Patent goo: self-replicating Paxil
Self-replicating chemicals are not merely hypothetical: since Cat's Cradle, scientists have discovered some real-world example of crystals that seed the environment, converting other forms (polymorphs) of the crystal into their own. The population of the original polymorph diminishes as it is converted into the new form: it is a “disappearing polymorph.” In 1996 Abbott Labs began manufacturing the new anti-AIDS drug ritonavir. In 1998 a more stable polymorph appeared in the American manufacturing plant. It converted the old form of the drug into a new polymorph, Form 2, that did not fight AIDS nearly as well. Abbott’s plant was contaminated, and it could no longer manufacture effective rintonavir. Abbott continued to successfully manufacture the drug in its Italian plant. Then American scientists visited, and that plant too was contaminated was contaminated and could henceforth only produce the ineffective Form 2. Apparently the scientists had carried some Form 2 crystals into the plant on their clothing.
Another instance of the “disappearing polymorph” may be the anti-depressant, Paxil (U.S. brand name for the chemical paroxetine hydrochloride). No, self-replicating Paxil doesn’t naturally spread into our brains and make people happy for free. It's not "happy goo." On the contrary, self-replicating Paxil converted, according to one of the parties in the ensuing lawsuit, an old, and now off-patent, form of Paxil into a new, patented form of Paxil. Once the new form, the hemihydrate form of Paxil, was created, its crystals started floating about, converting small fractions of the old form, anhydrous Paxil, into hemihydrate. Both forms of the drug work equally well as an anti-depressant, but it became impossible to manufacture the off-patent anhydrate without some of it being converted into the patented form. Call it "patent goo."
Apotex, a generic drug manufacturer, was all set up to manufacture the off-patent anhydrous generic Paxil when it discovered small fractions of it were being converted into the hemihydrate. They couldn’t remove the contamination. Smithkline, owner of the patent on the hemihydrate, sued them for patent infringement. Apotex argued that the hemihydrate form occurred naturally, so that Smithkline’s patent was invalid. Smithkline argued that it was a disappearing polymorph, that the hemihydrate form had not existed before they had created it in their labs, and that it was up to Apotex to remove the hemihydrate from its product or pay it a royalty. Apotex was unable to remove the hemihydrate and unwilling to pay a royalty.
Judge Richard Posner heard this case in the trial court and wrote an opinion that contains a good explanation of the self-replicating Paxil controversy. The Federal Circuit heard the appeal and decided that Smithkline’s patent on the hemihydrate was invalid as “inherently anticipated” because anhydrate naturally converts into hemihydrate. Normally, anticipation would require an actual reference describing the claimed chemical structure (in patent lingo that the hemihydrate was "taught in the prior art"). But Judge Rader held that inherent anticipation occurs when, more likely than not, an operation that is taught in the prior art would result in the claimed chemical. The anhydrate which was taught in the prior art would more than likely result in natural creation of some hemihydrate. Judge Gajarsa in concurrence argued that the drug was discovered not invented, making it unpatentable subject matter. Gajarsa’s opinion may have inspired the United States Supreme Court to raise the subject matter issue on its own (i.e., it had not been argued by the parties to the case) in Metabolite. The Supreme Court is considering whether to take the appeal on the self-replicating Paxil case as well.