Combinatorial approaches for treating neuropsychiatric social impairment

1. Understanding neural mechanisms of social behaviour

Social function is crucial to survival, and is evolutionarily conserved in many animals, from nematodes to humans [1–7]. Animals with simpler nervous systems engage in social behaviours such as courtship, mating, parenting and aggression, and animals with greater neural complexity exhibit more complex social behaviours such as alliance formation, cooperative hunting, empathy and altruism [8–10]. In humans, social behaviour relies on multiple processes that can be conceptualized into overlapping domains such as social anxiety, reward and attachment.

The importance of social behaviour is underscored by the significant impact of social isolation and social stress on human health [11,12]. Social isolation leads to pro-inflammatory gene expression [13], increases overall mortality [14–16], is linked to functional decline [17], and increases cardiovascular risk [18]. Aversive social environments have serious negative impacts on health [19,20] and can decrease the efficacy of mental health treatment [21,22]. Many psychiatric conditions such as depression, anxiety and schizophrenia show early social impairment and isolation, which further contribute to functional decline [23,24]. Other conditions such as autism spectrum disorder (ASD) and borderline personality disorder exhibit symptoms that are cardinal in the social domain [25–27].

There has been a concerted effort to understand the mechanisms that enable social cognition—the processing and integration of multisensory inputs to represent social information and control social behaviour. Many lines of research have established that social cognition requires the integration of multiple signalling networks such as those involving dopamine, serotonin, oxytocin and endocannabinoids [3,10,12,28,29]. Here, we focus on describing the social neurobiology of oxytocinergic and endocannabinoid signalling, which provides important principles to design combinatorial treatment approaches. We then expand on these principles to encompass various treatment modalities (such as psychotherapy and brain stimulation) and propose a network-based, mechanistic framework to design combinatorial treatment approaches with improved specificity and efficacy.

2. Oxytocin in the control of social behaviour

An extensive body of research indicates that oxytocin signalling plays a central role in mediating social behaviours across multiple species [2,30–33]. Oxytocin is a neuropeptide that is produced in the hypothalamus and acts both centrally and peripherally, via a combination of local and hormonal transmission, to facilitate critical social functions such as mother–child bonding and nurturing, pair-bonding, social recognition and social reward [30,34–39]. Central oxytocinergic signalling is thought to be composed of oxytocin neurons in the hypothalamus [40–42] and their synaptic and/or diffuse oxytocin-receptor-mediated actions in specific neural circuits spanning many cortical and subcortical regions [36,40,43–49]. Oxytocin is hypothesized to amplify the salience of social stimuli via sensory circuits, temper social anxiety via hypothalamic and limbic circuitry, and enhance social reward via signalling in ventral tegmental area and nucleus accumbens [36,50,51].

More recent cognitive and translational neuroscience studies have suggested that oxytocin impacts higher-order forms of social cognition in humans. In an experimental cooperative setting, intranasal oxytocin increases trust and willingness to accept social risks [52], depending on the kind of social information available [53]. Importantly, exogenous oxytocin does not always effect positive valence, as oxytocin administration in other contexts can increase aggression, envy, gloating, and both in- and out-group bias [54–56]. Further, the effect of oxytocin in the brain is modulated by epigenetic regulation of oxytocin receptors [57] and oxytocin can cause sexually dimorphic neuronal and behavioural effects [58]. Overall, oxytocin signalling shapes complex rodent, non-human primate, and human social behaviours and can amplify social salience and reward in specific social contexts [59,60].

However, as a treatment strategy, exogenously administered oxytocin has yielded mixed results in terms of clinical efficacy. Intranasal oxytocin improved emotional recognition in adults and children with ASD [61–63] and facilitated pro-social behaviour [64] and social responsiveness [65] in children with ASD; however, these results are inconsistent [66,67]. Oxytocin also enhanced social interaction in patients with social anxiety disorder, but this effect was specific to a social setting with explicit priming [68]. Overall, multiple clinical studies have been performed, with varying results [69], and studies of oxytocin have not yielded any U.S. Food and Drug Administration (FDA)-approved treatments for the social domain. In fact, oxytocin may be limited as a treatment modality as it does not cross the blood–brain barrier well, must be administered intranasally, and has short-lived effects [70–72]. However, given the effects seen with oxytocin administration across model systems, there remains enthusiasm that manipulating the oxytocin system holds promise to treat characteristic social symptoms in ASD and other psychiatric conditions [73,74]. Understanding how other signalling systems in the brain interact with oxytocin to modulate social behaviour and accordingly applying new technologies towards circuit-based treatment has emerged as a promising approach [75].

