In a previous episode chemical company Sanofi was granted exclusive access to certain yeast cells that produce a precursor to anti-malarial drug artemisinin. One of the charities making this all possible is the Bill and Melinda Gates Foundation. Another charity that has apparently entered into the drug business is the Clinton Health Access Initiative. Bill together with Rodger Stringham and David Teager report on an improved process for the conversion of artemisinin to artemether in Organic Process Research & Development (DOI).
Does the Clinton Health Access Initiative have a pilot-plant facility or even an organic lab? Unless it is all cramped in suite 400 on Dorchester Avenue in Boston, the article is not very explicit. The acknowledgements mention Mangalam Drugs and Organics.
Case at hand: artemether has the carbonyl group replaced by a methoxy group in a two-step reduction - methylation. So far so good. The point is that principal supplier Novartis reports up to 68% overall yields but that many Indian and Chinese suppliers working with the procedure generously supplied by same Novartis, report considerably lower figures (58-62%). But Why? And how can the process be improved?
Any organic chemist knows reported yields in the literature should be considered with caution. Chemists tend to be over-optimistic / self-delusionional but this scenario was not considered. No bottlenecks were encountered in step 1, the reduction with sodium borohydride. Only the beta form was isolated due to its poor solubility in the quench. Drying the product without heat prevented formation of one byproduct. Moving on to step two, the methylation with HCl in methanol was more troublesome. The byproducts lurking around the corner are the anomer and the elimination product. Co-solvent (co-reagent?) trimethyl orthoformate made all the difference. The critical element in the workup was first adding more methanol before adding the base quench otherwise you end up with a nasty gum. The new record yield for the improved synthesis is 72%.
But what have all these suppliers been doing wrong with the existing Novartis procedure? The answer to that question, remains unclear. The Novartis yield for step two with co-solvent methylacetate (not the formate) was confirmed so no surprise there.