Effect of P‐glycoprotein inhibition at the blood–brain barrier on brain distribution of (R)‐[¹¹C]verapamil in elderly vs. young subjects

Abstract

Aims

The efflux transporter P‐glycoprotein (ABCB1) acts at the blood–brain barrier (BBB) to restrict the distribution of many different drugs from blood to the brain. Previous data suggest an age‐associated decrease in the expression and function of ABCB1 at the BBB. In the present study, we investigated the influence of age on the magnitude of an ABCB1‐mediated drug–drug interaction (DDI) at the BBB.

Methods

We performed positron emission tomography scans using the model ABCB1 substrate (R)‐[¹¹C]verapamil in five young [26 ± 1 years, (mean ± standard deviation)] and five elderly (68 ± 6 years) healthy male volunteers before and after intravenous administration of a low dose of the ABCB1 inhibitor tariquidar (3 mg kg⁻¹).

Results

In baseline scans, the total distribution volume (V _T) of (R)‐[¹¹C]verapamil in whole‐brain grey matter was not significantly different between the elderly (V _T = 0.78 ± 0.15) and young (V _T = 0.79 ± 0.10) group. After partial (incomplete) ABCB1 inhibition, V _T values were significantly higher (P = 0.040) in the elderly (V _T = 1.08 ± 0.15) than in the young (V _T = 0.80 ± 0.18) group. The percentage increase in (R)‐[¹¹C]verapamil V _T following partial ABCB1 inhibition was significantly greater (P = 0.032) in elderly (+40 ± 17%) than in young (+2 ± 17%) volunteers. Tariquidar plasma concentrations were not significantly different between the young (786 ± 178 nmol l⁻¹) and elderly (1116 ± 347 nmol l⁻¹) group.

Conclusions

Our results provide the first direct evidence of an increased risk for ABCB1‐mediated DDIs at the BBB in elderly persons, which may have important consequences for pharmacotherapy of the elderly.

What is Already Known about this Subject

• P‐glycoprotein (ABCB1)‐mediated drug–drug interactions (DDIs) at the blood–brain barrier (BBB) may lead to adverse effects of drugs on the central nervous system.

• The expression and function of ABCB1 at the BBB decreases with healthy ageing.

What this Study Adds

• An assessment was carried out of the influence of age on the magnitude of an ABCB1‐mediated DDI between the model ABCB1 substrate/inhibitor pair (R)‐[¹¹C]verapamil and tariquidar using positron emission tomography imaging.

• A higher increase in the brain uptake of (R)‐[¹¹C]verapamil in elderly subjects was found following partial ABCB1 inhibition.

• Direct evidence was found of an increased risk of ABCB1‐mediated DDIs at the BBB of the elderly.

Tables of Links

TARGETS
Transporters 2
ABCB1

LIGANDS
Verapamil

These Tables list key protein targets and ligands in this article that are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 1, and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 2.

Introduction

The blood–brain barrier (BBB) is a sophisticated anatomical structure which protects the brain from potentially harmful blood‐borne substances 3. This protective mechanism is enhanced by the expression of efflux transporters in the luminal (blood‐facing) membrane of brain capillary endothelial cells 3. These transporters belong to the adenosine triphosphate (ATP)‐binding cassette (ABC) family and use the energy of ATP to translocate their substrates, against concentration gradients, from the endothelial cell into blood. The best characterized of these transporters is P‐glycoprotein (ABC subfamily B, member 1, humans: ABCB1, rodents: Abcb1a). ABCB1 restricts the brain distribution of a multitude of different, structurally unrelated drugs and may be implicated in drug–drug interactions (DDIs) at the BBB 4. In such ABCB1‐mediated DDIs, concomitant administration of two drugs which are recognized by ABCB1 may result in the partial inhibition of ABCB1 transport activity at the BBB and increases in the brain distribution of one of these drugs, compared with the administration of one of these drugs alone. Transporter‐mediated DDIs are of great concern in drug development as they may lead to central nervous system (CNS) adverse effects of drugs. Studies in mice have shown that genetic knockout of Abcb1a can result in dramatic increases in the brain distribution of drugs which are selective ABCB1 substrates (e.g. ivermectin, vinblastine, digoxin, cyclosporine A, loperamide) 5, 6, 7. However, for drugs that are recognized by other efflux transporters, in addition to ABCB1, at the BBB, in particular breast cancer resistance protein (ABCG2), much smaller increases in brain distribution will be achieved when only ABCB1 is inhibited, as the other non‐inhibited transporter(s) may compensate for the function of ABCB1 8, 9. According to current thinking, transporter‐mediated DDIs are unlikely to occur in humans as most drugs are not expected to achieve high enough unbound plasma concentrations to lead to significant efflux transporter inhibition at the human BBB 10. Transporter‐mediated DDIs at the BBB occur in most cases without changes in drug plasma pharmacokinetics, so that drug concentrations in the brain need to be considered in order to assess such DDIs 11, 12. Positron emission tomography (PET) is an imaging method which enables the non‐invasive measurement of the distribution of radiolabelled drugs within the body. It has been used before to assess transporter‐mediated DDIs in various organs, such as the brain 13, 14, 15, 16 and the liver 17.

