Safety and tolerability of intranasal insulin in Parkinson's disease.

全文（开放获取） · 英文原文

【Abstract】 BACKGROUND: Repurposing therapeutic treatments for diabetes may be a promising strategy for the treatment of neurodegenerative disorders. Intranasal insulin (INI) is shown to improve memory, attention, and cognition in Alzheimer's dementia. Associations between the etiology of Parkinson's Disease (PD), brain glucose hypometabolism, and central insulin resistance suggest INI as a promising treatment for PD. METHODS: A phase II single center, randomized, double-blind clinical trial was conducted in 30 participants with PD, aged 41-89, MoCA ≥ 16, on a stable dose of anti-PD medications. Participants were randomized to receive intranasal treatment for 21 days with either placebo, or 20, 40, or 80 IU of regular insulin (INI) using SipNose intranasal device. Primary outcomes of safety and tolerability were assessed in 30 participants by monitoring hypoglycemia, weight, and adverse events. Cognitive, mood, and motor measures were collected pre- and post-treatment for secondary outcome analysis and assessed in 28 participants. RESULTS: No significant changes in weight or serum insulin were observed during the study. No hypoglycemic values (<70 mg/dL) were recorded 15 or 30 min post-IN dose. Nasal irritation and nosebleeds were the most common adverse events, with most categorized as "mild." Several secondary outcomes (e.g., verbal fluency and recall) improved from baseline in the 40 IU/INI arm only, though not significantly more than in the placebo arm. No other outcomes assessed (mood, apathy, motor) improved under INI treatment. CONCLUSION: INI exhibits an acceptable safety profile in patients with PD. Further studies to investigate if INI has a positive impact on cognitive and motor symptoms in PD are warranted. CLINICAL TRIAL REGISTRATION: https://clinicaltrials.gov/study/NCT04251585 Identifier, NCT04251585. 【Introduction】 Parkinson’s disease (PD) is the second most prevalent neurodegenerative disorder, following Alzheimer’s disease. In addition to the motor symptoms, PD is recognized as a multisystem disorder, with a wide range of non-motor manifestations, including cognitive impairment, are commonly observed in affected individuals (). Despite their prevalence, cognitive symptoms in PD remain poorly managed, with limited pharmacological options available. Recent research has increasingly focused on the role of metabolic dysfunction in PD, particularly insulin signaling and insulin resistance (). Emerging evidence suggests a significant association between PD and type 2 diabetes mellitus, particularly through shared mechanisms () such as mitochondrial dysfunction, inflammation, and impaired glucose metabolism. Notably, authors in studies (, ) reported an approximately 38% increased risk of developing PD in patients with baseline diabetes compared to non-diabetic patients. Furthermore, type 2 diabetes mellitus has been linked to accelerated progression of PD, exacerbation of motor and cognitive symptoms, and molecular alterations in the striatum and cerebrospinal fluid (). A large cohort study by De Pablo-Fernandez et al. reported a higher incidence of PD among individuals with type 2 diabetes mellitus (), suggesting the possibility of shared genetic predisposition or overlapping pathogenic pathways, although the causal relationships remain incompletely understood. Mechanistically, insulin receptors are found in the basal ganglia and substantia nigra and there is evidence that insulin signaling plays a role in the regulation of neuronal survival, dopaminergic transmission, synaptic plasticity, and energy metabolism (, ). Disruption of this pathway in PD has been linked to altered insulin-like growth factor 1 (IGF-1) levels, insulin receptor substrate (IRS) phosphorylation, and reduced insulin receptor gene expression (, ). These changes contribute to impaired cerebral glucose metabolism (, ), as evidenced by fluorodeoxyglucose positron emission tomography (FDG-PET) studies showing hypometabolism in multiple cortical regions in PD patients including occipital, temporal, parietal, and frontal regions (, ). Given these insights, there is growing interest in repurposing antidiabetic agents for PD treatment. Several FDA-approved treatments for type 2 diabetes mellitus have shown promise in PD clinical trials, particularly for their effects on mitochondrial function (). Among these, glucagon-like peptide-1 (GLP-1) receptor agonists such as exenatide have demonstrated beneficial effects on motor and cognitive outcomes in PD patients (), without evidence to support a disease-modifying effect (). However, systemic side effects () such as weight loss and gastrointestinal symptoms () may limit the use of these agents, highlighting the need for alternative delivery strategies. Intranasal delivery offers a non-invasive route to deliver large molecules such as insulin directly to the brain while