Volume 65, Issue 1 p. 98-109
Free Access

Interaction between midazolam and clarithromycin in the elderly

Sara K. Quinney

Sara K. Quinney

Division of Clinical Pharmacology, Indiana University School of Medicine, Wishard Memorial Hospital, W7123 Myers Building, Indianapolis, IN, USA

Search for more papers by this author
Barbara D. Haehner

Barbara D. Haehner

Division of Clinical Pharmacology, Indiana University School of Medicine, Wishard Memorial Hospital, W7123 Myers Building, Indianapolis, IN, USA

Search for more papers by this author
Melissa B. Rhoades

Melissa B. Rhoades

Division of Clinical Pharmacology, Indiana University School of Medicine, Wishard Memorial Hospital, W7123 Myers Building, Indianapolis, IN, USA

Search for more papers by this author
Zhen Lin

Zhen Lin

Division of Clinical Pharmacology, Indiana University School of Medicine, Wishard Memorial Hospital, W7123 Myers Building, Indianapolis, IN, USA

Search for more papers by this author
J. Christopher Gorski

J. Christopher Gorski

Division of Clinical Pharmacology, Indiana University School of Medicine, Wishard Memorial Hospital, W7123 Myers Building, Indianapolis, IN, USA

Search for more papers by this author
Stephen D. Hall

Stephen D. Hall

Division of Clinical Pharmacology, Indiana University School of Medicine, Wishard Memorial Hospital, W7123 Myers Building, Indianapolis, IN, USA

Search for more papers by this author
First published: 22 October 2007
Citations: 38
Stephen D. Hall, Division of Clinical Pharmacology, Indiana University School of Medicine, Wishard Memorial Hospital, W7123 Myers Building, 1001 West 10th Street, Indianapolis, IN 46202-2879, USA.
Tel: + 1 317 630 8795
Fax: + 1 317 630 8185
E-mail:[email protected]

J.C.G. current address: Mylan Pharmaceuticals, Inc., 781 Chestnut Ridge Road, PO Box 4310, Morgantown, WV 26504-4310, USA.



• Clarithromycin is a mechanism-based inhibitor of CYP3A that has been shown to inhibit midazolam hydroxylation in young, healthy subjects.

• Elderly persons exhibit altered metabolism of a variety of drugs, including clarithromycin.

• However, the consequences of increased exposure to an inhibitor drug on CYP3A activity have not been addressed.


• This study utilized intravenous and oral midazolam as in vivo probes to examine the effect of clarithromycin on intestinal and hepatic CYP3A activity.

• Although clarithromycin concentrations are greater in elderly individuals than in young, healthy volunteers, intestinal and hepatic CYP3A enzymes are inhibited to a similar extent as in the young.


To assess the relative contribution of intestinal and hepatic CYP3A inhibition to the interaction between the prototypic CYP3A substrate midazolam and clarithromycin in the elderly.


On day 1, 16 volunteers (eight male, eight female) aged 65–75 years weighing 59–112 kg received simultaneous doses of midazolam intravenously (i.v.) (0.05 mg kg−1 over 30 min) and orally (p.o.) (3.5 mg of a stable isotope, 15N3-midazolam). Starting on day 2, clarithromycin 500 mg was administered orally twice daily for 7 days. On day eight, i.v. and p.o. doses of midazolam were administered 2 h after the final clarithromycin dose. Serum and urine samples were assayed for midazolam, 15N3-midazolam and metabolites by gas chromatography/mass spectometry.


Men and women exhibited similar i.v. (30.4 vs. 36.0 l h−1) and p.o. (119 vs. 124 l h−1) clearances of midazolam. Midazolam hepatic availability was significantly (P = 0.006) greater in men [0.79, 95% confidence interval (CI) 0.75, 0.84] than in women (0.66, 95% CI 0.59, 0.73), but midazolam intestinal availability (0.39 vs. 0.55) was not different. Following clarithromycin dosing, a significant decrease in systemic (33.2 l h−1 to 11.5 l h−1) and oral (121 l h−1 to 17.4 l h−1) midazolam clearance occurred. Oral, hepatic and intestinal availability was significantly increased after clarithromycin dosing from 0.34 to 0.72, 0.73 to 0.91 and 0.47 to 0.79, respectively. Clarithromycin administration led to an increase in the AUC of midazolam by 3.2-fold following i.v. dosing and 8.0-fold following p.o. dosing. Similar effects were observed for males and females.


Intestinal and hepatic CYP3A inhibition by clarithromycin significantly reduces the clearance of midazolam in the elderly.

