Background Niemann-Pick disease, type C1 (NPC1) is a lysosomal storage disorder characterised by progressive neurodegeneration. In preclinical testing, 2-hydroxypropyl-β-cyclodextrins (HPβCD) significantly delayed cerebellar Purkinje cell loss, slowed progression of neurological manifestations, and increased lifespan in mouse and cat models of NPC1. The aim of this study was to assess the safety and efficacy of lumbar intrathecal HPβCD. Methods In this open-label, dose-escalation phase 1–2a study, we gave monthly intrathecal HPβCD to participants with NPC1 with neurological manifestation at the National Institutes of Health (NIH), Bethesda, MD, USA. To explore the potential effect of 2-week dosing, three additional participants were enrolled in a parallel study at Rush University Medical Center (RUMC), Chicago, IL, USA. Participants from the NIH were non-randomly, sequentially assigned in cohorts of three to receive monthly initial intrathecal HPβCD at doses of 50, 200, 300, or 400 mg per month. A fifth cohort of two participants received initial doses of 900 mg. Participants from RUMC initially received 200 or 400 mg every 2 weeks. The dose was escalated based on tolerance or safety data from higher dose cohorts. Serum and CSF 24(S)-hydroxycholesterol (24[S]-HC), which serves as a biomarker of target engagement, and CSF protein biomarkers were evaluated. NPC Neurological Severity Scores (NNSS) were used to compare disease progression in HPβCD-treated participants relative to a historical comparison cohort of 21 NPC1 participants of similar age range. Findings Between Sept 21, 2013, and Jan 19, 2015, 32 participants with NPC1 were assessed for eligibility at the National Institutes of Health. 18 patients were excluded due to inclusion criteria not met (six patients), declined to participate (three patients), pursued independent expanded access and obtained the drug outside of the study (three patients), enrolled in the RUMC cohort (one patient), or too late for the trial enrolment (five patients). 14 patients were enrolled and sequentially assigned to receive intrathecal HPβCD at a starting dose of 50 mg per month (three patients), 200 mg per month (three patients), 300 mg per month (three patients), 400 mg per month (three patients), or 900 mg per month (two patients). During the first year, two patients had treatment interrupted for one dose, based on grade 1 ototoxicity. All 14 patients were assessed at 12 months. Between 12 and 18 months, one participant had treatment interrupted at 17 months due to hepatocellular carcinoma, one patient had dose interruption for 2 doses based on caregiver hardship and one patient had treatment interrupted for 1 dose for mastoiditis. 11 patients were assessed at 18 months. Between Dec 11, 2013, and June 25, 2014, three participants were assessed for eligibility and enrolled at RUMC, and were assigned to receive intrathecal HPβCD at a starting dose of 200 mg every 2 weeks (two patients), or 400 mg every two weeks (one patient). There were no dropouts in this group and all 3 patients were assessed at 18 months. Biomarker studies were consistent with improved neuronal cholesterol homoeostasis and decreased neuronal pathology. Post-drug plasma 24(S)-HC area under the curve (AUC8-72) values, an indicator of neuronal cholesterol homoeostasis, were significantly higher than post-saline plasma 24(S)-HC AUC8-72 after doses of 900 mg (p=0·0063) and 1200 mg (p=0·0037). CSF 24(S)-HC concentrations in three participants given either 600 or 900 mg of HPβCD were increased about two fold (p=0·0032) after drug administration. No drug-related serious adverse events were observed. Mid-frequency to high-frequency hearing loss, an expected adverse event, was documented in all participants. When managed with hearing aids, this did not have an appreciable effect on daily communication. The NNSS for the 14 participants treated monthly increased at a rate of 1·22, SEM 0·34 points per year compared with 2·92, SEM 0·27 points per year (p=0·0002) for the 21 patient comparison group. Decreased progression was observed for NNSS domains of ambulation (p=0·0622), cognition (p=0·0040) and speech (p=0·0423). Interpretation Patients with NPC1 treated with intrathecal HPβCD had slowed disease progression with an acceptable safety profile. These data support the initiation of a multinational, randomised, controlled trial of intrathecal HPβCD. Funding National Institutes of Health, Dana's Angels Research Trust, Ara Parseghian Medical Research Foundation, Hope for Haley, Samantha's Search for the Cure Foundation, National Niemann-Pick Disease Foundation, Support of Accelerated Research for NPC Disease, Vtesse, Janssen Research and Development, a Johnson & Johnson company, and Johnson & Johnson.