
Impurities Derived From Reduction By Sodium Alkoxides, And Formylating Reagents Derived From Ethereal Solvents
At the recent 20th SCI Process Development Symposium in Cambridge Dr R. Giles (GSK) presented work carried out on the development of a manufacturing route to Pranlukast (1), an orally active LTD-4 antagonist. The final stages of the synthesis were the coupling of aromatic ketone (2) with tetrazole ester (3) (prepared from sodium azide and ethyl cyanoformate, and shipped as the sodium salt) under basic conditions followed by cyclisation under acidic conditions to Pranlukast (1). The initial protocol was to react the aromatic ketone (2) and the tetrazole ester (3) with 6 equivalents of potassium tert-butoxide in THF. After acidification the intermediate (4) was isolated and washed with water. The water-wet diketone (4) was then cyclised with concentrated hydrochloric acid in methanol. The cyclisation reaction was heterogeneous throughout making it difficult to assess when the reaction was complete.

A telescoped version of these two stages in which methanol and concentrated hydrochloric acid were added at the end of the coupling reaction to effect the cyclisation worked well. The cyclisation reaction mixture was homogeneous and the drying of the final product (1) was quicker because there was no water removal required. Analysis revealed the presence of two major impurities (5 and 6/7), which led to an understanding of the side reactions taking place. A variety of different bases and solvents were investigated for the coupling step.

The des-tetrazole impurity (5) was impacted by the choice of solvent while the methylene-bridged impurities (6,7) were impacted by the choice of base. Up to 10% of the des-tetrazole impurity (5) was formed when the coupling was carried out with sodium m ethoxide in DMF as the solvent. Sodium methoxide can react with DMF to form methyl formate and sodium dimethylamide, but the equilibrium is very much in favour of the starting materials. When the tetrazole ester is added to the reaction mixture this can react with sodium dimethylamide to form the tetrazole amide (8) and so swing the equilibrium over forming appreciable amounts of methyl formate which can react with the substrate ketone (2) under the basic conditions to form β–ketoaldehyde (9) which can cyclise in the next stage to form the des-tetrazole impurity (5). Interestingly, no des-tetrazole impurity is formed when potassium tert-butoxide is used as base, and no formylation of (2), to give (8), is observed when the reaction is spiked with tert-butyl formate.
A variety of alternative solvents were considered instead of DMF, such as diglyme and THF, but even in these solvents up to 10% of the des-tetrazole impurity was observed. In both cases a formylating reagent could be produced via peroxide impurities in the solvent as shown.

The formation of the methylene-bridged impurities (MBI’s) (6,7) depends on the choice of base for the coupling reaction. MBI’s are formed when sodium alkoxides are used but not when potassium alkoxides are used. A Meerwein-Pondorff-Verley (MPV) reduction pathway was proposed to account for this whereby the aromatic ketone (2) is reduced to the alcohol while the alkoxide is oxidised to acetaldehyde or formaldehyde. The aldehyde can then react with two moles of (2) to form (6) or (7) depending on the alkoxide used (and so the aldehyde produced). The original literature on the MPV reduction (using aluminium tri-isopropoxide to catalyse the reduction of ketones to alcohols with isopropanol, effectively the reverse of the Oppenhauer oxidation) states that sodium alkoxides are inefficient MPV reagents but potassium alkoxides are not mentioned. This hypothesis is supported by the fact that no MBI’s are formed when sodium tert-butoxide is used as base until the tetrazole ester is added to the reaction mixture. Sodium tert-butoxide has no –protons which are required in the MPV reduction, but once the tetrazole ester is added to the reaction mixture ester exchange can talk place generating sodium ethoxide. Once it was known what to look for, the presence of small amounts of the alcohol (10) and the derived styrene (11).
















