Psychiatric Comorbidity in Children and Adults with Gluten-Related Disorders: A Narrative Review

Gluten-related disorders are characterized by both intestinal and extraintestinal manifestations. Previous studies have suggested an association between gluten-related disorder and psychiatric comorbidities. The objective of our current review is to provide a comprehensive review of this association in children and adults. A systematic literature search using MEDLINE, Embase and PsycINFO from inception to 2018 using terms of ‘celiac disease’ or ‘gluten-sensitivity-related disorders’ combined with terms of ‘mental disorders’ was conducted. A total of 47 articles were included in our review, of which 28 studies were conducted in adults, 11 studies in children and eight studies included both children and adults. The majority of studies were conducted in celiac disease, two studies in non-celiac gluten sensitivity and none in wheat allergy. Enough evidence is currently available supporting the association of celiac disease with depression and, to a lesser extent, with eating disorders. Further investigation is warranted to evaluate the association suggested with other psychiatric disorders. In conclusion, routine surveillance of potential psychiatric manifestations in children and adults with gluten-related disorders should be carried out by the attending physician.


Introduction
Gluten-related disorders include three pathologies caused by the ingestion of gluten-containing cereals grains, namely celiac disease (CD), non-celiac gluten sensitivity (NCGS) and wheat allergy (WA) [1]. Although all of them are due to the toxicity of gluten proteins in the sensitive subject, their respective pathogenetic mechanisms differ.
Celiac disease is a systemic autoimmune disease due to a permanent intolerance to gluten which causes villous atrophy of the intestinal mucosa. It involves both innate and adaptive immune responses that appear in genetically predisposed subjects exposed to gluten and, unlike food allergies, it is not mediated by an immediate hypersensitivity reaction. It is a polygenic multifactorial disorder whose development depends on the genetic constitution of the subject, on his/her exposure to gluten intake, and on different environmental factors [2,3]. To date, the only effective treatment for the disease is to observe a life-long strict gluten-free diet although other therapeutic approaches are being explored [4].
In relation to the genetic background of the disease, two HLA class II genes, the HLA-DQ2 and the HLA-DQ8 heterodimers are present in almost all CD patients and their simultaneous absence in a subject usually rules out a diagnosis of CD. However, these genes are also common in the general population and the implication of other non-HLA genes is being investigated by genome wide association studies [5]. Environmental factors that facilitate or, conversely, protect against the Wheat allergy is an IgE-mediated reaction to the proteins contained in wheat and in particular, although not exclusively, the omega-5-gliadin. WA can be developed by inhalation of wheat flour, the so-called baker's asthma and baker's rhinitis which are considered occupational diseases, or by wheat ingestion [21]. The latter case, which is the most frequent, may cause urticaria, angioedema and/or gastrointestinal symptoms such as nausea, vomiting, abdominal bloating, abdominal pain and diarrhea; in the most severe cases it can induce systemic anaphylaxis [18]. WA is especially frequent in children, being less commonly seen in adolescents and adults. The treatment is based on the avoidance of wheat-containing foods, being less restrictive compared to gluten-free diet in CD, as it does not require the restriction of rye and barley-containing foods [22].
Psychiatric disturbances have frequently been reported in patients with CD. Several narrative reviews of the literature undertaken in the last five years indicate that CD could be associated with a wide spectrum of psychiatric disorders, including anxiety disorders, dysthymia, major depression, bipolar disorders, schizophrenia, eating disorders, autism spectrum disorders, and attention-deficit hyperactive disorders [23][24][25][26][27]. However, these otherwise important reviews have several limitations. Several of them were focused on specific psychiatric disorders such as anxiety and depression [24], mood disorders and schizophrenia [25], or severe psychiatric disorders [27]. Some others, according to their objectives comprised the whole spectrum of psychiatric disorders, but they do not specify their search strategies and/or the biomedical literature database used for the review [23,26]. Finally, when specified, literature searches were almost restricted to PubMed, thus providing a limited review of the literature on this topic. Moreover, none of the previous have evaluated the association of psychiatric disorders in children and adults with gluten-related disorders separately. The aim of this manuscript is presenting a comprehensive review of the literature on the potential association of gluten-related disorders with the whole spectrum of psychiatric disorders using the most common literature databases for this kind of evaluation (namely, Medline, EMBASE and PyscINFO).

