J Am Soc Nephrol 11: 2344 2350, 2000
A Comparison of On-Line Hemodiafiltration and High-Flux Hemodialysis: A Prospective Clinical Study
RICHARD A. WARD,* BRBEL SCHMIDT, JEANNINE HULLIN, GNTHER F. HILLEBRAND, and WALTER SAMTLEBEN
*Department of Medicine, University of Louisville, Louisville, Kentucky; Department of Medicine I, Klinikum Grosshadern, University of Munich, Munich, and Kuratorium fuer Dialyse und Nierentransplantation, Neuried, Germany.
Abstract. Some of the morbidity associated with chronic hemodialysis is thought to result from retention of large molecular weight solutes that are poorly removed by diffusion in conventional hemodialysis. Hemodiafiltration combines convective and diffusive solute removal in a single therapy. The hypothesis that hemodiafiltration provides better solute removal than high-flux hemodialysis was tested in a prospective, randomized clinical trial. Patients were randomized to either on-line postdilution hemodiafiltration or high-flux hemodialysis. The groups did not differ in body size, treatment time, blood flow rate, or net fluid removal. The filtration volume in hemodiafiltration was L. Therapy prescriptions were unchanged for a 12-mo study period. Removal of both small (urea and creatinine) and large (2-microglobulin and complement factor D) solutes was significantly greater for hemodiIdentification of 2-microglobulin as the precursor of amyloid deposits in long-term hemodialysis patients (1) has focused attention on the need for renal replacement therapies that remove solutes with molecular weights in excess of 10 kD. Solutes of this size are not removed by conventional hemodialysis, and their removal by diffusion through high-flux hemodialysis membranes is also limited. In 1975, Henderson and colleagues (2) demonstrated greatly enhanced removal of highmolecular-weight solutes by convection through highly permeable membranes. This process, which became known as hemofiltration, involved infusion of a large volume of fluid into the blood entering the filter and its subsequent removal by ultrafiltration. Although hemofiltration provided good removal of high-molecular-weight solutes, it was less efficient than hemodialysis in removing small solutes, such as urea. This limitation led to the development of hemodiafiltration, a hybrid therapy that combined the convective clearance of hemofiltration with the diffusive clearance of hemodialysis (3). Initially, the ability to perform hemodiafiltration under routine clinical conditions
afiltration than for high-flux hemodialysis. The increased urea and creatinine removal did not result in lower pretreatment serum concentrations in the hemodiafiltration group. Pretreatment plasma 2-microglobulin concentrations decreased with time (P 0 0.001); however, the decrease was similar for both therapies (P 0.317). Pretreatment plasma complement factor D concentrations also decreased with time (P 0.001), and the decrease was significantly greater with hemodiafiltration than with high-flux hemodialysis (P 0.010). The conclusion is that on-line hemodiafiltration provides superior solute removal to high-flux hemodialysis over a wide molecular weight range. The improved removal may not result in lower pretreatment plasma concentrations, however, possibly because of limitations in mass transfer rates within the body.

was severely limited by the need for large volumes of sterile substitution solution. The development of systems that use sequential ultrafiltration to prepare sterile substitution solution on-line from water and concentrate (4) has removed the technical constraints to clinical implementation of hemodiafiltration. However, there have been few reports of controlled clinical trials that examine the putative therapeutic advantages of this therapy. Therefore, we compared hemodiafiltration with high-flux hemodialysis in a prospective clinical trial.

Materials and Methods

Study Design
This study was a single-center, prospective, randomized comparison of postdilution hemodiafiltration and high-flux hemodialysis. Patients who had been treated previously by conventional or high-flux hemodialysis at the Neuried KfH dialysis center were paired on the basis of body size, existing treatment time and blood flow rate, and predialysis serum 2-microglobulin concentration. Patients from each pair were randomized to either hemodiafiltration or high-flux hemodialysis and followed for 12 mo as described below. During the first 6 mo, additional patients were recruited to replace any patients who withdrew from the study.
Received September 29, 1999. Accepted April 20, 2000. Correspondence to Dr. Richard A. Ward, Kidney Disease Program, University of Louisville, 615 South Preston Street, Louisville, KY 40202. Phone: 502852-5757; Fax: 502-852-7643; E-mail: richard.ward@louisville.edu 1046-6673/1112-2344 Journal of the American Society of Nephrology Copyright 2000 by the American Society of Nephrology

Patients

Patients who had been stable on thrice weekly hemodialysis for at least 2 mo and who had a permanent blood access capable of delivering a blood flow rate of at least 250 ml/min were eligible for inclusion in the study. The study received ethics committee approval,
Comparison of On-Line Hemodiafiltration and High-Flux Hemodialysis
and informed consent was obtained from all patients before their enrollment in the study.

