Principles of Drug Therapy in Patients With Reduced Kidney Function

Abstract

Impaired kidney function leads to altered pharmacokinetics of many drugs. This may result in clinically significant changes in systemic drug exposure and response. Individualized drug dosing regimens may be designed with a sound understanding of basic pharmacokinetic principles. This chapter describes the influence of impaired kidney function on pharmacokinetics and provides a practical approach to drug dosage individualization for patients with reduced kidney function, and those receiving continuous renal replacement therapy (CRRT), peritoneal dialysis, or hemodialysis.

Keywords

kidney disease, pharmacokinetics, CRRT, hemodialysis, peritoneal dialysis, drug dosing adjustment

Reduced kidney function may be observed in many settings, including patients with chronic kidney disease (CKD), the elderly with age-related decline in glomerular filtration rate (GFR), and critically ill patients with acute kidney injury (AKI). In adults, these conditions are associated with significant medication use, making these patients particularly vulnerable to the accumulation of a drug or its active or toxic metabolites. Clinicians must have a thorough understanding of the impact of reduced kidney function (RKF) on drug disposition and the appropriate methods by which to individualize drug therapy as they strive to optimize the outcomes of their patients.

Individualization of therapy for those agents that are predominantly (>70%) eliminated unchanged by the kidney can be accomplished with a proportional dose reduction or dosing interval prolongation based on the fractional reduction in GFR or its more commonly evaluated clinical counterparts, creatinine clearance (CL _Cr) and estimated GFR (eGFR). However, RKF is associated with progressive alterations in the bioavailability, plasma protein binding, distribution volume, and nonrenal clearance (CL _NR; i.e., metabolism and transport) of many drugs. Thus a more complex adjustment scheme may be required for medications that are extensively metabolized by the liver or for which changes in protein binding and/or distribution volume have been noted. Patients with RKF may also respond to a given dose or serum concentration of a drug (e.g., phenytoin) differently from those with normal kidney function because of the physiologic and biochemical changes associated with progressive CKD.

Using a sound understanding of basic pharmacokinetic principles, the pharmacokinetic characteristics of a drug, and the pathophysiologic alterations associated with RKF, clinicians can design individualized therapeutic regimens. This chapter describes the influence of RKF resulting from CKD and, when information is available, from AKI, on drug absorption, distribution, metabolism, transport, and excretion. The chapter also provides a practical approach to drug dosage individualization for patients with RKF and those receiving continuous renal replacement therapy (CRRT), peritoneal dialysis, or hemodialysis.

Drug Absorption

There is little quantitative information about the influence of RKF in CKD patients on drug absorption. Several variables, including changes in gastrointestinal transit time and gastric pH, edema of the gastrointestinal tract, vomiting and diarrhea (frequently seen in those with stage 5 CKD), and concomitant administration of phosphate binders, have been associated with alterations in the absorption of some drugs, such as digoxin and many of the fluoroquinolone antibiotics. The fraction of a drug that reaches the systemic circulation after oral versus intravenous administration (termed absolute bioavailability ) is rarely altered in CKD patients. However, alterations in the peak concentration (C _max) and in the time to which the peak concentration is attained (t _max) have been noted for a few drugs, suggesting that the rate, but not the extent of absorption, is altered. Although the bioavailability of some drugs, such as furosemide or pindolol, is reported as being reduced, there are no consistent findings in patients with CKD to indicate that absorption is actually impaired. However, an increase in bioavailability resulting from a decrease in metabolism during the drug’s first pass through the gastrointestinal tract and liver has been noted for some β-blockers and for dextropropoxyphene and dihydrocodeine.

