, left panel, and a typical hemodialysis patients is shown in Figure 3B, right panel. The sample from the hemodialysis patients is characterized by thickened capillary endothelium and capillary basement membranes, with an enlarged interstitial space that was filled with electron-dense material. Mitochondrial density was also reduced in the adjacent muscle. Both the capillary endothelium and basement membranes of the patient on hemodialysis were significantly thicker in comparison with those in the controls (Figure 3, Table 6).

Figure 3.

Comparison of representative hemodialysis patient and healthy control electron micrograph of capillary and skeletal muscle. The photo in the left panel is from an individual without renal failure, and the photo in the right panel is from a patient on hemodialysis. The large black objects on the inner edges of the two photographs are red blood cells in capillaries. The white arrows depict the inner edges of the capillary endothelium, and the black arrows depict the outer edges of the capillary basement membrane. Magnification, ×5000.

Table 6.

Comparison of capillary structure using electron micrograph

Parameter	Patients on Hemodialysis, n=18	Healthy Control, n=7	P Value
Capillary endothelium, nm	780±450	460±120	0.01
Capillary basement membrane, nm	500±270	220±60	0.002
Total thickness, nm	1130±500	680±190	0.003

Data are presented as means±SD.

Discussion

We sought to identify the limitations in the oxygen cascade system in hemodialysis patients by examining the response to exercise training and normalization of hematocrit on components of the Fick Equation [VO₂ = cardiac output × (Ca − Cv)]. We found that both restoring hematocrit levels to within a normal range and intradialytic exercise training improve peak oxygen uptake and power, albeit by different mechanisms. The normalization of hematocrit markedly increased arterial oxygen concentration, widening the arteriovenous oxygen difference. In contrast, exercise training increased peak VO₂ through improved peak tissue diffusing capacity or greater ability to diffuse oxygen from the capillary to the mitochondria. Although the combination did provide for greater peak power and VO₂, the increases were far less than expected compared with changes observed in individuals not dependent on hemodialysis. A major finding was the discovery of markedly thickened capillary endothelium and electron-dense deposits in the interstitial space, which could explain the poor functional capacity found in hemodialysis patients and likely inhibits oxygen transport and removal of metabolic waste from working muscle.

Nonetheless, these improvements in functional capacity and maximal oxygen uptake are clinically relevant. The improvement in peak VO₂ after training and hematocrit normalization was equivalent to an approximately 1-MET increase in fitness from the baseline hematocrit untrained state (Δ+3.6 ml/kg per minute). Several studies have shown reductions in total mortality and cardiovascular events with increasing fitness.^22–24 Specifically, Blair and colleagues^22,23 and Gulati et al.²⁵ have shown that a 1-MET increase in exercise capacity is equivalent to a 12% improvement in survival in men and a 17% improvement in survival in women.

Effects of Exercise Training

In this study, intradialytic exercise training alone significantly increased peak power and VO₂, citrate synthase activity and HAD, and tissue diffusing capacity. The magnitude of the increase in peak VO₂ observed in our study is consistent with the results of the recent meta-analysis by Heiwe and Jacobson.²⁶ Thus, it seems that exercise training in hemodialysis patients shows a modest but significant improvement in peak VO₂ that is approximately two thirds of what is expected in healthy individuals who have received a similar dose of training and are not on hemodialysis.²¹

Effects of Normalization of Hematocrit

In this study, normalization of hematocrit resulted in an increase in peak VO₂ and arteriovenous oxygen difference as well as increases in arterial oxygen content and red blood cell volume. In healthy subjects (nondialysis), increasing hematocrit by 20% (from 42.7% to 50.8%) increases peak VO₂ by 7%.²⁷ In contrast, increasing hematocrit by 36% (from 30.8±1.9% to 41.8±2.1%) resulted in only a 9% increase in peak VO₂ in hemodialysis patients. These findings highlight the diminished response of peak VO₂ to increasing arterial oxygen content in hemodialysis patients. Furthermore, despite the large increase in arterial oxygen content, the skeletal muscle was only able to extract a small amount of the additional oxygen available.

