Threshold Pharmaceuticals — Update 28 April 2016

Last December, Threshold announced that evofosfamide had failed to meet the primary endpoints in its Phase III pancreatic cancer trial as well as its Phase III soft-tissue sarcoma trial. Following these results, the company discontinued enrollment in all its company-sponsored trials of evofosfamide and its partnership with Merck ended. Also, the company let go approximately 40 employees, representing around 60% of its total internal staff.

After a deeper analysis of the data, Threshold noted a significant survival benefit demonstrated in Japanese patients, who made up 116 of the 693 patients in the pancreatic cancer trial. The median overall survival for those patients was 13.6 months in the treatment arm versus 9.1 months in the control arm for a significant p-value of p=0.0106.

As a comparison, in a trial consisting of 34 Japanese patients with unresectable pancreatic cancer, therapy of gemcitabine combined with Abraxane, which was approved in Japan in 2014, exhibited a response rate of 58.8%, PFS of 6.5 months and OS of 13.5 months. While the Abraxane response rate and PFS data were better, the median overall survival numbers were nearly identical despite the fact that the evofosfamide trial treated generally sicker patients (only 38% had an ECOG performance status of zero, which denotes a fully functional patient with little impact of the disease upon their lives, compared to 62% in the Abraxane trial).1

The question then becomes whether the current data package is enough for filing and approval in Japan, or whether Threshold will have to run a prospective study. We will not know the answer until the company meets with the PMDA, which will likely be in a few months. The arguments for the data package being approvable are that the Japanese dataset is rather large at 116 patients, statistical significance was reached within this subgroup across response rate, PFS and OS endpoints, as well as a substantially similar toxicity profile to Abraxane (see Exhibit 2), with mostly manageable toxicities. Plus, in terms of toxicity, one advantage for evofosfamide over Abraxane is that no incidence of peripheral neuropathy was observed at a grade of 3 or above and only around 10% of any grade. Abraxane had 11.8% of patients with grade 3 and above and 82.4% of any grade.

Arguments against Threshold being able to file or gain approval on this data include that the Japanese subset was not prospectively defined and so this result could just be a statistical anomaly that was discovered due to data mining. In addition, the reason why evofosfamide is especially efficacious in Japanese patients is unknown. Evofosfamide is a prodrug of a nitrogen mustard alkylating agent, which is activated under hypoxic conditions by a metabolizing enzyme whose levels may vary between individuals. The company hypothesizes that it may be activated differently in Japanese patients due to pharmacokinetic data. However, as that did not equate to improved median PFS (the median PFS of the treatment arms are identical on an ITT and Japan-only basis), it is unclear why this difference would have led to improved median overall survival. Finally, while Abraxane was approved in Japan with only a 34-patient trial, it might have been aided by the fact that it had already been approved in the United States on the back of statistically significant results in a pivotal trial. The Japan study was basically just a bridging study to make sure the drug had activity in Japanese patients. However, while the PMDA will not be able to find comfort in a prior FDA approval or even a positive Phase III trial (on at ITT basis), evofosfamide plus gemcitabine does seem to be at the top of the range in terms of efficacy (both in PFS and OS) when compared to other treatment regimens previously tested in pancreatic cancer (see Exhibit 3).

Although Japan’s total population is almost a third the size of that of the United States, the number of new pancreatic cancer diagnoses is only 25% smaller, Japan has the highest incidence of pancreatic cancer in the world, nearly double that of the United States. This is largely due to the fact that the elderly are a disproportionately large percent of the total population in Japan.

Our model assumes that evofosfamide will be approved in Japan in 2018, eventually achieving 16% market share of all pancreatic patients (20% of those with advanced/metastatic disease). Also, due to a study by EvaluatePharma on pricing in Japan, which suggested a 40% discount to US pricing, we have adjusted down our original US pricing assumptions by that amount to come up with a starting price of $48,000 per patient. On that basis, we estimate peak sales of $379m by patent expiration at the end of 2027. However, due to the unknowns surrounding the approvability of subset data such as these, we are modelling a 30% chance of success for the program, which we will adjust upon greater clarity from the company, potential partners and/or the PMDA.

We expect Threshold to eventually partner this product as it would not be able to commercialize it itself and there are a number of Japanese pharma companies in need of marketed oncology products. However, as Threshold does not need a partner immediately to file or even to conduct additional clinical work in Japan, we are assuming a partnership sometime in 2017. We currently model the partnership with a 22.5% royalty, as well as a $30m upfront and $10m in milestones.

We have reduced our valuation for the company from $1,023m or $14.30 per share to $178m or $2.49 per share, mainly due to the suspension of all evofosfamide development by the company except for Japan. Hence, we are no longer valuing evofosfamide in soft-tissue sarcoma, non-small cell lung cancer, glioblastoma or multiple myeloma as well as the US/EU regions for pancreatic cancer. We have assigned a 30% probability of success for the pancreatic program in Japan due to the unknowns regarding the path forward. Also, we have cut the probability of success for tarloxotinib from 20% to 10% due to uncertainty regarding the direction of the company following the restructuring and as the company conserves cash ahead of clarity regarding evofosfamide in Japan.

As of year-end 2015, Threshold had $48.7m or $0.68 per basic share in cash and cash equivalents. Following the announcement of the suspension of enrollment in all clinical trials outside of tarloxotinib, the company has let go approximately 40 employees, representing around 60% of its staff. After these changes, our forecasted operating loss for 2016 has decreased to $18.3m from $38.9m (note that it was technically profitable in 2015 solely due to the accelerated revenue recognition of formerly amortized payments related to the Merck partnership). Based on our calculations, it should have enough cash to continue operations through 2017. With the potential for Japan approval in 2018, we currently expect profitability in 2020 (previously 2018) and a funding requirement of ~$30-35m (previously ~$54m).