Synthesis and Scale Up of the Tissue Selective Estrogene ZK 186619
Johannes Platzek spoke about the synthesis and scale up of the tissue selective Estrogene ZK 186619 (1) carried out at Schering AG. The medicinal chemistry route (see Scheme 1), 19 steps including 2 chromatographic purifications, had been successfully used to prepare 100g of product with an overall yield of 5%, but a more efficient route was required. Various scale up issues were identified – use of PCl5/SnCl4 for the cyclisation, demethylation with pyridine HCl requires temperature of 180oC, 5 equivalents of 4-methoxyphenylmagnesium bromide were required for the Grignard reaction, demethylation with NaSCH3 generates methane thiol as a by-product, and finally there were several protecting group changes in the synthesis.
Scheme 1
An alternative synthetic route was devised where the phenyl group was introduced prior to the cyclisation stage. The same starting material, 3-methoxybenzaldehyde, was used but in the new route was reacted with acetaldehyde to give 3-methoxycinnamaldehyde. The literature yield for this reaction was 10-20%, but initial optimisation studies improved this to 35% and then automated optimisation using a statistical approach gave a fully optimised yield of 80%. The intermediate 3-methoxycinnamaldehyde was not isolate but was reacted directly with phenylacetic acid to give a diene-acid intermediate which was isolated and then hydrogenated to give the cyclisation precursor. Cyclisation was effected by treatment with polyphosphoric acid (PPA), but the work-up of this reaction had to be modified for scale up. Quenching by addition of water was extremely exothermic, but the reaction mixture was too viscous to add to water. The answer was to cool the reaction mixture to 50oC, add 15% methanol to give a pumpable mixture, which could be quenched onto water.
Various methods of introducing the 4-methoxyphenyl group were investigated; including Grignard addition, aryl lithium addition, but the best method proved to be a Suzuki coupling with the enol sulphonate derived from the cycloheptanone. Some convergence was introduced to the synthetic route at this point by introducing the side chain in one piece (see Scheme 2).
A number of reagent combinations were evaluated for demethylation to give the phenol (48% HBr, HBr/HOAc, TMSCl/NaI/acetonitrile, AlCl3, BBr3/CH2Cl2) with boron tribromide/base working best. The nature of the base was critical to the success of the reaction. When pyridine was used a yield of 40% was obtained, but utilising a more hindered base, 2,6-lutidine gave much improved yield (85%), but use of an even more hindered base, 2,6-di-tert-butylpyridine was ineffective.
Introduction of the pentafluoro side chain was fairly straightforward although two routes were considered – introduction of the sulphide and then oxidation to the sulphoxide, or introduction of the side chain in one unit as the sulphoxide. This latter approach although introducing more convergence to the synthesis was less economic because of the cost of the 4,4,5,5,5-pentafluoropentan-1-ol used as starting material for this leg of the synthesis (the increased convergence meant this material had to be processed through more chemical steps). The final stages were therefore essentially as carried out in the medicinal chemistry synthesis – conversion of pentafluoropentanol in to the tosylate followed by displacement of the tosylate group with potassium thioacetate to give pentafluoropropyl thioacetate.
The demethylated chloro intermediate from the Suzuki coupling was converted in to the iodide in a Finklestein reaction in 2-butanone and this was then reacted with the thioacetate ester using sodium methoxide as base. The final oxidation was carried out with sodium metaperiodate. Hydrogen peroxide could also be used as the oxidant, but this gave a product with a poorer impurity profile.
Scheme 3
In summary the process route required 12 chemical steps (7 isolated steps) in 25% overall yield (lab, 20% on pilot plant scale) with no chromatography required.