T 0072/23 (Photostable roxadustat formulation/FIBROGEN) 23-01-2025

Reasons for the Decision

1. Admittance of the argument from appellant 2 contesting the technical effect shown in Table 7 of the patent (Article 12(4) and (6) RPBA)

1.1 Appellant 2 argued in its statement of grounds of appeal that the data in Table 7 of the patent did not conclusively demonstrate that red, orange and yellow dyes imparted better photoprotection of roxadustat than blue dyes. This was because, according to Table 6, the blue dye tested in Table 7 was present at a lower concentration than the yellow, orange and red dyes.

The respondent requested that this argument not be admitted into the appeal proceedings under Article 12(4) or (6) RPBA. In the opposition proceedings, the appellants had not contested the respondent's position that the data in Table 7 of the patent demonstrated that red, orange and yellow dyes were advantageous over blue dyes.

1.2 In the decision under appeal (page 12, first and second paragraphs), the opposition division concluded that a photostabilising agent comprising titanium dioxide and at least one dye selected from red, orange and yellow dyes reduced roxadustat photodegradation. This conclusion was derived from the data in the patent: Tables 2, 3, 4, 5 and 7 showed that roxadustat was less susceptible to photodegradation in compositions containing photostabilising agents as defined in claim 1 than in compositions with no photostabilising agent. Therefore the opposition division defined the objective technical problem as the provision of a solid pharmaceutical composition comprising roxadustat with improved photostability. The definition of the problem as an improvement was justified by the fact that the composition contained a photostabilising agent; the opposition division did not analyse the data in Table 7 specifically nor assess whether the selection of a red, orange or yellow dye over a blue dye was purposive.

In these appeal proceedings, the respondent has also defined the objective technical problem as an improvement. However, in contrast to the opposition division, the respondent has not justified the improvement only by the presence of a photostabilising agent. According to it, the improvement results from the selection of a photostabilising agent which contains a red, orange or yellow dye rather than a blue dye (reply to the statements of grounds of appeal, point 4.30). This technical effect was, it claimed, demonstrated by the data in Table 7 of the patent.

1.3 In view of the respondent's analysis of inventive step, the Board considered that it might be necessary, as an essential step for establishing the correct objective technical problem, to assess whether the data in Table 7 of the patent supported the allegation that the selection of a red, orange or yellow dye was advantageous. As the considerations from appellant 2 in that respect were not complex and were not detrimental to procedural economy, the Board decided to admit them. Nevertheless, the new argument ultimately had no bearing on the Board's conclusion on inventive step (see points 2.4 and 2.5 below). Like the opposition division, the Board acknowledged that the composition of claim 1 showed improved roxadustat photostability based on the presence of the photostabilising agent alone.

2. Main request - inventive step (Article 56 EPC)

2.1 Roxadustat is an inhibitor of hypoxia inducible factor (HIF) prolyl hydroxylase. It is a useful drug for treating or preventing conditions associated with HIF, such as anaemia, or ischaemia- and hypoxia-related disorders. The invention according to the patent is based on the observation that roxadustat degrades when exposed to light. Thus, the patent proposes protecting roxadustat with a photostabilising agent that comprises titanium dioxide and at least one additional dye (patent, paragraphs [0002] to [0004] and [0010]). In claim 1 of the main request, the additional dye is selected from a red dye, an orange dye, a yellow dye, and combinations thereof. The photostabilising agent is incorporated into a tablet comprising roxadustat and a pharmaceutically acceptable excipient.

2.2 The parties agreed that D1 can be taken as the closest prior art. D1 is directed to the use of roxadustat to increase or maintain the reticulocyte haemoglobin content in a subject in need thereof (page 2, first paragraph, and claim 1). It generally refers to oral formulations including tablets, pills, dragees, capsules, liquids, gels, syrups, slurries, suspensions and emulsions. The formulations contain roxadustat and excipients, e.g. colourants (page 7, lines 11 to 16 and 21 to 24). However, apart from the compositions used in Examples 1 and 2 for the oral administration of roxadustat, D1 does not disclose any actual formulation. As to the compositions used in Examples 1 and 2, D1 does not give any details beyond the roxadustat dose.

2.3 The parties did not dispute that the subject-matter of claim 1 differs from the teaching of D1 at least in that roxadustat is formulated together with a photostabilising agent that comprises titanium dioxide and a dye selected from a red dye, an orange dye, a yellow dye and combinations thereof.

