In April 2014 Celgene agreed to pay $710 million upfront for rights to GED-0301 (mongersen), an oral drug for Crohn’s disease. A Phase 2 trial had been positive. The Phase 3 trial found no benefit over placebo and was stopped in October 2017, and Celgene recorded an impairment of about $1.6 billion on the drug, largely offset by a release of about $1.4 billion in contingent payments, for a net pre-tax charge of about $411 million.
This page sets out what Celgene’s filings and the published papers say, including a 2022 analysis of manufacturing batches that offers a possible explanation, and the questions an audit would put to the evidence. Every figure below was read from the cited source. It does not claim that an audit would have prevented the loss.
| Apr 23, 2014 | Celgene signs a license agreement with Nogra Pharma for GED-0301, an antisense oligonucleotide targeting SMAD7, for Crohn’s disease and ulcerative colitis. Upfront payment of $710 million. Development and regulatory milestones of up to about $815 million, and sales milestones of up to $1,050 million if annual net sales reach $4,000 million. | Celgene 8-K, Apr 24 2014 |
| 2015 | Phase 2 results published. Patients took mongersen for 2 weeks. Clinical remission (the primary outcome, maintained for at least 2 weeks) was reached by 55% on 40 mg and 65% on 160 mg, versus 10% on placebo (P < 0.001). The study was funded by Giuliani. | Monteleone et al., NEJM 2015, PMID 25785968 |
| Dec 8, 2015 | Phase 3 trial REVOLVE starts, sponsored by Celgene. It enrolled 701 patients. | ClinicalTrials.gov NCT02596893 |
| Oct 19, 2017 | Celgene discontinues REVOLVE and its extension trial SUSTAIN after an October recommendation from the Data Monitoring Committee, which had reviewed an interim futility analysis. Celgene reports no meaningful safety imbalances. | Celgene press release, Oct 19 2017 |
| Q4 2017 | Celgene records a net pre-tax charge of about $411 million: an impairment of about $1,620 million on the GED-0301 in-process research and development asset, about $188 million of other one-time costs, and a reduction of about $1,397 million in contingent consideration liabilities. Its annual report notes that a Phase 2 trial in Crohn’s disease was complete when the drug was acquired. | Celgene 10-K, fiscal 2017 |
| 2020 | Phase 3 results published. Of 701 patients, clinical remission at week 12 was reached by 22.8% on GED-0301 and 25% on placebo (P = 0.6210). The authors conclude GED-0301 did not demonstrate efficacy versus placebo in active Crohn’s disease. | Sands et al., Am J Gastroenterol 2020, PMID 31850931 |
| 2022 | A new analysis of 27 mongersen batches finds marked differences in SMAD7 downregulation in a cell-based assay. Routine physicochemical tests made the batches appear identical; only phosphorus-31 NMR showed marked differences. The authors note the chirality of the 20 phosphorothioate linkages was not controlled in manufacturing and “tentatively suggest” this may be the origin of the different activity. | Arrico et al., Nucleic Acid Ther 2022, PMID 35263186 |
| 2022 | A review by the Phase 2 lead author and a co-author states that the majority of batches used in the Phase 3 study were chemically different from those used in earlier trials, with some unable to knock down SMAD7 in cultured cells. It calls for new trials using chemically homogeneous batches. | Monteleone & Stolfi, Pharmaceutics 2022, PMID 36678723 |
The net charge is smaller than the impairment because Celgene also reduced its contingent consideration liabilities by about $1.4 billion. The $710 million upfront payment is not itself the write-off.
These questions draw on the record as it developed. The Phase 2 paper was published in 2015, nearly a year after the April 2014 license; the Phase 3 result and the batch analysis came later still. None is a finding that the deal was wrong. The method behind them, tracing each load-bearing claim to its primary source and tagging each link supported, partial, contradicted or missing, is described on the main page.
Hindsight is cheap, and the 2022 NMR method may not have existed as a routine release test in 2014. What an audit can do is lay out, before the money moves, which links in the chain are strong and which rest on one trial, one funder, or one assumption that the product stays the same at scale.