Hemodiafiltration and High-Flux Hemodialysis
Postdilution hemodiafiltration was performed using a specifically designed system incorporating on-line preparation of substitution solution (AK 100 ULTRA, Gambro, Lund, Sweden) as described previously (4). Briefly, blood is passed through a high-flux filter, where it is subjected to dialysis with ultrafiltration at a rate in excess of that required to achieve the patients dry weight. Fluid balance is maintained by infusing sterile, nonpyrogenic substitution solution into the venous blood line. The substitution solution is derived from ultrapure dialysate by passing it through a single-use ultrafilter immediately before its infusion into the venous blood line. The dialysate is prepared by proportioning ultrafiltered water, liquid acid concentrate, and liquid bicarbonate concentrate made on-line from a dry powder cartridge. This dialysate is then rendered ultrapure by passage through a second ultrafilter. The water supplied to the AK 100 ULTRA for preparation of dialysate and substitution solution met the German microbiologic standard of less than 100 CFU/ml and less than 0.25 EU/ml of endotoxin. The dialysate contained 138 mmol/L sodium, 1 to 4 mmol/L potassium, 1.75 mmol/L calcium, 0.5 mmol/L magnesium, 32 mmol/L bicarbonate, 3 mmol/L acetate, and 1 g/L glucose. The present study used filters containing 1.7 m2 of polyamide membrane (Polyflux 17/17S, Gambro). During the first 6 mo of the study, filters were sterilized with ethylene oxide (Polyflux 17); thereafter, they were steam-sterilized (Polyflux 17S). At entry to the study, the ultrafiltration rate for each patient was set at 25% of the patients blood flow rate. The ultrafiltration rate was then increased until the rate that provided a stable transmembrane pressure of 200 mmHg was found. That ultrafiltration rate was used in all subsequent treatments, unless monitored transmembrane pressures indicated that a change was needed to keep the transmembrane pressure from exceeding 200 mmHg. The AK 100 ULTRA was set to prepare 500 ml/min of dialysate. Actual dialysate flow rates were reduced below 500 ml/min by the flow rate of substitution solution. Typical substitution solution flow rates ranged from 65 to 85 ml/min, so that actual dialysate flow rates during hemodiafiltration ranged from 415 to 435 ml/min. High-flux hemodialysis was performed using a dialyzer containing 1.4 m2 of steam-sterilized polyamide membrane (Polyflux 14S, Gambro) and a dialysate flow rate of 500 ml/min. Other aspects of the patients therapy prescription did not differ between the two groups. Treatment times and blood flow rates, which were individualized for each patient, were unchanged from those in use before entry into the study and remained unchanged throughout the 12 mo of the study. Anticoagulation was achieved using a loading dose and constant infusion of heparin. Net fluid removal was set on an individual basis according to the patients clinical need.

trations of electrolytes, urea, and creatinine were determined by routine clinical laboratory methods. 2-Microglobulin and Complement Factor D. Removal of 2microglobulin was determined at 6-wk intervals. The pre- to posttreatment reduction in plasma 2-microglobulin concentration was calculated using a posttreatment concentration corrected for hemoconcentration according to Bergstrm and Wehle (7). The clearance of 2-microglobulin was calculated using the method of Leypoldt et al. (8). Pretreatment plasma concentrations of complement factor D were determined at entry to the study and after 26, 39, and 52 wk of hemodiafiltration or high-flux hemodialysis using an enzyme-linked immunosorbent assay (9). Pre- to posttreatment reductions in plasma complement factor D concentration were also determined after correcting the posttreatment concentration for hemoconcentration using the method of Bergstrm and Wehle (7). However, care must be taken in interpreting these results because residual heparin interferes with the assay for complement factor D in the posttreatment sample (R. Deppisch and W. Beck, Hechingen, Germany, personal communication, April 3, 2000), possibly because of binding of heparin to factor D (10). Anemia Control. Hemoglobin and hematocrit were determined at 6-wk intervals using routine clinical laboratory methods. All patients received recombinant human erythropoietin. Erythropoietin doses were changed as required to maintain a hematocrit in the range of 30 to 36%. Quality of Life. The patients assessment of their quality of life was determined after 26 and 52 wk of the study using the Kidney Disease Questionnaire (11). (The questionnaire was not administered before entry into the study because a German language version of the instrument was unavailable then.) The Kidney Disease Questionnaire determines quality of life in five dimensions: physical symptoms, fatigue, depression, relationships with others, and frustration. A single interviewer administered the questionnaire to all patients.

Statistical Analyses

Changes in measured variables with time were assessed by repeated measures ANOVA, with the mode of treatment (hemodiafiltration or high-flux hemodialysis) as a between-subjects factor. All statistical testing was performed using the SPSS statistical package (version 8.0 for Windows, SPSS Inc, Chicago, IL). The multivariate statistic used was the Pillais Trace. Data are presented as mean SEM for n observations.

Results

Forty-four patients were randomized to hemodiafiltration or high-flux hemodialysis at the start of the study. Six additional patients were subsequently recruited to replace patients who withdrew from the study during the first 6 mo. Eleven of the 50 patients did not complete 12 mo of study. Three patients withdrew from the study because of worsening hypertension and a marked increase in BP from pre- to posttreatment after the initiation of hemodiafiltration. In these three patients, the average pretreatment BP increased from 156/86 mmHg before entry into the study to 173/93 mmHg in the month before their withdrawal from the study; postdialysis BP as high as 240/120 mmHg were observed. The worsening of hypertension was, however, limited to these three patients. Excluding these three patients, there was a slight but nonsignificant decrease in predialysis mean BP over the course of the study (P 0.103), which was independent of the mode of therapy (P 0.937)
Data Collection and Analysis
Electrolytes, Urea, and Creatinine. Predialysis concentrations of sodium, potassium, calcium, phosphate, bicarbonate, urea, and creatinine were measured at 6-wk intervals. Single-pool Kt/Vurea and eKt/V were calculated from pre- and posttreatment urea concentrations according to Daugirdas (5) and Daugirdas and Schneditz (6), respectively. Creatinine removal was estimated as the reduction in serum creatinine concentration from pre- to posttreatment. Pretreatment blood samples were drawn immediately after access needle insertion. Posttreatment samples were drawn from the arterial blood line 20 s after decreasing the blood flow rate to 80 ml/min. Concen-
Journal of the American Society of Nephrology
(data not shown). Again excluding the three patients with worsening hypertension, the change in mean BP from pre- to posttreatment was also small, did not differ between the two groups (P 0.969), and did not change over the course of the study (P 0.404) (data not shown). The other eight patients who withdrew from the study did so for reasons unrelated to the mode of therapy. These patients withdrew because of transfer to another facility (three patients), death (two patients), renal transplantation (one patient), apparent hypersensitivity to the polyamide membrane (one patient), and prolonged access problems that prevented compliance with the study protocol (one patient). Five of the 11 patients who withdrew from the study (4 from the hemodiafiltration group, including the 3 patients who withdrew because of worsening hypertension, and 1 from the high-flux hemodialysis group) did so within 10 wk of entering the study. These five patients are not included in the following analysis and presentation of data, which is based on 24 patients in the hemodiafiltration group and 21 patients in the high-flux hemodialysis group. Details of the two patient groups and their therapy prescriptions are presented in Table 1. Before entering the study, the patients had been treated by either high-flux (18 hemodiafiltration and 16 high-flux hemodialysis patients) or low-flux (6 hemodiafiltration and 5 high-flux hemodialysis patients) hemodialysis. The two groups did not differ with regard to gender, age, duration of previous dialysis therapy, weight, body mass index, treatment time, or blood flow rate. Five patients in the hemodiafiltration group and three patients in the high-flux hemodialysis group were diabetic. Four patients, two in each group, had some residual renal function (average creatinine clearance, 3.1 ml/min [range, 2.7 to 3.6 ml/min]) at entry to the study. The remaining patients had urine outputs less than 150 ml/d and were considered to have negligible residual renal function. Actual treatment times did not differ from those prescribed and were unchanged throughout the study. Actual blood flow rates were slightly greater than those prescribed in the hemo-