The cytochrome P450 3A enzymes (CYP3A4 and CYP3A5) participate in the biotransformation of approximately half of the drugs that undergo metabolic clearance in human adults. CYP3A4 is highly expressed in the liver and the epithelium of the small intestine and represents on average 30% of total hepatic CYP protein [1] and 33–87% of intestinal CYP [2]. CYP3A5 exhibits a polymorphic expression pattern with high expression in 10–30% of Whites [3, 4] and is clinically important in the disposition of some CYP3A substrates such as tacrolimus and sirolimus [5, 6]. CYP3A enzymes metabolize a broad array of structurally diverse compounds, including macrolide antibiotics, benzodiazepines, 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase inhibitors, calcium channel blockers and human immunodeficiency virus protease inhibitors, and consequently are at the centre of many clinically significant drug interactions. Clarithromycin is a widely used macrolide antibiotic that is a potent inhibitor of CYP3A4 in vitro and in vivo and appears to exert inhibition by an irreversible mechanism. We have previously shown that clarithromycin irreversibly inhibits midazolam hydroxylation in intestinal biopsy tissue obtained from pretreated, young, healthy subjects [7]. An important consequence of this irreversible mechanism is that the inhibition remains for up to 10 days after discontinuation of the inhibitor; this time delay is determined by the elimination half-life of CYP3A4 [8]. We have used the simultaneous intravenous (i.v.)/oral (p.o.) midazolam dosing paradigm to demonstrate that clarithromycin exerts a sex-dependent inhibition of intestinal and hepatic CYP3A in young, healthy subjects [9]. Clinically important consequences of CYP3A inhibition by clarithromycin include excessive sedation following benzodiazepine administration, increased concentrations of ciclosporin A [10–12] and increased risk of nifedipine-induced vasodilatory shock [13]. However the extent of inhibition of CYP3A at hepatic and intestinal sites by potent inhibitors has not previously been examined in the elderly.

Elderly patients are more likely to be taking multiple medications and there is a corresponding increase in the likelihood of drug–drug interactions. It has been estimated that >75% of drug–drug interactions occur in persons >50 years old [14]. Although expression of CYP3A proteins does not appear to change with age [15, 16], elderly persons may exhibit altered clearance of CYP3A substrates for a variety of reasons, including disease and altered tissue perfusion and structure. The disposition of CYP3A inhibitors may also be altered in the elderly particularly when renal excretion is an important route of elimination. The renal clearance and total clearance of clarithromycin are significantly decreased in elderly persons with reduced creatinine clearance [17]. However, the drug interaction consequences of the resulting increased exposure to inhibitor have not been previously addressed. Thus, this study was designed to examine the contribution of intestinal and hepatic CYP3A inhibition to the interaction between midazolam and clarithromycin in the elderly, and the effect of sex on the pharmacokinetics of midazolam and extent of interaction were evaluated.

Materials and methods


This study was approved by the Indiana University Purdue University Indianapolis and Clarian Health Partners institutional review board, and written informed consent was obtained from study volunteers prior to drug dosing. Healthy volunteers >65 years old with no significant medical history as assessed by physical examination and blood and urine chemistry screens were enrolled. Volunteers were excluded if they were taking medications known to influence CYP3A activity. Subjects with known allergies to either benzodiazepines or macrolide antibiotics were also excluded.

Study design

A fixed order study design was employed, because the duration of inhibition is unknown in this group and we wished to minimize interday variability in intestinal and hepatic CYP3A content which may occur during extended wash-out periods. Volunteers abstained from alcohol, grapefruit and grapefruit products, other citrus products and herbal supplements for at least 1 week prior to and during the study. However, participants were allowed to continue taking stable doses of medications which have not been previously identified as inducers, inhibitors or substrates of CYP3A4. Abstinence from alcohol and compliance with other dietary restrictions were assessed through volunteer questioning. All medications for the treatment of chronic conditions were withheld on the days of midazolam dosing. After an overnight fast, i.v. catheters were placed in one forearm for the withdrawal of blood samples and in the opposite forearm for the administration of drug. Prior to receiving the dose of midazolam, subjects emptied their bladder and a baseline blood sample was obtained. Subjects then simultaneously received midazolam (0.05 mg kg−1) intravenously over 30 min and 15N3-midazolam (3.0–4.0 mg) p.o. as a solution. Blood samples were obtained 5, 15, 30, 45 min and 1, 1.5, 2, 2.5, 3, 4, 5, 6, 8, 10, 12, and 24 h following drug administration. Serum was obtained and frozen at −20°C until analysis. Urine was collected over the following time intervals: 0–12 and 12–24 h post dose and frozen at −20°C until analysis. After the 24-h blood draw, p.o. clarithromycin (500 mg) was initiated twice daily for 7 days. On day 8, 2 h after the morning clarithromycin dose, the midazolam portion of the study was repeated. Compliance with clarithromycin was confirmed by pill count and quantification of clarithromycin serum concentrations.

