Synthesis and Physicochemical Characterization of the Process-Related Impurities of Eplerenone, an Antihypertensive Drug

Two unknown impurities were observed during the process development for multigram-scale synthesis of eplerenone (Inspra®). The new process-related impurities were identified and fully characterized as the corresponding (7β,11α,17α)-11-hydroxy- and (7α,11β,17α)-9,11-dichloroeplerenone derivatives 12a and 13. Seven other known but poorly described in the literature eplerenone impurities, including four impurities A, B, C and E listed in the European Pharmacopoeia 8.4 were also detected, identified and fully characterized. All these contaminants result from side reactions taking place on the steroid ring C of the starting 11α-hydroxy-7α-(methoxycarbonyl)-3-oxo-17α-pregn-4-ene-21,17-carbolactone (12) and the key intermediate (7α,17α)-9(11)-enester 7, including epimerization of the C-7 asymmetric center, oxidation, dehydration, chlorination and lactonization. The impurities were isolated and/or synthesized and fully characterized by infrared spectroscopy (IR), nuclear magnetic resonance spectroscopy (NMR) and high-resolution mass spectrometry/electrospray ionization (HRMS/ESI). Their 1H- and 13C-NMR signals were fully assigned. The molecular structures of the eight impurities, including the new (7β,11α,17α)-11-hydroxy- and (7α,11β,17α)-9,11-dichloroeplerenone related substances 12a and 13, were solved and refined using single-crystal X-ray diffraction (SCXRD). The full identification and characterization of these impurities should be useful for the quality control and the validation of the analytical methods in the manufacture of eplerenone.


Introduction
Eplerenone (2, Figure 1) is a cardiovascular drug indicated for the treatment of essential hypertension and congestive heart failure that, in contrast to its predecessor spironolactone 1 (Figure 1) demonstrates a high degree of selectivity for the aldosterone receptor and a low-binding affinity for progesterone and androgen receptors [1][2][3][4][5][6][7][8]. As a result of the presence of a 9,11-epoxide group in the eplerenone structure, its selectivity for the aldosterone receptor is enhanced and the drug minimizes the risk of adverse hormonal effects and provides important clinical benefits not previously available with spironolactone 1. Treatment with eplerenone is associated with reductions in blood pressure and improved survival (15% reduction in total mortality) for patients with heart failure who are in stable condition after a myocardial infarction. The product was originally developed by scientists at Eplerenone contains sensitive 9,11α-epoxide, 17α-γ-lactone and 7α-carbomethoxy moieties that, depending on the conditions, may hydrolyze or epimerize to afford degradation or epimerization products [9][10][11][12][13]. The SCXRD structure of eplerenone confirms the relative cis configuration of the 9α,11α-epoxide ring and the 7α-carbomethoxy substituent [14][15][16][17]. Manufacture of eplerenone is always accompanied by side reactions leading to the unwanted impurities that vary with the different starting materials and reaction conditions used. Most of the patents and literature information dealing with the preparation of eplerenone are based on the use of canrenone derivatives. As stereogenic centers in eplerenone precursors give rise to various process-related impurities, including diastereo-and regioisomers of starting materials, intermediates, by-products and the final drug substance, the manufacture of eplerenone with the required stereochemistry and pharmaceutical grade purity is a significant challenge. In designing a synthesis of eplerenone from canrenone derivatives, the principal challenges are the stereoselective introduction of the carbomethoxy substituent at the C-7α position of the steroid skeleton and the regioselective dehydration of the 11α-hydroxy group [1, 9,[16][17][18][19][20][21].
In 1984, Grob et al. [17,18] from Ciba-Geigy AG accomplished the first synthesis of eplerenone by employing Nagata hydrocyanation of Δ 9(11) -canrenone (3) as the key step (Scheme 1), but with moderate stereoselectivity in the 7α-cyano derivative 4 formation (7α/7β ≈ 4:1) necessitating tedious column chromatographic separations. The original EP 122232 B1 patent does not disclose any information with respect to the purity of the material obtained, its purification to pharmaceutical-grade or the removal of impurities. Scheme 1. The original eplerenone synthesis from Ciba-Geigy AG (Basel, Switzerland) [17,18].  Eplerenone contains sensitive 9,11α-epoxide, 17α-γ-lactone and 7α-carbomethoxy moieties that, depending on the conditions, may hydrolyze or epimerize to afford degradation or epimerization products [9][10][11][12][13]. The SCXRD structure of eplerenone confirms the relative cis configuration of the 9α,11α-epoxide ring and the 7α-carbomethoxy substituent [14][15][16][17]. Manufacture of eplerenone is always accompanied by side reactions leading to the unwanted impurities that vary with the different starting materials and reaction conditions used. Most of the patents and literature information dealing with the preparation of eplerenone are based on the use of canrenone derivatives. As stereogenic centers in eplerenone precursors give rise to various process-related impurities, including diastereoand regioisomers of starting materials, intermediates, by-products and the final drug substance, the manufacture of eplerenone with the required stereochemistry and pharmaceutical grade purity is a significant challenge. In designing a synthesis of eplerenone from canrenone derivatives, the principal challenges are the stereoselective introduction of the carbomethoxy substituent at the C-7α position of the steroid skeleton and the regioselective dehydration of the 11α-hydroxy group [1, 9,[16][17][18][19][20][21].
