INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients receiving chronic hemodialysis are more susceptible to tuberculosis (TB) and other infections than the general population. This is attributed to impaired cellular immunity and other immunologic abnormalities associated with end-stage renal disease (1). Published reports from the United States, Canada, England, and Japan in the late 1970s and early 1980s demonstrated that TB is approximately 10 to 15 times more common in chronic hemodialysis patients than their surrounding populations (2). Eight percent of dialysis centers surveyed across the United States, including 20% of dialysis centers in New York State, reported treating patients with active tuberculosis in 1995 (8). The diagnosis of TB may be complicated in patients with end-stage renal disease because they often have atypical clinical presentation of TB, negative skin tests, nonspecific symptoms that may be attributable to uremia, and a higher occurrence of extrapulmonary TB compared with other patient populations (1, 7, 9). Deaths due to TB represented 9% of the total mortality reported between 1994 and 1997 at a dialysis center in Turkey (13). Reports from Turkey and Saudi Arabia report incidences of 23.6% and 28% of TB in hemodialysis patients in areas with overall prevalences of 1% (14, 15). These higher incidences probably reflect the high prevalence of TB in these regions as well as awareness of the increased incidence of TB and its atypical presentation in hemodialysis patients. The authors of these reports suggest that this awareness results in earlier diagnosis and treatment of TB and enhanced patient outcomes.

Chronic hemodialysis complicates the management of antitubercular drug therapy. Rifampin (RIF) and isoniazid (INH) are hepatically metabolized, and dosing is generally unchanged in renal failure (16, 17). Pyrazinamide (PZA) is also hepatically metabolized, but there has been concern regarding the accumulation of its metabolites (pyrazinoic acid and 5-hydroxy-pyrazinamide) in renal failure. Dialyzability of INH and PZA has been demonstrated (18). Although data are lacking, RIF is believed not to be dialyzable because of its relatively high molecular weight and lipid solubility (4). Ethambutol (EMB) is approximately 80% renally cleared and may accumulate in patients with renal insufficiency causing toxicity (23). Dialyzability of EMB has been previously studied (24).

It is important to note that studies that addressed dialyzability of antitubercular drugs were done using slower blood flow rates than are commonly used today, and were done prior to the common use of high-flux hemodialysis membranes. Increases in blood flow rates increase hemodialysis clearance of low-molecular-weight compounds, and high-flux membranes are more permeable to larger molecular weight compounds (27). Therefore, hemodialysis clearances of antimycobacterial drugs reported in the literature may not be accurate for hemodialysis methods used today. Another shortcoming of many of these studies is that recovery of drug in dialysate was not measured. Calculations of hemodialysis clearance of drugs using methods that include drug recovery are more accurate than those that do not. The present study examines the effect of hemodialysis using high blood flow rates and high-flux membranes on the removal of antimycobacterial drugs from serum and reports drug recovery by dialysate collection.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Seven volunteers receiving chronic hemodialysis were recruited from the Rocky Mountain Kidney Center in Denver, Colorado between July 1994 and September 1997. Eligible subjects were those between the ages of 18 and 75 yr with end-stage renal disease requiring chronic hemodialysis. Potential subjects were excluded for any known allergies to antimycobacterial drugs or related agents, history of significant liver disease, baseline liver function tests (alanine aminotransferase [ALT], aspartate transaminase [AST], alkaline phosphatase, or total bilirubin) greater than three times the upper limit of normal, history or clinical evidence of heart failure, severe anemia (hematocrit less than 30%), or pregnancy/lactation. The study protocol was approved by the Colorado Multiple Institutional Review Board. Informed consent was obtained from each subject prior to participation in the study.

Hemodialysis Procedure

At the time of study participation, subjects received their usually prescribed hemodialysis regimens. Dialyzer type; blood, dialysate, and ultrafiltrate flow rates; and duration of hemodialysis were not changed for study purposes. All of these parameters and the number of times the dialyzer had been previously used were recorded. The system used for all patients was a Centrysystem 3 (Cobe Laboratories, Lakewood, CO) hemodialysis machine with a polysulfone, 1.8 m² Fresenius Hemoflow F80B (Fresenius USA, Inc., Walnut Creek, CA) high-flux dialyzer and single-pass dialysate flow.