Sample analysis

Serum samples were processed using a liquid–liquid extraction technique and quantified following derivatization with N-methyl-N-t-butyldimethylsilyl trifluoroacetamide containing 1% t-butyldimethylchlorosilane using gas chromatography/mass specrometry (GC/MS) (Hewlett Packard 5971 mass selective detector and 5890A gas chromatograph) as described previously [9]. The limit of quantification for midazolam and metabolite was 1 ng ml−1. The interday coefficient of variation at 5 and 50 ng ml−1 was 11% and 12%, and 4.6% and 5.2% for midazolam and 15N3-midazolam, respectively. Urine samples were processed as described above following deconjugation with β-glucuronidase (Sigma Chemical Co., St Louis, MO, USA). The midazolam concentration in the i.v. infusion solution was determined by high-performance liquid chromatography (HPLC) [18]. Additionally, serum samples were processed through a liquid–liquid extraction method and clarithromycin serum concentrations were estimated using HPLC with electrochemical detection as previously described [9, 17]. The limit of quantification for clarithromycin was 2.5 ng ml−1. The corresponding coefficients of variation and relative error for clarithromycin at 10 ng ml−1 were <8% and 9%, respectively.

Pharmacokinetic analysis

Standard model independent methods were used to determine the pharmacokinetic parameters of interest (WinNONLIN, v4.0; PharSight Inc., Mountain View, CA, USA). The terminal elimination rate constant (β) was determined using linear regression. The elimination half-life was determined as t1/2 = 0.693/β. The maximum concentration and time to reach the maximum concentration were determined by visual inspection of the data. The area under the concentration–time curve (AUC from zero to infinity) after i.v. and p.o. drug administration was determined using a combination of linear and logarithmic trapezoidal methods with extrapolation to infinity (Clast/β). The volume of distribution (Vdβ) was estimated by dividing the systemic blood clearance by β.

To determine the blood concentration (Cblood) to serum concentration (Cserum) ratio, blood from a volunteer was spiked with midazolam (10–2500 ng ml−1) and the serum concentrations were determined using GC/MS as described. The red blood cell (RBC) affinity, ρ, for midazolam was determined using Equation 1:


The haematocrit (Hct) for each subject was determined {mean, women 0.40 [95% confidence interval (CI) 0.39, 0.41]vs. men 0.44 (95% CI 0.42, 0.46); P = 0.0025} and the fraction unbound in serum, fu, of midazolam was assumed to be 0.024 [19, 20]. The relationship between RBC affinity, ρ, and total serum midazolam concentration was nonlinear and was completely described by the following empirical relationship [9] (Equation 2):


where ρi is equal to the RBC midazolam affinity at a given serum concentration (Ctotal). Following simultaneous p.o. and i.v. dosing, Ctotal reflects the sum of the labelled and unlabelled midazolam concentrations. The RBC affinity was estimated for each individual at each sample time. The blood to serum midazolam concentration ratio was then estimated using a rearrangement of Equation 1. Blood concentrations were calculated by multiplying the midazolam serum concentration following i.v. or p.o. administration by the blood to plasma partition ratio. Total blood clearance, CLiv,mdz, after an i.v. dose of midazolam was determined using Equation 3:


where the i.v. dose of midazolam was estimated as the infused solution concentration multiplied by the rate of infusion and the duration of infusion. The oral clearance of 15N3-midazolam, inline image, was determined using Equation 4:


The corresponding oral availability, Fpo, of midazolam was estimated using Equation 5:


The hepatic availability of midazolam following i.v. administration was determined using Equation 6 assuming that the liver was the sole organ of midazolam elimination:


Hepatic blood flow, QH, was estimated in men and women as 25.3 ml kg−1 or 25.5 ml kg−1 times body weight (kg), respectively [21]. The availability across the gut wall, Fgut, was estimated using Equation 7[22]:


where FABS represents the fraction of the dose absorbed into the intestinal wall. In light of the data obtained during the investigation, FABS is assumed to be unity (Table 2).