In 1984, Grob et al. [17,18] from Ciba-Geigy AG accomplished the first synthesis of eplerenone by employing Nagata hydrocyanation of ∆ 9(11) -canrenone (3) as the key step (Scheme 1), but with moderate stereoselectivity in the 7α-cyano derivative 4 formation (7α/7β ≈ 4:1) necessitating tedious column chromatographic separations. The original EP 122232 B1 patent does not disclose any information with respect to the purity of the material obtained, its purification to pharmaceutical-grade or the removal of impurities.  Eplerenone contains sensitive 9,11α-epoxide, 17α-γ-lactone and 7α-carbomethoxy moieties that, depending on the conditions, may hydrolyze or epimerize to afford degradation or epimerization products [9][10][11][12][13]. The SCXRD structure of eplerenone confirms the relative cis configuration of the 9α,11α-epoxide ring and the 7α-carbomethoxy substituent [14][15][16][17]. Manufacture of eplerenone is always accompanied by side reactions leading to the unwanted impurities that vary with the different starting materials and reaction conditions used. Most of the patents and literature information dealing with the preparation of eplerenone are based on the use of canrenone derivatives. As stereogenic centers in eplerenone precursors give rise to various process-related impurities, including diastereo-and regioisomers of starting materials, intermediates, by-products and the final drug substance, the manufacture of eplerenone with the required stereochemistry and pharmaceutical grade purity is a significant challenge. In designing a synthesis of eplerenone from canrenone derivatives, the principal challenges are the stereoselective introduction of the carbomethoxy substituent at the C-7α position of the steroid skeleton and the regioselective dehydration of the 11α-hydroxy group [1, 9,[16][17][18][19][20][21].
In 1984, Grob et al. [17,18] from Ciba-Geigy AG accomplished the first synthesis of eplerenone by employing Nagata hydrocyanation of Δ 9(11) -canrenone (3) as the key step (Scheme 1), but with moderate stereoselectivity in the 7α-cyano derivative 4 formation (7α/7β ≈ 4:1) necessitating tedious column chromatographic separations. The original EP 122232 B1 patent does not disclose any information with respect to the purity of the material obtained, its purification to pharmaceutical-grade or the removal of impurities.
Recently, we have developed an improved, scalable, cost-effective and environment-friendly technology for the industrial-scale synthesis of eplerenone (2) from commercially available 11α-hydroxy-7α-(methoxycarbonyl)-3-oxo-17α-pregn-4-ene-21,17-carbolactone (12), based on the last two steps of the general route described by Ng et al. (Scheme 2) [9]. During the process development, two new and seven known process-related impurities of eplerenone were observed, and/or synthesized and fully characterized, including four impurities A, B, C and E listed in the European Pharmacopoeia 8.4 [23].
Recently, we have developed an improved, scalable, cost-effective and environment-friendly technology for the industrial-scale synthesis of eplerenone (2) from commercially available 11α-hydroxy-7α-(methoxycarbonyl)-3-oxo-17α-pregn-4-ene-21,17-carbolactone (12), based on the last two steps of the general route described by Ng et al. (Scheme 2) [9]. During the process development, two new and seven known process-related impurities of eplerenone were observed, and/or synthesized and fully characterized, including four impurities A, B, C and E listed in the European Pharmacopoeia 8.4 [23].
The syntheses of the starting (7α,11α,17α)-11-hydroxyester 12 and the key intermediate (7α,17α)-9(11)-enester 7 are broadly described in the literature sources [9,[16][17][18][19][20][21]24]; however, their comprehensive structure elucidation and confirmation is still missing and SCXRD studies are reported here for the first time. As starting materials and intermediates in active pharmaceutical ingredient (API) synthesis often afford numerous impurities affecting the quality of the final drug product, their comprehensive structural elucidation and confirmation is essential for impurities identification and characterization. A complete physicochemical characterization, not only for an API, but also starting materials and key synthetic intermediates, became a requirement of both the U.S. Food and Drug Administration (FDA) and the European Medicine Agency (EMA). The five other impurities listed in Table 1, i.e., the isomeric (7β,17α)-9(11)-enester 7a and (7α,17α)-11(12)-enester 7b, the isomeric (7β,11α,17α)-9,11-epoxyester 2a (Imp. E) and (7α,11α,12α,17α)-11,12-epoxyester 2b (Imp. B) and the 7α,9:21,17-dicarbolactone 14 (Imp. A), were mentioned elsewhere in the literature, mainly as part of HPLC studies; however, their syntheses, methods of removal from the final product and comprehensive structural elucidation and confirmation were not disclosed [9,12]. The determination of a drug substance impurity profile, including starting materials, by-products, intermediates and potential degradation products, is critical for the safety assessment of API and manufacturing process thereof. In any API, it is necessary to study the impurity profile, and control it during the preparation of the pharmaceutical. As indicated in the ICH guidelines, any impurity, formed at a level of ≥0.10% with respect to the API, should be identified, synthesized and characterized thoroughly [25,26]. Only two eplerenone impurities, i.e., 7α,9:21,17-dicarbolactone 14 (Imp. A) and 11,12-epoxy ester 2b (Imp. B), are accepted at a level greater than 0.10%, i.e., maximum 0.3%, in accordance with pharmacopoeial acceptance criteria [22]. ingredient (API) synthesis often afford numerous impurities affecting the quality of the final drug product, their comprehensive structural elucidation and confirmation is essential for impurities identification and characterization. A complete physicochemical characterization, not only for an API, but also starting materials and key synthetic intermediates, became a requirement of both the U.S. Food and Drug Administration (FDA) and the European Medicine Agency (EMA). The five other impurities listed in Table 1, i.e., the isomeric (7β,17α)-9(11)-enester 7a and (7α,17α)-11(12)-enester 7b, the isomeric (7β,11α,17α)-9,11-epoxyester 2a (Imp. E) and (7α,11α,12α,17α)-11,12-epoxyester 2b (Imp. B) and the 7α,9:21,17-dicarbolactone 14 (Imp. A), were mentioned elsewhere in the literature, mainly as part of HPLC studies; however, their syntheses, methods of removal from the final product and comprehensive structural elucidation and confirmation were not disclosed [9,12]. The determination of a drug substance impurity profile, including starting materials, by-products, intermediates and potential degradation products, is critical for the safety assessment of API and manufacturing process thereof. In any API, it is necessary to study the impurity profile, and control it during the preparation of the pharmaceutical. As indicated in the ICH guidelines, any impurity, formed at a level of ≥0.10% with respect to the API, should be identified, synthesized and characterized thoroughly [25,26]. Only two eplerenone impurities, i.e., 7α,9:21,17-dicarbolactone 14 (Imp. A) and 11,12-epoxy ester 2b (Imp. B), are accepted at a level greater than 0.10%, i.e., maximum 0.3%, in accordance with pharmacopoeial acceptance criteria [22]. Table 1. Structures of the process-related impurities of eplerenone (2).

