articles
nature publishing group
Significantly Altered Systemic Exposure to Atorvastatin Acid Following Gastric Bypass Surgery in Morbidly Obese Patients IB Skottheim1, K Stormark2, H Christensen1, GS Jakobsen2, J Hjelmesæth2, T Jenssen3,4, JLE Reubsaet1, R Sandbu2 and A Åsberg1 The impact of gastric bypass on atorvastatin pharmacokinetics was investigated in 12 morbidly obese patients being treated with 20–80 mg atorvastatin each morning. Eight-hour pharmacokinetic investigations were performed the day before the surgery and at a median of 5 weeks (range 3–6 weeks) after the surgery. Gastric bypass surgery produced a variable effect on individual systemic exposure to atorvastatin acid (area under the plasma concentration vs. time curve from 0 to 8 h postdose (AUC(0–8))), ranging from a threefold decrease to a twofold increase (median ratio = 1.1, P = 0.99). Patients with the highest systemic exposure to atorvastatin before surgery showed reduced exposure after surgery (n = 3, median ratio = 0.4, range = 0.3–0.5, P < 0.01), whereas those with lower systemic exposure before surgery showed a median 1.2-fold increase in atorvastatin AUC(0–8) (n = 9, range = 0.8–2.3, P = 0.03) after surgery. This study indicates that the presurgical first-pass metabolic capacity influences the effect of gastric bypass on atorvastatin bioavailability. Because individual first-pass metabolic capacity is not readily assessable clinically, retitration up to the lowest effective dose should be performed after the surgery. Morbid obesity (body mass index (BMI) ≥40 kg/m2 or BMI 35–39 kg/m2 with comorbidity) is a growing global health problem.1,2 Bariatric surgery is the only approach that has demonstrated a long-term reduction in weight and comorbid medical conditions; this has led to a dramatic increase in the use of this surgical treatment for morbid obesity.3,4 Gastric bypass (Figure 1) is commonly performed worldwide and is the procedure of choice at our clinic.5 Following a gastric bypass, the redirected alimentary limb is connected directly to the remaining gastric pouch and the biliopancreatic secretions are delivered more distally (Figure 1). Physiological alterations that occur following a gastric bypass include reduction in the available gastrointestinal surface area and reduced retention of food and other exogenous compounds consequent to the reduced ventricular volume and diversion of the pylorus. The physio chemical changes that may occur as a result of these alterations include increase in the pH of the gastroventricular passage and increase in the osmotic load in the small intestine following a meal. The bypass of the proximal small intestine is likely to affect not only absorption of ingested nutrients and vitamins
but also the bioavailability of orally administered drugs, an issue only sparsely investigated so far. The published literature regarding the effects of bariatric surgery on drug bioavailability is presented mainly as case reports and data from small, uncontrolled pilot studies. Most of the reports concern gastric banding and jejunoileal bypass methods, the latter of which is no longer in use because of associated severe adverse events.6,7 In general, it has been reported that the bioavailability of orally administered drugs is decreased following gastric bypass. For example, increased dose requirements have been shown for cytochrome P450 3A (CYP3A) substrates such as tacrolimus, sirolimus, and cyclosporin A following gastric bypass surgery.8,9 The bioavailability of orally administered drugs is dependent on gastrointestinal absorption and is limited by intestinal and hepatic first-pass metabolism. In particular, CYP3A4 and CYP3A5 are involved in reducing the bioavailability of many drugs.10–12 The efflux transporter P-glycoprotein (P-gp) is another important restrictive mechanism of drug bioavailability.13,14 However, regarding the interplay between CYP3A enzymes and P-gp, it has recently been postulated (on the basis of data on
1Department of Pharmaceutical Biosciences, School of Pharmacy, University of Oslo, Oslo, Norway; 2Morbid Obesity Center, Vestfold Hospital Trust, Tønsberg, Norway;
3Medical Department, Rikshospitalet Medical Centre, University of Oslo, Oslo, Norway; 4Medical Department, Institute of Clinical Medicine, University of Tromsø, Tromsø, Norway. Correspondence: R Sandbu (
[email protected])