Table 2.
Pharmacokinetic parameters of midazolam following intravenous and oral administration in 16 healthy elderly volunteers before and after clarithromycin (500 mg bid ×7 days) administration. Presented as mean (95% confidence interval)
N Control
Female, 8 Male, 8 P-value, male vs. female All 16
C max,iv (µg l−1) 58 (48, 68) 59 (53, 64) 0.88 58 (53, 64)
AUCiv (µg l−1 h−1) 118 (96, 140) 142 (112, 171) 0.23 130 (111, 149)
CLiv (l h−1) 36.0 (29.5, 42.8) 30.4 (24.5, 36.3) 0.22 33.2 (28.7, 37.8)
t 1/2iv (h) 2.5 (2.0, 2.9) 4.0 (3.0, 5.0) 0.018 3.2 (2.6, 3.9)
Vdβ (l) 126 (100, 152) 162 (133, 192) 0.09 144 (123, 165)
t max,po (h) 0.88 (0.25, 2) 0.5 (0.5, 3) 0.88 0.75 (0.25, 3)
C max,po (µg l−1) 12 (7.2, 17) 12 (8.0, 15) 0.86 12 (9.0, 15)
AUCpo (µg l−1 h−1) 46 (27, 65) 48 (21, 75) 0.90 47 (31, 63)
CLpo (l h−1) 124 (71, 177) 119 (76, 161) 0.88 121 (88, 154)
Percent of dose recovered in the urine as 1'-hydroxymidazolam
Intravenous dose (%) 73 (63, 83) 63 (46, 80) 0.35 68 (58, 78)
Oral dose (%)* 70 (67, 74) 69 (56, 83) 0.89 70 (63, 77)
P.o./i.v. dose (%)§ 98 (85, 113) 115 (92, 145) 0.26 106 (93, 123)
N Clarithromycin treatment
Female, 8 Male, 8 P-value, male vs. female All 16 P-value, control vs. treatment
C max,iv (µg l−1) 70 (54, 85) 66 (50, 82) 0.75 68 (57, 79) 0.13
AUCiv (µg l−1 h−1) 409 (311, 507) 397 (291, 503) 0.87 403 (333, 473) <0.00001
CLiv (l h−1) 10.6 (8.6, 12.6) 12.4 (7.0, 17.9) 0.55 11.5 (8.7, 14.4) <0.00001
t 1/2iv (h) 13.5 (9.1, 17.9) 12.5 (8.5, 16.5) 0.74 13.0 (10.1, 15.9) <0.00001
Vdβ (l) 189 (153, 224) 180 (144, 215) 0.73 184 (160, 208) 0.011
t max,po (h) 0.63 (0.5, 1.5) 0.5 (0.5, 1.5) 0.88 0.5 (0.5, 1.5) 0.29
C max,po (µg l−1) 34 (28, 40) 31 (27, 35) 0.40 33 (29, 36) <0.00001
AUCpo (µg l−1 h−1) 304 (183, 426) 290 (205, 375) 0.86 297 (226, 369) <0.00001
CLpo (l h−1) 16.2 (11.6, 20.7) 18.6 (9.2, 28.0) 0.65 17.4 (12.3, 22.5) <0.00001
Percent of dose recovered in the urine as 1'-hydroxymidazolam
Intravenous dose (%) 31 (25, 37) 38 (24, 53) 0.33 34 (27, 42) 0.00001
Oral dose (%)* 38 (30, 45) 48 (41, 56) 0.07 43 (37, 49) <0.00001
P.o./i.v. dose (%)§ 124 (98, 156) 141 (115, 189) 0.51 132 (115, 151) 0.04
Extent of interaction
AUCiv ratio 3.6 (2.8, 4.4) 2.9 (2.0, 3.7) 0.22 3.2 (2.7, 3.8)
AUCpo ratio 8.2 (4.8, 11.6) 7.7 (4.5, 11) 0.85 8.0 (5.7, 10.2)
  • * Oral dose recovered as 1'-hydroxy- 15 N 3 -midazolam.
  • Normalized to a dose of 4 mg.
  • Median (range).
  • § Logarthmic average (95% CI) of individual oral to intravenous 1'-hydroxymidazolam recovery ratio.

Statistical analysis

The sample size of this study enabled detection of a 50% difference in midazolam AUC between control and clarithromycin treatment with 80% power. The effect of clarithromycin on the pharmacokinetic variables of midazolam (e.g. clearance, availability) was analysed using descriptive statistics, correlation analysis, linear regression analysis and Student's paired t-test. The effects of sex on the pharmacokinetic variables of midazolam were analysed using Student's t-test performed by the JMP computer program (v5.0.1; SAS Institute, Cary, NC, USA) at P < 0.05 level of significance.


Eight White male (ages 68–75 years) and eight White female (ages 66–80 years) healthy, nonsmoking volunteers participated in the study (Table 1). Seven men and four women did not take any medications. One male volunteer received spironolactone and enalapril for the treatment of hypertension. Two women were taking levothyroxine (0.05 or 0.1 mg once daily) for the treatment of hypothyroidism and one woman was receiving methimazole 10 mg once daily for the treatment of hyperthyroidism. One woman was prescribed hormone replacement therapy, consisting of medroxyprogesterone and conjugated oestrogens, which has been previously shown not to alter in vivo CYP3A activity [23]. Physical examination and blood and urine chemistry screening did not identify any additional significant medical conditions in these individuals. Volunteers were monitored for sedation following administration of midazolam. All volunteers completed the study and no unexpected adverse events occurred.

Table 1.
Demographic and clinical characteristics of individuals studied
N Female, 8 Male, 8 P-value, male vs. female All 16
Age (year)* 70 (66–80) 72 (68–75) 0.52 71 (66–80)
Weight (kg)* 71 (56–91) 97 (75–112) 0.002 84 (56–112)
Serum creatinine (mg dl−1) 0.8 (0.8, 0.9) 1.0 (0.9, 1.1) 0.0017 0.9 (0.9, 1.0)
Creatinine clearance (ml min−1) 50 (44, 57) 67 (60, 73) 0.0044 58 (52, 65)
  • * Mean (range).
  • Mean (95% confidence interval).
  • Creatinine clearance calculated using the Cockcroft and Gault equation.