Results and Discussion
Dehydration of 11-hydroxy steroids is commonly used for the introduction of the 9,11-double bond into the steroid skeleton. In a preferred technological embodiment, the (7α,17α)-9(11)-enester 7 is synthesized directly by in situ replacement of the 11α-hydroxy group of the ester (7α,11α,17α)-12 with halogen followed by thermal 9,11-dehydrohalogenation. The nucleophilic substitution of the 11α-hydroxy group is effected by reaction with sulfuryl halide at about −70 °C, after which a hydrogen halide scavenger is added. In manufacturing technology of eplerenone developed by the authors, the 11α-hydroxyester 12 and imidazole were dissolved simultaneously in anhydrous tetrahydrofuran and cooled to −10 °C. Sulfuryl chloride was added slowly and the reaction mixture was allowed to warm to room temperature, and then stirred for a time sufficient to complete the elimination reaction, typically about 1 h. The key intermediate 7 was isolated in a crude form by dichloromethane extraction followed by evaporation of the solvent and recrystallized twice, from 1 Impurities listed in the European Pharmacopoeia 8.4 [23].

Results and Discussion
Dehydration of 11-hydroxy steroids is commonly used for the introduction of the 9,11-double bond into the steroid skeleton. In a preferred technological embodiment, the (7α,17α)-9(11)-enester 7 is synthesized directly by in situ replacement of the 11α-hydroxy group of the ester (7α,11α,17α)-12 with halogen followed by thermal 9,11-dehydrohalogenation. The nucleophilic substitution of the 11α-hydroxy group is effected by reaction with sulfuryl halide at about −70 °C, after which a hydrogen halide scavenger is added. In manufacturing technology of eplerenone developed by the authors, the 11α-hydroxyester 12 and imidazole were dissolved simultaneously in anhydrous tetrahydrofuran and cooled to −10 °C. Sulfuryl chloride was added slowly and the reaction mixture was allowed to warm to room temperature, and then stirred for a time sufficient to complete the elimination reaction, typically about 1 h. The key intermediate 7 was isolated in a crude form by dichloromethane extraction followed by evaporation of the solvent and recrystallized twice, from 1 Impurities listed in the European Pharmacopoeia 8.4 [23].

Results and Discussion
Dehydration of 11-hydroxy steroids is commonly used for the introduction of the 9,11-double bond into the steroid skeleton. In a preferred technological embodiment, the (7α,17α)-9(11)-enester 7 is synthesized directly by in situ replacement of the 11α-hydroxy group of the ester (7α,11α,17α)-12 with halogen followed by thermal 9,11-dehydrohalogenation. The nucleophilic substitution of the 11α-hydroxy group is effected by reaction with sulfuryl halide at about −70 °C, after which a hydrogen halide scavenger is added. In manufacturing technology of eplerenone developed by the authors, the 11α-hydroxyester 12 and imidazole were dissolved simultaneously in anhydrous tetrahydrofuran and cooled to −10 °C. Sulfuryl chloride was added slowly and the reaction mixture was allowed to warm to room temperature, and then stirred for a time sufficient to complete the elimination reaction, typically about 1 h. The key intermediate 7 was isolated in a crude form by dichloromethane extraction followed by evaporation of the solvent and recrystallized twice, from 11α,12α-epoxy-7α-(methoxycarbonyl)-3-oxo-17α-pregn-4-ene-21,17-carbolactone 2b (Imp. B) 1 1 Impurities listed in the European Pharmacopoeia 8.4 [23].

Results and Discussion
Dehydration of 11-hydroxy steroids is commonly used for the introduction of the 9,11-double bond into the steroid skeleton. In a preferred technological embodiment, the (7α,17α)-9(11)-enester 7 is synthesized directly by in situ replacement of the 11α-hydroxy group of the ester (7α,11α,17α)-12 with halogen followed by thermal 9,11-dehydrohalogenation. The nucleophilic substitution of the 11α-hydroxy group is effected by reaction with sulfuryl halide at about −70 • C, after which a hydrogen halide scavenger is added. In manufacturing technology of eplerenone developed by the authors, the 11α-hydroxyester 12 and imidazole were dissolved simultaneously in anhydrous tetrahydrofuran and cooled to −10 • C. Sulfuryl chloride was added slowly and the reaction mixture was allowed to warm to room temperature, and then stirred for a time sufficient to complete the elimination reaction, typically about 1 h. The key intermediate 7 was isolated in a crude form by dichloromethane extraction followed by evaporation of the solvent and recrystallized twice, from ethanol and a mixture of dichloromethane/diethyl ether respectively, to give the pure (7α,17α)-9(11)-enester 7 (71% yield) as white crystals.