Received 4 March 2009; accepted 19 April 2009; advance online publication 3 June 2009. doi:10.1038/clpt.2009.82 Clinical pharmacology & Therapeutics | VOLUME 86 NUMBER 3 | SEPTEMBER 2009
311
articles drug interaction with P-gp) that changes in P-gp activity are of limited clinical importance.15 The expression of CYP enzymes and P-gp shows high interindividual variability and also varies over the length of the intestinal tract.16,17 Whereas the expression of CYP3A4 and CYP3A5 is highest in the duodenum and
decreases downstream in the intestine, the expression of P-gp increases from the stomach toward the colon.18 Atorvastatin, a 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase inhibitor (statin), shows low bioavailability (12%) because of considerable CYP3A4 and CYP3A5 metabolism in addition to being a P-gp substrate.19 Atorvastatin is administered in the active acid form20 but is partly interconverted in vivo to the inactive lactone form through non-CYP3A metabolism.21,22 Atorvastatin and its CYP3A-mediated hydroxymetabolites are present in both the acid and lactone forms.23 Given its extensive metabolism and P-gp-mediated transport, atorvastatin might be considered a suitable model drug for investigating processes involved in drug bioavailability. To our knowledge, no previous study has prospectively investigated the effects of bariatric surgery on the bioavailability of a drug in the same patients before and after surgery. The aim of this study was to investigate the effect of gastric bypass on the bioavailability of atorvastatin. Results Subject characteristics
All patients referred to the hospital for gastric bypass surgery were screened for inclusion in the study. All of the first 15 patients who fulfilled the inclusion and did not fill the exclusion criteria agreed to participate in the study. The data relating to 3 of the 15 patients were later excluded from the analysis. One patient was withdrawn from the study during the first pharmacokinetic investigation because of problems with clotting of the indwelling catheter (Venflon) that resulted in an insufficient number of blood samples from this patient. Two other patients were excluded because of deviating lengths of the alimentary
Figure 1 Schematic illustration of the gastroventricular tract following a gastric bypass procedure.
Table 1 Patient demographics BMI (kg/m2) Sexa
Age (years)
Before
After
AL (cm)
BPL (cm)
Dose (mg)
Time intervalb (weeks)
1
F
51
40
NA
100
100
40
6
2
M
50
34
31
120
100
20
4
3
F
47
42
NA
120
90
20
3
4
M
54
34
36
120
120
80
6
5
M
54
40
40
120
120
80
5
Patient
6
F
56
45
41
120
80
40
5
7
F
64
40
37
120
110
20
4
8
M
56
46
42
120
120
20
6
9
F
29
38
34
120
100
20
4
10
F
49
44
41
120
120
20
4
11
F
59
47
43
120
80
40
5
12
F
Mean
57
40
36
120
80
40
4
52
41
38
118
102
37
5
Minimum
29
34
28
100
80
20
3
Maximum
63
47
40
120
120
80
6
54
40
39
120
100
30
5
Median
8 F/ 4 M
AL, alimentary limb; BMI, body mass index; BPL, biliopancreatic limb; NA: not available. aFemale (F) and male (M). bThe time from the pre- to the postsurgery pharmacokinetic investigation (time interval).
312
VOLUME 86 NUMBER 3 | SEPTEMBER 2009 | www.nature.com/cpt
articles and biliopancreatic diversions, respectively. Demographic data of the 12 patients who completed the study and who were included in the data analysis are shown in Table 1. In four of the patients (33%), the drug was changed from simvastatin to atorvastatin before surgery. Despite having been asked to fast overnight, and although they fasted with respect to food, three of the patients took the atorvastatin dose on the morning of the pharmacokinetic investigation (presurgical investigation in two of the patients and postsurgical investigation in the third patient), and consequently C0 samples could not be obtained. Six patients were treated with 20 mg atorvastatin, four patients with 40 mg, and two patients with 80 mg daily. The mean (±SD) BMI prior to surgery was 44 ± 4 kg/m2, and at follow-up it was 38 ± 4 kg/m2. Two of the included patients (17%) were smokers. Comorbidities present in three or more of the patients were type 2 diabetes mellitus, hypertension, hypercholesterolemia, sleep apnea syndrome, and hypothyroidism. All patients were treated with esomeprazole or lansoprazole for 4 weeks following surgery. None of the patients received drugs known to interact with atorvastatin pharmacokinetics.
