Synthesis of European pharmacopoeial impurities A , B , C , and D of cabergoline †

For the use of analytics, European pharmacopoeial impurities A, B, C, and D of cabergoline were synthesized. Ergocryptine was chosen as a starting material and synthesis was accomplished via two approaches, different in length and stereochemical outcome. A longer, indirect approach was realized through otherwise problematic oxidations of the 9,10-dihidrolysergol derivative, to the corresponding aldehyde and carboxylic acid. This was achieved by the use of activated DMSO and a Pinnick oxidation sequence. All four synthesized impurities are used as analytical standards in cabergoline manufacturing processes.


Introduction
Cabergoline ( 1) is a natural product-based drug, representing a complex, branched amide of 6-N-allylated 9,10-dihydrolysergic acid (cabergolinic acid) A (Fig. 1).Primarily acting as a dopaminergic D 2 receptor agonist, cabergoline is most widely used for treatment of hyperprolactinaemic disorders and both early and advanced Parkinson's disease. 1 The structure of cabergoline also encodes for different biogenic amines than dopamine, therefore displaying modest pharmacodynamic properties towards adrenergic and serotonergic receptors. 1ne of the fundamental roles of the pharmaceutical industry is to develop new drugs that are safe, effective and of high quality when they reach the patients.With respect to this, delivering an impurity prole of an active pharmaceutical ingredient (API) is a must for fullling the aforementioned criteria. 2For the purpose of qualifying and/or quantifying the impurity prole of cabergoline (1) as the API, we have synthesized four known impurities, qualied by European Pharmacopoeia (Ph.Eur.) 3 as impurities A, B, C and D (Fig. 1).
The genesis of these impurities can be found in the syntheses and other manufacturing process activities of cabergoline (Scheme 1). 4 Since cabergoline (1), with the maximum weekly dosage of 2.0 mg, falls into low dosing drugs, the threshold for identication and qualication of impurities in API is usually 0.10% and 0.15% of mass, respectively. 2These limits are low and efficient production processes should not exceed them.Therefore production syntheses are usually neither suitable nor cost Scheme 1 Possible routes of cabergoline impurities formation.Some impurities (A, D) can represent syntheses intermediates or are formed via hydrolysis of amide bonds from cabergoline (1) or other impurities (B, C). effective means to produce impurities for the use of analytics.Oen, new synthesis routes have to be designed to obtain desired impurities and this was also our task.Synthesis routes for Ph.Eur.impurities A, B, C and D of cabergoline (1) were designed and results are presented in this paper.

Results and discussion
9,10-Dihydrolysergic acid (2) was selected as one of the key synthetic intermediates.Although there are several total syntheses of lysergic acid available, through which 9,10-dihydrolysergic acid (2) could be prepared, a-ergocryptine (3) was chosen as starting material.The reason is that a-ergocryptine (3) is biosynthetically produced at our production site and way of preparing 9,10-dihydrolysergic acid (2) from a-ergocryptine (3) is advantageous over total synthesis because it is shorter, easier and more economical.Also, there is a longstanding tradition of ergot alkaloids chemistry within Sandoz company, pioneered by Albert Hofmann.
