Evaluation of Vancomycin and Gentamicin Dialysis Clearance Using In Vivo and In Vitro Systems

Abstract

Advances in hemodialysis (HD) techniques have increased the potential for drug removal. Quantifying drug clearance in clinical studies for all possible dialysis conditions is impractical, given the variability in dialysis conditions. The purpose of this study was to determine the dialysis clearance (CL_D) of vancomycin and gentamicin using in vitro and in vivo methods and evaluate the applicability of in vitro data. In vitro dialysis was used to determine the CL_D of vancomycin and gentamicin under conditions of intermittent HD (IHD) and sustained low-efficiency dialysis (SLED). Two Fresenius polysulfone dialyzers were studied: F180NR for IHD and F50 for SLED. Data were compared with in vivo CL_D determined in patients with end-stage renal disease receiving IHD and from the literature for SLED. Under IHD conditions, in vitro CL_D of vancomycin and gentamicin was 131 ± 3 and 154 ± 3 mL/min, respectively, and under SLED condition it was 72 ± 9 and 84 ± 11 mL/min, respectively. These values were 11–27% higher than in vivo CL_D for IHD (103 ± 15 mL/min for vancomycin and 132 ± 25 mL/min for gentamicin) and SLED (63 mL/min for vancomycin and 76 ± 38 mL/min for gentamicin). There was a statistically significant difference in vancomycin clearance by IHD for the in vitro study compared with in vivo data (p = 0.012), but not for gentamicin (p = 0.18). In vitro methods overestimated in vivo CL_D, but are reasonable to assist with drug regimen design if one considers the limitations.

• gentamicin
• hemodialysis
• in vitro dialysis
• sustained low-efficiency dialysis
• vancomycin

INTRODUCTION

Technological advances in hemodialysis (HD) techniques to improve clearance of uremic toxins also affect drug elimination. Variability in dialysis conditions, such as the dialyzer used, blood and dialysate flow rates, and ultrafiltration rate (UFR), precludes a “one dose fits all” approach when designing drug dosing regimens for patients requiring renal replacement therapies. Most in vivo studies report drug clearances for select dialyzers under specific dialysis conditions, often quantified in a limited number of patients. Although this approach may have been satisfactory when patients were dialyzed under similar conditions, it is not valid with modern dialysis techniques. Today the conditions for dialysis vary based on the type of renal replacement therapy, for example, intermittent HD (IHD), continuous renal replacement therapies, or sustained low-efficiency dialysis (SLED).

In vitro dialysis systems have been used to a limited extent to quantify solute removal, but more extensive use of this technique may provide a more efficient method for determining drug removal under varying dialysis conditions.^1–6 Benefits of using an in vitro system include reduced time and cost of pharmacokinetic studies, reduced burden on patients, and the ability to alter dialysis conditions to simulate the variety of renal replacement modalities used in clinical practice.

Vancomycin and gentamicin are antibiotics commonly used in patients requiring dialysis and have characteristics conducive to removal by dialysis. Gentamicin has a relatively low molecular weight (∼477 Da), small volume of distribution (∼0.35 L/kg), and low degree of protein binding (<10%).⁷ Vancomycin is larger than gentamicin (1448 Da) and has a higher degree of protein binding [30–55% in patients with normal kidney function; 18% in patients with end-stage renal disease (ESRD)] making it less likely to be removed during dialysis.⁸ The difference in molecular weight allows for the assessment of removal of relative small and large molecular weight drugs.

The purpose of this study was to determine dialysis clearance (CL_D) of vancomycin and gentamicin by dialyzers used for IHD and SLED at our institution using in vitro and in vivo systems and explore whether the in vitro system is a predictive model of in vivo drug removal.

