Autism spectrum disorder is characterized by impaired social communication, repetitive behavior, and restricted interests (Sharma et al., 2018). The prevalence of autism is continuously increasing from 62/10,000 in 2012 to 65/10,000 in 2022 (Eom et al., 2023). More than 60% of autism patients have intact cognition and no language impairment, while adaptive functions are deficient (Saulnier and Klaiman, 2022). Because earlier interventions of adaptive behavior in autism patients are strongly correlated with the improvement of autism outcomes in adults (Saulnier and Klaiman, 2022), a novel therapy for improving those adaptive functions is needed.

The Vineland Adaptive Behavior Scale-II (VABS-II) is the adaptive behavior test to assess the individual’s social deficits and developmental retardation (Sparrow et al., 2011). It is widely used to evaluate the developmental progress and adaptive behaviors in individuals, including those with autism spectrum disorder, and assess the efficacy of the treatment (Sparrow et al., 2011; Klin et al., 2007). The standard version of Korea is the K-VINELAND-II and Korean-VABS-II Adaptive Behavior Composite (K-VABS-II-ABC) score. The K-VABS-II-ABC score is the calculated scores (standard scores) from four domains of communication, daily living skills, socialization, and motor skills, where motor skills are additionally evaluated in pediatrics less than the age of seven (Hwang et al., 2015; Oh et al., 2021). Adaptive behavior can be categorized into three classes with VABS-II score: severe impairment, <71; moderate impairment, 71–84; and mild to no impairment, >84 (Weitlauf et al., 2014). Previous research had reported that the minimal clinically significant difference in VABS-II score is 2–3.75 points (Chatham et al., 2018).

The pharmacological treatment of autism includes different classes of atypical antipsychotic drugs, antidepressants, and modulators of the sympathetic and parasympathetic (Aman et al., 2008). However, those medications show highly variable treatment responses between patients and only improve disruptive repetitive behaviors (Sharma et al., 2018). Given concerns of adverse events for pharmacological intervention in the long-term, concomitant non-pharmacological treatment involving music, cognitive-behavioral, and social-behavioral therapy are used to improve social interaction in autism patients (Sharma et al., 2018; Meza et al., 2022). One of the emerging treatment candidates is L-serine which has neuroprotective effects and a prominent safety profile (Ye et al., 2021; Garofalo et al., 2011; Gantner et al., 2019).

In a previous mouse model study, L-serine improved sociability by modulating SK channels to restore dopamine neuron firing in the ventral tegmental area impaired in autism spectrum disorder (Um et al., 2023). Also, L-serine has demonstrated clinical benefits in several neurological and metabolic disorders, such as Alzheimer’s disease, hereditary sensory neuropathy, and GRIN-related disorders, at doses of up to 850 mg/kg/day (Levine et al., 2017; Fridman et al., 2019; Krey et al., 2022; Yulug et al., 2023). These studies reported improvements in symptoms or slowed disease progression, with no significant safety concerns (Krey et al., 2022; Yulug et al., 2023; Julia-Palacios et al., 2024).

AST-001 is an L-serine syrup formulation developed to treat autism spectrum disorders. In a previous study, we evaluated the pharmacokinetic (PK) properties of AST-001 after single and multiple administration in healthy subjects (Lee et al., 2022). Orally administered AST-001 rapidly absorbed with a peak concentration reached in 0.75–2 h and showing a half-life of 7–14 h. Dose-linearity was observed in the range of 10–30 g, and endogenous L-serine production rate was estimated as 287 mg/h in population PK analysis (Lee et al., 2022). Based on the modeling and simulation, we suggested five weight ranges with a fixed-dose regimen that could achieve the target exposure to pediatrics. In a phase II clinical trial, 12 weeks of treatment of AST-001 twice daily improved functional adaptation in pediatric patients with autism (Kim et al., 2024; Kim et al., 2023). The primary efficacy endpoint, 12 weeks K-VABS-II-ABC score change from baseline in the ANOVA model adjusted for age and baseline score was significantly increased in the AST-001 high dose treatment group compared to the placebo group, showing the difference in score change between the two groups (LS mean ± SE) was 1.43 ± 0.69 points, which was statistically significant (P = 0.042) (Kim et al., 2023). As of December 2023, a phase III clinical trial is currently ongoing (KCT0008915) to confirm the efficacy of AST-001.

