More on Product-by-Process claims - what is the law, why is it, and what should it be?

My post on Wednesday morning has attracted a large number of comments, including many thoughtful contributions, and so I would like to follow up concerning the wider issue of product-by-process claims (independent of the specific Hospira v Genentech case), gathering up, and responding to, some of the comments.

In the meantime, readers interested in the decision itself may care to visit the report by Katfriend Suleman Ali (Holly IP) posted on the PatLit blog here, which has a more detailed analysis of the whole judgment and the claims actually at issue.

There is a lot of material, and so I will try to organise into subjects.

1) What kind of claim do you mean?

In my experience of discussing the law of product-by-process claims with a large number of people, it is important to have in mind the right kind of claim.  It won’t do if you are thinking of claim where the product is not actually new as such (such as erythropoietin in the Kirin Amgen case).  It may seem irrelevant, but actually I think that your legal instincts will be led astray if you are not actually thinking of a situation which satisfies the T 0150/82 criteria, namely:

Claims for products defined in terms of processes for their preparation (known as "product-by-process" claims) are admissible only if the products themselves fulfil the requirements for patentability and there is no other information available in the application which could enable the applicant to define the product satisfactorily by reference to its composition, structure or some other testable parameter.

This is the “EPO” standard that has been adopted in the UK.  Curiously, as one anonymous commenter helpfully pointed out, only the former of these two criteria appears in the corresponding EPO Guidelines, so at first instance at least the EPO seems to have resiled somewhat from the latter criterion.

Usually a product-by-process claim “X obtained/able by process Y” represents a novel subset of generic X.  It will necessarily have an ill-defined structure.  And pretty much always (I think), the claim will not define a single composition, but a genus of compositions, either because “process Y” itself contains a number of variables, or because “process Y” does not define absolutely all of the parameters that may affect the identity of the final product, and so by varying non-defined parameters, a number of products can be obtained.

A good example of the kind of claim that I think we should be considering was suggested by an Anon:

"A mixture of compounds obtainable by oxidising compound A under conditions x, y and z"

[This is now my comment, not Anon's] This may well produce a complex oxidised mixture that is difficult to characterise otherwise.  The claim is unlikely to define a single mixture, but, either because x y and z are defined with ranges, or because other conditions v and w also affect the outcome but are not specified, a number of mixtures are possible that fall within the scope of the claims.

The Anon went on to say

“Said mixture not being characterisable by techniques in common practice today, but identifiable by a 3rd party using their new fangled gadget, such that they are able to readily prepare said mixture quite easily.”  

I accept that this is possible in principle, and we should allow for the possibility, but I believe that in the majority of cases that would be unlikely to happen in the lifetime of the patent.

Some commenters pointed to the possibility of a claim to a single X, or something conceptually definable but not able to be structurally defined in particular circumstances.  One Anon stated:

I have a case to a new chemical entity that has been transferred post-drafting (getting my excuse in early). The claimed structure was in error, due to a re-arrangement taking place unbeknownst to the inventors. The characterising data all looked fine until recently when it was discovered otherwise.
New claims have now been drafted to the product obtainable by the disclosed manufacturing process. These claims, when granted, will be infringed by third parties making the same compound by a completely different route. Third parties will be well aware of the true structure because it is now in the public domain. In fact, third parties will only make this compound because they will be copying my client’s work once their product becomes successful. One day!

This approach may be successful on those specific facts (I don’t really want to adopt a view on that), but I don’t think it typical of the situation in general.

And I don’t think it should be overlooked that even when a product cannot be defined structurally, there are other means of characterisation that should be preferred over product-by-process definitions.

One Anon answered his own comment:
First comment:

Crystalline form of compound X obtainable by crystallisation from methanol, having crystal structure parameters a, b, c, melting point y, DSC peak at z, IR peaks at y1, y2, y3.
Fully characterised so that third parties can determine whether they fall within the claim when they make the same material by a different process.
e.g. Same crystal form obtained directly from the reaction mixture used to synthesise the compound, said crystal form crashing out of solution on work-up.

Follow up comment:

Question mark over my crystal form example would be whether the form is sufficiently characterised by the data, therefore not requiring the process step. However, we could add a functional feature not shown by all forms having the required data, yet would be testable by a third party to determine whether they infringe.

I think in general crystal forms can be defined without reference to process, and, even if a product-by-process limitation is additionally used, the spectroscopic characterisation in the claims allows third parties some greater clarity about whether an individual product infringes.

2) How plausible is it that a different process will achieve the same product?