Disturbances in BBB integrity are involved in normal ageing and contribute to the onset and progression of neurological diseases 18, 19, 20. There is evidence that ABCB1 expression at the BBB declines in healthy ageing and in Alzheimer's disease 21, 22, 23. PET imaging with (R)‐[¹¹C]verapamil can be used to measure non‐invasively the function of ABCB1 at the human BBB 24, 25. Results from PET studies suggest a moderate decrease in ABCB1 function at the human BBB with increasing age 26, 27, 28, 29. We hypothesized that an age‐associated reduction in ABCB1 expression and function at the BBB will lead to higher increases in the brain distribution of ABCB1‐selective substrates, when ABCB1 is partially (incompletely) inhibited, which could result in an elevated risk of ABCB1‐mediated DDIs at the BBB in elderly as compared with younger persons.

In the present study, we compared the effect of ABCB1 inhibition with the third‐generation ABCB1 inhibitor tariquidar 30 on the brain distribution of the model ABCB1 substrate (R)‐[¹¹C]verapamil in groups of young and elderly healthy volunteers.

Methods

Approvals, recruitment and screening

The study was registered with EudraCT (number 2013–001724‐19), approved by the ethics committee of the Medical University of Vienna (study number: 1456/2013) and conducted in accordance with the Declaration of Helsinki and its amendments. After giving written informed consent, subjects underwent clinical medical assessments, including medical history, use of medication and drugs of abuse, and a physical and brief neurological examination. All subjects showed normal values on laboratory screening and urinalysis. Additionally, elderly participants underwent a Mini‐Mental State Examination (MMSE) 31. Subjects' magnetic resonance imaging (MRI) scan, performed on an Achieva 3.0 T scanner (Philips Medical Systems, the Netherlands), was evaluated by a neuroradiologist and confirmed as normal. No medication interfering with ABCB1 (e.g. St John's wort, rifampicin, loperamide) or with metabolizing enzymes (e.g. ketoconazole) was allowed.

Imaging protocol and data analysis

Volunteers underwent two consecutive 40‐min dynamic (R)‐[¹¹C]verapamil PET scans (injected activity: 378 ± 19 MBq, containing 13 ± 7 nmol of unlabelled (R)‐verapamil) on an Advance scanner (General Electric Medical Systems, Milwaukee, WI, USA) with arterial blood sampling, as described previously 32. For construction of an arterial input function, total radioactivity in the arterial plasma was corrected for polar radiolabelled metabolites of (R)‐[¹¹C]verapamil using a previously described solid‐phase extraction protocol 33. In brief, arterial plasma samples were passed over a C18 solid‐phase extraction cartridge. The cartridge was then washed with water and eluted with methanol. Radioactivity contained in the methanol fraction represented (R)‐[¹¹C]verapamil and its lipophilic radiolabelled metabolites, whereas radioactivity in the other fractions represented polar radiolabelled metabolites of (R)‐[¹¹C]verapamil. Lipophilic radiolabelled metabolites of (R)‐[¹¹C]verapamil, which are transported by ABCB1, were included in the arterial input function 24. The ABCB1 inhibitor tariquidar (AzaTrius Pharmaceuticals, Mumbai, India) was administered as an intravenous infusion over 30 min at a dose of 3 mg kg⁻¹ body weight at the end of the first PET scan, 3 h before start of the second PET scan, as described previously 34. Whole‐brain grey matter, and the temporal lobe, parietal lobe, cerebellum and cingulate gyrus were outlined as regions of interest using the Hammersmith maximum probabilistic atlas of the human brain 32. Total distribution volume (V _T), which equals the brain to plasma concentration ratio of radiotracer at steady state, as well as the rate constants describing the transfer of radiotracer across the BBB, were estimated using a standard two tissue–four rate constant (2T4K) compartment model (Figure 1) 34. The vascular volume fraction in the brain was included as a fitting parameter. To obtain a model‐independent estimate of V _T, Logan graphical analysis was performed 35. Tariquidar plasma concentrations at the time of the second PET scan were determined using liquid chromatography tandem mass spectrometry 36. From concentration–time curves of radioactivity in plasma, normalized to injected dose per body weight and expressed as standardized uptake value (SUV), the area under the curve (AUC) was calculated.