The respondent correctly noted that the formulation of roxadustat in a tablet also constitutes a difference since D1 does not disclose any specific formulation.

2.4 The examples in the patent show that the technical effect associated with the photostabilising agent of claim 1 is the protection of roxadustat against photodegradation. It is apparent from Tables 2, 3 and 7 that the photodegradation of roxadustat in a formulation containing a photostabilising agent as defined in claim 1 is reduced compared with a formulation containing no photostabilising agent (uncoated formulation in Tables 2 and 3 or light control in Table 7).

With regard to the formulation of roxadustat as a tablet, the respondent has not alleged any particular technical effect.

2.5 Based on the improved photostability conferred by the photostabilising agent of claim 1, the respondent defined the objective technical problem as providing a composition comprising roxadustat having improved stability. The respondent contended that the problem should not refer to photostability or to a solid composition since this would point at the solution. As photostability was part of the total stability, improving photostability also implied improving stability in general.

The appellants did not agree with the respondent's definition of the problem. In their view, the objective technical problem should be to provide a photostable roxadustat composition. Nevertheless, at the oral proceedings before the Board, the appellants considered that this point was not critical because the subject-matter of claim 1 was also an obvious solution to the problem proposed by the respondent.

The Board decided to accept the respondent's definition of the objective technical problem. Given the conclusion on inventive step here below, the appellants were not negatively affected by this decision (see point 2.6.3, last paragraph).

2.6 The solution proposed in claim 1 is a tablet comprising roxadustat and an effective amount of a photostabilising agent comprising titanium dioxide and at least one dye selected from a red dye, an orange dye, a yellow dye and combinations thereof.

The formulation of roxadustat in a tablet is a customary choice that has not been shown to be associated with any surprising technical effect. It is also one of the options proposed in D1 (page 7, line 12). Therefore, the fact that roxadustat is formulated in a tablet does not involve any inventive step.

The parties focused their discussion on obviousness on the question of whether using a phostostabilising agent as defined in claim 1 to stabilise roxadustat was obvious. For the reasons set out in the following paragraphs, the Board agrees with the appellants that the answer to this question must be in the affirmative.

2.6.1 It was common ground that D1 provides no information on the photostability of roxadustat and is silent on the need to take photostability into account for roxadustat formulations. Although D1 mentions that colouring agents may be added to roxadustat formulations, it does not associate colouring agents with photostability (page 7, lines 15 and 23). The matter of dispute between the parties was whether the skilled person would arrive at the formulation of claim 1 in the light of common general knowledge.

2.6.2 The appellants cited common general knowledge which demonstrates that photostability testing is an essential part of drug development and that it has to be performed for each new drug and drug product.

D4, which is an excerpt from an encyclopedia of pharmaceutical technology, stresses the importance of testing the photostability of drugs and drug products because photodegradation during storage or use may lead to a loss of drug potency or even the occurrence of toxic degradation compounds (page 2859, left-hand column, first paragraph).

This common general knowledge is confirmed by D5 and D16. D5 is a guideline for photostability testing of new drugs and drug products proposed by the ICH Expert Working Group for adoption by the regulatory bodies of the European Union, Japan and the USA. D16 is the note issued by the European Medicines Agency implementing D5. These two documents teach the importance of testing the photostability of new drugs and drug products as an integral part of stress testing in the proceedings leading to regulatory approval to demonstrate that light exposure does not result in unacceptable change (D5, page 1, first and second paragraphs; D16, page 3, first and second paragraphs).

Thus the skilled person would necessarily test the photostability of roxadustat. By doing so, they would find that roxadustat degrades upon exposure to certain wavelengths in the UV-VIS range.

Against this conclusion, the respondent argued that developing a product for marketing authorisation required a research project and involved far more than routine testing. The data that had to be submitted to the regulatory authorities involved extensive experimentation on multiple aspects of drug development and focusing on photostability distorted the whole picture introducing unacceptable hindsight. Furthermore, even if photostability was tested, identifying whether a product was sensitive to light degradation was not a routine task but, rather, a complex one.

These arguments are not convincing. The Board accepts that developing a product for marketing authorisation generally requires a research project. However, photostability testing is an essential step for each new drug to be developed and, contrary to the respondent's submissions, does not require extensive experimentation. The tests are routine and, as indicated in D4 (page 2862, right-hand column, penultimate paragraph) a basic protocol proposing a reasonably simple test is provided in D5 (page 2). The test basically requires exposing the drug to a standard source of UV and visible light under controlled conditions and observing whether the drug shows degradation.