Effect of Exercise Training and Hematocrit Normalization

The effects of the combination of exercise training and hematocrit normalization significantly increased peak power, VO₂, and arteriovenous oxygen difference. However, the effects were modest, and the combination did not normalize exercise capacity, despite significant improvement in the ability of the muscle to extract oxygen and increased availability of oxygen. In contrast to our findings, Painter et al.¹⁵ found no improvement in peak VO₂ with exercise training and hematocrit normalization. The difference in results may be partially explained by study design; in this study, we used a crossover design, whereas Painter et al.¹⁵ randomized groups of different individuals. However, the findings from both studies do highlight the difficulty associated with improving peak VO₂ in hemodialysis patients. Neither increasing arterial oxygen content nor regular exercise was able to completely overcome the apparent barrier to oxygen transport from red cell to mitochondria.

Limitations in Oxygen Transfer in Patients on Hemodialysis

Although peak arteriovenous oxygen difference did increase with both normalization of hematocrit and the combination of hematocrit and exercise training, the absolute values were still well below those of age– and sex–matched healthy controls (Figure 2, Table 4). Tissue diffusing capacity, which is the maximal ability to transfer oxygen from red blood cells in the capillary to the mitochondria,²⁸ was also low in comparison with that observed in healthy individuals²¹ but did increase slightly with exercise training. This finding suggests that exercise training enhanced the transfer of oxygen from the microvasculature red cells to the mitochondria. Despite this increased ability to transfer oxygen, we did not observe any statistically significant changes in capillary density or capillary-to-fiber ratios. Our findings support previous studies^19,29,30 that have shown that capillary density and capillary-to-fiber ratios are reduced in hemodialysis patients.

Thus, the results from this study and others^15,31 strongly suggest that there is something unique to hemodialysis patients that limits oxygen transfer at the site of the capillary-muscle interface. Sala et al.³² conducted an elegant set of experiments, which also identified an impairment in maximal oxygen supply in young patients with ESRD. Similar to our findings, Sala et al.³² identified abnormally low oxidative diffusion conductance in the muscle but seemingly normal oxidative capacity within the muscle. A limitation of the study by Sala et al.³² was that they did not normalize hemoglobin levels; thus, their comparison group had significantly higher hemoglobin. In this study, we increased arterial oxygen content, which resulted in a modest improvement in peak VO₂ but failed to restore tissue diffusing capacity to expected levels.

We identified a possibly explanation for the modest increase in VO₂ typically observed after training in hemodialysis patients. In a subset of patients, we examined electron micrographs from the biopsies of skeletal muscle. The patients’ electron micrographs were characterized by markedly thickened capillary endothelium and capillary basement membranes, with an enlarged interstitial space that was filled with electron-dense material (Figure 3, Table 6). The etiology of the thickened capillary endothelium and basement membrane is not clear. However, we hypothesize two possible explanations for the accumulation of the material in the skeletal muscle: (1) chronic renal insufficiency or hemodialysis itself creating an inflammatory reaction in capillary endothelium or (2) the incomplete removal of waste products by hemodialysis, which may have resulted in the accumulation of these unidentified waste products in the skeletal muscle, limiting oxygen diffusion in the muscle. Additional studies are required to confirm these preliminary findings.

Special Considerations

This study investigated the limitations of oxygen transport in patients on hemodialysis. Patients on hemodialysis dialyze via an arteriovenous shunt created surgically by either anastomosing an artery (typically in the upper extremity) to a large superficial vein, which enlarges (“arteriolyzes”) the veins, or inserting a subcutaneous graph between artery and vein. This arteriovenous fistula was a benefit for our study by allowing technically easy access to the arterial circulation for measurements of arterial blood gases and arterial pressures during rest and peak exercise with no additional subject burden, because patients would go on to dialysis after the exercise testing. Whereas, in normal (non-hemodialysis) populations, multiple arterial cannulations carry an unwarranted risk to the subject and are rarely performed. A potential confounding explanation for the lower than expected arteriovenous oxygen difference measurements at rest is that a percentage (perhaps as much as 25%³³) of resting cardiac output flows through the shunt, where there is presumably no offloading of oxygen. However, the magnitude of the shunt is likely constant at rest and peak exercise under the four conditions of our experiment (UN30, TR30, UN42, and TR42); therefore, the presence of the shunt fails to account for the less than expected modest increase in arteriovenous oxygen difference observed in this study to the perturbations of normalizing hematocrit and exercise training.

