Review Of Recent EPO Decisions On Polymorphs

The European Patent Office (EPO) has established a fairly consistent and settled approach to the patentability of crystalline polymorphs of small molecule pharmaceuticals: it is generally required that the polymorph has an unexpected advantage compared to known forms of the drug in order for inventive step to be acknowledged. This approach is derived from leading cases such as T777/08 and T1422/12. Difficulties in obtaining patent protection for polymorphs of pharmaceuticals at the EPO often arise because no advantage has been demonstrated or because the advantage observed is thought to be expected for crystalline forms as compared with other forms, both as confirmed in T777/08.

To assess the ongoing approach of the EPO Boards of Appeal in this area, we have reviewed Technical Board of Appeal decisions relating to polymorph inventions published between 1 January 2020 and 31 March 2026. The outcomes of a selection of those decisions are discussed below, with our comments on how aspects of these decisions highlight any developments or changes to the framework the Boards of Appeal have established for the assessment of polymorph inventions. The decisions reviewed are grouped by the primary patentability provisions at issue in the cases in question.

The key conclusion is that polymorph inventions remain patentable at the EPO, with the Boards of Appeal in many cases taking a patentee-friendly approach. T1684/16 states explicitly that the existence of routine polymorph screening does not necessarily make a given advantageous polymorph obvious. Several decisions have also emphasised that a balance of properties can support inventive step. More care will need to be taken, however, when it comes to reliance on post-filed data (in view of G2/21) or in cases where the sole advantage relied on is stability (as discussed in T41/17).

National courts may adopt a different approach to the EPO in the assessment of the validity of polymorph inventions, for instance focussing on what the skilled person would have actually done in practice. However, it remains the case that first instance tribunals and Boards of Appeal at the EPO continue to grant and uphold patents for polymorph inventions when applying the well-established problem-solution approach. It will be interesting to see whether the Unified Patent Court will adopt an approach in line with that of the EPO, consistent with what has been seen in other areas.

Background

Before considering recent decisions, a reminder of the principles of the two leading cases mentioned above is provided.

• T777/08 (24 May 2011) considered a claim to a polymorph of atorvastatin. The Board found that, starting from the amorphous form, an improvement in filterability and drying characteristics for the polymorphic form described would have been obvious because crystalline polymorphs were known generally to have improved filterability and drying characteristics compared with amorphous forms. Consequently, the Board of Appeal concluded that “[t]he arbitrary selection of a specific polymorph from a group of equally suitable candidates cannot be viewed as involving an inventive step”¹. This decision has often been seen as establishing a negative outlook for polymorph inventions at the EPO. However, it also makes clear that a lack of inventive step arises “in the absence of any technical prejudice and in the absence of any unexpected property”².
• T1422/12 (11 April 2013) later confirmed that an “unexpected property” of a crystalline form could form the basis of a finding of inventive step, even if that property was only shown to be an improvement relative to the amorphous form. This decision concerned an application for a specific polymorph of tigecycline. The application was refused during examination for lack of inventive step over the amorphous form of tigecycline. Comparative data had been filed during examination showing that the claimed polymorph was more stable than amorphous tigecycline with respect to epimerisation, but this was not accepted by the examining division (in some ways foreshadowing the approach adopted following G2/21). The Board of Appeal disagreed and found that, starting from the amorphous form, it would not have been obvious to the skilled person that formation of a polymorph of tigecycline would overcome the known issue of epimerisation and the decision of the examining division was overturned. This case demonstrates that by re-formulating the technical problem to be solved (e.g. “the provision of a solid form more stable with respect to epimerisation” rather than “the provision of a more stable solid form”) inventive step may be demonstrated, even without providing data comparing the claimed polymorph with other polymorphs.

Inventive step

The main consideration determining the outcome for a polymorph invention is often inventive step. Consequently, inventive step is the focus of many polymorph decisions at the EPO, with their approach still ultimately based on the test set out in T777/08.