Drug Distribution

The volume of distribution of many drugs is significantly altered in patients with stages 4 or 5 CKD ( Table 36.1 ), and changes in patients with oliguric AKI are also reported. These changes are predominantly the result of altered plasma protein or tissue binding or of volume expansion secondary to reduced kidney sodium and water excretion. The plasma protein binding of acidic drugs, such as warfarin and phenytoin, typically is decreased in patients with CKD because of decreased concentrations of albumin. Changes in the conformation of albumin binding sites and accumulation of endogenous inhibitors of binding may also contribute to decreased protein binding. In addition, the high concentrations of some drug metabolites that accumulate in CKD patients may interfere with the protein binding of the parent compound. Regardless of the mechanism, decreased protein binding increases the free or unbound fraction of the drug. On the other hand, the plasma concentration of the principal binding protein for several basic drug compounds, α ₁-acid glycoprotein, is increased in kidney transplant patients and in hemodialysis patients. For this reason, the unbound fraction of some basic drugs (e.g., quinidine) may be decreased, and as a result, the volume of distribution in these patients is decreased.

Table 36.1

Volume of Distribution of Selected Drugs in Patients With Stage 5 Chronic Kidney Disease

Drug	Normal (L/kg)	Stage 5 CKD (L/kg)	Change From Normal (%)
Amikacin	0.20	0.29	45
Azlocillin	0.21	0.28	33
Cefazolin	0.13	0.17	31
Cefoxitin	0.16	0.26	63
Cefuroxime	0.20	0.26	30
Clofibrate	0.14	0.24	71
Dicloxacillin	0.08	0.18	125
Digoxin	7.3	4.0	−45
Erythromycin	0.57	1.09	91
Gentamicin	0.20	0.32	60
Isoniazid	0.6	0.8	33
Minoxidil	2.6	4.9	88
Phenytoin	0.64	1.4	119
Trimethoprim	1.36	1.83	35
Vancomycin	0.64	0.85	33

CKD, Chronic kidney disease.

The net effect of changes in protein binding is usually an alteration in the relationship between unbound and total drug concentrations, an effect frequently encountered with phenytoin. The increase in the unbound fraction, to values as high as 20% to 25% from the normal of 10%, results in increased hepatic clearance and decreased total concentrations of phenytoin. Although the unbound concentration therapeutic range is unchanged, the therapeutic range for total phenytoin concentration is reduced to 4 to 10 µg/mL (normal, 10 to 20 µg/mL) as GFR falls. Therefore the maintenance of therapeutic unbound concentrations of 1 to 2 µg/mL provides the best target for individualizing phenytoin therapy in patients with RKF.

Altered tissue binding may also affect the apparent volume of distribution of a drug. For example, the distribution volume of digoxin is reported as being reduced by 30% to 50% in patients with severe CKD. This may be the result of competitive inhibition by endogenous or exogenous digoxin-like immunoreactive substances that bind to and inhibit membrane adenosine triphosphatase (ATPase). The absolute amount of digoxin bound to the tissue digoxin receptor is reduced, and the resultant serum digoxin concentration observed after administration of any dose is greater than expected.

Therefore, in CKD patients, a normal total drug concentration may be associated with either serious adverse effects secondary to elevated unbound drug concentrations or subtherapeutic responses because of an increased plasma-to-tissue drug concentration ratio. Monitoring of unbound drug concentrations is suggested for drugs that have a narrow therapeutic range, those that are highly protein bound (>80%), and those with marked variability in the bound fraction (e.g., phenytoin, disopyramide).

Drug Metabolism and Transport

CL _NRof drugs includes all routes of drug elimination excluding kidney excretion. Several metabolic enzymes and active transporters collectively constitute the primary pathways of CL _NR. Alterations in the function of and interactions between them can significantly affect the pharmacokinetic disposition and corresponding patient exposure to drugs that are substrates of nonrenal pathways. The effect of CKD on the expression or function of many of these pathways has been characterized in experimental models of kidney disease. For example, in rat models of end-stage kidney disease, hepatic expression of several cytochrome P450 (CYP) enzymes, including CYP3A1 and CYP3A2 (equivalent to human CYP3A4), is reduced by as much as 85%. CYP2C11 and CYP3A2 activity is also significantly reduced, but CYP1A1 activity is unchanged. CYP functional expression is also decreased in the intestine; CYP1A1 and CYP3A2 are reduced up to 40% and 70%, respectively.

