Quantitative Structure-Interaction Relationship Analysis of 1,4-Dihydropyridine Drugs in Concomitant Administration with Grapefruit Juice
Total Page：16
File Type：pdf, Size：1020Kb
ORIGINAL ARTICLES Clinical Pharmaceutics Laboratory, Department of Pharmaceutics, Meiji Pharmaceutical University, Tokyo, Japan Quantitative structure-interaction relationship analysis of 1,4-dihydropyridine drugs in concomitant administration with grapefruit juice Y. Uesawa, K. Mohri Received July 11, 2011, accepted August 16, 2011 Dr. Yoshihiro Uesawa, Clinical Pharmaceutics Laboratory, Department of Pharmaceutics, Meiji Pharmaceutical University, 2-522-1, Noshio, Kiyose, Tokyo 204-8588, Japan [email protected] Pharmazie 67: 195–201 (2012) doi: 10.1691/ph.2012.1101 Quantitative structure-interaction relationship (QSIR) analyses of 1,4-dihydropyridine drugs were per- formed on grapefruit juice interaction potentials to characterize the interaction and evaluate drugs not yet tested in clinical research. AUC ratios of drugs with and without grapefruit juice ingestion were esti- mated as grapefruit juice interaction potentials from clinical studies on dihydropyridine drugs such as amlodipine, azelnidipine, benidipine, cilnidipine, felodipine, manidipine, nicardipine, nifedipine, nimodip- ine, nisoldipine, and pranidipine. The minimal energy conformation in each dihydropyridine drug was searched for using Merck Molecular Force Field (MMFFaq), and then geometry optimization was performed by density-functional-theory (DFT) calculation (B3LYP/6-31G**). The geometric, electronic, and physico- chemical features including molecular size, dipole moment, total energy, HOMO/LUMO energies, and logP values were then obtained. Dragon descriptors were also calculated by optimized 3D-structures. The rela- tion between the potentials and over 1000 of the molecular properties was investigated using statistical techniques including partial least squares analysis with genetic algorithm (GA-PLS) to a variable subset selection. Some PLS regression equations including logP values and dragon descriptors as explanatory variables were constructed in which the maximal contribution coefficient was 94%. These models could be applied to estimate the interaction potentials of other dihydropyridine drugs that have gone unreported in interactions with drugs such as aranidipine, barnidipine, clevidipine, lemildipine, lercanidipine, niguldipine, niludipine, and nilvadipine. In the assessment of major dihydropyridines, amlodipine was found to be the safest drug to avoid interactions among the drugs investigated in the present study. 1. Introduction GFJ (Dresser et al. 2000; Bailey and Dresser 2004). Thir- teen types of DHPs have been reported to be related with the Grapefruit juice (GFJ) evokes pharmaceutical interactions with above interactions in clinical studies (Uesawa 2008; Uesawa and increase in concentration of a variety of drugs in the sys- Mohri 2008b). In the findings, interaction-strength varied widely temic circulation (Uesawa 2008). Furanocoumarin derivatives according to the DHP. For example, amlodipine showed little such as bergamottin and 6’,7’-dihydroxybergamottin, potent increment in plasma concentrations (Josefsson 1996). On the CYP3A inhibitors, are identified as causative ingredients of GFJ other hand, AUC of azelnidipine was increased more than three (Tassaneeyakul 2000; Paine 2004). These compounds inhibit times following GFJ administration compared with the con- drug-metabolizing activities of intestinal CYP3A, a major drug- trol subjects that administrated the drug with water (Hirashima metabolizing enzyme in the gastrointestinal tract (Obach 2001). 2006). The chemical structures of DHPs consist of a common Because this enzyme functions as a barrier in the absorption of dihydropyridine skeleton and a great variety of residues. The dis- low-molecular substances in the intestinal mucosa, breakage of crepancy in the interaction potentials among DHPs is likely due the system by GFJ drinking causes leakage of substrate drugs to the differences in structures. In a previous paper, we reported into the blood from the digestive tract (Lown 1997; Schmiedlin- that the strength of pharmacokinetic interaction was related to Ren 1997). As a result, GFJ is able to cause increases in plasma the lipophilicities of DHPs (Uesawa and Mohri 2008b). That drug concentrations that may result in adverse effects of drugs is, logP values (logarithmic octanol/water distribution coeffi- (Dresser 2000). Dihydropyridine calcium channel antagonists cients) were calculated from the DHP structures by several (DHPs), used to treat hypertension and angina pectoris, are canonical algorithms and confronted with the interaction poten- in the highest category of drugs that undergo GFJ interac- cies of DHPs. As a result, significant relationships were found tions (Uesawa 2008). Patients