A Review from the 6th International Conference on Organic Process Research and Development

Development of an Efficient Process for the Preparation of PNU-96391A, A Large Scale Implementation of a Suzuki Coupling

presented by Dr Mike Lipton (Pharmacia)

The meeting was opened by Dr Mike Lipton from Pharmacia who described development work carried out on PNU-96391, a chiral 3-arylpiperidine with potential indications for diskinesia and schizophrenia. Initial work centred on the synthesis of 3-bromophenyl methyl sulphone (1) which was prepared in 2/3 steps starting from 1,3-dibromobenzene. Reaction with sodium methanethiolate gave 3-bromophenyl methyl sulphide; it was found that this worked best if the reaction was not pushed beyond 90% conversion as the over thiomethylated product caused downstream processing problems. The oxidation to the sulphone with peracetic acid was extremely exothermic, so the reaction was run in 2 stages, allowing full conversion to the sulphoxide to occur before trying to effect the slower conversion to the sulphone. Ultimately this part of the chemistry was outsourced and the compound was manufactured via alternative technology.

The next challenge was the Suzuki coupling of 3-bromophenyl methyl sulphone and a 3-pyridyl derivative. The crystalline diethylborane derivative (2) was chosen and this could be readily synthesized in 76-83% yield from 3-pyridyl lithium (derived from 3-bromopyridine and n-BuLi, <-45^oC). There was some concern about the stability of 3-pyridyl lithium on a manufacturing scale, but it was inadvertently discovered that the lithio intermediate was stable below –45^oC for at least 7 hours.

The initial conditions for the Suzuki coupling used 3-4 mol% Pd(PPh₃)₄ (tetrakis), 60-70wt% tetra-n-butylammonium bromide (TBAB) and sodium hydroxide in aqueous THF. However the cost of the tetrakis and contamination of the product with TBAB led to improvements to this protocol. The final process was run at 100⁰C and used greatly reduced catalyst loadings (0.3 mol%), K₂CO₃ as base and toluene – water as the solvent with the product being isolated as the methanesulphonate salt (3) in 92-96% yield. This process was operated at “several hundred” kilo scale.

The reduction of the pyridine ring was relatively straight forward, hydrogenation over initially platinum oxide, which was subsequently changed to palladium on carbon. The reaction could not be run on the mesylate or hydrochloride salts due to the materials of construction of the autoclaves available so the biaryl mesylate was converted via the free base to the phosphate salt. The piperidine (4) was not isolated but was resolved directly with L-tartaric acid, which gave material of >97% de after 2 recrystallisations. The tartrate salt was not isolated in the final process, but was propylated under Gribble conditions (propionic acid / sodium borohydride) to give the final compound which was isolated as the hydrochloride salt.

New Generation PDE5 Inhibitors – A Process R&D Case History of UK-357,903
presented by Dr Phil Levett (Pfizer)

Dr Phil Levett from Pfizer talked about the development of UK-357,903 a second generation erectile dysfunction compound and in particular the synthesis of the substituted pyridine fragment (1).

Structurally (1) is very similar to the equivalent sildenafil (Viagra) unit, the main difference being a pyridine ring in place of a benzene ring. However the sildenafil approach (to chlorosulphonate the hydroxy acid, see Org Proc Res Dev., 2000, 4, 17) was not applicable as sulphonatioin of 2-hydroxynicotinic acids does not work according to the literature. Over 200kg of material was prepared via the medicinal chemistry route.

An alternative route starting with 2-chloronicotinic acid (an agrochemical intermediate) was investigated – reaction with sodium ethoxide gives ethyl 2-ethoxynicotinate, which can be nitrated in the 5-position and then reduced to ethyl 5-amino-2-ethoxynicotinate, and the derived diazonium salt can be converted to the sulphonic acid by reaction with SO₂ and CuCl₂ in acetic acid and then on to the desired intermediate (1)- but the sulphonation step was not amenable to scale up. The fact that 2-aminopyridine is sulphonated in the original route, suggested that the literature might be wrong.

Reaction of 2-hydroxynicotininc acid with 30% oleum at 140^oC gives the sulphonic acid (2) in 90% yield. This reaction and the subsequent quench are both exothermic and care had to be taken when deciding the optimal protocol for carrying out the reaction. Initially the reaction was quenched on to water and the intermediate trichlorinated to give 2-chloro-5-chlorosulphonylnicotinoyl chloride followed by selective hydrolysis of the acid chloride. On scale up significant hydrolysis of the sulphonyl chloride group was a problem, but this was solved by quenching the sulphonation reaction on to ethanol to generate the ethyl ester of the carboxylic acid, which could be dichlorinated cleanly.

