On 30 October 2026 the US Court of Appeals for the Federal Circuit reversed a $9,151,133 award of attorneys’ fees against ChromaDex, Inc. and the Trustees of Dartmouth College, outright and without a remand. The case is ChromaDex, Inc. v. Elysium Health, Inc., No. 2025-1227, on appeal from the District of Delaware (No. 1:18-cv-01434-CFC-JLH, Chief Judge Colm F. Connolly). Judge Chen wrote for a panel of Judges Prost, Chen and Stoll — the same panel that, in February 2023, had affirmed the summary judgment holding ChromaDex’s asserted claims ineligible under 35 U.S.C. § 101 as directed to a product of nature (ChromaDex I, 59 F.4th 1280). The disposition is marked nonprecedential, so it binds nobody; what it shows is how this panel reads Octane Fitness when a losing party is asked to pay.
The patents, US 8,197,807 and 8,383,086, claimed dietary supplement compositions containing isolated nicotinamide riboside (NR), a form of vitamin B3 that occurs in trace, non-bioavailable amounts in cow’s milk. After losing the merits appeal, Elysium moved under § 285 on a single theme: that ChromaDex had litigated "as if the Myriad case did not exist." Chief Judge Connolly agreed, finding in March 2024 that the holding of Association for Molecular Pathology v. Myriad Genetics, 569 U.S. 576 (2013), was "crystal clear" and dispositive, and noting that ChromaDex’s summary-judgment opposition mentioned Myriad only once. He fixed the award in August 2024 at $9,151,133. The Federal Circuit held that treating Myriad as the beginning and the end of the eligibility analysis was legal error, and that a district court basing a ruling on an erroneous view of the law necessarily abuses its discretion. ChromaDex’s position had not rested on isolation alone; it rested on an analogy to Natural Alternatives International v. Creative Compounds, 918 F.3d 1338 (Fed. Cir. 2019), which required the court to work through markedly-different-characteristics, the quantity of NR the claims actually recited, and NR’s causal role in NAD+ biosynthesis. On the panel’s assessment, that was not "open and shut in light of Myriad."
Wrong and exceptionally meritless are two different findings
The useful line in the opinion is the one the court borrowed from SFA Systems v. Newegg, 793 F.3d 1344: a party’s position on a question of law need not be correct in order not to stand out. Exceptionality under § 285 turns on substantive strength, not on the eventual outcome, and the panel restated the bar in the terms used in Realtime Adaptive Streaming v. Sling TV, 113 F.4th 1348 — the position must be "exceptionally meritless." ChromaDex lost its patents, lost its appeal, and was still not exceptional.
That matters most to whoever is deciding, right now, whether to assert a composition claim over a natural product in the face of adverse Supreme Court authority. The honest reading is that a losing § 101 position survives a fee motion when the court had to do real work to reject it. The corollary is harsher: if your eligibility argument reduces to the single distinction that the adverse case already rejected, you are in the zone where fees are available, because no additional analysis was required to beat you.
The record that saved ChromaDex was made years before the fee motion
Nothing ChromaDex argued in 2025 won this appeal. What won it was material already in the district court file: an affirmative eligibility theory tied to a favourable Federal Circuit decision, and an expert report opining that the claimed compositions had utility that naturally occurring NR does not. The panel cited both. Treat that as the operative instruction. The time to build an exceptionality defence is while you are briefing summary judgment, not after you have lost — and the content of the defence is a second, non-derivative ground for your position, evidenced, on the record.
The flip side is the trap ChromaDex walked into and spent two and a half years climbing out of. Chief Judge Connolly counted the single mention of Myriad in the opposition brief and read the silence as unreasonableness. He was reversed, but the reversal took an appeal. Distinguish the controlling adverse case expressly and at length, even when you believe your argument sits outside it, because a district judge reading a thin treatment of the case against you will supply the inference that you had no answer.
Fee exposure runs from the complaint, not from the day the case got weak
The amount is the part portfolio owners should sit with. ChromaDex objected that roughly $3.9 million of the award covered work done before the Markman hearing — before the claim construction that made § 101 fatal. The district court included it anyway, reasoning that the outcome would not have changed under ChromaDex’s own proposed construction. So the realistic downside case on a weak eligibility assertion is not the fees your opponent runs up after the turning point; it is the fees for the whole action. For a plaintiff deciding whether to file, that is the number that belongs in the memo, and it is the argument for resolving eligibility exposure through a hard-nosed patent invalidation analysis of your own claims before the complaint rather than after discovery.
There is one shot at the exceptionality record, and it belongs to the movant
Footnote 5 is where the opinion bites. Elysium asked for a remand so the exceptionality question could be revisited. The court refused, holding that neither the district court’s analysis nor the record and argument Elysium had actually presented could support an exceptionality finding, and quoting Fox v. Vice, 563 U.S. 826, for the proposition that a fee determination should not become a second major litigation. A movant who overstates the case therefore does not get a second attempt at a narrower theory. Costs on the appeal were taxed to ChromaDex’s side.
Read against the reflex the decision describes — move for fees after every win, because what is there to lose — the answer is now visible: nine million dollars of expectation, a second appeal, and the costs. Plead the exceptionality theory you can actually prove. "They ignored the controlling case" is not that theory unless the docket shows they genuinely had nothing else.
Two drafting and strategy points the eligibility analysis exposes
First, the claims failed in ChromaDex I partly for reasons a drafter controls. The panel noted the claims required no specific quantity of isolated NR and did not attribute the increased NAD+ biosynthesis to the isolated NR, while the comparable claims in Natural Alternatives recited an amount effective to produce the asserted effect. If you are claiming an isolated natural product, recite an effective amount and tie the functional result to the isolated component. That difference is what separated the two cases.
Second, the claim-construction history is a warning about relying on a PTAB win. Elysium petitioned for inter partes review of claims 1–3 of the ‘807 patent in July 2017; the Board construed "isolated" to require NR at about 25% purity by weight and, in January 2018, denied institution because milk does not anticipate claims so construed. ChromaDex then sued, and the district court declined that construction, reading "isolated" as separated or substantially free from at least some of the other source components — which, as the Federal Circuit explained, left the NR potentially still bound to whey protein and therefore not necessarily bioavailable. The construction that defeated the anticipation challenge was not available to defend eligibility, and the Board’s reading did not travel to Delaware. An institution denial is a win on one ground under one construction; it is not a validity shield, and the construction that saves a claim from prior art can be the construction that loses it on § 101. Anyone building a validity position across both forums should test it under every construction the claim can bear, which is the work a litigation-grade prior art search is for.
Because the disposition is nonprecedential it cannot be cited as binding authority, and the eligibility holding of ChromaDex I is untouched — the patents remain invalid. What changed is the price of having asserted them.