Scientific Foundations for the Use of Arbaclofen in Autism Spectrum Disorder

This white paper summarizes the mechanistic, nonclinical, and human pharmacodynamic evidence supporting the scientific rationale for arbaclofen in ASD within FDA's substantial-evidence framework (U.S. Food and Drug Administration, 2026). Clinical efficacy results and statistical analyses are presented separately in the briefing document and are intentionally not repeated here.

The organizing principle is convergence. Bottom-up studies (those that begin with a defined molecular pathophysiology) test a mechanistic prediction in defined genetic disorders; top-down studies (those that begin with ASD-like behavioral phenotypes) test whether GABA-B-sensitive rescue generalizes across biologically distinct ASD-relevant animal models; and human pharmacodynamic studies (those that begin with biomarkers of altered brain physiology) test whether corresponding physiological abnormalities are drug-responsive in autistic people. Together, these approaches ask whether GABA-B-sensitive synaptic and circuit dysfunction represents a shared drug-responsive process in ASD.

ASD is defined behaviorally and arises in individuals with widely differing genetic and environmental backgrounds. Hundreds of genetic variants are associated with ASD risk, and environmental factors are also known to contribute. The resulting heterogeneity makes it unlikely that a single proximal molecular lesion accounts for ASD across the population (Rylaarsdam and Guemez-Gamboa, 2019).

Etiologic heterogeneity, however, is compatible with convergence at downstream levels of biological organization. Distinct perturbations can influence overlapping processes governing synaptic protein synthesis, synaptic plasticity, neuronal excitability, inhibitory control, and the operation of neural circuits supporting social behavior, cognition, sensory processing, and behavioral flexibility. A therapy acting on such a downstream process need not reverse the initiating lesion. It need only modify a downstream physiological state that contributes materially to functional impairment.

The rationale therefore does not require excessive excitation or dysregulated protein synthesis to be universal in ASD. It requires only that distinct ASD-associated causes can produce synaptic or circuit dysfunction that remains responsive to GABA-B receptor modulation.

GABA (gamma-aminobutyric acid) is the major inhibitory neurotransmitter in the brain, and GABA-B receptors are metabotropic G-protein-coupled receptors with both presynaptic and postsynaptic functions. Their activation can reduce neurotransmitter release, including glutamate release at excitatory synapses, and alter neuronal excitability through regulation of ion channels and intracellular signaling. These properties place GABA-B signaling in a position to influence the balance and gain of excitatory and inhibitory transmission and, more broadly, activity-dependent synaptic function (Bettler et al., 2004).

Arbaclofen is the pharmacologically active R-enantiomer of baclofen and is a selective GABA-B receptor agonist (Henderson et al., 2012). The relevance to ASD emerged from mechanistic work in FXS and from the broader hypothesis that abnormalities of synaptic signaling and E/I regulation contribute to neurodevelopmental phenotypes. The important therapeutic concept is modulation rather than nonspecific neuronal suppression: in disease models, arbaclofen has corrected abnormal biochemical and synaptic measures toward wild-type values while also rescuing functional phenotypes (Henderson et al., 2012; Qin et al., 2015).

The recurrent 16p11.2 microdeletion is a well-established genetic cause of neurodevelopmental impairment and is strongly associated with ASD and intellectual disability (Hanson et al., 2015). Because its genetic basis is distinct from FXS, responsiveness to arbaclofen provides a test of whether GABA-B modulation can generalize beyond loss of FMRP.

Stoppel et al. evaluated R-baclofen in two independently generated 16p11.2 deletion mouse lines. Treatment rescued deficits in object recognition memory and fear conditioning in one model and object location memory and social interaction in another (Stoppel et al., 2018). The findings demonstrated beneficial effects across distinct deletion constructs, genetic backgrounds, laboratories, and behavioral assays.

A later multicenter preclinical validation effort extended this work across three independently generated 16p11.2 deletion mouse lines studied in four laboratories. Arbaclofen reproducibly rescued cognitive deficits in two deletion lines in traditional recognition-memory paradigms; an unsupervised machine-learning analysis in one laboratory also identified rescue of genotype-associated exploratory behavior. The multicenter design directly addressed pervasive concern about reproducibility in mouse behavioral studies and strengthens the inference that the effect is not peculiar to a single line, laboratory, or assay (Gundersen et al., 2023). This body of evidence provides an important bridge: the therapeutic hypothesis generated from FXS predicts benefit in a genetically distinct ASD-risk condition, and the prediction is borne out across independent models and sites.

Arbaclofen has also been tested across ASD-relevant models with very different etiologies. These include the inbred BTBR and C58 strains; genetic models involving Fmr1, 16p11.2, Cntnap2, and Pcdh10; NMDA-receptor hypofunction; and prenatal valproate exposure. Across these models, GABA-B receptor activation improves behavioral, synaptic, or circuit phenotypes despite the different initiating perturbations (Silverman et al., 2015; Gandal et al., 2012; Port et al., 2017; Möhrle et al., 2021; Jiang et al., 2022).

The valproate model is particularly informative because it extends the evidence beyond genetically defined models. Chronic treatment with arbaclofen (STX209) improved sociability and social novelty preference, novel-object recognition, locomotor/exploratory behavior, and marble burying, with accompanying changes in hippocampal dendritic morphology, spine density, and GABA-B receptor 2 expression (Jiang et al., 2022).