In practice the chlorination, coupling with N-ethylpiperazine, reaction with sodium ethoxide and subsequent hydrolysis were all telescoped. The coupling with the other part of the molecule was initially carried out using hydroxbenzotriazole and a water-soluble carbodiimide, but a switch ro carbonyl diimidazole was found to be preferable. Cyclisation and introduction of the 2-methoxyethoxy side chain could be effected in a single step by reaction with potassium tert-butoxide in 2-methoxyethanol. Impurity problems arising from the mole of potassium hydroxide generated during the reaction were avoided by adding an ester to mop up the hydroxide ion, with 2-methoxyethoxy pivalate giving the best results.

The final form of the drug substance was an issue, as the monohydrate of the benzenesulphonate salt had been chosen from the original screen, but uncontrolled dehydration was a problem, but this was solved by switching to the anhydrous tosylate salt.

This work will be published in Organic Process Research and Development later this year.

The Chemical Development of Irbesartan
presented by Professor Bertrand Castro (Sanofi-Synthelabo) 

Irbesartan, Sanofi-Synthelabo’s 4th generation anti-hypertensive drug was discovered almost by accident. The researcher fortuitously discovered the compound while seeking a me-too compound to Dupont-Merck’s Losartan.

From a retrosynthetic viewpoint there are two possible strategies – make the biaryl C-C bond as the final stage or form the tetrazole ring as the final stage. Sanofi-Synthelabo chose the latter approach and this led back to o-tolylbenzonitrile (o-TBN) and spiro-cyclopentanone fragment (1).

The original syntheses of o-TBN were based on forming the biaryl C-C bond by reacting a Grignard reagent with a 2-methoxyaryloxazoline as pioneered by Evans. The final route was a Grignard coupling between o-tolylmagnesium chloride and 2-chlorobenzonitrile, catalysed by Manganese (II) chloride. As a result over a period of 8 years or so the cost of o-TBN dropped by a factor of 10. The spiro-cyclopentanone (1) fragment was synthesized from cyclopentanone using the route shown below.

Tetrazoylation was originally performed using tri-n-butyltin azide, but because of phase separation problems, waste problems and cost a cheaper reagent was needed. Sodium azide / triethylamine hydrochloride in DMF, as solvent, worked well in the laboratory, but needed a thorough hazard evaluation. The reaction was not exothermic and all attempts to make the reaction “run away” were unsuccessful. During the reaction, which is carried out at 130 ^ºC, some reflux is observed and some solid condenses on the cooler. The boiling liquid was identified as triethylamine and the solid is believed to be triethylammonium azide (Et₃NHN₃). Et₃NHN₃ is a shock and heat stable substance, unlike ammonium azide! So the safety of the reaction is believed to arise from the formation of Et₃NHN₃ in the reaction of sodium azide and triethylamine hydrochloride with precipitation of sodium chloride. When dissociation of Et₃NHN₃ to Et₃N and HN₃ occurs, the two liquids reflux, but because some of the HN₃ coordinates to the solvent there is always an excess of Et₃N in the head-space of the reactor, or more to the point there is never any free HN₃ present in the headspace.

There followed some discussion of the changing regulatory situation regarding what is and what is not be regarded as a regulatory starting material. Various suggestions have come out of the FDA, but a new official guideline is still awaited. Sanofi-Synthelabo’s strategy is to register the synthesis so that there is a set of regulatory starting materials, one key intermediate (preferably the final intermediate) and the drug substance. In the case of Irbesartan the starting materials are o-TBN and the spiro-cyclopentanone (1), and the key intermediate is the final nitrile precursor.

Sodium Dichlorobromate(I) Solution: A Wolf in Sheep’s Clothing (NaBrCl₂)
presented by Dr Hassan Elnagar (Albemarle) 

Bromine chloride is an unstable, corrosive liquid that can be used as a brominating reagent or as a bromochlorinating reagent. It is a more powerful brominating reagent than bromine itself. Sodium dichlorobromate (NaBrCl₂) is a safer more convenient form, which can be prepared by reacting bromine chloride with sodium chloride or by oxidising sodium bromide with chlorine. Like bromine sodium dichlorobromate will brominate activated aromatic compounds, but with higher regioselectivity than bromine (e.g. bromination of anisole gives >98:2 p/o selectivity). It can be used to brominate sensitive substrates such as 2,6-diisopropylaniline, which give tars under conventional brominating conditions, in good yield (>96%). The reagent can also be used to oxidise sulphides such as thioanisole to the corresponding sulphoxide without any ring bromination occurring if methanol is used as solvent. Similarly phosphines are oxidised to phosphine oxides.