Starting with a-ergocryptine (3), the double bond between C9-C10 was diastereoselectively hydrogenated and the reduced ergopeptine 4 was cleaved by alkaline hydrolysis to give 9,10dihydrolysergic acid (2). 5 To increase solubility of 9,10-dihydrolysergic acid (2) in organic solvents, the corresponding methyl ester 5 was prepared (Scheme 2). 6n the next steps allylic substituent on 6-N of 5 had to be introduced.To introduce the allylic substituent on 6-N of 5, 9,10-dihydrolysergic acid (2) had to be N-demethylated rst.2,2,2-Trichloroethyl chloroformate (TrocCl) was used, in a transformation mechanistically analogues to the von Braun reaction, 7 to N-demethylate 9,10-dihydrolysergic acid methyl ester (5).Subsequent zinc mediated reductive cleavage of 6-N-Troc substituted compound 6 gave the desired 6-N-demethylated intermediate 7.This N-demethylation sequence was already reported earlier and slight procedure modications were used in our case. 8We have noticed that intermediate 7 decomposes slowly in acetic acid and is photolabile and therefore, the reaction yield depends on the swiness of work-up and isolation.6-N-Allylation of 7 was then realized by reaction with allylbromide and potassium carbonate in dimethylformamide, giving allylated intermediate 8 in a high yield (Scheme 3). 9 At this point, intermediate 8 served us for the synthesis of impurities A and D. Alkaline hydrolysis of methyl ester 8 gave cabergolinic acid, termed as impurity A by Ph.Eur., in 82% yield.The impurity D was prepared directly from methyl ester 8 as well.By reacting 8 with 3-(dimethylamino)-1-propylamine in acetic acid, direct amidation of ester gave impurity D in 74% yield (Scheme 4). 4 Compared to A and D, both impurities B and C have an ethylcarbamoyl functionality on the indole nitrogen.Since such functionalization of indole nitrogen requires the use of base, epimerization at the C8 position of ester 8 was possible.
Although the stereocenter at C8 in 9,10-dihydrolysergic acid ( 2) is less prone to epimerization under basic conditions than in lysergic acid, due to the absence of C9-C10 double bond, epimerization does occur. 10,11For example, it is known from the literature, that exposing compound 5 to 4 equivalents of NaH in THF at 20 C, app.30% of the opposite, thermodynamically less favourable, axial epimer is observed, and by exposing compound 8 to 4 equivalents of LDA in THF at À20 C, almost 71% of axial epimer is generated. 11Therefore, for installing the ethylcarbamoyl fragment on the indole nitrogen N-1 we have envisioned two different approaches.In the rst one, a direct functionalization of N-1 of 8, under basic conditions with EtNCO was planned.Being aware of the potential epimerisation problems under these conditions, a second indirect approach was envisioned.In this, indirect approach it was planned to reduce C8 carbonyl group of 8 to corresponding alcohol, do the functionalization of N-1 with EtNCO under basic conditions and then reoxidise the alcohol back to the corresponding carboxylic acid.Although longer, this reduction/oxidation approach, in comparison to the rst direct one, would ensure us with the epimerisation free preparation of the desired carboxylic acid.Since both approaches for functionalization of N-1, direct and indirect one, have advantages and disadvantages, we have decided to test both of them.Subjecting methyl ester 8 to the 1.1 equivalents of NaH and EtNCO afforded the desired epimer 9a in 6 : 1 diastereomeric ratio and with 47% isolated yield (Scheme 5). 12With the methyl ester 9a in hand we were only one step short of the key intermediate 9,10-dihydrolysergic acid 10, which would enable us to prepare impurities B and C. Luckily saponication of the methyl ester 9a proceeded selectively with respect to the N-1 ethylcarbamoyl fragment.This has provided us with the desired 9,10-dihydrolysergic acid derivative 10 in 77% yield.We found 9,10-dihydrolysergic acid 10 rather difficult to isolate, since it displays modest solubility in both organic solvents and water, but with acid 10 in hand, the door towards completion of the synthesis plan was open (Scheme 5).
For the indirect approach, methyl ester 8 had to be reduced rst.Reacting ester 8 with the LiAlH 4 in THF proceeded smoothly and in high yield to give 9,10-dihydrolysergol derivative 11. 13 Hydroxy group in 9,10-dihydrolysergol derivative 11 was protected as TBDMS ether 12 and a carbamoyl moiety was introduced by reacting 12 with sodium hydride and ethylisocyanate to give TBDMS-protected 9,10-dihydrolysergol derivative 13 (Scheme 6).
