Scientific paper Synthesis and Identification of Some Impurities of Irbesartan Stanislav Radl,* Jan Stach, Jaroslav Havlicek, Marcela Tkadlecova, and Lukas Placek Zentiva, U kabelovny 130, 102 01 Prague 10, Czech Republic * Corresponding author: E-mail: stanislav.radl@zentiva.cz Received: 17-09-2008 Dedicated to Professor Blanko Stanovnik on the occasion of his 70'' birthday Abstract Synthesis of two principal impurities of irbesartan prepared via its W-trityl derivative is described. The impurities were isolated and unambiguouesly identified by NMR techniques. Spectral characteristics (IR, UV, MS) of these compounds are also given. Keywords: Irbesartan, Impurities, Identification, NMR, LC MS, IR, UV. 1. Introduction Irbesartan (1, SR 47436) is a member of a modern therapeutic group of drugs known as Angiotensin II Receptor Antagonists (AIIRAs) used in treatment of hyper-tension.1'2 Clinical results have shown that the blood pressure lowering ability of irbesartan is significantly better than that of losartan (Cozaar®), the first marketed AIIRA. Irbesartan is marketed by two different companies under the names Avapro® (Bristol-Myers Squibb) and Karvea® (Sanofi-Winthrop). One of the principal parts of documentation of any active pharmaceutical ingredient (API) is description of impurities and/or degradation products which can be present. The specified impurities can be either identified or unidentified. Identified impurities should be included in the specification when they are present at a level higher than the identification threshold, which is usually 0.10%. These impurities must be not only identified but also independently synthesized. In general, the impurities could be either process-related or formed by degradation of the drug. The originally described methods1,3 of preparation of irbesartan are based on N-alkylation of spiroimidazolo-ne 2 with biphenyl derivatives 3 and the formed intermediates 4 are then converted in several steps into irbesartan (Scheme 1). oK X refs 1,3 N Bu N^n n^nh Scheme 1 X = COO'Bu, CN Scheme 2 More recent procedures4-6 use 5-(4'-(bromomethyl) biphenyl-2-yl)-1-trityl-1H-tetrazole (5, BBTT), a common intermediate used in the sythesis of various sartans (Scheme 2). Intermediate tritylirbesartan (6) is then de-tritylated, usually under acid conditions, to provide irbe-sartan (1) and trityl alcohol. Recently we have developed more advantageous method of detritylation in boiling methanol providing irbesartan (1) and methyl trityl ether.7 2. Results and Discussion Irbesartan APIs produced by these different procedures have different impurity profiles. While two papers describing impurities related to the process shown in Scheme 1 have been published,89 no such report on impurities of the process of Scheme 2 has appeared. In irbesar-tan API prepared by the procedure, two principle impuri- ties are frequently detected by HPLC. The HPLC-MS data suggested two isomeric structures 7 and 8. Formation of these impurities is easily explainable by partial deprotec-tion of either 5 or 6 during the alkylation of 2 and the following alkylation of the tetrazole ring providing finaly the mixture of possible tritylated isomers 9 and 10. The final detritylation then provides impurities 7 and 8. Structures of impurities 7-10 are given in Fig. 1. For more precise HPLC determination of the impurity profile of irbesartan we needed to prepare these impurities for establishing their correction factors. Synthesis of both isomers was quite straightforward; it started with ir-besartan (1), which was alkylated