diafiltration group (ml/min versus ml/min) and slightly less than those prescribed in the high-flux hemodialysis group (ml/min versus ml/min). Actual blood flow rates did not change with time, but the difference between groups was significant (P 0.022). Weight and body mass index did not differ between the groups, and neither changed over the course of the study. Net fluid removal averaged 2.7 0.2 kg in the hemodiafiltration group and 2.9 0.2 kg in the high-flux hemodialysis group and did not change over the course of the study or differ between the two groups.
Electrolytes, Urea, and Creatinine
Average pretreatment concentrations of electrolytes for the two groups are presented in Table 2. Pretreatment serum concentrations of sodium, potassium, inorganic phosphorus, and calcium did not differ between the groups. There were statistically significant changes in the pretreatment serum concentrations of sodium, potassium, and calcium over the duration of the study; however, the magnitudes of these changes were very small and of no clinical significance (data not shown). Pretreatment serum bicarbonate concentrations were significantly higher in the hemodiafiltration group than in the high-flux hemodialysis group (P 0.001); however, the magnitude of this difference was independent of the duration of the study (P 0.275). Single-pool Kt/Vurea for the hemodiafiltration group was significantly higher than for the high-flux hemodialysis group (1.58 0.09 versus 1.39 0.09, P 0.020). Results for eKt/V were similar (1.37 0.08 versus 1.21 0.07, P 0.023). At entry to the study, pretreatment serum urea concentrations were significantly higher in the high-flux hemodialysis group than in the hemodiafiltration group (P 0.017). The magnitude of this difference did not change during the study (P 0.372). Pre- to posttreatment reduction in serum creatinine concentration was significantly higher with hemodiafiltration than
Table 1. Patient demographics and treatment prescriptionsa
Hemodiafiltration (n 24) High-Flux Hemodialysis (n 21)
Gender (M:F) Age (yr) Cause of ESRD
Duration of dialysis (mo) Weight (kg) Body mass index (kg/m2) Treatment time (min) Blood flow rate (ml/min) Ultrafiltration volume (L)b
15:3 GN (6), HTN (4), DNS (3), IgA nephropathy (3), PCKD (2), amyloidosis (1), urolithiasis (1), pyelonephritis (1), unknown (3) 66.7 2.9 22.9 0.1
14:3 GN (9), PCKD (5), DNS (3), Balkan nephritis (1), reflux (1), unknown (2) 66.6 2.8 22.7 0.4 2.9 0.2
DNS, diabetes; GN, glomerulonephritis; HTN, hypertension; PCKD, polycystic kidney disease. Prescribed filtration volume after maximization based on transmembrane pressure (see text).
Table 2. Average predialysis serum electrolyte, urea, and creatinine concentrations over the 12-mo study period
Hemodiafiltration (n 24) High-flux hemodialysis (n 21) Pa
Sodium (mmol/L) Potassium (mmol/L) Calcium (mmol/L) Inorganic phosphorus (mg/dl) Bicarbonate (mmol/L) Urea (mg/dl) Creatinine (mg/dl)

5.8 0.1 2.30 0.02 4.8 0.2 21.3 0.5 9.7 0.4
5.8 0.1 2.28 0.03 4.9 0.3 19.6 0.5 11.5 0.4
0.238 0.740 0.616 0.767 0.001 0.001 0.005
P values indicate the significance of the difference between hemodiafiltration and high-flux hemodialysis.
with high-flux hemodialysis (versus 60 1%, P 0.007). Pretreatment serum creatinine concentrations, which were higher in the high-flux hemodialysis group than in the hemodiafiltration group at entry to the study (P 0.005), decreased significantly during the study (P 0.001); however, the decrease was similar in both groups (P 0.565). Hemodiafiltration effected a greater removal of 2-microglobulin than did high-flux hemodialysis as indicated by a significantly higher pre- to posttreatment change in plasma concentration (73 1% versus 58 1%, P 0.001) and clearance (versus ml/min, P 0.001). Pretreatment serum 2-microglobulin concentrations decreased during the study (P 0.001) and were slightly but significantly lower in the hemodiafiltration group than in the highflux hemodialysis group (P 0.045; Figure 1). However, the decrease in pretreatment plasma 2-microglobulin concentrations with time did not differ between the two therapies (P
2-Microglobulin and Complement Factor D
0.317), despite the apparent difference in removal of 2-microglobulin. Pretreatment plasma complement factor D concentrations decreased significantly during the study (P 0.001; Table 3). The magnitude of the change depended significantly on the mode of therapy (P 0.010); hemodiafiltration was associated with a 21% decrease in concentration after 12 mo compared with a 13% decrease with high-flux hemodialysis. Hemodiafiltration was associated with a significantly greater pre- to posttreatment decrease in plasma concentration of complement factor D than was high-flux hemodialysis (33 6% versus 2 8%, P 0.001). However, as indicated earlier, these latter data must be interpreted with caution because of uncertainties in determining the posttreatment concentration.

