A short-term high fat diet increases exposure to midazolam and omeprazole in healthy subjects

ABSTRACT

Objectives: Knowledge of factors contributing to variation in drug metabolism is of vital importance to optimize drug treatment. This study assesses the effects of a short-term hypercaloric high fat diet on metabolism of five oral drugs, which are each specific for a single P450 isoform: midazolam (CYP3A4), omeprazole (CYP2C19), metoprolol (CYP2D6), S-warfarin (CYP2C9) and caffeine (CYP1A2).

Methods: In 9 healthy volunteers, pharmacokinetics of the five drugs were assessed after an overnight fast at two separate occasions: after a regular diet and after 3 days of a hypercaloric high fat diet (i.e. regular diet supplemented with 500 mL cream [1715 kcal, 35% fat]). Pharmacokinetic parameters (mean [SEM]) were estimated by non-compartmental analysis.

Results: The high fat diet increased exposure to midazolam by 19% from 24.7 (2.6) to 29.5 (3.6) ng ml-1h-1 (p=0.04) and exposure to omeprazole by 31% from 726 (104) to 951 (168) ng ml-1h-1 (p=0.05). Exposure to metoprolol, caffeine and S-warfarin was not affected by the high fat diet.

Conclusion: A short-term hypercaloric high fat diet increases exposure to midazolam and omeprazole, possibly reflecting modulation of CYP3A4 and CYP2C19.

KEYWORDS:

• Cytochrome P450
• drug metabolism and pharmacokinetics
• high fat diet
• nutrition

1. Introduction

Many drugs are metabolized by hepatic cytochrome P450 (P450) enzymes before excretion via urine or feces. There is a wide variation in P450 enzyme activity not only between but also within individuals, which may account for adverse events or decreased efficacy of drugs.[1] In order to optimize drug treatment, it is important to study factors that could contribute to inter- and intra-patient variability in drug metabolism.

P450 enzyme activity is regulated via the nuclear receptors pregnane X-receptor (PXR) and constitutive androstane receptor (CAR).[2] These nuclear receptors recognize endogenous and exogenous ligands and transduce these internal and environmental signals into upregulation or downregulation of target genes, including P450 genes. Experimental studies have revealed numerous ligands of PXR and CAR, such as certain drugs, lipids, bile acids, and hormones.[3] Moreover, target genes of PXR and CAR are involved in lipid, as well as in drug metabolism.[4] These common molecular pathways suggest that interventions that affect lipid metabolism, such as a high-fat diet (HFD), can affect drug metabolism.

Animal studies have revealed differential effects of hypercaloric HFD on gene expression of different cytochrome P450 enzymes. In mice, a 14-week HFD (containing 60% kcal fat) decreased hepatic mRNA levels of CYP3A11, 2B10, and 2A4.[5] Interestingly, a study in pigs showed that a continuous HFD increased hepatic CYP2E1 activity, whereas a HFD on alternate weeks induced hepatic CYP3A4.[6] In male rats, a HFD increased CYP2E1 and CYP4A protein levels.[7] These studies point to an interaction between dietary composition of macronutrients and drug metabolism, although the effects differ among species and P450 isoforms and may depend on the type of diet.

Studies on the effect of food on pharmacokinetics in humans have mainly been focused on drug absorption,[8] oral bioavailability,[9] and on the effects of specific food components such as grape fruit juice on cytochrome P450 metabolism.[10] Isocaloric diets with a high fat percentage in healthy volunteers did not demonstrate an effect on drug metabolism.[11] However, these isocaloric diets probably did not induce steatotic changes in the liver. The accumulation of fatty acids in the liver, i.e. hepatic steatosis, might be required for activation of hepatic nuclear receptors in drug metabolism to occur.

Since a short-term hypercaloric HFD induces steatotic changes in the liver,[12] it may influence drug metabolism. Therefore, this study aimed to assess the effects of a short-term hypercaloric HFD on pharmacokinetics in humans. Since preclinical studies have shown a differential effect on P450 isoforms, we assessed the pharmacokinetics of five oral probe drugs (a validated cocktail approach), which are each specific for a single cytochrome P450 isoform: midazolam (CYP3A4), omeprazole (CYP2C19), metoprolol (CYP2D6), S-warfarin (CYP2C9), and caffeine (CYP1A2).[13]

2. Patients and methods

2.1. Subjects

Nine healthy, nonsmoking, young men were enrolled. Women were excluded since hormonal status or use of contraceptives may influence their lipid metabolism. Screening at baseline included medical history, vital signs, and laboratory assessments of liver and kidney function. During the total study period, they did not take any medicines or dietary supplements.