The Solvent Effect on the Reaction of Bromine with p-Toluate esters. The Preparation of Methyl and Ethyl 4-(Bromomethyl)benzoates (4-MBMB & EBMB)
presented by Dr Hassan Elnagar (Albemarle) 

The side chain bromination of toluate esters can be carried out with a variety of reagents – N-bromosuccinimide or dibromodimethylhydantoin with AIBN in carbon tetrachloride or using bromine and light in the same solvent, and there is one reference in the literature to a thermal bromination at 185-190^oC. In concentrated solution the bromination is also accompanied by some de-esterification giving toluic acid and a-bromo-p-toluic acid as by-products. The best reaction conditions were found to be using chlorobenzene as solvent and an ester concentration < 50 wt% to minimise hydrolysis. In order to minimise benzal bromide formation the reaction is not pushed to completion and the product is purified by distillation.

The Hoffman Reaction in the Synthesis of Pyridines and Pyrazines: Some Case Studies
presented by Dr Michael Hassler (Rutger’s Chemicals) 

The Hoffman reaction (the conversion of a carboxylic acid amide to an amine: RCONH₂ ---> RNH₂) is an effective way of introducing and amino group in certain positions particularly in substrates such as pyridines and pyrazines where nitration is difficult. The reaction can be carried out using sodium or potassium hypochlorite or hypobromite and it is important to use freshly prepared reagent. Whether the potassium or sodium reagent gives best results depends almost entirely on the substrate and needs to be determined on a case by case basis. In general the cheaper hypochlorite reagent works well in most cases, but the stability of N-halogeno intermediate cannot be predicted (see reaction pathway below)

RCONH₂ + NaOCl ---> RCONHCl
RCONHCl + NaOH ---> RCON:
RCON: ---> RN=C=0
RN=C=O + H0- ---> RNHCO₂H
RNHCO₂H ---> RNH₂ + CO₂

Typically the first step is carried out cold (0-5 ^oC) and then the subsequent steps require heat to initiate but are all exothermic. The classical way to carry out the second part of the reaction is to dose the N-halo intermediate in to hot aqueous NaOH or KOH, but in some cases the whole reaction can be carried out at 30-40⁰C by adding the starting amide to a mixture of hypohalite and excess hydroxide (sodium or potassium). Unfortunately this method dopes not work in all cases.

A typical example of a synthesis employing a Hoffman reaction is that shown for 2-amino-3-methoxy-5-methylpyrazine (MMAP).

Although the yields for the individual steps were all >80% there were a number of challenges to be overcome in order to produce large quantities of MMAP efficiently. In particular the amide required extraction with diethyl ether and the bromination step was run at very low concentration. The bromination could be improved by using dibromodimethylhydantoin (DBDMH) in place of bromine but this led to safety issues caused by the presence of unreacted DBDMH in the filter cake, so an alternative route was developed.

Commercially available diethyl aminomalonate was used as the starting material and was converted to the diamide in the first step. The reaction with the alpha-ketoaldehyde was regioselective, but using the bisulphite adduct of the aldehyde was more convenient. The Hoffman reaction was found to go best under dilute conditions using potassium hypochlorite. In this particular case the whole reaction, including the rearrangement proceeded at 0-5⁰C. The next stage was simply to react the lactam with a chlorinating agent, however all standard methods gave poor yields and it was found that the methoxide displacement reaction did not work well. Serendipity played a part when Vilsmeier conditions (phosphorus oxychloride in dimethyl formamide, DMF) were tried. An unknown crystalline product was isolated, which was shown to be the chloro-imine. This reacted smoothly with methoxide to give, after aqueous work-up, MMAP.

In another example, 3-amino-2-chloro-4-alkylpyridine the identification of a cheap intermediate from the dyestuff industry (1) and the use of the Hoffman reaction allowed a potentially dangerous exothermic and non-selective nitration step to be avoided.

Safe Execution of a Large Scale Ozonolysis: Preparation of the Bisulfite Adduct of 2-Hydroxyindan-2-carboxaldehyde and Utility in a Reductive Amination
presented by Dr John Ragan (Pfizer) 