A key part of the EPO’s approach to assessing inventive step is identifying the closest prior art. For a polymorph invention, this is often a disclosure of the compound per se, an amorphous solid form, another crystalline solid form, or a solid form of unknown/unspecified character. Sometimes the claimed polymorph is also that of a different salt (or free base) form of the compound (or salt), as compared with the substance disclosed in the prior art, which adds another differentiating feature to this assessment. Each of these situations gives rise to slight variations in the application of the above principles.

Routine screening for polymorphs

Opponents often present arguments that it would have been routine for the skilled person to have screened for polymorphs of a given drug compound and, as a result, to have identified a polymorph with favourable properties. T1684/16 (3 March 2020) considered inventive step for bosutinib monohydrate and made it explicit that the EPO does not necessarily see the routine nature of polymorph screens as a barrier to inventive step.

The Board of Appeal in T1684/16 directly addressed an argument by an opponent that polymorph screening was a routine activity and that there was a reasonable expectation of success in securing a polymorph with good properties. The Board dismissed this general argument, noting that: “The fact that the skilled person is taught in the prior art to investigate polymorphs in order to isolate the crystalline form having the most desirable properties is in itself not necessarily sufficient to consider a specific polymorphic form having a certain desired property obvious”³. This approach is consistent with that taken by Boards in earlier decisions. However, it is helpful to have this point confirmed explicitly in T1684/16.

Balance of properties supporting inventive step

Arguing for inventive step for a crystalline form based on the presence of a favourable balance of properties has generally been met with success before the EPO Boards of Appeal. The objective technical problem can then be formulated as providing that balance, with the prior art making no suggestion of which polymorph would have the specific combination of properties observed.

In T1354/23 (14 May 2024), the Board emphasised that the balance can be based on relative qualities, without requiring the claimed polymorph to be the most favourable in all aspects. In that case, a pharmaceutical composition comprising crystalline Form B of apalutamide was claimed, with Form B being defined by XRPD, unit cell parameters, and/or the same XRPD after periods of storage under specified conditions. The technical effects relied upon in formulating the objective technical problem were improved hygroscopicity, “high” (as opposed to improved) thermodynamic stability and “high” (as opposed to improved) polymorphic stability. The Board found that Form B did not need to show an improvement in stability in order for the effect to be taken into account and went on to decide based on this balance of beneficial properties that Form B was inventive.

When arguing for the formulation of the technical problem solved by the invention, it should generally therefore be considered whether there is a combination of advantages for the claimed polymorph, rather than simply picking the single “best” property.

A potentially stricter approach?

An expanded 5-member Board of Appeal handed down a decision revoking a patent for a polymorph invention in T41/17 (14 January 2020). This decision appears to be an outlier in some respects and has been relied on by opponents in many cases since.

The decision concerned a polymorphic form of sorafenib tosylate which was the most thermodynamically stable form and which demonstrated good mechanical stability to grinding as measured by a lack of interconversion between forms. One particular citation in those proceedings appeared to demonstrate a link between thermodynamic stability and stability to grinding. Prior art was also cited which suggested that the primary purpose of polymorphic screening was to identify the most thermodynamically stable form. Based on these citations and the specific facts of the case, the Board found that it would have been obvious to arrive at the claimed polymorph through routine polymorph screening because it was the most thermodynamically stable.

Arguments citing this decision since have sought to link other beneficial properties of polymorphs at issue with it being the most thermodynamically stable, but these arguments have not typically succeeded. As noted above, T1684/16 comments that the mere fact that screening for polymorphs is a routine activity cannot on its own deprive a polymorph of inventive step. T1684/16 is therefore often cited as a rebuttal of the T41/17 arguments and tends to be successful in doing so.

Marcus Müller (the chairman of Board 3.3.02 which hears the majority of appeals related to polymorph inventions) has also clearly differentiated most later cases from T41/17, highlighting that a specific set of facts gave rise to the revocation in T41/17 and limiting its general applicability. In T41/17 itself, a distinction is drawn between other cases and the facts at hand in T41/17. For instance, it is noted that the patent as maintained in T2114/13 (20 June 2017) related to a form of febuxostat with improved polymorphic stability rather than improved stability under mechanical stress as in T41/17.