2.6.3 Considering that the skilled person would necessarily find that roxadustat degrades when exposed to light, the next question to be answered is whether the addition of a photostabilising agent comprising titanium dioxide together with a red dye, an orange dye, a yellow dye or a combination thereof was an obvious measure to reduce photodegradation.

D6, which is an excerpt from a textbook on pharmaceutical photostability, sets out the basic principles for the photostabilisation of tablet formulations (section "Photostabilisation of Tablets" starting on page 329). It states that, in order to protect photosensitive drugs, the amount of undesirable radiation penetrating the drug molecules must be reduced. One way of doing this is photostabilisation by spectral overlay. This method relies on the principle that excipients with absorption spectra similar to that of the photosensitive drug can be added to the formulation to reduce the amount of radiation interacting with the drug molecules (paragraph bridging pages 329 and 330 and Figure 12). D6 also indicates that in pharmaceutical practice food colorants can be used as photostabilising excipients for oral dosage forms (sentence bridging pages 330 and 331). In addition to excipients with UV-VIS spectra overlapping the UV-VIS spectrum of the photosensitive drug, tablets may contain opacifiers as photostabilising agents. Opacifiers are compounds that scatter and absorb light. A typical example is titanium dioxide, although it is known that it has an absorption gap between 400 and 420 nm (page 332, first paragraph).

The principles in D6 are confirmed by D9 which teaches in its introduction that photosensitive substances are usually protected by additives which have an absorption spectrum that overlaps that of the substance to be protected, e.g. food colourants (page 85, left-hand column, fourth paragraph). D9 also teaches that titanium dioxide is a very good photostabilising agent but that its absorption gap between 400 and 420 nm has to be considered prior to use. Similarly, D11 discloses the usefulness of the principle of photoprotection by spectral overlay and the use of food colourants for that purpose (abstract, last sentence; conclusion, first paragraph and Figure 9).

Thus it follows from common general knowledge that the use of photostabilising agents to reduce the photodegradation of a drug was obvious. Typical photostabilising agents are dyes with a UV-VIS absorption spectrum overlapping that of the drug and an opacifier which absorbs and scatters light, preventing it from penetrating into the formulation. Furthermore, it appears that these two types of photostabilising agents were commonly combined to improve the photoprotection provided by each. For instance, D6 refers to nifedipine, which is a drug sensitive to wavelengths between 290 and 450 nm (page 333, first paragraph). Nifedipine was protected using the spectral overlay principle by a combination of tartrazine (a yellow dye) and titanium dioxide. This combination reduced light transmittance in the relevant wavelength range and imparted greater protection than either of the individual photostabilising agents did. Similarly, D26, which is a chapter from a textbook and also represents common general knowledge, refers to nifedipine and other cases in which the combination of a yellow dye (iron oxide) and titanium dioxide provided the best photoprotection (page 189, last paragraph and page 190).

Consequently, once the photostability problems and the UV-VIS absorption spectrum of roxadustat were known from the mandatory photostability testing at the outset of drug development, it was obvious which dyes could be used for stabilising it in accordance with the principle of spectral overlay. The UV-VIS spectra of the dyes commonly used for photoprotection were known and the respondent never contested that the UV-VIS spectra of red, orange and yellow dyes greatly overlap with the UV-VIS spectrum of roxadustat. In other words, the respondent did not contest that the dyes in claim 1 protected roxadustat in accordance with the principle of spectral overlay. Furthermore, it was known that titanium dioxide is broadly used as a photostabilising opacifier and that it is commonly combined with dyes to enhance photostabilisation. Therefore the solution proposed in claim 1 does not involve an inventive step.