For the next steps, TBDMS-protected alcohol group in 13 had to be removed and free alcohol reoxidized back to carboxylic acid.Deprotection of silyl ether in 13 with TBAF gave 9,10dihydrolysergol derivative 14 in very good yield, although some temperature dependent chemoselectivity was observed.When silyl deprotection was performed at 0 C, slow but clean conversion of 13 to 14 was achieved.On the other hand, when the reaction was done at room temperature, removal of the ethylcarbamoyl group was observed in addition to silyl deprotection.9,10-Dihydrolysergol derivative 14 was then used for subsequent oxidation reaction.Ideally a one-step oxidation procedure from primary alcohol 14 to the corresponding carboxylic acid would be a method of choice.But surprisingly, while searching the literature, we learned that examples covering oxidations of 9,10-dihydrolysergols and their derivatives to the corresponding 9,10-dihydrolysergals and/or 9,10dihydrolysergic acids are very scarce.In the most recent literature, Křen et al. reported on their efforts to oxidize N-1-protected lysergol derivatives to aldehydes and acids. 14Beside that, patent literature covers two examples of 9,10-dihydrolysergol oxidations.In the rst one BAIB/TEMPO was used for direct oxidation of 9,10-dihydrolysergol derivative to the corresponding 9,10dihydrolysergic acid. 15Unfortunately, BAIB/TEMPO mediated oxidation did not work in our case, giving only traces of the desired acid, and failure to oxidize the alcohol 14 is in agreement with the results reported in the literature. 14In the second literature example, slightly modied Parikh-Doering oxidation (SO 3 $TEA, DMSO, TEA) was successfully used to oxidize 9,10dihydrolysergol to the 9,10-dihydrolysergal, 16 but as reported by Křen, they were unsuccessful in repeating this type of oxidation under standard Parikh-Doering conditions (SO 3 $Py, DMSO, TEA). 14Since 9,10-dihidrolysergol derivatives are poorly soluble in various solvents, we anticipated that DMSO mediated oxidations could be a good choice amongst the plethora of alcohol oxidation methods.DMSO should address the solubility issue of 9,10-dihydrolysergol derivative and concurrently serve as an oxidant.Since contrary reports about effectiveness of the Parikh-Doering oxidation of 9,10-dihidrolysergols were available in the literature, we have decided to try other analogues reaction.Luckily one can chose among different DMSO mediated oxidations.Indeed, when alcohol 14 was subjected to Ptzner-Moffatt oxidation conditions, a clean and high yielding conversion to the corresponding 9,10-dihydrolysergal 15 was observed. 17Here again temperature dependence on reaction selectivity was observed.Below 5 C the reaction was clean, giving only aldehyde as a product, while at room temperature or higher several side products could be observed by TLC.Since we were successful with Ptzner-Moffatt oxidation of 14, the question was raised, whether the outcome of Parikh-Doering reaction, on the same alcohol substrate would be analogues?Oxidation of the alcohol 14, under the Parikh-Doering conditions, proceeded smoothly, affording aldehyde 15 in slightly lower yield than via Ptzner-Moffatt reaction (Scheme 7). 18n added value of this oxidation transformations also lies in the 9,10-dihydrolysergal itself.Knowing that aldehydes of this type are difficult to prepare and that pharmacological activity of synthetically modied ergot alkaloids varies mainly because of the nature and conguration of substituents on C8, such aldehydes also represent valuable medicinal chemistry intermediates.4a,16a,19 Oxidations of 9,10-dihydrolysergals to the corresponding 9,10dihydrolysergic acids are hardly reported in the literature.This is mainly due to the fact that 9,10-dihydrolysergic acid derivatives are easily accessible from naturally occurring lysergic acid.The literature reports on oxidation of 9,10-dihydrolysergal with Tollens's reagent or MCPBA, but both methods have their limitations. 14The rst one unfortunately gives poor yields while the second one suffers from chemoselectivity.Peroxy acid oxidizes not only the aldehyde, but also the nitrogen on position 6 to the corresponding N-oxide.Although reduction of N-oxide to the desired 9,10-dihydrolysergic acid was demonstrated, 14 this method suffers from functional group compatibility, since allylated lysergal derivatives are not tolerated by peroxy acids nor catalytic hydrogenation.On the other hand there are some examples of lysergal oxidations to lysergic acid derivatives.In the nal steps of total synthesis of lysergic acid by Fujii and Ohno, Pinnick oxidation was used to oxidize 1-N-6-N-ditosylated isolysergal into the corresponding lysergic acid. 20Indeed, upon subjecting aldehyde 15 to Pinnick oxidation, the desired 9,10dihydrolysergic acid 10 was obtained, in 69% yield (Scheme 7). 21,10-Dihydrolysergic acid 10 was then used to synthesize both impurities B and C. The coupling of 10 with 3-(dimethylamino)-1-propylamine by the use of CDI furnished the impurity B in 54% yield.For the synthesis of impurity C a more complex N-acylurea appendage on C8 had to be installed.In the original synthesis of cabergoline, this side chain was installed with the use of EDC$HCl, by exploiting the usually undesired ability of carbodiimide coupling reagents to form N-acylureas as intramolecular rearrangement side products. 22Reacting cabergolinic acid A with EDC$HCl, in the absence of an external nucleophile, Mantegani and co-workers synthesized cabergoline (1). 4 Although the use of EDC$HCl gives the desired N-acylurea appendage in one step, a disadvantage of using this asymmetric carbodiimide is concomitant formation of regioisomeric N-acylurea and tedious separation of both isomers.