It does remain to be seen, however, whether T41/17 could be used successfully to argue that the specific advantage of improved thermodynamic stability is, on its own, not considered sufficient to establish inventive step, even if this advantage is shown relative to other polymorphic forms. Any such future decision would also depend on the specific facts at issue, but in the meantime it is worth considering whether it can be argued that a combination of advantageous properties (e.g. thermodynamic stability and maintenance of a reasonable dissolution rate) applies for a claimed polymorph and/or whether a more specific form of stability can be relied upon (e.g. stability with respect to epimerisation, as in T1422/12).

It is also notable that most of the later decisions citing T41/17 draw a distinction between the facts in those later cases and those in T41/17, often on the basis of an advantageous combination of effects.

Post-filed data

Post-filed data (i.e. experimental results not present in the application as filed) is commonly used in polymorph cases as further evidence of asserted technical effects or of the nature of a solid form obtained by a prior art process. Enlarged Board of Appeal decision G2/21 (23 March 2023) has set out certain requirements for post-filed data to be admitted in support of inventive step. Some decisions issued subsequently have shed light on how G2/21 can be applied to polymorph cases.

Briefly, the test in G2/21 requires that a technical effect being supported by post-filed data is encompassed in the technical teaching of the application and is embodied by the same originally disclosed invention.

One of the first cases applying G2/21 to a polymorph invention was T1994/22 (15 April 2024). The patent at issue related to Form II of selexipag. The only discussion of advantages of the claimed polymorph in the application as filed was a general statement that the (three) polymorphs disclosed in the application were a “thing of a high quality for which constant effect can always be shown and of a form which is handled easily industrially”. The patentee then sought to rely on post-filed data showing that the claimed form (Form II) had improved photostability compared with Form III, which was also disclosed in the application. While improved photostability may well be an unexpected effect, the Board found that the “sweeping statement” about “high quality” was not sufficient to support the technical effect of improved photostability. On this basis, the Board held that the post-filed data could not be relied on for inventive step and, in particular, could not be used to formulate the objective technical problem solved by the invention. Form II was accordingly found to be “an arbitrary selection from the host of alternatives covered by the closest prior art. Such an arbitrary selection without any unexpected balance of properties being produced cannot contribute to inventive step”⁴.

Another case in which post-filed data was not taken into consideration is T852/20 (27 November 2023) relating to Form 1 of vemurafenib. Post-published data was presented at first instance proceedings before the Opposition Division showing that Form 1 had improved solubility and bioavailability compared to Form 2, leading to rejection of the opposition. At appeal, though, applying the principles of G2/21 (issued between the first instance decision and appeal decision), the post-filed data was not taken into account and Form 1 was found to be merely an alternative (and obvious) solid form of vemurafenib. The passage relied on by the patentee to foreshadow the technical effect demonstrated by the post-filed data was interpreted by the Board as relating to amorphous, rather than crystalline, vemurafenib. A further argument by the patentee based on the interpretation of DSC data in the application as filed was not admitted to proceedings as it was made too late, being first raised at the appeal oral proceedings.

In contrast, T672/21 (15 April 2024, concerning the parent of the patent at issue in T1994/22) found in favour of the patentee that Form I of selexipag was inventive. This was based on an advantageous balance of four properties, three of which were evidenced in the patent itself and one of which (thermodynamic stability) was evidenced by post-filed data, which post-filed data was taken into account for inventive step (presumably as the general property of “stability” could be linked to the invention based on the application as filed, unlike the more specific issue of “photostability” as in T1994/22 above). The objective technical problem was formulated ambitiously by the Board and included reference to all of these properties. Arguments based on T41/17 were dismissed (with reference to T1684/16), based on the objective technical problem defining a balance of properties.

There are still uncertainties about how G2/21 will be applied by the Boards of Appeal. However, it is clear that it is important to spell out in the application as filed what advantages are thought to be associated with the invention and to what polymorph(s) those advantages apply. It may not be possible to rely on post-filed data for inventive step for a polymorph in an application which includes only very general statements (or no statements) about the advantages of the polymorph claimed. The best case remains that in which the experimental results relied on are included in the application as filed.

