PHARMACEUTICAL INDUSTRY IN SWITZERLAND 640 CHIMIA 2004, 58, No. 9 Chimia 58 (2004) 640–648 © Schweizerische Chemische Gesellschaft ISSN 0009–4293 Broad Spectrum Chemistry as Practised by Novartis Process Research Stuart J. Mickel*, Reto Fischer, and Wolfgang Marterer Abstract: Three actual examples from the current product palette within Novartis Process Research will demon- strate the some of the variety and challenges encountered in modern chemical development. Keywords: Aldol · Dakin-West · Dimroth · Discodermolide · Nitration 1. Introduction how mechanistic considerations can be of The synthesis from research (Scheme 1) immense value in attaining a safe repro- was relatively straightforward, however, for Modern chemical development faces a ducible scalable process. the following main reasons, we elaborated formidable challenge in converting the an improved, alternative synthesis for wonderful set of reactions and reagents scale-up in the pilot plant and towards fu- which are published in the chemical litera- 2. CGP59326 New Process ture production: ture every day, into workable reliable 1) The use of a benzyl protecting group for processes. Normally this ‘state of the art’ Pyrrolo[2,3-d]pyrimidine-derivative 1 the pyrrole amine functionality was chemistry is described to function, usually (Fig.1) was manufactured in 100 kg scale mandatory for the outlined synthetic on a very small scale, but its conversion to for clinical investigation as an exploratory strategy, however, deprotection turned a scaleable reproducible process provides anti-tumour agent [1]. out to be technically unfavourable be- the development chemist with a set of se- cause it required a large excess of alu- vere problems. The variety of structures minium trichloride which subsequently that can be encountered by the development had to be quenched and disposed of. chemist is vast and ranges from antibiotics 2) The classical, three-component pyrrole- to tertasaccharides. The array of issues with forming condensation reaction was low- which a development chemist is confronted yielding. is equally as large, from environmen- 3) The pyrimidine cyclisation step in boil- tal/health and safety concerns to polymor- ing formic acid raised safety concerns phism and the properties of solids. In this and resulted in dark coloration of the article we would like to highlight some of product which made cumbersome pu- these challenges by referring to a selection rification procedures necessary. of three recent projects from our depart- 4) The hydroxy- as well as the chloropy- ment. These will demonstrate how some of rimidine intermediates had extremely the issues encountered in the daily business Fig. 1. poor solubility which resulted in dilute of chemical development may be solved. In the first example a short overview of the development history of a relatively sim- ple compound will show how the elabora- tion of alternative synthesis concepts – even in the field of well-known heterocycles – can indeed lead to economically as well as ecologically superior manufacturing processes. It also demonstrates the value of re-examining ‘old’ reactions and shows *Correspondence: Dr. S.J. Mickel Chemical and Analytical Development Novartis Pharma AG CH–4002 Basel Tel.: + 41 61 696 29 52 Fax: + 41 61 696 29 57 E-Mail: [email protected] Scheme 1. PHARMACEUTICAL INDUSTRY IN SWITZERLAND 641 CHIMIA 2004, 58, No. 9 reactions and a large excess of phos- to form the acylamino ketone (Scheme 3) literature. The results were convincing: 1 phorus oxychloride being used for the [6]. equiv. of acetic acid made it possible to run chlorination step. After some of our own investigations the reaction in an addition-controlled mode confirmed the mechanism outlined above, with carbon dioxide evolution being directly 2.1. Development Synthesis we concluded that the following kinetic in- linked to alanine addition. As an extra bene- Strategy terpretation would at least partly explain the fit, acetic anhydride and pyridine excess For the synthesis of 1,we concluded peculiarities of the Dakin-West reaction: as could now be significantly reduced. that 2-amino-3-cyano-4,5-dimethylpyrrole we see in Scheme 3, the ring-opening hy- In a second step, the reaction was suc- (2) (Fig. 2) was the intermediate of choice. drolysis of acylated azlactone formally re- cessfully adapted to the modified Dakin- Its simple synthesis from 3-amino-2-bu- quires 1 equiv. of water. Since there are cer- West procedure with triethylamine/DMAP, tanone and malonodinitrile was already de- tainly not sufficient quantities of water in a introduced by Steglich and Höfle in 1969 scribed more than 40 years ago [2] and it is boiling reaction mixture of excess anhy- [7] which requires much smaller amounts sufficiently stable towards air oxidation to dride and pyridine, this ‘water-equivalent’ of anhydride and base and is therefore eco- allow normal handling in a production fa- has to come from the formation of acetic nomically more advantageous. Eventually, cility. anhydride