Figure 1.

Diagram of the two tissue–four rate constant (2T4K) compartment model used for the kinetic modelling of (R)‐[¹¹C]verapamil PET data. C _p is concentration of radiotracer in arterial plasma; C ₁ and C ₂ denote the radiotracer concentrations in the first and second brain tissue compartment. K ₁ (ml (g*min)⁻¹) and k ₂, k ₃ and k ₄ (min⁻¹) are first‐order rate constants describing the transfer of radiotracer between the plasma, the first and the second brain tissue compartments. Total distribution volume (V _T) is calculated as (1 + (k ₃/k ₄) * (K ₁/k ₂) and corresponds to the brain to plasma concentration ratio of the radiotracer at steady state

Statistical analysis

Our study was exploratory, so no sample size calculation was performed. All data are presented as mean ± standard deviation (SD). The Mann–Whitney U test was used for statistical comparisons of measurements in the young and elderly, and the Wilcoxon signed‐rank test for comparison of outcome parameters within one group (scan 1 and scan 2). To assess correlations, the Spearman rank coefficient (r) was calculated. Statistical analysis was performed using Statistica 6.1 (StatSoft, Tulsa, OK, USA). Statistical significance was set at P < 0.05.

Results

Five young men (mean age: 26 ± 1 years) and five elderly men (mean age: 68 ± 6 years) were included in the study. There was no significant brain atrophy present on MRI scans, and the MMSE scores for the elderly group (30 ± 0.4) were within normal range. All volunteers were medication free at time of the PET study, except for one elderly volunteer, who was on stable medication with tamsulosin, an alpha‐1 antagonist for the treatment of benign prostatic hyperplasia.

In all subjects, (R)‐[¹¹C]verapamil PET scans were performed before (scan 1) and after (scan 2) infusion of a low dose of tariquidar. In both groups, tariquidar administration did not significantly change the plasma exposure to (R)‐[¹¹C]verapamil and its lipophilic radiolabelled metabolites (Figure 2A). The mean AUC of unmetabolized (R)‐[¹¹C]verapamil and its lipophilic radiolabelled metabolites in the plasma, calculated 3.5–40 min after radiotracer injection (unit: SUV*min), was 21 ± 3 in scan 1 and 25 ± 4 in scan 2 in the young group and 20 ± 3 in scan 1 and 20 ± 4 in scan 2 in the elderly group (Figure 2A). The fraction of (R)‐[¹¹C]verapamil and its lipophilic radiolabelled metabolites relative to total plasma radioactivity at 40 min after radiotracer injection was not significantly different between the young and the elderly group, either in scan 1 and scan 2 (scan 1: young: 0.70 ± 0.02, elderly: 0.77 ± 0.06; scan 2: young: 0.73 ± 0.06, elderly: 0.76 ± 0.03) (Figure 2B). As the main outcome parameter of (R)‐[¹¹C]verapamil brain distribution, we chose V _T, which equals the brain to plasma concentration ratio of radiotracer at steady state, and which was found in previous studies to be sensitive to ABCB1 inhibition 16, 25, 34. In scan 1, (R)‐[¹¹C]verapamil V _T values in whole‐brain grey matter were not significantly different (P = 0.968) between the elderly (V _T = 0.78 ± 0.15) and the young (V _T = 0.79 ± 0.10) group. In response to ABCB1 inhibition with tariquidar, V _T was significantly increased in elderly subjects (V _T = 1.08 ± 0.15, P = 0.043), but not in young subjects (V _T = 0.80 ± 0.18, P = 0.999) (Figure 3A). Figure 4 shows representative PET images in young and elderly subjects. The percentage change in (R)‐[¹¹C]verapamil V _T in scan 2 was significantly greater (P = 0.032) in elderly (+40 ± 17%) than in young (+2 ± 17%) volunteers (Figure 3B). V _T values determined with Logan analysis were in good agreement with V _T values calculated with the 2T4K model (Table 1). Similar results as in whole‐brain grey matter were obtained for the other brain regions which had been selected based on previous studies to assess ABCB1 function at the BBB (Table 1) 26. Other than V _T, the influx rate constant of radioactivity from the plasma into the brain, K ₁, was significantly increased in the elderly but not in the young group following ABCB1 inhibition (Table 1).