Salts and free bases

T1825/21 (13 March 2023) is notable for the apparent breadth of claim 1 maintained by the Board of Appeal: “A crystalline form of [ceritinib free base]”. Two forms (A and B) of ceritinib free base were identified in the application as filed. The closest prior art (Example 7 of D1) described the production of ceritinib hydrochloride and it was uncontested that the ceritinib hydrochloride produced was amorphous. Two distinctions over D1 therefore arose: that ceritinib was in free base form (rather than the hydrochloride salt) and that ceritinib was crystalline (rather than amorphous).

With reference to T777/08, which set out those advantages the skilled person would expect for crystalline forms over amorphous forms, the Board concluded that the application credibly shows that crystalline forms of ceritinib free base demonstrated improved stability and ease of drying, formulating the objective technical problem on that basis. At this stage, it might have been expected that the Board would have concluded the solution provided was obvious, as it highlighted these advantages for a polymorph as being part of the common general knowledge. However, taking the specific fact pattern into account, the Board reached a positive finding with respect to inventive step: experimental evidence was filed showed that attempts to prepare the hydrochloride salt consistently led to an amorphous form and the Board felt that there was no motivation for the skilled person at that point to try instead to produce a crystalline form of the free base form. This was supported by D14 which commented that the formation of salts could be used to improve physical characteristics of a compound, effectively teaching away from the free base.

Novelty

The assessment of novelty of a polymorph of a known drug is often complicated by the potential for implicit disclosure of a specific polymorph in the prior art, even if there is no actual characterisation of a crystalline form. Typically, this arises where the prior art describes a synthetic protocol for a compound which concludes with production of a solid.

It can be the case that the specific conditions under which the solid form of the prior art is obtained are not fully described. In this scenario, the EPO’s requirement that the substance must be the inevitable result of the process described in order to be novelty-destroying can assist patentees, if they can demonstrate that variations in conditions such as temperature and speed of concentration affect the solid form obtained.

T0479/24 (15 January 2026) confirms the importance of exactly repeating the prior art in order to attack the novelty of a claim for a polymorph based on the disclosure of an unspecified solid form in the prior art. The invention related to Form A of roxadustat and two opponents argued that the claimed form inevitably arose from the prior art teaching. However, the experimental reports relied on to support this argument included additional steps not disclosed in the prior art and also produced arguably contradictory results (particularly in light of the patentee’s own repetition of the prior art which did not always produce Form A) leading the Board to conclude: “the disclosure of D1 leaves such a degree of latitude already as regards the concentration in vacuo that the repetition of example D-81 of D1 does not necessarily result in a solid form of roxadustat, let alone roxadustat form A”⁵. The Board then proceeded to decide that Form A involved an inventive step, noting the beneficial combination of properties and commenting that “The mere suggestion in the prior art to screen for solid forms with advantageous properties, contrary to the appellants’ argument, is not in itself sufficient to deny roxadustat form A an inventive step”⁶.

A lack of novelty over an alleged implicit disclosure was, however, found in T1416/21 (12 September 2023), although it appears that the requirement for prior art to provide an enabling disclosure and the requirement for an implicit disclosure to be an inevitable result of the teaching of the prior art were somewhat conflated. The case is an appeal against revocation by the opposition division of a patent relating to crystalline forms of 2’-O-fucosyllactose (2’-FL). The closest prior art, D1, which was also concerned with preparing crystalline forms of the compound, described three methods for preparing crystals and described crystalline 2’-FL as being non-hygroscopic, but did not disclose any XRPD peaks. The opponent filed data (as D21) relating to preparation of crystals by a method based on method 2 of D1, which matched the XRPD peaks of the claimed form. The patentee did not submit their own data in response, but instead relied on statements in the application as filed that the inventors had failed to reproduce the experiments of D1. The Board did not appear to consider the question of whether the results in D21 demonstrated that the claimed form was an “inevitable result” of the method in D1, focussing instead on whether method 2 of D1 was suitably enabled. The Board found in favour of the opponent on this point based on the skilled person’s common general knowledge, so the claimed polymorph was found to lack novelty over D1. Ultimately, the patentee’s position would have been significantly improved if they had filed their own specific repetition of the prior art showing that the claimed form was not produced.