receiving DHP treatment might with correlation coefficients of about 0.6. However, in the sim- experience hypotension related side effects such as nausea and ple regression analyses, relationships with a single descriptor of stagger if they administer the medication concomitantly with the compounds were accompanied by large deviations. Because Pharmazie 67 (2012) 195 ORIGINAL ARTICLES Table 1: Reported pharmacokinetic interactions of dihydropy- ridine derivatives following concomitant consumption of grapefruit juice in human subjects DHP Dose (mg) N GFJ (ml) AUC ratio Amlodipine 5 12 250 1.14 Azelnidipine 8 8 250 3.28 Benidipine 4 6 200 1.59 Cilnidipine 10 6 200 2.27 Efonidipine 40 19 250 1.67 Felodipine 5 6 250 2.51 Manidipine 40 6 250 2.36* Nicardipine 40 6 300 1.43 Nifedipine 10 6 250 1.35 Nimodipine 30 8 250 1.51 Nisoldipine 20 12 250 1.76 Nitrendipine 20 9 150 2.25 Pranidipine 2 16 250 1.73 N: number of subjects in the clinical studies. AUC ratios: (AUCDHPswith GFJ)/(AUCDHPs without GFJ) * Average ratio between R- and S-manidipine Fig. 1: Scatter plots between observed CIS and HATSp for 12 DHPs. Asterisk indicates a point of nitrendipine contribution of structural factors other than logP in the interac- 2.2. Simple regression analysis tions was presumable, a further quantitative structure-interaction relationship (QSIR) analysis was performed to construct better The calculation of Ghose-Crippen logP (LogPC) value using predictive models on the GFJ-interaction strength of untested Spartan software was unsuccessful only with the efonidip- DHPs. ine structure, which is among the 13 DHPs for which GFJ interactions have been reported. LogPC was one of the most significant descriptors in the remaining 12 DHPs (r = 0.705, p = 0.0105). That is because the structure of this DHP is unique 2. Investigations and results in that it has a phosphorus atom directly connected to the 2.1. Clinical interaction strength (CIS) DHP ring. This atom is not found in the structures of all other marketed DHPs. Therefore, efonidipine was excluded from the Thirteen DHPs, amlodipine (Josefsson et al. 1996), azelnidipine analysis objects as its unique characteristic might not be useful (Hirashima et al. 2006), benidipine (Ohnishi 2006), cilnidip- to predict interactions of DHPs that have not been reported to ine (Ohnishi et al. 2006), efonidipine (Yajima 2003), felodipine have GFJ interactions in clinical studies. (Bailey 1991), manidipine (Sugawara 1996), nicardipine (Uno Statistically significant parameters were extracted from the sim- 2000), nifedipine (Bailey et al. 1991), nimodipine (Bailey et al. ple regression analysis between all descriptors calculated by 1991), nisoldipine (Bailey 1993), nitrendipine (Soons 1991), and Spartan and Dragon software from the 12 kinds of DHP struc- pranidipine (Hashimoto 1998), on which there were confirmable tures and CSIs from the literatures. As a result, 98 kinds of sig- reports of pharmacokinetic interactions with GFJ were selected nificant parameters were discovered from a total of 1409 param- for the analysis (Table 1). However, efonidipine was eliminated eters. Investigation of scatter plots between these significant in the analysis because of reasons mentioned in the results sec- descriptors and CISs revealed that the plot of nitrendipine was tion. CISs were defined as common logarithmic values of the an outlier in cases with many descriptors. The diagram between AUC increasing ratio, in which the AUC of each DHP with GFJ HATSp and CIS is presented in Fig. 1 as an example. Sim- consumption was divided by the corresponding control AUC. ple regression analyses with 11 DHPs excluding nitrendipine showed much more significant parameters than with 12 DHPs = CIS log[(AUCDHP with GFJ)/(AUCDHP without GFJ)] (1) including nitrendipine. That is, 185 kinds of significant variables were found in the relationship with CIS. Furthermore, there were The first report of a significant interaction with GFJ intake for 21 kinds of very significant variables (p < 0.01), despite finding each drug referred to the AUC value in order to avoid the vari- only 3 variables in the DHP dataset with nitrendipine. These ations of CIS in publication bias (Uesawa 2010) (Table 1). findings suggest that use of the dataset for 11 DHPs excluding Table 2: High correlation descriptors (r > 0.8) in single regression with CIS and static values Descriptor a b n r RMSE F p HOv 1.629 ± 0.347 −1.660 ± 0.409 11 0.843 0.0774 22.1 0.00112 HATSp 0.598 ± 0.132 −1.578 ± 0.407 11 0.833 0.0796 20.4 0.00145 HATSv 0.661 ± 0.151 −1.488 ± 0.400 11 0.825 0.0814 19.1 0.00179 LogPC 0.088 ± 0.021 0.061 ± 0.054 11 0.814 0.0836 17.6 0.00230 HOp 1.339 ± 0.329 −1.432 ± 0.417 11 0.805 0.0854 16.6 0.00281 ALOGP 0.096 ± 0.024 −0.066 ± 0.084 11 0.803 0.0857 16.4 0.00290 LogPC2 0.019 ± 0.005 0.136 ± 0.040 11 0.802 0.0859 16.3 0.00296 a, slope; b, intercept