MATERIALS AND METHODS

In Vitro Study

The in vitro dialysis model consisted of a “drug reservoir” containing 5 L of Sorensen’s phosphate buffer solution (pH 7.4), vancomycin (100 mg/L), and gentamicin (50 mg/L). The drug reservoir also contained bovine albumin (2 g/dL) to incorporate protein for binding. The marker compounds urea (250 mg/dL) and creatinine (40 mg/dL) were added to allow comparison of clearance data for these compounds with that provided in the package inserts for each dialyzer. The drug reservoir was maintained at 37°C and continuously mixed with a magnetic stirrer. Polyvinyl chloride tubing connecting the drug reservoir to the dialyzer served as the “arterial” line and that connecting the dialyzer back to the drug reservoir served as the “venous” line, thus forming a closed-loop, fixed volume system. Dialysate lines were connected from the dialysate supply containing standard bicarbonate dialysate to the dialyzer and from the dialyzer into the drainage system. All dialysis procedures were performed using a Gambro Phoenix^® dialysis machine (Gambro, Lakewood, CO, USA).

The dialysis was performed using two polysulfone dialyzers, the Optiflex^® F180NR (K_uf = 60 mL/h/mmHg, surface area = 1.8 m², KoA of urea = 1239 mL/min) and the Hemoflow F50 (K_uf = 29 mL/h/mmHg, surface area = 1.0 m², KoA of urea = 589 mL/min), both manufactured by Fresenius (Fresenius Medical Care, Waltham, MA, USA).^9–11 These dialyzers are used at our institution for IHD (F180) and SLED (F50).

Conditions for simulated IHD were a reservoir flow rate (Q_R) of 200 mL/min and a dialysate flow rate (Q_D) of 500 mL/min. Because the in vivo study was not completed at the time of initiation of the in vitro study, we chose flow rates predicted to correspond to the average plasma flow rate and dialysate flow rates typically achieved in our inpatient IHD population. SLED conditions were a Q_R of 150 mL/min and a Q_D of 350 mL/min. The ultrafiltration flow rate was set to the minimum rate for the dialysis machine (∼3 mL/min) for all procedures. In vitro dialysis was performed for 90 min and repeated three times using a new dialyzer for each procedure. Reservoir samples were collected from the “arterial” line at the start of the procedure and simultaneously with “venous” line samples at 5, 15, 30, 60, and 90 min. In vitro vancomycin and gentamicin concentrations were analyzed using fluorescence polarization immunoassays (TDx: Abbott Diagnostics, North Chicago, IL, USA). The lower limit of detection was 2.0 mg/L for vancomycin and 0.27 mg/L for gentamicin. Creatinine concentrations were determined by the kinetic Jaffe method (Stanbio Laboratories, Boerne, TX, USA) and urea concentrations were measured by the modified Berthelot method (Stanbio Laboratories).

The CL_D of vancomycin, gentamicin, urea, and creatinine was calculated as (1)

where C_a is the “arterial” concentration and C_v is the “venous” concentration.

The mean CL_D was calculated using data from each of the five simultaneous arterial and venous samples. The CL_D values of vancomycin and gentamicin from the IHD procedures were compared with data from the in vivo portion of this study. The CL_D values from the in vitro SLED procedures were compared with values reported in the literature under similar conditions.^12,13

In Vivo Study

This was a prospective, open-label study. Adult patients (18–80 years) with ESRD requiring IHD who were receiving vancomycin or gentamicin or both as part of their clinical care during hospitalization were eligible for inclusion. Patients were excluded if they had a hemoglobin level less than 8 g/dL or a hematocrit (HCT) level less than 24%, were dialyzed using a dialyzer other than the F180NR, were unable to give informed consent, received IHD outside of the inpatient dialysis unit, or had received vancomycin or gentamicin more than 72 h before their scheduled dialysis session. This study was approved by the Methodist University Hospital institutional review board and procedures followed were in accordance with the Helsinki Declaration. All subjects provided written informed consent.