One issue that led to lively debate was whether the situation that “obtainable by” language would lead to a different outcome in infringement from “obtained by” was likely to be a common or a rare occurrence.  Perhaps there has been some misunderstanding here, so I shall set out my thoughts from the beginning.

As I stated above, the usual situation is that a product-by-process claim “X obtained/able by process Y” represents a novel subset of generic X.  The whole of X is not new, otherwise the limitation defined by the process would not be needed.  These situations are quite common.  I gave examples of “the products of processing or purifying complex mixtures, products of polymerisation, metallurgical compositions, and the like.”  A later anon added “oxidation process, re-arrangement reactions, extrusion processes, desalination processes, osmotic processes, filtrations processes, hardening, softening, annealing, centrifugation, mixing, stirring, blending, injecting, stretching, not-stretching, crystallisation, spray-drying, etc.”

If the test for infringement of an “obtainable by” claim is that set out by Birss J in Hospira v Genentech, namely that it "must have every single characteristic which is the inevitable consequence of that process", then I do believe with the greatest of respect to the Anon who clearly strongly disagrees with me, that the cases where a patentee may be able to prove infringement of an “obtainable by” claim, where an “obtained by” claim would not have been infringed, will be rare.  That is, I think it will be very difficult for a patentee to prove that the material produced by the infringer’s process is identical in all material respects to material that could be obtained by a different process that is defined in the claim.  That is not just from my experience, I think that there are conceptual reasons why this is likely to be so as well.

The product by process claim is in the domain X, that already encompasses some prior art compositions X(prior), which the process limitation excludes.  So the whole of X is not new, only the part that is effectively delineated by the process feature – let us call this X(claimed).  To prove infringement by the accused composition X(infringe), the patentee has to both a) prove that X(infringe) has every single characteristic which is the inevitable consequence of a composition of X(claimed) but also b) ensure that it does not stray close to X(prior).  If the process is completely different, I think that in most cases this will be an impossible task.

On the other hand, if the process is only slightly different from that defined in the claims (which I expect would be a more likely scenario) then there is a better prospect of demonstrating infringement.  However, in that case, I think that there is also a realistic prospect of infringement of an “obtained by” claim under normal principles of purposive construction.

In summary therefore, I think that situations where it will actually make a difference which language is used will be rare.  I have still not been persuaded otherwise.

3) What are the principles that are at stake?

An early commenter suggested two principles:

1) The 'Angora Cat' principle of construction, which neatly explains why the same interpretation of the claim should be adopted for infringement as for novelty. A patentee can't adopt a broad construction for infringement having argued for a narrow one for novelty (this protects the public) BUT, by the same token (and by general judicial principles: equal treatment of parties etc etc), one (in particular an infringer/challenger) should not be able to argue for a broader interpretation for novelty than for infringement - this protects the patentee. Deciding arbitrarily that a 'reverse Angora' scenario is fine is unjust and unfair on patentees/inventors.
2) The protection conferred by a patent should correspond to the patentee's contribution to the art. Post Kirin-Amgen, to be allowable, a product-by-process claim must be directed to a NOVEL product. The contribution to the art is hence the product per se, not the process, and the scope of protection should reflect that.

I would first point out that neither of these is a statutory principle.  The over-reaching statutory principle is Article 69 and the Protocol, to which I will return later.

Even though I referred to an anti-Angora Cat situation, actually the symmetry is not perfect, and so I don’t think that what has happened is “deciding arbitrarily that a 'reverse Angora' scenario is fine”.  There are cogent reasons why, in the special case of product by process claims, the symmetry between novelty and infringement is broken.  But there are already such situations – such as lack of enablement of the prior art (for example in the UK case of Synthon v SmithKline Beecham) – where the symmetry between novelty and infringement is broken, so actually this is not a fundamental principle at all.  (Other commentators may consider that the axiom that what infringes if later, anticipates if earlier was laid to rest even earlier, in the terfenidine case)

I don’t think it follows that just because the product has to be novel per se that the contribution to the art is the product.  On reflection over the last few days, I think actually what the contribution to the art is may not be the same for all kinds of product-by-process claims.  But at least in some cases, I think it will really only be the process, even if the product of the process does happen to be new.
My view of the applicable principles is probably easiest explained by reproducing exchange with Norman:

Norman:

I don’t have a realistic example, but here is a conceptual example, just to make sure I understand your position. Suppose the patentee produces a new product by a certain process, and with the available technology, the product cannot be described by except by that process. The patentee claims the product “obtainable by” the process and separately “obtained by” the process. After grant, but before expiry, advances in the technology used to characterize products of that nature allows the product to be fully characterized and described by structure. Another party, armed with knowledge of the structure, then produces the exactly the same compound by a different process. Does that party infringe either of the claims? As a matter of policy, in your view, should it be considered to infringe?