minimizing systemic exposure (, ). Recent studies with INI using PET scan shows increased brain uptake in multiple brain regions including hippocampus, caudate and thalamus (, ). There is also evidence for detection of insulin in the CSF after intranasal delivery (, ). These findings provide evidence of central nervous system target engagement and support the investigation of INI as a therapeutic strategy for neurodegenerative disorders, including PD. Clinical investigations of intranasal insulin in healthy adults, Alzheimer’s dementia (AD) have demonstrated therapeutic effects impacting memory, attention, and cognition without significantly altering serum insulin or glucose levels (). Although a longer-duration trial in individuals with mild cognitive impairment and Alzheimer’s disease did not show cognitive or functional benefits (), INI treatment was associated with reduced progression of white matter hyperintensities in deep and frontal brain regions (). To date, only one small randomized, double-blind study has evaluated INI in PD, which reported good tolerability and preliminary improvements in verbal fluency and improvement in motor function measured by UPDRS motor (part III) scores (). Rodent models of PD have also demonstrated that INI can improve motor function and promote dopaminergic cell survival (, ). Moreover, insulin receptors and glucose transporters are co-localized in memory-related brain regions, suggesting that enhanced cerebral glucose utilization may support cognitive function (, ). Previous INI studies have used doses ranging from 10 to 160 IU/INI across various populations. The Novak et al. (2019) study tested only one dose (40 IU) of intranasal insulin (), which did show preliminary efficacy in select domains. It is unclear whether the chosen dose is optimal for clinical benefit, or whether a higher dose would be safe in this population. In the present study, we evaluated the safety and tolerability of three different doses (20 IU, 40 IU, and 80 IU) of intranasal insulin compared to placebo over a 21-day period in individuals with PD. Additionally, we explored the preliminary effects of INI on cognition, mood, apathy, and motor performance. 【Methods】 【Study design】 This study was a single center, randomized, double-blinded, placebo-controlled, phase 2 trial to determine dose finding, safety and tolerability of three doses of intranasal (IN) regular insulin (Novolin R) compared to placebo (saline) in patients with PD (). The study was conducted at HealthPartners Neuroscience Center in St. Paul, Minnesota. The study was approved by local institutional review board (IRB) and registered on www.clinicaltrials.gov (NCT04251585). Study design and visits. 【Study participants】 A total of 97 participants were contacted regarding study participation; of these, 31 were enrolled, 30 initiated treatments, and 29 completed treatments. One participant who signed the consent was deemed ineligible prior to randomization and did not initiate treatment. A total of 28 participants were included in the primary analytic cohort; one participant who completed treatment was removed following the revision of eligibility criteria around Montreal cognitive Assessment (MoCA ()) score but is included in the safety analysis, as is the participant who did not complete treatment (, consort). The eligibility criteria for the MoCA score were revised from ≥10 to ≥16 after completion of the first study participant. This modification was made in consultation with a neuropsychologist for two primary reasons, one, to minimize potential ceiling effects in cognitive testing, and second, to address the difficulty participants with MoCA scores below 16 experiences in completing the study’s cognitive assessments. Study CONSORT diagram. Participants were identified through a combination of targeted recruitment during scheduled clinic visits, internal promotion through organizational webpages, and newsletters distributed to Parkinson’s patients. Written informed consent was obtained from participants and assessed using study eligibility criteria. Participants were scheduled for a baseline visit if they were eligible at the screening visit. Study participants had a diagnosis of idiopathic Parkinson’s Disease, confirmed by a movement disorders specialist, following MDS 2015 clinical diagnostic criteria (Hoehn and Yahr stage of </=3), were between 40 and 90 years of age (inclusive), and had a MoCA score of >/= 16 at screening. All study participants had a prior magnetic resonance imaging (MRI; obtained clinically prior to study participation) and agreed to remain stable on anti-parkinsonian medications for 30 days prior to randomization and throughout the treatment period. Participants with a history of significant head trauma, oto-pharyngeal surgery, chronic sinusitis, or severe deviated symptoms, insulin allergy were excluded from the trial. 