Anemia Control

Neither hematocrit (30 1% for both groups) nor hemoglobin (10.3 0.2 g/dl versus 10.4 0.2 g/dl for hemodiafiltration and high-flux hemodialysis, respectively) differed between the two groups at entry to the study. Moreover, there was no change in hematocrit or hemoglobin over the course of the study (P 0.307 for hematocrit and P 0.360 for hemoglobin). Because dosing patterns of erythropoietin varied from patient to patient, changes in erythropoietin dose were assessed on the basis of the total weekly dose of erythropoietin received by each patient, regardless of the route or frequency of administration. Overall, the average weekly dose of erythropoietin

Table 3. Predialysis plasma complement factor D concentrationsa
Time (wk) Hemodiafiltration (n 24) High-Flux Hemodialysis (n 21)
Figure 1. Pretreatment serum concentration of 2-microglobulin in patients who were treated with hemodiafiltration (F) and high-flux hemodialysis (). Pretreatment serum concentrations decreased significantly during the study (P 0.001); however, this decrease did not depend on the choice of therapy (P 0.317).
10.4 0.5 9.2 0.5 7.1 0.4 8.0 0.6
9.5 0.6 8.6 0.5 9.1 0.7 8.2 0.6
a Pretreatment concentrations decreased significantly during the study (P 0.001). The decrease was significantly greater for the hemodiafiltration group than for the high-flux hemodialysis group (P 0.010).
increased slightly but significantly over the course of the study (to U/wk, P 0.031). However, this increase was independent of the mode of therapy (P 0.123).

Quality of Life

The patients in both groups had similar perceptions of their quality of life as assessed by the Kidney Disease Questionnaire (Table 4). The patients assessment of their physical symptoms showed a significant improvement during the course of the study (P 0.001); however, this increase did not depend on the mode of therapy (P 0.230). None of the other dimensions of the Kidney Disease Questionnaire showed a change over the course of the study.

Discussion

The results of this study confirm the experience of other investigators that routine on-line hemodiafiltration can be performed safely in a large group of patients for an extended period (1214). Our results also show that hemodiafiltration provides superior solute removal to high-flux hemodialysis over a wide range of solute sizes for blood flow rates in the range of 250 to 300 ml/min. The improvement in solute removal with hemodiafiltration was relatively small for urea and creatinine; however, it may be helpful in treating large patients who tend to have a lower delivered Kt/Vurea than patients with a smaller body size (15,16). The difference in solute removal between the groups was more marked for 2-microglobulin. However, this apparent difference in removal did not result in lower predialysis plasma concentrations with hemodiafiltration compared with high-flux hemodialysis after 1 yr of treatment. The pre- to posttreatment change in concentration of a solute is a good indicator of removal for solutes distributed in a single pool that includes plasma. A substantial rebound in plasma 2-microglobulin concentrations, postdialysis, has been reported (17 20), suggesting that a single-pool model is not adequate to describe 2-microglobulin kinetics, particularly in the face of efficient removal of 2-microglobulin. In this case, the pre- to posttreatment change in concentration will overestimate actual 2-microglobulin removal (21). That intrabody mass transfer rates limit 2-microglobulin removal is supported by the results of other hemodiafiltration studies in which both longer

follow-up periods and filtration volumes of up to 60 L have failed to lower pretreatment 2-microglobulin concentrations below 18 to 20 mg/L (14,22,23). Additional studies of 2microglobulin kinetics during highly efficient therapies are needed to determine the point at which intrabody mass transfer begins to limit 2-microglobulin removal. Increasing 2-microglobulin removal beyond that point will not result in lower serum 2-microglobulin concentrations unless treatment time or frequency is also increased (21). Support for the importance of increased frequency and treatment time over clearance comes from a recent report by Raj et al. (19), who switched patients from conventional thrice-weekly high-flux hemodialysis to nocturnal hemodialysis six nights per week with smaller surface area dialyzers and lower blood and dialysate flow rates. Over 9 mo, they observed a reduction in mean pretreatment serum 2-microglobulin from 27.2 mg/L to 13.7 mg/L. Failure to find a difference in pretreatment 2-microglobulin concentration between high-flux hemodialysis and hemodiafiltration may also result from greater than anticipated removal of 2-microglobulin by high-flux hemodialysis. In practice, highflux hemodialysis represents a form of hemodiafiltration by virtue of the internal filtration and back-filtration that can occur in a dialyzer. Back-filtration flow rates are estimated to be up to 30 ml/min (24). Thus, back-filtration may generate 7 to 8 L of filtrate and substitution fluid flow within the dialyzer in a 4-h treatment, in addition to net fluid removal. That backfiltration in high-flux hemodialysis yields comparable 2-microglobulin removal to hemodiafiltration at low filtration volumes is suggested by the observation of Lornoy et al. (25) that 2-microglobulin removal with hemodiafiltration did not exceed that with high-flux hemodialysis until substitution fluid volumes exceeded approximately 10 L. This study and others fail to show an advantage for hemodiafiltration over high-flux hemodialysis in terms of serum 2-microglobulin concentrations. However, it should not be concluded from these data that hemodiafiltration is without benefit in terms of removing large-molecular-weight solutes. Complement factor D is a 24 kD protein involved in regulating the alternative pathway of complement. Serum complement factor D concentrations are increased in chronic renal failure (26), and at these elevated concentrations it enhances activity of the alternative pathway of complement (27) and inhibits
Table 4. Patient assessment of quality of life based on the Kidney Disease Questionnaire (11)a
Hemodiafiltration 6 Mo 12 Mo High-Flux Hemodialysis 6 Mo 12 Mo P (Time) P (Mode)
Physical symptoms Fatigue Depression Relationships Frustration
3.9 0.3 4.6 0.3 5.6 0.2 5.1 0.3 5.3 0.3
4.8 0.3 4.9 0.4 5.8 0.2 5.2 0.3 5.2 0.4