Me:

In the situation that it subsequently becomes possible to structurally describe characterise the product, then I think under UK law as it stands, an “obtained by” claim is not infringed and an “obtainable by” claim is infringed.
I hope it will be clear to other readers that such a situation is difficult to imagine in reality. Usually things that cannot be characterised stay uncharacterisable for a long time.
But to answer your last point, as a matter of policy I think that it should be considered NOT to infringe. Essentially I suppose I am arguing that “obtainable by” language should not be able to be used by the patentee to expand the monopoly to cover things that only afterwards can be ascertained to infringe, because I think that it completely offends the Protocol to Article 69 – there is no certainty for third parties. It also offends the principle that the monopoly should be commensurate with the technical contribution – the real technical contribution is the process, even if the product is actually novel. (Here I completely disagree with ANON from Wednesday, 10 December 2014 at 10:12:00 GMT)
Otherwise, if “obtainable by” claims are allowed to cover the same product made by a different process, then the patentee has used his difficulty – “I cannot describe my product any other way” not only to get a product claim (rather than only a process claim), but to get a product claim that covers things that the patent itself gives no guidance as to whether they infringe or not. In your example, the patent has provided a process A, and an uncharacterised product X. Someone else, possibly the putative infringer himself, has to establish the structure X, the different process B, and that the different process B produces X, and not X’ or Y. In my view, the burden on third parties and the putative infringer is intolerable, compared with the little that the patent has contributed (really only A). If patentee’s difficulty to describe his product produces adverse consequences, these should be shouldered by the patentee, not by third parties.
It can even be framed as an estoppel point. To get the claim at all, the patentee has represented that the product cannot be described except by the process. But then to assert infringement, the patentee must represent that the product can be characterised in another way, necessarily not revealed in the patent, but only later. I don’t consider it far-fetched to say that the patentee should be estopped from this change in position, even if it is technically justified by interim developments.

Norman:

Your point regarding certainty for third parties is well taken. To expand my example, suppose that before the advance in characterizing technology, the second party develops the same product, but by an entirely different process. There will be no way for the second party to know whether it infringes, and indeed, the second party might also get an “obtainable by” patent for what is in fact exactly the same compound.
However, I’m not sure the principle that the monopoly should be commensurate with the technical contribution is offended, because I’m not convinced that the true contribution is necessarily only the process. I do agree that in some cases the true contribution would be only the process – eg where the desirability of isolating a compound is well known, and the patentee is the first to do it – and in such cases I agree that infringement should be limited to the particular process. But I’m not sure that is always the case. Suppose that the technical insight is that a certain plant compound has medicinal properties, and the only known way to obtain the compound is to extract it from the plant, but once the decision to extract is made, the extraction itself requires only well-known methods. The technology needed to characterize the product might subsequently advance by incremental unpatentable steps, until the point where the product can be readily characterized and synthesized, in which case the subsequent party would have contributed nothing inventive. Even if the characterizing technology or synthesizing technology were patentable, I would have no trouble with the subsequent party getting a patent on their particular novel process, but still, the thought of synthesizing the compound would never have occurred to them without the original invention. In this situation, I would say that the real technical contribution is the compound itself.
So, I don’t see this as a (relatively) easy case in which all policy considerations point in the same direction. It strikes me as a hard case where there is a real tension between the need for due notice to subsequent parties, and the need to reward the patentee for their contribution.

I would finally note that the disconnect between novelty and infringement has been endorsed in the USA by the CAFC in the case Amgen v Hoffmann-La Roche:

“In determining validity of a product-by-process claim, the focus is on the product and not on the process of making it”
“In determining infringement of a product-by-process claim, however, the focus is on the process of making the product as much as it is on the product itself”.