(Table 2 and Figure 2). The bioequivalence criteria were not fulfilled, as the 90% confidence intervals for AUC(0–8) and Cmax ratios were (0.72–1.39) and (0.67–1.66), respectively. There was no significant change in atorvastatin acid time to reach Cmax after surgery (Table 2, P = 0.39). The median terminal half-life was 12 h (range 5–30 h) before surgery. 350
300
AUC (ng·h/ml)
250
200
150
100
Pharmacokinetics of atorvastatin acid and lactone
Gastric bypass surgery showed a variable effect on systemic exposure to atorvastatin acid, ranging from a 2.9-fold decrease to a 2.3-fold increase (median ratio = 1.1, P = 0.99, Table 2). This was consistent with a median maximum plasma concentration (Cmax) ratio of 1.1 (ranging from 0.2 to 3.0, P = 0.83). Absolute area under the plasma concentration vs. time curve from 0 to 8 h postdose (AUC(0–8)) values ranged from 9 to 315 ng·h/ml before surgery to 17 to 107 ng·h/ml after surgery
50
0
Before
After
Figure 2 Individual atorvastatin area under the plasma concentration vs. time curve from 0 to 8 h postdose (AUC(0–8)) (ng·h/ml) changes from before the surgery to after the surgery.
Table 2 Atorvastatin acid pharmacokinetic parameters AUC(0–8) (ng·h/ml) Patient
Dose (mg)
Before
Cmax (ng/ml)
After
Ratio
Before
tmax (h)
After
Ratio
Before
After
Difference
9
20
9
21
2.3
2
4
1.9
1.1
3.0
1.9
10
20
14
16
1.2
5
4
0.7
0.6
0.5
−0.1
8
20
18
38
2.2
4
9
2.2
2.7
0.5
−2.1
7
20
19
21
1.1
5
4
0.8
0.7
1.3
0.5
3
20
42
48
1.1
8
12
1.6
0.8
3.0
2.2
2
20
43
35
0.8
7
7
1.0
1.0
0.8
−0.2
12
40
37
54
1.5
9
12
1.3
1.1
3.0
1.9
1
40
63
65
1.0
11
15
1.5
1.3
3.2
1.9
6
40
138
65
0.5
33
20
0.6
1.0
0.5
−0.5
11
40
163
63
0.4
117
21
0.2
1.0
1.5
0.5
5
80
35
61
1.7
6
17
3.0
5.9
1.6
−4.4
4
80
315
107
0.3
134
29
0.2
1.6
3.1
1.5
Median
30
40
51
1.1
7
12
1.1
1.1
1.5
0.5
Minimum
20
9
16
0.3
2
4
0.2
0.6
0.5
−4.4
Maximum
80
315
107
2.3
134
29
3.0
5.9
3.2
2.2
Mean
37
75
50
1.2
28
13
1.2
1.6
1.8
0.3
P value
0.99
0.83
0.39
AUC, area under the plasma concentration vs. time curve; Cmax, maximum plasma concentration; tmax, time to reach Cmax. Clinical pharmacology & Therapeutics | VOLUME 86 NUMBER 3 | SEPTEMBER 2009
313
articles The three patients with the highest systemic exposure to a torvastatin acid before surgery showed reductions in AUC(0–8) after gastric bypass (n = 3, median ratio = 0.4, range = 0.3– 0.5, P = 0.01). On the other hand, patients with low systemic exposure prior to surgery showed a median 1.2-fold increase in AUC(0–8) (n = 9, range = 0.8–2.3, P = 0.03) of atorvastatin acid. Of these nine patients, only one showed a reduction (19%)
in AUC(0–8). Gastric bypass reduced interpatient variability, with the coefficient of variation for atorvastatin acid AUC(0–8) decreasing from 121 to 52%. The effect of gastric bypass on atorvastatin lactone AUC(0–8) and Cmax levels in general showed the same pattern as observed for atorvastatin acid, although not all patients showed parallel alterations in lactone and acid exposure levels (Table 3). Absolute lactone AUC(0–8) values ranged from 4 to 79 ng·h/ml before surgery and from 5 to 33 ng·h/ml after surgery (Table 4). The median ratio of atorvastatin lactone AUC(0–8) was 0.9 (range = 0.4–1.8, P = 0.43) The median lactone/acid ratio in plasma was 0.4 (range = 0.1–0.9) before surgery and 0.3 (0.1–0.7) after surgery (Figure 3, P = 0.08). Only two patients showed an increase in the lactone/acid ratio; both had had high atorvastatin exposure before surgery. The only covariate that showed a significant association with change in systemic exposure of atorvastatin acid was presurgical systemic exposure to atorvastatin acid (ln-transformed AUC(0–8), univariate regression analysis, P < 0.001). Other covariates investigated were length of the biliopancreatic limb or the alimentary limb, total limb length, sex (fixed factor), BMI before surgery, BMI after surgery, time elapsed after surgery, and daily dose of atorvastatin.