Medications

Six subjects were given single oral doses of INH (UDL Laboratories, Rockford, IL) 300 mg, RIF (Rimactane; Novartis Pharmaceuticals, Summit, NJ) 600 mg, PZA (Lederle Standard Products, Philadelphia, PA) 1,000 mg (which ranged from 9 to 13 mg/kg), and EMB (Myambutol; Lederle Laboratories, Philadelphia, PA) 25 mg/kg and instructed to take these all at once on an empty stomach 2 h before a regularly scheduled hemodialysis session. Subjects were contacted by telephone at the time study medications were due and reminded to take them. The seventh subject was being treated for TB with INH 600 mg, RIF 600 mg, and PZA 2,000 mg (20 mg/kg) daily. He was studied with his usual doses and was not given EMB.

Sample Collection

Blood samples (8 ml each) to be analyzed for INH, RIF, PZA, and EMB concentrations were taken from ports in the arterial and venous tubing of the dialyzer at the start of the hemodialysis session and hourly, with final samples taken as hemodialysis was discontinued. Samples were collected in plain red-top vacuum tubes and placed on ice, centrifuged within 60 min of collection, and serum was harvested into labeled polypropylene tubes and stored at 70° C until assayed. Samples were frozen within 90 min of collection. For the duration of the hemodialysis session, all dialysate/ultrafiltrate fluid coming out of the system was collected in 20-L containers. When each container approached full, it was removed from the system and the total volume collected was recorded. From each container, an 80-ml dialysate/ultrafiltrate sample was collected, frozen promptly at 70° C, and the rest of the fluid was discarded.

Sample Analysis

All samples were assayed for drug concentrations at the Infectious Disease Pharmacokinetics Laboratory at National Jewish Medical and Research Center in Denver, Colorado according to validated procedures.

INH

All INH assays were performed using a validated high-performance liquid chromatography (HPLC) assay on a Waters (Milford, MA) 510 pump and Model 680 gradient controller with a solvent select valve, a Spectra Physics (San Jose, CA) Model 8875 fixed-volume autosampler, a Waters Model 486 ultraviolet detector, a Macintosh IIci computer (Apple Computers, Inc., Cupertino, CA), and the Rainin (Woburn, MA) Dynamax HPLC data management system. The six-point standard curves for the INH assays ranged from 0.5 to 20 µg/ml, with linearity extending well above this range. The absolute recovery of INH from serum was 61%, as determined by comparing peak height counts across four serum curves to an unextracted solvent curve. The within-day precision (percentage of coefficient of variation [%CV]) of validation quality control (QC) samples was 1 to 6%, and the overall validation precision was 6 to 10%. QC sample concentrations were 0.8, 6, and 13 µg/ml. The dialysate method was a modification of the serum method, with similar recovery and reproducibility.

RIF

All HPLC assays used the same equipment as for INH. The six-point standard curves for the RIF assays ranged from 0.5 to 50 µg/ml, with linearity extending well above this range. The absolute recovery of RIF from serum was 95.5%, as determined by comparing peak height counts across four serum curves to an unextracted solvent curve. The within-day precision (%CV) of validation QC samples was 2.4 to 4.6%, and the overall validation precision was 6.3 to 7.1%. QC sample concentrations were 3, 8, and 26 µg/ml. The dialysate method was a modification of the serum method, with similar recovery and reproducibility.

PZA

All assays were performed using a validated assay on a Hewlett-Packard (Wilmington, DE) Model 5890 Series II gas chromatograph (GC) with a Hewlett-Packard Model 5971A mass selective detector. The six-point serum standard curves for PZA assays ranged from 5 to 100 µg/ml, with linearity extending well above and below this range. The absolute recovery of PZA from serum was 100.5%, as determined by comparing area counts across four serum curves to an unextracted solvent curve. The within-day precision (%CV) of validation QC samples was 2.2 to 3.2%, and the overall validation precision was 2.8 to 3.3%. QC sample concentrations were 16, 32, and 66 µg/ml. The dialysate method was a modification of the serum method, with similar recovery and reproducibility.