Study procedures were performed during the IHD session that followed the administration of vancomycin or gentamicin. Doses of these agents were determined by the prescriber and the dialysis prescription was determined by the prescribing nephrologist. During HD, 5 mL of blood was collected immediately before IHD and then simultaneously from the arterial and venous ports at regular intervals during the IHD procedure (maximum of five sets of arterial and venous samples) for the measurement of vancomycin and gentamicin concentrations. Blood samples were processed immediately after the IHD session by the laboratory at Methodist University Hospital using SYNCHRON LX 20 PRO^® (Beckman Coulter, Fullerton, CA, USA) for the measurement of vancomycin and gentamicin concentrations by a particle-enhanced turbidimetric inhibition immunoassay. The lower limits of quantification for this assay are 3.5 mg/L for vancomycin and 0.5 mg/L for gentamicin.

The delivered blood flow rate (Q_b), dialysate flow rate (Q_D), UFR, and the HCT level on the day of dialysis were also recorded.

Dialysis drug clearance (CL_D) at each time point was calculated as follows: (2)

where C_a and C_v represent the arterial and venous drug concentrations, respectively.

The correction of the blood flow rate based on the HCT level provides the plasma flow rate. The mean CL_D was calculated using data from each of the simultaneous arterial and venous samples.

Statistical Analysis

Demographic data were expressed as mean ± SD. The in vitro vancomycin and gentamicin clearances were compared with in vivo clearances determined in this study using Student’s t-test. A p-value <0.05 was considered statistically significant.

RESULTS

In Vitro Study

The mean CL_D values for vancomycin, gentamicin, urea, and creatinine under each set of conditions tested are shown in Table 1. In vitro CL_D values for urea and creatinine reported by the manufacturer were determined at higher blood and dialysate flow rates limiting a direct comparison with results from this study. Urea and creatinine clearances reported by the manufacturer for the F180NR at a Q_b of 300 mL/min and Q_D of 500 mL/min were 274 and 251 mL/min, respectively.⁹ Urea and creatinine clearances for the F50 dialyzer at a Q_b of 200 mL/min and Q_D of 500 mL/min were 175 and 158 mL/min, respectively.¹⁰

Table 1.  In vitro conditions and solute clearances for IHD and SLED.

Conditions	IHD	SLED
Dialyzer	F180	F50
QR (mL/min)	200	150
QD (mL/min)	500	350
CLD (mL/min)
Vancomycin	131 ± 2	72 ± 9
Gentamicin	154 ± 2	84 ± 11
Urea	195 ± 2	129 ± 17
Creatinine	187 ± 3	120 ± 14

Notes: CL_D, dialysis clearance reported as mean ± SD (mL/min); IHD, intermittent hemodialysis; Q_D, dialysate flow rate (mL/min); Q_R, reservoir flow rate (mL/min); SLED, sustained low-efficiency dialysis; F180, surface area 1.8 m², K_uf 60 mL/h/mmHg; F50, surface area 1.0 m², K_uf 29 mL/h/mmHg.

In Vivo Study

A total of 16 patients were recruited for the in vivo dialysis study. Two patients were excluded because the majority of vancomycin concentrations were below the lower limit of detection for the assay and could not be used for the calculation of CL_D. Of the remaining 14 patients, 10 received vancomycin and 7 received gentamicin. Three patients received concomitant vancomycin and gentamicin. Baseline characteristics and dialysis parameters of the 14 patients are reported in Tables 2 and 3. No patient had received a dose of vancomycin or gentamicin within 8 h of the scheduled IHD session; therefore, distribution of drug following administration was complete. The mean vancomycin and gentamicin doses administered before HD were 800 ± 400 mg (or 10 mg/kg) and 177 ± 145 mg (or 2 mg/kg), respectively. The mean time between drug administration and the start of dialysis was 36 ± 12 h for vancomycin and 30 ± 14 h for gentamicin. Blood samples were collected during dialysis according to the designated sampling schedule. The mean time of dialysis during which there were detectable drug levels was 1.8 ± 1 h for vancomycin and 0.93 ± 0.19 h for gentamicin. The mean CL_D for vancomycin was 103 ± 15 mL/min and for gentamicin was 132 ± 25 mL/min at a mean blood flow rate of 357 ± 45 mL/min for patients receiving vancomycin and 336 ± 54 mL/min for patients receiving gentamicin. These rates correspond to a plasma flow rate of 240 ± 34 mL/min for vancomycin and 225 ± 37 mL/min for gentamicin.