The [M + Na] + values, m/z 439.2110 and 439.2080, obtained for both 11-hydroxyesters 12 and 12a correspond to C 24 H 32 O 6 Na. The NMR data for epimeric hydroxyesters 12 and 12a are given in Table 2. The 1 H-/ 13 C-NMR chemical shifts assignment was made after careful and precise 2D NMR spectra analysis and the data obtained for the known isomer 12 are in full compliance with those presented by Preisig et al. [24]. The 2D NOESY experiments allowed discrimination between epimeric structures 12 and 12a and fully confirmed the stereochemistry at the C-7 atom in both isomers. Blue arrows in Figure 2 show the most important NOE effects involving H7 proton, simultaneously clearly indicating the 7α or 7β positioned carbomethoxy group. The strong H7-H8 NOE effect is observed for the compound 12 with β positioned H7 proton, whereas α situated H7 proton in 12a is involved in two significant H9-H7 and H7-H14 NOE effects. Additionally, the strong NOE effect between H9 and H14 protons for both 12 and 12a epimers is observed. Comparison of the NMR data for epimeric structures 12 and 12a revealed that some of the 1 H/ 13 C nuclei shieldings within the steroid rings B, C and D are related to the α or β configuration of the C-7 atom. The 1 H shielding increase of 0.5 ppm for H7, 0.71 ppm for H9 and 0.28 ppm for H14 nuclei is observed when passing from the compound 12 with 7α positioned carbomethoxy group to its 7β epimer 12a. Additionally, the change of configuration at the C7 is accompanied by H8 (0.35 ppm) shielding decrease. Simultaneously, both diasterotopic H15 protons of the epimer 12a with 7β positioned carbomethoxy group became equal having the same proton chemical shift (1.49 ppm). In the case of 13 C-NMR data the opposite effect is observed, the transition from 12 to 12a leads to the shielding decrease of 6.6 ppm for C7, 4.8 ppm for C9 and 3.4 ppm for C14 nuclei. Surprisingly, the change of configuration from 7α in 12 to 7β in 12a is related with no change of the shielding of the C8 nucleus and decreasing of shielding by 2.7 ppm for C23 carbon of the carbomethoxy substituent.  The NMR structure assignments of the isomeric 11-hydroxyesters (7α,11α,17α)-12 and (7β,11α,17α)-12a were confirmed by X-ray analysis ( Figure 3 and Table 3). The SCXRD structures confirmed the relative cis-and trans-configurations of the 11α-hydroxy and 7α-carbomethoxy substituents in 12 and 12a, respectively. The carbomethoxy group is situated at the C-7α position in 12 and the C-7β position in 12a, whereas the hydroxy group adopts the C-11α position in both 12 and 12a isomers. The dehydration of 11α-hydroxy steroids leads predominantly to the formation of the double bond between C-9 and C-11 carbons of the steroid skeleton. As expected, the (7α,17α)-9(11)-enester 7 was the main product of the 11-hydroxyester 12 dehydration with sulfuryl  The NMR structure assignments of the isomeric 11-hydroxyesters (7α,11α,17α)-12 and (7β,11α,17α)-12a were confirmed by X-ray analysis ( Figure 3 and Table 3). The SCXRD structures confirmed the relative cisand trans-configurations of the 11α-hydroxy and 7α-carbomethoxy substituents in 12 and 12a, respectively. The carbomethoxy group is situated at the C-7α position in 12 and the C-7β position in 12a, whereas the hydroxy group adopts the C-11α position in both 12 and 12a isomers. The dehydration of 11α-hydroxy steroids leads predominantly to the formation of the double bond between C-9 and C-11 carbons of the steroid skeleton. As expected, the (7α,17α)-9(11)-enester 7 was the main product of the 11-hydroxyester 12 dehydration with sulfuryl chloride in the presence of imidazole; however, a considerable amount of the regioiosmeric (7α,17α)-11(12)-enester 7b, as a result of competitive 11,12-dehydration of 12, was also isolated from the recrystallization mother liquors of 7.