The cardiac outputs at rest and during exercise were elevated in these hemodialysis patients compared with sedentary controls. We have previously shown that, for a given VO₂, cardiac output is elevated in patients dialysis.¹⁹ The increase in cardiac output is partially caused by the presence of the arteriovenous fistula, which can increase cardiac output by as much as 25%.^33,34 We speculate that the combination of hemodialysis and the arteriovenous shunt over time may act to train cardiac output. In sedentary individuals, there is little training of cardiac output; despite this, the cardiac output-VO₂ relationship is maintained across a wide range of fitness levels. In addition, there were also differences in the exercise testing protocols: the hemodialysis patients were tested on a bike ergometer (Monark) with a ramp protocol, whereas the sedentary control group were tested upright on a treadmill using a modified Astrand–Saltin incremental protocol. These factors may account for the higher resting and peak cardiac outputs observed in these subjects.

In conclusion, We found that both correcting the anemia of renal disease and intradialytic exercise training produced significant, clinically meaningful improvements in peak oxygen uptake and power. The improvements observed would likely translate into improvements in the ability to perform and sustain a variety of activities of daily living. However, there remains a significant limitation in the ability of the hemodialysis patients to extract oxygen from the red blood cell into the working tissue. We discovered electron microscopic evidence that suggests that the impaired ability to normalize oxygen transport and extraction may be at the site of the capillary wall and interstitial space. We hypothesize that this observation may explain the impaired ability to normalize oxygen transport in hemodialysis patients, even after markedly increasing arterial oxygen carrying capacity and despite a rigorous exercise training program, which resulted in a 43% increase in citrate synthase activity. Additional work is required to identify the electron-dense material in the interstitial space and the etiology of the thickened capillary endothelium to determine (1) whether the changes can be reversed and (2) if removal, reduction, or prevention of accumulation of the material restores the response to training and/or the functional capacity of patients with ESRD to normal.

Concise Methods

Study Participants

We recruited 50 patients undergoing maintenance hemodialysis from outpatient dialysis units in Dallas, Texas. Eligible patients were ≥18 years of age, were on hemodialysis for >3 months, and had hematocrit of 30%±3%. Patients were excluded from the study if they had orthopedic limitations to pedaling a cycle ergometer, unstable heart disease, poor dialysis compliance, or likelihood of renal transplant within 6 months. All participants provided written informed consent to a protocol approved by the Institutional Review Board of the University of Texas Southwestern Medical Center. This study followed the guidelines set forth in the Declaration of Helsinki. After enrollment, three patients were deemed ineligible for randomization (n=1 did not achieve target hematocrit and n=2 did not have ability to pedal a cycle ergometer) (Supplemental Material).

Study Design

This study had a crossover design (Figure 1). After enrollment and baseline assessment, patients were randomized to either group A or B. Group A participants’ hematocrit levels were increased to 42% (UN42), whereas group B levels were maintained at 30% (UN30). After a minimum of 12 weeks, the testing battery was repeated (with a stable hematocrit maintained for a minimum of 4 weeks). Intradialytic exercise training (36 exercise sessions) was then performed by group A with hematocrit levels at 42% (TR42) and group B with hematocrit levels at 30% (TR30). Training was then ceased; hematocrit decreased to 30% in group A and increased to 42% in group B. The testing battery was repeated again before both groups completed 12 weeks of exercise training. The final testing battery was administered on completion of the second 12 weeks of intradialytic exercise training.

Interventions

The data collection for this study was completed before the black box warning issued by the Food and Drug Administration regarding hematocrit and hemoglobin targets for patients with ESRD.