4.3 0.3 4.7 0.3 5.6 0.3 5.1 0.3 5.3 0.3
4.8 0.4 4.9 0.3 5.6 0.3 5.1 0.3 5.4 0.4
0.001 0.083 0.086 0.077 0.648
0.657 0.910 0.684 0.904 0.851
Each dimension of the Kidney Disease Questionnaire is scored on a seven-point scale, in which 1 is the worst possible score and 7 is the best possible score. P values represent the significance of changes with time (Time) or between hemodiafiltration and high-flux hemodialysis (Mode).
neutrophil degranulation (28). In renal failure, complement factor D accumulates in the intravascular compartment (29). As a result, the impact of intrabody mass transfer on solute removal will be minimal and should not limit the ability of hemodiafiltration to provide superior removal to high-flux hemodialysis. Indeed, we observed significantly greater pre- to posttreatment changes in serum complement factor D concentrations and decreased pretreatment serum concentrations in patients who were treated with hemodiafiltration relative to patients who were treated with high-flux hemodialysis. Unfortunately, heparin interferes with the assay used to measure complement factor D in this study, and the magnitude of the pre- to postdialysis concentration changes must be viewed with caution. A recent report suggested that hemodiafiltration may improve anemia control with reduced erythropoietin doses (14). Such improvement has been ascribed to increased Kt/Vurea (30) or better removal of large-molecular-weight toxins. We could not confirm this observation; however, anemia control was not a primary outcome variable in our study and was not assessed rigorously. Three patients withdrew from the study for reasons that may have been related to hemodiafiltration. These patients developed intratreatment hypertension that was not evident before their entry into the study. Two of the patients had a history of long-standing hypertension controlled by multiple-drug therapy, and the third had a history of borderline hypertension. Early in their experience with hemodiafiltration, Wizemann et al. (31) reported similar problems in a few patients. The reasons for the hypertension are not clear. Removal of sodium may be lower in hemofiltration and hemodiafiltration than in hemodialysis when the substitution fluid sodium concentration is the same as the dialysate sodium concentration (32,33). Thus, on-line hemodiafiltration may be associated with sodium retention, relative to high-flux hemodialysis. Sodium retention could cause vascular volume expansion during a treatment, particularly in fluid-overloaded patients. Alternatively, hemodiafiltration may efficiently remove antihypertensive drugs or endogenous vasodilators, leading to an increase in total peripheral resistance and intratreatment BP. Whatever the reason for hypertension in the three patients, there was no evidence of a generalized association between hemodiafiltration and hypertension. Given the small number of patients and limited follow-up time of this study, it was not possible to address the question of whether the enhanced solute removal associated with hemodiafiltration improves clinical outcomes. There are indications, however, that enhancing the removal of larger solutes does improve outcomes. Leypoldt et al. (34) recently reported an analysis of the 1991 Case Mix Adequacy Study of the United States Renal Data System. After adjustments for case mix, comorbidities, and Kt/Vurea, they found that a 10% increase in vitamin B12 clearance was associated with a significantly reduced relative risk of mortality. Other investigators (3537) have shown that therapies that increase the removal of 2microglobulin postpone the onset of 2-microglobulin amyloid disease compared with conventional low-flux dialysis. Taken together, these data support the need for long-term studies,

involving large numbers of patients, to address the hypothesis that increased removal of large-molecular-weight solutes improves patient outcomes. The results of the present study indicate that on-line hemodiafiltration may be the best means of providing increased removal of large-molecular-weight solutes for these studies.

Acknowledgments

The authors thank Christina Duhr and Jrgen Kbler for their help in sample collection and analysis and Walter Grgen and the nursing staff of the KfH Neuried dialysis center.

References

1. Gejyo F, Yamada T, Odani S, Nakagawa Y, Arakawa M, Kunitomo T, Kataoka H, Suzuki M, Hirasawa Y, Shirahama T, Cohen AS, Schmid K: A new form of amyloid protein associated with chronic hemodialysis was identified as 2-microglobulin. Biochem Biophys Res Commun 129: 701706, 1985 2. Henderson LW, Colton CK, Ford CA: Kinetics of hemodiafiltration. II. Clinical characterization of a new blood cleansing modality. J Lab Clin Med 85: 372391, 1975 3. Leber HW, Wizemann V, Goubeaud G, Rawer P, Schtterle G: Hemodiafiltration: A new alternative to hemofiltration and conventional hemodialysis. Artif Organs 2: 150 153, 1978 4. Ledebo I: On-line hemodiafiltration: Technique and therapy. Adv Ren Replace Ther 6: 195208, 1999 5. Daugirdas JT: Second generation logarithmic estimates of singlepool variable volume Kt/V: An analysis of error. J Am Soc Nephrol 4: 12051213, 1993 6. Daugirdas JT, Schneditz D: Overestimation of hemodialysis dose depends on dialysis efficiency by regional blood flow but not by conventional two pool urea kinetic analysis. ASAIO J 41: M719 M724, 1995 7. Bergstrm J, Wehle B: No change in corrected 2-microglobulin concentration after cuprophane haemodialysis. Lancet 1: 628 629, 1987 8. Leypoldt JK, Cheung AK, Deeter RB: Single compartment models for evaluating 2-microglobulin clearance during hemodialysis. ASAIO J 43: 904 909, 1997 9. Oppermann M, Baumgarten H, Brandt E, Gottsleben W, Kurts C, Gotze O: Quantitation of components of the alternative pathway of complement (APC) by enzyme-linked immunosorbent assays. J Immunol Methods 133: 181190, 1990 10. Keil LB, Jiminez E, Guma M, Reyes MD, Liguori C, DeBari VA: Biphasic response of complement to heparin: Fluid-phase generation of neoantigens in human serum and in a reconstituted alternative pathway amplification cycle. Am J Hematol 50: 254 262, 1995 11. Laupacis A, Muirhead N, Keown P, Wong C: A disease-specific questionnaire for assessing quality of life in patients on hemodialysis. Nephron 60: 302306, 1992 12. Canaud B, Bosc JY, Leray H, Stec F, Argiles A, Leblanc M, Mion C: On-line haemodiafiltration: State of the art. Nephrol Dial Transplant 13(Suppl 5): 311, 1998 13. Pizzarelli F, Cerrai T, Dattolo P, Tetta C, Maggiore Q: Convective treatments with on-line production of replacement fluid: A clinical experience lasting 6 years. Nephrol Dial Transplant 13: 363369, 1998 14. Maduell F, del Pozo C, Garcia H, Sanchez L, Hdez-Jaras J, Albero MD, Calvo C, Torregrosa I, Navarro V: Change from