The [M + Na] + values, m/z 421.1986, 421.2006, and 421.1988, obtained for the free isomeric enesters 7, 7a and 7b correspond to C 24 H 30 O 5 Na. The full NMR data confirming the structure of the known enester 7 were reported by Grob et al. [17]; however several 1 H and 13 C signals were assigned reversibly. The proper 1 H/ 13 C chemical shift assignment for isomeric enesters 7, 7a and 7b was based on the careful analysis of 2D NMR experiments and is given in Table 4. Similarly to the epimeric hydroxyesters 12 and 12a, the NOE effects involving H7 proton allowed distinction between 7α (7 and 7b) and 7β (7a) diastereoisomers ( Figure 4). The strong H7-H8 NOE effect was noted for isomers 7 and 7b with β positioned H7 proton, whereas α situated H7 in 7a is involved in H7-H14 NOE effect disturbed by H7-H15 interaction. Additionally, the significant H9-H14 NOE effect for 7b regioisomer was observed, while H8 proton of the epimers 7 and 7b is involved in H8-H6 interaction. Similarly to the epimeric hydroxyesters 12 and 12a, some of the 1 H/ 13 C nuclei shieldings within the steroid rings B, C and D are related to the α or β configuration of the C-7 atom. The 1 H shielding increase of 0.7 ppm for H7 and 0.22 ppm for H8 nuclei was observed when passing from the compound 7 with 7α positioned carbomethoxy group to its 7β epimer 7a. Similarly to 12a, both diasterotopic H15 protons of the epimer 7a with 7β positioned carbomethoxy group became equal having the same proton chemical shift (1.54 ppm). In the case of 13 C-NMR the opposite effect is observed, the transition from 7 to 7a leads to the shielding decrease of 5.9 ppm for C7 and 2.4 ppm for C14 nuclei. The change of configuration from 7α in 7 to 7β in 7a is related with no change of the shielding of the C8 nucleus and decreasing of shielding by 2.3 ppm for C23 carbon of carbomethoxy substituent. The dehydration of the epimeric 11-hydroxyesters 12/12a to the corresponding 9(11)-enesters 7/7a caused the strong 1 H shielding decrease of 89.7 ppm/84.7 ppm for C9 and 49.8 ppm/51.9 ppm for C11 carbons, respectively. The transition from 12/12a to 7/7a also resulted in medium shielding increase of 10.5/10.8 ppm for C12 and shielding decrease of 8.8/7.3 ppm for C19 and 3.4/3.4 ppm for C8, whereas for C10 (1.1/1.3 ppm), C13 (1.7/1.9 ppm) and C14 (2.2/3.2 ppm) a weak increasing effect was observed. The competitive dehydration of 12 to the regioisomeric 11,12-enester 7b resulted in the strong 1 H shielding decrease of 57.3 ppm for C11 and 90.4 ppm for C12 nuclei. Similarly to the hydroxyesters 12 and 12a, the change of configuration from 7α in 7 to 7β in 7a is related with no change of the shielding of the C8 nucleus and decreasing of shielding by 2.3 ppm for C23 carbon of the carbomethoxy substituent. Additionally, the shielding increase of 4.5 ppm for C8 carbon was observed when passing from epimeric 9(11)-enesters 7 and 7a to regioisomeric 11-enester 7b.  and 421.1988, obtained for the free isomeric enesters 7, 7a and 7b correspond to C24H30O5Na. The full NMR data confirming the structure of the known enester 7 were reported by Grob et al. [17]; however several 1 H and 13 C signals were assigned reversibly. The proper 1 H/ 13 C chemical shift assignment for isomeric enesters 7, 7a and 7b was based on the careful analysis of 2D NMR experiments and is given in Table 4. Similarly to the epimeric hydroxyesters 12 and 12a, the NOE effects involving H7 proton allowed distinction between 7α (7 and 7b) and 7β (7a) diastereoisomers (Figure 4). The strong H7-H8 NOE effect was noted for isomers 7 and 7b with β positioned H7 proton, whereas α situated H7 in 7a is involved in H7-H14 NOE effect disturbed by H7-H15 interaction. Additionally, the significant H9-H14 NOE effect for 7b regioisomer was observed, while H8 proton of the epimers 7 and 7b is involved in H8-H6 interaction. Similarly to the epimeric hydroxyesters 12 and 12a, some of the 1 H/ 13 C nuclei shieldings within the steroid rings B, C and D are related to the α or β configuration of the C-7 atom. The 1 H shielding increase of 0.7 ppm for H7 and 0.22 ppm for H8 nuclei was observed when passing from the compound 7 with 7α positioned carbomethoxy group to its 7β epimer 7a. Similarly to 12a, both diasterotopic H15 protons of the epimer 7a with 7β positioned carbomethoxy group became equal having the same proton chemical shift (1.54 ppm). In the case of 13 C-NMR the opposite effect is observed, the transition from 7 to 7a leads to the shielding decrease of 5.9 ppm for C7 and 2.4 ppm for C14 nuclei. The change of configuration from 7α in 7 to 7β in 7a is related with no change of the shielding of the C8 nucleus and decreasing of shielding by 2.3 ppm for C23 carbon of carbomethoxy substituent. The dehydration of the epimeric 11-hydroxyesters 12/12a to the corresponding 9(11)-enesters 7/7a caused the strong 1 H shielding decrease of 89.7 ppm/84.7 ppm for C9 and 49.8 ppm/51.9 ppm for C11 carbons, respectively. The transition from 12/12a to 7/7a also resulted in medium shielding increase of 10.5/10.8 ppm for C12 and shielding decrease of 8.8/7.3 ppm for C19 and 3.4/3.4 ppm for C8, whereas for C10 (1.1/1.3 ppm), C13 (1.7/1.9 ppm) and C14 (2.2/3.2 ppm) a weak increasing effect was observed. The competitive dehydration of 12 to the regioisomeric 11,12-enester 7b resulted in the strong 1 H shielding decrease of 57.3 ppm for C11 and 90.4 ppm for C12 nuclei. Similarly to the hydroxyesters 12 and 12a, the change of configuration from 7α in 7 to 7β in 7a is related with no change of the shielding of the C8 nucleus and decreasing of shielding by 2.3 ppm for C23 carbon of the carbomethoxy substituent. Additionally, the shielding increase of 4.5 ppm for C8 carbon was observed when passing from epimeric 9(11)-enesters 7 and 7a to regioisomeric 11-enester 7b.    The NMR structure assignments of the isomeric (7α,17α)-9(11)-enester 7, (7β,17α)-9(11)-enester 7a and (7α,17α)-11(12)-enester 7b were confirmed by X-ray analysis ( Figure 5 and Tables 5 and 6). The SCXRD structures confirmed the presence of the 9,11-double bond in 7 and 7a and the 11,12-double bond in 7b. The carbomethoxy group is situated at the C-7α position in 7 and 7b and at the C-7β position in 7a.   13 C-NMR spectral data for esters (7α,17α)-9(11)-ene 7, (7β,17α)-9(11)-ene 7a and (7α,17α)-11-ene 7b. The NMR structure assignments of the isomeric (7α,17α)-9(11)-enester 7, (7β,17α)-9(11)-enester 7a and (7α,17α)-11(12)-enester 7b were confirmed by X-ray analysis ( Figure 5 and Tables 5 and 6). The SCXRD structures confirmed the presence of the 9,11-double bond in 7 and 7a and the 11,12-double bond in 7b. The carbomethoxy group is situated at the C-7α position in 7 and 7b and at the C-7β position in 7a.      Sulfuryl chloride used for dehydration of 11α-hydroxy steroids is also known as a chlorinating agent and, under some conditions, applied for chlorination of steroid double bonds [27,28]. Indeed, the novel (7α,11β,17α)-9,11-dichloro derivative 13, formed as a result of competitive chlorine addition to the newly formed 9,11-double bond of the key intermediate (7α,17α)-9(11)-enester 7, was isolated chromatographically from the recrystallization mother liquors of 7. It was also synthesized by chlorination of the pure enester 7 with sulfuryl chloride in the presence of pyridine in chlorobenzene, and then purified by column chromatography with 5-30% EtOAc/CH 2 Cl 2 gradient elution to give the pure (7α,11β,17α)-9,11-dichloro derivative 13 (41% yield) as a white solid. The novel 9,11-dichloro impurity 13 was formed by the nucleophilic attack of a chloride anion on the intermediate chloronium cation, a structural analogue of the eplerenone epoxide ring. No other isomers of 13 were obtained.