Sufficiency

Sufficiency does not often come up in relation to claims to polymorphs per se. One recent decision in which an objection of lack of sufficiency was raised by the opponent is T1354/23 (14 May 2024, also discussed above under inventive step). The opponent alleged that conversion of some forms into Form B was not sufficiently taught and it would require a research program to prepare Form B. The Board was quick to dismiss this argument, since the opponent had not raised serious doubts substantiated by verifiable facts.

Claim 1 defined the polymorph by a list of five alternatives: (a) XRPD as shown in a particular figure, (b) XRPD as defined by certain peaks, (c) unit cell parameters and (d, e) the same XRPD after periods of storage under specified condition. Another angle the opponent sought to argue was based on the alleged breadth of claim 1. It was argued that the list of ways to define the polymorph meant that claim 1 was too broad and potentially covered more than Form B. The Board disagreed, noting that claim 1 related to Form B and not other unmentioned polymorphs, so there was no issue with breadth of claim. Again, it seems that the opponent would have needed to show that other (presumably non-enabled) forms of the compound were embraced by the claims to succeed with this attack.

Clarity and definition of the crystalline form

Sometimes polymorphs (and pharmaceutical compositions more broadly) can be defined by features relating to their storage stability. It can be challenging to formulate these features appropriately to meet the EPO’s requirements for the skilled person to know whether they are working within the scope of the claims and to avoid defining the claimed subject-matter as a result to be achieved. Clarity is not a ground for opposition at the EPO, so features in granted claims cannot be challenged on clarity grounds, but amendments to the claims based on features from the description can be objected to for lacking clarity in post-grant proceedings.

The Board in T1065/18 (30 January 2023) found the feature “characterized as being an anhydrous form having a water content, when stored at 20°C at ambient pressure in an environment from 0% up to 90% relative humidity, of below 0.1% according to Karl Fischer (KF)” to lack clarity because it covered any storage time and a wide range of relative humidity. The Board also rejected an argument by the proprietor that this meant that the water content must be below 0.1% both at the beginning and at the end of the storage period. This highlights that such features are best defined specifically, or at least with appropriate specific fallback positions available.

The claim at issue in T461/22 (26 February 2024) defined a hydrate form of dapagliflozin with reference to the “XRPD pattern shown in Figure 1”. Although EPO examiners often object to clarity of claims defining features by reference to figures, polymorph claims can be an exception to this general rule: during prosecution, the examiner of this case had actually suggested that the definition include reference to the figure. As the reference to the figure was present in the claims as granted, its clarity was not addressed on appeal. No other issues were found to be associated with the reference to the figures and the claimed hydrate form was also found to be inventive.

Conclusion

It remains the case that the EPO is willing to grant and uphold patents directed to polymorphs of small molecules which show an unexpected advantage over the prior art. In addition to those cases where a single unexpected advantage is enough to support inventive step, a number of decisions confirm that a favourable balance of properties can lead to a finding of inventive step. However, in light of G2/21, the content of the application as filed may need to be assessed to determine whether the properties relied on for inventive step can be supported only by post-filed data.

One decision which appears to be a slight outlier is T41/17, which arguably suggests that thermodynamic stability on its own is not an unexpected advantage, even when demonstrated in relation to other polymorphs. However, as with any case, the outcome was highly dependent on the facts at issue. Further, it does not appear to date that T41/17 is being applied generally by the Boards of Appeal in connection with polymorph inventions.

Footnotes

1. T777/08, Headnote 2.

2. T777/08, Headnote 1.

3. T1684/16, Catchword

4. T1994/22, reason 1.8

5. T0479/24, reason 6.7

6. T0479/24, reason 7.4