This study aimed to develop a population PK-PD model to describe changes in K-VABS-II-ABC scores with natural autism progression and their drug effects. As PK samples were not obtained from pediatric patients, the PD model was linked to the adult PK model using an allometric scale. The established PK-PD model was used to determine the weight-tiered efficacious doses of AST-001 in pediatric patients with autism for a phase III trial.

Data from a multi-center, randomized, double-blind, placebo-controlled Phase II trial were used in this analysis (Kim et al., 2024; Kim et al., 2023). This study consisted of two parts: primary and extended treatment periods. The patients were randomly allocated to the placebo, low-dose, or high-dose groups and the respective treatments were administered for 12 weeks during the primary study period. Based on the previous population PK analysis, 2, 4, 7, 10, and 14 g twice daily (BID) fixed-dose regimens for each weight range of 10–14 kg, 15–24, 25–37, 38–51, and 52–60 kg were administered in the high-dose group, which showed similar exposure to 400 mg/kg/day (Lee et al., 2022). Half of those weight-tiered doses were administered for the low-dose group, which was equivalent to 200 mg/kg/day exposure. In the high-dose group, there was a two-week titration period with a low dose followed by treatment with a high dose for the remaining 10 weeks. During the extension treatment period, the low- and high-dose groups continuously received each treatment for an additional 12 weeks, whereas the placebo group received a high dose of AST-001 including the titration period. For follow-up assessment, patients visited the center at 36 weeks, which is 12 weeks after the last dose of AST-001. The K-VABS-II-ABC score was obtained at baseline, week 12, week 24, and week 36 to evaluate the PD response.

The trial was conducted following Korean Good Clinical Practice and the Declaration of Helsinki and was approved by the Ministry of Food and Drug Safety. The study information was registered in the Korean Open Clinical Trial Registry (KCT0007519), and all patients or their legally acceptable representatives provided written informed consent before study enrollment.

Due to the absence of PK samples from pediatric patients, the previously published population PK model of AST-001 was used to link the PD model. Briefly summarizing the PK model, a two-compartment model, zero-order absorption, and linear elimination combined with zero-order L-serine production were used (Lee et al., 2022). As we changed the modeling software, we re-estimated the population PK model and confirmed that the parameter estimates and precisions were similar to the previous report. We used the final PK model that excluded endogenous production because endogenous L-serine levels interfered with parameter estimation related to drug effects by externally administered AST-001. By applying the empirical allometry coefficient of 0.75 and 1 in clearance and volume of distribution, respectively, we assumed the PK profile of pediatrics. All PK parameters were fixed using final model estimates.

The final PK-PD models of the AST-001 and K-VABS-II-ABC scores are presented in Figure 1. The PD model is best described as an effect-compartment model with linear drug effects. The goodness-of-fit plot showed an adequate model structure without any apparent misspecifications (Supplementary Figure S1). The prediction-corrected visual predictive check showed that the model generally described the observations well (Supplementary Figure S2). The precision of the parameters was generally acceptable, with relative standard errors (RSEs) of less than 25%, except for the linear slope of the drug effect (Deff). The final estimates were similar to the median bootstrap values and included in the 95% confidence intervals of the bootstrap results, implying that the model was robust and stable (Table 1).

Figure 1. Structure of the population pharmacokinetic and pharmacodynamic model of AST-001 and K-VABS-II-ABC score. D1, duration of zero-order absorption; CL/F, apparent clearance; V1/F, apparent central compartment volume; Q/F, apparent inter-compartmental clearance; V2/F, apparent peripheral compartment volume; ke0, effect-site first order elimination rate constant; E0, baseline VABS score; Deff, linear slope of drug effect regarding effect site concentration; Ce, effect site concentration, Kprog, slope of natural VABS score progression; K-VABS-II-ABC, Korean-Vineland Adaptive Behavior Scale-II Adaptive Behavior Composite.

Table 1. Final parameter estimates of population pharmacodynamic model of K-VABS-II-ABC score.

The slope of the natural K-VABS-II-ABC score progression (Kprog) was estimated to be 0.015 day−1 with a standard deviation of random effects of 0.018 in a normal eta distribution (Table 1). The Deff was estimated as 0.0022 L/μg and the equilibration half-life to the effect compartment was approximately 15 weeks (Table 1). Age was identified as a significant covariate for the baseline K-VABS-II-ABC score (E0) and was included in the model using a power function. There was a positive correlation between E0 and Kprog, suggesting that patients with poorer adaptive behavior at baseline (lower E0) tend to exhibit less natural improvement in their behavior over time.