To compensate for disadvantage of applying EDC$HCl, different synthetic strategy was utilized in which excess ethylisocyanate was reacted with compound D to install N-acylurea appendage.Unfortunately the use of ethylisocyanate is also the root cause for formation of impurities B and C as side products in cabergoline (1) synthesis (Scheme 1). 4 Other methods for N-acylurea installation in cabergoline (1) synthesis were also developed, although none of them introduces N-acylurea appendage in one step from the corresponding 9,10-dihydrolysergic acid. 23Therefore, compromising lower yield, we have used EDC$HCl in DCM 24 for the synthesis of impurity C, thus preparing the desired impurity and its regioisomer 16 in ratio 5 : 1 respectively. 25Aer separation of regioisomers, the impurity C was obtained in 58% yield (Scheme 8).
For the synthesis of C, another synthetic strategy would be viable, by following linear route 10 / B / C. By reacting B with an excess of EtNCO, impurity C could be prepared.In the original synthesis of cabergoline (1) up to 40 equivalents of EtNCO were reacted with compound D to get the desired API.The amount of EtNCO was later reduced in consequent synthesis optimizations, by the use of CuI and Ph 3 P catalysis, to as low as 3 equivalents.Under these, optimised conditions cabergoline (1) was prepared along with the impurities B and C in the ratio of 82 : 6 : 12 respectively. 26As mentioned earlier, other syntheses of cabergoline (1) were developed to avoid the use of toxic EtNCO and they are all based on the functionalization of the carboxamide side chain of D. 23 In our case the impurity B was prepared from precursor 10 in a modest 54% yield.Therefore, if the reaction of B with EtNCO, or any other suitable reagent, would work quantitatively, the impurity C, starting linearly from 10, would be prepared in not greater than 54% yield.Following this, we have never tried to prepare impurity C from 10 via impurity B. Instead, the impurity C was prepared in one step, directly from carboxylic acid 10 with 58% yield (Scheme 8).
The structure of proper N-acylurea appendage in the impurity C was determined using 2D COSY NMR technique, where expected interactions between ethylamido protons were observed (Fig. 2).The proton NMR of the impurity C is also in accordance with the previous report of Mantegani and co-workers. 4

Conclusions
A synthesis of four European pharmacopoeial impurities A, B, C and D of cabergoline (1) was demonstrated and the synthesized impurities are used as HPLC analytical standards.To the best of our knowledge, this is the rst report on selective synthesis of these impurities.From a-ergocryptine (3), 9,10-dihydrolysergic acid methyl ester (5) was prepared, which served as the key starting material.Using 5, N-1 ethylcarbamoyl 9,10-dihydrolysergic acid 10 was prepared via two different routes.First route, a direct one, proceeds through direct N-1 ethylcarbamoylation of 10, causing partial epimerisation at C8 position.In spite of epimerisation, this route is shorter and therefore advantageous over the indirect one.Second route, an indirect one, follows reduction-oxidation of 9,10-dihydrolysergic acid derivative 8, for epimerisation free installation of N-1 ethylcarbamoyl moiety.Notwithstanding the fact that oxidations of 9,10-dihydrolysergol derivatives to the corresponding 9,10-dihydrolysergals and further on to 9,10dihydrolysergic acids are very rare, we have shown that by combination of activated DMSO and Pinnick reactions this oxidation sequence can be achieved successfully.Furthermore, such oxidation method of preparing otherwise scarcely presented 9,10-dihydrolysergals, can be useful in medicinal chemistry, for preparing novel pharmacologically active C8substituted derivatives of ergoline.