by protected tetrazole derivative (5) in acetonitrile/^-ethyldiisopropylamine to give a mixture of protected isomers 9 and 10, containing 53% of 9 and 35% of 10 (HPLC). This result is quite surprising since similar ethylation of candesartan cilexetil with iodoethane10 gave a mixture of the N-1 and N-2 iso- Figure 1. Structures of the discussed impurities 7-10. mers in a ratio of 2 : 3 and with much bulkier (2'-(1-trityl-1H-tetrazol-5-yl)biphenyl-4-yl)methyl substituent we expected compound 10 as the major product. Attempts to separate compounds 9 and 10 by flash chromatography provided mixed fractions containing both compounds and only small amount of 9. A part of the mixture was subjected to centrifugally accelerated axial chromatography to get a minor fraction containing 9, major mixed fractions containing 9 and 10, and small amount of 10. Both compounds 9 and 10 were recrystalized and fully characterized (m.p., 1H NMR, 13C NMR, IR, UV, HRMS). Deprotection of compound 9 in boiling methanol provided the correspon- Figure 2. Numbering used for the discussed impurities 7-10. Table 1. 'H and 13C NMR assignments for impurities 7 and 9. Position 8(c) 1 9 [ppm] 7 8(H) [ppm] multiplicitya 9 7 Integ. H J(H,H) 9 7 1 161.35 162.32 2 186.69 186.78 3 76.55 76.44 4 37.40 37.31 1.80-2.10 m 1.80-2.10 m 1.75 m 1.92 m 2H 2H 5 26.06 25.98 1.80-2.10 m 1.85 m 4H 6 28.74 28.39 2.28 t 2.19 t 2H 7.6 7.5 7 27.68 27.66 1.57 qn 1.51 qn 2H 7.6 7.9 8 22.27 22.10 1.32 sx 1.30 sx 2H 7.6 7.5 9 13.68 13.55 0.86 t 0.85 t 3H 7.3 7.3 10 43.18 43.23 4.66 s 4.68 s 2H 11 136.59 135.93 12 127.20 126.85 7.10 m 7.03 d 1H 8.3 13 129.20 129.47 7.10 m 7.11 d 1H 8.3 14 138.41 138.41 15 141.03b 141.07 16 130.20 130.54 7.52 dd 7.58 dd 1H 7.7, 1.3 7.8, 1.2 17 131.61 131.81 7.60 dt 7.70 dt 1H 7.7, 1.3 7.6, 1.3 18 128.10 128.19 7.37 dt 7.54 dt 1H 7.7, 1.3 7.6, 1.3 19 131.29 130.97 7.25 m 7.42 dd 1H 7.8, 1.1 20 122.70 122.38 21 154.38 154.52 22 50.54 50.44 4.70 s 4.87 s 2H 23 131.62 132.29 24 127.20 128.14 6.55 d 6.65 d 2H 8.1 8.3 25 129.60 129.31 6.94 d 6.99 d 2H 8.1 8.3 26 141.59 139.88 27 141.04b 140.69 28 130.50 130.40 7.26 m 7.37 dd 1H 7.8, 1.1 29 130.01 131.09 7.48 m 7.59 dt 1H 7.6, 1.3 30 127.78 128.07 7.44 m 7.51 dt 1H 7.6, 1.3 31 130.30 130.72 7.89 dd 7.78 dd 1H 7.5, 1.6 7.8, 1.2 32 126.23 123.41 33 163.86 155.49 34 82.91 - - 35 141.19 - - 36 130.18 - 6.94 d - 6H 8.1 37 127.59 - 6.55 d - 6H 8.1 38 128.25 — 7.31 t - 3H 7.4 a - Standard abbreviation used: s = singlet, d = doublet, t = triplet, qn = quintet, sx = sextet, m = complex multiplet; b - Interchangeable. ding compound 7. The combined fractions were evaporated, the residue was deprotected by refluxing in methanol, and the formed mixture of 7 and 8 was successfully separated by flash chromatography on silica gel to provide pure compounds 7 and 8. Structures of these complex tetra-zole isomers were determined by NMR experiments; the used numbering is shown in Fig. 2. For the assignment of protons and carbons, several advanced 1D and 2D NMR techniques (COSY, HSQC, HMBC, ROESY, 1D NOESY) were used. D NOESY of impurities 7 and 8 are shown in Figs 3 and 4, respectively. Chemical shifts of protons and carbons are given in Tables 1 and 2, respectively. The differentiation of both impurities of irbesartan were performed using 1D NOESY experiments. In both cases proton 22 was excitated. While for impurity 7 signals of protons 13, 19 and 24 were found (Fig. 3), in the case of impurity 8 (Fig. 4) signals of aliphatic protons 6, 8, 9, in addition to the signals of protons 24, appear. 