0.9
Ratio of atorvastatin lactone over acid
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0
Genotyping Before
The frequencies of occurrence of genotypes CYP3A5 and ABCB1 are summarized in Table 4. Three of the patients were hetero zygotic for CYP3A5*1/*3 (25%). With regard to ABCB1, no patient expressed mutation alleles of G1199A. Six patients were homo zygotic for 3435T (50%), whereas three were heterozygotic (25%).
After
Figure 3 Individual changes in atorvastatin lactone/acid ratio from before the surgery to after the surgery.
Table 3 Atorvastatin lactone pharmacokinetic parameters AUC(0–8) (ng·h/ml) Patient
Cmax (ng/ml)
tmax (h)
Dose (mg)
Before
After
Ratio
Before
After
Ratio
Before
After
9
20
4
7
1.8
1
2
2.7
1.1
3.0
1.9
10
20
5
5
1.2
1
1
0.9
0.6
3.0
2.3
8
20
10
11
1.1
2
2
1.2
6.7
3.2
−3.4
7
20
16
15
0.9
3
2
0.7
1.5
5.1
3.6
3
20
18
18
1.0
4
4
1.1
1.8
5.7
4.0
2
20
10
8
0.8
2
1
0.8
6.0
0.8
−5.3
12
40
11
10
1.0
2
2
1.0
2.0
2.0
0.0
1
40
16
15
0.9
4
3
0.6
12.6
3.0
−9.6
6
40
79
33
0.4
12
8
0.7
2.9
1.5
−1.4
11
40
28
21
0.7
10
7
0.6
1.4
1.5
0.1
5
80
25
33
1.3
7
8
1.3
5.9
2.1
−3.8
4
80
19
11
0.6
5
3
0.5
1.6
3.1
1.5
Median
30
16
13
0.9
4
2
0.9
1.9
3.0
0.0
Minimum
20
4
5
0.4
1
1
0.5
0.6
0.8
−9.6
Maximum
80
79
33
1.8
12
8
2.7
12.6
5.7
4.0
Mean
37
20
16
1.0
4
4
1.0
3.7
2.8
−0.8
P value
0.43
0.44
Difference
0.70
AUC, area under the plasma concentration vs. time curve; Cmax, maximum plasma concentration; tmax, time to reach Cmax. 314
VOLUME 86 NUMBER 3 | SEPTEMBER 2009 | www.nature.com/cpt
articles Table 4 Genotyping frequency of CYP3A5 and ABCB1 Allele
Number of patients
CYP3A5*3 *3/*3
9
*1/*3
3
ABCB1 C1236T C/T
7
C/C
3
T/T
2
C3435T C/T
5
C/C
1
T/T
6
G2677A/T G/T
6
T/T
3
G/G
3
A/G
0
Two patients expressed the homozygote TTT haplotype of ABCB1 (16.7%). No associations were found between the different genotypes and changes in atorvastatin bioavailability, investigated as fixed factors in a univariate regression analysis. Two of the three patients expressing the CYP3A5*1/*3 genotype were patients with high systemic exposure to atorvastatin acid before the surgery. The metabolite profile of atorvastatin in the three patients who expressed CYP3A5 did not differ from that observed in the overall study population (data not shown). The different ABCB1 gene SNPs did not show any correlations with the effect of gastric bypass surgery on atorvastatin bioavailability. Safety
No serious adverse events were reported during the study period. Discussion
Laparoscopic gastric bypass induced an increased atorvastatin bioavailability in 8 of 12 patients (Figure 2). However, the three patients who had by far the highest presurgical systemic exposure to atorvastatin acid showed reduced bioavailability, a median 2.6-fold decrease. The mechanism for this variable effect of gastric bypass surgery on atorvastatin bioavailability is not evident from our study design. Extrapolation to the general population should be done with caution, given the relatively low power of the study. Bypassing the proximal small intestine reduces the surface area available for absorption. However, because the CYP content is greater in the proximal small intestine,18,24–26 bypass of this segment will also cause a relatively large reduction in overall gastrointestinal metabolic activity. This may increase the bioavailability of drugs, such as atorvastatin, that are subject to first-pass metabolism in the intestine. These two opposing
processes may, to different degrees, affect the bioavailability of orally administered drugs, depending on the drug and the individual patient characteristics. In patients with low gastrointestinal metabolic capacity, who are therefore likely to achieve a high systemic exposure, a restriction in absorption area can potentially reduce bioavailability. This could explain the net reduction in systemic exposure seen in the three patients with the highest dose-adjusted presurgical exposures to atorvastatin acid. In the other patients, the increase in bioavailability following gastric bypass might have been caused by a reduced metabolic elimination consequent to bypassing the CYP enzyme–containing proximal small intestine. We could not link factors such as diarrhea, vomiting, level of physical activity, or the use of drugs or herbal medicines to the differences in outcome in this study. However, other factors may also