2-Hydroxyindan-2-carboxaldehyde was needed in multi-kilo quantities as a synthetic intermediaite. Two synthetic routes to this intermediate were studied. The original route was to convert 2-indanone to the cyanohydrin, or trimethylsilyl protected cyanohydrin and then to carry out a Dibal reduction, but this gave a modest yield of material heavily contaminated with inorganic impurities. Isolation of the product a a bisulphite adduct minimised lactol dimer formation.
The second synthesis involved reaction of 2-indanone with vinylmagnesium bromide followed by ozonolysis. The Grignard reaction proceeded efficiently if toluene was used as solvent, if THF was used large amounts of unreacted starting material were isolated, because enolisation of the ketone was a competing reaction. A detailed hazard study was carried out on the ozonolysis step and showed that the reaction was highly exothermic, but could be controlled by adjusting the addition rate of the ozone. The reaction was optimised in methanol, which gave a methylhydroperoxide intermediate rather than a secondary ozonide, which was produced when dichloromethane is used as solvent. Although a more stable intermediate was produced, the atmosphere in the headspace of the reactor had to be controlled to ensure a flammable atmosphere was not produced. This could be achieved by diluting the air / ozone feed x 19 with nitrogen. In practice it was more convenient to use air as the oxygen source and then dilute x 3 with nitrogen. In this way 9kg of bisulphite adduct was produced in 3 runs.

New Syntheses of the Dibenzoazepinone, Oxcarbazepine via Anionic Ring Closure
presented by Dr Olivier Lohse (Novartis)

Most synthetic routes to dibenzoazepinones start from iminostilbene which is produced by a 4 step synthesis from o-nitrotoluene – oxidative dimerisation, reduction of the nitro goups to amino groups, followed by a high temperature (~500 ⁰C) cyclisation and finally a high temperature (~600⁰C) dehydrogenation over an iron catalyst. Three of the 4 steps require dedicated equipment and this does not make iminostilbene a particularly attractive starting material. However the initial syntheses of oxcarbazepine did start from iminostilbene. The second generation process is shown below. Apart from the problem of the starting material, this route like the first generation route which contained a phosgenation step, necessitated the use of toxic gasses (Cl-CN, N₂O₄).

Alternative routes were considered with anionic and cationic ring closure strategies looking the most promising. The anionic route was developed further.

A number of parameters were studied, such as the leaving group, the protecting group on nitrogen and the base to use for the cyclisation. An amide group was chosen as the cyclisation terminus and after various dialkyl amides had been investigated the dimethylamide was selected. A carbamate was identified as the optimal protecting group for nitrogen and then the base to effect the cyclisation was investigated. LDA / TMEDA in THF looked promising as the reaction proceeded at –10⁰C and gave ~90% yield, but ultimately it was found that the reaction could be run just as efficiently in the absence of TMEDA. The ketone was protected as the enol ether by reaction with trimethyl orthoformate, the carbamate protecting group was removed and replaced with the desired urea and then the ketone was unmasked to give a synthesis that was cheaper, shorter and more ecologically friendly.

Challenges in Perfluoro-Organometallic Chemistry
presented by Dr. Ulf Tilstam (Schering AG) 

Perfluoro compounds exhibit quite different reactivity from their proto-counterparts, and present an unusual set of challenges. The reaction in question is quite simple, in principle at least, the addition of a perfluoroethyl group to the 17-keto group of a steroid, followed by hydrolysis of the ketal group at C-3.

The discovery chemistry method used 7 equivalents of perfluoroethyl iodide and 7 equivalents of methyl lithium – lithium bromide complex under Barbier reaction conditions in diethyl ether at –78^oC. The reduction of the excess reagents and a change of solvent were the most important goals, and these were successfully achieved. Using toluene as solvent and 2.9 equivalents of the reagents an improved yield of 65% was obtained (compared to 53%) with no chromatography required. However when the hazard testing was carried out the reaction mixture formed a gel which was unsuitable for scale up. A search for alternative solvents was undertaken. The reaction proceeded smoothly in THF and dimethoxyethane but gave the wrong product (the 17 alpha-methyl compound), whilst MTBE and diisopropyl ether exhibited poor solubility and/or gel formation. 2-Methyl-THF, benzotrifluoride, and diethoxymethane were also unsuitable. It was not until mixed solvent systems were studied that a solution to the problem was found – using MTBE/THF (80/20) gave a clear solution throughout and the product was obtained in 70% yield.

Retrosynthetic analysis of LY518674 suggests two possible strategies – one linear, the other convergent. The convergent approach looked promising initially, benzaldehyde was reductively coupled with hydrazine to give the benzylhydrazine (1) and then converted to a mixture (3:1 to 85:15) of semicarbazides. Attempts to take this through gave a mixture containing the desired product, but this was not a route that was going to be preparatively useful in the given time frame.

The linear approach quickly paid off, with the synthesis of the cyclisation precursor (3) proving straightforward, but the key step was yet to come. Attempts to carryout the cyclisation under basic conditions were unsuccessful, however the use of sulphonic acids in toluene gave an 85:15 mixture of the desired product and (2). The reaction conditions / acid were varied and optimised using camphorsulphonic acid in ethyl acetate at reflux and this gave a much improved product / des-urea ratio of 15.45:1.