Questioning the role of palmitoylethanolamide in psychosis: a systematic review of clinical and preclinical evidence

Introduction The endocannabinoid (eCB) system disruption has been suggested to underpin the development of psychosis, fueling the search for novel, better-tolerated antipsychotic agents that target the eCB system. Among these, palmitoylethanolamide (PEA), an N-acylethanolamine (AE) with neuroprotective, anti-inflammatory, and analgesic properties, has drawn attention for its antipsychotic potential. Methods This Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020-compliant systematic review aimed at reappraising all clinical and preclinical studies investigating the biobehavioral role of PEA in psychosis. Results Overall, 13 studies were eligible for data extraction (11 human, 2 animal). Observational studies investigating PEA tone in psychosis patients converged on the evidence for increased PEA plasma (6 human) and central nervous system (CNS; 1 human) levels, as a potential early compensatory response to illness and its severity, that seems to be lost in the longer-term (CNS; 1 human), opening to the possibility of exogenously supplementing it to sustain control of the disorder. Consistently, PEA oral supplementation reduced negative psychotic and manic symptoms among psychosis patients, with no serious adverse events (3 human). No PEA changes emerged in either preclinical psychosis model (2 animal) studied. Discussion Evidence supports PEA signaling as a potential psychosis biomarker, also indicating a therapeutic role of its supplementation in the disorder. Systematic review registration https://doi.org/10.17605/OSF.IO/AFMTK.


. . Objectives
The effect of PEA over neuroinflammation and glutamate signaling may represent a promising biobehavioral mechanism underlying its clinical utility in psychosis. This systematic review aimed to collect and discuss all available clinical and preclinical data generated by studies investigating the role of PEA in non-affective and affective psychoses. We reviewed all interventional and observational studies, employing either retrospective or prospective methodological approaches with any PEA neuro-biological correlates investigated in psychosis.
. Materials and methods

. . Inclusion and exclusion criteria
All clinical and preclinical evidence about the topic was gathered and systematically reviewed according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines (41). Inclusion criteria were defined as follows: 1. analytic, observational, and interventional studies; 2. studies assessing (i) acute or long-term effects of palmitoylethanolamide (PEA) administration over psychosisrelated biological underpinnings (e.g., neuroimmune disruption, hypothalamus-pituitary-adrenal axis dysregulation); and behavioral features (e.g., negative psychotic symptoms, manic symptoms); or (ii) PEA and PEA signaling-related molecular marker (e.g., other endocannabinoids/acylethanolamines, PEArelated enzymes) modulation in peripheral tissues (e.g., plasma, serum), or in the central nervous system (e.g. cerebrospinal fluid, brain tissue) in psychosis and related conditions. Exclusion criteria were defined as outlined: 1. studies in which (i) PEA was not the intervention or outcome of interest (e.g., studies evaluating only exogenous cannabinoid administration or assessing endogenous cannabinoid levels); and (ii) PEA bio-behavioral correlates were not investigated with reference to psychosis; and (iii) PEA biobehavioral correlates were not directly reported on; 2. reviews; 3. systematic reviews; 4. meta-analyses.

. . Search strategy and data extraction
A literature search was performed using electronic databases (PubMed, Scopus, and Web of Science) for any published original study written in English, on 16 January 2023. In order to be as much inclusive as possible, a combination of broad-meaning terms describing and/or concerning PEA ("palmitoylethanolamide, " "palmitylethanolamide, " "N-2-hydroxyethyl-hexadecanamide, " "N-2-hydroxyethyl-palmitate, " "N-palmitoylethanolamine, " "PEA, " and "palmitoyl-ethanolamine") and psychosis ("schizophreni * , " 'psychosis, " "psychoses, " "psychotic, " "bipolar, " "mania, " "manic") was adopted. Reference lists of all selected studies were scrutinized to identify any adjunctive eligible evidence. Data screening and extraction were conducted according to a two-step selection process (conventional double-screening), performed by two researchers (RB and MC) independently from each other. In the instances of conflicting opinions regarding papers' inclusion, a consensus was sought through discussion with a third senior reviewer (MB).

