Patent applications are no fun to write. You have to figure out just what you're trying to cover (and how wide a space around it you want to try to clear), and the lawyers have to whip up language that casts just the right legal spell. The chemists have to write up detailed experimental procedures for all the important compounds and procedures, gather the matching analytical data, and make sure that it all fits together. Just getting the numbers assigned to the compounds right (and keeping them right through all the revisions) is a tedious job in itself. You always have to go through more drafts than you thought. No one enjoys it.
So maybe it's not surprising that things sometimes, well, slip a little. But how about when they slip a lot? Take this morning's example from Merck (a company that pitchforks out patent applications by the pile). Their WO2009091856 just published recently, directed at bicyclic beta-lactamase inhibitors. And everything looks normal for quite a while - 120 pages or so, in fact. Then the text suddenly snaps into bold face, and an authorial voice makes itself heard:
It appears that the data in Table 3 were generated in the same manner using the same enzymes as in Table 2 (Table 2 is unchanged from the provisional filing. I plan to DELETE the entries in Table 3 for Ex. 2,6,7 and 8 because this data duplicates the data in Table 2. . .Also, the entries in Table 2 for Ex.8 are both "1.6", not "16" as shown in Table 3. Please clarify these differences. . .
Is the data shown for Ex 1A data generated in a separate run, or is it supposed to be the same as for Ex. 1?. . .You don't want to include synergy data for these compouds (sic). It would be helpful to include it, at least for some of the examples (could put in a separate table). Recommend we include it for Ex. 14, since this is a likely backup candidate.
Now that's not supposed to be in there! What you're reading are the comments of someone in Merck's legal department - the sorts of comments that every patent draft collects as it's written, the sorts of comments that are supposed to be excised before you send in the application. Not this time! So if you were wondering which compound in this application represents the real candidate, and which ones are the backups to it, well, wonder no more. That query about including synergy data, for example, is an attempt to make it harder to figure out the most preferred compound itself - in vain, as it turns out. Oh, and those corrections that the comments say should be made? They weren't. So you'll want to fill in the correct numbers yourself.
That sort of thing goes on all the time in patent writing. You have to disclose your best compound - and in fact, you have to "teach toward" it in the claims. But you don't have to spray-paint it orange, and there's no sense in making things easy for your competition to figure out. A careful analysis of a patent application's claim structure will narrow down what a patent's authors are really interested in protecting, but there's often still some doubt about which exact compound is the winner. There can be other clues. Sometimes it'll be the compound with the most extensive biological characterization, or sometimes, if you look through all the experimental procedures, you'll notice that everything's being done on 50-milligram scale until one prep jumps to twenty grams. Bingo! Careful preparation of an application can scrub most of this stuff out. But all is for naught if your legal team's strategy comments are included in the Special Bonus Director's Cut version of the application.
Oh well, bonus dormitat Homerus. Anyone who's interested in beta-lactamase inhibitors (which, I should add, I'm not) now has more data to work with. The patent analysts at Thomson-Reuters are the people who caught this mistake (a colleague forwarded their writeup on to me). As they note, the wayward legal paragraphs also mention the possibility of comparing compounds to "MK-8712". That MK designation is Merck's usual method of showing a compound that's been recommended for the clinic, but this is the first that anyone seems to have heard about this one. But we can be pretty sure of something: someone in Merck's legal department has had a very bad day of it within the past couple of weeks. . .
1. D. on August 5, 2009 9:17 AM writes...
This is interesting. There's also been some work recently on clavulanic acid, which is in IIa right now for depression and ED, assuming Rexahn doesn't run out of cash. Their proposed indications are "depression and anxiety disorders, Parkinson's disease, Alzheimer's and neurodegenerative illnesses, neuroprotection and biodefense uses."
Permalink to Comment2. TT on August 5, 2009 10:14 AM writes...
It is more likely that the synergy comments were an effort to provide data, which would trump a possible obviousness rejection.
Permalink to Comment3. Chrispy on August 5, 2009 2:29 PM writes...
A lot of us in Wonder Drug factories have the eye-opening privilege of trying to reproduce work from patents and the literature, and it is absolutely astounding how much of it is not reproducible. I am not sure which is worse, but patents do have more gross, cut-and-paste type errors.