The study was approved by the human research ethics committee of the Academic Medical Center Amsterdam and conducted in accordance with the Declaration of Helsinki.

2.2. Study design

The study was a crossover intervention trial. Pharmacokinetics of the five probe drugs were assessed at two different occasions: after the subject’s own regular diet (control visit) and after a hypercaloric HFD for a period of 3 days consisting of the subject’s own regular diet supplemented with 500 ml cream (1715 kcal, 35% fat) after dinner (HFD visit). Participants were asked to keep food diaries and to take similar meals during the 3 days before both visits. Cream supplementation was scheduled after dinner to ensure that food intake during the day was not affected. We have previously documented by MR spectroscopy that this regimen doubles hepatic triglyceride content.[12] The order of interventions was determined by balanced assignment with a washout period of at least 4 weeks.

Before both visits, subjects were restricted to a regimen that excluded intensive physical activity (defined as more than 1 h a day) during the preceding 3 days. The consumption of grapefruits and alcohol during the preceding 2 days and caffeine containing beverages was not allowed the day prior to the study days. In order to exclude the effect of prior food intake on drug absorption, the drug cocktail was administered after an overnight fast on both occasions and a standardized liquid meal containing 25% of their estimated daily energy expenditure (25 kcal kg⁻¹ day⁻¹) was consumed 4 h after drug ingestion. The dietary restrictions ended 8 h after drug administration. Compliance to the study protocol was assessed by food diaries, phone calls, text messages, and e-mail contact.

2.3. Cocktail and pharmacokinetic sampling

The oral drug cocktail was administered after an overnight fast at 8.00 a.m. and contained 20 mg omeprazole (CYP2C219) (20 mg capsule, Teva Pharmachemie, Haarlem, The Netherlands), 100 mg metoprolol (CYP2D6) (100 mg tablet, Teva Pharmachemie, Haarlem, The Netherlands), 0.03 mg kg⁻¹ midazolam (CYP3A4) (1 mg ml⁻¹ oral solution, UMCG, Groningen, The Netherlands), 5 mg racemic warfarin (CYP2C9) (5 mg tablet, Crescent Pharma Ltd, Hampshire, United Kingdom), and 100 mg caffeine (CYP1A2) (10 mg ml⁻¹ ampoules, VUMC, Amsterdam, The Netherlands). These drugs are highly specific for the individual P450 isoforms and have been demonstrated not to interact with each other.[13]

Blood samples of 4 ml were collected in ethylenediaminetetraacetic acid tubes at t = 0, 1, 2, 3, 4, 5, 6, 8, 10 and 24, 48, 168, and 336 h after administration of the cocktail. After centrifugation, plasma samples were stored at −80°C until liquid chromatography/tandem mass spectrometry (LC–MS/MS) analysis.