Figure 2.

Area under the concentration–time curve [AUC (mean ± standard deviation)] of unmetabolized (R)‐[¹¹C]verapamil and its lipophilic radiolabelled metabolites in the plasma 3.5–40 min after radiotracer injection (A), and the fraction of (R)‐[¹¹C]verapamil and its lipophilic radiolabelled metabolites relative to total plasma radioactivity (mean ± standard deviation) at 40 min after radiotracer injection (B) in young and elderly subjects for scan 1 and scan 2. SUV, standardized uptake value

Figure 3.

(R)‐[¹¹C]verapamil distribution volume [V _T, (mean ± standard deviation)] calculated from the two tissue–four rate constant model in whole‐brain grey matter for young and elderly subjects before (scan 1) and after (scan 2) ABCB1 inhibition (A) and percent change of V _T (mean ± standard deviation) in response to P‐glycoprotein inhibition (B). *P < 0.05, Mann–Whitney U test or Wilcoxon signed‐rank test

Figure 4.

Axial planes of standardized uptake value (SUV) positron emission tomography average images (0–40 min) for young and elderly subjects before (baseline) and after P‐glycoprotein (ABCB1) inhibition

Table 1.

Modelling outcome parameters for different brain regions

	Temporal lobe		Parietal lobe		Cerebellum		Cingulate gyrus		Whole‐brain grey matter
	Scan 1	Scan 2	Scan 1	Scan 2	Scan 1	Scan 2	Scan 1	Scan 2	Scan 1	Scan 2
K1 (ml (g*min)−1) young	0.05 ± 0.01 (8)	0.06 ± 0.02 (7)	0.05 ± 0.01 (7)	0.06 ± 0.02 (13)	0.06 ± 0.02 (14)	0.07 ± 0.03 (12)	0.06 ± 0.01 (19)	0.06 ± 0.01 (13)	0.05 ± 0.01 (6)	0.07 ± 0.02 (10)
K1 (ml (g*min)−1) elderly	0.05 ± 0.01 (13)	0.07 ± 0.02 (8)*	0.04 ± 0.01 (10)	0.06 ± 0.02 (7)*	0.05 ± 0.02 (13)	0.08 ± 0.03 (12)	0.05 ± 0.01 (22)	0.07 ± 0.02 (14)*	0.05 ± 0.01 (11)	0.06 ± 0.02 (8)*
k2 (min−1) young	0.28 ± 0.12 (30)	0.27 ± 0.13 (60)	0.17 ± 0.04 (29)	0.25 ± 0.13 (47)	0.47 ± 0.37 (52)	0.43 ± 0.35 (54)	0.34 ± 0.16 (71)	0.17 ± 0.10 (66)*	0.22 ± 0.06 (24)	0.46 ± 0.24 (49)
k2 (min−1) elderly	0.36 ± 0.19 (50)	0.33 ± 0.18 (43)	0.25 ± 0.14 (42)	0.25 ± 0.15 (30)	0.24 ± 0.23 (47)	0.30 ± 0.23 (57)	0.28 ± 0.16 (93)	0.26 ± 0.18 (70)	0.32 ± 0.19 (41)	0.21 ± 0.08 (49)
k3 (min−1) young	0.28 ± 0.09 (26)	0.44 ± 0.18 (76)	0.16 ± 0.05 (41)	0.37 ± 0.22 (80)	0.31 ± 0.19 (52)	0.48 ± 0.26 (84)	0.33 ± 0.14 (94)	0.31 ± 0.21 (128)	0.21 ± 0.07 (28)	0.59 ± 0.14 (41)*
k3 (min−1) elderly	0.40 ± 0.15 (38)	0.40 ± 0.15 (62)	0.28 ± 0.16 (45)	0.33 ± 0.22 (49)	0.18 ± 0.15 (86)	0.28 ± 0.17 (65)	0.32 ± 0.16 (105)	0.29 ± 0.16 (95)	0.32 ± 0.15 (40)	0.28 ± 0.08 (70)
