Lilly Confirmed the Calculations Were Wrong. A Jury Is Deciding What That Cost Nektar.

Nektar Therapeutics and Eli Lilly are in front of a jury in San Francisco over a drug Lilly handed back in 2023. One fact is not in dispute: Lilly confirmed it miscalculated efficacy data it had presented at a medical congress. Everything else is contested, and Nektar’s damages expert has put a number on it.

The arithmetic is agreed. The consequence is not.

Lilly confirmed the errors. That distinguishes this case from most partnership disputes, where the parties argue about what the data meant. Here both sides accept that what was presented at EADV in 2022 was wrong.

Nektar’s account of the mechanism is specific. It alleges the atopic dermatitis analysis miscalculated against the validated 72-point EASI scoring system and excluded patient data available at the interim, and that the psoriasis endpoints were similarly miscalculated against the 72-point PASI system.

The corrections run in one direction. Every restated figure is more favourable to the drug than the original. Mean EASI improvement at the top dose rose from 66% to 83%, and the placebo-adjusted figure roughly doubled from 17% to 36%.

Lilly’s position is that the termination decision rested on the failed lupus trial rather than on the dermatology numbers, and that it walked away on the data. Its executives have argued at trial that Nektar was seeking to salvage the programme after setbacks elsewhere in its pipeline. Those are the two cases the jury is weighing.

The drug Lilly returned went on to hit its endpoints

This is what changes the damages question from theoretical to arithmetic. After regaining rights, Nektar took rezpeg into a Phase 2b study in biologic-naive patients with moderate to severe atopic dermatitis. The readout was positive across endpoints, and Nektar shares rose sharply on it.

A jury assessing what a terminated licence was worth normally has to imagine a counterfactual. Here a substantial part of the counterfactual has already been run, by the smaller party, after the larger one let the asset go.

That does not establish that Lilly acted improperly. A company can decline an asset that later succeeds, and doing so is not misconduct. It does mean the jury is not being asked to value a hypothesis.

The transferable lesson is about who controls the analysis. Under this structure Lilly ran the studies, performed the statistics and presented the results, while Nektar owned the molecule and bore the consequences of how those results read. Nektar did not find the errors until the data were transferred back on termination, which is to say it could not audit the arithmetic behind its own asset while the partnership was live. Any licence where the partner runs the trials should specify who holds the raw data, on what terms the originator may re-analyse it during the collaboration, and what verification precedes a congress presentation carrying the originator’s asset name. Those provisions cost nothing to negotiate at signing and are unobtainable afterwards.

Two procedural details worth holding

Lilly counterclaimed in March 2024, alleging Nektar breached confidentiality provisions and defamed it. In October, Lilly voluntarily dismissed that counterclaim. A party that withdraws its own claim before trial is narrowing the dispute to the defence, and juries are not told why.

The trial itself was scheduled for 27 October 2025 and postponed by the federal government shutdown. That delay pushed a damages claim of this size back by roughly ten months, during which the Phase 2b result arrived. The sequencing is accidental and it is not neutral.

What to watch

Whether the jury separates the confirmed miscalculation from the termination decision. Lilly has conceded the first while defending the second, and the entire case turns on whether those are treated as one act or two.

Whether the near $1 billion figure survives cross-examination. Damages experts in licence disputes work from projected milestones, royalties and lost time, and the Phase 2b result makes the underlying asset easier to value than it was in 2023.

And what the verdict does to congress data practices. If a jury attaches a nine-figure number to statistics presented at a dermatology meeting, sponsors will look again at who signs off on the numbers before a partnered asset is presented publicly.

Neither side disputes that the published figures were wrong or that the corrected ones favour the drug. What remains is whether that arithmetic changed the outcome for the asset, and whether Lilly owes anything for it. That question sits with a jury in San Francisco and has not yet been answered.