2.6.4 The respondent contended that the addition of a photostabilising agent as defined in claim 1 was not obvious for two reasons. First, the addition of a photostabilising agent was not the solution of choice against photodegradation. D4 stated that the method used most commonly to protect photosensitive drugs is to place the drug product in a protective market pack or in a coloured or amber immediate container (page 2862, right-hand column, second paragraph, first sentence). This teaching was in line with the flow chart in D5 (page 3) and D16 (page 4), which shows that the immediate solution against photodegradation was putting the drug product into light-resistant packaging. A reformulation of the product to add a photostabilising agent was a last-resort measure and should be avoided because reformulation implied repeating a large number of the tests required by the regulatory authorities for the reformulated product. Second, the principle of spectral overlay was not reliable because a formulation could contain photosensitising excipients, i.e. excipients which absorb energy at wavelengths outside the absorption range of the drug but which subsequently transfer the absorbed energy to the drug causing its degradation. Therefore the photostability of a drug in a formulation could not be predicted only from the absorption spectrum or the stability studies of the drug in a pure solvent (D4, page 2859, right-hand column, first paragraph). Furthermore, titanium dioxide would not be selected as the opacifier because it has an absorption gap at the range between 400 and 420 nm. Other opacifiers would be preferred, e.g. iron oxides.

The Board disagrees.

With regard to the first argument, D4 teaches that it is essential to obtain information about the photoreactivity of a drug as early as possible in the formulation process. D4 also refers to D5 as a document providing a reasonably simple photostability test (D4, page 2862, right-hand column, penultimate paragraph). Thus routine photostability testing is performed at the pre-formulation stage. It is at that stage that, if needed, the principle of spectral overlay is applied and photostabilising agents are added. The flow chart in D5 and D16 does not deal with the photostability testing of drugs. Instead, it relates to the photostability testing of drug products. This means that the flow chart does not deal with photostability testing at the outset of drug development but at the final stage, shortly before marketing. At that stage, photostabilising agents have already been added to the product if initial photostability testing revealed that the drug was photosensitive. The late photostability testing of drug products depicted in D5 and D16 is required because, as argued by the respondent, the formulated product might contain excipients that act as sensitisers, in which case additional photostabilising measures would need to be taken. A reformulation of the product at a late stage would be undesirable because extensive testing would need to be repeated for the reformulated product. As indicated in the flow chart of D5 and D16, the immediate solution would be to use light-resistant packaging. However, the fact that light-resistant packaging is a preferred solution when photostability issues are detected shortly before marketing does not negate that adding photostabilising agents is an obvious measure when a drug is found to be photosensitive at the pre-formulation stage.

With regard to the second argument, the principle of spectral overlay is applied at the outset of drug development based on the results of routine photostability testing on the drug rather than the drug product. Therefore, the respondent's arguments on the photostability testing of drug products and the possible presence of photosensitising excipients are not convincing.

With regard to the use of titanium dioxide, it is known that titanium dioxide has an absorption gap between 400 and 420 nm, but this does not make it unsuitable for photoprotection, especially if combined with a dye absorbing in the gap range, e.g. a yellow dye (see Figure 12 in D6). Such combinations are known in the art (see e.g. D6, page 333, first paragraph; D26, page 190). The respondent also referred to the fact that titanium dioxide might be incompatible with some active compounds, as stated in D8 (paragraph bridging pages 8742 and 8743). However, this does not point away from titanium dioxide either. The incompatibility described in D8 was the exception rather than the rule. Titanium dioxide was a broadly used opacifier and the fact that other opacifiers might be preferred depending on the circumstances does not render titanium dioxide a less-obvious option.

2.6.5 The respondent cited several decisions to argue that the solution proposed in claim 1 could not be obvious.

It referred in particular to T 814/19, T 2591/18 and T 91/22 to argue that identifying and overcoming the photostability problems of roxadustat was not a matter of routine and that the skilled person had no incentive to combine roxadustat with the photostabilising agent of claim 1. These decisions, however, do not support the respondent's case.

T 814/19, T 2591/18 and T 91/22 deal with several aspects of drug formulation but not with the particular issue of drug photostability. In T 814/19 (Reasons 2.6 and 2.7) and T 2591/18 (Reasons 3.7.2) the competent boards, taking into consideration the circumstances of the case, came to the conclusion that an undue amount of experimentation was required to identify and overcome the formulation issues arising in each case. This was the case even if the tests that had to be performed were part of the process of drug development. In T 91/22 (Reasons 6.6), the competent board held that the skilled person had no incentive to combine the drug of the closest prior art with an anti-nucleating agent because it was not known that the drug converted to a less-soluble form under certain acidic conditions.