k4 (min−1) young	0.08 ± 0.01 (10)	0.17 ± 0.06 (25)*	0.08 ± 0.01 (14)	0.16 ± 0.08 (46)	0.08 ± 0.02 (16)	0.16 ± 0.04 (53)*	0.13 ± 0.10 (59)	0.28 ± 0.17 (78)*	0.09 ± 0.01 (10)	0.16 ± 0.07 (20)*
k4 (min−1) elderly	0.09 ± 0.04 (21)	0.11 ± 0.04 (24)*	0.09 ± 0.03 (17)	0.12 ± 0.01 (32)	0.08 ± 0.02 (44)	0.10 ± 0.03 (22)	0.15 ± 0.09 (86)	0.12 ± 0.06 (36)	0.09 ± 0.03 (15)	0.12 ± 0.03 (23)*
VT model (ml g−1) young	0.82 ± 0.10 (1)	0.84 ± 0.20 (2)	0.78 ± 0.11 (2)	0.80 ± 0.17 (3)	0.81 ± 0.10 (4)	0.81 ± 0.17 (2)	0.76 ± 0.11 (6)	0.85 ± 0.21 (4)	0.79 ± 0.10 (2)	0.80 ± 0.18 (1)
VT model (ml g−1) elderly	0.82 ± 0.16 (2)	1.14 ± 0.16 (2)*	0.76 ± 0.14 (3)	1.04 ± 0.13 (2)*	0.78 ± 0.14 (6)	1.06 ± 0.15 (4)*	0.76 ± 0.16 (5)	1.16 ± 0.20 (5)*	0.78 ± 0.15 (2)	1.08 ± 0.15 (2)*
VT Logan (ml g−1) young	0.82 ± 0.11 (1)	0.83 ± 0.20 (1)	0.77 ± 0.10 (1)	0.80 ± 0.16 (1)	0.79 ± 0.09 (2)	0.80 ± 0.17 (1)	0.74 ± 0.10 (10)	0.85 ± 0.19 (0)	0.79 ± 0.10 (1)	0.81 ± 0.17 (0)
VT Logan (ml g−1) elderly	0.81 ± 0.16 (2)	1.13 ± 0.16 (1)*	0.75 ± 0.13 (2)	1.03 ± 0.14 (0)*	0.77 ± 0.14 (1)	1.05 ± 0.15 (2)*	0.75 ± 0.14 (2)	1.14 ± 0.22 (3)*	0.78 ± 0.15 (1)	1.07 ± 0.15 (0)*

Outcome parameters are given as mean ± standard deviation (n = 5 per group). The value in parentheses represents the precision of parameter estimates (expressed as their coefficient of variation as a percentage), averaged over the study subjects

K ₁, k ₂, k ₃, k ₄, rate constants for exchange of radioactivity between arterial plasma and brain tissue compartments; V _T, total distribution volume; V _T Logan, V _T calculated using Logan analysis

^* P < 0.05 for comparison with scan 1 using the Wilcoxon signed‐rank test

Tariquidar plasma concentrations during scan 2 were not significantly different (P = 0.222) between the young (786 ± 178 nmol l⁻¹) and the elderly (1116 ± 347 nmol l⁻¹) group (Table 2). There was a significant positive correlation between the percentage change of (R)‐[¹¹C]verapamil V _T in scan 2 and tariquidar plasma concentrations in the elderly (r = 1.000, P = 0.017) but not in the young (r = −0.200, P = 0.783) group (Figure 5). In a subset of three young (subjects 3, 4 and 5) and three elderly (subjects 8, 9 and 10) subjects, in whom tariquidar plasma concentrations at the time of the PET scans were comparable (Table 2), there was a significantly greater change (P = 0.025, Student's t‐test) in (R)‐[¹¹C]verapamil V _T in scan 2 in elderly (+28 ± 9%) than in young (−6 ± 14%) subjects.

Table 2.

