Schizophrenia is classically defined as a neurodevelopmental psychiatric disorder (Lewis and Levitt, 2002). However, the heterogeneous illness presentation, course, and outcomes have hindered the development of novel effective treatments. Schizophrenia is also a neuropsychiatric disorder that results in significant economic burden (Chong et al., 2016), caregiver responsibility (Szkultecka-Debek et al., 2016), and global health disability, since most patients are unable to achieve complete functional recovery (e.g., consistent paid employment, living independence, etc.).

Characteristic symptoms of schizophrenia include positive (delusions, auditory, and visual hallucinations) (Morris et al., 2012), negative (amotivation, anhedonia, apathy, inappropriate affect) (Rabinowitz et al., 2012), and dysfunction in multiple neurocognitive domains including attention (Fioravanti et al., 2005), learning and memory (Goldman-Rakic, 1994; Gold et al., 1997; Manoach, 2003), executive functioning (Hutton et al., 1998), processing speed (Rodriguez-Sanchez et al., 2007), and IQ (Zammit et al., 2004). More recent studies have consistently reported deficits in additional neuropsychiatric phenotypes including social cognition and functioning (Nuechterlein et al., 2004; Fett et al., 2011), prediction error and reward learning (Kapur, 2003; Corlett et al., 2007), and sensory gating (Braff and Geyer, 1990; Hazlett et al., 2015).

The etiology of schizophrenia resembles most chronic diseases, an interaction of complex environmental and genetic risk factors. Psychosocial stressors including poor socioeconomic status (Werner et al., 2007), migration status (Cantor-Graae and Selten, 2005), lack of social relationships (Jones et al., 1993; Schenkel et al., 2005), in conjunction with multiple genetic risk loci (Ripke et al., 2014) and epigenetic modifications (Roth et al., 2009) leads to alterations in neurotransmitter systems (Tsai et al., 1995; Howes and Kapur, 2009), neuroimmune activation (Bayer et al., 1999), brain structure (Suddath et al., 1990; Ho et al., 2003), and brain functional connectivity (Gur et al., 1985; Andreasen et al., 1994).

Publications implicating premature mortality as a characteristic of schizophrenia have been ongoing in the literature for the past few decades. Some of these past reports have suggested that attenuated lifespan may be independent of schizophrenia symptom chronicity. During the 1960's the primary causes of mortality in schizophrenia were thought to be due to the following diseases or illnesses: infection, cardiovascular-renal, neoplasm, endocrine and metabolic, suicide/accident or other external causes, and other disease and unspecified causes (Niswander et al., 1963).

Current estimates suggest that the lifespan for people with schizophrenia is approximately 28.5 years shorter than the general population (Olfson et al., 2015). Interestingly, the marked increase in mortality for people with schizophrenia continues to largely be a consequence of immune or metabolic illness exacerbation (Saha et al., 2007) as reported decades ago. Most recent studies suggest that cardiovascular disease (Hennekens et al., 2005; Olfson et al., 2015), type II diabetes mellitus (Olfson et al., 2015), sepsis (Seeman, 2007; Olfson et al., 2015), gastrointestinal or digestive disease (Dickerson et al., 2016), autoimmune disorders (Dickerson et al., 2016), influenza and pneumonia (Olfson et al., 2015) are the major causes of mortality in schizophrenia. To further complicate matters, these immune and metabolic disorders have complex convergent and divergent biological mechanisms (Hotamisligil, 2006). Notably, the gut is the key functional organ for many of these disorders.

Comparatively little scientific effort has been focused on modifying gut-neuropsychiatric pathways in schizophrenia. This is in part due to our limited knowledge of gastrointestinal (GI) functioning in relation to schizophrenia disease onset, illness course, or comorbidities. The extant literature suggests that multiple immune and metabolic makers are likely to mediate the relationships between gut functioning and neuropsychiatric outcomes.