3. Endocannabinoids in the control of social behaviour

Another promising signalling target for impacting social function is the endocannabinoid system. While cannabinoids have long been used in social ritual [76,77], only more recently has the important role of endocannabinoid signalling in the control of specific aspects of social behaviour been mechanistically demonstrated. In animal and human studies, evidence for its roles span social stress reactivity, social reward and social development, and its dysregulation is linked to neuropsychiatric social impairment. The molecular and systems neurobiology of endocannabinoid signalling provides evidence of its importance to social behaviour and provide context for the insights from human studies on cannabinoids. For more comprehensive reviews on cannabinoid-based drug development as well as the neurobiology of endocannabinoid signalling in the control of social behaviour, see Wei et al. [77,78]; here we focus on clinical studies and network principles important to inform therapeutic development in the social domain, particularly towards combinatorial approaches. The endocannabinoid system is a modulatory neurotransmitter system consisting of three main components: (i) two lipid-derived local messengers—2-arachidonoyl-sn-glycerol (2-AG) and anandamide; (ii) enzymes and transporters that mediate their formation and elimination; and (iii) cannabinoid receptors that are activated by endocannabinoids and regulate neuronal activity (mainly, albeit not exclusively, via CB₁ cannabinoid receptors (CB₁) in the brain [79,80].

The endocannabinoid system mediates the effects of the psychoactive molecule in Cannabis sativa, a plant that contains a family of chemically related terpene-like molecules, which include delta-9-tetrahydrocannabinol (Δ-9-THC). Δ-9-THC has a complex set of pharmacological properties stemming from its binding to cannabinoid receptors expressed throughout the human body. The binding of Δ-9-THC to CB₁ receptors in the brain causes a characteristic acute mental state often subjectively described as a combination of enhanced sociability, quickened mental associations, heightened sensitivity to certain sensory stimuli (e.g. sounds or colours), alterations in the perception of time and space, and increased appetite for sweet and fatty foods [76]. It has also been shown to lead to states of anxiety or paranoia [81]. These behavioural effects are broadly consistent with the demonstrated roles of the endocannabinoid system.

The brain distribution of components of the endocannabinoid system is consistent with its role in social behaviour. CB₁ receptors are highly expressed in associational cortical regions of the frontal lobe and subcortical structures such as the amygdala and hippocampus that underpin human socio-emotional functioning [82,83]. However, the action of CB₁ receptors is determined by many factors, including their cellular subtype, activity of nearby interacting proteins such as cannabinoid receptor-interacting protein 1A (CRIP1A), and associations with varying G-coupled protein receptors [84,85]. CB₁ receptors are also expressed throughout regions implicated in reward and addiction, including the central and basolateral amygdala, prefrontal cortex (PFC), hippocampus, dorsolateral striatum, ventral tegmental area, and, to a lesser extent, the nucleus accumbens [82,86]. These regions have also been considered key parts of the brain that mediate aspects of social cognition, based on imaging and network studies [83]. The regional distribution of the aforementioned endocannabinoid-synthesizing and -degradative enzymes is associated with that of CB₁ receptors [86–88]. The specificity of endocannabinoid signalling likely relies on the convergence of receptor distribution, ligand chemical identity (both endogenous and exogenous), the differential expression of regulatory enzymes, as well as downstream second-messenger cascades [89].