Several hepatic reductase enzymes are also affected by kidney disease. Gene and protein expression of carbonyl reductase-1, aldo-keto reductase-3, and 11β-hydroxysteroid dehydrogenase-1 is decreased by as much as 93% and 76%, respectively, in CKD rats. Hepatic expression of the conjugative enzymes N -acetyltransferases (NAT) is also decreased, while uridine diphosphate-glucuronosyltransferases (UGT) are unchanged. Similarly, functional expression of several intestinal and hepatic transporters is altered in experimental models of kidney disease. The expression and corresponding activities of the efflux transporters P-glycoprotein (P-gp) and multidrug resistance-associated protein 2 (MRP2) are reduced by as much as 65% in the intestine, but the uptake transporter organic anion transporting polypeptide (OATP) is not affected. Conversely, in the liver, protein expression of P-gp, MRP2, and OATP is increased, unchanged, and decreased, respectively.

In humans with kidney disease, the activities of CYPs and reductases appear to be relatively unaffected. It was previously reported that CYP3A4 activity was reduced, but recent data indicate that OATP uptake activity is reduced, and thus the perceived changes in CYP3A4 activity were likely due to altered transporter activity, not an alteration in CYP activity. The reduction of CL _NRof several drugs that exhibit overlapping CYP and transporter substrate specificity in patients with stages 4 or 5 CKD supports this premise ( Table 36.2 ). These studies must be interpreted with caution, however, because concurrent drug intake, age, smoking status, and alcohol intake were often not taken into consideration. Furthermore, pharmacogenetic variations in drug-metabolizing enzymes and transporters that may have been present in the individual before the onset of AKI or progression of CKD must be considered, if known. For these reasons, prediction of the effect of RKF on the metabolism and/or transport of a particular drug is difficult, and a general quantitative strategy to adjust dosage regimens for drugs that undergo extensive CL _NRhas not yet been proposed. However, some qualitative insight may be gained if one knows which enzymes or transporters are involved in the clearance of the drug of interest and how those proteins are affected by a reduction in kidney function.

Table 36.2

Major Pathways of Nonrenal Drug Clearance and Selected Substrates

CL _NRPathway	Selected Substrates
*Oxidative Enzymes*
CYP
1A2	Polycyclic aromatic hydrocarbons, caffeine, imipramine, theophylline
2A6	Coumarin
2B6	Nicotine, bupropion
2C8	Retinoids, paclitaxel, repaglinide
2C9	Celecoxib, diclofenac, flurbiprofen, indomethacin, ibuprofen, losartan, phenytoin, tolbutamide, S -warfarin
2C19	Diazepam, S -mephenytoin, omeprazole
2D6	Codeine, debrisoquine, desipramine, dextromethorphan, fluoxetine, paroxetine, duloxetine, nortriptyline, haloperidol, metoprolol, propranolol
2E1	Ethanol, acetaminophen, chlorzoxazone, nitrosamines
3A4/5	Alprazolam, midazolam, cyclosporine, tacrolimus, nifedipine, felodipine, diltiazem, verapamil, fluconazole, ketoconazole, itraconazole, erythromycin, lovastatin, simvastatin, cisapride, terfenadine
*Reductase Enzymes*
11β-HSD	Bupropion, daunorubicin, prednisone, warfarin
CBR	Bupropion, daunorubicin, haloperidol, warfarin
AKR	Bupropion, daunorubicin, haloperidol, ketoprofen, nabumetone, naloxone, naltrexone, warfarin
*Conjugative Enzymes*
UGT	Acetaminophen, morphine, lorazepam, oxazepam, naproxen, ketoprofen, irinotecan, bilirubin
NAT	Dapsone, hydralazine, isoniazid, procainamide
*Transporters*
OATP
1A2	Bile salts, statins, fexofenadine, methotrexate, digoxin, levofloxacin
1B1	Bile salts, statins, fexofenadine repaglinide, valsartan, olmesartan, irinotecan, bosentan
1B3	Bile salts, statins, fexofenadine, telmisartan, valsartan, olmesartan, digoxin
2B1	Statins, fexofenadine, glyburide
P-gp	Digoxin, fexofenadine, loperamide, irinotecan, doxorubicin, vinblastine, paclitaxel, erythromycin
MRP
2	Methotrexate, etoposide, mitoxantrone, valsartan, olmesartan
3	Methotrexate, fexofenadine