The urinary recovery of the oral 15N3-midazolam dose as 1'-hydroxy-15N3-midazolam averaged 70% (95% CI 63, 77). This value was not significantly (P = 0.76) different from the urinary recovery of 1'-hydroxymidazolam following i.v. midazolam administration (68%, 95% CI 58, 78) and is consistent with previous reports [9, 23, 24]. The extent of urinary 1'-hydroxy-15N3-midazolam recovery following oral 15N3-midazolam dosing indicates that the oral dose was completely absorbed into the gastrointestinal epithelium (Table 2). The urinary recovery of 1'-hydroxymidazolam was not significantly different between elderly men and women following i.v. (P = 0.35) or p.o. (P = 0.89) doses of midazolam (Table 2).

The i.v. dose of midazolam was significantly (P = 0.029) greater in men (4.3 mg, 95% CI 3.6, 5.0) than in women (3.2 mg, 95% CI 2.6, 3.8) because of a significant (P = 0.0024) difference in body weight between men (96.4 kg, 95% CI 75, 112) and women (70.3 kg, 95% CI 56, 91). Therefore, concentrations were normalized to a midazolam dose of 4 mg for all individuals. The Cmax following i.v. dosing was equivalent in men and women (58 µg l−1, 95% CI 53, 64). A sex difference was not observed in the systemic clearance of midazolam [30.4 l h−1 (95% CI 24.5, 36.3) in men vs. 36.0 l h−1 (95% CI 29.5, 42.8) in women; P = 0.14], or the V (162 l[95% CI 133–192 l]vs. 126 l[95% CI 100–152 l], p = 0.09). However, the half-life of midazolam following intravenous dosing was significantly (P = 0.018) longer in male (4.0 h, 95% CI 3.0–5.0 h) than in female volunteers (2.5 h, 95% CI 2–2.9 h) because men exhibited a slightly greater Vdβ and smaller clearance than women (Table 2).

In the absence of clarithromycin, a significant (P = 0.006) sex difference in hepatic availability (FH) was observed, but oral bioavailability (Fpo) and intestinal availability (FG) were not different (Table 3). As expected from our previous report, in the absence of clarithromycin, a significant correlation (r = 0.94) was observed between oral bioavailability and intestinal availability (Figure 1A), but not between hepatic and oral availability (Figure 1B). In addition, hepatic and intestinal availability was poorly correlated (Figure 1C), suggesting that the factors controlling CYP3A activity in these organs are different.

Table 3.
Mean (95% confidence interval) midazolam oral availability, hepatic extraction, hepatic availability, and intestinal availability in 16 elderly volunteers before and after clarithromycin (500 mg bid ×7 days) administration
N Control
Female, 8 Male, 8 P-value, male vs. female All 16
F po 0.37 (0.26, 0.48) 0.31 (0.21, 0.41) 0.47 0.34 (0.27, 0.42)
E H 0.34 (0.27, 0.41) 0.21 (0.16, 0.25) 0.006 0.27 (0.22, 0.33)
F H 0.66 (0.59, 0.73) 0.79 (0.75, 0.84) 0.006 0.73 (0.67, 0.78)
F G 0.55 (0.42, 0.69) 0.39 (0.28, 0.50) 0.098 0.47 (0.38, 0.57)
N Clarithromycin treatment
Female, 8 Male, 8 P-value, male vs. female All 16 P-value, control vs. treatment
F po 0.72 (0.59, 0.84) 0.72 (0.63, 0.81) 0.97 0.72 (0.64, 0.79) <0.00001
E H 0.10 (0.08, 0.13) 0.09 (0.05, 0.12) 0.46 0.09 (0.07, 0.12) <0.00001
F H 0.90 (0.87, 0.92) 0.91 (0.88, 0.95) 0.46 0.91 (0.88, 0.93) <0.00001
F G 0.80 (0.66, 0.94) 0.79 (0.69, 0.88) 0.89 0.79 (0.71, 0.87) 0.00002
Details are in the caption following the image

Relationship between the oral and intestinal availability of midazolam prior to (closed symbols, N = 16) and after 7 days of clarithromycin (500 mg bid, open symbols, N = 16) (A). The relationship between hepatic and intestinal availability of midazolam before and after 7 days of clarithromycin (500 mg bid) (B) and the relationship between hepatic and oral availability of midazolam before and after 7 days of clarithromycin (500 mg bid) (C). Circles and squares represent values from women and men, respectively. The correlation coefficient (r) is provided for each comparison

Complete compliance with clarithromycin dosing was noted based on the day 7 pill counts and by measuring the clarithromycin serum concentrations in each volunteer (Figure 2). All volunteers had measurable clarithromycin concentrations throughout the dosing interval. The clearance of clarithromycin following oral dosing varied 7.6-fold from 2.5 l h−1 to 19 l h−1 with a mean of 10.9 l h−1 (95% CI 8.5, 13.2, Table 4). The mean steady-state concentration of clarithromycin (inline image) was 3.2 µm (95% CI 2.8, 3.0). No sex difference in steady-state concentration, oral clearance or elimination half-life of clarithromycin was observed (Table 4). No significant differences between men and women were noted in the inline image of the 14-hydroxy- and N-desmethylclarithromycin metabolites (Table 4).