EMB

All assays were performed using the same equipment as the PZA assays. The serum standard curves for EMB ranged from 0.2 to 10 µg/ ml. The absolute recovery of EMB from serum was 95.8%. The within-day precision (%CV) of validation QC samples was 2.2 to 4.1%, and the overall validation precision was 2.8 to 3.3%. QC sample concentrations were 0.4, 2, and 6 µg/ml. The dialysate method was a modification of the serum method, with similar recovery and reproducibility.

For INH, RIF, and EMB, drug concentrations in dialysate were too low to be quantified. Therefore, samples were concentrated tenfold by lyophilizing 10-ml aliquots of the dialysate and reconstituting the samples to a volume of 1.0 ml. These assay results were divided by a factor of 10 to determine the dialysate concentrations. PZA concentrations in dialysate were determined by the same method used for the PZA serum samples and did not require concentration for accurate quantitation.

Hemodialysis Clearance Determination

Previously published equations for calculating hemodialysis clearance were used (28). As calculations based on extraction ratios derived from arterial-venous (A-V) concentration differences (Equations 1 and 2) have fallen out of favor to methods based on dialysate recovery of drugs, they were used primarily to facilitate comparison of this study's results to existing literature.

(1)

(2)

where Cl_HDb = hemodialysis clearance from blood, Cl_HDs = hemodialysis clearance from serum, _b = blood flow rate, A_s = arterial serum concentration, V_s = venous serum concentration, and Hct = hematocrit.

These equations were applied to samples collected 1 and 2 h into hemodialysis and at the end of hemodialysis. The median of the three values was taken for each subject to determine a median value for the group.

Calculations based on drug recovery in dialysate were made using Equations 3 and 4.

(3)

(4)

where Cl_HD = hemodialysis clearance, R = total amount of drug recovered in dialysate, AUC_s = area under the serum concentration versus time during hemodialysis, _d = dialysate flow rate, and AUC_d = area under the dialysate concentration versus time curve.

Area under the curve (AUC) calculations were made using the linear trapezoidal rule programmed into Microsoft Excel 97 (Microsoft Corp., Redmond, WA) spreadsheets. Drug recovery (R) in dialysate was calculated by multiplying the concentration of each dialysate sample by the total volume of the corresponding collection. Drug recovery as a percentage of the dose administered was determined by dividing the total amount of drug recovered in dialysate by the administered dose. No corrections for expected bioavailability were made because bioavailability of the drugs studied specific to this patient population has not been studied.

Statistical Analysis

All statistical analyses were performed using JMP version 3.2 software (SAS, Inc., Cary, NC). Means with standard deviations and medians with ranges were determined for patient demographics and hemodialysis procedure characteristics as well as hemodialysis clearance calculations.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient demographics and hemodialysis characteristics are summarized in Table 1. Because most subjects had been receiving chronic hemodialysis for longer than 1 yr, it is unlikely that they had significant residual renal function.

Although subjects were instructed to take study medications 2 h before hemodialysis, the median time from dosing to start of hemodialysis was 2.2 h, ranging from 1.3 to 3.2 h. The drugs studied typically reach their maximal serum concentration (Cmax) within 1 to 3 h (Tmax) after oral administration (31). The apparent Cmax and Tmax for each drug studied along with normal values are summarized in Table 2. The apparent Cmax was observed with the second or later samples in two subjects for INH, seven for RIF, two for PZA, and four for EMB; indicating that absorption was not complete prior to initiating hemodialysis in these subjects. One subject is not included in the PZA analyses because PZA was not detectable in any of this subject's serum samples. Because PZA is very reliably absorbed after oral administration, it appears that this subject did not take the PZA dose as instructed.

Calculated hemodialysis clearances and drug recovery in dialysate are summarized in Tables 3456.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Limitations and assumptions of hemodialysis clearance calculations are reviewed elsewhere (29). Briefly, Equation 1 is internally inconsistent because it combines whole blood flow rate and serum extraction ratio. Equation 2 uses the Hct to estimate serum flow making the equation internally consistent. Calculations based on dialysate recovery, such as Equations 3 and 4, are more accurate because they are not affected by serum/red cell equilibrium (which may affect Equation 1) or ultrafiltration (which may affect Equations 1 and 2).

