In families that carry a mutation in the granulin gene, the arithmetic is cruelly simple. One working copy instead of two means roughly half the normal supply of a protein called progranulin, from birth, in every cell that makes it. For decades nothing seems to happen. Then, usually somewhere between the forties and the sixties, the personality starts to shift, or the words stop arriving, and a scan shows the frontal and temporal lobes thinning. The obvious fix is to put the protein back. The obstacle is equally obvious: progranulin given into a vein largely stays in the blood, because the blood-brain barrier is very good at its job.
DNL593 is Denali Therapeutics' attempt to talk its way past that barrier rather than break it down. The molecule links progranulin to a modified antibody Fc domain engineered to bind the human transferrin receptor, the same receptor the brain normally uses to pull iron across its vessel walls. In work published in Cell on 26 August 2021, Denali scientists and collaborators reported that this construct, called PTV:PGRN, improved delivery of progranulin into the brains of progranulin-deficient mice and corrected lysosomal abnormalities that accompany the loss, including a deficiency in the endolysosomal lipid BMP. That work was done in mice and cells. It is the reason the programme exists; it is not evidence about people.
The human study, registered as NCT05262023, was designed in three pieces and described that way from the start: Part A, single ascending intravenous doses in healthy volunteers; Part B, multiple doses over 25 weeks in participants with frontotemporal dementia carrying a GRN mutation; Part C, an optional 18-month open-label extension for those who complete Part B. It is multicentre, randomised, placebo-controlled and double-blind, and it has been listed among the frontotemporal lobar degeneration trials recruiting through UCSF. Dosing in Part A began in March 2022. In an interim announcement on 1 November 2022, Denali said single doses produced dose-dependent increases in progranulin in cerebrospinal fluid and were generally well tolerated, which is what cleared the way into Part B. Those were interim, company-reported, healthy-volunteer findings, in a small early-phase cohort, and they have not appeared as a peer-reviewed publication.
This year the programme has moved in ways that have little to do with biology. On 3 April 2026 Denali said Takeda had ended the co-development collaboration, returning full rights to Denali; the company stated the decision was driven by strategic considerations and "is not related to efficacy or safety data". At that point results were promised by the end of 2026. Four months later, in its second-quarter update on 6 August 2026, Denali reported that enrolment in the Phase 1/2 study is complete with 40 participants with FTD-GRN, that the FDA had granted DNL593 orphan drug designation in August 2026, and that results are now expected in 2027, a change the company attributed to allowing a longer observation period. Orphan designation is a regulatory status tied to rare disease development incentives. It says nothing about whether a drug works.
For a family watching this, the honest translation is that the delivery question and the treatment question are two different questions, and only the first one has anything resembling an early answer. Getting more progranulin into cerebrospinal fluid is a measurement. Whether restoring progranulin in an already-symptomatic brain changes how a person thinks, speaks, behaves or lives is a different matter, and 40 people over 25 weeks in an open-ended early-phase study is a sample built to detect safety signals and biomarker movement, not to prove benefit. Other groups are pursuing the same target by other routes, which is a reasonable hedge in a disease where no approved therapy slows progression.
What the delay actually buys is time on drug and time in follow-up, which in a slowly declining condition is not nothing. It also means another year of waiting for people who have watched a parent go through this and know their own odds. Enrolment closing is the point at which a trial stops being a recruitment poster and starts being a dataset. The dataset is now being assembled.