4) Words from another commenter (with whom I generally agree)
I was very grateful to Freddie Noble who expressed some principles clearly (and I think in a refreshing way), and I so I would like to repeat some of his contributions here:

Freddie:

I do like this decision. And I do not think it necessarily breaches the "Angora Cat" principle.
To start with you have a claim, and have to decide what it means. Fundamental to that is that is whether the claim is for a product or a process. If it is ambiguous in that respect then a clarity objection is proper.
Clearly this type of claim is for a product. So the question is - is the product novel? i.e. is it a new product? If the product is identical to a prior art product (albeit obtained by a different process) then the claim is not novel. I think this is what Birss means by a "rule of novelty" - notice there is no "construction" of "obtained by" at this stage, the "process" is just not relevant to the novelty of *the product*, apart from in so much as it defines the features of the product.
Then we come to infringement, and the question is "has the alleged infringer done what the claim says he must not do?" The "obtained by" claim says that the product, specified in terms of how it is manufactured, is the monopoly. If the product is not manufactured like that, then it does not infringe.
Alternatively, if the claim says "obtainable by" rather than "obtained by", then the alleged infringer has to make something which really is "obtainable by" the specified process. In other words, if it is made by a different process then it must be indistinguishable from the "obtained by" product.
The implication seems to be that it is (very slightly) better to say "obtainable by" than "obtained by" in your claim. And irrespective of whether the EPO might allow an "obtainable by" claim where there is an alternative, if there is an alternative way of defining the novel and inventive features of the product then that alternative is probably going to be better for the patentee.

Anon: “More than “very slightly” I would suggest”

Freddie:

Yes, you are right in that "obtainable by" instead of "obtained by" could make the difference between infringement and not, which in an important case could be worth an awful lot.
What I meant was that cases where it does make that difference are likely to be rare, bearing in mind the narrow definition of "obtainable by" - it "must have every single characteristic which is the inevitable consequence of that process".

Me:

I completely agree with you both in the sense that this is what the UK law is, following Kirin Amgen and as admirably now expounded by Birss J, and also as I think that it should be.
It is only at first blush that a fundamental precept of claim construction is breached; on careful consideration, as you have done, it is clear that this is not so.
Moreover, in my opinion, this approach is the only one that satisfies the Art 69 EPC requirement for an claim interpretation "which combines a fair protection for the patent proprietor with a reasonable degree of legal certainty for third parties". The "expansive" approach gives complete uncertainty for third parties, as Tim Roberts notes that such a claim is unconstruable.

What Tim Roberts wrote was:

"Product A obtained by process X" is a claim the scope of which can be decided by evidence (about the process used). What about "Product A obtainable by process X"? What does this include? It clearly includes cases (A') where A has been obtained by process X. But suppose A (A'')has been obtained by another process (Y,not X)? How do you tell whether A'' could have been been obtained by process X? In what sense could it have been so obtained, when it wasn't?
Clearly if the patentee can produce a product A''' produced by process X that is indistinguishable from A'', then A'' infringes. But suppose A''' and A'' differ slightly - or more than slightly? How do you construe the claim? Infringement (no doubt) is a matter of substance. But how do you tell what is an important difference and what is a trivial one? I would argue that the claim is unconstruable. It is no clearer than a claim to 'A substantially as described with reference to the description' - and we know what the EPO thinks of those.

[Clarifying the possible disconnect between novelty and infringement]

Freddie:

When novelty is being assessed, we deny that there is any construction of "obtained by" and instead say there is a "rule of novelty" which says that the product as such must be new. Part of the reason I gave for this was that the claim is either a product or a process - it can't be both - and if it's for a product then it is the novelty of the product which must be assessed.
Then when infringement is being assessed, we say that it is not enough for it to be the same product, it has to be really obtained by the process as stated. In other words, there is a "rule of infringement" which is not the same as the "rule of novelty".
There may well be some room to criticise this state of affairs - where infringement is assessed differently from novelty - but whether it is right or not I maintain that it is reconcilable with consistent construction of the claim, which is important.

…

As to whether all that is "right" in the sense that it should be the law, as well as in the sense that it is the law, well I think it is, because:
When we assess novelty of an "obtainable by" claim, we insist that the product is new, because it is the product which is being claimed.
When we assess infringement of an "obtainable by" claim, we insist that the infringing product really is obtainable by that process (which is really to say no more than that the infringing product has to be identical to the claimed product), because anything else leaves the scope of the claim uncertain and indeed subject to change in the face of new technology.
If it's an "obtained by" claim then, in certain circumstances (and still I have seen no convincing real-world examples) an identical product will not infringe, which might seem harsh, but then nobody forced the patentee to say "obtained by" instead of "obtainable by" , and if the anon comment @1213 is right, then the patentee has known very well for years that he had better say "obtainable by" to avoid that.

Thanks again to all the commenters for such excellent contributions.  I hope and expect that our dear readers may have further comments on this tricky and controversial topic where, perhaps surprisingly, the law is still not clearly fixed.

Addendum 12 December at 21:45 to add this picture of Claim 2 of EP0309477B, referred to in the comments below