Hematocrit Management

To increase hematocrit, the patient’s maintenance dose of recombinant ESA was increased by 150%; after 2 weeks, the dose was doubled every 2 weeks until the target hematocrit (42%±3%) was obtained. The ESA dose was then titrated to maintain hematocrit within the target range. A placebo was administered to maintain or lower hematocrit to 30%±3%. Hematocrit levels were managed by one of the primary investigators (J.R.T.). This investigator (J.R.T.) did not share this knowledge of the subject’s group assignment with any patients, staff, or investigators in the study. The staff in the dialysis unit were provided with identically labeled syringes with the patients’ dosages of ESA adjusted with diluent so that the volumes were identical, and they were administered by a clinical nurse. This investigator (J.R.T.) had no involvement with study subject training or exercise testing. While participating in the study, ESA for the patient was provided complimentary by the sponsor. Staff at the dialysis units were unaware of the specific ESA dose (or placebo) delivered to the patients in the study and the patient’s hematocrit.

Exercise Training

Supervised exercise training was performed during the second hour of the dialysis session three times per week for 12 weeks. Target workloads were determined from the initial exercise test, and each patient cycled at 75% of his/her peak power output. Exercise duration and intensity were gradually increased, such that, by workout 24, participants were cycling for approximately 60 minutes at the highest tolerable workload that could be sustained. From week 9, high–intensity interval training was introduced with every third workout consisting of an interval session of either 1 minute at 200% of the pre-training peak power output or 2- to 3-minute intervals at 150%–200% of pre-training peak power. Each session was supervised by a trainer who was blinded to the subject’s hematocrit level, and heart rate (Polar) was monitored throughout the session along with cadence, which was maintained at 50 revolutions per minute.

To quantify exercise training, the absolute amount of work performed (kilojoules) in each session (kilojoules per session) was calculated along with TRIMPs of the workout (TRIMPs per session).³⁵ The TRIMP method multiplies the duration of a training session by the average heart rate achieved during that session weighted for exercise intensity as a function of heart rate reserve. Using this method, exercise sessions of longer duration and/or greater intensity, such as interval workouts, are assigned relatively higher TRIMP values (higher heart rate and higher weighting factor) than sessions of shorter duration and/or lower intensity. TRIMP is a commonly used technique to assess and monitor exercise training load.³⁶ The goal for each patient was to complete at least 33 good sessions (92% of all sessions) before post-training testing was performed. Post-testing was scheduled at the end of each phase when an individual subject had completed the goal of that phase (Figure 1) (either attained the target hematocrit or completion of ≥33 good training sessions). The actual time for completion of study phases was 17±10 weeks for phase 1, 15±2 weeks for phase 2, 18±10 weeks for phase 3, and 15±4 weeks for phase 4. Thus, the patients spent an average of 16 months enrolled in the study.

Measurements and Outcomes: On–Dialysis Day Assessments

Resting Metabolic and Hemodynamic Assessment

Approximately 44 hours after the last dialysis session, patients attended our laboratory. On arrival, body weight was measured before subjects were instrumented, and they laid quietly for 60 minutes in a resting metabolic suite. After 30 minutes of supine rest, resting O₂ utilization was determined via the Douglas bags method. Resting heart rate, BP, and cardiac output (acetylene rebreathing) were also measured. An arterial blood gas sample was drawn and analyzed for oxyhemoglobin, pO₂, pCO₂, pH, and HCO₃⁻ on an ILS-1306 Blood Gas Analyzer and an ILS-482 Co-Oximeter. Arterial oxygen content was calculated from these data. Arterial lactate concentration was measured on a Yellow Springs 23L Lactate Analyzer (Yellow Springs, Yellow Springs, OH).

Blood Volume

Blood volume was measured using the Evan Blue dye dilution technique.³⁷ Blood volume was calculated using ratios of 0.91 and 0.96 for venous to body hematocrit and plasma trapping, respectively. The volume of red cell mass was determined by subtracting plasma volume from blood volume. Values were indexed to body weight and presented as milliliters per kilogram of body mass.

