FIELD OF THE INVENTION

The present invention relates to a novel route of synthesis for the production of enantiomerically pure Cobimetinib, new intermediates in the synthesis of Cobimetinib and an amorphous Cobimetinib hemifumarate salt comprising a high salt content.

BACKGROUND

Cobimetinib is a kinase inhibitor and belongs to the group of MEK-inhibitors. The compound is clinically used in the form of the hemifumarate salt

2 and the treatment usually includes the combined administration together with a BRAF- Inhibitor like Vemurafenib to treat progressive melanoma comprising a BRAF-V600- mutation. The clinical effects are based on the interaction of the drug with the MAPK-signal pathway by inhibition of MEK-kinase. Cobimetinib is able to positively increase the efficiency of Vemurafenib and additionally reduces possible unintended side effects of the BRAF-inhibitor.

Several different routes of Cobimetinib synthesis can be found in the scientific as well as in the patent literature.

EP 19 341 74 Bl for instance teaches the use of azetidines as MEK inhibitors for the treatment of proliferative diseases. The patent document discloses compounds according to the following Formula and pharmaceutically acceptable salts and solvates thereof. Such compounds are MEK inhibitors and are useful in the treatment of proliferative diseases, such as cancer. Also disclosed are pharmaceutical compositions containing such compounds as well as methods of using the compounds and compositions of the invention in the treatment of cancer. This patent document teaches the separation of enantiomers of Cobimetinib intermediates by derivatisation with chiral auxiliary agents (e.g. to form a Mosher ester) and separation by flash chromatography. This procedure is only workable in small scale and not suitable for efficient large-scale production due to the applied working procedures (flash

chromatography). Further, this route requires high costs for the chiral auxiliary agent.

EP 29 091 88 Bl teaches the introduction of chirality at an early stage of the synthesis by a (3S,5R,8aS)-3-phenyl-hexahydro-oxazolo[3,2-a]pyridine-carbon itrile followed by five subsequent steps including two chromatographic purifications.

Furthermore, WO 2014 059 422 Al disclose a process of making a compound according to the following formula

wherein the compound is an inhibitor of MEK and thus can be used to treat cancer.

In addition, WO 2017 004 393 Al relates to crystalline fumarate salts of (S)-[3,4-difiuoro-2- (2-fluoro-4-iodophenylamino)phenyl] [3-hydroxy-3-(piperidin-2-yl) azetidine-l-yl]- methanone. This document also relates to pharmaceutical compositions comprising the crystalline fumarate salt of (5)-[3,4-difluoro-2-(2-fluoro-4-iodophenylamino)phenyl] [3- hydroxy-3-(piperidin-2-yl) azetidine-l-yl]-methanone and methods of treating cancers comprising administering to a patient in need thereof the crystalline fumarate salt of (5)-[3,4- difluoro-2-(2-fluoro-4-iodophenylamino)phenyl] [3-hydroxy-3-(piperidin-2-yl) azetidine-1 - yl]-methanone.

Thus, the use for treatment and at least some routes for synthesis of Cobimetinib or derivatives are disclosed in the literature. Nevertheless, it has been found that it is difficult to find suitable ways to achieve stereochemically well defined material.

WO 2017 096 996 Al, for instance, describes a route of synthesis using very expensive chiral starting material. However, such route is disadvantageous, because from the beginning, costly material has to be used and, consequently, losses during the complex synthesis occur by epimerization. Such losses add to the overall synthesis costs.

CN 106 045 969 A, CN 105 330 643 A and CN 106 220 607 A also attempt to solve the problem by using expensive chiral starting material. Consequently, during the several synthesis-steps costly losses have to be considered, based on the work-up steps and the risk of epimerisation.

Another patent, CN 104 725 352 A uses chiral chromatography for resolution of enantiomers, which is an inefficient and expensive procedure for largescale manufacturing.

Nevertheless, besides the existing ways of synthesis, there is still the need for further reliable routes of synthesis, which are able to deliver under conditions suitable for large scale production high yields of enantiomerically pure material at moderate synthesis conditions and costs.

BRIEF DESCRIPTION OF THE INVENTION

Above mentioned problem is solved by a process for the production of Cobimetinib (5)[3,4- Difluoro-2-(2-fluoro-4-iodophenyl-amino)phenyl] [3-hydroxy-3-(piperidin-2-yl]azetidine- 1 - yl)methanone, characterized in that an activated 3,4-difluoro-2-[(2-fluoro-4- iodophenyl)amino]benzaldehyde-derivative (compound A) according to the following formula

compound A wherein Act is an activating moiety selected from the group consisting of CI, Br, Imidazoyl, O- Succinyl, OH, O-benzotriazoyl, 0(CO)R\ wherein R ¹ is selected from the group consisting of C1-C5 Alkyl or Benzyl, is reacted with a 3-(2-Z-piperidin-2-yl)azetidine-3-ol-derivative (compound B), wherein the Pg are independently the same or different protection groups or H and Z is selected from the group consisting of H or cyano; or the cyclization product thereof tetrahydro-5'//-spiro[azetidine- 3,l'-[l,3]oxazolo[3,4-a]pyridin]-3'-one (compound B')

compound B compound B ' to yield [3,4-Difluoro-2-(2-fluoro-4-iodo-phenyl-amino)phenyl] [3-hydroxy-3-(piperidin-2- yl]azetidine-l-yl)methanone (compound C) or l-(3,4-difluoro-2-((2-fluoro-4- iodophenyl)amino)benzoyl)tetrahydrospiro-[azetidine-3,r-oxaz olo[3,4-a]pyridin]-3'(5'ii)-one, respectively (compound C)

compound C compound C followed by optionally de -protection of compound C or ring-opening of compound C to yield essentially enantiomerically pure Cobimetinib free base or a pharmaceutically acceptable salt thereof, wherein the essentially enantiomerically pure Cobimetinib is obtained by kinetic resolution of one or more synthesis intermediates or the final product. Surprisingly it has been found that above given synthesis route is able to provide Cobimetinib free base or salts in high yields, high purity and only small amounts of side -products, which can easily be separated from unwanted by-products by standard purification techniques. This can be attributed, at least in part, to the overall gentle processing conditions compared to other state-of the art processes. Furthermore, the overall reaction scheme is environmentally friendly with respect to the used solvents and reagents, the reaction times are short and the synthesis is readily up-scalable. A further advantage of the process is that enantiomerically pure forms can be achieved by means of kinetic resolution of selected intermediates or the end product. The resolution of bases by means of simple and cost effective kinetic resolution with chiral acid is an established process in the pharmaceutical industry, but however, no kinetic resolution process is known, targeted to the intermediates or the end product in the synthesis of Cobimetinib until today. In the available literature, much more sophisticated and expensive approaches for the resolution of the enantiomers have been suggested and patent documents reveal, that it is seems impossible to achieve suitable salts (e.g. WO2017/004393A1, paragraph [0024]). Therefore, the literature suggests alternative routes based on chiral starting material or chiral chromatography, only. For the first time we report in this application a straight forward resolution procedure. Such resolution process is very surprising, because it was found, that Cobimetinib or the synthesis intermediates thereof, do not easily form ionic salts, but this is a pre -requisite for an efficient resolution process. Without being bound by the theory it is assumed that the difficulties in salt formation can, at least in part, be attributed to the compound structure,

wherein it seems that the hydrogen atom sterically shields the N(l)-atom, preventing or hindering effective salt formation. The same is also true for the intermediates containing azetidine rings with a chiral centre at the tertiary hydroxy-group. The proposed kinetic resolution further comprises the advantage, that epimerisation of the undesired enantiomer of compound B or compound B' is possible and an increase of the overall efficiency of the resolution process can be obtained. The basis of this synthesis route is provided by the reaction of the compound A and either compound B or compound B', wherein the compound B and B' are interconvertible. The processes can be performed in parallel, both reaction arms can be performed separately or the arms of the reaction schemes can be crossed in order to yield the core ring-structure of compound B to B' or vice versa. With respect to the synthesis outcome there are no significant differences with respect to yield, purity and workability. Therefore, the proposed reaction forms a single synthesis concept.