Tariquidar plasma concentrationsa and percentage change in V _T b in scan 2 in individual young and elderly subjects

Young			Elderly
Subject number	Tariquidar in plasma (nmol l−1)	% V T change	Subject number	Tariquidar in plasma (nmol l−1)	% V T change
1	611	27	6	1423	56
2	601	2	7	1554	60
3	806	–11	8	920	39
4	914	−17	9	905	24
5	997	10	10	781	22

^aMean of two samples taken in the beginning and in the end of the 40‐min positron emission tomography scan

^bCalculated using the two tissue–four rate constant model

Figure 5.

Correlation between percentage change in (R)‐[¹¹C]verapamil distribution volume (V _T) in whole‐brain grey matter following P‐glycoprotein (ABCB1) inhibition and tariquidar plasma concentrations at the time of the positron emission tomography scan in young (A) and elderly (B) subjects. r, Spearman rank coefficient

Discussion

To our knowledge, this was the first study to examine directly the influence of age on the magnitude of an ABCB1‐mediated DDI at the human BBB. As the model ABCB1 substrate, we used (R)‐[¹¹C]verapamil, which was shown to be selectively transported by ABCB1, and not by ABCG2, at the BBB 37. As the model ABCB1 inhibitor, we used the nonmarketed third‐generation ABCB1 inhibitor tariquidar 30, which inhibits ABCB1 at the human BBB effectively, resulting in up to fourfold increases in (R)‐[¹¹C]verapamil brain uptake 25. Tariquidar also inhibits ABCG2, but at considerably higher concentrations than needed for ABCB1 inhibition 9, 38. In contrast to previous studies, in which we had administered tariquidar at doses up to 8 mg kg⁻¹ 16, 25, in the present study we administered only a low dose of tariquidar (3 mg kg⁻¹) to achieve incomplete inhibition of ABCB1 at the BBB. Incomplete ABCB1 inhibition is a more realistic scenario for the clinic than complete inhibition, as clinically used drugs are not expected to lead to complete ABCB1 inhibition at the human BBB at clinically relevant plasma concentrations. In baseline scans without ABCB1 inhibition, we failed to detect significant differences in (R)‐[¹¹C]verapamil V _T values in the small sample of young and elderly subjects investigated in the present study (Figure 3A). This stands in contrast with previous (R)‐[¹¹C]verapamil PET studies, conducted without administration of an ABCB1 inhibitor, which found moderately increased V _T values (i.e. by 15–18%) in the brains of elderly subjects as compared with young subjects 26, 27, 28. Our failure to reveal differences in ABCB1 function in baseline scans may have been related to the limited sensitivity of high‐affinity ABCB1 substrates such as (R)‐[¹¹C]verapamil to detect moderate changes in ABCB1 expression at the BBB. This has been attributed to the fact that ABCB1 is a high‐capacity transporter, which is able to compensate functionally for moderate (<50%) decreases in its expression so that the brain distribution of ABCB1 substrates may remain unchanged 10, 39. This phenomenon is, in fact, similar to the functional interplay between ABCB1 and ABCG2 at the BBB, resulting in only very small increases in the brain distribution of dual ABCB1/ABCG2 substrates, when ABCB1 alone is inhibited 9. It has been estimated that the BBB needs to be depleted by more than 90% of ABCB1 to result in large (>2‐fold) increases in the CNS penetration of ABCB1 substrates 10. In agreement with this, we previously found that the brain distribution of (R)‐[¹¹C]verapamil was increased by only 1.5‐fold in heterozygous Abcb1a knockout mice, which have a 50% reduction in Abcb1a density at the BBB as compared with wild‐type mice, whereas a 3.9‐fold increase in (R)‐[¹¹C]verapamil brain distribution was observed in homozygous Abcb1a knockout mice, which completely lack Abcb1a at the BBB 39. Our previous results indicate that partial inhibition of ABCB1 increases the sensitivity of (R)‐[¹¹C]verapamil to measure changes in ABCB1 function at the BBB 40, 41. In line with this, we found that, following ABCB1 inhibition, (R)‐[¹¹C]verapamil V _T values were significantly higher in elderly than in young subjects (Figure 3A). Despite the small sample size of our study, remarkably clear differences could be detected between the two groups. Tariquidar plasma concentrations at the time of the PET scans ranged from 601 nmol l⁻¹ to 1554 nmol l⁻¹ (Table 2), which was below the previously determined in vivo half‐maximal effect concentration of tariquidar to enhance the brain uptake of (R)‐[¹¹C]verapamil in young healthy volunteers (2248 nmol l⁻¹) 25. In two elderly subjects, tariquidar plasma concentrations were higher than in the young group (Table 2). When comparing only subjects with similar tariquidar plasma concentrations (n = 3 per group), (R)‐[¹¹C]verapamil V _T increases in scan 2 were still significantly higher in the elderly group, suggesting that the observed group differences were not caused by differences in tariquidar plasma concentrations.