Findings from studies of chronic schizophrenia and bipolar patients indicate serum elevation of bacterial markers (Severance et al., 2013) also present on gut microbes, implicating increased bacterial translocation from the gut. Toxoplasma gondii (bacteria that infects the GI tract) seropositive status has been linked to development or progression of multiple neuropsychiatric diseases, including schizophrenia (Severance et al., 2016b). Inflammatory GI diseases, especially colitis, are thought to be highly prevalent (in over 90% of samples) in schizophrenia based on post mortem biopsy (Hemmings, 1990, 2004). Sex specific GI dysfunction may also be present in males with schizophrenia due to Candida albicans exposure (Severance et al., 2016a).

The relationships among schizophrenia illness, diet, and gut based immune function, is also thought to be present through the c1q component of the complement pathway (Severance et al., 2012). Notably, the complement pathway makers and associated genes are recognized as potential predictors of schizophrenia genetic risk (Sekar et al., 2016), excessive synaptic pruning (Inta et al., 2016), and other biological outcomes (Nsaiba et al., 2015). Taken together, these findings support the role of diet as a possible environmental factor that contributes to the biochemical and genetic variation observed in schizophrenia. Moreover, dietary intake patterns and dietary interventions are increasingly being explored for their ability to reduce inflammation and metabolic disease risk, with a minimal likelihood of adverse effects. Therefore, these and other gut based treatments that have the potential to target converging immune and metabolic pathways could be most beneficial for people with schizophrenia.

The remainder of this article will provide an overview of key observations and treatments associated with metabolic syndrome and type II diabetes, cardiovascular disease, and inflammation as pertinent to schizophrenia. This is followed by a brief analysis of dietary intake and the hypothesized neurobiological mechanisms for unhealthy dietary intake patterns in schizophrenia. Empirically based dietary modifications that have been tested in schizophrenia or are potentially relevant to schizophrenia, short chain fatty acids (SCFA), immune, and metabolic dysfunction will be reviewed. Then, the production of SCFA in the colon, their systemic transport, and findings of the SCFA in the brain and links to immune and metabolic function most germane to schizophrenia will be examined. Lastly, alternative dietary modifications, such as a high-fiber, Mediterranean style diet, that enriches production of SCFA, will be discussed as a potential adjunctive treatment for schizophrenia.

Metabolic syndrome is a combination of three of the following physiological factors: (1) abdominal obesity, (2) high triglyceride levels, (3) elevated high density lipoprotein (HDL) levels, (4) blood pressure, and (5) insulin resistance. It is well recognized that the incidence of metabolic syndrome, along with the incidence of type II diabetes that typically follows metabolic syndrome, is 20% higher in chronic schizophrenia patients that the general population (Dixon et al., 2000; Mitchell et al., 2013). Metformin is now being investigated and implemented as an adjunctive therapy to help mitigate antipsychotic induced metabolic syndrome (Jarskog et al., 2013).

Although, certain antipsychotic medications (McEvoy et al., 2005; De Hert et al., 2006), illness chronicity, lifestyle habits such as diet, smoking, etc., and aging related factors (Subramaniam et al., 2003) contribute to incidence of metabolic syndrome; a higher prevalence of metabolic syndrome and type II diabetes during early stages of (Correll et al., 2014) and antipsychotic medication naïve (Ryan et al., 2003; Fernandez-Egea et al., 2009; Pillinger et al., 2017) schizophrenia patients has been reported. Besides the profound effects metabolic syndrome and type II diabetes have on premature mortality in schizophrenia, they have also been associated with poor school performance in adolescence (de Nijs and Pet, 2016), sensory gating deficits (Micoulaud-Franchi et al., 2015) and other neurocognitive outcomes (Lindenmayer et al., 2012; Goughari et al., 2015).

