[Updated, 2/16/18, 9:10 a.m. See below.] Using software to speed up drug discovery isn’t a new idea, but it’s generating more interest thanks to artificial intelligence. BioXcel Therapeutics is trying to parlay that momentum into an initial public stock offering to fund human studies on a pair of drugs it found with the help of AI.
Branford, CT-based BioXcel set a preliminary IPO target of $69 million, though that figure is a placeholder and likely to change as the offering creeps closer. The company has applied to the Nasdaq for a listing under the stock symbol “BTAI.”
BioXcel uses AI to discover medicines for neurological conditions and cancers. The company uses machine learning techniques to help find new therapeutic applications for drugs either already approved or tested in human studies. After BioXcel’s proprietary technology identifies a drug candidate, the company then files patent applications covering new methods or formulations of the drug. By working with drugs that already have safety and efficacy data from earlier studies, BioXcel hopes to take a speedier FDA review path. While this approach is meant to reduce the time and cost of drug development, it may also expose the company to patent infringement suits, BioXcel says in its IPO prospectus.
A growing number of companies are applying AI to drug development. NuMedii, a Menlo Park, CA, company, mines genetic data, clinical trial results, and other studies to develop drugs or find new uses for old ones. Engine Biosciences of San Francisco recently raised $10 million to back its AI-based drug discovery technology. Both NuMedii and Engine Bio use their AI systems to find drugs for larger pharmaceutical companies that are their partners. NuMedii also has a drug pipeline of drugs for rare inflammatory diseases and cancers. BioXcel aims to develop and test its own drugs after AI points the way. [Sentence added to reflect NuMedii’s drug pipeline.]
BioXcel’s two drugs are reformulated versions of previously developed treatments. BXCL501 is a different form of dexmedetomidine (Precedex), a now generic injectable drug sold by Teva Pharmaceutical (NYSE: TEVA) and others for anxiety and pain, and also as a sedative. BioXcel says the drug, reformulated as a sublingual film or an intravenous infusion, might treat the agitation caused by neurological and psychiatric disorders like Alzheimer’s disease and schizophrenia. Data from Phase 1 studies testing the IV versions in Alzheimer’s and schizophrenia patients are expected in the second half of the year. The company plans to start Phase 2 studies later this year.
A second candidate, BXCL701, is a version of an experimental cancer drug, a so-called dipeptidyl peptidase (DPP) inhibitor called talabostat, that failed a Phase 3 trial in lung cancer run by now-defunct Boston biotech Point Therapeutics. BioXcel says its drug is more precise than other DPP blockers, and is developing its version for pancreatic cancer, a rare form of prostate cancer, and possibly as a combination candidate with other immunotherapy treatments. The company plans to start Phase 2 studies for the cancer drug this year; preliminary data are expected in the first half of 2019.
BioXcel was formed last March by Connecticut-based BioXcel Corp., which owns the AI technology used to discover the startup’s drugs. The parent owns 93 percent of BioXcel, and is expected to own more than half the company’s shares even after the IPO, according to the filing.
BioXcel spent $2.6 million on R&D in 2017 and had just $887,000 in cash at the end of the year. BioXcel will use the IPO cash to take its two drug candidates through mid-stage studies, and perhaps further. It also needs to make payments to its parent for licensing the two therapies. BioXcel would pay $1 million upon completion of the IPO, plus another $1 million pegged to clinical trial milestones. If BioXcel brings any of its drugs to the market and reaches $50 million in sales, it would owe its parent another $5 million. The parent company also stands to gain royalties from BioXcel’s drug sales.
Image by VLADGRIN via Depositphotos.