2.4. Plasma drug concentration measurements

A LC–MS/MS method for the simultaneous determination of plasma concentrations of the five drugs used in the P450-probe cocktail was developed and validated. The plasma samples were spiked with internal standards (caffeine-D9, omeprazole-D3, warfarin-D5, metoprolol-D7, and midazolam-D5) [TRC-Canada]) in acetonitrile: methanol (420:80, v:v). Chromatography was performed on a Shimadzu LC-30 Nexera high-performance liquid chromatography (HPLC) system using Thermo Scientific Hypersil Gold HPLC column (50 mm × 2.1 mm, 1.9 μm). Compounds were separated using gradient elution (Table 1) at a flow rate of 0.4 ml min⁻¹. The HPLC column was coupled to a 5500 QTrap mass spectrometer (ABsciex). The MS/MS system was operated using an electrospray in positive ionization mode. For the chiral compound, S-warfarin, an Astec Chirobiotic V column (150 mm × 2.1 mm, 5 μm), was used and separation was performed using the same gradient elution and flow rate as for the other compounds. The MS/MS system for S-warfarin was operated using an electrospray in negative ionization mode. The precursor-to-product ion reactions monitored are presented in Table 2. The respective lower and upper limits of quantification (LLOQ and ULOQ) were 50–5000 ng ml⁻¹ for caffeine, 1–200 ng ml⁻¹ for metoprolol, 0.5–100 ng ml⁻¹ for midazolam, 2–500 ng ml⁻¹ for omeprazole, and 4–1000 ng ml⁻¹ for S-warfarin. Calibration standards and quality control (QC) samples were interspersed throughout each batch. Accuracy and precision were calculated following the measurement of LLOQ, method limit of quantification, and ULOQ samples during six consecutive runs. During the first run, six replicates for all levels were determined. During the next five runs, a single sample of each level was determined. Mean accuracy and within-run precision were calculated from the results of the first run (n = 6). Inter-assay precision was calculated from the results of run one through six (n = 6). The accuracy of the method was determined by comparing the means of the measured concentrations of the QC samples with their theoretical concentrations. The accuracy results demonstrated calculated deviation of mean value from nominal values in the range of −6–12%. The inter-assay results demonstrated a relative standard deviation for the QC samples of <13%. For all analytes, no significant interfering peaks were detected in matrix blanks.

Table 1. Gradient profile of HPLC mobile phase.

Step	Time (min)	Ultrapure water (%)	Acetonitrile (%)	2% Acetic acid/2% ammonium formate in water (%)	Flow (ml min^−1)
0	0.00	93.0	2.0	5.0	0.4
1	4.00	5.0	90.0	5.0	0.4
2	4.01	90.0	2.0	5.0	0.4
3	5.00	90.0	2.0	5.0	0.4

Table 2. Mass transition monitored for each probe and its internal standard.

Drug	Transition monitored	IS name	Transition monitored
Caffeine	194.85 → 138.00	Caffeine-D9	204.25 → 144.15
Metoprolol	268.05 → 191.00	Metoprolol-D7	275.25 → 191.10
Omeprazole	345.85 → 197.95	Omeprazole-D3	349.15 → 198.05
Midazolam	326.10 → 291.1	Midazolam-D5	331.1 → 296.1
R- and S-warfarin	307.00 → 160.85	Warfarin-D5	312.10 → 160.85

2.5. Pharmacokinetic analysis

For each drug, plasma concentration-versus-time profiles were plotted. Values under the LLOQ were not taken into account. The pharmacokinetic parameters area under the curve (AUC), maximum plasma concentration (C_max), and plasma half-life (t_1/2) were calculated by non-compartmental pharmacokinetic analysis software (‘PK Solver’ [14]) using the log linear trapezoidal setting. Oral clearance was estimated by the formula: dose/AUC_0–>inf and oral volume of distribution by the formula: plasma half-life × oral clearance/0.693.

2.6. Statistical analysis

The baseline biochemical parameters and the pharmacokinetic parameters at both visits were tested using the Wilcoxon signed-ranks test (IBM SPSS Statistics version 20.0).

Since only a relatively large difference between exposure to the probe drugs would be clinically relevant and because each subject serves as his own control in this paired study design, the sample size could be limited to nine subjects. Sample size was based on the pharmacokinetic parameters of midazolam as studied by Turpault et al. [13] – because CYP3A4 is the most important P450 enzyme in drug metabolism. When the sample size is nine, a single-group repeated measures analysis of variance with 0.05 significance level will have 80% power to detect a difference in the AUC of midazolam characterized by an effect size of 0.6367 (e.g. a variance of means, V = Σ(μ_i − μ)²/M, of 7.29, a standard deviation at each level, σ, of 10.7, and a between level correlation, ρ, of 0.900).

3. Results

All nine subjects (mean [SD] age 22.8 [3.1] years, weight 78.8 [7.3] kg, height 185.5 [6.1] cm) completed the study without the occurrence of any adverse events. During the study, INR (International Normalized Ratio) levels were monitored for safety reasons but did not increase after administration of a single dose of 5 mg warfarin. Table 3 shows that the baseline biochemical characteristics of participants at both visits did not differ.

Table 3. Effects of 3 days of a high-fat diet versus control diet on biochemical parameters in healthy men after an overnight fast.