Table 2.  Patient characteristics and dialysis conditions for vancomycin in vivo study.

Subject number	Age (years)	Gender	Weight (kg)	HD access	Hct level (%)	Qb (mL/min)	Qp (mL/min)	QD (mL/min)	UFR (mL/min)	CLD (mL/min)
1	21	M	67	Fistula	34.5	350	229	600	17	99
2	39	M	93	Graft	27.1	400	292	600	17	136
3	34	F	99	Vascath	30.1	350	245	600	6	95
4	62	M	88	Graft	33.8	400	265	600	14	114
5	48	F	79	Fistula	34.6	300	196	600	6	103
6	58	M	70	Graft	36.4	400	254	600	35	108
7	66	F	97	Fistula	30.4	400	278	600	12	103
8	30	F	91	Fistula	39.8	350	211	700	22	79
9	60	F	60	Vascath	29.7	270	190	600	16	87
10	53	M	72	Permacath	29.8	350	246	700	20	106
Mean ± SD	47 ± 15	–	82 ± 14	–	32.6 ± 3.8	357 ± 45	240 ± 34	620 ± 42	17 ± 8	103 ± 15

Notes: CL_D, dialysis clearance; F, female; Hct, hematocrit; HD, hemodialysis; M, male; Q_b, blood flow rate; Q_D, dialysate flow rate; Q_p, plasma flow rate; UFR, ultrafiltration rate.

Table 3.  Patient characteristics and dialysis data for gentamicin in vivo study.

Subject number	Age (years)	Gender	Weight (kg)	HD access	Hct level (%)	Qb (mL/min)	Qp (mL/min)	QD (mL/min)	UFR (mL/min)	CLD (mL/min)
1	58	M	76	Graft	33.8	400	265	600	12	172
2	55	F	83	Fistula	25.2	300	224	500	12	141
3	30	F	91	Fistula	39.8	350	211	700	22	124
4	60	F	60	Vascath	29.7	270	190	600	16	86
5	53	M	72	Permacath	29.8	350	246	700	20	140
6	52	F	139	Permacath	32.7	400	269	500	22	128
7	53	M	89	Vascath	39.3	280	170	600	17	133
Mean ± SD	52 ± 10	–	87 ± 25	–	32.9 ± 5.3	336 ± 54	225 ± 37	600 ± 82	17 ± 4	132 ± 25

Notes: CL_D, dialysis clearance; F, female; Hct, hematocrit; HD, hemodialysis; M, male; Q_b, blood flow rate; Q_D, dialysate flow rate; Q_p, plasma flow rate; UFR, ultrafiltration rate.

Comparison of In Vitro and In Vivo Data

Dialysis clearances determined under conditions of IHD in the in vitro study overestimated in vivo clearance by 27% for vancomycin and 17% for gentamicin (CL_D data shown in Tables 1–3). There was a statistically significant difference in vancomycin clearance by IHD for the in vitro study compared with in vivo data (p = 0.012), but not for gentamicin (p = 0.18). Dialysis clearances of vancomycin and gentamicin by SLED determined from other studies are reported in Table 4. Clearance values from the in vitro phase of this study overestimated the reported values by 14% for vancomycin and 11% for gentamicin.

Table 4.  In vivo SLED data previously reported.

Study	Vancomycin^12	Gentamicin^13
Number of subjects	10	8
Dialyzer	F60^a	F50^b
Qb (mL/min)	160	200
QD (mL/min)	160	300
CLD (mL/min)	63	76

Notes: CL_D, dialysis clearance reported as mean ± SD (mL/min); Q_b, blood flow rate; Q_D, dialysate flow rate; SLED, sustained low-efficiency dialysis.

^aF60, Fresenius polysulfone dialyzer; surface area 1.3 m², K_uf = 38 mL/h/mmHg.