【Study intervention】 Insulin (Novolin-R, Novo Nordisk Inc) and placebo (0.9% bacteriostatic sodium chloride, 200 μL, each nostril, twice daily) were delivered using the SipNose intranasal delivery device (SipNose, Ltd., Israel ()) for 21 days. Three different doses of insulin were used, with the highest number of participants (n = 9) in the group with the highest dose (80 IU/INI), a model previously used by Finger et al. (). Dosing for the first two groups of this study is based on previously conducted INI studies in neurodegenerative disorders (, ), using daily doses of 20 and 40 IU of INI. Higher doses have been found to be safe in healthy adults (). Prior studies performed have demonstrated favorable effects of this regimen in the MCI/AD population without peripheral hypoglycemia (). Use of placebo control allows for clearer interpretation of the safety and tolerability of INI, given the studied population may have some baseline level of symptoms and/or deterioration in condition that could otherwise confound the results. Participants in the active treatment arms received one of three different doses of insulin: 20 IU/INI (1O IU [100 μL]), one nostril, twice daily); 40 IU/INI (10 IU [100 μL], each nostril, twice daily); or 80 IU/INI (20 IU ([200 μL], each nostril, twice daily). The principal investigator, coordinator, and study staff who assessed outcomes as well as participants were blinded to the treatment assignment. Two members of the study team were unblinded and were responsible for preparing, dispensing, and collecting the study drug, as well as addressing any questions regarding the device or study drug, to maintain blinding for the rest of the study personnel. The first dose was completed in-person with unblinded staff to ensure correct use of the device without compromising blinding of the remaining study team. Allocation concealment was maintained using a centralized randomization scheme, with treatment assignments accessible only to designated unblinded study staff. To minimize the potential confounding effects of gender and MoCA score on the treatment effect, we used a 2 × 2 stratified randomization scheme with a block size of 4 when assigning patients to one of the four study groups. The stratification factors were gender (male/female) and MOCA score (low <22/30 and high>23/30). However, use of multiple stratification factors can (and did) lead to unequal treatment group sizes, especially with multiple treatment arms and small strata. Six participants were randomized to both placebo and 20 IU/INI arms, 7 to the 40 IU/INI arm, and 11 to the 80 IU/INI arm. 【Outcomes】 Key outcomes included a composite safety event of either a reduction of fasting glucose to <70 mg/dL or an unintended reduction of weight >5% between any of screening, baseline or post treatment visits; any significant pre-post change in fasting glucose or weight; and adverse events (AEs) and serious adverse events (SAEs). Secondary outcomes included pre-post changes in cognition assessed with a neurocognitive battery (see ) (, ); motor symptoms measured with the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale III [MDS-UPDRS; ()]; and mood measured by the Beck Depression Inventory- Second Edition [BDI; ()] and apathy as measured by the Apathy Scale (). 【Statistical analysis】 The repeated measures ANOVA (RM-ANOVA) model was used to test changes in body weight, fasting blood glucose, and insulin levels over time by treatment arm for significance; post-hoc pairwise comparisons were made using a Bonferroni correction. General linear mixed models with random person-specific intercepts were used to test for within-arm pre-post differences in any of the cognitive measures, mood, apathy and motor performance. In addition, a difference-in-differences (DiD) analysis framework using estimates from the Generalized Linear Mixed Model (GLMM) was used to test whether the active treatment arms improved significantly more than the placebo arm; this was accomplished by subtracting the pre-post placebo effect (meant to represent the natural within-person variation in responses associated with time and/or learning effects) from the pre-post treatment effects associated with each active arm. Appropriate to the exploratory nature of the secondary outcomes, we employed the False Discovery Rate (Benjamini-Hochberg) correction to control the proportion of false positives, while retaining high power to uncover potential relationships. All within-arm pre-post () cognitive assessments () were treated as one set of tests (4 × 19 = 76), and all DiD () cognitive assessments () were treated as another set (3 × 19 = 57); the BH correction was applied separately to each set. The procedure was repeated for non-cognitive tests (e.g., mood, apathy, and motor performance). 【Randomization】 Stratified block randomization with a block size of 4 was used to assign participants to one of the 4 groups. The stratification factors were gender (male/female) and MoCA score (low <22/30 vs. high>23/30). The target sample size was 7 for each of the placebo, 20 IU/INI, and 40 IU/INI arms, and 9 for the 80 IU/INI arm. 