. . Risk of bias assessment
In light of the methodological heterogeneity of collected evidence, quality of studies assessment was conducted in accordance to an adapted and suitably flexible set of criteria suggested by the Agency for Healthcare Research and Quality (AHRQ) guidance (42), in line with previous research in the field (38)(39)(40). Risk of systematic bias across human studies was ruled Bortoletto et al. .
/fpsyt. . out by screening all papers for potential confounding factors (e.g., gender, age, smoking status, level of education). Furthermore, factors possibly accounting for similarities and differences between all studies were assessed, extracting information about study characteristics, including study design (e.g., observational, interventional), defined study population (for human studies: e.g., schizophrenia (SCZ) patients, clinical high-risk (CHR) subjects; for animal studies: e.g., mouse or rat model), age or developmental stage, gender, adequate psychosis model (for animal studies only: e.g., maternal deprivation, methylazoxymethanol acetate (MAM) prenatal exposure), PEA measure (e.g., PEA dosage and administration route, PEA assessment in tissues), adequate PEA evaluation (e.g., time of exposure, single or multiple assessments), defined control group, comparability of subjects (for human studies only), exclusion criteria/adjusting factors (for human studies only), statistical analyses, and declaration of fundings/sponsorship. The full study protocol is available at https://doi.org/10.17605/OSF.IO/ AFMTK.

. . Study selection
Overall, 418 studies were retrieved through the initial data search. After removing duplicates as well as excluding articles owing to article type (e.g., non-systematic reviews, systematic reviews, meta-analyses), titles, abstracts, or full texts of all records were examined against the inclusion and exclusion criteria following a three-step screening process ( Figure 1). A final list of thirteen studies was used for systematic analysis in this review, including 11 studies conducted only in human populations and two studies performed in animal models, exploring various aspects of palmitoylethanolamide (PEA) signaling pathway (Table 1). These include (i) in vivo PEA add-on treatment exposure in humans with different types of psychoses (e.g., non-affective psychosis, affective psychosis) or psychotic symptoms (e.g., hallucinations) (3 studies; Table 1); (ii) PEA quantitative blood assessment in humans with psychosis clinical high-risk (CHR) state (1 study; Table 1); (iii) PEA quantitative blood assessment in humans with different types of psychoses (e.g., non-affective psychosis, affective psychosis) at different stages of illness (5 studies; Table 1); (iv) PEA quantitative central nervous system (CNS; e.g., brain tissue, cerebrospinal fluid) assessment in humans with schizophrenia (SCZ; 2 studies; Table 1); (v) PEA quantitative brain tissue assessment in animal models of SCZ (2 studies; Table 1). Additional data on methodological quality of studies conducted in humans and animals are reported in Tables 2, 3. A brief synthesis of the main results is presented below and summarized in Table 1.  1. Effect on eCBs/AEs levels: (a) 2-AG levels: SCZ, ↑; brain region; SCZ x brain region interaction, NS; (b) AEA levels: SCZ, ↓; brain region, NS; SCZ x brain region interaction, NS; (c) DHEA levels: SCZ, ↓; brain region; SCZ x brain region interaction; (d) PEA, (e) LEA levels: SCZ, ↓; brain region; SCZ x brain region interaction, NS; (f) OEA levels: SCZ, NS; brain region; SCZ x brain region interaction, NS; 2. PEA brain tissue levels:              .      . . In vivo palmitoylethanolamide (PEA) add-on treatment exposure in humans with di erent types of psychoses or psychotic symptoms Three human studies have addressed this area (Table 1) (51)]. Apart from a single study lacking a controlled condition (51), all studies adopted a randomized, double-blind, placebo-controlled design (52,53). Overall, results indicated a beneficial effect of PEA adjunctive therapy on residual negative and general psychopathological symptoms, but not positive symptoms, of risperidone-treated SCZ patients (52), as well as on manic symptoms of lithium-plus risperidone-treated BPAD patients (53). Coherent data emerged regarding depressive symptomatology, which did not appear to be improved in both SCZ and BPAD patients treated with PEA add-on as compared to placebo (52,53). A single study addressing the effect of PEA over nonmotor symptoms among levodopa-treated PD patients, showed no reduction in the number of subjects presenting with hallucinations and psychosis (51). Noteworthy, PEA was well-tolerated, in the absence of extrapyramidal symptoms or any other relevant side effect across the three studies, and for the entire duration of the compound administration.