The metabolic syndrome risks associated with schizophrenia may not be limited to factors that are observed post illness onset. Maternal type II diabetes is considered a risk factor for fetal neurodevelopment disorders, including schizophrenia (Cannon et al., 2002). The hypothesized biological mechanism for this risk factor is altered docosahexaenoic acid (DHA) transfer to the fetus (Judge et al., 2016). Therefore, it is not only important to address metabolic syndrome and diabetes in individuals who have already been diagnosed with schizophrenia, but to manage metabolic syndrome and type II diabetes in pregnant women to reduce the subsequent fetal neurodevelopmental risk. Notably, the SCFA butyrate has been shown to decrease metabolic impairments in pregnant mice (Li et al., 2013). Future studies should also consider the potential fetal neuroprotective effects of SCFA in models of maternal diabetes and risk for psychosis.

The etiology of cardiovascular disease as relevant to schizophrenia is multifactorial and complex, with antipsychotic medications (Peet, 2004), increased incidence of smoking (McCreadie, 2003; Hennekens et al., 2005), unhealthy dietary and excess sodium intake (Brown et al., 1999; Teasdale et al., 2016), sedentary behavior (McCreadie et al., 1998), each having a substantial role. The primary cardiovascular disease risk markers that are routinely investigated in schizophrenia include metabolic syndrome, Framingham 10-year Relative Risk score, C-reactive protein, and dyslipidemia.

The Framingham 10-year relative risk score is a recognized tool for predicting future coronary events and has been validated in various populations (Lakoski et al., 2007). Although, modifications of the Framingham 10-year relative risk score calculator have been developed for various research and treatment programs, the commonly recognized calculator is comprised of the following factors: sex, age, HDL, and total cholesterol levels, smoking status, and systolic blood pressure. Compared to individuals without a psychiatric disorder, the Framingham 10-year relative risk score is significantly higher in people with schizophrenia (Goff et al., 2005; Jin et al., 2011).

C reactive protein (CRP), an acute phase protein and cardiovascular disease risk marker (Ridker, 2001), triggers other detrimental cardiovascular outcomes such as increased clotting, generation of oxygen radicals and plaque destabilization (Prasad, 2006). CRP levels are modulated by the inflammatory cytokines Interleukin-6 (IL-6) and Tumor Necrosis Factor-α (TNF-α) that is secreted by macrophages and adipose cells (Puglisi and Fernandez, 2008). Notably, elevated CRP levels have been reported in schizophrenia by multiple research groups (Dickerson et al., 2013; Sicras-Mainar et al., 2013).

Although, cross sectional survey in the general population indicates a positive relationship between plasma CRP levels and Framingham 10-year Relative Risk score (Albert et al., 2003), the exact relationship between Framingham relative risk and CRP levels in schizophrenia remains unclear. Some research groups have found a positive association between elevated CRP levels and Framingham risk score (Sicras-Mainar et al., 2013), whereas other reports did not observe a significant association (Joseph et al., 2015). In schizophrenia, both CRP and Framingham risk have been linked to higher body mass index (BMI) (Miller et al., 2014; Joseph et al., 2015), psychiatric symptom severity (Barzilay et al., 2016; Dimitrov et al., 2016), and other dysregulated metabolic factors including fasting glucose and hemoglobin A1c levels (Dieset et al., 2012; Joseph et al., 2015). A correlation between CRP levels and routinely prescribed antipsychotic medication treatments has also been observed by some groups (Stefanovic et al., 2015). However, these findings have not been consistent as other recent findings suggest that antipsychotic medications do not target elevated CRP levels in schizophrenia (Fernandes et al., 2016).

In addition to elevated Framingham relative risk scores and CRP levels, prominent dyslipidemia has been observed in schizophrenia (Hennekens et al., 2005; Nasrallah et al., 2006). In people experiencing early illness stages of schizophrenia, significant elevations of triglycerides and other non-HDL lipids has been reported (Correll et al., 2014). In chronic patients, lipid levels have been linked to psychiatric symptom severity (Solberg et al., 2015). Pilot studies of pravastatin and simvastatin, lipid lowering medications, have been conducted in schizophrenia patients. The preliminary findings from these trials also indicate these medications also provide a temporary reduction in inflammation, positive, and negative symptoms (Chaudhry et al., 2014; Vincenzi et al., 2014). However, larger scale clinical trials that consider their long-term effects are necessary to replicate and confirm long-term beneficial outcomes.