As with oxytocin, multiple lines of research have demonstrated diverging effects on social interaction, likely owing to differential recruitment of circuits such as for social anxiety and social reward. Early studies in rodents demonstrated the ability of Δ-9-THC to modulate aggressive behaviour, fight or flight responses, and sexual behaviour, depending on the sex, dose and behavioural context [90–92]. In an early experiment in rhesus monkeys in which one observer was required to respond based on the facial expression of a ‘stimulus’ conspecific, Δ-9-THC enhanced discrimination of facial expressions in observers while impairing the ability to display facial expressions in ‘stimulus’ conspecifics [93]. Later studies employed genetic approaches to evaluate the role of this system in social behaviour. CB₁ receptor knock-out (CB₁-KO) mice showed increased aggression in a familiar environment but decreased sociability in an unfamiliar environment [94]. CB₁-KO mice also showed increased social anxiety and enhanced levels of social memory [95]. CB₁-KO mice also showed increased aggression, but giving a CB₁ agonist ACEA was sufficient to decrease aggression in OF1 mice [96]. A more recent study showed altered ultrasonic vocalization communication and deficits in social interest and social interaction in CB₁-KO mice [97]. Interestingly, over-expressing the CB₁ receptor in the medial prefrontal cortex (mPFC) decreased social interaction with an unfamiliar animal [98]. A recent study on capuchin monkeys found that pharmacological disruption of CB₁ signalling reduced spontaneous social interaction, corroborating a conserved role for endocannabinoid signalling in social behaviour [99]. While studies of cellular specificity are lacking, knockout in ventral hippocampus GABA neurons has been shown to enhance social investigation while knockout in glutamatergic neurons reduced interaction [100,101]. While wholesale manipulation of CB₁ receptors produce dichotomous results on social interaction, more consistent effects at the ligand- and region-specific level support this hypothesis of circuit selectivity. For example, a direct CB₁ receptor agonist reduced social play while an indirect cannabinoid agonist enhanced social play [102].

The role of endocannabinoids has also been elucidated through interruption of the enzymes involved in the synthesis and/or degradation of its signalling molecules. For example, targeted deletion of diacylglycerol lipase α (DGLα), the primary synthetic enzyme of 2-AG, decreased levels of 2-AG and led to deficits in social interaction [103]. Deletion of FAAH (the primary degradative enzyme for anandamide), thereby potentiating anandamide signalling, increased time spent in social interaction and enhanced social reward [104]. This effect was driven by circuit-selective oxytocin signalling and stands in contrast to global reduction of 2-AG signalling, which generalized to impair other natural rewards such as food (and was not driven by oxytocin) [105]. Particularly in contrast to receptor agonism, these studies highlight the flexibility of this system to modulate social behaviour based on ligand-, context- and circuit-specific mechanisms.

Neurodevelopmental studies of the endocannabinoid system also underscore its important role in social development and properly tuned network signalling. Developmental exposure to cannabinoid agents persistently influences the expression of social behaviour later in adulthood, which is particularly salient for addictive and psychotic disorders [106,107]. The expression of CB₁ receptors throughout regions of the social brain network is remarkably dynamic—it reaches a peak during adolescence, and decreases into adulthood [106] (though some rodent studies suggest it can increase to adulthood [108]). Rather than simply modulating the development of other social brain systems, endocannabinoid signalling appears to play a central role in mediating the social transition between adolescence and adulthood [109–111]. Although CB₁ receptor expression does not necessarily determine its function, this change implies an important role of changing receptor expression in development. Indeed, enhanced CB₁ activity in rats prolonged the adolescent behavioural repertoire, including increased impulsivity and social play, into adulthood [112]. While the developmental trajectories of endocannabinoids are less clear, endocannabinoid levels have been shown to be dependent on age as well as region of activity [109]. Lastly, endocannabinoid activity regulates processes such as neuronal genesis, migration and differentiation, as well as in synaptic pruning and glial cell formation [106,113]. Together, these studies suggest that endocannabinoid signalling is crucial for social development and appropriately flexible network signalling.

Human studies also emphasize context dependence and network selectivity. Early studies showed that cannabis significantly decreased feelings of hostility and increased socio-emotional communication, social interaction time, and group problem solving; however, its effect on social cognition was dependent on dose, environment and situation [114–117]. Later studies showed that this effect on reducing threat perception and enhancing socio-emotional regulation is likely mediated through enhancing the connectivity in the mPFC/ACC–amygdala network [118]. Interestingly, functional impairments in this same network have been linked to socio-emotional dysregulation and aggression in borderline personality disorder, a condition characterized by impaired social cognition [119–121]. Borderline personality disorder has also been associated with elevated FAAH binding and decreased endocannabinoid tone in the PFC [122].