^bF50, surface area 1.0 m², K_uf 29 mL/h/mmHg.

DISCUSSION

Vancomycin and gentamicin clearances were determined to provide specific clearance information for the dialyzers used at our institution and to serve as model solutes for the in vitro study. The in vitro and in vivo gentamicin CL_D values for the IHD procedures were higher than the rates previously reported.^14,15 Dialysis conditions were similar in these studies; however, different dialyzers were used. Sowinski et al.¹⁴ reported a median gentamicin clearance of 104 mL/min in patients dialyzed using the cellulose acetate high-performance dialyzer [CAHP-210 (Baxter International, Inc., Deerfield, IL, USA) ; surface area 2.1 m², K_uf 13.2 mL/h/mmHg]. Amin et al.¹⁵ reported a mean gentamicin clearance of 116 ± 9 mL/min using a high-flux polysulfone dialyzer [F80 (Fresenius Medical Care, Waltham, MA, USA); surface area 1.6 m², K_uf 52 mL/h/mmHg]. The in vitro and in vivo vancomycin CL_D values for IHD procedures were in agreement with the values reported from other in vivo studies with polysulfone membranes (108–130 mL/min).^16–18 For SLED procedures the mean in vitro gentamicin CL_D was 11% higher than that reported by Manley et al.¹³ Similarly, the in vitro vancomycin CL_D during simulated SLED was 14% higher than that reported by Kielstein under similar dialysis conditions.¹² We designed this study to simulate the SLED conditions used at our institution, which were not identical to the comparative studies. The dialysate flow rate of 350 mL/min used for the in vitro SLED simulation was above that used in the comparative in vivo studies (Table 4). The plasma flow rates used in the in vivo studies were not reported; however, our plasma flow rate of 150 mL/min was likely higher because the blood flow rates from the in vivo studies were 160 and 200 mL/min (blood flow rate is higher than plasma flow rate).

The overestimation of CL_D by our in vitro system may be explained in part by differences in dialysis conditions and limitations of an in vitro system. One limitation of our in vitro system is the lack of protein binding. We attempted to provide some level of protein binding by adding albumin to the drug reservoir. The relatively low albumin concentration of 2 g/dL used in the in vitro study was selected based on the fact that often our inpatient IHD population has albumin levels below normal. Because gentamicin is less than 10% bound to plasma protein it is unlikely that incorporating higher concentrations of albumin would significantly alter the CL_D. Protein binding of vancomycin is between 30% and 55% in patients with normal kidney function; however, the extent of binding is reduced to approximately 18% in patients with ESRD.⁸ We also did not evaluate binding of drug to the dialyzer during the in vitro studies, an effect that would contribute to overestimation of CL_D. Ultrafiltration was also minimized during the in vitro procedures by setting the transmembrane pressure at the minimum level allowed for the dialysis machine, which yielded an UFR of approximately 3 mL/min, a much lower setting than that used in clinical practice for IHD. Omission of the ultrafiltration component of dialysis removes the convective portion of CL_D of vancomycin that has been reported to play a role in the removal of larger molecules.¹ The magnitude of this role, however, is less with high-flux dialyzers.¹⁹

The overestimation of drug clearance with the in vitro system is not surprising, given these differences in protein binding, volume of distribution, and convective clearance between the in vitro and in vivo systems. The extent to which in vitro methods overestimate true in vivo drug clearance requires further study. Although there were limitations in using an in vitro dialysis system, this system provided reasonable estimates of in vivo removal of two commonly used antibiotics in the ESRD population. Variation in observations is likely due to limitations of in vitro conditions, but estimates are practical to assist with drug regimen design if one considers such limitations. Clearance data determined for individual dialyzers using this in vitro method may be used in conjunction with other pharmacokinetic parameters to predict the likelihood of drug removal during dialysis. This approach to predict drug disposition may serve as a useful tool to determine dosing regimens in patients with kidney disease requiring HD.

ACKNOWLEDGMENT

Financial Support. This study was funded by the Methodist Healthcare Foundation.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.