H-NMR (1) 13 C-NMR
The [M + Na] + value, m/z 491.1386 obtained for the 9,11-dichloroderivative 13 corresponds to C 24 H 30 O 5 NaCl 2 . The stereochemistry at the C9 and C11 carbons of the dichloro derivative 13 was established on the basis of 1D/2D NOESY experiments, which indicated the relative trans configuration of the C-9α and C-11β positioned chlorine atoms. The addition of chlorine to the 9,11-double bond of enester 7 caused significant changes in the 1 H/ 13 C nuclei shieldings within the steroid rings A, B, C and D of the dichloro derivative 13 (Table 7). The strong shielding increase of 55.7 ppm for C9 and 59.5 ppm for C11 nuclei was observed when passing from 7 to 13. Minor shielding increase of 4.1 ppm for C1, 2.9 ppm for C5, 3.3 ppm for C6, 2.9 ppm for C7 and 1.8 ppm for C14 and shielding decrease of 6.6 ppm for C10 nuclei were also noted. The introduction of chlorine into C9 and C11 positions of 7 also resulted in shielding increase of 0.97 ppm for H11 and shielding decrease of 0.18 ppm for H4, 0.42 ppm for H8 and 0.97 for H14 nuclei in dichloro derivative 13. Noteworthy, the diasterotopic effect observed for H15 protons of 7 (0.4 ppm) increased to 0.75 ppm for the same protons in 13. Similarly to the other compounds with 7α-carbomethoxy group, the β positioned H7 proton in dichloro derivative 13 is involved in strong H7-H8 NOE effect ( Figure 6). The NMR structure assignment of the novel (7α,11β,17α)-9,11-dichloro derivative 13 was confirmed by X-ray analysis (Figure 7 and Table 5). The SCXRD structure confirmed the presence of the two chlorine atoms at the C-9α and C-11β positions of the steroid ring C in the relative trans configuration and the C-7α positioned carbomethoxy substituent. Similarly to the other compounds with 7α-carbomethoxy group, the β positioned H7 proton in dichloro derivative 13 is involved in strong H7-H8 NOE effect ( Figure 6). The NMR structure assignment of the novel (7α,11β,17α)-9,11-dichloro derivative 13 was confirmed by X-ray analysis (Figure 7 and Table 5). The SCXRD structure confirmed the presence of the two chlorine atoms at the C-9α and C-11β positions of the steroid ring C in the relative trans configuration and the C-7α positioned carbomethoxy substituent. The acidic conditions applied for the dehydration of the 11α-hydroxy group in 12 resulted in competitive lactonization between the hydroxyl and carbomethoxy groups and lead to 7α,9:21,17-dicarbolactone 14 (Imp. A), which tends to be formed even in the presence of traces of free water. The dicarbolactone 14 was isolated chromatographically from the recrystallization mother liquors of 7. It was also synthesized independently from the mesylate of the starting 11α-hydroxyester 12 by reaction with acetic acid in the presence of sodium acetate, and then purified by column chromatography with 5-30% ethyl acetate/dichloromethane gradient elution to afford the pure 7α,9:21,17-dicarbolactone 14 (83% yield) as a white solid.
The [M + Na] + value, m/z 407.1847, obtained for the dicarbolactone 14 corresponds to C23H28O5Na. The 1 H-/ 13 C-NMR chemical shifts assignment for dilactone 14 is presented in Table 7. The detailed analysis of 2D NMR spectra, especially 1 H-13 C HMBC and NOESY experiments, unambiguously confirmed the structure consisting of the two γ-lactone rings. The second γ-lactone moiety results from internal lactonization between 9α-hydroxy and 7α-carbomethoxy groups of Similarly to the other compounds with 7α-carbomethoxy group, the β positioned H7 proton in dichloro derivative 13 is involved in strong H7-H8 NOE effect ( Figure 6). The NMR structure assignment of the novel (7α,11β,17α)-9,11-dichloro derivative 13 was confirmed by X-ray analysis (Figure 7 and Table 5). The SCXRD structure confirmed the presence of the two chlorine atoms at the C-9α and C-11β positions of the steroid ring C in the relative trans configuration and the C-7α positioned carbomethoxy substituent. The acidic conditions applied for the dehydration of the 11α-hydroxy group in 12 resulted in competitive lactonization between the hydroxyl and carbomethoxy groups and lead to 7α,9:21,17-dicarbolactone 14 (Imp. A), which tends to be formed even in the presence of traces of free water. The dicarbolactone 14 was isolated chromatographically from the recrystallization mother liquors of 7. It was also synthesized independently from the mesylate of the starting 11α-hydroxyester 12 by reaction with acetic acid in the presence of sodium acetate, and then purified by column chromatography with 5-30% ethyl acetate/dichloromethane gradient elution to afford the pure 7α,9:21,17-dicarbolactone 14 (83% yield) as a white solid.