In the course of reaction an activated 3,4-difluoro-2-[(2-fluoro-4-iodophenyl)amino]benzaldehyde- derivative (compound A) is used, wherein Act is an activating moiety selected from the above mentioned group. The task of the activating group is to provide a leaving group for the coupling of compound B or B' at mild reaction conditions. This reaction can easily be controlled and is highly selective, building the bases for the overall reproducible outcome and the low amount of unwanted side -products.

Compound A is reacted with an 3-(2-Z-piperidin-2-yl)azetidine-3-ol-derivative (compound B), wherein the Pg are independently the same or different protection groups or H. This means, that the Nitrogen-atoms of compound B may either be protected by one or two protective groups, wherein the protection groups can be the same or different, or the nitrogens in the compound B are unprotected. Therefore, it is possible to tailor the reaction either by using already the protected educts, by reaction of only a single N-protected educt or to use a non N-protected educt and perform any N-protection on the reaction product of compound A and B, respectively. Suitable protection groups Pg can for instance be selected from the group consisting of Boc, Cbz, Fmoc Benzyl, biphenylmethyl, acetyl (Ac), benzoyl (Bz), p-methoxybenzyl carbonyl (Moz), benzyl (Bn), carbamate, tosyl (Ts), trichloroethyl chloroformate (Troc) or mixtures thereof. Methods of N-protection and de-protection are known to the skilled artisan.

Compound B also comprises a group Z, wherein this moiety is selected from the group consisting of H or cyano. If this group is a cyano-group it is especially possible to build a ring structure by cyclization of this functional group. This reaction inter alia is responsible for the interchangeability of the compound B and B' in the course of a reaction with compound A.

The de-protection of compound C or ring-opening of compound C may be performed by methods known to the skilled artisan. The de -protection reaction is especially selected with respect to the used protection group and may for instance include acid or base treatment or other reactions suitable to selectively cleave the protection group-nitrogen bond, e.g. by hydrogenation.

Within the disclosed process it is possible to obtain essentially enantiomeric pure Cobimetinib free base or a pharmaceutically acceptable salt thereof. The chemical purity of this compound should be more than 99%, preferentially more than 99.5 % and further preferred more than 99.9%%. The compound is essentially enantiomerically pure, if the compound comprises a chiral purity of more 99% ee, preferentially more than 99.5 % ee and further preferred more than 99.9%% ee. The purity and the enantiomerically purity determination can be performed via HLPC and chiral HPLC methods, respectively. The exact methods are known to the skilled artisan.

For obtaining the essentially enantiomerically pure Cobimetinib a kinetic resolution of one or more synthesis intermediates or the final product is performed. Kinetic resolution is a process, wherein a chiral resolution of a mixture of enantiomers is achieved. The process includes a separation of compounds containing a chirality centre by means of a resolution agent. The resolution agents have to contain at least one chiral centre and are used preferably in enantiomeric pure form. As result a diastereomeric salt is formed with enrichment of one enantiomer of the to-be -resolved compound and the chiral resolution agent. For the compounds at hand, i.e. at least in part chiral amines, resolution can be performed by reaction with chiral acids. The chiral acids preferably are selected from the group of chiral organic acids like, but not limited to, tartaric acid, o-Benzoyl-tartaric acid, lactic acid camphour- 10-sulphonic acid and mandelic acid. The ratio between the chiral base and the chiral acid can vary in the range of 0.25 to 10 eq. of chiral acid related to the chiral base. The final product is preferably Cobimetinib free base.

In a preferred embodiment of the process Z in compound B can be H and compound B can be obtained via a process at least comprising the steps of a) reacting piperidine-l-carboxylic acid with a protecting agent PG to yield N-protected piperidine-1- carboxylic acid;

b) reacting the N-protected piperidine-l-carboxylic acid and an activating reagent to yield the corresponding activated acid (compound Bl)

Bl c) reacting compound Bl with a functionalized Cl-synthon to yield N-protected 2-[2-(chloro- methyl)oxiran-2-yl]piperidine (compound B2)

Bl B2 d) addition of benzylamine to compound B2 to yield N-protected l-(benzylamino)-3-chloro-2- (piperidin-2-yl)propan-2-ol (compound B3) and cyclization of compound B3 to yield N-protected 1- benzyl-3-(piperidin-2-yl)azetidine-3-ol (compound B4)

B2 B3 B4 e) optionally followed by selectively de -protection of the azetidine-ring of compound B4 to yield an azetidine N-unprotected 3-(piperidin-2-yl)azetidine-3-ol (compound B)

This route of reaction for obtained compound B has been found useful, because compound B is obtainable by only a small number of reaction steps, which can be performed at moderate reaction conditions. Furthermore, by using this reaction path high yields are obtainable and unwanted side- products can easily be removed from the desired product by standard separation techniques.

In step a) piperidine-1 -carboxylic acid is reacted with a protecting agent Pg. Suitable protecting agents are known to the skilled artisan and can for instance be selected from above given list of suitable protecting agents.

In step b) the carboxylic acid functional group is activated by changing the hydroxyl-functionality of the carboxylic acid group with another function, especially leaving group, which is also referred to as an activating group. Suitable leaving groups are for instance halogens. Therefore, suitable activating agents can be for instance halogenation reagents. Suitable halogenation reagents are known to the skilled artisan and can for instance be selected from the group consisting of thionylchloride, oxalylchloride, phosphorus tricloride, phosphorus pentachloride, triphenylphosphine/tetrachlorocarbon (Appel reaction).

In reaction step c) compound Bl is reacted with a functionalized Cl-synthon. A Cl-synthon is any organic group able to add one carbon atom to the functionalized carboxy group. This synthon further bears a functional group which is able to form the ring structure as depicted in compound B2. Therefore, possible functionalizes CI synthons may be selected from the group of the di-halo methanes, wherein the halogen atoms may be the same or different. A possible functionalized Cl-synthon may for instance be a chloro-functionalized Cl-synthon according to the structure CI-CH2-I.

In a further preferred embodiment of the process Z in compound B can be H and compound B can be obtained via a process at least comprising the steps of a) esterification of compound B l with R ² -OH to yield R ² -piperidine-2-carboxylate (compound Bl.l), wherein R ² is C1-C5 alkyl or CH ₂ -Ph;

b) reacting compound Bl.l and a functionalized Cl-synthon to yield N-protected 2-halo-l-(piperidin- 2-yl)ethanone (compound B 1.2)

1.1 B1.2 c) reacting compound B1.2 with a functionalized Cl-synthon to yield N-protected l,3-halo,chloro-2- (piperidin-2-yl)propan-2-ol (compound B 1.3)

B1.2 B 1.3 d) reacting compound B1.3 with benzylamine to yield N-protected l-(benzylamino)-3-halo-2- (piperidin-2-yl)propan-2-ol (compound B3)

e) cyclization of compound B3 to yield N-protected l-benzyl-3-(piperidin-2-yl)azetidine-3-ol (compound B4) and

B3 B4 f) optionally de-protection of compound B4 at the azetidine-ring to yield azetidine-unprotected compound B. This route of reaction has been found very efficient for the synthesis of compound B. The reaction can be performed at moderate reaction condition and result in high yields even at such moderate reaction conditions. The amount of unwanted side -products is very low and high outputs are possible at short reaction times.

In step a) of above displayed reaction compound Bl is reacted with an alcohol to yield the Bl -ester respectively. Especially the lower CI -C5 alkyl- or CH ₂ -Ph-ester has been found very suitable for further reaction with the functionahzed Cl-synthon. The ester-cleavage of these esters is easily achievable and quantitative, resulting in high reaction rates in reaction with the Cl-synthons.

Within a further aspect of the process the synthesis of compound B3 in step d) can be performed stepwise, wherein in a first reaction step dl) compound B1.3 is reacted in the presence of a base to the epichlorohydrin-derivative compound B2

B1.3 B2 followed by a second reaction step d2), wherein compound B2 is reacted by addition of benzylamine to the double N-protected compound B3

B3

This reaction route is suitable to deliver compound B3 in high yields with only a very low amount of unwanted side products.

In step dl) the compound B 1.3 is reacted in the presence of a base to yield the epichlorohydrin-derivative B2. Suitable bases for the cyclization can be organic or inorganic bases, wherein the inorganic bases are preferred. The reaction can for instance be performed in the presence of the alkali or alkaline earth hydroxides. It is further possible to use bases selected from the group consisting of NaHCC , KHCO ₃ ,

In another aspect of the process Z in compound B can be H and compound B can be obtained via a process at least comprising the steps of a) coupling of bromomagnesium-pyridine and Pg-N-protected azetidinone to yield Pg-N-protected 3- (pyridin-2-yl)azetidine-3-ol

b) reduction of the coupling product obtained in step a) in the presence of a metal catalyst to yield Pg- N-protected 3-(piperidin-2-yl)azetidine-3-ol c) Pg'-N -protection of the piperidine-N to yield the double N-protected 3-(piperidin-2-yl)azetidine-3-ol

and

d) optionally selectively N-de -protection of the azetidine-N to yield the Pg' -N-protected 3-(piperidin-2- yl)azetidine-3-ol

This route starts from a Grignard-functionalized pyridine, wherein the Grignard-moiety is used for coupling of the protected azetidine -moiety. This reaction can be performed at short reaction times nearly quantitative.

In step b) the coupling product is reduced to the non-aromatic piperidine ring in the presence of a metal. Suitable metal catalysts for the hydrogenation reaction are principally known to the skilled artisan and can for instance be selected from the group comprising Pd/C, Pt/C, Raney Nickel, formic acid/formaldehyde, L1AIH ₄ or mixtures thereof.

In step c) the piperidine-N is protected with a protection group Pg', wherein the Pg' denotes that the protection group is a different protection group than Pg. Consequently, both protection groups can be cleaved at different conditions or to a different extend. In certain cases it is helpful, that Pg and Pg' are orthogonal protection group in the meaning that the protection groups may be cleaved under orthogonal reaction conditions. It is for instance feasible that one of the protection groups is acid labile and the other protection group prone to base hydrolysis. Within such embodiment it is possible to attach and cleave such groups independently. Other orthogonal protection groups are known to the skilled artisan. B sed on the usage of different protection groups for independent protection of the different nitrogen it is possible in step d) to selectively N-de -protect the azetidine-N. The exact de -protection conditions can be selected as a direct function of the used protection group chemistry. Principally Pg-moieties might be selected from the group consisting of Boc, Cbz, Fmoc, biphenylmethyl, acetyl (Ac), benzoyl (Bz), p- methoxybenzyl carbonyl (Moz), Benzyl (Bn), carbamate, tosyl (Ts), trichloroethyl chloroformate (Troc), preferentially Cbz, Fmoc, Bn and Bz and principally Pg'-moieties might be selected from the group consisting of Boc, Cbz, Fmoc, biphenylmethyl, Acetyl (Ac), Benzoyl (Bz), p-methoxybenzyl carbonyl (Moz), Benzyl (Bn), carbamate, Tosyl (Ts), Trichloroethyl chloroformate (Troc), preferentially Boc. For the use in the subsequent steps the orthogonal protection properties have to be ensured by appropriate protective group selection.

In another preferred characteristic of the process Z in compound B can be cyano, the two protection groups Pg may be orthogonal protection groups Pg and Pg' and compound B can be obtained via a process at least comprising the step of: reacting N-protected azetidinone and orthogonal N-protected cc-cyano-piperidine to yield the double N- protected 2-(3-hydroxyazetidine-3-yl)-2 -cyano piperidine

This reaction can be performed at moderate reaction conditions without the use of pharmaceutically not acceptable solvents. Suitable pairs of orthogonal protection groups (Pg/Pg') might be selected from the pairs of Cbz/Boc, Boc/Cbz Boc/Bn, Bn/Boc, Fmoc/Boc, Boc/Fmoc, Ac/Boc, Boc/ Ac Boc/Triphenylmethyl, Triphenylmefhyl/Boc, Boc/Trifluoroacetyl, Trifluorocateyl/Boc.

Within a further preferred embodiment the cyano-group can be cleaved from the above disclosed addition product by a reducing agent, followed by subsequent kinetic resolution using chiral acids to obtain the enantiomerically pure (S)-stereo-isomer:

It has been found that especially this reaction product can be efficiently kinetically resolved and can be obtained at high chiral purity. However, it seems, that a better resolution is possible in cases, wherein the complete synthesis cycle is performed, i.e. coupling of the cyano-compound according to the above described scheme, de-cyanation of the product, followed by kinetic resolution. Suitable reducing agents for cleavage of the cyano-group may be selected from agents like complex borohydrides, complex Aluminium hydrides, activated Raney Nickel and alkali metals in ammonia. It is further preferred to use strong reducing agent selected from the group consisting of LiBH4, NaBH4, KBH4, Zn(BH4)2, NaBH(OAc ₃ ), NaCN(BH ₃ ), LiAlH ₄ , Activated Raney Nickel and Li or Na in NH ₃ . It is especially preferred to perform the kinetic resolution according to the following scheme, wherein the asterisks denote two chirality centres at the tartaric acid:

It seems that the tartaric acid is an efficient salt former with this intermediate. Very fast and efficient kinetic resolution processes have been performed according to the below mentioned scheme using a specific tartaric acid enantiomer

In a further embodiment of the process compound B' can be obtained via a process at least comprising the steps of a) reacting N-protected azetidinone and orthogonal N-protected a-cyano-piperidine to yield the cyclic N-protected 3'-oxotetrahydro-8a'H-spiro[azetidine-3, -[l,3]oxazolo[3,4-a]pyridine]-8a'-carbonitrile

b) followed by cleavage of the cyano group to yield compound B' in presence a reducing agent. Step b) of this reaction alternative is performed in the presence of a reducing agent. Suitable reducing agents are known to the skilled artisan, but it has especially been found that it is possible to significantly reduce the amount of side -products by the use of strong reducing agents like the alkali metals in ammonia. It is further preferred to use strong reducing agent selected from the group consisting of Li or Na in NH ₃ . In another aspect of the process different N-protection groups can be used for the educts and the azetidine-N can be protected by a Boc-group and the piperidine-N can be protected by a Cbz-group and the reaction step a) in above displayed reaction for achieving compound B' is performed in a one-step reaction in the presence of a complexing reagent and a non nucleophilic base. Especially the reaction in step a) of above displayed scheme (reacting N-protected azetidinone and orthogonal N-protected oc- cyano-piperidine to yield the cyclic N-protected 3'-oxotetrahydro-8a'H-spiro[azetidine-3,l'- [l,3]oxazolo[3,4-a]pyridine]-8a'-carbonitrile) can suitably be performed by using Boc and Cbz for orthogonal protection. The overall reaction can be performed as a one-step reaction in the presence of a complexing reagent and a non nucleophilic base. Suitable non nucleophilic bases may for instance be selected from the group consisting of lithiumdiisopropylamide, butyllithium, lithium tetramethylpiperidide, silicon-based amides, such as sodium and potassium bis(trimethylsilyl)amide or mixtures thereof. Lithiumdiisopropylamide and butyllithium being preferred.

Suitable chelating agents may for instance be EDTA or tetramethylethylendiamine (TMEDA), DBU, DMPU, wherein tetramethylethylendiamine is preferred.

Within another characteristic of the process the azetidine-N can be protected by a Cbz-group and the piperidine-N can be protected by a Boc-group and the reaction in step a) can be performed stepwise, wherein in a first reaction al) the azetidine and the piperidine are reacted in the presence of a complexing reagents and a non nucleophilic base to yield the double N-protected 2-(3-hydroxyazetidine-3-yl)piperidine-2-carbonitrile

followed by a2) cyclization of the 2-(3-hydroxyazetidine-3-yl)piperidine-2-carbonitrile in the presence of a base

In a further preferred embodiment of the process Z in compound B can be cyano and compound C can be obtained via a process at least comprising the steps of a) cyclization of a double N-protected compound B by addition of a base to yield tetrahydro-5'H- spiro[azetidine-3,l'-[l,3]oxazolo[3,4-a]pyridin]-3'-one (compound B')

B'

followed by b) coupling of compound B' and compound A to yield compound C

compound C c) followed by selectively opening of the oxazolidine-ring by treatment with a base to yield compound D. This reaction route is able to deliver compound C in high quantities and, additionally, is easily up- scalable. Furthermore, this step can be performed in a one pot reaction. Compound C is very stable and can for instance be stored before further processing to Cobimetinib free base or a pharmaceutically acceptable salt, like the Cobimetinib hemifumarate.

The base in step a) of above mentioned scheme can for instance be any inorganic or organic base, wherein the inorganic bases are preferred. Especially preferred are nucleophilic hydroxide ion containing or generating bases. Suitable bases can for instance be selected from the group consisting of NaBH ₄ , AI2O3, L12CO3, Na ₂ C0 ₃ , K ₂ C0 ₃ , Cs ₂ C0 ₃ , teri-BuOKor mixtures thereof.

The selectively opening of the oxazolidine-ring to yield compound D can be performed by contacting compound C with a base. Suitable bases which are able to selectively perform the ring-opening without generating further side -products are for instance LiOH, NaOH, KOH, LiH, NaH, KH or mixtures thereof. After obtaining compound D, Cobimetinib as a free base, it is further possible to resolve compound D with chiral acids like, but not limited to, tartaric acid o-Benzoyl-tartaric acid, lactic acid camphour-10- sulphonic acid, mandelic acid to result in an enantiomerically pure compound D. For chiral resolution especially L-tartaric acid is highly preferred. It is also possible to optionally perform a resolution step of compound B' and use the chirally defined intermediate in the subsequent synthesis steps.

In another characteristic of the process compound C can be treated after step b) with chiral acids to result in an enantiomerically pure compound C

followed by selectively opening of the oxazolidine-ring by treatment with a base to yield the enantiomerically pure form of Cobimetinib

It is also possible to introduce a chiral separation step on a process intermediate. Within the above mentioned alternative compound C is resolved to achieve the enantiomerically pure form. Possible reagents for achieving the enantiomerically pure forms are chiral acids. Possible chiral acids can be selected from the group consisting of tartaric acid o-Benzoyl-tartaric acid, lactic acid, camphour-10- sulphonic acid or mandelic acid. Suitable bases for selectively opening of the oxazolidine-ring can be selected from the group consisting of NaBtU, LiOH, NaOH, KOH, LiH, NaH, KH or mixtures thereof.

Within an additional aspect of the process prior to the reaction of compound B and compound A, compound B can be treated with chiral acids in order to obtain the essentially enantiomerically pure S- form, resulting in the formation of the enantiomerically pure form of compound D

In general, it is possible to process compound B to a pure enantiomerically form before performing the reaction with compound A in order to yield compound C. This reaction can be performed easily by reacting the racemic mixture with above mentioned chiral acids. In addition, this route of chiral resolution is able to be applied in all presented synthesis alternatives.

In a further preferred embodiment of the process prior to the reaction of compound B' and compound A, compound B' can be treated with chiral acids in order to obtain the essentially enantiomerically pure form of compound B', resulting in the formation of the enantiomerically pure form of compound D

The same process of chiral resolution is also applicable in the alternative processing route with compound B'. The enantiomerically pure form is obtained after coupling and ring opening. The same chiral acids can for instance be used for resolution and highly enantiomerically pure forms are obtainable.

In an additional aspect of the process the reaction can be performed in presence of a chiral catalyst to yield essentially the R-configurated cyano-piperidine

The step of chiral resolution can further be performed on different reaction intermediates. It is for instance possible to resolve the cyano-piperide by using a chiral acid, e.g. tartaric acid, o-Benzoyl- tartaric acid, lactic acid, camphour-10-sulphonic acid or mandelic acid.

In a further preferred embodiment of the process any educt or intermediate comprising a chirality center is kinetically resolved with chiral acids to yield the enantiomerically pure stereo-isomer. Surprisingly it has been found that it is possible to achieve enantiomerically pure Cobimetinib by using a simple kinetic resolution step comprising chiral acids. By this process a fast and cost efficient resolution is possible compared to the proposed resolution steps in the literature. Such finding is surprising, because the educts and intermediate are known for forming salt intermediates only sparingly.

In another preferred embodiment either the intermediate compounds C, C, B, B' or the resulting cobimetinib is kinetically resolved with chiral acids to yield the enantiomerically pure stereo-isomer. The kinetic resolution is especially effective for these compounds, rendering the overall resolution process very cost effective and easy.

An intermediate in the synthesis of pharmaceutically acceptable Cobimetinib comprising

is further within the scope of invention, wherein Pg are independently the same or different protection groups or H; or an essentially enantiomerically form or salt thereof. The protection groups can be independently selected from the group consisting of Boc, Cbz, Fmoc, biphenylmethyl, acetyl (Ac), benzoyl (Bz), p-methoxybenzyl carbonyl (Moz), Benzyl (Bn), carbamate, tosyl (Ts), trichloroethyl chloroformate (Troc).

An intermediate in the synthesis of pharmaceutically acceptable Cobimetinib comprising

or an essentially enantiomerically pure form or salt thereof is further within the scope of invention. The chemical purity of this intermediate should be more than 99%, preferentially more than 99.5 % and further preferred more than 99.9%%. The intermediate is enantiomerically pure if the intermediate comprises a chiral purity of more 99% ee, preferentially more than 99.5 % ee and further preferred more than 99.9%% ee. The purity and enantiomerically purity determination can be performed via HLPC and chiral HPLC methods, respectively. The exact methods are known to the skilled artisan. The same definition also applies for the following compounds and process intermediates.

In addition, also the compound N-tert-butyloxycarbonyl-3-(2-cyano-piperidin-2-yl)-N'- benzyloxycarbonyl-azetidin-3-ol according to the following formula

or an essentially enantiomerically pure form or salt thereof is within the scope of invention.

Furthermore, it is within the scope of the invention to disclose a tartrate salt of N-tert-butyloxycarbonyl- 3-(piperidin-2-yl)-N'-benzyloxycarbonyl-azetidine-3-ol or an essentially enantiomerically pure form thereof. Preferably, the tartrate salt may be in the form of an (L)-tartrate salt.

Furthermore, an intermediate in the synthesis of a pharmaceutically acceptable Cobimetinib comprising

or an essentially enantiomerically pure form thereof is within the scope of invention.

Furthermore, an intermediate in the synthesis of a pharmaceutically acceptable Cobimetinib comprising

or an essentially enantiomerically pure form thereof, is within the scope of invention. Also the enantiomerically pure isomers or a racemic mixture of the above displayed compound are within the scope of the invention.

Additionally, an intermediate in the synthesis of a pharmaceutically acceptable Cobimetinib comprising

or an essentially enantiomerically pure form thereof is within the scope of invention. In addition, an intermediate in the synthesis of a pharmaceutically acceptable Cobimetinib comprising

or an essentially enantiomerically pure form thereof is within the scope of invention.

Furthermore, the compound tetrahydrospiro[azetidine-3,l'-oxazolo[3,4-a]pyridin]-3'(5'H )-one according to the following formula

or an essentially enantiomerically pure form thereof is within the scope of invention.

In addition, an intermediate in the synthesis of a pharmaceutically acceptable Cobimetinib comprising

or an essentially enantiomerically pure form thereof is within the scope of invention.

Furthermore, the compound 3'-oxohexahydrospiro[azetidme-3,l'-oxazolo[3,4-a]pyridi carbonitrile according to the following formula

or an essentially enantiomerically pure form thereof is within the scope of invention.

In addition, an intermediate in the synthesis of a pharmaceutically acceptable Cobimetinib comprising

or an essentially enantiomerically pure form thereof is within the scope of invention.

Also the compound l-(3,4-difluoro-2-((2-fluoro-4-iodophenyl)amino)benzoyl)tetr ahydrospiro- [azetidine-3,l'-oxazolo[3,4-a]pyridin]-3'(5'H)-one according to the following formula

or an essentially enantiomerically pure form thereof is within the scope of invention.

It is additionally within the scope of invention to disclose a process for the production of amorphous Cobimetinib hemifumarate salt at least comprising the steps of a) dissolving Cobimetinib free base in a non-polar solvent,

b) dissolving fumaric acid in a polar solvent,

c) combining the two solutions and

d) evaporation of the solvents.

It is very difficult to obtain Cobimetinib hemifumarate in a solid form and it is even more difficult to obtain Cobimetinib hemifumarate in a defined macroscopic form, i.e. either in a crystalline or a salt form. Several tests to precipitate the salt form from different solvents failed, either resulting in no or only insufficient precipitation. Surprisingly it was found that is possible to yield amorphous material at fairly high yield by above described precipitation process. In addition, the amorphous material did show a much better dissolution kinetics compared to a crystalline variant, indicating that bioavailability of the amorphous may be superior compared to the crystalline form.

Suitable non-polar solvents in step a) can be selected from the group of ether, ester, chlorinated hydrocarbons, ketones or mixtures thereof. Suitable polar solvents in step b) can be alcohol, formamide, water, sulfoxide or mixtures thereof. The combining of the solutions in step c) can be performed by rapid addition of the Cobimetinib solution in the fumaric acid solution under stirring. It is preferred that the combined solutions are at least stirred for further 30 minutes. This might increase the precipitation rate and yield and results in the amorphous form.

The evaporation of the solvent can be performed at atmospheric or reduced pressure. It is preferred to perform the evaporation at reduced pressure at elevated temperatures in the range of room temperature up to 70°C, preferably in the range of 30°C to 50°C.

It is further within the scope of invention to disclose cobimetinib hemifumarate salt obtained by the inventive process comprising at least a salt content of > 50% and < 100% as determined by XPS- analysis. It has additionally been found, that not all precipitation steps yield Cobimetinib hemifumarate in a salt state. Very often it was found that not a true salt precipitated but a physical mixture of (separated) Cobimetinib free base comprising clusters of fumaric acids. By using the above described precipitation process it is possible to generate the true and not only the physical mixture. The salt content of the precipitate can be obtained by XPS -analysis by differentiating the signals of N Is (1) of non protonated specie vs. N Is (3) of protonated specie (see experimental section below). The ratio of Cobimetinib hemifumarate in the salt state and Cobimetinib hemifumarate comprising larger amount of Cobimetinib free base and separated clusters of fumaric acid might drastically influence the solubility, the stability and the bioavailability of the drug form.

In another aspect the use of the inventively synthesised and processed Cobimetinib hemifumarate salt for manufacturing of a pharmaceutical composition is within the scope of invention. Especially, the inventively proposed route of synthesis and, in addition, also the inventively proposed route of preparation of the amorphous Cobimetinib hemifumarate render the obtainable product suitable for the use in a pharmaceutical composition. The suitability might be attributed to the low amount of unwanted side products, the reproducibility and, in addition, also the defined and reproducible salt content of the final product.

In addition, an oral dosage form comprising a Cobimetinib hemifumarate salt inventively synthesised and/or processed is within the scope of invention.

Furthermore, the use of an oral dosage form comprising a Cobimetinib hemifumarate salt inventively synthesised and/or processed for the treatment of metastatic melanoma is within the scope of invention. BRIEF DESCRIPTION OF THE DRAWINGS

The invention at hand is further exemplified in certain embodiments, wherein the figures 1 to 9 show:

Fig. 1 the inventive route of synthesis for obtaining Cobimetinib free base;

Fig. 2 one possible route of synthesis for obtaining compound B via intermediates B 1-B4, wherein

Z = H;

Fig. 3 one possible route of synthesis for obtaining different chiral forms of compound B';

Fig. 4 two possible routes of synthesis for obtaining compound B', wherein Z CN;

Fig. 5 one possible route of synthesis for obtaining compound B;

Fig. 6 a PXRD diffraction pattern of amorphous Cobimetinib hemifumarate prepared according to the invention;

Fig. 7 a peak-fitting result of an XPS-pattern of amorphous Cobimetinib hemifumarate prepared according to the invention;

Fig. 8 a ^-NMR-sopectrum of amorphous Cobimetinib hemifumarate prepared according to the invention;

Fig. 9 a DSC-spectrum of amorphous Cobimetinib hemifumarate prepared according to the invention.

Figure 1 displays the inventive route of synthesis. In this route a compound A is reacted either with compound B or compound B' in order to yield either compound C or compound C. In a later step the compounds C or C are converted to Cobimetinib free base or a pharmaceutically acceptable salt thereof. The processing of defined chiral compounds is also possible in every step of the synthesis either by using chiral educts or by chiral work-up of the products. The process routes A + B→C→D or A + B' → C→ D are parallel routes only differing in the chemistry of educt B. It is possible to convert the core compounds B to B' and vice versa as depicted by the dotted arrow. In certain cases, it might be necessary to adjust the protective groups afterwards.

Figure 2 shows on possible route of synthesis in order to obtain double N-protected compound B in the variant Z=H. This synthesis is a 5 step synthesis starting from the educt piperidine-l-carboxylic acid and yields compound B via the intermediates B1-B4. This route includes functionalization of the carboxy-group, followed by building an oxirane-ring, ring opening of the oxirane-ring and azetidine- ring formation. The piperidine ring protection can be introduced in every synthesis step or at the end of the synthesis. Figure 3 exhibits a possible reaction scheme to obtain compound B or B', respectively. Two orthogonal N-protected educts (an azetidine derivative and a piperidine derivative, wherein Z = cyano) are reacted either under chiral or non-chiral conditions. Consequently, either a defined stereo-isomer or a racemic mixture will be obtained after the first reaction step. The intermediate may either be reacted under cyclization (left and right arm of the reaction tree) to yield compound B' either in a defined stereo- conformation or as a racemic mixture. In the case of the formation of a racemic mixture it is possible to obtain stereo-isomeric defined products by treatment of the mixture with chiral acids. It is also possible to cleave the cyano-group after the first step from the piperidine -ring and selectively de -protect the azetidine-N to obtain the mono-N-protected compound B either in a defined stereo-conformation or as a racemic mixture.

Figure 4 displays two possible ways of reacting the azetidine and the cyano-functionalized piperidine educt. The reaction can either be performed in one or two steps. The one step reaction directly yields the cyclization product, wherein in the two step reaction firstly the different ring-structures are linked and the cyclization is performed in the second reaction step.

Figure 5 displays a possible Grignard -route for obtaining compound B. A halo-functionalized pyridine is treated with a Grignard-reagent in order to obtain the Grignard functionalized pyridine. This compound is reacted with a Pg-N-protected azetidinone followed by reduction of the aromatic pyridine ring. In a next step the piperidine-N can be protected followed by de -protection of the azetidine-N protection group.

Figure 6 displays a X-ray powder diffractogram (PXRD) of the amorphous Cobimetinib hemifumarate obtained by a process according to the invention. The measurement conditions are described below in the methods -section. Based on the broad halo it is clearly visible that the obtained Cobimetinib hemifumarate is amorphous comprising no ordered, crystalline structures.

Figure 7 displays the N Is fitting result of a XPS-spectrum of amorphous Cobimetinib prepared according to the invention. This experiment is performed to demonstrate that the inventive route of amorphous salt generation is able to yield the hemifumarate salt and not only a physical hemifumarate/Cobimetinib mixture as is obtained by most processing routes. Cobimetinib comprises three different nitrogen -atoms (displayed below) and, in addition, one of the nitrogen atom is prone to protonation. The different nitrogen -atoms can be distinguished in an XPS-spectrum, based on an absorption at a different energy. The different nitrogen atoms and their designations are depicted in the following formula:

The N Is (1) nitrogen in the piperidine ring can be attributed to a signal at 398.6 eV (https://srdata.nist. gov/xps/query_chem_name_detail.aspx?ID_NO=1224&CName=pip eridinp).

The N Is (2) at the diphenylamine can be attributed to a signal at 400.2 eV (https://srdata.nist. gov/xps/query_chem_name_detail.aspx?ID_NO=20459&CName=2- %28diphenylamino%29-5-%28%283-ethyl-2 283H 29-benzothiazolylidene%29ethyli-dene 29- 4 285H 29-thiazolone).

The N Is (2) at the benzamide can be attributed to a signal at 400.2 eV (https://srdata.nist. gov/xps/query_chem_name_detail.aspx?ID_NO=2422&CName=ben zamide). Therefore, the diphenylamine N Is and the benzamide N Is are forming only one single peak at the same energy.

The N Is (3) for the protonated specie can be attributed to a signal at 401.4 eV (dx.doi.org/10.1021/mp300371 f I Mol. Pharmaceutics 2013, 10, 211-224)

For the peak-fitting of the XPS-spectrum symmetric Gaussian-Lorentzian peaks (GL(30), i.e. GL(100) is purely lorentzian and GL(0) is purely gaussian) were used and a Shirleey background for all regions. The data were used as recorded without further smoothing operations. In addition, further major restrictions in the peak-fitting were omitted. The fitting routine includes the following assumptions:

- FWHM are equal for all specie in the same spectral region,

- only 1 F ls-specie;

- only one I 3d5/2-spezie,

- three O ls-spezie (excluding oxygen contamination, O Is (1) for C-0 at 531,5 - 532 eV, O Is (2) for C=0 at approx. 533 eV and O ls(3) for a deprotonated oxygen at the fumaric acid (designation according http://xpssimplified.com elements/oxygen.php).

The quantification is based using an average of 5 scans for the respective specie. The following intensities can be attributed to the different nitrogen: Specie peak position %At cone.

N Is (1) 397.89 + 0.19 eV 9.83 + 0.96

N Is (2) 399.83 ± 0.24 eV 69.23 + 2.65

N Is (3) 401.31 + 0.28 eV 20.94 + 2.06

Based on the peak-fitting results of the N ls-region it is obvious that Cobimetinib fumaric acid salt is obtained. The region attributed to the protonated specie increases in intensity and the intensity in the region attributed to the piperidine N Is specie is reduced, clearly demonstrating the protonation of the piperidine. Based on the ratio of the different N ls-specie a degree of salt formation can be calculated. The ratio of Nls(l) to Nls(3) is approximately 32% to 68%, indicating that the degree of salt formation is in the range of 70%.

Figure 8 displays a ^-NMR-spectrum of amorphous Cobimetinib hemifumarate. The integrated signal intensities of Cobimetinib (8.52/8.49 ppm) to fumaric acid (6.43 ppm) indicate a stoichiometric ratio of approximately 1 :0.5.

Figure 9 displays a DSC-curve of the melting behaviour of amorphous Cobimetinib hemifumarate. The Data are obtained in a temperature interval from approximately 20°C to 350°C and the temperature gradient is lOK/min. The visible phase transitions of the amorphous Cobimetinib hemifumarate can be attributed to amorphous - glass transition - crystallization. The peak at 300 °C indicates the decomposition of the compound.

EXPERIMENTAL EXAMPLES

I. Methods I.a PXRD-diffraction

The samples for powder X-ray diffractometry (PXRD) were filled in standard glass vials (0 = 0.7 mm). All measurements were performed at room temperature using a D8 Bruker Advance Diffraktometer (Cu- Kal = 1.54059 A, Johansson primary beam monochromator, position sensitive detector) in transmission geometry and a rotating sample. Data were collected in a range of 2Theta of 3-40° at 40 kV and 20 mA.

I.b ¾ -NMR ^: H-NMR measurements were performed at room temperature in a deuterated solvent (d6 -DMSO) using a Bruker Avance DRX-400 as well as a Bruker Avance DPX-300 spectrometer Bruker, Rheinstetten (¾ 400 or 300 MHz). The chemical shift δ is given in ppm relative to the used frequency, wherein for the ^-NMR spectra the remaining signal of the deuterated solvent was used as internal standard.

I.c XPS-measurements

XPS-measurements were performed using a specially equipped near ambient pressure XPS (NAP-XPS), Specs Surface Nano Analysis GmbH, using a hemispheric analyzer (Phoibos 150) and a 2D DLD detector, wherein a 300 μιη Skimmer was used as analyzer-input. The angular dispersion was 22° and the distance between sample and skimmer was set to 300 μηι. The X-ray source (μ -Focus 500) was oriented 55° with respect to the analyzer-axis and operated at 15 kV, 9.2 mA, collimator- alignment to spot 1 at UL= 1500 V, monochromatic X-rays (Al-Ka 1486.71 eV). The resolution was 0.706 eV FWHM of the Ag 3d5/2 at 368.235 eV. For recording and analysis Specs Lab Prodigy Version 4.14.1- r52157 und CasaXPS Version 2.3.17 PR1.1 was used.

II. Synthesis

7¾ri-Butyl 2-cyano-l-piperidinecarboxylate was synthesized from piperidine-2-carboxylic acid via modified procedures [Org. Prep, and Procedures Int., 39 (5) 503-508 (2007); Synthetic Communication, 37: 3793-3799, (2007); European Journal of Medicinal Chemistry 75, 111-122, (2014)] allowing to get higher yields at milder conditions, compared to the original ones.

N-Cbz-azetidin-3-one was synthesized starting from epichlorohydrin and benzylaniine via oxidation of the corresponding alcohol [Org. Process Res. Dev. 15, 462-466, (2011)].

Compound A - 3,4-Difluoro-2-(2-fluoro-4-iodophenylamino)benzoic acid, was synthesized from 2- fluoro-4-iodoaniline and 2,3,4-trifluorobenzioc acid according to Nat. Chem. 7, 7, 554-561 (2015). l-teri-Butoxycarbonyl-2-piperidinecarboxylic acid

To a stirred mixture of 500g of D,L-piperidine-2-carboxylic acid (leq.), 3000mL of ethanol and 750mL of water 495g (1.53eq.) of sodium bicarbonate and 1216g (1.44eq.) of B0C ₂ O were added. Reaction mixture stirred at ambient temperature overnight, ethanol concentrated in vacuum and to the residue 2800mL of ethyl acetate were added. Mixture was cooled to 20°C and 2235g of 3N HC1 aqueous solution added dropwise, maintaining the temperature below 30C, followed by concentration to ca. 2/5 of the initial volume. To the residue 2500mL of n-heptane were added and concentrated to ca. 3/5 of the volume. Mixture was cooled to 0°C and precipitated white solid filtered off with suction, dried in vacuum to give 884.8g of the title product (100% HPLC peak area) with a yield of 99.6%. tert-Buty\ 2-carbamoyl-l-piperidinecarboxylate

To a mixture of 410.6g (leq.) of l-tert-Butoxycarbonyl-2-piperidinecarboxylic acid, 228.8g (1.62eq.) of ammonium bicarbonate and 605.9g (2.77eq.) of Boc20 377.9 g (2.67eq.) of pyridine and 2000mL of dioxane were added. Reaction mixture stirred overnight at ambient temperature and 1653.3g of 3N HC1 aqueous solution was added drop wise followed by addition of 2100mL of DCM, stirring and phase separation. Organic phase was concentrated and 1700mL of n-heptane were added to the residue and stirred at ambient temperature. Precipitated white solid was filtered off with suction, filter cake washed with n-heptane and dried in vacuum to give 304.5g of the title product (99% HPLC peak area) with a yield of 88.2%. tert-Butyl 2-cyano-l-piperidinecarboxylate

To a solution of 50g (leq.) of ierf-Butyl-2-cyano-l-piperidinecarboxylate in 1150mL of DCM 88.64g (4eq.) of triethylamine was added with stirring and solution was cooled down to 0°C followed by dropwise addition of a solution of 92g (2eq.) TFAA in 265mL of DCM, maintaining internal temperature at 0-3 °C. After complete addition reaction mixture allowed to warm to ambient temperature and washed with water two times, then concentrated to ca. 1/5 of the initial volume and to the residue 250mL of n- Heptane added. 150mL of the solvent was distilled off and mixture was cooled down to 0°C while stirring. Precipitated material was filtered off with suction, the filter cake washed with cold n-heptane and dried in vacuum to give 33.48g of the title compound (97.7% HPLC peak area) with a yield of 73%.

tert-Butyl 2-[l-(benzyloxycarbonyl)-3-hydroxy-3-azetidinyl]-2-cyano-l-p iperidinecarboxylate

A mixture of 13.4 g of N-Boc-2-cyano-piperidine (1 eq., 92% HPLC purity), 6.83 g TMEDA (1 eq.) and 100 mL THF was cooled to -78°C under vigorous stirring. 51 mL of LDA 1.5 M solution (1.3 eq.) in THF were added dropwise in the course of 50 min, maintaining the internal temperature < -73 °C, and additionally stirred for 15 min at -78 °C. To the solution 12.06 g (1 eq.) N-Cbz-azetidine-3-one in 50 mL THF was added dropwise in the course of 35 min, also maintaining internal temperature < - 73 °C. The reaction mixture was stirred for 2 h at -78 °C, followed by quenching with 20 mL of ammonium chloride saturated aqueous solution. The reaction mixture was allowed to warm to ambient temperature. The phases were separated and the aqueous phase three times extracted with 10 mL ethyl acetate. The combined organic washings were dried over magnesium sulfate, filtered and concentrated to a volume of approx. 50 mL. A precipitate formed upon solvent removal. 50 mL MTBE was added and the forming red-brown solution was stored at -15°C for 16 h. The precipitated solid was filtered off applying mild suction and the forming filter cake was washed five times with 20 mL MTBE until the cake became almost white. The cake was dried in vacuum to give 14.73 g of the title product (93.2% HPLC peak area) with a yield of 56.6%. Additional 2.6 g of the material (with lower HPLC peak area) could be isolated from the filtrate. tert- utyl 2-[l-(benzyloxycarbonyl)-3-hydroxy-3-azetidinyl]-l-piperidin ecarboxylate

To a heated solution (70 °C) of 3.8 g (4.09 eq.) acetic acid in 50 mL of anhydrous ethanol a suspension of 6.43g (1 eq.) terf-Butyl 2-[l-(benzyloxycarbonyl)-3-hydroxy-3-azetidinyl]-2-cyano-l -piperidine- carboxylate and 1.95 g (2 eq.) sodium cyanoborohydride in 80 mL of anhydrous ethanol is added dropwise in the course of 10 min. The reaction mixture is further stirred vigorously at 70 °C. The reaction progress is monitored via HPLC and a full conversion of the starting material was achieved in 30 h. The solvent was evaporated and the residue suspended in 50 mL of toluene and washed with 50 mL of sodium hydroxide 10% aqueous solution. The phases were separated and the aqueous phase discarded. The organic phase was washed with water until the pH of the aqueous washings became 8. The organic phase was dried over magnesium sulfate, filtered and evaporated to dryness to give 5.57 g of the title product as yellowish sticky mass (yield 80%).

(R,S) tert-Butyl 2-(3-hydroxy-3-azetidinyl)-l-piperidinecarboxylate ((R,S) compound B)

A mixture of 2.11 g ferf-Butyl 2-[l-(benzyloxycarbonyl)-3-hydroxy-3-azetidinyl]-l -piperidine- carboxylate, 0,41 g Pd C and 30 mL ethanol was stirred under 3.5 bar ¾ pressure. After 2 h reaction time a HPLC showed full conversion of the starting material. The catalyst was filtered off, the filtrate evaporated in vacuum to give 1.27g of the title compound as white solid (yield 95.8%).

(R,S) tert-Butyl 2-{l-[3,4-difluoro-2-(2-fluoro-4-iodophenylamino)benzoyl]-3- hydroxy-3- azetidinyl}-l-piperidinecarboxylate ((R,S)- Compound C)

A mixture of 0.052 g (1.25 eq.) HOBt hydrate, 0.062 g (1.18 eq.) EDC hydrochloride, 0.13 g (1.2 eq.) compound A (Act = OH) and 10 mL acetonitrile was stirred at ambient temperature under nitrogen for 1 h followed by addition 0.07 g (1 eq.) feri-Butyl 2-(3-hydroxy-3-azetidinyl)-l-piperidinecarboxylate suspended of in 10 mL acetonitrile. The solution was stirred overnight at ambient temperature. HPLC of the reaction mixture showed full conversion of the starting materials. The solvent was evaporated in vacuum and the residuum was suspended in 10 mL DCM and washed three times with 10 mL of water. The organic phase was dried over magnesium sulfate, filtered and evaporated to dryness to give 0.22 g of a brownish powder. After carbon treatment 0.17 g of the title compound was obtained as beige solid with 96.35% HPLC peak area (yield 95%).

Synthesis of (/?,S)-Cobimetinib hydrochloride

To a solution of 0.17 g ferf-Butyl 2-{ l-[3,4-difluoro-2-(2-fluoro-4-iodophenylamino)benzoyl]-3- hydroxy-3-azetidinyl}-l-piperidinecarboxylate in 7 mL methanol 0.23 mL of 36% aqueous HCl was added and stirred for 3 h. The solvent was evaporated and a yellowish solid residue triturated with diethyl ether to result in 0.15 g of the title product as beige powder with 96.2% HPLC peak area (yield 98%). Synthesis of (/i,S Cobimetinib free base

To a stirred suspension of 1.26 g of cobimetinib hydrochloride in 25 mL of diethyl ether sodium bicarbonate aqueous solution added (1 g sodium bicarbonate in 30 mL water). Biphasic solution stirred vigorously for 2 h. Phases separated, aqueous phase washed twice with 15 mL of diethyl ether. Combined organic washings dried over magnesium sulfate, filtered and evaporated in vacuum to give 1.1 lg of beige solid with 85% HPLC peak area for the title compound (Yield 95%). Further recrystallization from DCM/acetonitrile 1:3 mixture gave product with 98.9% HPLC peak area.

tert-Butyl S-2-(3-hydroxy-3-azetidinyl)perhydropyridine-l-carboxylate (LR,2 ?)-tartrate salt (Compound B L- tartrate)

To a suspension of lg (2eq.) of the compound B in 25mL of THF, 5mL of methanol were added with stirring to form a clear, colorless solution followed by addition of a solution of 0,293g (leq.) of L-tartaric acid in 2mL of methanol. Reaction mixture became turbid in 10 minutes then stirred for 20 minutes and stayed for 17 h at room temperature. Precipitated white solid filtered off with suction, washed three times with overall 15 mL of THF and dried in vacuum affording 0,65g of a crystalline solid with 63,7% EE of the title compound. Above material was suspended in 25mL of THF, heated to reflux and 14,5mL of methanol were added dropwise until slightly turbid solution was formed. The mixture was stirred under reflux for 20 minutes, then stirring stopped and allowed to cool to ambient temperature. After staying for 19 h at room temperature, precipitated white crystalline solid was filtered off with suction, washed with overall 15mL of THF and dried in vacuum to give 0,31g of the title product (EE 98,85%, yield 48%). tert-Butyl (7i)-2-(3-hydroxy-3-azetidinyl)perhydropyridine-l-carboxylat e (2S,2S)-tartrate salt (Compound B D-tartrate)

To a suspension of lg (2eq.) of the compound B in 25mL of THF, 5mL of methanol were added with stirring to form a clear, colorless solution followed by addition of a solution of 0,293g (leq.) of D- tartaric acid in 2mL of methanol. Reaction mixture became turbid in 3 hours then stopped and allowed to stay for 17 h at room temperature. Precipitated white solid filtered off with suction, washed three times with overall 15 mL of ethyl acetate and dried in vacuum affording 0,6g of a crystalline solid with 71% EE of the opposite enantiomer. Material precipitated from the filtrate was isolated by suction, washed with overall 15mL of ethyl acetate and dried in vacuum to give 0,23g of almost racemic material. The remaining filtrate was evaporated to dryness. Residual viscous mass washed with diethyl ether to give 0,4g of a white solid with 93% EE of the title compound. The above material taken up in lOmL of THF under reflux and solution of 0.084g of L-tartaric acid in lmL of methanol added with stirring. The mixture became turbid in 5 minutes. The mixture was allowed to cool to room temperature and stayed for 19h. Precipitated solid filtered off with suction affording 0,19g of the title product (EE 100%).

tot-Butyl (S)-2-{l-[3,4-difluoro-2-(2-fluoro-4-iodophenylamino)benzoyl ]-3-hydroxy-3-azetidi- nyl}perhydropyridine-l-carboxylate (S-Compound C)

A mixture of 0.087g (2,5 eq.) HOBt hydrate, 0.102g (2,36 eq.) EDC hydrochloride, 0.178g (2 eq.) compound A (Act = OH) and 10 inL acetonitrile was stirred at ambient temperature under nitrogen for 1 h followed by cooling to 5°C and addition of 0.15g (1 eq.) of tert-Butyl (lS)-2-(3-hydroxy-3- azetidinyl)perhydropyridine-l -carboxylate (lR,2J?)-tartrate salt suspended of in 10 mL of acetonitrile. The solution was stirred overnight at ambient temperature. HPLC of the reaction mixture showed full conversion of the starting materials. The solvent was evaporated in vacuum and the residue was suspended in 10 mL DCM and washed three times with lOmL of water. The organic phase was dried over magnesium sulfate, filtered and evaporated to dryness to give 0.31 g of a beige viscous mass which was used in the next step without further purification.

Synthesis of (S)-Cobimetinib hydrochloride

To a solution of 0.29 g (leq.) tert-Butyl (lS)-2-{ l-[3,4-difluoro-2-(2-fluoro-4- iodophenylamino)benzoyl]-3-hydroxy-3-azetidinyl}perhydropyri dine-l-carboxylate, prepared by the above procedure, in 10 mL methanol 0.38 mL of 36% aqueous HC1 (lOeq.) was added and stirred for 24h. The solvent was evaporated and a yellowish solid residue triturated with diethyl ether to result in 0.2 g of the title product as beige powder with 85,4% HPLC peak area (yield 65,6%).

Synthesis of (S)-Cobimetinib free base

To a mixture of 0,2g of Cobimetinib hydrochloride (leq.), prepared by the above procedure, and 0,06g of sodium bicarbonate (2eq.) 10 mL of methanol were added with stirring to form a light-pink turbid solution. Reaction mixture stirred for 15min and evaporated to dryness. To the residue lOmL of water added and extracted three times with overall 30mL of ethyl acetate. Combined organic washings dried over magnesium sulfate, filtered and evaporated to dryness to give 0,18 g of a crude product. Recrystallisation from 2mL DCM/acetonitrile 1 :3 mixture afforded 0, lg of pure product (HPLC peak area of the title compound 98,73%, EE 99,5%). Synthesis of compound B'

To a solution of 4.76 g (1 eq.) feri-Butyl 2-[l-(benzyloxycarbonyl)-3-hydroxy-3-azetidinyl]-2-cyano-l- piperidinecarboxylate in 40 mL THF a solution of 2.8g (2.3 eq.) of sodium bicarbonate in 10 mL of water was added and the biphasic mixture heated under reflux with vigorous stirring. The reaction progress was monitored by TLC and after 20 h full conversion of the starting material was achieved. The reaction mixture was cooled to ambient temperature, the phases were separated and the aqueous phase extracted with 10 mL THF. The combined organic washings were dried over magnesium sulfate, filtered and concentrated in vacuum. A brownish residue was obtained and washed three times with MTBE. The residue was dried in vacuum to give 2.91 g of the title product as beige solid in a yield of 65%.

l-(benzyloxycarbonyl)-2-piperidinecarboxylic acid

A mixture of 5.8 g (1 eq.) of D,L-pipecolinic acid and 234 mL of a 10% NaOH aqueous solution was stirred until a clear colorless solution formed. The solution was cooled to 0 °C and a solution of 11.52 g (1,5 eq.) benzylchlorformate in 234 mL THF was added dropwise. The internal temperature was not higher than 0 °C. The reaction mixture was allowed to warm to ambient temperature and the solution was further stirred for 3 h. The reaction mixture was washed with diethyl ether two times and the ethereal washings were discarded. The aqueous phase was acidified with 225 mL of 2.5N HC1 and three times extracted with 200 mL ethyl acetate. The combined organic washings were dried over magnesium sulfate, filtered and evaporated in vacuum to yield 9.21g of the title product as yellow-orange oil (yield 75%).

1-benzyl 2-methyl 1,2-piperidinedicarboxylate

To a solution of 9.41 g (1 eq.) l-(Benzyloxycarbonyl)-2-piperidinecarboxylic acid in 200 mL DCM 0.6 g HOBt (0.11 eq.) and 7.45 g (1.1 eq.) EDC hydrochloride were added and the mixture was stirred vigorously at ambient temperature for 1.5 h. To the solution was added 40 mL methanol. The mixture was stirred overnight, extracted with water, and the organic phase dried over magnesium sulfate. The dry solution was filtered and concentrated in vacuum. The resulting residue was purified by chromatography (ethyl acetate/nTieptane 1 : 1), yielding 6.6 g of the title compound as colorless oil (yield 61%). benzyl 2-(2-chloroacetyl)-l-piperidinecarboxylate

To a cooled solution (-78°C) of 0.42 g (2.2 eq.) chloro-iodo-methane in 10 mL THF, 1.58 mL (2.2 eq.) of LDA 1.5 M solution in THF was added dropwise in a nitrogen atmosphere. The temperature was maintained below -72°C. After stirring for additional 10 min at -78 °C a solution of 0.3 g (1 eq.) of 1- benzyl 2-methyl 1,2-piperidinedicarboxylate in 3 mL THF was added dropwise. The internal temperature was maintained not higher than -72 °C. After stirring for additional 5 min 1.08 mL (1.5 eq.) of LDA 1.5 M solution in THF was added dropwise. The internal temperature was maintained below - 72°C. After stirring for further 30 min at -78°C the reaction mixture was quenched with a solution of 1.2 g acetic acid in lOmL THF. The reaction mixture was allowed to warm to ambient temperature and 20 mL of water was added. The organic and aqueous phases were separated and the aqueous phase 3 times extracted with 25 mL of ethyl acetate. The combined organic washings were dried over magnesium sulphate, filtered, the solvents evaporated in vacuum and flashed chromatographed using heptane/dioxane 4: 1 mixture to give 0.15 g of the title product as almost colorless oil (Rf 0.12, yield 47%).

Preparation of amorphous Cobimetinib hemifumarate

Amorphous Cobimetinib hemifumarate was obtained by dissolving Cobimetinib free base (500.3 mg) in ethanol (28 ml) to form a solution I. Fumaric acid (54,826 mg, resulting in a molar ratio Cobimetinib: fumaric acid of 2: 1) was completely dissolved in ethanol (4 ml) to form a solution II. Solution II was combined with solution I and in order to assure the complete transfer of solution II the vial was rinsed with further 1 ml ethanol. No precipitation was visible. The combined solution was stirred for 30 minutes and the solvent was removed at reduced pressure and higher temperature using a rotary evaporator. A precipitate was obtained and dried at 20 h at 50°C in vacuo. 206.5 mg of the white solid were treated in a mill (3x10 minutes) and at the end of the procedure 120.3 mg of a white, amorphous solid were obtained.