Genetic studies show that single nucleotide polymorphisms (SNPs) within the genes for CB₁ receptor and FAAH are associated with altered social gaze, reactivity to threat, and altered neuronal signalling in the PFC and amygdala [123–125]. Likewise, decreased serum levels of endocannabinoids have been found in individuals with depression [126,127] and children with ASD [128] while increased levels have been found in schizophrenia [129]. Indeed, dysregulated endocannabinoid signalling has now been implicated in a variety of psychiatric conditions [125,130–133].

Preclinical studies have examined the therapeutic potential of targeting the endocannabinoid system to ameliorate social dysfunction. Consistently, selective potentiation of either anandamide or 2-AG signalling—through pharmacological inhibition of their respective degradative enzymes (FAAH and MGL)—ameliorated social deficits in various monogenic, polygenic, immunogenic and other animal models related to ASD [134–137]. Cannabidiol (CBD), another psychoactive compound found in cannabis, whose mechanism of action is poorly understood, is hypothesized to enhance endocannabinoid signalling through inhibition of the reuptake and enzymatic degradation of anandamide [138] (though it is thought to also influence activity at a number of receptors, including vanilloid receptors and serotonin receptors 5-hydroxytryptamine [139]) and has been shown to enhance social interaction [140] and reduce isolation-induced aggression [141] in rodents.

Supporting therapeutic potential, clinical studies indicate that enhancement of endocannabinoid signalling may treat social anxiety and post-traumatic stress disorder (PTSD) as well as several facets of ASD, and these approaches are undergoing clinical development [142–144]. Findings thus far, while mixed, are promising, as cannabinoids appear to improve ASD-associated symptoms such as caregiver-rated social responsiveness and decrease polypharmacy burden [143]. Recently, administration of a selective FAAH inhibitor resulted in significant improvements in social anxiety over a placebo in a randomized control trial with 149 patients [145]. However, FAAH inhibitors, CBD and compounds like nabiximols (purified plant extract of 1 : 1 ratio of CBD and D⁹-THC) have controversial results with mixed evidence of efficacy in social anxiety and ASD [139,142,143] and may have different mechanisms of action in brains of patients with ASD [146]. In a related line of work, plasma anandamide concentrations were found to be lower in children with ASD [147].

In sum, multiple lines of evidence suggest that the endocannabinoid system is important in social development and behaviour across rodents, non-human primates and humans. Various molecular and circuit mechanisms of endocannabinoid signalling have been demonstrated in the regulation of social behaviour. A theme in this literature is that endocannabinoid potentiation leads to variable effects, which is likely due to its context-, dose- and circuit-dependence—a better understanding of the mechanisms selective for ameliorating social impairment may be harnessed to design more effective treatment strategies.

4. Designing combination treatments for social impairment

As described, therapeutic engagement of the oxytocin or endocannabinoid system exhibits varying efficacy and specificity based on the social and environmental context. This highlights that multifunctionality and redundancy are continuing challenges of creating targeted therapies for social cognition [148]. One general solution to improve efficacy and enhance specificity of treatment in medicine is well-validated, combination pharmacological approaches [148]. Here, we expand on this principle to encompass multimodal combination therapies beyond dual pharmacology approaches. We propose that pharmacology, brain stimulation and psychosocial treatments can be combined in a network-based, mechanistic framework to enhance treatment specificity and efficacy. Here, we describe biological principles of multisystem signalling, describe current combination treatment approaches used in psychiatry today, and provide examples of modalities that can be used in combination approaches for social impairment and their potential network mechanisms.

Various aspects of social behaviour have been shown to rely on an overlapping network of signalling involving molecules such as oxytocin, dopamine, serotonin, opioids and endocannabinoids acting across multiple brain regions (which are also involved in non-social states such as general anxiety) [3,24,33,36]. This combinatorial biology suggests that effective treatment of social dysfunction may require combination approaches that parallel social brain physiology. Neural networks for motivated behaviours, including brain regions such as the amygdala, hypothalamus, ventral tegmental area and nucleus accumbens, also process social information [149], and advances in connectomics and brain-wide imaging reveal diverse multisystem connectivity for social cognition [150,151]. This overlapping connectivity has been demonstrated in principle to result in synergistic effects in potentially clinically relevant treatments targeting the social domain. For example, in an experiment in which non-human primates were given oxytocin alone or oxytocin in combination with an opioid antagonist, the combination treatment was found to enhance social attention in a supralinear manner when compared with oxytocin alone [152]. This provides one example of combining oxytocin with engagement of other signalling networks in order to enhance specific effects on social cognition [153].