. . Palmitoylethanolamide quantitative blood assessment in humans with psychosis clinical high-risk state This systematic reappraisal identified a single human study specifically assessing peripheral blood PEA levels in individuals suffering from CHR state, as compared to healthy controls (HC) (47) ( Table 1). PEA levels tended to be elevated in CHR patients, even though just approaching statistical significance. Intriguingly, PEA levels appeared to be significantly higher in those who were both CHR and had been exposed to childhood trauma (CT), compared to individuals having none of the above-mentioned risk factors or one risk factor alone. Furthermore, a significant positive correlation between PEA levels and the severity of CHR state and CT was observed (47).
. . Palmitoylethanolamide quantitative blood assessment in humans with di erent types of psychoses at di erent stages of illness Most of the studies retrieved investigated peripheral blood PEA levels in patients with non-affective [e.g., schizophrenia (SCZ)  (Table 1). Two studies converged on the evidence of higher PEA levels in SCZ patients than in healthy controls (HC) (46,48), one of which further suggesting significantly higher PEA plasma concentration in SCZ patients compared to patients meeting criteria for cannabis use disorder (CUD) or dual diagnosis of CUD and SCZ (48). Interestingly, compared to HC, unaffected twin siblings of SCZ patients also showed increased PEA levels, that did not differ from those of patients (46). Remarkably, antipsychotic (AP)-naïve first-episode psychosis (FEP) patients compared to HC showed a tendency to elevated PEA plasma levels and a significantly higher PEA/2-arachidonoylglycerol (2-AG) ratio, with both that subsided an average of 0.6 and 5.1 years after the initiation of AP treatment (49). The modulating effect of AP treatment over acylethanolamines (AEs) levels was also investigated through a double-blind, randomized, parallel-group, controlled clinical trial, showing elevated serum PEA concentration in cannabidiol (CBD)-treated SCZ patients, compared with those treated with the antipsychotic amisulpride (44). Studies measuring PEA levels among BPAD patients showed a less pronounced increase in affected and unaffected siblings of illness-discordant twin couples, when compared to HC (46). Further, a higher PEA plasma concentration was found in BPAD patients having first episode as depression than in those who had their first episode as mania (50) as well as in those who had at least one depressive episode than in patients who had no prior depressive episodes (50). Finally, PEA levels were increased according to the number of depressive episodes and the presence of depressive mood and anxiety, while inversely correlating with the number of hypomanic episodes, the number of hospitalizations, the duration of valproate (VPA) treatment, sexual desire, presence of flight of ideas, delusion, and grandiosity (50).
. . Palmitoylethanolamide quantitative central nervous system assessment in humans with schizophrenia Two studies analyzed PEA levels in the CNS of patients with psychosis (43,45) (Table 1). In particular, PEA levels were reported to be elevated in the cerebrospinal fluid (CSF) of SCZ and schizophreniform psychosis patients compared with healthy controls (HC) (43). Conversely, a study on postmortem brain samples from subjects diagnosed with SCZ compared to controls indicated lower PEA levels in the cerebellum of antipsychotic (AP)free patients only (45). No significant differences in PEA brain quantification were detected among all study groups in the other brain areas investigated (45).