Considering the maximum administration dose of 560 mg/kg/day in the phase 2 trial, we established weight-tiered doses of 10–13 kg (2 g BID), 14–20 kg (4 g BID), 21–34 kg (6 g BID), 35–49 kg (10 g BID), and >50 kg (14 g BID) for the simulation. These regimens correspond to 308–400 mg/kg/day, 400–571 mg/kg/day, 353–571 mg/kg, 408–571 mg/kg, and less than 560 mg/kg. Our simulation results showed that 12 weeks of twice-daily administration of AST-001 at 2, 4, 6, 10, and 14 g with each weight band improved K-VABS-II-ABC scores from 1.7 to 3.7 compared to the baseline (Table 2). In contrast, the placebo group showed 0.9 to 1.2 increased K-VABS-II-ABC scores overall. The virtual pediatric population had improved K-VABS-II-ABC scores in the weight-band fixed-dose treatments compared to the placebo group (Figure 2). Target attainment was higher in the high-dose weight band group (Figure 2; Table 2). Although the proportion of participants in the 10–13 kg (2 g BID) group showed a lower target attainment of 40.5%, it was twice as high as that in the placebo group (Table 2). In addition, the proportions of target attainment were 40.5%, 51.5%, 54%, 65%, and 73%, similar to the simulation results of the phase 2 trial weight-band dose (Table 2).

Table 2. Simulation results of changes in K-VABS-II-ABC score at 12 weeks from baseline according to the weight-tiered dose in placebo group and treatment group with different dose scenario.

Figure 2. Histogram plots for changes in K-VABS-II-ABC score at 12 weeks from baseline in 200 virtual pediatric patients without AST-001 treatment and after (A) 2 g (10–13 kg), (B) 4 g (14–20 kg), (C) 6 g (21–34 kg), (D) 10 g (35–49 kg), and (E) 14 g (more than 50 kg) twice daily dose treatment. The dashed line represents the median changes in K-VABS-II-ABC score at 12 weeks in each treatment. K-VABS-II-ABC, Korean-Vineland Adaptive Behavior Scale-II Adaptive Behavior Composite.

The suggested final weight-tiered fixed dose was compared with 200 mg/kg twice daily for each weight range in the pediatric population. Overall, the two regimens showed similar improvements in K-VABS-II-ABC scores (Figure 3). Although the 200 mg/kg twice daily regimen in the 21–34 kg and 35–49 kg weight groups led to greater median score improvements of 20% and 31% respectively, the increases were not statistically significant.

Figure 3. Box plots for changes in K-VABS-II-ABC score at 12 weeks from baseline (ΔK-VABS-II-ABC) in 200 virtual pediatric patients after AST-001 treatment with 200 mg/kg twice daily (BID) regimen and (A) 2 g BID (10–13 kg), (B) 4 g BID (14–20 kg), (C) 6 g BID (21–34 kg), (D) 10 g BID (35–49 kg), and (E) 14 g BID (more than 50 kg) in each weight ranges. The dotted line represents the ΔK-VABS-II-ABC score of 2 and 4. K-VABS-II-ABC, Korean-Vineland Adaptive Behavior Scale-II Adaptive Behavior Composite.

To explore the effects of covariates on apparent clearance (CL/F) and baseline K-VABS-II-ABC scores in virtual pediatric patients, we analyzed forest plots across different weight and age range categories. The CL/F of AST-001 was 34% lower in the lower-weight group (10–13 kg) than in the group weighing >50 kg (Figure 4). As age and weight were positively correlated, the 2–5 years age group exhibited a 41% decrease in CL/F compared to the 10–12 years age group. Pediatric patients weighing more than 21 kg showed relatively comparable E0 values, whereas those weighing less than 20 kg displayed a 35%–53% increase in E0. Similarly, the 6–12 years age group showed comparable E0 values, whereas the 2–5 years age group demonstrated a 45% increase in E0 (Figure 4).