Details are in the caption following the image

The mean (±SD) blood concentration vs. time curve of clarithromycin, 14-hydroxyclarithromycin and N-desmethylclarithromcyin following oral administration of 500 mg twice daily for 7 days in 16 healthy volunteers. Mean (±SD) clarithromycin serum concentrations, (○); mean (±SD) 14-hydroxyclarithromycin concentration, (□); mean (±SD) N-desmethylclarithromycin concentrations on day 7 of clarithromycin dosing, (Δ)

Table 4.
Mean (95% confidence interval) pharmacokinetic parameters of clarithromycin and metabolites following oral administration of 500 mg clarithromycin for 7 days in 16 healthy elderly volunteers
Male (n = 8) Female (n = 8) P-value, male vs. female All (n = 16)
 AUC0−12m h−1) 38.9 (32.3, 45.5) 38.5 (28.6, 48.4) 0.95 38.7 (33.0, 44.4)
 CLpo (l h−1) 10.5 (7.1, 13.9) 11.2 (7.7, 14.7) 0.79 10.9 (8.5, 13.2)
t1/2 (h) 10.7 (6.7, 14.7) 11.6 (5.8, 17.4) 0.80 11.2 (7.8, 14.6)
14-hydroxyclarithromycin AUC0−12m h−1) 5.2 (4.4, 6.0) 6.27 (4.9, 7.7) 0.21 5.7 (4.9, 6.6)
N-desmethylclarithromycin AUC0−12m h−1) 5.1 (3.4, 6.7) 5.9 (3.3, 8.6) 0.58 5.5 (4.0, 7.0)

Following treatment with clarithromycin for 1 week, systemic exposure to midazolam was increased for both i.v. and p.o. dosing (Figure 3). The mean systemic clearance of midazolam decreased from 33.2 l h−1 (95% CI 28.7, 37.8) to 11.5 l h−1 (95% CI 8.7, 14.4) in the presence of clarithromycin (P < 0.00001, Table 2). The volume of distribution (Vdβ) of midazolam was not significantly altered by clarithromycin and consequently the elimination half-life was significantly prolonged (Table 2). The oral clearance of midazolam (121 l h−1, 95% CI 88, 154) was significantly reduced by clarithromycin (17.4 l h−1, 95% CI 12.3, 22.5). Sex differences in the pharmacokinetic parameters of midazolam disposition were not observed following clarithromycin dosing.

Details are in the caption following the image

The mean (± SD) blood concentration vs. time curve of midazolam and 15N3-midazolam after simultaneous intravenous and oral (n = 16) administration. Mean blood concentrations on day 1, (●); mean blood concentrations following the oral administration of clarithromycin (500 mg bid) for 7 days, (○)

Following treatment with clarithromycin, oral, intestinal and hepatic availabilities of midazolam were significantly increased (P < 0.00001, Table 3). The relative fold-increase in midazolam hepatic or intestinal availability was significantly correlated with the hepatic (r = 0.93) or intestinal (r = 0.82) availability of midazolam in the absence of clarithromycin (Figure 4). A significant correlation was observed between the oral and intestinal availability of midazolam before and after clarithromycin treatment (Figure 1A). However, no significant correlation was observed between the oral and hepatic availabilities, or the intestinal and hepatic availabilities of midazolam in the presence of clarithromycin (Figure 1B,C).

Details are in the caption following the image

Relationship between the fold change in hepatic or intestinal availability and initial intestinal or hepatic availability. The correlation coefficient (r) is given for each comparison. Solid squares and circles represent the values for men and women, respectively

In the presence of clarithromycin, the fraction of dose excreted into the urine as the 1'-hydroxymidazolam metabolite over a 24-h period was similar for i.v. (34%, 95% CI 27, 42) and p.o. (43%, 95% CI 37, 49, Table 2). However, administration of clarithromycin for 7 days significantly reduced the fraction of the dose recovered in the 24-h urine collection as 1'-hydroxymidazolam following i.v. and p.o. midazolam administration compared with the fraction of the doses collected in the absence of clarithromycin (Table 2).

The extent of interaction between midazolam and clarithromycin (as determined by the ratio of inline imageof midazolam, normalized for dose, in the presence and absence of clarithromycin) was not significantly different between men and women after i.v. or p.o. midazolam dosing (Table 2).


Elderly individuals are more likely to take multiple concomitant medications, leading to an increased risk of drug–drug interactions. The Food and Drug Administration's publication Guideline for Industry Studies in Support of Special Populations: Geriatrics emphasizes the importance of studying drug–drug interactions in the elderly [25]. Drug disposition may change with advancing age and drug–drug interactions may differentially affect elderly and younger adults. However, few studies have specifically examined the inhibition of cytochrome P450s in elderly populations. Loi et al. reported that age and gender do not influence the extent of inhibition of CYP1A2 assessed by the reduction in clearance of theophyline in the presence of ciprofloxacin and cimetidine [26], but corresponding studies for CYP3A-mediated metabolism are lacking.