. Discussion
This is the first systematic review of all evidence exploring the biobehavioral correlates of palmitoylethanolamide (PEA) in psychosis across human and animal studies. Unlike previous research in this field (38)(39)(40), the greatest majority of records included consisted of human studies. Existing reviews focusing on the role of the major phytocannabinoid cannabidiol (CBD) as a potential treatment for schizophrenia (SCZ) patients (20, 56, 57) gathered still preliminary evidence supporting its antipsychotic (AP) efficacy, while highlighting its advantageous side effect profile and good tolerability. Targeting similar pathways, PEA may be considered as a viable alternative to CBD with implications for many therapeutic areas, due to its established safety and the development of formulations maximizing its bioavailability (33,37).
Overall, the present review demonstrated that PEA may be involved in different psychotic phenotypes. Also, it found initial evidence that PEA levels may reflect the severity of the disorder as well as the stage of illness. Evidence was obtained from both interventional studies addressing the AP potential of PEA supplementation, and observational studies of PEA tone in peripheral blood and in the central nervous system (CNS) in the context of clinical high-risk (CHR) for psychosis and psychotic disorders at different stages of illness.
Some important findings from this systematic review deserve to be highlighted. First, despite its promise, evidence regarding the therapeutic potential of PEA supplementation for psychosis is still limited (51)(52)(53) and provides findings about selective efficacy on specific symptoms. In particular, PEA add-on to conventional psychotropic medications did not significantly reduce positive psychotic symptoms (51)(52)(53), while ameliorating negative psychotic (52) and acute manic (53) symptoms. Importantly, while not being the focus of this review, results presented here are inconclusive about a potential role of PEA in ameliorating depressive symptoms among psychosis patients (53). Also, in only one study participants were clearly asked to avoid forms of cognitive behavioral therapies during the trial period (52), thus requiring future interventional studies to clearly . /fpsyt. . rule out a potential masking effect of add-on psychotherapy over symptoms. Second, a line of research investigating endocannabinoids (eCBs) tone in blood among subjects with genetic vulnerability to psychosis (i.e., unaffected twins of SCZ patients) (46), CHR individuals (47), untreated first-episode psychosis (FEP) patients (49), and longer-course SCZ patients (46,48), converged on the evidence of increased PEA plasma levels as compared to healthy subjects. Also, among CHR patients, the more severe the clinical picture (i.e., greater symptom severity) and the risk profile (i.e., more severe childhood trauma), the higher were the PEA levels (47). Based on PEA-related neurobiological mechanisms, the finding of augmented PEA release across different populations may reflect an endogenous attempt to restore homodynamic balance under disease conditions (28,29). However, follow-up studies revealed that PEA plasma levels are no longer heightened in AP-treated SCZ patients after 0.6 years of treatment, and further decreased at 5.1 years from baseline (49), potentially suggesting that such biological self-regulation of PEA levels is lost as the diseases progresses. Interestingly, a 2-to 4-week CBD treatment among SCZ patients was associated with higher PEA levels (44) when compared with a 2-to 4-week AP treatment (44), possibly indicating a CBD-specific property in sustaining PEA tone.
Further, differently from SCZ spectrum disorders, PEA plasma level increase was less pronounced in bipolar disorder (BPAD) patients and unaffected twins of BPAD patients, when compared to healthy subjects (46), perhaps accounting for existing phenotypical discrepancies within the affective psychosis group. In fact, PEA plasma levels were greater among BPAD patients having first episode as depression and increased consistently as a function of the lifetime number of depressive episodes (50). Instead, the occurrence of positive psychotic symptoms, manic symptoms, and hypomanic episodes, was correlated with reduced PEA tone (50). Based on evidence that BPAD patients with manic onset have an older age at diagnosis and a longer duration of untreated illness than those with depressive onset (58), and that polarity of episodes over time often reflects polarity at onset (59), such different findings among BPAD patients may suggest a changing pattern in PEA levels over time. In other words, PEA plasma levels would be higher in the first phases of the disorder, while declining because of disease progression, in line with what observed for non-affective psychosis, and thus strengthening the rationale for its supplementation.
Third, CNS PEA levels were found to be augmented in the cerebrospinal fluid (CSF) of young adults with SCZ and schizophreniform disorder (43), whereas reduced in postmortem cerebellar samples of AP-free middle-aged SCZ patients (45), as compared to healthy controls. Similarly, brain PEA levels did not differ between AP-treated middle-aged SCZ patients and healthy controls (45). Again, it can be assumed that PEA levels are elevated in the early stages of psychosis, potentially reflecting an innate compensatory mechanism, before dropping concomitantly to disease progression. With reference to AP treatment, while on one hand it did not appear to increase PEA levels, on the other it is not clear if it prevented a more pronounced longer-term decrease (45).
Only two studies assessed PEA levels in the prefrontal cortex (PFC) of preclinical models of psychosis. They did not show any significant differences as compared to the control condition (54,55). Further animal studies will need to yield a more robust understanding of the neurobiological mechanisms involving PEA in psychosis.
The findings of this review should be seen considering some strengths and limitations. Despite supporting PEA tone alterations in psychosis and effectiveness of PEA supplementation as a therapeutic strategy, research in this field needs to be expanded, especially to fully comprehend the potential of PEA supplementation as add-on therapy for psychosis across all its symptoms dimensions. Indeed, evidence points toward a beneficial effect of PEA over negative psychotic symptoms and acute manic symptoms, reasonably due to the protective role of the compound against altered neuroinflammation and related mechanisms (28,29), while PEA effect over positive psychotic symptoms remains to be clarified. Further, future follow-up studies will have to investigate whether PEA effect is sustained in the longer-term. Besides, to date, evidence of PEA effect as monotherapy in the clinical stages preceding full-blown psychosis onset to prevent the risk of progression is totally lacking and is worth of exploration. Finally, PEA levels appear to be altered in psychosis. However, further clarification is needed as to whether PEA tone is altered to a different extent depending on the stage of illness and whether this can be considered a biomarker of psychosis. In line with this, whether any AP treatment may be beneficial through PEA tone modulation requires additional investigation. Biobehavioral comparisons between CBD and PEA are also worth of consideration, especially whether CBD antipsychotic effects are mediated by PEA signaling and similar effects can be obtained directly supplementing PEA. The latter would be inevitably more advantageous, due to its safer profile and shorter biological distance from the therapeutic target.

. Conclusions
This systematic review provided a first overview of all observational and experimental studies of PEA and its biobehavioral correlates in psychosis. Although in its infancy and still limited, research in this field is primarily carried out on humans and provides evidence for both alterations in PEA signaling, implications for psychosis-related behavioral features, and benefits from PEA supplementation. In particular, PEA may be useful to improve negative psychotic symptoms and manic symptoms. Noteworthy, no serious adverse events were reported across all human studies investigating its administration, further supporting PEA potential effectiveness and elevated safety as a therapeutic intervention in psychosis.