General experimental details
Where stated, anhydrous reactions were carried out in vacuum oven dried glassware, under stream of nitrogen, passed through DrieriteÔ drying column.Anhydrous solvents were purchased form Sigma-Aldrich and all reagents were commercial grade, used without any purication.Exception was a-ergocryptine which was acquired through in-house biofermentation process and was puried by crystallization from toluene.Reactions were monitored by thin layer chromatography (TLC) performed on pre-coated aluminum-baked silica gel (60 F 254 , Merck) or aluminum oxide (60 F 254 , Merck) plates.Developed TLC plates were visualized under UV light and/or by aqueous cerium ammonium molybdate and aqueous potassium permanganate.Flash chromatography was performed on ZEOPrep 60 Eco 40-63 mm silica gel and aluminum oxide activated, basic, with indicated mobile phase.Lower-temperature reactions were performed on ice/water bath or on Mettler-Toledo EasyMaxÔ synthesis workstation.Infrared spectra were recorded on a Bruker Alpha (Platinum ATR) FT-IR spectrometer and are reported in reciprocal centimeters (cm À1 ).Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance III 500 and Bruker Avance DPX 300 spectrometers.Chemical shis for 1 H NMR spectra are reported in parts per million from tetramethylsilane, with the solvent resonance as the internal standard (CDCl 3 , d 7.26 ppm, DMSO-d 6 , d 2.50 ppm).Data are reported as follows: chemical shi, multiplicity, (s ¼ singlet, d ¼ doublet, t ¼ triplet, q ¼ quartet, qn ¼ quintet, m ¼ multiplet, br ¼ broad), coupling constant in Hz and integration.Chemical shis for 13 C NMR spectra are reported in parts per million from tetramethylsilane, using solvent central peak as the internal standard (CDCl 3 , d 77.16 ppm, DMSO-d 6 , d 39.52 ppm).Where stated 2D NMR spectra were recorded.For pH measurements of water solutions, Seven Easy Mettler-Toledo pH meter was used.Optical rotations were determined at 589 nm at ambient temperature on Perkin-Elmer 2400 241MC polarimeter.Data reported as follows: [a] D , concentration (c in g per 100 mL) and solvent.Elemental microanalyses were performed on Perkin-Elmer Analyser 2400.Mass spectra were obtained using an AutoSpecQ, Q-TOF Premier and Agilent 6224 Accurate TOF LC/MS spectrometers.Melting points were determined on a Koer micro hot stage and are uncorrected.9,10-Dihydro-a-ergocryptine (4).In a 1000 mL round-bottom ask a-ergocryptine (3) was weighed (30 g, 52 mmol) and dissolved in MeOH (500 mL) at 35 C.Then, while stirring, Pd (3.9 g, 10% on carbon) suspended in MeOH (143 mL) was added and reaction mixture was purged with N 2 .Aer that reaction mixture was evacuated, purged with H 2 and put under H 2 atmosphere (balloon).Reaction mixture was stirred at 35 C for 3 h and aer that it was purged with N 2 and heated to 55 C. At 55 C the Pd catalyst was ltered off and ltrate was ran through aluminum oxide (30 g, basic) column and column was washed with MeOH (2 Â 200 mL) heated to 55 C. Combined MeOH fractions were evaporated at reduced pressure to dryness and resuspended in MeOH (450 mL).Suspension was heated to 55 C and while stirring, H 2 O (900 mL) was added.Stirred suspension was cooled to ambient temperature and further to 0 C, at which it was stirred for 0.5 h.Precipitate, was ltered off and dried in vacuum to give the titled compound as an off white solid (28.5 g, 95%).Optionally, product could be further crystallized from MeOH.R f : 0.