In the present case, the Board has explained that the skilled person would necessarily have found that roxadustat was photosensitive. Once this was known, combining roxadustat with the photostabilising agent of claim 1 was an obvious measure in the light of common general knowledge. Thus, unlike in the cases dealt with in T 814/19, T 2591/18 and T 91/22, in the present case the skilled person would arrive at the claimed invention without conducting undue experimentation and prompted by common general knowledge. The Board agrees with the appellants that the circumstances of the present case are closer to those of T 393/18 (Reasons 2.3.1 and 2.3.2). In T 393/18, the Board agreed with the parties that stability testing was part of the pre-formulation studies of new drugs and that the skilled person would identify stability issues as a matter of routine.

At the oral proceedings before the Board, the respondent also referred to decision T 783/22, which relates to the photostability of a drug in a tablet formulation. As noted by the appellants, this decision was based on different facts and the conclusions therein are not applicable to the case at hand. In particular, the parties to T 783/22 never discussed the principle of spectral overlay, which is a crucial aspect in the present appeal proceedings.

The respondent also cited T 1349/19 (Reasons 1.27) to argue that holding the invention obvious would introduce unacceptable hindsight because it implied working backwards the steps taken by the respondent to arrive at the invention. Points 2.6.2 and 2.6.3 above make it clear that the claimed invention is obvious because the skilled person would inevitably find that roxadustat is photosensitive and, therefore, they would be prompted by common general knowledge to add a photostabilising agent as defined in claim 1. Thus, the circumstances of T 1349/19 are not applicable, either.

3. Auxiliary request 3-A - inventive step (Article 56 EPC)

Claim 1 of auxiliary request 3-A differs from claim 1 of the main request in that the additional dye has been limited to a red dye.

At the oral proceedings before the Board, the respondent stated that roxadustat is yellow and therefore, in accordance with the principle of spectral overlay, the skilled person would have selected a yellow dye for the photoprotection of roxadustat. In addition, it was generally known that yellow dyes had a maximum absorbance at wavelengths covering the range of 400 to 420 nm at which titanium dioxide had its absorption gap (see D6, Figure 12). Therefore the skilled person would have chosen a yellow dye rather than a red dye. It was unexpected that the combination of titanium dioxide with a red dye was suitable to improve roxadustat photostability.

The Board disagrees. The fact that roxadustat is yellow does not mean that the UV-VIS absorption spectrum of a red dye does not substantially overlap with that of roxadustat or, at least, that it does not absorb at the problematic wavelengths. As explained for the main request, the skilled person would find the suitable dyes by knowing the problematic wavelength range and comparing the absorption spectrum of roxadustat with that of the commonly used dyes. The respondent has not denied that the spectra of red dyes and roxadustat substantially overlap, especially at the problematic wavelength range. The respondent may be right that a yellow dye could have been regarded as a better option. Indeed, Table 7 of the patent shows that the combination of titanium dioxide with a red dye provides less protection than the combination with a yellow dye. However, there are no apparent reasons why the skilled person would discard a red dye as a suitable photoprotecting agent.

Therefore the subject-matter of claim 1 of auxiliary request 3-A does not involve an inventive step either.

4. Auxiliary request 5-A - inventive step (Article 56 EPC)

Claim 1 of auxiliary request 5-A differs from claim 1 of auxiliary request 3-A in that the red dye has been specified to be Allura Red AC aluminum lake.

The respondent argued that the spectral overlay principle may suggest a colour range, but not a specific dye within that colour range. There were different dyes of each colour and each dye produced a different result. According to the respondent, a comparison of the results for Orange#3 and Orange#4 in Table 7 of the patent, which differ only in that Orange#3 contains red iron oxide and Orange#4 contains Allura Red AC (see Table 6), demonstrated that the latter was advantageous over other red dyes. There was no pointer in the prior art suggesting that Allura Red AC might have a higher photostabilising effect on roxadustat than other red dyes.

The respondent's arguments are not convincing. On the one hand, it cannot be conclusively derived from a comparison of the results of Orange#3 and Orange#4 in Table 7 that Allura Red AC provides greater photoprotection of roxadustat than red iron oxide. In addition to titanium dioxide and the red dye, Orange#3 and Orange#4 contain a yellow dye. It is uncertain whether the difference in photoprotection between Orange#4 and Orange#3 comes from the red dye alone or whether there are interactions between the different components of the photostabilising agent. On the other hand, different dyes have different absorption spectra and can be expected to provide different degrees of photoprotection within a certain range. There is no evidence that the photoprotection provided by Allura Red AC is surprisingly better than that conferred by other red dyes.

Therefore, the subject-matter of auxiliary request 5-A does not involve an inventive step, either.