	Control visit Median [range]	High-fat diet Median [range]	p Value
Glucose (mmol l^−l)	5.1 [4.65–5.45]	5.1 [4.95–5.35]	0.78
β-Hydroxybutyrate (mmol l^−1)	0.1 [0.0–0.15]	0.0 [0.0–0.1]	0.08
Acetoacetate (mmol l^−1) (N = 8)	0.03 [0.02–0.08]	0.03 [0.02–0.04]	0.21
Triglycerides (mmol l^−1)	0.82 [0.57–1.42]	1.18 [0.79–1.88]	0.11
Free fatty acids (mmol l^−1) (N = 7)	0.50 [0.28–0.70]	0.37 [0.21–0.48]	0.17
C-Reactive protein (mg l^−1)	0.4 [0.3–0.65]	0.5 [0.3–1.35]	0.55

The mean (±95% confidence interval) pharmacokinetic parameters (AUC_0–>inf, AUC_0–>t, plasma half-life, C_max, oral clearance, oral volume of distribution) of each individual drug in each condition are shown in Table 4.

Table 4. Pharmacokinetic parameters (mean [95% CI]) of cocktail substances after control diet and after a high-fat diet.

	Control	High fat	p Value
Midazolam
AUC0–>t (ng ml^−1 h^−1)	22.5 (17.6–27.4)	26.6 (20.5–32.7)	0.07
AUC0–>inf (ng ml^−1 h^−1)	24.7 (19.6–29.8)	29.5 (22.4–36.6)	0.04
t1/2 (h)	2.6 (2.2–3.0)	2.7 (2.5–2.9)	0.37
Cmax (ng ml^−1)	9.2 (7.4–11.0)	10.0 (7.8–12.2)	0.59
Tmax (h)	0.99 (0.92–1.06)	0.99 (0.94–1.04)	0.95
Vd/F (l)	323 (239–407)	329 (266–392)	0.68
Cl/F (l h^−1)	103 (81.8–124.2)	88.8 (67.2–110.4)	0.04
Omeprazole
AUC0–>t (ng ml^−1 h^−1)	721 (517–925)	947 (618–1276)	0.05
AUC0–>inf (ng ml^−1 h^−1)	726 (522–930)	951 (622–1280)	0.05
t1/2 (h)	0.9 (0.8–1.0)	0.9 (0.8–1.0)	0.52
Cmax (ng ml^−1)	315 (217–413)	544 (332–756)	0.03
Tmax (h)	2.00 (1.95–2.05)	1.68 (1.38–1.98)	0.52
Vd/F (l)	45.5 (26.9–64.1)	34.1 (22.9–45.3)	0.11
Cl/F (l h^−1)	32.5 (23.3–41.7)	28.9 (16.7–41.1)	0.31
Caffeine
AUC0–>t (ng ml^−1 h^−1)	11,689 (10,121–13,257)	10,741 (9408–12074)	0.33
AUC0–>inf (ng ml^−1 h^−1)	17,615 (13,971–21,259)	14,370 (12,073–16,667)	0.16
t1/2 (h)	5.7 (4.9–6.5)	4.5 (3.9–5.1)	0.07
Cmax (ng ml^−1)	1971 (1812–2130)	1965 (1720–2210)	0.89
Tmax (h)	0.99 (0.92–1.06)	1.11 (0.90–1.32)	0.59
Vd/F (l)	47.7 (44.6–50.8)	47.4 (42.0–52.8)	0.89
Cl/F (l h^−1)	6.3 (4.9–7.7)	7.4 (6.0–8.8)	0.26
Metoprolol
AUC0–>t (ng ml^−1 h^−1)	1268 (488–2048)	884 (229–1539)	0.31
AUC0–>inf (ng ml^−1 h^−1)	1294 (510–2078)	940 (285–1595)	0.44
t1/2 (h)	4.1 (2.7–5.5)	3.6 (2.4–4.8)	0.17
Cmax (ng ml^−1)	230 (154–306)	170 (115–225)	0.21
Tmax (h)	1.45 (1.05–1.85)	1.23 (0.92–1.54)	0.44
Vd/F (l)	653 (459–847)	827 (525–1129)	0.17
Cl/F (l h^−1)	157 (70–244)	223 (81–365)	0.21
S-warfarin
AUC0–>t (ng ml^−1 h^−1)	13,097 (11,370–14,824)	12,838 (11,017–14,659)	0.59
AUC0–>inf (ng ml^−1 h^−1)	14,110 (12,093–16,127)	13,893 (12,041–15,745)	0.86
t1/2 (h)	76.3 (66.9–85.7)	82.3 (70.5–94.1)	0.09
Cmax (ng ml^−1)	265 (243–287)	249 (225–273)	0.26
Tmax (h)	0.99 (0.92–1.06)	0.99 (0.94–1.04)	0.95
Vd/F (l)	20.0 (17.2–22.7)	21.7 (19.0–24.4)	0.21
Cl/F (l h^−1)	0.19 (0.17–0.21)	0.19 (0.17–0.21)	0.95