cause interindividual variability in drug bioavailability. Differences in gastric emptying and peristalsis can lead to variability in transit time and dissolution of the administered formulation, and inflammatory conditions of the intestine early after surgery could possibly affect drug bioavailability. As the solubility of atorvastatin acid increases with increasing pH, with an optimum at pH 6, an increased gastroventricular pH may result in increased solubility of atorvastatin.23 However, a shift of the equilibrium of atorvastatin toward the ionized form will also influence the passive absorption of the drug. The acid/lactone equilibrium may also be altered by changes in pH, affecting active and passive transport across the enterocyte membrane. It is therefore likely that absorption of atorvastatin could be affected by pH changes, but to what degree and in which direction is difficult to predict. Gastric bypass surgery has previously been reported to decrease the bioavailability of tacrolimus and cyclosporin A.8,9 Like atorvastatin, both tacrolimus and cyclosporin A are typical CYP3A and P-gp substrates, and it is therefore surprising that similar results were not obtained in this study. However, the previous reports on cyclosporin A and tacrolimus included only three and four patients, respectively, and for tacrolimus only postsurgical values were reported. The possibility cannot be ruled out that other factors such as formulation effects, body weight of patients, and time elapsed since surgery may be involved in determining the different effects of gastric bypass on these drugs. In a recent study, the bioavailability of simvastatin was found to be higher when administered as a delayed-release formulation rather than as an immediate-release formulation.27 This situation mirrors the surgical procedure with respect to bypass of the gastric ventricle and the proximal segments of the small intestine and supports the hypothesis that the loss of available intestinal absorption area by gastric bypass surgery may not be of critical importance in all subjects. Gastric bypass reduced variability in systemic exposure to atorvastatin acid in this study. CYP3A activity accounts for a great proportion of the variability in atorvastatin bioavailability, as shown in previous interaction studies performed with CYP3A inhibitors and inducers.28,29 The tendency of CYP3A inhibition to reduce interindividual variability in systemic exposure, as the outliers are moved toward the mean
Clinical pharmacology & Therapeutics | VOLUME 86 NUMBER 3 | SEPTEMBER 2009
315
articles population, parallels the decreased variability observed in our study. The coefficient of variability for atorvastatin lactone was also decreased following the bypass, although to a lesser extent than for the acid form. The interconversion between atorvastatin acid and lactone is not primarily mediated in the intestine and will therefore be less sensitive to the bypass. The lactone/ acid ratio showed substantial interindividual variability, possibly as a result of the direct interconversion between acid and lactone forms. It is therefore difficult to ascribe alterations in lactone/acid ratio to a direct relationship with the effects of gastric bypass on intestinal absorption and metabolism. Different genotypes for both CYP3A5 and P-gp can influence the variability of the metabolism as well as efflux transport for many drugs.30–32 Our study was too small and insufficiently powered to reveal any associations, if present, between these genotypes and the effect of gastric bypass on atorvastatin pharmacokinetics. In conclusion, this study revealed the existence of great interindividual variability in the effect of gastric bypass on systemic exposure to atorvastatin. The effect is complex but likely to depend on both available absorption area and the metabolic capacity of the bypassed small intestine. At present, it is not possible to predict the effect of gastric bypass in an individual patient without comprehensive pharmacokinetic investigations, and the requirements with regard to dose alterations cannot be generalized. Patients who have undergone gastric bypass should therefore be treated on a de novo basis and retitrated to the lowest effective doses. Further investigations are needed to explore the mechanisms behind the differential effects of gastric bypass on atorvastatin bioavailability. Methods