AKR, Aldo-keto reductase; CYP, cytochrome P450 isozyme; CBR, carbonyl reductase; 11β-HSD, 11β-hydroxysteroid dehydrogenase; MRP, multidrug resistance-associated protein; NAT, N -acetyltransferase; OATP, organic anion-transporting polypeptide; P-gp, P-glycoprotein; UGT, uridine 5′-diphosphate glucuronosyltransferase.

The effect of CKD on the CL _NRof a particular drug is difficult to predict, even for drugs within the same pharmacologic class. The reductions in CL _NRfor CKD patients have frequently been noted to be proportional to the reductions in GFR. In the small number of studies that have evaluated CL _NRin critically ill patients with AKI, residual CL _NRwas higher than in CKD patients with similar levels of CL _Cr, whether measured or estimated by the Cockcroft-Gault equation. Because an AKI patient may have a higher CL _NRthan a CKD patient, the resultant plasma concentrations will be lower than expected and possibly subtherapeutic if classic CKD-derived dosage guidelines are followed.

Kidney Excretion of Drugs

Kidney clearance (CL _K) is the net result of glomerular filtration of unbound drug plus tubular secretion, minus tubular reabsorption. An acute or chronic reduction in GFR results in a decrease in CL _K. The degree of change in total body drug clearance is dependent on the fraction of the dose that is eliminated unchanged in individuals with normal kidney function, the intrarenal drug transport pathways, and the degree of functional impairment of each of these pathways. The primary kidney transport systems of clinical importance with respect to drug excretion include the organic anionic (OAT), organic cationic (OCT), P-gp, breast cancer resistance protein, and multidrug resistance-associated protein transporters. Diuretics, β-lactam antibiotics, nonsteroidal antiinflammatory drugs, and glucuronide drug metabolites are eliminated by the family of OAT transporters. The OCT transporters contribute to the secretion and excretion of cimetidine, famotidine, and quinidine. The P-gp transport system in the kidney is involved in the secretion of cationic and hydrophobic drugs (e.g., digoxin, Vinca alkaloids). The clearance of drugs that are secreted by the kidney (CL _K>300 mL/min) may be reduced from impairment in one or more of these kidney transporters.

Despite the different mechanisms involved in the elimination of drugs by the kidney and the availability of several methods for determining kidney function, the clinical estimation of CL _Crremains the most commonly used index for guiding drug dosage regimen design. The importance of an alteration in kidney function on drug elimination usually depends primarily on two variables: (1) the fraction of drug normally eliminated by the kidney unchanged and (2) the degree of GFR loss. There are a few drugs for which a metabolite is the primary active entity; in that situation, a key variable is the degree of CL _Kof the metabolite. The calculation of CL _Crfrom a timed urine collection has been the standard clinical measure of kidney function for decades. However, urine is difficult to collect accurately in most clinical settings, and the interference of many commonly used medications with creatinine measurement limits the utility of this approach. Use of radioactive markers ([ ¹²⁵I]iothalamate, ⁵¹Cr-EDTA, or ^99mTc-DTPA) or nonradioactive GFR markers (iohexol, iothalamate, and inulin), although scientifically sound, is clinically impractical, because intravenous or subcutaneous marker administration and multiple timed blood and urine collections make the procedures expensive and cumbersome.