from acetic acid. Acetic acid it- over 0.5 kmol alanine were successfully From this pyrrole we intended to modi- self is formed in the first two steps of the re- converted to acetylaminobutanone 3 in a fy the pyrimidine ring formation in such a action sequence, namely N-acylation and 400 l reactor in over 90% yield with carbon way that no protection was necessary and azlactone condensation. Therefore its ini- dioxide evolution (~ 11’000 l!) being in a the need for formic acid was eliminated. tial concentration in the reaction mixture is virtually linear relation to alanine addition. very low and hence azlactone hydrolysis is The crude product solution was concentrat- very slow. Once the mixture is heated and ed by evaporation of acetic acid, acetic an- N-acylation of alanine starts to increase the hydride, and triethylamine. This concen- acetic acid concentration increases rapidly trate was diluted with water and malono- and therefore azlactone hydrolysis is accel- dinitrile and the resulting solution added to erated. The following decarboxylation is 30% sodium hydroxide whereby pyrrole 2 virtually spontaneous. This results in the re- was formed and precipitated in an extreme- action having a typical, self-accelerating ki- ly fast, exothermic but fully addition-con- netic, which is exactly what you do not trolled reaction [8]. Crystalline pyrrole 2 want to have in a pilot or production plant, was directly isolated by centrifugation, particularly not if equivalent amounts of washing with water and drying and did not Fig. 2. gas are formed during a reaction. require further purification [9]. Neither dis- To test this hypothesis, we performed ex- tillate nor mother liquor required any spe- periments where we added various amounts cial treatment. This procedure for the syn- of acetic acid to the initial mixture of acetic thesis of pyrrole 2 (Scheme 2) is very effi- 2.2. Pyrrole Synthesis anhydride and pyridine – an idea which sur- cient, high-yielding (83–85% over 2 steps α-Acylamino ketone 3 [3] is not com- prisingly has never been investigated in the calculated from alanine) and requires only mercially available but easily accessible two standard reactors. from alanine and acetic anhydride, a reac- tion that was initially described by Dakin and West in 1928 [4] and therefore is known as the Dakin-West reaction. Since (racemic) alanine is very cheap, this seemed to be an attractive approach to the synthesis of pyrrole 2 (Scheme 2). However, there is one major drawback to the Dakin-West reaction regarding its scale-up to plant scale: it is accompanied by an uncontrolled formation of carbon diox- Scheme 2. Pyrrole synthesis ide, which is absolutely prohibitive on tech- nical scale. Our initial attempts to control formation of carbon dioxide by variation of reaction parameters all failed and addition- ally never produced yields which were at least comparable to the standard protocol. So we decided to examine the reaction mechanism in more detail. Even though the Dakin-West reaction can hardly be considered well-known, there is quite a lot of published work in- cluding some very detailed mechanistic investigations particularly by Steglich [5]. It is generally accepted that the mecha- nism involves dehydrative formation of an azlactone (oxazolinone) which is then C- acylated (in equilibrium with O-acyla- tion), and then undergoes ring-opening hydrolysis followed by decarboxylation Scheme 3. Published mechanism of the Dakin-West reaction PHARMACEUTICAL INDUSTRY IN SWITZERLAND 642 CHIMIA 2004, 58, No. 9 2.3. Pyrimidine Ring Synthesis and upon isolation identified as amidine 4c able for production in standard equipment In 1964, Taylor and Hendess reported a which now seemed to be the reactive of a multi-purpose plant [16]. In addition – procedure to convert an unprotected amino- species towards the formation of the pyrim- to the best of our knowledge – this is the cyano-pyrrole into an aminopyrrolopyrimi- idine ring [12]. This optimised and simpli- first application of a Dakin-West reaction dine derivative with trimethyl orthoformate fied process was run in the pilot plant on on a technical scale. and ammonia [10]; however the very scarce 0.45 kmol scale. Chloroaniline and triethyl use of this method in the newer literature orthoformate were dissolved in ethanol and seemed to indicate a somewhat limited the pH of the solution was adjusted to 5.5 3. Discodermolide scope. We reasoned that in principle, it by addition of acetic acid. Pyrrole 3 was should be possible to apply this type of then added over a few hours at 45–50 °C; The second example demonstrates the pyrimidine ring formation to pyrrole 3 un- after cooling, precipitated pyrrolopyrimi- power of modern organic chemistry and al- der carefully optimised conditions which dine 5 was centrifuged off, washed and so highlights some of its limitations. We avoid side reactions at the pyrrole ring ni- dried. The yield in the pilot plant was over were asked to prepare a significant amount trogen.