Less than six months af­ter re­port­ing pos­i­tive Phase III re­sults for its lead pro­gram in treat­ing ag­i­ta­tion re­lat­ed to schiz­o­phre­nia and bipo­lar dis­or­der, small AI biotech BioX­cel fol­lowed up Tues­day with top-line da­ta in an ear­ly stage tri­al for ag­i­ta­tion in de­men­tia and Alzheimer’s dis­ease.

The ex­per­i­men­tal drug, known as BX­CL501, met its pri­ma­ry and sec­ondary end­points in the high­er evalu­able dose of a 54-pa­tient Phase Ib/II tri­al, BioX­cel an­nounced, in­duc­ing a sta­tis­ti­cal­ly sig­nif­i­cant calm­ing ef­fect com­pared to place­bo with no se­vere or se­ri­ous side ef­fects. With the da­ta in hand, BioX­cel plans to meet with the FDA some­time in the first half of 2021 be­fore launch­ing a late-stage study some­time af­ter­wards.

BioX­cel $BTAI shares ticked up by as much as 10% in pre-mar­ket trad­ing Tues­day, but af­ter the open­ing bell were down about 7%.

BX­CL501 is a re­for­mu­la­tion of Pfiz­er’s 21-year-old se­da­tion drug Pre­cedex. BioX­cel cre­at­ed a dis­solv­ing film-based, sub­lin­gual ver­sion of the drug — think Lis­ter­ine strips, but placed un­der the tongue — that pa­tients ad­min­is­tered them­selves. It es­sen­tial­ly pro­duces a calm­ing ef­fect with­out knock­ing a pa­tient out and ac­ti­vates the al­pha-2 re­cep­tor, the path­way through which nor­ep­i­neph­rine, among oth­er neu­ro-chem­i­cals, trav­els.

Tues­day’s da­ta come from a dou­ble-blind­ed and ran­dom­ized study con­duct­ed en­tire­ly with­in as­sist­ed liv­ing fa­cil­i­ties. Pa­tients in the drug co­horts re­ceived ei­ther a 30 μg, 60 μg or 90 μg dose, though BioX­cel did not mea­sure ef­fi­ca­cy or safe­ty in the 90 μg arm due to high­er ex­po­sure lev­els. Six­teen pa­tients re­ceived the 30-μg dose, 20 took the 60-μg dose, four be­gan the 90-μg dose lev­el and 14 were ran­dom­ized to the place­bo group.

It’s the 60 μg lev­el where BX­CL501 dif­fer­en­ti­at­ed it­self from place­bo at a sig­nif­i­cant lev­el. The pro­gram hit on all three scales used to mea­sure change in ag­i­ta­tion from base­line, the PEC, PAS, and Mod-CMAI. For the PEC, the p-val­ue was p=0.0011, while the oth­er two p-val­ues reg­is­tered at p<0.0001.

The can­di­date al­so showed a rapid on­set, BioX­cel said, hit­ting sta­tis­ti­cal sig­nif­i­cance com­pared with place­bo in the PEC and PAS just one hour af­ter pa­tients took the film strips, with the com­pa­ny not­ing it didn’t mea­sure Mod-CMAI af­ter this time pe­ri­od. Calm­ing ef­fects at this dos­ing lev­el re­mained at sta­tis­ti­cal­ly sig­nif­i­cant lev­els through eight hours.

BX­CL501’s 30-μg dose did not hit sta­tis­ti­cal sig­nif­i­cance in any of the three scales, clock­ing in at p=0.0813 for PEC, p=0.0961 for PAS and p=0.0591 in the Mod-CMAI.

The most com­mon side ef­fect was som­no­lence, or drowsi­ness. Half of the 16 pa­tients in the 30-μg co­hort self-re­port­ed mild drowsi­ness, while 11 of 20 in­di­vid­u­als in the 60-μg co­hort did so. There was one case of mod­er­ate drowsi­ness in the 60-μg group as well.

In ad­di­tion, there were two cas­es of hy­poten­sion in the 60-μg co­hort, one mild and one mod­er­ate, as well as one case of mild dizzi­ness.

Jef­feries an­a­lyst Chris How­er­ton took a pos­i­tive view of the da­ta, writ­ing that the 60-μg dose ap­pears to “check all the box­es.” The safe­ty at this dose lev­el looked “clean,” though How­er­ton not­ed be­cause Pre­cedex is known to be more po­tent in el­der­ly pa­tients, he wouldn’t be sur­prised if BioX­cel al­so looked at a 45-μg dose in their piv­otal tri­al.

Vi­mal Mehta

The is­sue of a 45-μg dose al­so came up in a con­fer­ence call BioX­cel held Tues­day morn­ing with in­vestors. CEO Vi­mal Mehta said that al­though BioX­cel will like­ly pri­or­i­tize 60 μg for the Phase III giv­en Tues­day’s top-line re­sults, they aren’t rul­ing out al­so test­ing 45 μg.

Mehta al­so said on the call that he doesn’t an­tic­i­pate much dis­rup­tion in Phase III en­roll­ment re­lat­ed to the on­go­ing Covid-19 pan­dem­ic.

BioX­cel is one in a slate of biotechs that has raised mil­lions in re­cent years on the promise of us­ing ar­ti­fi­cial in­tel­li­gence to speed the drug de­vel­op­ment process. The com­pa­ny filed for a $69 mil­lion IPO in 2018 less than a year af­ter be­ing found­ed, and has aimed to use its tech to find com­pounds val­i­dat­ed to some de­gree by oth­er com­pa­nies and re­pur­pose them.

Back in Ju­ly, BioX­cel re­port­ed that in two near­ly 400-per­son Phase III tri­als for bipo­lar dis­or­der and schiz­o­phre­nia, the lead drug ef­fec­tive­ly calmed pa­tients’ ag­i­ta­tion.

As more and more biopharmas develop artificial intelligence platforms, the drug discovery process is being reshaped to fit new goals on cutting down the prodigious amount of time, energy and money that go into a drug program. Now one of the most ambitious players in the drive to improve on ROI, AstraZeneca, is marking a milestone on that front by adding the first target generated by AI to its portfolio.