Estimation of Kidney Function for Drug Dosing Purposes

The estimation of kidney function by various estimating equations for drug dosing purposes is a critically important issue. In contrast to measured approaches, estimation of CL _Cror GFR requires only routinely collected laboratory and demographic data. The Cockcroft-Gault equation for CL _Crand the Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) equations for eGFR correlate well with CL _Crand GFR measurements in individuals with stable kidney function and average body composition (see Chapter 3 ). The traditional approach of estimating CL _Crand using it as a continuous variable of kidney function for drug dosing adjustment is now being supplemented and, in some institutions, replaced by eGFR. Caution is warranted since the use of eGFR as a guide for drug dosage adjustment, has not been systematically validated. Currently, there are limited prospective pharmacokinetic data and corresponding dosing recommendations based on GFR estimating equations. Because nearly all of the primary published literature to date has used CL _Crto derive the relationship between kidney function and kidney and/or total body clearance of a drug, CL _Cris still the standard metric for drug dosing purposes. Nevertheless, widespread availability of automatically reported eGFR affords clinicians a tool that, if validated for drug dosing, could easily be incorporated into clinical practice. Furthermore, use of eGFR for management of kidney disease and drug dosing, and harmonization of practice in this regard between physicians, pharmacists, and other clinicians, would be ideal and warrants further evaluation.

Several issues should be considered by clinicians when assessing CL _Crand eGFR data for drug dosing. First, the automatically reported eGFR value provides an estimate that is normalized for body surface area (BSA) in units of mL/min per 1.73 m ². When used for drug dosing, the eGFR value should be individualized (i.e., not normalized for BSA) and converted to units of mL/min, particularly in patients whose BSA is considerably larger or smaller than 1.73 m ². The individualized value should be compared with CL _Crestimates (mL/min). Second, when presented with various kidney function estimates that potentially translate into different drug dosing regimens, clinicians should choose the regimen that optimizes the risk-benefit ratio, given the patient-specific clinical scenario. For drugs with a narrow therapeutic range, typically more conservative kidney function estimates and corresponding doses should be used, particularly if therapeutic drug monitoring is not readily available. Because CL _Crestimates are more conservative and indicate the need for dose adjustment more often than eGFR, they may be preferred when dosing narrow therapeutic window drugs, especially in high-risk subgroups such as the elderly. The use of eGFR and a more aggressive dosing strategy may be acceptable for drugs with a wide therapeutic range and a broader margin of safety. Third, when estimating equations are not expected to provide accurate measures of kidney function (i.e., because of altered creatinine generation or unstable serum creatinine concentrations) and therapeutic drug monitoring is not available, it may be reasonable to obtain an accurately timed urine collection to calculate CL _Cr, particularly for narrow therapeutic window drugs with high toxicity. Fourth, the limitations and the study population of the original trials from which the eGFR equations were developed, and subsequent populations in which they have been validated, must be considered before applying them to a specific patient. All of these methods are poor predictors of kidney function in individuals with liver disease, and their use is not recommended for such patients. Finally, although several methods for CL _Crestimation in patients with unstable kidney function (e.g., AKI) have been proposed, the accuracy of these methods has not been rigorously assessed, and at the present time their use cannot be recommended.

Strategies for Drug Therapy Individualization

Design of the optimal dosage regimen for a patient with RKF depends on the availability of an accurate characterization of the relationship between the pharmacokinetic drug parameters and kidney function. Before 1998, there was no consensus regarding the criteria for characterization of the pharmacokinetics of a drug in CKD patients. An industry guidance report issued by the US Food and Drug Administration (FDA) in May 1998 provided guidelines regarding when a study should be considered, provided recommendations for study design, data analysis, and assessment of the impact of the study results on drug dosing, and recommended use of dose adjustment categories derived from CL _Cr. Currently the FDA is considering including dosing tables based on eGFR and CL _Crin a revised version of the 1998 FDA guideline. In the future, drug dosing recommendations based on eGFR in addition to CL _CRmay be included in FDA-approved drug dosing labels. However, for drugs already approved by the FDA with existing dose adjustment recommendations based on CL _CR, it is unlikely drug manufacturers will provide additional eGFR-based dosing recommendations.