Taken together, these findings support the hypothesis that distinct ASD-associated etiologies can produce shared downstream pathophysiology that is responsive to GABA-B receptor modulation. This does not imply a uniform E/I abnormality across models or across ASD; E/I regulation varies by circuit, cell type, developmental stage, and method of measurement. The common feature demonstrated across these models is not a particular direction of E/I imbalance, but the responsiveness of downstream synaptic and circuit abnormalities to GABA-B receptor modulation.

The principal strength of the preclinical case is that the two evidentiary paths are not simply repetitions of the same experiment. Bottom-up studies begin with mechanistic knowledge and test the prediction that GABA-B receptor activation should correct defined disease biology. Top-down studies begin with diverse ASD-relevant models and ask empirically whether the same pharmacology produces functional benefit without requiring a common prespecified molecular abnormality.

The evidence therefore does not depend on a single theory of ASD pathophysiology. Across defined genetic disorders and the broader model portfolio, GABA-B receptor modulation improves molecular, synaptic, circuit, and behavioral abnormalities arising from distinct initiating perturbations. Together, these independent approaches support the hypothesis that different upstream causes of ASD can produce shared downstream pathophysiology that is responsive to GABA-B receptor modulation.

The integrated model is therefore one of shared downstream drug sensitivity rather than etiologic uniformity. Diverse ASD-associated perturbations can disturb synaptic and circuit function through different proximal mechanisms while leaving downstream abnormalities susceptible to GABA-B receptor modulation.

This model neither predicts uniform pathophysiology nor requires uniform clinical response. It predicts only that GABA-B modulation can produce functional benefit across more than one ASD etiology when downstream drug-sensitive physiology contributes materially to impairment.

Several limitations must be acknowledged. First, behavioral phenotypes in animal models are not homologous to the full human ASD phenotype, and no individual model captures ASD in its entirety. Second, the concepts of E/I imbalance and altered proteostasis should not be treated as uniform features of ASD; their expression varies across models, circuits, developmental stages, and experimental methods. Third, pharmacological rescue of a phenotype does not by itself establish the precise molecular mechanism responsible for rescue. Fourth, the relevance of acute effects observed in some preclinical paradigms to sustained clinical treatment must be interpreted carefully. Finally, although the breadth of responsiveness across etiologically distinct preclinical models suggests that the potential therapeutic effect of arbaclofen may extend across etiologies and symptom domains, the biological heterogeneity of ASD makes it unlikely that all individuals meeting diagnostic criteria will share the relevant GABA-B-sensitive pathophysiology or respond similarly to treatment.

Within these limitations, the findings are consistent across independent experimental systems. Distinct ASD-associated perturbations produce molecular, synaptic, circuit, or behavioral abnormalities that are responsive to GABA-B receptor modulation, and corresponding GABA-B-sensitive physiological abnormalities are demonstrable and pharmacologically modifiable in people with ASD.

The evidence summarized here is relevant to FDA's framework for demonstrating substantial evidence of effectiveness. FDA's June 2026 revised draft guidance states that, when scientifically sound and clinically appropriate, confirmatory evidence used with a single adequate and well-controlled clinical investigation may include mechanistic evidence derived from pharmacodynamic endpoints in clinical studies or from nonclinical studies (U.S. Food and Drug Administration, 2026).

FDA also sets the demanding standard that disease pathophysiology should be well understood, and the drug's mechanism should be clearly understood and shown to directly target major driver(s) of that pathophysiology. ASD does not fit the guidance's simplest single-gene or single-enzyme paradigm at the disorder-wide level. The relevant question for arbaclofen is therefore whether evidence across distinct ASD-associated etiologies demonstrates that the drug consistently corrects abnormalities relevant to ASD pathophysiology through GABA-B receptor activation (U.S. Food and Drug Administration, 2026).

The arbaclofen evidence addresses that question at multiple levels. In genetic models with a defined molecular pathophysiology, arbaclofen corrects disease-associated molecular, physiological, and behavioral phenotypes (Henderson et al., 2012; Stoppel et al., 2018). GABA-B agonism also produces functional rescue across a broader and biologically heterogeneous portfolio of ASD-relevant animal models (Silverman et al., 2015; Gandal et al., 2012; Port et al., 2017; Möhrle et al., 2021; Jiang et al., 2022). In autistic participants, arbaclofen shifts selected disease-associated neurophysiological measures toward non-autistic values, with different or opposite effects in non-autistic participants (Huang et al., 2022; Huang et al., 2026).

This evidence goes beyond general biological plausibility: it is mutually reinforcing, spans molecular mechanism to human pharmacodynamics, and was generated independently of the clinical efficacy analysis. Whether FDA considers it sufficient confirmatory evidence necessarily depends on the strength of the accompanying adequate and well-controlled clinical evidence and the clinical context.

Accordingly, the mechanistic, nonclinical, and human pharmacodynamic evidence summarized here may provide confirmatory support for the clinical evidence of arbaclofen’s effectiveness in ASD. This evidence does not establish effectiveness on its own. Rather, it strengthens the clinical evidence by showing that arbaclofen acts on disease-relevant biology across multiple ASD-associated etiologies and produces corresponding pharmacodynamic effects in people with ASD.

Taken together, these findings support the specific therapeutic hypothesis that etiologically diverse forms of ASD can share GABA-B-sensitive synaptic and circuit dysfunction that remains pharmacologically modifiable. This hypothesis does not require a single pathophysiology for ASD; rather, it is supported by parallel streams of mechanistic, nonclinical, and human pharmacodynamic evidence.