Study design. The study was designed as a prospective, open, con-
trolled, nonrandomized, single-center study. It was conducted in accordance with international and national laws and guidelines. Approvals were obtained from the Regional Committee for Medical and Health Research Ethics (REK Sør-Øst) and all relevant Norwegian authorities. The study is registered at ClinicalTrials.gov (NCT00331565). Fifteen patients over the age of 18 years, assigned to undergo bariatric surgery by the laparoscopic gastric bypass procedure, were included in the study after giving their written informed consent. All patients were treated with statins on clinical indication prior to inclusion. Patients who were taking statins other than atorvastatin were switched to atorvastatin (Lipitor) not less than 2 weeks prior to the investigation. Atorvastatin was administered once daily (each morning), and the dose was kept unchanged during the study period. The only exclusion criterion for the study was concomitant treatment with substances that could affect the pharmacokinetics of atorvastatin: gemfibrozil, erythromycin, clarithromycin, carbamazepine, rifampicin, ketoconazole, fluconazole, itraconazole, voriconazole, diltiazem, verapamil, dexamethasone, prednisolone, fenytoin, ritonavir, indinavir, nelfinavir, bosentan, telithromycin, nefazodone, St John’s wort, orlistat, and sibutramine. Comorbidity and concomitant medications used by the patients were recorded in the case report forms on each investigation day. Patients were asked to keep a drug diary, including dose administration time points, to ensure compliance. On the pharmacokinetic investigation days, patients met after an overnight fast (avoidance of food, drugs, nicotine, and caffeine; however, they were allowed to take water). A standard reduced-fat hospital breakfast was served 2 h after atorvastatin administration, and thereafter there were no further 316
food restrictions. Concomitant drugs ordinarily administered in the morning, where applicable, were given along with the breakfast. Pharmacokinetic investigations were performed the day prior to surgery and 3–6 weeks after surgery. Before surgery, patients were subjected to a 24-h pharmacokinetics investigation. Blood samples were collected before atorvastatin administration (0 h) and at 0.5, 1, 1.5, 2, 3, 5, 6, 8, 10, 12, 22, and 24 h after the dose. At the follow-up investigation, blood samples were collected at the corresponding time points, but only up to 8 h after the dose. The blood samples were drawn into 7-ml heparin vacutainer tubes (prechilled and kept on ice), centrifuged for 10 min at 4 °C (1,800 g), plasma decanted, and frozen at −80 °C within 1 h. One whole blood sample (4 ml ethylenediaminetetraacetic acid) was drawn for genotyping before surgery and stored at −20 °C. Surgical procedure. The bariatric procedure undergone by the patients
in this study was laparoscopic Roux-en-Y gastric bypass. A gastric pouch of ~10–30 ml, and an antecolic, antegastric Roux limb of either 100 or 120 cm and a biliopancreatic limb of 80–120 cm were created. Both the gastrojejunostomy of the Roux limb and the jejunojejunostomy were formed with a linear stapled 45 mm anastomosis. The length of the common limb was not measured. Atorvastatin analysis. Plasma samples were analyzed for atorvastatin
and its metabolites with a validated high-performance liquid chromato graphy method using tandem mass spectrometry detection as previously described.33 With the introduction of atorvastatin-d5 acid and atorvastatin-d5 lactone as internal standards, and stabilization of the temperature of the column at 40 °C, the performance of the method was evaluated by a partial validation. Briefly, plasma samples were prepared by solid-phase extraction with 1 ml C18 (100 mg) cartridges (Varian, Harbor City, CA). Atorvastatin and its metabolites were separated chromatographically on an Omnisphere C18 (3 µm, 30 × 2 mm2) analytical column with a Chromguard (5 µm, 10 × 2 mm2) guard column (both columns from Varian, Harbor City, CA) before electrospray tandem mass spectrometric detection. A linear gradient ranging from 100% mobile phase A (30:70 acetonitrile:1 mmol/l formic acid vol/vol) to 70% mobile phase A and 30% mobile phase B (60:40 acetonitrile:1 mmol/l formic acid vol/vol) was