I hope that someday, Derek, you'll do a post about the irreproducibility of literature. I know you've covered a few examples of out-and-out fraud, but the amount of stuff which gets out there which is just plain wrong (or they got "lucky") really makes me not believe anything I read these days.
Permalink to Comment4. milkshake on August 5, 2009 4:28 PM writes...
You can tell when there is "prophetic" example baloney - I never even bother with those. A vague-sounding procedure with no yields, no solvent volumes and no NMR data is always a warning sign of crappiness.
My guess is that some people crank up compounds in hurry, without bothering to write down much in their notebooks. Two years later when the time comes to write-up on how they made their analogs, they don't remember much so they 'decipher' their own cryptic notes into some kind of plausible procedure that often has few key details missing...
Permalink to Comment5. Me on August 5, 2009 8:57 PM writes...
It is extremely ironic that it is Merck involved in this. Merck usually includes exactly no biodata in their patents. They describe the method of testing, but do not tell what any compound actually did. I understand the motivation of not disclosing data for individual compounds so that your competition doesn't know what the good compounds are. However, they are quite likely to publish a paper in BOMCL about the same time which actually includes IC 50s. Why does the patent office allow this? OR better still, why don't their competitors complain that they are not disclosing their "best mode"?
Permalink to Comment6. Pat Pending on August 5, 2009 11:16 PM writes...
All I can say is there but for the grace of God goes I.
Milkshake a prophetic example would be in present tense while an example that was actually performed would be in past tense
Me, the patent office (USPTO) does not care in that Examiners do not examine or look for "best mode" when examining a patent. However, in litigation, best mode will come up, especially if the best mode is not disclosed or there are efforts to hide the best mode of carrying out the invention. The penalty for hiding the best mode is that the patent will be invalidated by the court. The competitor that will complain would be the generic trying to invalidate the patent,arguing that the best mode was concealed.
Permalink to Comment7. LFree on August 6, 2009 12:28 AM writes...
love the latin and 'indeed' literary devices eh
Permalink to Comment8. Petros on August 6, 2009 1:00 AM writes...
Me's comments indicate what Merck's current philosophy is but there was a time when they belived in much more extensive disclosure (ca 20 years ago). Why this changed I don't know.
Permalink to Comment9. NoName on August 6, 2009 8:34 AM writes...
You may want to check this one ...
WO2004/091480 page 35 bottom
Jim, have we made any of these geometric isomers ...
see also page 38.
Permalink to Comment10. marklar on August 6, 2009 11:02 AM writes...
Here's another one:
WO2004/098589 p147 entry 0281
How was your weeked?(sic)
Permalink to Comment11. z on August 6, 2009 3:34 PM writes...
About reproducibility. A lot of it isn't necessarily fraud, or prophecy, or even sloppy notebook-keeping (although all of the above happen), but some chemistry is touchy, and one person's way of doing things is not the same as another person's. These differences aren't usually important so we don't usually write them down with extreme detail. In the cases where they are important, we don't know that because nobody else is trying to repeat the same chemistry. We may suspect a reaction of being more touchy but not have a way of saying that--a hunch that this one might be more sensitive, so I'll be extra careful with the flame-drying and adding reagents under inert gas streams, etc. Maybe we should write more of this stuff down? But if we don't know (and don't want to spend time playing around to find out), then it would give the person reading it the impression that we know more about a reaction's touchiness than we really do.
Permalink to Comment12. BLA on August 6, 2009 7:49 PM writes...
More examples, please!
Permalink to Comment13. Hap on April 6, 2010 7:30 PM writes...
Still no news on Merck's wayward drug candidate - no one else refers to "MK-8712" other than people referring to In The Pipeline.
Remind me, though, if I need any free info from BNet, it's not worth it. If our torpeodes and missiles were as persistent as their ad popovers, we wouldn't have too many wars.
Permalink to Comment14. Hap on November 15, 2011 3:27 PM writes...
Finally - here's a a paper in BMCL which reveals exactly what MK-8712 is. I don't know if it matters now, though.
Permalink to Comment