The [M + Na] + value, m/z 407.1847, obtained for the dicarbolactone 14 corresponds to C23H28O5Na. The 1 H-/ 13 C-NMR chemical shifts assignment for dilactone 14 is presented in Table 7. The detailed analysis of 2D NMR spectra, especially 1 H-13 C HMBC and NOESY experiments, unambiguously confirmed the structure consisting of the two γ-lactone rings. The second γ-lactone moiety results from internal lactonization between 9α-hydroxy and 7α-carbomethoxy groups of The acidic conditions applied for the dehydration of the 11α-hydroxy group in 12 resulted in competitive lactonization between the hydroxyl and carbomethoxy groups and lead to 7α,9:21,17-dicarbolactone 14 (Imp. A), which tends to be formed even in the presence of traces of free water. The dicarbolactone 14 was isolated chromatographically from the recrystallization mother liquors of 7. It was also synthesized independently from the mesylate of the starting 11α-hydroxyester 12 by reaction with acetic acid in the presence of sodium acetate, and then purified by column chromatography with 5-30% ethyl acetate/dichloromethane gradient elution to afford the pure 7α,9:21,17-dicarbolactone 14 (83% yield) as a white solid.
The [M + Na] + value, m/z 407.1847, obtained for the dicarbolactone 14 corresponds to C 23 H 28 O 5 Na. The 1 H-/ 13 C-NMR chemical shifts assignment for dilactone 14 is presented in Table 7. The detailed analysis of 2D NMR spectra, especially 1 H-13 C HMBC and NOESY experiments, unambiguously confirmed the structure consisting of the two γ-lactone rings. The second γ-lactone moiety results from internal lactonization between 9α-hydroxy and 7α-carbomethoxy groups of intermediate ester, formed by water addition to the 9,11-double bond of enester 7 during the dehydration step of 12. This significant change in the steroid structure entails numerous changes in the 1 H/ 13 C nuclei shielding. The shielding increase of 46.8 ppm for C11 and 16.5 ppm for C12 and decrease of 37.4 ppm for C9 nuclei were observed when passing from hydroxyester 12 to the dicarbolactone 14. Minor shielding increase of 7.7 ppm for C1 and 6.2 ppm for C5 was also noted, whereas for C8 shielding decrease of 8.1 ppm was observed. Similarly to the other compounds of the series with 7α positioned carbomethoxy group, the H7 proton of the dicarbolactone 14 is involved in two strong H7-H8 and H6-H7 NOE effects ( Figure 6).
The NMR data of eplerenone were presented in literature sources many times; however the full and correct 1 H/ 13 C chemical shifts assignment confirming its structure has never been done. Although the complete assignment of 1 H-/ 13 C-NMR signals was presented by Grob et al. [17], some of the signals were assigned reversibly. The careful analysis of 2D NMR experiments, including HSQC, HMBC and NOESY measurements, allowed to assign signals to the corresponding protons and carbons of the isomeric epoxy esters 2, 2a and 2b unambiguously, and thus confirm the structure of eplerenone and its isomers (Table 8). Blue arrows in Figure 8 show the most important NOE effects involving H7 proton, simultaneously indicating the 7α or 7β positioned carbomethoxy group. The strong H7-H8 NOE effect is observed for the compounds 2 and 2b with β positioned H7 proton, whereas α situated H7 proton in 2a is involved in the strong interaction with H14. The epoxidation of the 9,11-double bond caused several shielding effects observed for 1 H/ 13 C nuclei within the steroid rings A, B and C. The strong shielding increase of 77.1/75 ppm for C9, 67.6/66.8 ppm for C11 carbons and 2.5/2.4 ppm for H11 protons was observed when passing from enesters 7/7a to epoxides 2/2a, respectively. Minor shielding increase of 7/6.5 ppm for C1, 5.9/6.1 ppm for C14, 2.6/3.4 for C7, 1.8/1.6 ppm for C8, 2/2 ppm for C12 and 0.7/0.4 for C13 carbons was also noted. The transition from enesters 7/7a to epoxides 2/2a caused shielding increase of 0.76/0.66 ppm for one of the H1 protons and 0.36/0.31 ppm for one of the H12 protons, whereas for H-14 protons shielding decrease of 0.4/0.14 ppm was noted. Similarly to the epimeric hydroxyesters 12/12a and enesters 7/7a, some of the 1 H/ 13 C nuclei shielding within the steroid rings B, C and D are related to the α or β configuration of the C7 atom. The 1 H shielding increase of 0.22 ppm for H7 and 0.25 ppm for H14 nuclei is observed when passing from the compound 2 with 7α positioned carbomethoxy group to its 7β epimer 2a. Additionally, the change of configuration at the C7 is accompanied by weak shielding decrease of 0.1 ppm for H8. Simultaneously, both diasterotopic H15 protons of the epimer 2a with 7β positioned carbomethoxy group became equal having the same proton chemical shift (1.51 ppm). In the case of 13 C-NMR data the opposite effect is observed, the transition from 2 to 2a leads to