【Results】 【Study participants】 Twenty-nine participants who met all the revised inclusion criteria (MoCA ≥ 16) were randomized, and 28 completed treatments. One participant did not complete treatment because of persistent nosebleeds; they are included in the safety outcomes analysis, but not the efficacy outcomes analysis. The average age of the 28 participants was 67.3 (SD = 5.5), with the average group age ranging from 63.4 (5.1) to 69.2 (6.1). Most participants were white (100%), non-Hispanic (96.4%) males (57.1%). The average MoCA score of all participants was 25.7 (2.7), and the average score in the placebo and 80 IU/INI arms was consistent with the mild cognitive impairment cutoff of <26; however, despite stratified randomization by MoCA, arms are somewhat imbalanced on MoCA score, complicating the interpretation of any cognitive improvements observed. MDS-UPDRS III and HbA1c were similarly distributed across groups, while average LEDD, a surrogate for PD severity, varied more widely (). Demographics by treatment group and overall. Continuous measures reported as Mean (SD), categorical as % (N). N, Number of participants; MoCA, Montreal Cognitive Assessment, MDS-UPDRS III, Movement Disorders Society–Unified Parkinson’s Disease Rating Scale; LEDD, Levodopa Equivalent Daily Dose in mg; HbA1c, (Hemoglobin A1c). 【Safety outcomes】 Overall, intranasal insulin was relatively safe and well tolerated in PD. 【Change in weight】 No participant experienced unintended weight loss of >5% of weight measured at screening. When analyzing body weight across the 3 timepoints, there were no main effects of treatment arm, time, nor interaction of the two. Post-hoc pairwise testing with Bonferroni adjustment showed that the 20 IU/INI group had a significantly lower body weight across all time points compared to all other study arms (p = 0.01); this was unrelated to the study drug and is instead a result of natural imbalance between the arms (). 【Change in fasting blood glucose level】 Fasting blood glucose level was measured at baseline and post-treatment visits (). There was a statistically significant decline in fasting glucose over time for all groups (i.e., a main effect for time, p = 0.004). The average decline across all groups was 3.5 mg/dL, a clinically insignificant level. There was not a significant difference between groups in terms of glucose level, nor was there a significant interaction between groups and time such that patterns differed between groups. 【Change in point of care glucose (pre-post dose) level】 No blood glucose values of <70 mg/dL were observed at pre-dose, 15-min post-dose and 30 min post-dose (). There was a statistically significant decline in glucose levels over time for all groups (i.e., a main effect for time, p = 0.0006). There was not a significant difference between groups in terms of glucose level, nor was there a significant interaction between groups and time such that patterns differed between groups. 【Change in insulin level】 Insulin levels were measured at visits 2 and 3. Baseline insulin levels were within physiological range, though imbalance was observed (). Given that within-person change in insulin was evaluated rather than improvement against a threshold, this imbalance was not concerning, particularly in the context of a randomized trial. There were no main effects of treatment arm, time, nor interaction of the two, meaning that arms did not differ at either visit 2 or visit 3; nor did insulin level change from visit 2 to visit 3 across all arms (). 【Adverse events】 There were 112 adverse events (AEs), 0 serious adverse events (SAEs), and 10 protocol deviations. Every participant reported at least one adverse event, with an average of 3.7 AEs per person. The average number of adverse events per person was nearly identical in the placebo and 20 IU/INI arms, and highest in the 40 IU/INI arm (5.0 per person). The average number of AEs per person in the 80 IU/INI arm was 3.8. No statistical comparisons are made between groups with respect to adverse event rates. All participants experienced at least one mild adverse event, while a minority experienced at least one moderate adverse event (9–17% across arms). Only 1 participant (placebo arm) experienced a severe adverse event. The percentage of patients experiencing an adverse event that was probably related to the study drug was lowest in the placebo arm (67%) and similar across active treatment arms (82–86%). The 40 and 80 IU/INI arms were likelier to experience adverse events that were probably not related to the study drug (64–71%) than were the placebo or 20 IU/INI arms (33%). Adverse event details are provided in . 【Nasal adverse events】 Virtually all participants experienced at least one nasal adverse event (nasal irritation or nosebleeds, 83–100% by arm). However, nosebleeds an