The benzodiazepine midazolam is commonly used as an in vivo probe of CYP3A activity. Midazolam is exclusively metabolized by CYP3A to two metabolites, 1'-hydroxymidazolam and 4-hydroxymidazolam, with the formation of 1'-hydroxymidazolam being the principle route of in vivo and in vitro metabolism; <1% of the administered midazolam dose is excreted unchanged in the urine. Importantly, and unlike many other CYP3A substrates, midazolam does not undergo p-glycoprotein-mediated transport [20, 27]. Therefore, the systemic and oral clearances of midazolam in vivo reflect only routes of CYP3A-mediated biotransformation. Thummel and coworkers have demonstrated that when midazolam is administered intravenously, the systemic clearance of midazolam is correlated with hepatic CYP3A protein expression and activity [28, 29]. Likewise, the expression of CYP3A in the intestine plays a major role in the first-pass extraction of midazolam following p.o. administration [9, 30]. Therefore, administration of i.v. and p.o. midazolam enables one to separate the contribution of CYP3A from the intestine and liver in an individual. A potentially confounding factor in the execution of these studies is the day-to-day variability in CYP3A activity due to environmental factors (e.g. diet). Our approach circumvents this issue by using the simultaneous dosing of i.v. midazolam and stably labelled p.o. 15N3-midazolam [9].

In this study it was found that, in agreement with data from young subjects, oral midazolam was completely absorbed from the gastrointestinal tract, as indicated by the percentage of dose recovered in the urine as 1'-hydroxymidazolam (Table 2) [9]. Similarly, the bioavailability of midazolam (34%) indicated extensive first-pass metabolism of midazolam at both intestinal and hepatic sites that was not different from that observed in young subjects [9]. As noted previously [31–33], the systemic clearance of midazolam in elderly subjects (33.2 l h−1) was similar to the clearance in young volunteers (27.8 l h−1 kg−1) [9]. There was no difference between men and women in the systemic or oral clearance of midazolam. Our earlier work found sex-dependent differences in systemic and oral clearance in young subjects, but this has not been a consistent finding [19, 30, 34]. Thus, the disposition of midazolam appears to be unaffected by advancing age in the context of the specific subject characteristics of this study. Nevertheless the possibility remains that older, less healthy or ‘fragile’ elderly individuals will display significant alterations in effective CYP3A activity.

Clarithromycin significantly reduced both the i.v. and p.o. clearances of midazolam to 35% and 14% of control values, respectively (Table 2). The greater reduction in oral clearance reflects the contribution of first-pass metabolism, particularly in the intestinal wall. Unlike the younger population previously studied [9], women did not exhibit a significantly greater degree of inhibition than men. Following clarithromycin treatment, the percentage of the midazolam dose recovered in urine as 1'-hydroxymidazolam over 24 h decreased by 50% and 39% for i.v. and p.o. midazolam, respectively. This change in urinary recovery of 1'-hydroxymidazolam is consistent with decreased catalytic activity of CYP3A towards midazolam and a lack of alternative clearance pathways for midazolam as demonstrated by the increased half-life of midazolam (Table 2). While the amount of 1'-hydroxymmidazolam excreted into the urine following p.o. administration is essentially the same as that following i.v. administration, this p.o. to i.v. ratio increases following inhibition by clarithromycin (Table 2). This may reflect that first-pass formation of 1'-hydroxymidazolam by CYP3A in the gut wall is not inhibited to the same extent as hepatic clearance of midazolam.

Exposure to clarithromycin and its metabolites, 14-hydroxyclarithromycin and N-desmethylclarithromycin, was similar between men and women (Table 4). When compared with data obtained previously, the clearance of clarithromycin was decreased in the elderly (10–27 l h−1; Table 4) compared with young healthy volunteers, where clearance ranged from 36 to 105 l h−1[9]. Similar reductions in the oral clearance of clarithromycin in the elderly, with a corresponding increase in AUC, have been documented previously [17, 35, 36]. The average calculated creatinine clearance of our study population was 58 ml min−1 (Table 1) and the reduction in the oral clearance of clarithromycin probably reflects this age-related decrease in renal function. The renal clearance of clarithromycin following seven 500-mg doses is approximately 120 ml min−1 and accounts for 30% of the systemic clearance [37]. Chu et al. have reported that the renal clearance of clarithromycin in the elderly decreases by approximately 50% compared with healthy young volunteers [17]. This corresponds to a 30% decrease in total oral clearance.