Figure 4. Forest plot showing age and weight effect on model-predicted apparent clearance (CL) of AST-001 and baseline K-VABS-II-ABC score (E0). The blue symbols represent the median model predicted CL and E0 ratio, and the whiskers represent the 95% confidence interval. K-VABS-II-ABC, Korean-Vineland Adaptive Behavior Scale-II Adaptive Behavior Composite.

In the Phase II clinical trial, the least-squares mean change from baseline in K-VABS-II-ABC scores at 12 weeks was 2.82 ± 0.50 for the low-dose group and 1.73 ± 0.51 for the placebo group (Kim et al., 2024). In comparison, the high-dose and placebo groups showed corresponding changes of 3.08 ± 0.49 and 1.66 ± 0.48, respectively (Kim et al., 2024). The mean changes in K-VABS-II-ABC scores at 12 weeks from baseline were significantly higher in the high-dose group than in the placebo group, showing the difference in score change between the two groups was 1.43 ± 0.69 points (p = 0.042) (Kim et al., 2024; Kim et al., 2023). These results were also well predicted by our model, in which the high-dose group showed an increase in K-VABS-II-ABC scores ranging from 1.67 to 3.73. Similarly, significant improvements were observed in the Communication and Motor Skills domains of the K-VABS-II in the high-dose group, while Clinical Global Impression-Severity scores were significantly reduced in both dose groups (Kim et al., 2024).

In our model analysis, age was identified as a significant covariate affecting baseline K-VABS-II-ABC scores. Consistent with a previous report (Klin et al., 2007), a negative correlation between age and baseline K-VABS-II-ABC score was also observed in our study (Pearson correlation coefficient = −0.2793, p = 0.0007). As shown in the simulation results (Figure 4), the 6–12 years age group showed comparable E0 values, whereas the 2–5 years age group exhibited a 45% increase in the baseline score. Given that the K-VABS-II-ABC score includes motor skills assessments (Hwang et al., 2015), particularly in children younger than 7 years, it is plausible that higher scores are observed in the younger age group. Therefore, we established the efficacy end point as the change in scores at 12 weeks from the baseline.

Our study is the first to report a population PK-PD model linking L-serine treatment with K-VABS-II-ABC scores in pediatric patients with autism spectrum disorder. Patients with higher baseline values exhibited a greater response to L-serine treatment and a better prognosis for adaptive behavior. In our virtual fixed-dose regimen simulation, the group with E0 less than 45 demonstrated a median Deff of 1.45 (mL/μg) and a kprog of 0.06 (week−1). Conversely, groups with higher baselines exhibited more pronounced drug effects and score progression, with a median Deff of 1.65 (mL/μg) and kprog of 0.14 (week−1). This finding is consistent with Phase II results, where pediatric patients aged 2–6 years with higher baseline scores demonstrated significant improvements following AST-001 treatment (Kim et al., 2024).

Currently, we are developing an AST-001 syrup formulation containing 2 g of L-serine in each 20 mL pouch. Considering convenience for pediatric patients, we developed a syrup formulation. The phase III dose was selected based on the doses provided in these pouches to give accurate doses in individuals. Several weights and dose ranges were investigated using the final model. The optimal weight-band doses determined to achieve sufficient improvement in adaptive behavior were 10–13 kg (2 g BID), 14–20 kg (4 g BID), 21–34 kg (6 g BID), 35–49 kg (10 g BID), and >50 kg (14 g BID). These regimens improved the K-VABS-II-ABC score by more than 2 points overall compared to the placebo group, and their effects were similar to those of the 200 mg/kg BID dose regimen. By applying a weight-tiered fixed-dose regimen, we could offer a convenient dosing option for pediatric patients, along with effective treatment results.

A limitation of our study is that we hypothesized the PK in pediatric patients using empirical allometry scaling. Although AST-001 has a linear PK profile with a relatively short half-life in healthy subjects, patients with autism may have different PK characteristics. In addition, the adjustments to the Phase III weight-tiered dosing regimen were based on formulation constraints and dosing convenience considerations, which were not fully addressed during the transition from Phase I to Phase II study. Further phase III study data, including L-serine trough concentration and efficacy data, can provide a more elaborate exposure-response relationship.

The datasets presented in this article are not readily available because of the confidentiality of clinical trial data. Requests to access the datasets should be directed to Su-Kyeong Hwang, c2tod2FuZ0Bhc3Ryb2dlbi5jby5rcg==.