established within the first 5 min after injection. This composition was held for 14 min and reversed to 100% mobile phase A before re-equilibration at 0.4 ml/min for 5.5 min and 1 min at 0.2 ml/min at starting conditions. The flow rate at all other times was 0.2 ml/min. High-performance liquid chromatography analyses were performed on a Dionex Ultimate 3000 system (Instrument Teknikk AS, Østerås, Norway) equipped with an LPG-3000 pump fitted with a degasser, and an Ultimate 3000 autosampler coupled to an LC 20A Prominence CTO 20AC column oven (Shimadzu, Duisburg, Germany). The other high-performance liquid chromatography–mass spectrometry/mass spectrometry equipment used has been previously described.33 The lower limit of quantification (LLOQ) was 0.5 ng/ml for o-OH atorvastatin acid and 0.2 ng/ml for atorvastatin acid, atorvastatin lactone, p-hydroxyatorvastatin acid, p-hydroxyatorvastatin lactone, and o-hydroxyatorvastatin lactone. Validation showed linearity from 0.5 to 30 ng/ml for o-hydroxyatorvastatin acid and 0.2 to 30 ng/ml for atorvastatin acid and all other metabolites (R2 > 0.99). The intercepts were not significantly different from zero. The average extraction recoveries were >60% for all of the compounds. Intra- and interday errors in accuracy and precision were within 15% for low, middle, and high concentrations and within 20% at LLOQ for atorvastatin acid and lactone. For the majority of concentrations of p-hydroxy- and o-hydroxyatorvastatin acids, the intraday and between-day errors in accuracy and precision were within 15% for low, middle, and high concentrations and within 20% at LLOQ; all were within 18%, except for a 30% error in accuracy at LLOQ for o-hydroxyatorvastatin acid. Similarly, for p-hydroxy- and o-hydroxyatorvastatin lactones, the majority of concentrations showed intraday and between-day errors in accuracy and precision within 15% for low, middle, and high concentrations; within 20% at LLOQ; and within 21% for all concentrations. VOLUME 86 NUMBER 3 | SEPTEMBER 2009 | www.nature.com/cpt
articles Genotyping procedures. DNA was extracted from whole blood with
QIAamp (Qiagen, Valencia, CA) for determination of the patient CYP3A5*2 and *3 and MDR-1 (G1199A, C1236T, G2677T, G2677A, G2677G, and C3435T) genotypes, using specific primers, restrictive enzyme digestion, and separation on 3% agarose gels, as previously described.34,35 Positive controls were kindly supplied by D. Katz, Abbott Laboratories, Abbott Park, IL (MDR-1), and R. van Schaik, Department of Clinical Chemistry, Erasmus MC, The Netherlands (CYP3A5). Pharmacokinetic calculations. The Cmax and the time to reach Cmax
values reported are the actual measured values. AUC(0–8) was calculated using the trapezoidal method. The elimination rate constant (kel) was determined as the slope of the semilogarithmic plot of at least three points of the terminal elimination phase from the presurgical 24-h pharmacokinetic investigation. AUC(0–8) was adjusted for differences in actual sampling time by using kel from the presurgical measurement, assuming first-order elimination in the terminal phase. Missing C0 samples were substituted with C24 values from the presurgical measurements. The changes in bioavailability were evaluated from the postsurgery to presurgery AUC(0–8) ratio. Statistical analyses. Sample size was based on (i) the assumption that a
50% relative change in bioavailability following surgery was clinically relevant and (ii) a relative standard deviation of 40%.36 To ensure a power of 80% at the 5% significance level, 10 evaluable patients were needed. To allow for dropouts, 15 patients were included. The criteria of bioequivalence were used for assessing changes in bioavailability according to the European Medicines Agency guidelines.37 In short, the data were logarithmically (ln) transformed, the 90% confidence intervals for the AUC(0–8) and Cmax ratios (post-/presurgery) were calculated, and the acceptance range for the ratios (0.80–1.25) was applied. The effects of bariatric surgery on atorvastatin AUC(0–8) and Cmax were also evaluated using univariate regression on ln-transformed data, with the respective ratios as dependent variables. Time to reach Cmax was analyzed using the Wilcoxon signed-rank test on untransformed data. Differences were considered statistically significant if P values were