conventional haemodiafiltration to on-line haemodiafiltration. Nephrol Dial Transplant 14: 12021207, 1999 Sherman RA, Cody RP, Solanchick JC: Body weight and adequacy of hemodialysis: Results of an ESRD network study. ASAIO J 39: 933935, 1993 Ward RA, Buscaroli A, Schmidt B, Stefoni S, Gurland HJ, Klinkmann H: A comparison of dialysers with low-flux membranes: Significant differences in spite of many similarities. Nephrol Dial Transplant 12: 965972, 1997 Maeda K, Shinzato T, Ota T, Kobayakawa H, Takai I, Fujita Y, Morita H: Beta-2-microglobulin generation rate and clearance rate in maintenance hemodialysis patients. Nephron 56: 118 125, 1990 Yasuhiro I, Eiichi N, Mineo O, Muneto Y, Toshihiro I, Masahiro F, Yoshifumi N, Masahito M, Ryuichi M: Removal of serum beta-2 microglobulin using high-performance membranes and analysis of changes in serum BMG levels after dialysis. Am J Nephrol 18: 228 232, 1998 Raj DSC, Ouwendyk M, Francoeur R, Pierratos A: 2-microglobulin kinetics in nocturnal haemodialysis. Nephrol Dial Transplant 15: 58 64, 2000 Leypoldt JK, Cheung AK, Deeter RB: Rebound kinetics of 2-microglobulin after hemodialysis. Kidney Int 56: 15711577, 1999 Clark WR, Leypoldt JK, Henderson LW, Mueller BA, Scott MK, Vonesh EF: Quantifying the effect of changes in the hemodialysis prescription on effective solute removal with a mathematical model. J Am Soc Nephrol 10: 601 609, 1999 Kerr PB, Argils A, Flavier JL, Canaud B, Mion CM: Comparison of hemodialysis and hemodiafiltration: A long-term longitudinal study. Kidney Int 41: 10351040, 1992 Wizemann V, Lotz C, Techert F, Uthoff S: On-line haemodiafiltration versus low-flux haemodialysis: A prospective randomized study. Nephrol Dial Transplant 15[Suppl 1]: 43 48, 2000 Ronco C, Brendolan A, Feriani M, Milan M, Conz P, Lupi A, Berto P, Bettini MC, La Greca G: A new scintigraphic method to characterize ultrafiltration in hollow fiber dialyzers. Kidney Int 41: 13831393, 1992 Lornoy W, Becaus I, Billiouw JM, Sierens L, van Malderen P: Remarkable removal of beta-2-microglobulin by on-line hemodiafiltration. Am J Nephrol 18: 105108, 1998
26. Volanakis JE, Barnum SR, Giddens M, Galla JH: Renal filtration and catabolism of complement protein D. N Engl J Med 312: 395399, 1985 27. Pascual M, Paccaud JP, Macon K, Volonakis JE, Schifferli JA: Complement activation by the alternative pathway is modified in renal failure: The role of factor D. Clin Nephrol 32: 185193, 1989 28. Balke N, Holtcamp U, Hrl WH, Tschesche H: Inhibition of degranulation of human polymorphonuclear leukocytes by complement factor D: FEBS Lett 371:300 302, 1995 29. Pascual M, Steiger G, Estreicher J, Macon K, Volanakis JE, Schifferli JA: Metabolism of complement factor D in renal failure. Kidney Int 34: 529 536, 1988 30. Ifudu O, Feldman J, Friedman EA: The intensity of hemodialysis and the response to erythropoietin in patients with end-stage renal disease. N Engl J Med 334: 420 425, 1996 31. Wizemann V, Birk HW, Techert F: Effects of a modified hemodiafiltration method on low-molecular-weight protein composition in plasma. Blood Purif 8: 4551, 1990 32. Gotch FA, Evans MC, Keen ML: Measurement of the effective dialyzer Na diffusion gradient in vitro and in vivo. Trans Am Soc Artif Intern Organs 31: 354 357, 1985 33. Pedrini LA, Ponti R, Faranna P, Cozzi G, Locatelli F: Sodium modeling in hemodiafiltration. Kidney Int 40: 525532, 1991 34. Leypoldt JK, Cheung AK, Carroll CE, Stannard DC, Pereira BJG, Agodoa LY, Port FK: Effect of dialysis membranes and middle molecule removal on chronic hemodialysis patient survival. Am J Kidney Dis 33: 349 355, 1999 35. Kchle C, Fricke H, Held E, Schiffl H: High-flux hemodialysis postpones clinical manifestation of dialysis-related amyloidosis. Am J Nephrol 16: 484 488, 1996 36. Koda Y, Nishi S, Miyazaki S, Haginoshita S, Sakurabayashi T, Suzuki M, Sakai S, Yuasa Y, Hirasawa Y, Nishi T: Switch from conventional to high-flux membrane reduces the risk of carpal tunnel syndrome and mortality of hemodialysis patients. Kidney Int 52: 1096 1101, 1997 37. Locatelli F, Marcelli D, Conte F, Limido A, Malberti F, Spotti D: Comparison of mortality in ESRD patients on convective and diffusive extracorporeal treatments. Kidney Int 55: 286 293, 1999