Overall, treatment for dyslipidemia, diabetes, and hypertension has been underutilized in schizophrenia (Nasrallah et al., 2006). It is important to implement existing treatments and develop novel interventions directed at reducing cardiovascular disease risk at early illness stages to improve lifespan and outcomes in schizophrenia. Therefore, high fiber diets (Ma et al., 2006) and pharmacological interventions to target cardiovascular risk factors such as novel anticoagulants, lipid lowering agents, beta-adrenoreceptor antagonists, and angiotensin converting enzyme (ACE) inhibitors (Prasad, 2006) are important adjunctive treatments to consider for people with schizophrenia.

Numerous clinical and epidemiological studies indicate that the rates of obesity and morbid obesity for people with schizophrenia are significantly higher than the general population (Hsiao et al., 2004; Susce et al., 2005). While elevated BMI is commonly associated with second generation antipsychotic medication effects (Grunder et al., 2016); ethnic/racial differences, diet, physical activity, and other lifestyle factors also contribute to the prevalence of obesity in schizophrenia (Brown et al., 1999; Norlelawati et al., 2012).

Current studies of dietary intake in schizophrenia suggest that most patients have a poor diet largely characterized by increased fast and processed foods (Strassnig et al., 2003), increased sodium and cholesterol intake (Nunes et al., 2014), and higher saturated fat and lower fiber content than non-psychiatric comparison subjects (Brown et al., 1999; Henderson et al., 2006). Although, increased sugar and processed diet consumption is thought to be characteristic of people with schizophrenia originating from Western and European countries (Peet, 2003; Stokes and Peet, 2004), similar findings have also been observed in patients from Eastern countries (Sugawara et al., 2014; Ito et al., 2015). The interaction of sex specific effects for BMI and dietary consumption in schizophrenia may also be present with female patients reported to have a higher BMI and unhealthier diet than males (Amani, 2007; Carliner et al., 2014). Homelessness and poor socioeconomic status is also likely to have a major influence on nutritional status and dietary intake patterns in schizophrenia. However, this has not been taken into consideration for many studies.

Nutritional changes including ω-3 or vitamin D deficiencies (Dealberto, 2007) may increase risk for developing psychosis. A lower intake of ω-6, phytosterols, vitamin A, and vitamin E (α-tocopherol) has also been reported in relation to BMI and demographically matched comparison subjects (Nunes et al., 2014). In addition, consumption of fruit and vegetable portions is significantly lower that the recommended daily allowance (Heald et al., 2015). However, other reports indicate that it is primarily the overall, rather than specific type of, caloric intake that is increased in schizophrenia (Strassnig et al., 2003). Additional scientific investigations are needed to clarify the relationships between lifestyle factors, dietary intake, and nutritional status in schizophrenia patient populations. Elucidating the specific role of dietary compounds and nutritional factors in relation to symptom and behavioral outcomes will require combined efforts from animal model testing and human clinical trials.

The neurobiological and neurocognitive mechanisms that lead to unhealthy dietary intake by people with schizophrenia are hypothesized to be related to dysregulated reward circuitry (Elman et al., 2006): a hyperdopamineric mesolimbic pathway combined with poor cognitive control (Kapur, 2003). This pathway has also been implicated in the context of obesity (Vucetic and Reyes, 2010), food cravings (Blum et al., 2011), eating disorders (Wagner et al., 2007) and addiction (Volkow et al., 2012), with altered reward circuitry being a major common pathway linked to comorbid substance abuse in schizophrenia (Chambers et al., 2001).

