MSN Made a Different Cabozantinib Crystal and Still Lost the Cabometyx Case

Cabozantinib, sold as Cabometyx, is Exelixis's tyrosine kinase inhibitor for kidney, liver, and differentiated thyroid cancer. The drug substance is a malate salt, and like many pharmaceutical salts it does not exist as a single fixed structure. Dissolve it, crystallize it under different conditions, and the molecules can pack into different repeating arrangements: different polymorphs, each with its own density, melting point, solubility, and moisture behavior even though the chemical formula never changes. Exelixis's own 2015 FDA submission said as much: cabozantinib (L)-malate had been found in two closely related crystalline forms, named N-1 and N-2, and "no other forms were identified" at the time.

That single sentence is where this case starts. Exelixis's later patents (US 11,091,439, 11,091,440, and 11,098,015, collectively the Malate Salt Patents) did not limit their claims to N-1 or N-2 specifically. Claim 1 of the '439 patent claims cabozantinib malate salt "wherein said salt is crystalline," full stop. That is a genus claim covering every crystalline form the salt could take, not just the two the inventors had actually isolated and measured.

MSN Laboratories, the Hyderabad-based generic manufacturer, filed its own abbreviated new drug application in 2019 using a different crystalline form it calls Form S, on which MSN separately holds its own patent. Having built a genuinely different crystal, MSN's natural argument would be non-infringement. It never got there. Because the asserted claims covered crystallinity generically, MSN conceded infringement outright and pivoted its entire defense to invalidity: the patent, MSN argued, could not adequately describe a genus of every possible crystalline form when the specification only ever measured two of them.

MSN's strongest remaining argument was that N-1 and N-2 have measurably different densities, melting points, solubilities, and stability profiles, and that those differences show a skilled chemist could not have predicted the properties of an undisclosed form like MSN's own Form S. The court rejected this on a narrow but important ground: the district court had never relied on those specific physical properties to define genus membership in the first place, only on the shared chemical formula and crystallinity. MSN's argument attacked a foundation the ruling was not actually built on. The court also noted, without dispute from MSN, that a "pure polymorph" of this particular salt can only take a maximum of fourteen possible crystalline forms, a chemically bounded universe that made this a much easier case than the sprawling antibody-sequence or Markush superfamily disputes where genus claims usually run into trouble. In one of those harder cases, AbbVie v. Janssen, the disclosed antibodies shared only 50 percent sequence similarity with the accused ones, against 90 percent shared among the disclosed set: a concrete, measured gap. MSN never produced an equivalent number for its own Form S.

A separate patent in the same litigation, covering cabozantinib formulations kept below 200 parts per million of a genotoxic synthesis byproduct, went the other way for Exelixis: the district court found MSN's product did not infringe that claim, and once Exelixis dropped its own cross-appeal on the point, the Federal Circuit vacated the underlying validity ruling as moot rather than reaching it on the merits. The opinion's own footnote records that Exelixis has since obtained a continuation patent without that impurity limitation and has already sued MSN, along with Azurity Pharmaceuticals and Slayback Pharma, over it. That case is active and unresolved, and nothing here should be read as a view on how it comes out.