INH

Hemodialysis clearance of INH has been reported as 90.0 ml/ min (using Equation 1) resulting in 73.3% removal of a 5 mg/kg dose (19). As this study did not report dialysate recovery, it is likely that the reported removal reflects both hepatic metabolism and hemodialysis removal. Our median INH recovery in dialysate was only 9.2% of the dose, suggesting that hepatic metabolism remains the primary mechanism of clearing INH in patients receiving hemodialysis.

RIF

Our finding that RIF is not significantly removed by hemodialysis was expected owing to this drug's large molecular weight, wide distribution into tissues, 80% protein binding, and rapid hepatic clearance (23).

PZA

PZA was significantly removed by hemodialysis, with a median dialysate recovery of 45.1% of the dose. This is slightly higher than a previous report of 31.5% removal, which was estimated from calculated clearance rather than dialysate recovery (22).

Pyrazinoic acid, the primary metabolite of PZA, may contribute to the antibacterial activity and toxicity of PZA, although the precise role has not been defined (21, 35). Pyrazinoic acid has been shown to accumulate in patients with renal failure and to be significantly removed by hemodialysis (21, 22). We did not determine pyrazinoic acid concentrations.

EMB

EMB is normally 80% renally eliminated and may accumulate in renal failure, predisposing to optic neuritis (23). Our median hemodialysis clearance (using Equation 2) was higher than previously reported at 56 to 81 ml/min (26). Our median dialysate recovery of 1.5% of the dose is lower than previously reported at 6.9 to 16.7% (26). In both studies, dialysate recoveries are lower than expected from estimated extraction ratios and hemodialysis clearances. Potential explanations for this discrepancy may include EMB's relatively large volume of distribution, incomplete absorption of drug prior to starting dialysis, or adherence of drug to the dialyzer membrane or tubing.

Ideally, subjects would have been given study medications intravenously to avoid problems associated with incomplete absorption and to allow complete distribution prior to starting dialysis. Injectable forms of PZA and EMB are not available in the United States, and it was not practical to administer only INH and RIF intravenously to our study patients. Using oral agents may have enhanced clinical relevance of our results, because oral agents are routinely used.

Drug absorption was incomplete prior to starting hemodialysis in several subjects. Drug recoveries and AUCs may have been slightly higher if serum concentrations were higher during hemodialysis. Because AUCs during dialysis were used only in combination with corresponding dialysate concentrations, hemodialysis clearances calculated from AUCs should still be valid.

Due to logistical constraints, we were unable to determine drug clearances between hemodialysis sessions or post-hemodialysis rebound in drug concentrations. However, these limitations do not alter our conclusions, which are based on the recovery of drug in dialysate.

The median observed Cmax and Tmax of RIF were 5.5 µg/ml and 3.3 h, compared with expected values of 11.8 µg/ml and 1.5 h (31). This suggests that chronic hemodialysis patients have delayed and decreased absorption of RIF. This may be the result of underlying disease states in some patients or food ingestion near the time of dosing (33). Data in Table 2 may suggest that INH and PZA absorption are also decreased in these patients. However, we did not sample early enough to assess Cmax for these drugs. Because our study used a lower dose of PZA (1,000 mg versus 1,500 mg), the comparison study data were adjusted by a factor of two-thirds.

Dosing Recommendations

Our results support previous recommendations that INH and RIF be given in their usual daily doses of 300 and 600 mg respectively (16, 17). Supplemental dosing of INH, RIF, or EMB to account for hemodialysis removal does not seem warranted. Our results support previous recommendations that PZA be dosed at 25 to 30 mg/kg three times a week (17) and that EMB be dosed at 15 to 25 mg/kg three times a week (16, 17). This should provide adequate Cmax, avoid accumulation of PZA metabolites, and avoid accumulation of EMB. PZA should be administered after hemodialysis to avoid premature drug removal. All four drugs may be administered after hemodialysis to facilitate directly observed therapy.