3.1. Midazolam (CYP3A4)

The HFD significantly increased the AUC_0–>inf of midazolam by 19% from 24.7 (19.6–29.8) to 29.5 (22.4–36.6) ng ml⁻¹ h⁻¹ (p = 0.04). In accordance, the oral clearance was significantly decreased by a HFD. The mean plasma concentration versus time plots for both interventions are shown in Figure 1. The HFD did not affect the C_max, t_1/2, nor the oral volume of distribution.

Figure 1. Mean (95% confidence interval) plasma concentration versus time plots for each individual drug after the high fat diet versus the control diet.

3.2. Omeprazole (CYP2C219)

The HFD significantly increased the AUC_0–>inf of omeprazole by 31% from 726 (522–930) to 951 (622–1280) ng ml⁻¹ h⁻¹ (p = 0.05). Figure 1 shows the mean plasma concentration versus time plots for both interventions. Notably, this effect is reflected by a significant increase in C_max after a HFD. There was no significant difference in oral clearance, plasma half-life, and oral volume of distribution.

3.3. Caffeine (CYP1A2)

The peak concentrations after oral caffeine administration varied from 1270 to 2860 ng ml⁻¹. The participant with a pre-administration concentration level of 1010 ng ml⁻¹ was excluded from this analysis since this would have a great effect on the AUC level. Ingestion of the HFD did not significantly affect the AUC_0–>inf of caffeine, nor the other pharmacokinetic parameters. The plasma concentrations versus time plots are shown in Figure 1.

3.4. Metoprolol (CYP2D6)

The HFD did not affect the AUC_0–>inf of metoprolol, nor the other pharmacokinetic parameters. The plasma concentrations versus time plots are shown in Figure 1.

3.5. S-warfarin (CYP2C9)

Warfarin is a racemic mixture of R and S enantiomers, of which S-warfarin is primarily metabolized by CYP2C9. The HFD did not significantly affect the pharmacokinetic parameters of S-warfarin. The plasma concentrations versus time plots are shown in Figure 1.

4. Discussion

This study investigated the effects of a short-term hypercaloric HFD on drug metabolism by assessing differences in exposure to five P450 specific probe drugs. To our knowledge, this is the first study assessing the effects of a short-term hypercaloric HFD on drug metabolism in humans. We have demonstrated that the 3-day hypercaloric HFD increases exposure to midazolam and omeprazole. In contrast, the HFD does not significantly alter exposure to S-warfarin, caffeine, and metoprolol. Considering these drugs as probes for individual P450 enzymes, this indicates that the HFD might possibly decrease CYP3A4 (midazolam) and CYP2C19 (omeprazole) activity, whereas this diet does not alter CYP2C9 (S-warfarin), CYP1A2 (caffeine), and CYP2D6 (metoprolol) activity.

This study did not directly address the mechanisms by which the preceding nutritional conditions alter pharmacokinetics of the probe drugs. Although lipids can be effective ligands for PXR and CAR, and therefore, one could expect upregulation of drug metabolism by increased lipid intake; in this study, we have demonstrated that a short-term hypercaloric HFD decreases drug metabolism. The increased lipid exposure in healthy volunteers induces hepatic accumulation of triglycerides,[12,15] which also involves an increase in the release of inflammatory cytokines such as tumor necrosis factor and interleukin-6.[16,17] Because an increase in these cytokines is associated with downregulation of P450 enzyme activity,[18] one may speculate that such an inflammatory response might be involved in the decreased activity of CYP3A4 and CYP2C19. In accordance, in vitro studies have also demonstrated reduced CYP3A4 activity in human steatotic livers.[19]

CYP1A2, CYP2D6, and CYP2C9 activity were not altered in the current study. It has been suggested previously that individual P450 enzymes could be selectively modulated by degree of liver disease.[20] This may implicate that at higher grades of steatosis, other P450 enzymes might be affected as well. Further studies on human liver tissue after a HFD might reveal underlying biochemical mechanisms.