the shielding decrease of 5.1 ppm for C7, 1.9 ppm for C9 and 2.2 ppm for C14 nuclei. The change of configuration from 7α in 2 to 7β in 2a is also related with shielding decrease of 2.1 ppm for C23 carbon of the carbomethoxy substituent. The competitive epoxidation of the 11,12-enester 7b to the 11,12-epoxyester 2b resulted in the strong shielding increase of 76.1 ppm for C11 and 77.1 ppm for C12 nuclei, whereas for C9 (3.3 ppm), C18 (4.4 ppm) and C14 (6.7 ppm) medium effects were observed. The strong shielding increase of 2.57 and 2.84 ppm was noted for H11 and H12 protons, respectively. Minor changes in increasing of shielding for H7 (0.1 ppm), H8 (0.24 ppm) and H9 (0.4 ppm) protons were also observed. Similarly to the enesters 7, 7a and 7b, the shielding increase of 3.6 ppm and 3.8 ppm for C8 carbon was noted when passing from epimeric 9,11-epoxides 2 and 2a to regioisomeric 11,12-epoxyester 2b. Table 8. 1 H-and 13 C-NMR spectral data for (7α,11α,17α)-9,11-epoxyester 2, (7β,11α,17α)-9,11-epoxyester 2a and (7α,11α,12α,17α)  The [M + Na] + values, m/z 437. 1941 and 437.1942, obtained for the two isomeric epoxyesters 2 and 2b correspond to C24H30O6Na. The [M + H] + value, m/z 415.2113, obtained for the third isomeric epoxyester 2a corresponds to C24H31O6. The NMR structure assignments of the isomeric (7β,11α,17α)-9,11-epoxyester 2a and (7α,11α,12α,17α)-11,12-epoxyester 2b were confirmed by X-ray analysis ( Figure 9 and Table 9). The SCXRD structures confirmed the presence of the 9α,11α-epoxide ring in 2a and 11α,12α-epoxide ring in 2b. The carbomethoxy group is situated at the C-7α position in 2b and at the C-7β position in 2a.

Optical Rotation
Optical rotations were measured with a Perkin Elmer 341 automatic polarimeter (Perkin Elmer, Norwalk, CT, USA) in CH 2 Cl 2 solutions as the solvents with percent concentrations.

Melting Point
Melting points were determined with a MEL-TEMP II capillary melting point apparatus (Laboratory Devices, Holliston, MA, USA).

FT-IR Spectroscopy
FT-IR spectra were taken for KBr pellets on a Nicolet Impact 410 FT-IR spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

NMR Spectroscopy
The NMR spectra of all the compounds were measured in CDCl 3 solutions with a Varian VNMRS-600 (600 MHz for 1 H-NMR and 150 MHz for 13 C-NMR; Varian Inc., Palo Alto, CA, USA) at temperature 298 K using TMS as internal standard. All 1 H-/ 13 C-NMR resonance signals were assigned using results of 2D experiments [g-COSY ( 1 H-1 H), g-HSQC ( 1 H-13 C) and g-HMBC ( 1 H-13 C)] in gradient versions. The 1 H-NMR chemical shifts were determined as centres of the correlation spots in the 1 H domain of the 2D 1 H-13 C HSQC experiments. Relative configuration and stereochemistry at the C-7 and other atoms was established on the basis of 1D and 2D NOESY spectra. Concentration of all solutions used for measurements was about 20-30 mg of compounds in 0.6 mL of solvent.

Method 2
Imidazole (1.31 g, 19.20 mmol) was added to a solution of 11-hydroxyester (7β,11α,17α)-12a (2.0 g, 4.80 mmol) in THF (25 mL) and cooled to −20 • C. SO 2 Cl 2 (0.8 mL, 10.08 mmol) was added dropwise and the mixture was stirred for 30 min. at −20 • C. The cooling bath was then removed and the stirring was continued for 1 h at room temperature. The reaction mixture was diluted with H 2 O (10 mL) and the product was extracted with CH 2 Cl 2 (3 × 25 mL).

Method 1
Eplerenone (3.1 g, 7.48 mmol) was added to a suspension of MeONa (1.41 g, 26.17 mmol) in anhydrous MeOH (25 mL) and then refluxed for 24 h. The mixture was cooled to room temperature and 3 M aqueous HCl solution (150 mL) was added. The product was extracted with CH 2 Cl 2 (3 × 25 mL).
The combined organic phases were dried over anhydrous Na2SO4, filtered and concentrated under reduced pressure to give a yellowish white solid. The crude mixture of epimers 2 and 2a was The combined organic phases were dried over anhydrous Na 2 SO 4 , filtered and concentrated under reduced pressure to give a yellowish white solid. The crude mixture of epimers 2 and 2a was purified by column chromatography over silica gel with 1-10% Me 2 CO/CH 2 Cl 2 gradient elution to give (7β,11α,17α)-9,11-epoxyester 2a (1.52 g, 49% yield) as a white solid, a mixture of 2 and 2a (0.55 g) and (7α,11α,17α)-9,11-epoxyester 2 (0.77 g).
IR, NMR, HRMS/ESI and SCXRD. All the impurities resulted from side reactions taking place on the steroid rings B and C of the starting (7α,11α,17α)-11-hydroxyester 12 and the key intermediate (7α,17α)-9(11)-enester 7, including epimerization of the C-7 asymmetric center, oxidation, dehydration, chlorination and lactonization. The full identification and characterization of the impurities should be useful for the quality control and the validation of the analytical methods in the manufacture of eplerenone.