Most dosage adjustment reference sources for clinical use have proposed the use of a fixed dose or interval for patients with a broad range of kidney function. Indeed, “normal” kidney function has often been ascribed to anyone who has a CL _Crgreater than 50 mL/min, even though many individuals (e.g., hyperfiltering early diabetics) have values in the range of 120 to 180 mL/min. The “moderate kidney function impairment” category in many guides encompasses a fivefold range of CL _Cr, from 10 to 49 mL/min, whereas severe kidney function impairment or end-stage kidney disease is defined as a CL _Crof less than 10 to 15 mL/min. Each of these categories encompasses a broad range of kidney function, and the calculated drug regimen may not be optimal for all patients within that range.

If specific literature recommendations or data on the relationship of the pharmacokinetic parameters of a drug to CL _Crare not available, then these parameters can be estimated for a particular patient with the method of Rowland and Tozer, provided that the fraction of the drug that is eliminated unchanged by the kidney (f _e) in normal subjects is known. This approach assumes that the change in drug clearance is proportional to the change in CL _Cr, kidney disease does not alter the drug’s CL _NR, any metabolites produced are inactive and nontoxic, the drug obeys first-order (linear) kinetic principles, and it is adequately described by a one-compartment model. If these assumptions are true, the kinetic parameter or dosage adjustment factor (Q) can be calculated as follows:

Q=1−[fe(1−KF)]Q=1−[fe(1−KF)]
Q = 1 − [ f e ( 1 − KF ) ]

where KF is the ratio of the patient’s CL _Crto the assumed normal value of 120 mL/min. As an example, the Q factor for a patient who has a CL _Crof 10 mL/min and a drug that is 85% eliminated unchanged by the kidney would be

Q=1−[0.85(1−10/120)]Q=1−[0.85(0.92)]Q=1−0.78Q=0.22Q=1−[0.85(1−10/120)]Q=1−[0.85(0.92)]Q=1−0.78Q=0.22
Q = 1 − [ 0.85 ( 1 − 10 / 120 ) ] Q = 1 − [ 0.85 ( 0.92 ) ] Q = 1 − 0.78 Q = 0.22

The estimated clearance rate of the drug in this patient (CL _PT) would then be calculated as

CLPT=CLnorm×QCLPT=CLnorm×Q
CL PT = CL norm × Q

where CL _normis the respective value in patients with normal kidney function derived from the literature.

For antihypertensive agents, cephalosporins, and many other drugs for which there are no target values for peak or trough concentrations, attainment of an average steady-state concentration similar to that in normal subjects is appropriate. The principal means to achieve this goal is to decrease the dose or prolong the dosing interval. If the dose is reduced and the dosing interval is unchanged, the desired average steady-state concentration will be near normal; however, the peak will be lower and the trough higher. Alternatively, if the dosing interval is increased and the dose remains unchanged, the peak, trough, and average concentrations will be similar to those in the patients with normal kidney function. This interval adjustment method is often preferred because it is likely to yield significant cost savings due to less frequent drug administration. If a loading dose is not administered, it will take approximately five half-lives for the desired steady-state plasma concentrations to be achieved in any patient; this may require days rather than hours, because of the prolonged half-life of many drugs in patients with RKF. Therefore, to achieve the desired concentration rapidly, a loading dose (D _L) should be administered for most patients with RKF. D _Lcan be calculated as follows:

DL=(Cpeak)×(VD)×(Body weight in kilograms)DL=(Cpeak)×(VD)×(Body weight in kilograms)
D L = ( C peak ) × ( V D ) × ( Body weight in kilograms )

The loading dose is usually the same for patients with RKF as it is for those with normal kidney function. However, if the V _Din patients with RKF is significantly different from the V _Din patients with normal kidney function (see Table 36.1 ), then the modified value should be used to calculate the D _L.

The adjusted dosing interval (τ _RKF) or maintenance dose (D _RKF) for the patient can then be calculated from the normal dosing interval (τ _n) and normal dose (D _n), respectively:

τRKF=τn/QτRKF=τn/Q
τ RKF = τ n / Q

DRKF=Dn×QDRKF=Dn×Q
D RKF = D n × Q

If these approaches yield a time interval or a dose that is impractical, a new dose can be calculated with a fixed, prespecified dose interval (τ _FPDI) such as 24 or 48 hours, as follows: