T 1728/07 (Florfenicol synthesis/SCHERING) 11-10-2010

Reasons for the Decision

1. The appeal is admissible.

2. Amendments (Articles 100(c), 123(3), 123(2) EPC)

2.1 Amendments of the structural formulae

2.1.1 The sole objection raised under Article 100(c) EPC concerned the question of whether or not the amendments of the structural formulae representing oxazoline derivatives, found to be allowable under Rule 88 EPC 1973 (now Rule 139 EPC 2000) during examination proceedings, introduced subject-matter extending beyond the content of the application as filed. Since these formulae are still present in the main request (cf. points I and X above), it must be decided whether these amendments represent corrections of an obvious error within the meaning of Rule 139 EPC.

2.1.2 Throughout the patent application as originally filed, the compounds of formula (i) are depicted as bearing an amino substituent at position 4 of the oxazoline ring, as illustrated by the following formula appearing in claim 1 as originally filed (see also claims 3, 4; page 2, line 1; page 6, lines 1, 2; page 8, lines 12 to 14, page 9, lines 10 to 12):

FORMULA/TABLE/GRAPHIC

In the corresponding formulae disclosed in the main request, and in the specification of patent in suit, the amino substituent at position 4 of the oxazoline ring (H2N-) has been replaced by hydrogen (H-) (cf. points I and X above, and also patent in suit, paragraphs [0003], [0013], [0022], [0024]).

According to Rule 139 EPC, second sentence, "the correction must be obvious in the sense that it is immediately evident that nothing else would have been intended than what is offered as the correction". This means that it must be immediately apparent to the skilled person that (i) an error had occurred and (ii) how it should be corrected.

2.1.3 Concerning requirement (i), it is noted that the application as originally filed discloses the oxazoline compounds of formula (i) as being obtained by reaction of an aminodiol of formula (III) with a nitrile of formula (IV) (see claim 1, step (b); page 3, lines 1 to 7; page 6, lines 1 to 3 in combination with page 7, lines 7 to 20). In examples 1 and 2, the aminodiol sulfone, referred to as ADS (see page 5, lines 11 to 13), is reacted with benzonitrile and dichloro acetonitrile, respectively.

Mapping atoms from the starting material to product, the skilled person would immediately recognise that there was an inconsistency in the substitution pattern of the carbon atom bearing an amino substituent in formula (III) and the corresponding carbon atom in formula (i), that is, between the H2N?H fragment in the former and the H2N?N fragment in the latter.

Furthermore, in the application as originally filed, it is stated that "the present invention relates to intermediates to florfenicol" (see page 1, lines 7, 16), which is a known compound having the following structure (reproduced from document (2)):

FORMULA/TABLE/GRAPHIC

On examining the structure of florfenicol (see 2'-position) and comparing it to that of formula (III), the skilled person would have immediately recognised that the H2N?H fragment was intended to be conserved in the desired end product, and would therefore have identified the H2N?N fragment in formula (i) as being erroneous.

This would have further been confirmed by the fact that the configuration at the oxazoline ring is specified to be "4R trans" in the application as originally filed (page 2, lines 4, 5 and page 11, lines 7, 8). The term "trans" in this context can only be intended to denote the relative arrangement on opposite sides of the ring of a substituent at position 4 and a substituent at position 5. The skilled person would therefore note a further contradiction, since the designation "trans" does not make any chemical sense for the structure depicted for formula (i) as originally filed, owing to the presence of two substituents at position 4.

The respondent's argument that it could not be excluded that formula (i) as originally filed was in fact the intended structure owing to the use of the terms "analogs" and "comprising" is not convincing:

There is no mention whatsoever in the originally filed application of the presence of an oxidising agent or of the possibility of an additional oxidising step. It is noted that the error in the oxazoline formula is also present in the examples, where an oxidising agent is certainly not present. Moreover, the wording used in claim 1, step (b), namely, "reacting a compound of formula III, with a compound of the formula IV ... to obtain a compound of the formula I" (emphasis added) makes it clear that the compound of formula (i) is the direct product of the reaction between the compounds of formulae (III) and (IV) (see also page 3, lines 3 to 7). The skilled person would therefore recognise that an oxidation of the carbon atom at position 4 of the oxazoline ring was not intended and disregard this as being a chemical nonsense, particularly in view of the stated aim of providing "an efficient and economical process for preparing florfenicol" (application as originally filed, page 3, lines 8 to 10). As explained above, an oxazoline intermediate containing an H2N?N fragment would not be suited to this aim.

Therefore, the board concludes that the skilled person would have no doubt that an error had occurred in the structure depicted for formula (i) as originally filed (cf. requirement (i)).

2.1.4 With respect to requirement (ii), it must be decided whether the corrected feature is directly and unambiguously derivable from the content of the application as originally filed taken as a whole.

The application as originally filed contains the following cross-reference to document (9) (see page 1, lines 17 to 20): "The intermediates described in the present specification can be used to prepare florfenicol as can be seen, for example, in U.S. Pat. No. 4,876,352, which is hereby incorporated by reference". The only intermediates disclosed in document (9) that match those disclosed in the present application as originally filed are the oxazolines of formula (II) (cf. claim 1 and column 2). Specific embodiments thereof are disclosed in document (9) in Table I, claims 10 to 13, and as starting materials in the examples, for example, in examples 7 and 8 for the synthesis of florfenicol. In all these intermediates, the oxazoline ring bears a hydrogen atom and not an amino substituent at the position 4. Based on this information, the person skilled in the art would have no doubt that the H2N- substituent present in the pictorial representations of formula (i) was a misprint that should be replaced by H-, thus removing all the contradictions outlined above.

The respondent's argument that the skilled person would not know how to correct the error once detected, is not considered to be convincing. It is true that the formation of a second regioisomer is possible in the reaction of an aminodiol of formula (IV) with a nitrile of formula (III) (cf. e.g. document (2), page 5292, scheme, compounds 6 and 8; document (8), column 1, lines 25 to 30; document (21), formula (i)). However, it is evident from the cross-reference to document (9) that this is not the intended product, since said second regioisomer is not disclose therein. Moreover, the skilled person would a priori not consider the structure of the second regioisomer as a possible correction since it is structurally remote from the erroneous structure and does not include a stereocentre at position 4 of the oxazoline ring; the required "4R trans" configuration is therefore not possible for this structure.

Thus, the second requirement for allowing a correction in the sense of Rule 139 EPC is also fulfilled in the present case.

2.1.5 Accordingly, the amendment to formula (i) is allowable under Rule 139 EPC and does not introduce subject-matter that extends beyond the content of the application as originally filed. The ground for opposition according to Article 100(c) EPC does not therefore prejudice the maintenance of the patent on the basis of the main request.

2.2 The respondent did not raise any further objections under Articles 123(2) and 123(3) EPC resulting from the amendments made to the main request (cf. point X above), and the board sees no reason to differ. The amended set of claims of the main request is therefore also considered to meet the requirements of Articles 123(2) and 123(3) EPC.

3. Sufficiency of disclosure (Articles 100(b), 83 EPC)

3.1 In order to assess whether the requirement of sufficiency of disclosure is fulfilled in the present case, it must be assessed whether the patent in suit as a whole, that is, the claims and the description (including the examples), disclose the claimed synthesis of florfenicol in a manner sufficiently clear and complete for it to be carried out by a person skilled in the art, in the light of the general common knowledge of the technical field involved.

Present claim 1 relates to a process for producing florfenicol, which is an antibacterial agent of known structure (cf. patent in suit, paragraph [0002]; see also formula on page 14 above). The first step of the synthesis (step (a)) involves the reduction of a phenylserine derivative of formula (II) in a protic solvent to obtain the aminodiol sulfone of formula (III), also known as ADS (cf. patent in suit, paragraph [0012], last line). Then, in the same reaction vessel, ADS is reacted with a nitrile of formula (IV) to obtain an oxazoline of formula (i) (step (b)), which is converted to florlenicol (step (c)).

The patent in suit provides two preparative examples illustrating the conversion of the d-threo stereoisomer of a phenylserine derivative of formula (II) into an oxazoline intermediate of formula (i) (examples 1 and 2). Further, details of suitable reaction conditions for steps (a) and (b) are given in paragraphs [0014] to [0017]. With respect to the further processing to florfenicol, the patent in suit refers to document (9) (see page 2, lines 12 to 14), and this cross-referenced document discloses present step (c) in examples 7 and 8.

3.2 The respondent's arguments are not considered to be persuasive for the following reasons:

As outlined under point 3.1, the stereoisomer to be used as starting material of formula (II) is defined in examples 1 and 2. Moreover, the configuration of the intermediate oxazoline of formula (i) is defined without ambiguity as being "4R trans". There appears to be no reason to doubt that, based on this information, the skilled person would be able to obtain the final product florfenicol having the correct absolute stereochemistry.

The objections of the respondent with respect to discrepancies or ambiguities in the pictorial representation of the stereochemistry in the formulae appearing in claim 1 relate to the issue of lack of clarity (Article 84 EPC) rather than to that of insufficient disclosure (Article 83 EPC). Since the features objected to were already present in the claims as granted, they are not open to objection under Article 84 EPC, which does not constitute a ground of opposition under Article 100 EPC.

The argument of the respondent with respect to the breadth of R" is also not convincing. According to document (9), the oxazoline of formula (i) wherein R" is phenyl is converted into florfenicol by fluorination of the free hydroxyl group, hydrolysis of the oxazoline group to give the corresponding 2-amino-3-fluoro-1-propanol, followed by a dichloro acetylation reaction (see example 7 for first step, and example 8 for second and third steps). Thus, in the second of these steps, the R" group is removed by hydrolysis. There is therefore no reason to suppose that the exact nature of this group is critical in the context of the present reaction.

Finally, the respondent did not provide any evidence to support its attack with respect to the reaction conditions used in present steps (a) and (b). This objection is to be rejected as being unsubstantiated in the absence of evidence to the contrary.

3.3 In view of the above considerations, the board sees no reason to doubt that the skilled person would be in a position to select appropriate starting materials of formula (II) and convert these into florfenicol according to the method defined in claim 1.

Consequently, the requirement of sufficiency of disclosure is considered to be met by the main request and the objection under Article 100(b) EPC is to be rejected.

4. Novelty (Articles 52(1) and 54 EPC)

The respondent maintained its novelty objection with respect to document (15), mainly based on the scheme on page 9. This scheme depicts the enzymatic resolution of phenylserine esters of formulae (i) and (II) and their further processing by various alternative routes to a dichloroacetylated aminodiol derivative of formula IV. When Y is -SO2CH3, the compound of formula (IV) is known as thiamphenicol (page 10, lines 24 to 27), which serves as a starting material for preparing florfenicol (page 1, lines 15, 16). One of the routes disclosed in this scheme involves a reduction to produce an aminodiol of formula (III) followed by formation of an oxazoline of formula (v). However, there is no direct and unambiguous disclosure in document (15) that this specific reaction sequence is to be performed with compounds wherein Y is -SO2CH3. The only specific disclosure in this respect can be found on page 10, lines 21 to 27, wherein thiamphenicol is prepared by direct dichloroacetylation of the aminodiol of formula (III), and not via the oxazoline of formula (v). For this reason alone, said scheme cannot provide the basis for a successful novelty attack.

Furthermore, as pointed out by the appellant, several additional features of present claim 1 are not to be found in document (15), such as the requirement that the reduction and oxazoline formation are to be performed "in the same reaction vessel". Contrary to the assertions of the respondent, there is no disclosure of this feature, implicit or otherwise, in document (15). Indeed, separate paragraphs in document (15) deal with the reduction and oxazoline-formation steps (see page 10, lines 7 to 20 and page 11, lines 18 to 24, respectively).

Thus, the subject-matter of claim 1, and that of dependent claims 2 and 3, are novel over document (15).

None of the remaining cited prior art documents disclose a process according to present claim 1.

Accordingly, the subject-matter of the main request meets the requirements of novelty.

5. Inventive step (Articles 52(1) and 56 EPC)

5.1 The subject-matter of claim 1 relates to a process for producing florfenicol comprising two key steps, namely, reduction to obtain the aminodiol sulfone ADS, followed by oxazoline formation, in the same reaction vessel.

The board understands the feature "in the same reaction vessel" to mean that the intermediate ADS is formed and undergoes further reaction in a single reaction vessel without being isolated, in other words, in a so-called "one-pot" process. These two terms were used interchangeably by the parties during the appeal proceedings.

The board considers, in agreement with the appellant and respondent, that document (1) represents the closest state of the art. Document (15), suggested as a possible alternative by the respondent, relates to similar subject-matter to that dealt with in document (1), but is a less suitable starting point since it only provides a general disclosure of the relevant reaction steps.

Document (1) relates to a process for the synthesis of florfenicol (see title and introduction). In the first step of the second reaction sequence disclosed on page 892, methyl d-threo-3-(4-methylthiophenyl)serinate hydrochloride ((2S,3R)-5.HCl) is reduced to give the corresponding (methylthiophenyl) aminodiol 6. The detailed experimental procedure for this step is given on page 894, left-hand column. The reaction is performed on a milligram scale (377 mg of (2S,3R)-5.HCl), and the work-up procedure involves a filtration step, an extraction of the aqueous phase with dichloromethane, drying of the combined organic layers, and evaporation to give a white solid in 92% yield. A portion of this is recrystallised from methanol to provide white crystals in 75% yield.

In the second step of the reaction sequence, the aminodiol 6 is then subjected to condensation with benzonitrile followed by oxidation with peracetic acid, according to the method described in reference [8] to give a methylsulfonyl oxazoline derivative 7, which corresponds to present formula (i) wherein R" is phenyl. The corresponding experimental procedure in reference [8] (document (2) in present proceedings) can be found on page 5293, right-hand column, "Method B". In Scheme I of document (2), this reaction sequence corresponds to the conversion 3 -> 5 -> 7. According to "Method B" this sequence is performed starting with 50.0 g of 3, without isolation of 5, to yield 7 in 93% yield. In document (2) the further conversion of 7 to florfenicol (1) is also disclosed (cf. Scheme I).

Thus, document (1), in combination with document (2) referenced therein, discloses a complete synthesis of florfenicol starting from the serinate sulfide (2S,3R)-5.HCl. The reduction thereof to the aminodiol is performed in a first step (document (1)), and the cyclisation and oxidation to the oxazoline sulfone of present formula (i) in a second step (document (2)).

No precise yield is available for this two-pot procedure, since the two steps according to documents (1) and (2) are performed on very different scales (cf. previous two paragraphs). However, a rough estimate can be obtained by simple multiplication of yields, namely, 85% (92% × 93%) or 64% (92% × 75% × 93%), depending on whether the intermediate compound taken over to the second step is recrystallised or not.

5.2 The problem to be solved in the light of the closest prior art can be seen, as submitted by the appellant, as lying in the provision of an improved, industrially useful procedure for producing florfenicol.

The solution as defined in claim 1 relates to a combination of measures wherein the amino acid employed as starting material is a sulfone rather than a sulfide, and wherein the two key steps of reduction and cyclisation are performed in the same reaction vessel, that is, in a one-pot synthesis.

5.3 As a next step, it has to be decided whether it has been rendered plausible that the problem defined under point 5.2 has been successfully solved with respect to the closest prior art.

The two distinguishing features listed under point 5.2 may be seen as providing advantages of relevance in developing an industrially useful procedure: as a result of the first measure, an oxidising agent is no longer required, given that the phenyl substituent is already in the correct oxidation state for producing florfenicol (CH3SO2-); as a result of the second measure, less solvent and equipment is required, since the work-up procedure disclosed in document (1) is avoided. However, that these features can actually be seen as providing an improvement presupposes that the resulting procedure is efficient in the sense of providing satisfactory yields, since it would otherwise not be of industrial relevance.

According to the examples of the patent in suit, the one-pot synthesis starting with 5 g of d-threo-(p-methylsulfonylphenyl)serine ethyl ester provides a yield of 81% for the compound of formula (i) wherein R" is phenyl (example 1), and 65% when R" is -CHCl2 (example 2) (cf. specific compounds reproduced on page 3 above). In Annex B, the appellant repeated the one-pot synthesis on the smaller scale and with the reagents used in document (1) (368.5 mg), and obtained a yield of 53.3%. This reaction was repeated by Professor Dötz of the University of Bonn, commissioned by the appellant: yields of ca. 85% and 94% were obtained (Annex C, experiment (b), entries 1 and 2). During the first instance proceedings, a third party submitted results according to which it had obtained yields of 55, 59 and 63% on repeating the process disclosed in example 1 of the patent in suit (cf. point VI above, and observations under Article 115 EPC dated 28 February 2007, received on 8 March 2007, page 6, last paragraph and page 7, first paragraph).

Summarising these results it may be concluded that yields varying between 53 and 94% have been obtained for the process according to the patent in suit, depending on scale, operator and reaction conditions. These yields can be considered to be satisfactory in the context of the present reaction, in the sense that they are comparable to those estimated for the prior art (cf. point 5.1, last paragraph).

Having regard to the considerations outlined above, the board is therefore satisfied that the problem posed has been credibly solved by the reaction defined in claim 1.

5.4 It remains to be investigated whether the proposed solution would have been obvious to the skilled person in the light of the prior art.

As already outlined under point 4 above, one of the routes generally disclosed in the scheme on page 9 of document (15) involves a reduction to produce an aminodiol of formula (III) followed by formation of an oxazoline of formula (v). The substituent Y, corresponding to present substituent R, is inter alia defined as being -S-R**(2) or -SO2R**(2) wherein R**(2) is alkyl (page 3, lines 7, 8). Thus, based on the teaching of document (15), the skilled person would have expected

the methylsulfide (CH3S-) and methylsulfonyl (CH3SO2-) derivatives to behave in a similar manner in the context of the disclosed reactions.

However, the appellant has demonstrated by means of comparative tests that this is not the case:

Thus, as outlined above under point 5.1 above, document (1) discloses the reduction of the serine ester bearing a CH3S- substituent to give the corresponding crude aminodiol 6 in a yield of 92%, following filtration, extraction and drying. The yield for the subsequent crystallisation step is 75%. The overall yield is therefore 69% (92% × 75%). A similar overall yield of 68% was obtained by the appellant in Annex D.

In contrast, in Annex A, the corresponding results obtained for ADS starting from the serine ester bearing a CH3SO2- substituent were 61.6% for crude ADS, and 28.22% for the recrystallisation step, which amounts to an overall yield of 17.4%. According to Annex C, method (a), the corresponding results as repeated by Professor Dötz were (see entries 1 and 2, respectively): 64.3 and 52.7% for crude ADS, and an overall yield of 19.3% and 12.4%. Similarly, in Annex E, an overall yield of 10.3% was obtained for this process.

Therefore, the comparative data outlined in the previous two paragraphs convincingly demonstrate that, by simply substituting the starting methylsulfide amino acid according to document (1) by the methylsulfonyl amino acid, according to the teaching of document (15), much worse yields would in fact be obtained.

Concerning the second distinguishing feature according to present claim 1, namely, that the steps (a) and (b) are performed "in the same reaction vessel", no information can be found in the cited prior art as to whether this type of reaction sequence would be amenable to a one-pot procedure, in the sense of producing satisfactory yields. The skilled person would therefore have no reason to expect that, by implementing this further modification, the problems of the poor yields obtained for ADS, as outlined above, could be overcome.

Consequently, no teaching can be found in the prior art that would have led the skilled person to the present modifications of the closest prior art reaction as a solution to the problem posed.

5.5 The respondent's arguments with respect to inventive step do not hold for the following reasons:

It cannot be accepted that the variability in the yields obtained is such that the data submitted by the appellant should be disregarded. A certain amount of variation in yield depending on scale, reaction conditions and skill of operator is to be expected. As explained above under points 5.3 and 5.4, the data in question cannot be regarded as being irrelevant, since consistent trends and meaningful conclusions can be derived therefrom.

Moreover, it is true that a one-pot synthesis may, in principle, be a desirable strategy that would be considered by the skilled person involved in process development. However, this does not mean that the skilled person would expect that this could be successfully implemented regardless of the multistep process under consideration. In the present case, document (1) teaches that the insolubles generated after hydrolysis of the reducing agent should be removed by filtration, before the aminodiol is taken through to the subsequent step (see page 894, left-hand column). There is no suggestion in the prior art that comparable yields would be obtained if the filtration step were to be omitted. In other words, although there may be no prejudice against combining present steps (a) and (b), there was also no expectation that satisfactory yields would be obtained on doing so.

5.6 Hence, the subject-matter of claim 1 of the main request involves an inventive step. The same is true of dependent claims 2 and 3.

Accordingly, it is concluded that the subject-matter of the claim set according to the main request meets the requirements of Articles 52(1) and 56 EPC.

6. Remittal (Article 111(1) EPC)

The description has yet to be adapted to the allowable claims according to the main request. For this purpose, the board exercises its discretion under Article 111(1) EPC and remits the case to the department of first instance.