PRACTICES AND TECHNOLOGIES DESIGNED TO PROTECT BIRDS
Charles (Rick) Johnson Senior Research Biologist ABR, Inc.
INTRODUCTION More than 20 years have passed since the first production well began pumping oil in Prudhoe Bay. Over the intervening years, our understanding of the impacts of oil development on birds and the mitigation practices to avoid and minimize those impacts have progressed substantially. In this brief description of the practices used to protect birds, I first review the development issues that potentially effect the bird communities in oilfields on the Arctic Coastal Plain, then the species of birds that receive most attention, and finally research and monitoring methods that guide effective mitigation. DEVELOPMENT ISSUES The issues on the Arctic Coastal Plain are general and probably are germane to development elsewhere, but I am going to address specific issues on the coastal plain, which is a breeding area for a diverse assemblage of long-distant migrants and a few resident species. For a full discussion of development impacts and mitigation measures at a recently developed oilfield, I refer readers to the Alpine Project Environmental Evaluation Document (ARCO 1997). The following list of issues pertain to most oilfield developments: 1. habitat loss or modification, either long- or short-term, usually results from placement of gravel pads and roads, airstrips, pipelines and powerlines, and other infrastructure; 2. disturbance from noise, vehicles, aircraft, predators, or people may change habitat use, affect behavior, decrease nest attendance, increase risk of predation, and increase energetic costs. People and predators usually elicit the greatest disturbance responses from birds; 3. death and injury from bird collisions with vehicles and structures such as powerlines claim an unknown number of birds. A growing concern is the potential for collisions of migrating flocks of birds, particularly eiders, that fly along the coast. In fog or conditions of reduced visibility, birds sometimes collide with buildings, towers, and powerlines. Such collisions are infrequent but can kill large numbers of birds in single incidents; 4. predation is a major factor limiting the productivity of ground nesting birds on the coastal plain. Arctic and red foxes, brown bears, Glaucous Gulls, Common Ravens, and several spp. of jaegers are endemic to the coastal plain and prey on birds and their eggs. Nesting colonies of brant on the Colville Delta of over 1,000 nests have suffered near complete failure from a single bear and similar failures have occurred from bears and arctic foxes at Howe Island, a brant and snow goose colony on the
Sagavanirktok River Delta. Foxes, bears, gulls, and ravens are attracted to human food sources, which are available when waste management is ineffective; 5. hydrocarbons and byproducts of oil production potentially can contaminate birds and cause death, injury, decreased productivity, or reduced health; and 6. marine oilspills can be devastating to bird life depending on the location, timing, volume, and containment of a spill. Marine spills are a major concern for offshore oil development. Not all species of birds that occur on the coastal plain receive the same attention when it comes to oilfield development. Two species are federally listed threatened species Stellers Eiders and Spectacled Eidersthat require special clearance before new facilities can be constructed. There are several species that are rare locally (i.e., do not breed in all the available habitat): Yellow-billed Loons, Bar-tailed Godwits, and Peregrine Falcons. Species that are known or suspected to have declining populations, either regionally or globally, are granted more protection than other species, and this list changes as our knowledge improves: Brant, King Eiders, Oldsquaws, Red-throated Loons, and Buff-breasted Sandpipers. Finally, there are species that are protected because they have special subsistence or economic values: Tundra Swans, Greater White-fronted Geese, and Snow Geese. PRACTICES The specific practices and technologies employed for bird research and protection are somewhat dependent on the stage of oil development. The most effective protection for birds and wildlife in general is to incorporate baseline information on distribution, abundance, and habitat use into the design and location of oilfield facilities. A recent example of this strategy is the Alpine development project, which used seven years of baseline studies on the Colville River Delta (Smith et al. 1993, 1994; Johnson 1995, Johnson et al. 1996, 1997, 1998, 1999a) as technical data to identify preferred habitats and specific nest and brood-rearing sites for the species of concern in the area. Using GIS and habitat modeling techniques (see Murphy 2000, in this proceeding; Johnson et al. 1999a), baseline data on site specific use were analyzed to map habitat preferences for individual species and the species data were then integrated into maps of species diversity for different sets of species (e.g., rare species, subsistence species) (ARCO 1997). Through this process specific areas and habitats are identified that may be considered more sensitive to oilfield development. Pads and roads locations can be modified then to avoid or minimize their incursion into valuable bird habitats and specific use areas. Baseline data on the regional distribution, abundance, and habitat use for large showy species are usually collected during surveys from aircraft timed for important periods in the breeding cycle of birds. Site specific surveys are conducted in the proposed location of the oilfield and support facilities for nests and broods of species that are difficult to see from the air are conducted with intensive ground-based searches (Johnson et al. 1999a, 1999b). Although the above techniques are employed prior to development in the planning stages, they also may be employed, along with other techniques, during the construction and operational stages of oilfield development to identify responses of birds. A brief and incomplete description of some of these other techniques follows. Radio and satellite telemetry are used to follow movements of individual birds within oilfields (to identify

nesting and brood-rearing areas as well as bird movements relative to facilities) and beyond (to identify molting, staging, and wintering areas) (TERA 1996). Capture and banding of Brant and Snow Geese are conducted to identify migration routes, staging and wintering areas, and estimate survival rates (Anderson et al. 1999). Radar is used to identify flight paths and flight elevations of migrating birds in places where structures could lead to collisions (Day et al. 1998). Time-lapse photographs or videotapes are used to monitor the effects of disturbance and predation at nest sites (Anderson et al. 1999, Johnson et al. 1999b). Temperature sensors implanted in artificial eggs are used to monitor nesting behavior as part of disturbance studies in oilfields (Anderson et al. 1999, Johnson et al. 1999b). All these techniques are most effective at measuring oilfield impacts when they are conducted before and after construction in both affected and reference sites (Stewart-Oaten et al. 1986), but a number of techniques can be effectively used in paired plot and blocked designs or in gradient analyses after construction or operation has begun (Murphy and Anderson 1993, TERA 1993, Ellis and Schneider 1997). Studies of bird responses to development have provided invaluable information that can be applied to minimize potential impacts in oilfields. Another effective practice to protect nesting birds is the reduction of disturbance and predation. Seasonal restrictions on aircraft, vehicles, noise, and people on foot in nesting and brood-rearing areas help maintain avian productivity. Winter construction, particularly of roads, pads, and pipes, eliminates much of the disturbance related to heavy equipment. Maintaining minimum flight altitudes is an effective measure to reduce aircraft disturbance. The level of predation can be controlled with effective waste and food management. Eliminating the availability of human food to predators reduces the attraction of predators to oilfields and thus reduces the level of predation on birds and their nests. Research is an essential element in the array of practices used to protect and maintain the bird community in the vicinity of oilfields. Continued research and monitoring into habitat selection, the effects of disturbance, and new approaches to mitigation are needed to improve our ability to protect birds and to search for cost-effective management practices. More species are likely to come under concern as global and regional modifications in habitat and climate cause populations to decline. Regardless of where problems occur in a species range, protection of species with declining populations will be a necessity on their breeding grounds. Conclusions The most effective protection of avian communities during oil and gas development is through the design and siting of projects. Baseline data on species abundance and distribution can be integrated into a habitat evaluation through GIS that can take into account multiple rare, sensitive, or socially valued resources. These tools make it possible for development to avoid specific nest sites as well as habitat that has a high potential for use. Another important protection is an effective mitigation program to minimize the attraction of predators and the impacts of disturbance. Controlling the availability of human food to foxes, bears, gulls, and ravens is essential to maintaining a healthy productive community of breeding birds in a developed area. Minimizing disturbance through managing aircraft, vehicle, and pedestrian traffic during the nesting season and during