Deficits in reward learning have been observed in multiple schizophrenia patient studies (Juckel et al., 2006) and have been consistently linked to co-occurring negative symptoms (Strauss et al., 2011; Gold et al., 2012). Yet, the ability to alter responses with the use of prediction error is thought to be intact for people with schizophrenia (Gold et al., 2012). This could potentially be utilized as a cognitive strategy to help modify and improve the quality of dietary intake. However, the specific relationships between reward circuits and dietary intake patterns in schizophrenia remain unclear. Novel investigations are needed to determine the neurocognitive and neurobiological factors that may be contributing to unhealthy dietary intake and nutritional status in schizophrenia.

The short chain fatty acids (SCFA) acetate, propionate, and butyrate are the primary metabolic products derived from the colonic fermentation of dietary fibers by gut microbes and is estimated to be produced in a molar ratio of 60:20:20 respectively (Wong et al., 2006). Table 1 summarizes the major dietary carbohydrates, primary sources for these carbohydrates, and identified gut microbiota that ferment these carbohydrates to produce SCFA.

Short chain fatty acids bind to the free fatty acid receptors FFAR2 (GPR41) and FFAR3 (GPR43). Propionate binds to human GPR41 and GPR43 with equal affinity. Acetate has shown to be selective for GPR43, whereas butyrate preferentially binds GPR41 (Le Poul et al., 2003). GPR41 expression has been detected in the autonomic sensory ganglia, pancreas, spleen, lymph nodes, bone marrow, adipose tissue, and peripheral blood mononuclear cells including monocytes (Brown et al., 2003; Le Poul et al., 2003; Nohr et al., 2015). GPR43 expression has primarily been observed in the colon, ileum, adipose tissue, and immune cells especially monocytes and neutrophils (Brown et al., 2003; Le Poul et al., 2003; Nilsson et al., 2003). Expression of these receptors in human or animal brain models has not yet been well characterized.

Most of the butyrate that is produced in the colon is taken up by colonocytes as an energy source (Wong et al., 2006). However, SCFA are also then transported from the colon via the hepatic portal vein to the liver. The predominant SCFA that is absorbed by the liver is propionate for gluconeogenesis (Cummings et al., 1987). From the liver, SCFA enter the systemic circulation. In healthy, BMI and lipid level matched, men and women, serum concentrations of SCFA ranged from 20–190, 1.7–8.4, and 0.0–7.6 μmol/L for acetate, propionate, and butyrate, respectively (Wolever and Bolognesi, 1996). Circulating levels of SCFA are able to cross the blood brain barrier, and the primary SCFA uptake by the brain is for butyrate, followed by propionate and acetate (Oldendorf, 1973). In the human brain, butyrate and propionate concentrations are estimated to be 17.0 and 18.8 pmol/mg, respectively (Bachmann et al., 1979).

Sodium butyrate is anti-inflammatory against LPS induced inflammation in rat primary microglia, hippocampal cultures, and neuronal co-cultures of microglial cells, astrocytes and cerebellar granule neurons (Huuskonen et al., 2004). Butyrate treatment in hippocampal slice cultures also resulted in the downregulation of NF-κB-binding capacity induced by LPS (Huuskonen et al., 2004).

In addition to targeting brain derived inflammation, SCFA have the ability to modulate multiple immune and epigenetic pathways including obesity induced inflammation (Meijer et al., 2010), IL-6 and TNF-α release from macrophages (Kim et al., 2014), inhibiting cytokine induced NF-κB activation (Tedelind et al., 2007), and histone deacetylase inhibition (Wong et al., 2006). These biological pathways have also been shown to be dysregulated in schizophrenia (Fan et al., 2007; Sharma et al., 2008; Song et al., 2009; Miller et al., 2011). Therefore, the potential of SCFA as an adjunctive treatment may have significant beneficial outcomes. However, the role of SCFA in schizophrenia risk, onset, and comorbid illness outcomes remains unknown.