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphar.2024.1452526/full#supplementary-material

Aman, M. G., Farmer, C. A., Hollway, J., and Arnold, L. E. (2008). Treatment of inattention, overactivity, and impulsiveness in autism spectrum disorders. Child. Adolesc. Psychiatr. Clin. N. Am. 17 (4), 713–738. doi:10.1016/j.chc.2008.06.009

PubMed Abstract | CrossRef Full Text | Google Scholar

Chatham, C. H., Taylor, K. I., Charman, T., Liogier D'ardhuy, X., Eule, E., Fedele, A., et al. (2018). Adaptive behavior in autism: minimal clinically important differences on the Vineland-II. Autism Res. 11 (2), 270–283. doi:10.1002/aur.1874

PubMed Abstract | CrossRef Full Text | Google Scholar

Eom, H., Kim, M., Yoo, J. K., Bae, E., Yi, J., and Lee, J. Y. (2023). Prevalence of pediatric autism spectrum disorder in Korea using national statistics including the COVID-19 pandemic, 2011 to 2021. Ann. Child Neurology 31 (1), 32–28. doi:10.26815/acn.2022.00360

CrossRef Full Text | Google Scholar

Fridman, V., Suriyanarayanan, S., Novak, P., David, W., Macklin, E. A., McKenna-Yasek, D., et al. (2019). Randomized trial of l-serine in patients with hereditary sensory and autonomic neuropathy type 1. Neurology 92 (4), e359–e370. doi:10.1212/WNL.0000000000006811

PubMed Abstract | CrossRef Full Text | Google Scholar

Gantner, M. L., Eade, K., Wallace, M., Handzlik, M. K., Fallon, R., Trombley, J., et al. (2019). Serine and lipid metabolism in macular disease and peripheral neuropathy. N. Engl. J. Med. 381 (15), 1422–1433. doi:10.1056/NEJMoa1815111

PubMed Abstract | CrossRef Full Text | Google Scholar

Garofalo, K., Penno, A., Schmidt, B. P., Lee, H. J., Frosch, M. P., von Eckardstein, A., et al. (2011). Oral L-serine supplementation reduces production of neurotoxic deoxysphingolipids in mice and humans with hereditary sensory autonomic neuropathy type 1. J. Clin. Invest. 121 (12), 4735–4745. doi:10.1172/JCI57549

PubMed Abstract | CrossRef Full Text | Google Scholar

Hwang, S., Cho, S., Bae, S., Kim, J., and Hong, S. (2015). Standardization study of the Korean Vineland adaptive behavior Scales-Ⅱ(K-Vineland-II). Korean J. Clin. Psychol. 34 (4), 851–876. doi:10.15842/kjcp.2015.34.4.002

CrossRef Full Text | Google Scholar

Julia-Palacios, N., Olivella, M., Sigatullina, B. M., Ibáñez-Micó, S., Muñoz-Cabello, B., Alonso-Luengo, O., et al. (2024). L-serine treatment in patients with GRIN-related encephalopathy: a phase 2A, non-randomized study. Brain 147 (5), 1653–1666. doi:10.1093/brain/awae041

PubMed Abstract | CrossRef Full Text | Google Scholar

Kim, H.-W., Kim, J.-H., Chung, U.-S., Kim, J., Shim, S. H., Park, T. W., et al. (2023). AST-001 vs placebo for social communication in children with ASD: an RCT. J. Am. Acad. Child and Adolesc. Psychiatry 62 (10), S320. doi:10.1016/j.jaac.2023.09.514

CrossRef Full Text | Google Scholar

Kim, H. W., Kim, J. H., Chung, U. S., Kim, J. I., Shim, S. H., Park, T. W., et al. (2024). AST-001 versus placebo for social communication in children with autism spectrum disorder: a randomized clinical trial. Psychiatry Clin. Neurosci. doi:10.1111/pcn.13757

CrossRef Full Text | Google Scholar

Klin, A., Saulnier, C. A., Sparrow, S. S., Cicchetti, D. V., Volkmar, F. R., and Lord, C. (2007). Social and communication abilities and disabilities in higher functioning individuals with autism spectrum disorders: the Vineland and the ADOS. J. Autism Dev. Disord. 37 (4), 748–759. doi:10.1007/s10803-006-0229-4