the brood-rearing season in specific habitats allows the birds opportunities to produce young without additional stress. Research and monitoring into habitat selection, the effects of disturbance, and effective approaches to mitigation continue to be needed to supply information for bird management and protection. More species are likely to decline from global or regional modifications in populations and habitat, and these species will need protection on their breeding grounds. Finally, the cost of proactive strategiescollecting baseline data, effective design and siting of projects, effective mitigation programs, and continued researchmight seem expensive. But it may be cost effective when we consider the costs of increased oversight and regulation by resource agencies, lawsuits and injunctions, missed construction seasons and deadlines, and delayed production. The Alpine oilfield and Tarn development in Kuparuk are examples of new development where this strategy has worked.
Bibliography Anderson, B. A., R. J. Ritchie, A. A. Stickney, and A. M. Wildman. 1999. Avian studies in the Kuparuk Oilfield, 1998. Final report prepared for ARCO Alaska, Inc., Anchorage, by ABR, Inc., Fairbanks, AK. 100 pp. ARCO. 1997. Alpine development project environmental evaluation document. Prepared for U.S. Army Corps of Engineers, Anchorage, AK, by ARCO Alaska, Inc., Anadarko Petroleum Corp., and Union Texas Petroleum Alaska, Corp., Anchorage, AK. Day, R. H., J. R. Rose, and B. A. Cooper. Evaluation of ornithological radar for monitoring eider migration at Point Barrow, Alaska. Unpubl. report prepared for Northern Alaska Ecological Services, U.S. Fish and Wildlife Service, Fairbanks, AK, and Dept. Wildlife Mgmt. North Slope Borough, Barrow, AK, by ABR, Inc., Fairbanks, AK. 42 pp. Ellis, J. I. and D. C. Schneider. 1997. Evaluation of a gradient sampling design for environmental impact assessment. Environmental Monitoring and Assessment 48:157172. Johnson, C. B. 1995. Abundance and distribution of eiders on the Colville River Delta, Alaska, 1994. Final report prepared for ARCO Alaska, Inc., Anchorage, by ABR, Inc., Fairbanks, AK. 12 pp. Johnson, C. B., M. T. Jorgenson, R. M. Burgess, B. E. Lawhead, J. R. Rose, and A. A. Stickney. 1996. Wildlife studies on the Colville River Delta, Alaska, 1995. Fourth annual report prepared for ARCO Alaska, Inc., Anchorage, and Kuukpik Unit Owners by ABR, Inc., Fairbanks, AK. 154 pp. Johnson, C. B., B. E. Lawhead, J. R. Rose, A. A. Stickney, and A. M. Wildman. 1997. Wildlife studies on the Colville River Delta, Alaska, 1996. Fifth annual report prepared for ARCO Alaska, Inc., Anchorage, and Kuukpik Unit Owners by ABR, Inc., Fairbanks, AK. 139 pp.
Johnson, C. B., B. E. Lawhead, J. R. Rose, M. D. Smith, A. A. Stickney, and A. M. Wildman. 1998. Wildlife studies on the Colville River Delta, Alaska, 1997. Sixth annual report prepared for ARCO Alaska, Inc., Anchorage, and Kuukpik Unit Owners by ABR, Inc., Fairbanks, AK. 144 pp. Johnson, C. B., B. E. Lawhead, J. R. Rose, M. D. Smith, A. A. Stickney, and A. M. Wildman. 1999a. Wildlife studies on the Colville River Delta, 1998. Seventh annual report prepared for ARCO Alaska, Inc., Anchorage, and Kuukpik Unit Owners by ABR, Inc., Fairbanks, AK. 102 pp. Johnson, C. B., W. Lentz, J. R. Rose, A. A. Stickney, and A. M. Wildman. 1999b. Alpine Avian Monitoring Program, 1998. Annual report prepared for ARCO Alaska, Inc., Anchorage, and Kuukpik Unit Owners by ABR, Inc., Fairbanks, AK. 46 pp. Murphy, S. M., and B. A. Anderson. 1993. Lisburne Terrestrial Monitoring Program: The effects of the Lisburne Development Project on geese and swans, 19851989. Final synthesis report prepared for ARCO Alaska, Inc., Anchorage, by Alaska Biological Research, Inc., Fairbanks. 202 pp. Smith, L. N., L. C. Byrne, C. B. Johnson, and A. A. Stickney. 1994. Wildlife studies on the Colville River Delta, Alaska, 1993. Final report prepared for ARCO Alaska, Inc., Anchorage, by Alaska Biological Research, Inc., Fairbanks. 95 pp. Smith, L. N., L. C. Byrne, and R. J. Ritchie. 1993. Wildlife studies on the Colville River Delta, Alaska, 1992. Final report prepared for ARCO Alaska, Inc., Anchorage, by Alaska Biological Research, Inc., Fairbanks. 69 pp. Stewart-Oaten, A., W. W. Murdoch, and K. A. Parker. 1986. Environmental impact assessment: psuedoreplication in time? Ecology 67:929940. TERA. 1993. Bird use of the Prudhoe Bay oil field. Unpubl. report prepared for BP Exploration (Alaska) Inc., Anchorage, by Troy Ecological Research Associates, Anchorage, AK. 58 pp. TERA. 1996. Distribution and abundance of Spectacled Eiders in the vicinity of Prudhoe Bay, Alaska: 1994 status report. Unpubl. report prepared for BP Exploration (Alaska) Inc., Anchorage, by Troy Ecological Research Associates, Anchorage, AK. 13 pp + figures.