Although, the abovementioned studies demonstrate that SCFA are present in the brain and modify inflammation in a beneficial manner, administration of valproic acid, a medication commonly prescribed for symptoms associated with bipolar disorder and epilepsy, inhibits the transport of SCFA across the blood brain barrier in rodents (Adkison and Shen, 1996). In addition, in vitro studies reveal that free fatty acids in the intestine can have cytotoxic properties (Penn and Schmid-Schonbein, 2008). Therefore, there is still much to be learned about the compounds that modulate SCFA and the types and expression of the receptors that SCFA target. In addition, the role of SCFA in the brain and their relationship to neurobiological factors and pathways including neurotransmitter circuits, neurotrophic factors and other brain metabolites remains largely unknown.

Mediterranean based diet treatment trials have led to reductions in overall cardiovascular disease risk (Estruch et al., 2013) compared to most Western diets. It is thought that the Mediterranean diet primarily ameliorates cardiovascular disease by providing a more optimal ω-6/ω-3 ratio (Simopoulos, 2002). In addition, adherence to a Mediterranean style diet reduces levels of CRP and TNF-α (Koloverou et al., 2016; Neale et al., 2016), immune markers routinely linked to poor cardiovascular outcomes. In type II diabetes patients, the Mediterranean style diet significantly reduced hemoglobin A1c levels (Elhayany et al., 2010), suggesting that a Mediterranean style diet will likely benefit schizophrenia patients with comorbid type II diabetes.

In individuals who were defined as healthy based on screening for autoimmune, cancer, and digestive diseases, a higher Mediterranean diet score was associated with increased abundance of health beneficial gut microbiota and coincided with higher fecal concentrations of propionate and butyrate (Gutierrez-Diaz et al., 2016). Studies comparing the effects of Mediterranean style, ketogenic, and low fat diets in obese individuals indicate that Mediterranean diets are equally effective as the other diets for weight loss while having the added benefit of maintaining glycemic and lipid control (Shai et al., 2008). To our knowledge, studies of Mediterranean diets in schizophrenia have not yet been conducted. Therefore, incorporating a high-fiber, ω-3 rich, Mediterranean style diet into patient lifestyle management and treatment could very likely improve the metabolic and immune outcomes that have consistently linked to premature mortality in schizophrenia. Higher fermentable fiber intake through a modified Mediterranean diet lifestyle should increase circulating levels of SCFA and may also directly mitigate significant constipation, a gastrointestinal side effect associated with some commonly prescribed antipsychotic medications (De Hert et al., 2011).

Dietary modifications and interventions provide an opportunity to directly target gut, immune, and metabolic markers. This remains highly underexplored in the context of psychiatric disorders, especially schizophrenia. Dietary and nutritional investigations that have been conducted in schizophrenia are few, with randomized controlled trials of dietary modification being scarce, thereby leaving a significant gap in the schizophrenia literature.

Schizophrenia patients are most likely to benefit from the implementation of individualized dietary interventions due to co-occurring nutritional deficiencies (Hoffer, 2008; Kale et al., 2010). Many people with schizophrenia vary in insight with regards to their illness (Aleman et al., 2006), dietary intake habits (Heald et al., 2015), social support (Buchanan, 1995), and everyday functioning (Sabbag et al., 2012). For dietary interventions in schizophrenia patient populations to be successful, the combined support from scientists, dieticians, family members, and neuropsychiatric clinicians who have been successful in implementing behavioral modifications will be necessary.

SCFA can be assayed from blood and fecal samples, and enriched through high fiber dietary modification. To date, the role of SCFA in relation to dietary intake, immune, and metabolic outcomes has not been investigated in schizophrenia. Therefore, it is unknown whether high-fiber, Mediterranean type, dietary modification enriched for SCFA production, could have a direct effect on improving schizophrenia psychopathology, an additive effect via concurrent modification of immune and metabolic markers that may trigger schizophrenia symptoms, or primarily an indirect effect. Translational animal studies and human clinical trials will be needed to determine the exact SCFA and Mediterranean dietary components that can improve short and long term physiological and behavioral outcomes for people with schizophrenia.

JJ developed the hypotheses and wrote the manuscript. CD assisted with hypothesis development. PS, KC, and GS edited the manuscript.