Exercise Testing

Maximal exercise testing was performed on a cycle ergometer using a ramp protocol (Monark 818E). Measures of ventilatory gas exchange were made using the Douglas bag technique. Heart rate was monitored continuously by a three-lead ECG (telemetry unit; Transkinetics) throughout the test. At the end of each stage, BP (STBP-780; Colin Medical Instruments, South Plainfield, NJ), arterial blood gases (ILS-1306 Blood Gas Analyzer and ILS-482 Co-Oximeter), and cardiac output (modification of foreign gas rebreathing method³⁸) were measured. Peak values for oxygen uptake and power were the highest values obtained during the test. Data collected from the exercise test included power output, ventilation, oxygen uptake, carbon dioxide production, cardiac output, heart rate, systolic and diastolic BPs, MAP, arterial oxygen content, arterial pH, and arterial lactate concentration. Data calculated from the collected data included respiratory quotient, stroke volume, MAP, systemic vascular resistance, arteriovenous oxygen difference, mixed venous oxygen content, and tissue diffusing capacity (peak data only). Briefly, the calculation of tissue diffusing capacity allows the assessment of oxygen transport: VO₂ maximum = diffusing capacity × (mean muscle capillary pO₂ − mitochondrial pO₂).^28,40 To calculate tissue diffusing capacity, the following variables were imputed: muscle blood flow (assumed to equal cardiac output), arterial blood gases (pO₂, pCO₂, pH, and O₂ saturation), estimated tissue venous blood gases (on the basis of the Fick principle), hemoglobin, hemoglobin p50, and blood temperature. Muscle capillary pO₂ is then calculated using a custom–designed computer program.⁴⁰

Off–Dialysis Day Assessments

Quadriceps Biopsy

The day after a dialysis session, the patient would report to the laboratory for a muscle biopsy. Muscle biopsies were performed on the right vastus lateralis. All specimens from a given subject were analyzed on the same run of a given assay.

Image Analyses

The percentages of each fiber type and fiber areas were assessed in duplicate from three to four full fields at ×20 by two experienced technicians blinded to group assignment. Capillary density and the capillary-to-fiber ratio were read at ×40. Inter- and intraobserver variabilities were <10%. Fiber type is expressed as a percentage of total fibers counted. Fiber area is expressed in square micrometers. Capillary density is expressed as capillaries per square micrometer and a ratio of capillaries to fiber.

Electron Microscopy

Capillary endothelium and capillary basement membrane thickness were measured in six radial directions, taking care to avoid areas including endothelial nuclear material. Measurements were averaged and expressed in nanometers. Seven subjects (five males and two females) with normal renal function and similar age to our patients had biopsies analyzed in the same manner for comparison.

Enzyme Activities

Activities for citrate synthase, HAD, and phosphofructokinase were measured with a flurometer on muscle homogenates at pH 7.0 and room temperature. Values are expressed as micromoles of NADH per minute per gram dry weight of tissue.

Tissue Buffering Capacity

Tissue buffering capacity was measured on muscle homogenates by the method described by Nakao et al.⁴¹ Values are expressed as micromoles of H⁺ per pH unit per gram dry weight of tissue.

Muscle Myoglobin Concentration

Myoglobin concentrations were measured on muscle homogenates by RIA using a standardized kit. Crossreactivity/cross-sensitivity values are expressed as nanograms of myoglobin per gram wet weight of tissue.

Reference Control Group

For comparison, healthy sedentary age–matched control data are presented from previous published studies in our laboratory,⁴² and the effect of the normal training response is from the literature.²¹ Exercise testing was performed to determine the response to maximal exercise. It should be noted that the healthy sedentary cohort was predominantly tested using a treadmill in contrast to the cycle ergometer used in this study.

Statistical Analyses

There was no effect of order/carryover effect; therefore, we combined groups A and B for analysis under each condition of interest (Figure 1). An unpaired t test and a chi-squared test were used to compare baseline characteristics in randomized patients and those who completed the study. A two–way repeated measures ANOVA was used to determine the effect of normalizing hematocrit and exercise training. An unpaired t test was used for the comparison of capillary wall thickness between renal subjects and healthy subjects. Data presented as means and SDs. Commercially available statistical software was used for analysis (SAS, version 6.02; SAS Institute Inc., Cary, NC and GraphPad Prism, version 6; GraphPad Software, La Jolla, CA); α was considered significant if P≤0.05.

Disclosures

The salary of J.S.-G. was partially supported by Amgen, Inc. (Thousand Oaks, CA) during design and data acquisition phases of this study. Financial support was received from Amgen, Inc.