The oral probe drugs used in the current study have been widely used to phenotype individual drug metabolism and to predict exposure to other drugs metabolized by the corresponding P450 enzyme.[21] However, it should be noted that there might be confounding factors for drug exposure (AUC) or oral clearance of probe drugs as a measure of specific P450 enzyme activity.

First, metabolization of the probe drugs is not a single transformation but involves multiple conversions. For instance, omeprazole is metabolized to 5-hydroxyomeprazole by CYP2C19 for about 80% and to hydroxysulfon by CYP3A4 for approximately 20%. This would mean that the effect on omeprazole could also be caused by altered CYP3A4 activity. However, our results show that a HFD has a greater impact on omeprazole exposure than on midazolam exposure. This indicates that the effect on omeprazole could only be partly due to changes in CYP3A4 activity but is most likely to be caused by CYP2C19. Another possibility to distinguish between CYP3A4 and CYP2C19 activity could be to include 5-hydroxyomeprazole into the analysis. However, 5-hydroxyomeprazole itself is also metabolized to hydroxysulfon by CYP3A4.[22] Furthermore, the measurement of 5-hydroxyomeprazole is not validated in combination with the other drugs used in our cocktail approach. The addition of 5-hydroxyomeprazole to our analysis would therefore not make omeprazole a better probe for CYP2C19.

Second, oral clearance is not only determined by P450 activity but also by drug protein binding (fraction unbound) and bioavailability. An increase in drug protein binding (fraction unbound decreases) or in bioavailability might also be responsible for the observed increase in exposure to midazolam and omeprazole. Since the drugs were ingested after an overnight fast at both visits, an effect of the HFD on drug absorption seems unlikely. The HFD might have an effect on bioavailability by altering the first-pass effect in the liver. This effect could be P450 enzyme dependent. It is known that the bioavailability of omeprazole increases from 60% to 90% in chronic liver disease due to a decrease in the first-pass effect of omeprazole in patients with chronic liver disease.[23] Assuming that the temporary-induced hepatic steatosis of the HFD mimics the effects of chronic liver disease, this is in accordance with our finding of a significant increase in the C_max of omeprazole after a HFD.

Nutritional effects on drug protein binding have not been investigated previously in humans. A recent study on pharmacokinetics of atazanavir (metabolized by CYP3A4) in rats has demonstrated a significant lower fraction of drug unbound and a lower clearance in HFD obese rats (4.2%) compared to controls (14%).[24] Since both midazolam and warfarin have a very high-plasma protein binding, the fraction unbound is very small (1–3%). A further decrease would quantitatively be a minor change but could relatively be important.

The additional caloric value of 500 ml cream used in this study (1715 kcal) is in accordance with that of a fast-food meal and induces an increase in hepatic fat content within 3 days.[12] Moreover, up to 25% of the general population in some western countries have hepatic steatosis.[25] Approximately 50% of all marketed drugs are metabolized by CYP3A4,[26] which was one of the enzymes that are possibly affected by a short-term HFD in this study. Therefore, additional studies should focus on other drugs metabolized by CYP3A4. We speculate that especially drugs with a small therapeutic range, such as oncolytic (i.e. tacrolimus, docetaxel, cyclophosphamide, imatinib, sunitinib) and antiretroviral (i.e. ritonavir, nelfinavir) drugs are relevant candidate drugs to be investigated for untoward effects of HFD or hepatic steatosis.

5. Conclusion

In conclusion, this study has shown that a HFD increases exposure to midazolam and omeprazole. Since these drugs are well-known probes for CYP3A4 and CYP2C19 activity, it is possibly that the HFD decreases the activity of CYP3A4 and CYP2C19. Further studies are needed to unravel underlying biochemical mechanisms, but this study provides proof of concept for nutritional conditioning, in this case for the effects of a HFD, of drug metabolism in healthy subjects.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Acknowledgments

The authors thank M Pistorius for his specific contribution to the laboratory analysis of this study.