First data with NantKwest’s cancer vaccine ‘unclear’

Patrick Soon-Shiong’s NantKwest unit has reported the first data for its “cancer memory vaccine,” but it has some analysts scratching their heads.

Phase 1b results with the immuno-oncology vaccine were presented at the Society for Immunotherapy of Cancer (SITC) meeting late last week and showed clinical responses in a handful of patients with heavily pretreated pancreatic cancer, triple-negative breast cancer (TNBC) and head and neck squamous cell carcinoma (HNSCC).

However, according to Biren Amin of Jefferies, “it’s unclear if the vaccine is driving meaningful responses,” so future readouts will be needed to get a clearer picture of the immunotherapy’s efficacy.

The word cocktail is often overused when it comes to combination drug treatment, but in this case it seems apt. The NantKwest’s vaccine contains no fewer than six components delivered over a 21-day cycle, including low-dose chemotherapy, cancer antigens delivered using viral and yeast vectors, an interleukin-15 agonist (N-803), a PD-1 inhibitor (Merck KGaA/Pfizer’s Bavencio) plus high-affinity natural killer (haNK) cells directed against CD16.

The aim is to rev up the patients’ own cell-based immune assault on the tumor, supplementing it with additional haNK cells, to try to overcome resistance.

In the phase 1b studies, there was a 90% overall response rate in 10 evaluable patients with pancreatic cancer who had been treated with at least three prior lines of therapy, all stabilized disease (SD). The median overall survival (OS) was 9.5 months, which was said to compare favorably with first-line chemotherapies, although Jefferies says that should be interpreted carefully.

“We caution against this type of assessment since it can be driven by post-progression therapy in addition to a small sample size of 10 patients,” write Amin and colleagues in a research note.

There was an 80% response rate in TNBC (four of five patients), including two SD and two immune-related partial responses (irPR), while in three HSNCC patients there was one SD and one complete response (irCR).

Jefferies is also uncomfortable with the use of the immune-related response criteria rather than the RECIST 1.1 measure acknowledged as the industry standard.

Billionaire biotech mogul Soon-Shiong—who made his fortune at Abraxane developer Abraxis after selling the company to Celgene for $2.9 billion in 2010—reckons that the cancer memory vaccine could “induce T cell memory and long-term durable response” and lead “to a chemo-free treatment of cancer patients, early in the course of their disease and ultimately to a path of cancer prevention.”

Jefferies’ base case interpretation of NantKwest is that while nearly all its NK projects are in early clinical or preclinical development, “we see potential for some programs to be on the market by 2021.”

Investors seemed to be excited by the results initially, driving shares in NantKwest up around 25%, but the stock quickly retreated again to pre-announcement levels.