In fact, combination pharmacological approaches are widely used in psychiatry for depression, anxiety, schizophrenia and other conditions in cases in which the use of one agent is ineffective [154–157]. In some cases, treatment resistance to a pharmacological agent with a particular molecular mechanism is thought to arise as a result of compensatory mechanisms that engage a second molecular mechanism [158]. In this case combination approaches can (a) target such alternative mechanisms, (b) potentiate an existing mechanism of action or (c) mitigate potential adverse or dependence effects of the primary pharmacological agent to allow its titration [159–162]. More broadly, treatment resistance in this context can be expanded to also include psychological resistance, which is an expected part of the therapeutic process [163,164]. In this multi-factorial context, a network-based view for convergent combination therapy would provide a grounding framework to account for and overcome such resistance holistically and enhance treatment strategies (figure 1).

• Download figure
• Open in new tab
• Download PowerPoint

5. Non-invasive brain stimulation for treating social impairment

Other tools in psychiatry can also be incorporated in a combinatorial approach such as brain stimulation. Non-invasive brain stimulation (NIBS) is an example of a demonstrably effective non-pharmacological option for conditions such as depression, bipolar disorder, schizophrenia, schizoaffective disorder, catatonia, neuroleptic malignant syndrome, obsessive compulsive disorder and smoking cessation [165,166]. While general non-invasive interventional approaches include transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), magnetic seizure therapy (MST), electroconvulsive therapy (ECT) and photobiomodulation, none of these interventions is FDA-approved for the treatment of symptoms in the social domain [167–169]. However, there is evidence that some of these approaches have the potential to improve social functioning [170,171]. Here, we will focus on the NIBS methods TMS and tDCS as they are pain-free, non-invasive, well-tolerated and have high acceptance from patients [169,172].

TMS is a non-invasive tool that applies changing magnetic fields to induce an electric current that modulates neuronal excitability in a localized area of the brain. TMS has demonstrated its ability to disrupt and enhance function of neural circuitry and alter social behaviour. TMS applied to the superior temporal sulcus (STS) alters social gaze [173] while application to the ventrolateral prefrontal cortex (vlPFC) reduces social pain [174]. TMS application to the temporoparietal junction (TPJ) enhanced perspective-taking ability [175] and disrupted moral judgements [176], while application to the dorsolateral prefrontal cortex (dlPFC) altered reciprocal fairness [177] and moral decision-making [178] in healthy participants. Thus, TMS has a demonstrated ability to modulate complex aspects of social cognition in healthy adults. As a treatment, TMS applied to the dlPFC and STS has been used to improve social and executive function and reduce repetitive behaviours in individuals with ASD [171,179–181]. For instance, TMS applied to the dlPFC successfully reduced impairment in social relating and social anxiety in individuals with ASD [182]. Additionally, TMS administered to the TPJ improved associative learning in ASD patients [183], and a multisite randomized control trial is currently assessing TMS in the TPJ to treat ASD [184]. However, results using TMS as a treatment modality for some domains of ASD are heterogeneous and vary based on stimulus location and stimulation parameters adjusted to modulate inhibitory or excitatory networks [185].

tDCS uses a direct low-intensity current flowing between a pair of scalp electrodes to modulate cortical activity in a localized area of the brain. tDCS has also shown potential as a treatment for social dysfunction [186]. tDCS applied to the vlPFC bidirectionally modulated social pain [187] and reduced aggression after social exclusion. tDCS applied to the dlPFC reduced or enhanced emotional contagion [188,189], while tDCS applied to the inferior frontal cortex (IFC) enhanced imitation during social interaction [190]. tDCS applied to the TPJ has also been shown to improve sociocognitive processing in healthy participants [191] and application to the lPFC bi-directionally modulates compliance to social sanctions [192]. As a treatment, tDCS applied to the TPJ led to lasting improvements in social function in a patient with ASD [193] while tDCS stimulation of the ventromedial prefrontal cortex (vmPFC) significantly improved theory of mind in children with ASD. Lastly, stimulation to the dlPFC significantly decreased social attentional bias in patients with social anxiety disorder [194].