3. Experimental 2-Butyl-1,3-diazaspiro[4.4]non-1-en-4-one (2) and 5-(4' -(bromomethyl)biphenyl-2-yl)-1-trityl-1H-te- Table 2. 'H and '3C NMR assignments for impurities 8 and 10. Position ?(c) 10 [ppm] 8 ?(H) [ppm] multiplicitya 10 8 Integ.H 10 8 1 161.66 164.76 2 186.61 184.95 3 76.45 75.63 4 37.36 37.36 1.80-2.10 m 1.80-2.10 m 1.43 m 1.84 m 2H 2H 5 26.02 22.50 1.80-2.10 m 1.80-2.10 m 1.61 m 1.84 m 2H 2H 6 28.73 28.67 2.36 t 1.76 t 2H 7.6 6.9 7 27.68 27.60 1.61 qn 1.26 m 2H 7.6 8 22.22 22.50 1.37 sx 1.23 m 2H 7.6 9 13.65 13.46 0.91 t 0.82 t 3H 7.3 7.0 10 43.32 43.31 4.71 s 4.76 s 2H 11 135.30 134.59 12 126.30 125.69 7.10 d 7.12 d 2H 8.2 7.9 13 129.68b 130.02 7.18 dc 7.33 d 2H 7.9 14 140.43 141.66 15 141.30 141.07 16 130.75 130.84 7.46 m 7.44 m 1H 17 129.91d 130.44 757 m 7.58 m 1H 18 127.55c 128.09 7.49 m 7.54 m 1H 19 130.45 129.90 7.83 dd 8.17 m 1H 7.8, 1. 1 20 126.25 125.80 21 165.41 164.85 22 56.12 55.13 5.59 s 5.61 s 2H 23 131.78 133.54 24 127.50 125.91 7.05 d 6.74 d 2H 8.2 8.6 25 129.72b 129.53 7.18 dc 7.03 d 2H 8.2 26 141.73 139.46 27 141.20 140.66 28 130.65 130.44 7.43 m 7.39 dd 1H 7.9, 0.8 29 129.94d 131.25 7.56 m 7.61 dd 1H 7.5, 1.2 30 127.70 128.25 7.53 m 7.57 m 1H 31 130.20 131.21 8.02 dd 7.75 dd 1H 7.4, 1. 8 7.7, 1.1 32 126.12 123.90 33 163.80 154.58 34 82.90 - - 35 141.11 - - 36 130.17 - 6.94 d - 6H 7.3 37 127.55d - 7.27 t - 6H 8.0 38 128.18 — 7.35 m - 3H 7.6 a - Standard abbreviation used: s = singlet, d = doublet, t = triplet, qn = quintet, sx = sextet, m = complex multiplet; b -Interchangeable; c - Overlap; d - Interchangeable. Figure 3. 1D NOESY of impurity 7. Figure 4. 1D NOESY of impurity 8. trazole (5, BBTT) were obtained from Zhejiang Tianyu Pharmaceutical Company (http://www.tianyupharma. com). Other chemicals used in the synthesis were purchased from Sigma-Aldrich and were used without purification. Melting points were measured on a Kofler block and are uncorrected. The IR spectra were measured on a Perkin Elmer Spectrum BX FT-IR machine by the diffuse reflectance method (KBr), wavenumbers are given in cm-1. The UV spectra were recorded on a Hew- lett-Packard 8452A spectrophotometer (ethanol) in the range 190-400 nm. NMR experiments were carried out on a Bruker Avance 500 at 500.13 MHz (1H), 125.77 MHz (13C) and 50.70 MHz (15N). Reference for 1H 5 (CDClj) = 7.26 ppm, for 13C 5 (CDClj) = 77.0 ppm. All experiments were performed in CDCl3 at 298 K. COSY, HSQC, 1H,13C HMBC and 1H, 15N HM4BC spectra were recorded using pulse programs from the Bruker NMR standard library. At 500 MHz, standard 5 mm TXO (triple-nucleus X-observe) and TBI (triple-broadband in- verse) probeheads equipped with z-gradient coils were employed for all measurements. For the 'H-'3C HSQC, a dataset was acquired with 8 scans for each t1 increment at a resolution of 2048 and 256 points in the F2 and Fj dimensions, respectively. The time domain data were zero-filled to 2048 and 1024 data points in F2 and F1 dimensions, and multiplied with a sinusoidal squared sine-bell window function in both dimensions prior to Fourier transform. The gradient-selected 'H-'3C HMBC data sets were recorded with 4 K and 512 points in the F2 and F1 dimensions, respectively. The magnetization transfer in the 1H-13C HMBC experiment was optimized for a three-bond coupling constant 3J(C,H) of 8 Hz. The data was subsequently processed employing zero-filling to 2 K and 1 K data points in the F1 and F2 dimensions, using a sinusoidal squared sine-bell window function for apodization prior to Fourier transform in both dimensions. The ROESY spinlock was 200 ms. The mixing