PubMed Abstract | CrossRef Full Text | Google Scholar

Krey, I., von Spiczak, S., Johannesen, K. M., Hikel, C., Kurlemann, G., Muhle, H., et al. (2022). L-serine treatment is associated with improvements in behavior, EEG, and seizure frequency in individuals with GRIN-related disorders due to null variants. Neurotherapeutics 19 (1), 334–341. doi:10.1007/s13311-021-01173-9

CrossRef Full Text | Google Scholar

Lee, S., Hwang, S. K., Nam, H. S., Cho, J. S., and Chung, J. Y. (2022). Population pharmacokinetic model of AST-001, L-isomer of serine, combining endogenous production and exogenous administration in healthy subjects. Front. Pharmacol. 13, 891227. doi:10.3389/fphar.2022.891227

PubMed Abstract | CrossRef Full Text | Google Scholar

Levine, T. D., Miller, R. G., Bradley, W. G., Moore, D. H., Saperstein, D. S., Flynn, L. E., et al. (2017). Phase I clinical trial of safety of L-serine for ALS patients. Amyotroph. Lateral Scler. Front. Degener. 18 (1-2), 107–111. doi:10.1080/21678421.2016.1221971

PubMed Abstract | CrossRef Full Text | Google Scholar

Meza, N., Rojas, V., Escobar Liquitay, C. M., Pérez, I., Aguilera Johnson, F., Amarales Osorio, C., et al. (2022). Non-pharmacological interventions for autism spectrum disorder in children: an overview of systematic reviews. BMJ Evid. Based Med. 28, 273–282. doi:10.1136/bmjebm-2021-111811

PubMed Abstract | CrossRef Full Text | Google Scholar

Oh, M., Song, D. Y., Bong, G., Yoon, N. H., Kim, S. Y., Kim, J. H., et al. (2021). Validating the autism diagnostic interview-revised in the Korean population. Psychiatry Investig. 18 (3), 196–204. doi:10.30773/pi.2020.0337

PubMed Abstract | CrossRef Full Text | Google Scholar

Saulnier, C. A., and Klaiman, C. (2022). Assessment of adaptive behavior in autism spectrum disorder. Psychol. Sch. 59 (7), 1419–1429. doi:10.1002/pits.22690

CrossRef Full Text | Google Scholar

Sharma, S. R., Gonda, X., and Tarazi, F. I. (2018). Autism spectrum disorder: classification, diagnosis and therapy. Pharmacol. Ther. 190, 91–104. doi:10.1016/j.pharmthera.2018.05.007

PubMed Abstract | CrossRef Full Text | Google Scholar

Sparrow, S. S. (2011). “Vineland adaptive behavior scales,” in Encyclopedia of clinical neuropsychology. Editors J. S. Kreutzer, J. DeLuca, and B. Caplan (New York: Springer), 2618–2621.

Google Scholar

Um, K. B., Kwak, S., Cheon, S. H., Kim, J., and Hwang, S. K. (2023). AST-001 improves social deficits and restores dopamine neuron activity in a mouse model of autism. Biomedicines 11 (12), 3283. doi:10.3390/biomedicines11123283

PubMed Abstract | CrossRef Full Text | Google Scholar

Weitlauf, A. S., Gotham, K. O., Vehorn, A. C., and Warren, Z. E. (2014). Brief report: DSM-5 “levels of support:” a comment on discrepant conceptualizations of severity in ASD. J. Autism Dev. Disord. 44 (2), 471–476. doi:10.1007/s10803-013-1882-z

PubMed Abstract | CrossRef Full Text | Google Scholar

Ye, L., Sun, Y., Jiang, Z., and Wang, G. (2021). L-serine, an endogenous amino acid, is a potential neuroprotective agent for neurological disease and injury. Front. Mol. Neurosci. 14, 726665. doi:10.3389/fnmol.2021.726665

PubMed Abstract | CrossRef Full Text | Google Scholar

Yulug, B., Altay, O., Li, X., Hanoglu, L., Cankaya, S., Lam, S., et al. (2023). Combined metabolic activators improve cognitive functions in Alzheimer's disease patients: a randomised, double-blinded, placebo-controlled phase-II trial. Transl. Neurodegener. 12 (1), 4. doi:10.1186/s40035-023-00336-2

PubMed Abstract | CrossRef Full Text | Google Scholar