Medications have also been shown to modulate the efficacy of the intervention techniques described above [195] and informed combinatorial approaches could be used to augment pharmacological approaches with stimulation and vice versa. For example, TMS was found to augment and accelerate the onset of action and response to amitriptyline in severe depression [196]. Likewise, there is evidence that tDCS might be most effective when combined with pharmacological agents or cognitive behavioural therapies [197]. To this end, one study suggested that tDCS is most effective for treatment response in depression when combined with selective serotonin reuptake inhibitors [198].

While these techniques hold promise and have shown varying effects that enhance aspects of social behaviour in conditions like ASD, the results are heterogeneous and reflect varying protocols, study design and a lack of longitudinal experimental protocols with a sufficient follow-up period after treatment [171,199]. This underscores the need for more research in this area to inform combined therapeutic approaches. To that end, new stimulation parameters are demonstrating improved efficacy, shortened protocol times and improved circuit precision [200], which may improve these interventions. The current research indicates that the candidate target brain regions for TMS or tDCS that will likely play a role in treatments of social behaviour are the various subregions of the PFC (mPFC, dlPFC, vlPFC), the TPJ and the STS [184,201] (figure 1).

6. Psychological approaches for treating social impairment

While NIBS and pharmacotherapy can address aspects of social cognition and behaviour, one of the most powerful tools in psychiatry for addressing specific aspects of social behaviour is psychotherapy. Psychotherapy itself is inherently a social phenomenon that depends on the dynamic social cognition of both the therapist and patient; group therapy further relies on social interaction, altruism, and attachment processes and related brain networks [202–204]. Combination approaches have shown psychotherapy can enhance the effects of pharmacological approaches and vice versa [205]. In fact, psychotherapy has been shown to specifically improve social function in patients with depression when added to pharmacological approaches [206,207]. In social anxiety disorder, psychotherapy in combination with phenelzine was found to be more effective than phenelzine alone [208]. More recently, the resurgence of psychedelics as therapeutic agents within psychiatry has also led to studies illustrating the improved benefits of psychedelic-assisted psychotherapy for social anxiety in ASD, PTSD, anxiety, and depression when compared with psychotherapy alone [209,210]; similar approaches have also enhanced the results of group psychotherapy [211,212]. The prominence of such enhancement effects would suggest mechanistic specificity beyond coincident or summative effects.

Moreover, the use of the right pharmacological agent could significantly enhance the social dynamics of therapy. For instance, the entactogen 3,4-methylenedioxymethamphetamine (MDMA), in combination with psychotherapy, has been shown to rapidly improve symptoms of PTSD in phase III trials [213]. Translational studies have investigated possible mechanisms by which MDMA might enhance the therapeutic process and have found that MDMA enhances emotional empathy and prosociality [214], attention to social cues [215], and responses to social reward [216], and leads to a sense of social connection and ‘openness’ characteristic of mystical experience [217,218]. In treatment, MDMA-assisted psychotherapy has also shown to lead to rapid and durable improvement in social anxiety symptoms in adults with ASD [219]. MDMA-assisted therapy has illustrated the power of combining a psychotherapeutic context with the acute administration of a pharmacological agent [220].

These ligand-, anatomical- and augmentation-based approaches echo the combinatorial systems neurobiology described for the oxytocin and endocannabinoid systems. This parallel can be harnessed to derive more network-selective and effective treatments. For instance, oxytocin and endocannabinoids could evoke a neuromodulatory physiology that primes for more targeted psychotherapeutic interventions. Experiential therapies focused on overcoming resistance could use the anxiolytic properties of endocannabinoids, and transference-focused therapies could utilize the rewarding properties of oxytocin to build trust. TMS or tDCS can induce plasticity in networks that could then facilitate the impact of psychotherapy. For instance, psychotherapy has been shown to alter neuronal network activity in brain regions relevant to social behaviour such as the mPFC, TPJ, amygdala, hippocampus and anterior cingulate cortex [221–223] and also leads to synchrony in the TPJ between therapist and patient [224]. Some of these changes have been correlated with symptom improvement [225].

In addition to the modalities described above, several other approaches may have potential as well. For example, social cognitive interventions such as social skills training helps teach individuals how to interact with peers by providing explicit instructions and opportunities to develop or improve social interaction, social performance, and interpersonal skills and has been effective in schizophrenia and ASD [226–229]. Administering social cognitive interventions with recovery-oriented individual and group therapy that targets dysfunctional socialization attitudes or behaviours [230–232] could have further benefits than psychotherapy alone. Proof-of-concept pharmacotherapy-aided approaches have been extended to other practices such as mindfulness as well with promising results [233–235]. In addition, aerobic exercise has pro-social cognitive effects [236] and music therapy activates cortical regions and improves social cognition and behaviour in children with ASD and might be usefully combined with social cognitive interventions [237–240].

7. Trans-diagnostic social domains for targeting combinatorial strategies

Using multiple pharmacological agents or combining pharmacological agents with stimulation, psychotherapy, skills-based work, music therapy and mindfulness all hold promise. Based on the available literature and the postulated circuit neurobiology of oxytocin and endocannabinoid signalling as detailed, we hypothesize that these combination approaches in social settings can be distinguished in mechanism, which has important implications for designing treatment approaches. We propose the following trans-diagnostic social domains as a framework to guide combinatorial treatment: modulation of (a) social anxiety, (b) attachment processes that result in potentiating or protective effects, (c) reward processes that reinforce social engagement, (d) recognition of social cues and (e) mentalization of ongoing social dynamics (figure 2). Previous models have posited that deficits in processes underlying social cognition and motivation could be targeted for treatment [241–243]. This framework provides more specificity by relying on available circuit-based evidence and is also centred on the social processes in which oxytocin and endocannabinoid signalling are likely to play an important role. We also recognize that these are broad and interrelated mechanisms—for example, empathy requires tempering of social anxiety, recognition of social cues and mentalization of social dynamics. However, we believe that conceptual distinction of these processes is important to inform combinatorial treatment approaches.

• Download figure
• Open in new tab
• Download PowerPoint

Ultimately, in order to design more targeted treatments for social impairments, the optimal approach may be more holistic and/or multimodal, combining pharmacology, psychotherapy, skills, music and mindfulness with stimulation. Neurostimulation (e.g. tDCS) has shown potential additive value in combination with cognitive remediation [244], and might also be usefully combined with social cognitive treatments [245]. This comprehensive approach would provide top-down cognitive effects through interventions like psychotherapy, tDCS and TMS acting on cortical networks [246], while pharmacological agents provide a bottom-up effect through altering the levels of neuromodulatory signalling molecules that shape network activity (figure 1). For example, individuals with an impaired ability for recognizing social cues and social anxiety could use oxytocin and endocannabinoids to enhance the salience of social information and reduce social stress, respectively. This increase in salience and reduction in social stress, in turn, would increase the ability of subjects to learn higher-level social cognitive skills during psychotherapy or social skills interventions. Evidence for the viability of this approach can already be seen as studies have shown improvements of several symptoms, including social anxiety, when CBD-based medicine is given as an adjunct to pharmacotherapy or psychotherapy [247]. In addition, rTMS delivered to the dlPFC in depressed patients led to changes in peripheral 2-AG which were correlated to symptom improvements, but this was not seen for anandamide [248]. Although these types of early findings require further elaboration, they indirectly support our hypothesis that brain stimulation and pharmacology could be combinatorially designed to converge on circuits (specific to anandamide versus 2-AG, or differential oxytocin/vasopressin receptor expression) in a mechanism that is fundamentally not possible via each modality alone.

8. Oxytocinergic and endocannabinoid system-based combination treatments for social impairment

We have discussed several different approaches that together suggest combination therapy may improve efficacy. Yet, not all combinatorial approaches result in higher efficacy, but rather consist of redundant or burdensome effects. One driving force of a desired efficacy may be specificity. While there has been active thought into the principles of such treatment approaches individually, we suggest applying tools that converge upon functional connectivity within specific neuronal populations and networks. Here, we describe approaches centred on oxytocin and endocannabinoids to inform these types of potential approaches.

It has been postulated that socially driven oxytocinergic signalling interacts with neuromodulatory systems such as the endocannabinoid system to enhance functional plasticity and connectivity [51]; preclinical mechanistic evidence as well as clinical translational evidence support a role for driving combinatorial oxytocin and endocannabinoid signalling to modulate specific aspects of behaviour.

Beyond previously described studies on within-system context specificity, circuit-selective recruitment can also be achieved across neuromodulator systems. In animal models, chemically distinct endocannabinoid signalling has been found to be involved in distinct aspects of oxytocin-dependent and -independent signalling pathways. For instance, disrupting endocannabinoid signalling alters expression of the oxytocin receptor [249] and using a CB₁ receptor antagonist reduces oxytocin-induced analgesic behaviour [250]. Anandamide-mediated endocannabinoid signalling was found to mediate the effects of oxytocin to enhance and consolidate the rewarding properties of social interactions [78], but, by contrast, 2-AG-mediated endocannabinoid signalling was found to be oxytocin-independent and generalized to other non-social natural rewards [105]. In prairie voles, oxytocin's effect on pair bonding was shown to be dependent on endocannabinoid signalling through the CB₁ receptor. It was also shown that blocking CB₁ receptors after pair bonding led to a specific form of social rejection towards a partner but not an unfamiliar individual [251]. That these effects can range widely from the proposed facets of social behaviour (e.g. social reward, attachment) to analgesic behaviour suggests this type of across-system recruitment could be a powerful approach to circuit-selective therapeutic development.

Further, while global cannabinoid receptor activation causes dichotomous effects on social behaviour, on-demand potentiation of endocannabinoid signalling consistently corrects social impairment in ASD-related animal models [77,134,135,137]. Based on these results, oxytocin-driven signalling is likely the result of social stimulation driving a hypothalamic–accumbens circuit that recruits endocannabinoid signalling. We would hypothesize that chemically distinct endocannabinoid mediators would be recruited to control distinct aspects of a given circuit, which may or may not be oxytocin-driven. This hypothesis would support a treatment approach that emphasizes selective and on-demand endocannabinoid mobilization in response to specific social settings or activation of oxytocin signalling.

Clinical translational evidence also supports the possibility that circuit-selective endocannabinoid signalling would be more responsible for specific aspects of processing social information and controlling related behaviour. For instance, a randomized control trial was designed to treat patients with social phobia or panic disorder by combining CBD with exposure therapy, with the specific hypothesis that CBD's role in fear extinction and social behaviour would enhance exposure therapy (a central mechanism for which is promotion of fear extinction) [252]. Lastly, a preliminary study for ASD suggested that a neuroendocrine treatment using melatonin, oxytocin and CBD led to specific social effects through combined targeting of the oxytocin and endocannabinoid system [253].

Taken together, these lines of evidence on oxytocin and endocannabinoid neurobiology, clinical effects and combinatorial treatment approaches suggest that endocannabinoid augmentation of oxytocin-driven or socially driven therapies could be selectively recruited, mobilized on demand, and thus be more likely to act in a state- and context-dependent manner. Harnessing such physiological properties with combinatorial treatment approaches could improve selective targeting of disease entities or trans-diagnostic social domains (figure 2). A design that considers oxytocin or endocannabinoid signalling specific to social settings and brain networks would represent an evolution of current treatment approaches, which typically use mono-modal, context-independent or non-synergistic polypharmacy implementation of treatments.

In conclusion, targeted treatments for social impairment in neuropsychiatric conditions are critically needed. While clinical development of neuromodulators such as oxytocin and endocannabinoids are promising, we have proposed a framework grounded in systems neurobiology (figure 1) and categorization of trans-diagnostic social processes based on known endocannabinoid and oxytocin physiology (figure 2) to guide immediately actionable development of combinatorial treatment approaches. We have described how these approaches may be related to wider multimodal approaches as well as psychosocial considerations (box 1). We believe that these considerations can aid mechanistically driven strategies that target various components of social behaviour to improve the specificity and efficacy of treatment.

Data accessibility

This article has no additional data.

Authors' contributions

D.W.: conceptualization, writing—original draft, writing—review and editing; S.T.: conceptualization, writing—review and editing; J.C.M.: conceptualization, supervision, writing—review and editing; A.Z.A.S.A.A.: conceptualization, data curation, supervision, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interest.

Funding

We received no funding for this study.