time for 1D NOESY was optimized to 800 ms. The Mass spectra (MS/MS; ionization mode AP-CI(+)) were measured on an API 3000 PE machine (Sciex Instruments, Applied Biosystems). The purity of the prepared substances was evaluated by TLC on silica gel (FP KG F 254, Merck) and by UPLC system Waters Acquity with UV detection (column length: 0.1 m, internal diameter 2.1 mm, stationary phase: UPLC BEH-C8, temperature: 35 °C). Gradient elution with mobile phase A (phosphate buffer [1.32 g (NH4)2HPO4 diluted in1000 mL of H2O, pH adjusted to 3.0 with 50%c phosphoric acid), and mobile phase B (acetonitrile) was used. Flash chromatography was performed on silica gel Merck, particle size 0.04-0.063 mm. Centrifugally accelerated axial chromatography was done using Cyclo-graphTM instrument (Analtech) with silica gel pre-scra-ped rotors. 2-Butyl-3-((2'-(1-((2'-(1-trityl-1H-tetrazol-5-yl)bip-henyl-4-yl)methyl)-1H -tetrazol-5-yl)biphenyl-4-yl)methyl)-1,3-diazaspiro[4.4]non-1-en-4-one (9) and 2-Butyl-3-((2'-(1-((2'-(2-trityl-2H-tetrazol-5-yl)biphenyl-4-yl)methyl)-1H-tetrazol-5-yl)biphenyl-4-yl)methyl)-1,3-diazaspiro[4.4]non-1-en-4-one (10). A mixture of irbesartan (1; 2.5 g, 5.8 mmol), 5-(4'-(bromomethyl)biphenyl-2-yl)-1-trityl-1H-tetrazole (5; 3.4 g, 6.1 mmol) and ^-ethyldiisopropylamine (1.7 g, 13 mmol) in acetonitrile (30 mL) was stirred at 90 °C under nitrogen for 5 hrs. The mixture was evaporated, the residue was dissolved in ethyl acetate (100 mL), the cloudy solution was washed with water (3 x 20 mL) and dried with magnesium sulfate. The residue after evaporation (5.17 g) contained 4.2% of 5, 53.1% of 9 and 35.5% of 10 (HPLC). The mixture was subjected to flash chromatography (silica gel; dichloromethane - ethyl acetate 50 : 1) to provide 5 (0.15 g), a mixture of 9 and 10 (3.4 g) and 9 (0.6 g). A part of the mixture (1.0 g) was subjected to cen-trifugally accelerated axial chromatography using Cyclo- graphTM (silica gel, from dichloromethane to dichloromet-hane - ethyl acetate 20 : 1) to give 10 (0.1 g), a mixture of 9 and 10 (0.75 g), and 9 (0.1 g). All mixed fractions were combined, evaporated (4.1 g) and used for further reaction. Both fractions of 9 (0.7 g) were crystallized from methylcyclohexane to give 0.55 g of white crystals. Compound 10 was characterized as obtained without further purification. Data for compound 9: m.p. 88-92 °C (from methylcyclohexane). 1H and 13C NMR (CDCl3): See Table 1. IR (KBr): 697 (5Ar), 746 (5Ar), 1006 (8 CH), 1027 (5_Ce), 1343 (sCH), 1444 (sCH), 1630 (VAr), 1720 (vC_O), 2848 (VCH), 2924 (vCH) cm-1. UV (EtOH), Àmax (log e): 206.0 (4.99); Àjnfj 252.0 (4.23). HRMS m/z calcd for C58H53N1oO [M+H]+ 905.44038, found 905.43988. Data for compound 10: m.p. 73-75 °C (from methylcyclohexane); 1H and 13C NMR (CDCl3): See Table 2. IR (KBr): 697 (5Ar), 746 (5Ar), 1006 (5_CH), 1032 (5_ch), 1342 (5ch), 14444 (5CH),1629 (vAr), 17_2^ (vC_O), 2870 (vch), 2960 (vch) cm-1. UV (EtOH), À^ax (log e): 206.0 (4.94). HRMS m/z calcd for C58H53N1oO [M+H]+ 905.44038, found 905. 43951. 3-((2'-(1-((2'-(1H-Tetrazol-5-yl)biphenyl-4-yl)methyl)-1H-tetrazol-5-yl)biphenyl-4-yl)methyl)-2-butyl-1,3-diazaspiro[4.4]non-1-en-4-one (7) and 3-((2'-(2-((2'-(1H-Tetrazol-5-yl)biphenyl-4-yl)methyl)-2H-te-trazol-5-yl)biphenyl-4-yl)methyl)-2-butyl-1,3-diazaspi-ro[4.4]non-1-en-4-one (8). A mixture of 9 and 10 (4 g, 4.4 mmol) was reflu-xed in methanol (75 mL) for 6 h, the mixture was evaporated and the residue was separated by flash chromato-graphy (silica gel; dichloromethane - methanol 20 : 1) to provide compound 7 (1.55 g, 53%) and 8 (0.95 g, 33%). Data for compound 7: m.p. 218-223 °C (methylcyclohexane - ethyl acetate 1 : 1); 1H and 13C NMR (CDCl3): See Table 1. IR (KBr): 697 (5Ar), 746 (5Ar), 1006 (5_Cjj), 1032 (5_ch), 1337 (5ch), 1403 (5CH), 1620 (vAr), 1_729 (vC_O), 42857 (VCH), 2962 (vCH) 4cm-1. UV (EtOH), (log e): 206.4 (4.77), 248.8 (4.36). HRMS m/z calcd for C39H39N100 [M+H]+ 663.33083, found 663.33044. Data for compound 8: m.p. 111-114 °C (methylcyclohexane); 1H and 13C NMR (CDCl3): See Table 2. IR (KBr): 691 (5Ar), 758 (5Ar), 1006 (5_CH), 1033 (5_CH), 1345 (5ch), (5ch), 1622 (vAr), 1732 (vC_O) :=871 (VCH), 2956(vch) cm-1. UV (EtOH), (log e): 206.6 (4.84), 252.3 (4.42). HRMS m/z calcd for C39H39N10O [M+H]+ 663.33083, found 663.33026. 3-((2'-(2-((2'-(1H-Tetrazol-5-yl)biphenyl-4-yl)methyl)-2H-tetrazol-5-yl)biphenyl-4-yl)methyl)-2-butyl-1,3-diazaspiro[4.4]non-1-en-4-one (7). A mixture of 9 (0.4 g, 0.44 mmol) and methanol (10 mL) was refluxed for 6 h. The mixture was evaporated and separated by CyclographTM (silica gel, from dichloromet-hane to dichloromethane-ethyl acetate 10 : 1) to provide 0.25 g (85%) of 8. 4. Conclusion 6. References Two principal impurities of irbesartan API prepared via its ^-trityl derivatives are described. The impurities were identified as isomeric N-1 and N-2 (2'-(1H-tetrazol-5-yl)biphenyl-4-yl)methyl derivatives of irbesartan. Both compounds were unambigouesly identified by NMR techniques. For assignments of proton and carbon NMR spectra, COSY, HSQC, and HMBC experiments were used. The differentiation of both impurities of irbesartan was performed using 1D NOESY experiments. In both cases proton H-22 was excitated. While for impurity 7 signals of protons 13, 19 and 24 were found, in the case of impurity 8 signals of aliphatic protons 6, 8, and 9 appear. Spectral characteristics (IR, UV, MS) of these compounds are also given. 5. Acknowledgements This work was supported by Zentiva Prague. The authors' thanks are also due to Mr. Tomas Pekarek for measuring and interpreting the UV and IR spectra. 1. C. A. Bernhart, P. M. Perreaut, B. P. Ferrari, Y. A. Muneaux, J.-L. A. Assens, J. Clement, F. Haudricourt, C. F. Muneaux, J. E. Taillades, M.-A. Vignal, J. Gougat, P. R. Guiraudou, C. A. Lacour, A. Roccon, C. F. Cazaubon, J.-C. Brelihre, G. Le Fur, D. Nisato, J. Med. Chem. 1993, 36, 3371-3380. 2. K. F. Croom, M. P. Curran, K. L. Goa, Drugs 2004 64, 999-1028. 3. C. Bernhard, J.-C. Breliere, J. Clement, D. Nisato, P. M. Perreaut, C. F. Muneaux, (Elf Sanofi) US 5 270 317; Chem. Ab-str. 1993, 119, 95560. 4. S. Chava, M. Bandari, K. S. Mathuresh, (Matrix Laboratories) WO 2005/122699; Chem. Abstr. 2005, 144, 88292. 5. S. Zupančič, A. Pečavar, R. Zupet, (Krka) WO 2006/073376; Chem. Abstr. 2006, 145, 124576. 6. C. V. Kavitha, S. L. Gaonkar, J. N. Chandra, S. Narendra, C. T. Sadashiva, K. S. Rangappa, Bioorg. Med. Chem. 2007, 15, 7391-7398. 7. S. Radl, J. Stach, O. Klecan, (Zentiva) WO 2005/021535; Chem. Abstr. 2005, 142, 298118. 8. B. Satyanarayana, Y. Anjaneyulu, P. Veerasomaiah, P. P. Reddy, Heterocycl. Commun. 2007, 13, 223-228. 9. V. V. Korrapati, P. Rao, R. Dandala, V. K. Handa, I. V. S. Rao, A. Rani, A. Naidu, Synth. Commun. 2007, 37, 2897-2905. 10. J. Havlicek, Z. Mandelova, R. Weisemann, I. Strelec, S. Radl, Collect. Czech. Chem. Commun. 2009, 77, 347. Povzetek Opisana je sinteza dveh glavnih nečistoč irbesartana prek njegovega ^-tritilnega derivata. Nečistoče so bile izolirane in nedvoumno identificirane z NMR spektroskopskimi tehnikami. Podana je tudi IR, UV in MS spektroskopska karakteri-zacija omenjenih spojin.