The significantly higher increases in (R)‐[¹¹C]verapamil brain distribution following partial ABCB1 inhibition found in elderly as compared with young subjects (Figure 3B) may have been caused by lower expression levels of ABCB1 in the brain capillaries of elderly persons 22. The observed differences between elderly and young subjects are expected to disappear in the case of complete ABCB1 inhibition at the BBB; however, this is unlikely to occur in the clinic. Our data provide the first evidence for an increased risk of ABCB1‐mediated DDIs at the BBB in elderly persons, which is of clinical relevance as elderly persons often take several different drugs on a regular basis 42. Our results suggest that particular caution is warranted in the elderly, when coadministering drugs which inhibit ABCB1 (e.g. cyclosporine A or quinidine) with ABCB1 substrates, to avoid the occurrence of CNS adverse effects. This is unlikely to be relevant for drugs which are transported by other efflux transporters at the BBB, in addition to ABCB1, such as dual ABCB1/ABCG2 substrates. The ABCB1‐mediated DDI risk may be increased further in certain patient groups, in whom ABCB1 density at the BBB is reduced. For example, it has been shown that, in addition to several other changes occurring at the BBB 18, the expression of ABCB1 is reduced at the BBB of patients with Alzheimer's disease 21.

In the present study, a model ABCB1 substrate was used at subtherapeutic doses (<20 μg) in combination with a potent, experimental ABCB1 inhibitor. Further studies, with larger sample sizes, including both male and female subjects 26, are needed to find out the extent to which our results are also valid for clinically relevant combinations of ABCB1‐substrate drugs and (less potent) ABCB1‐inhibiting drugs, ideally used at clinical doses. A limitation of the present study was the relatively small number of subjects. Inclusion of a greater number of subjects into our study was hampered by our complex study protocol, involving two consecutive PET scans, with arterial blood sampling and intravenous infusion of tariquidar, and also by the limited availability of tariquidar for clinical use. Another limitation of our study was that we did not measure cerebral blood flow (CBF) in study participants. It has been suggested that the brain distribution of [¹¹C]verapamil is dependent on CBF, and that [¹¹C]verapamil K ₁ needs to be normalized to CBF to measure ABCB1 function at the BBB 43. However, it appears unlikely that the age‐associated differences in (R)‐[¹¹C]verapamil brain distribution following ABCB1 inhibition observed in our study were caused by a higher CBF in the elderly, as CBF does not increase with age 44. In addition, tariquidar administration does not lead to changes in CBF 15.

In conclusion, we found significantly higher increases in the brain distribution of the model ABCB1 substrate (R)‐[¹¹C]verapamil following partial ABCB1 inhibition in elderly vs. young subjects. This confirms previous findings of reduced ABCB1 expression/function at the BBB of the elderly and provides the first evidence of an increased risk for ABCB1‐mediated DDIs at the BBB in elderly persons. The results of our study need to be confirmed in larger cohorts of subjects and need to be extended to other, clinically relevant combinations of ABCB1 substrates and inhibitors.

Competing Interests

There are no competing interests to declare.

This work was supported by the Austrian Science Fund (FWF) projects F 3513‐B20 (expired) and KLI 480‐B30 (to Oliver Langer). The authors would like to thank Maria Weber from the Department of Clinical Pharmacology, Harald Ibeschitz and Ingrid Leitinger from the Division of Nuclear Medicine and Peter Marhofer from the Department of Anaesthesiology for supporting this study. Vanessa Berger‐Kulemann from the Department of Biomedical Imaging and Image‐guided Therapy is acknowledged for help in MRI evaluation.

Bauer, M. , Wulkersdorfer, B. , Karch, R. , Philippe, C. , Jäger, W. , Stanek, J. , Wadsak, W. , Hacker, M. , Zeitlinger, M. , and Langer, O. (2017) Effect of P‐glycoprotein inhibition at the blood–brain barrier on brain distribution of (R)‐[¹¹C]verapamil in elderly vs. young subjects. Br J Clin Pharmacol, 83: 1991–1999. doi: 10.1111/bcp.13301.