The extent of interaction between clarithromycin and midazolam was not different from that observed in younger individuals. The ratios of midazolam's AUC in the presence to its AUC in the absence of clarithromycin were 3.2 (95% CI 2.7, 3.8) in elderly (Table 2) vs. 3.1 (95% CI 2.4, 3.8) in young volunteers [9] following i.v. administration and 8.0 (95% CI 5.7, 10.2) in elderly and 8.1 (95% CI 5.7, 10.5) in young following p.o. administration. Calculation of hepatic extraction incorporates hepatic blood flow (QH, Equation 6). As no information is available regarding hepatic blood flow in a healthy elderly population, we calculated hepatic blood flow using information derived from a young, healthy population in which estimation of QH is dependent on body weight [21]. In the event that QH is reduced in elderly individuals, FH would be reduced. However, a 10% reduction in QH resulted in an insignificant (<5%) reduction in FH. In this elderly population, women exhibited a significantly lower body weight than men, which may bias the calculation of QH. However, the changes in hepatic and intestinal availability in this elderly population were in agreement with our previous report of midazolam inhibition in the young [9]. In both the elderly and young populations, midazolam bioavailability increased 2.1-fold following clarithromycin dosing. Intestinal availability increased 1.7-fold and hepatic availability increased 1.2-fold following clarithromycin treatment. This change in intestinal and hepatic availability corresponds to that found in young, healthy volunteers (2.0 and 1.2, respectively) [9]. These changes in availability correspond, on average, to a 75% reduction in the intrinsic clearance of midazolam at both hepatic and intestinal sites [9]. Thus, despite the greater functional contribution of the gut wall metabolism to the increase in midazolam bioavailability, the inhibition of CYP3A activity at intestinal and hepatic sites is equivalent. This calculated change in intestinal wall CYP3A activity is in good agreement with the 75% reduction in midazolam 1'-hydroxylation in intestinal biopsy specimens obtained from subjects who received the same regimen of clarithromycin [7].

The method of analysis that we employed is based upon the blood clearance of midazolam. As demonstrated by Equation 2, the relationship between RBC affinity and total serum midazolam concentration was nonlinear. However, as midazolam is >97% bound to plasma proteins and <1.5% of midazolam in blood partitions into RBCs, this nonlinearity is of relatively minor importance. An alternative approach was employed by Thummel et al.[30], in which partitioning into RBCs was considered insignificant at the observed plasma midazolam concentrations. Thummel et al. subsequently used systemic plasma clearance and liver plasma flow to determine hepatic availability. Application of the Thummel method to our data results in essentially identical conclusions.

Despite the increased systemic exposure to clarithromycin in our elderly volunteers, the extent of inhibition of CYP3A activity was essentially equivalent to that in young volunteers [9]. This is not unexpected for intestinal wall CYP3A, because there is no reason to believe that the elderly enterocytes were exposed to a higher clarithromycin concentration. The failure of increased systemic concentrations of clarithromycin to cause greater inhibition of hepatic CYP3A might reflect the fact that maximal inhibition of midazolam hydroxylation has been achieved at the lower concentrations seen in the young volunteers. This would be equivalent to assuming that the fraction of the hepatic intrinsic clearance of midazolam that is dependent on CYP3A enzymes is approximately 75%. Although possible, it is generally assumed that at least 90% of the midazolam intrinsic clearance depends on CYP3A enzymes and is therefore potentially inhibited by clarithromycin. In this study population of Whites, the potential for differential inhibition of CYP3A4 and CYP3A5 is unlikely to contribute to the outcome because it is likely that only 10–30% of the volunteers expressed CYP3A5 [3, 6].

A likely explanation for the lack of greater CYP3A inhibition in the elderly may be found in the mechanism by which clarithromycin exerts an inhibitory effect. We and others have shown that clarithromycin is a weak competitive inhibitor, but forms a metabolite intermediate complex (MIC) in vitro and in vivo. The MIC is a drug–enzyme complex that results in inactive CYP3A4 and is irreversible under physiological conditions. At the clarithromycin concentrations experienced after 7 days dosing at 500 mg bid (3.2 µm, 95% CI 2.8, 3.0) the rate of formation of MIC would be maximal (kinact) and independent of concentration. Thus, the concentration difference between young and elderly subjects would be expected to be inconsequential. Despite operating at the maximal rate of inactivation, there is not necessarily complete inhibition of CYP3A4 because it is the balance between the maximal rate of inhibitor-driven inactivation (kinact) and the rate constant of CYP3A4 degradation (kdeg) that determines the extent to which the active pool of enzyme is diminished.

In conclusion, a significant interaction was observed between clarithromycin and midazolam, resulting in a >50% reduction in systemic clearance and doubling of the systemic availability of midazolam in elderly subjects. Thus, midazolam and clarithromycin should be coadministered with caution, because increased sedation is to be expected. Likewise, clarithromycin would be expected to inhibit the clearance of other CYP3A substrates such as ciclosporin and nifedipine. Caution on the part of the clinician is of particular importance when clarithromycin is coadministered with drugs having a narrow therapeutic index. A greater degree of interaction would be expected for drugs that undergo extensive first-pass elimination in the intestine, such as felodipine [38], tacrolimus [39] and ciclosporin. In comparison with our previous study on the interaction of clarithromycin and midazolam in a younger population, the extent of interaction was similar in our relatively healthy elderly group. Thus, CYP3A4 inhibition studies that are conducted in younger populations may be used to predict the extent of drug–drug interactions in healthy elderly individuals.

Supported by NIH Grants T32GM08425, GM067308, and by M01 RR00750 to the Indiana University General Clinical Research Center.