Search Strategy
We searched the medical literature for published studies indexed in the Medline (1966 to January 2018), EMBASE (1947 to January 2018), and PsycINFO (1967 to January 2018). The search strategy included terms of 'celiac disease' or 'gluten-sensitivity related disorders' combined with terms of 'mental disorders' as described in Supplementary Table S1. No limits or restrictions were applied. Retrieved references were pooled and managed using EndNote X8 (Clarivate Analytics, Philadelphia, PA, USA).

Inclusion Criteria
We included studies that investigated the prevalence, incidence or the likelihood of presenting mental or psychiatric disorders in patients with CD or gluten-sensitivity related disorders. For that purpose, comparative observational or interventional studies, including meta-analysis, assessing the aforementioned objectives as part of their primary or secondary objectives were included. Only studies published in English, Spanish, French, Portuguese, or Italian were included. Case-reports, case-series, abstracts and editorials were excluded. The relationship between CD and psychiatric disorders may be bidirectional. Our purpose was to assess the comorbidity between gluten-related disorders and psychiatric manifestations; thus, those studies assessing the prevalence, incidence or likelihood of presenting CD or gluten-related disorders in patients with diagnosed psychiatric disorders were excluded.
Study eligibility was independently evaluated by the three investigators (MS, EPC, FRV). Discrepancies in the evaluation were resolved by consensus among study investigators.

Data Extraction
Standardized data collection forms were used to extract data that included: (1) name of the first author; (2) year of publication; (3) country where the study was conducted; (4) study objective(s); (5) study design; (6) assessment tools used in psychiatric comorbidities evaluation; (7) Disease diagnostic criteria; (8) sample size and demographic characteristics; and (9) summary of outcomes. Data extraction was independently completed by two investigators (MS and FRV). Discrepancies in data extraction were solved by consensus.

Study Selection
Our systematic search strategy identified 1375 potentially relevant articles (730 articles from EMBASE, 453 articles from MEDLINE and 192 articles from PsycINFO). After removing 461 duplicate articles, 914 articles underwent title and abstract screening. Seven hundred and eighty-eight articles were excluded as they were case-reports, editorials, animal studies, basic science studies, did not include comparator group, or were published in a language other than those specified in the inclusion criteria, leaving 126 articles for a full-text screening. Two studies were excluded because we were unable to obtain their full text [28,29]. A total of 77 were excluded following full-text review because they were either published in abstract form, did not meet the specific objectives set for our current review or did not report outcomes of interest, leaving a total of 47 articles that were included in our review, of which 28 studies were conducted in the adult population, 11 studies were conducted in the pediatric population and eight studies included both adults and children. Mixed studies (including children and adults, n = 8) were classified under the corresponding population group with a larger sample size (pediatrics (n = 4), adults (n = 4)) ( Figure 1).
Nutrients 2018, 10, x FOR PEER REVIEW 4 of 28 extraction was independently completed by two investigators (MS and FRV). Discrepancies in data extraction were solved by consensus.

Study Selection
Our systematic search strategy identified 1375 potentially relevant articles (730 articles from EMBASE, 453 articles from MEDLINE and 192 articles from PsycINFO). After removing 461 duplicate articles, 914 articles underwent title and abstract screening. Seven hundred and eighty-eight articles were excluded as they were case-reports, editorials, animal studies, basic science studies, did not include comparator group, or were published in a language other than those specified in the inclusion criteria, leaving 126 articles for a full-text screening. Two studies were excluded because we were unable to obtain their full text [28,29]. A total of 77 were excluded following full-text review because they were either published in abstract form, did not meet the specific objectives set for our current review or did not report outcomes of interest, leaving a total of 47 articles that were included in our review, of which 28 studies were conducted in the adult population, 11 studies were conducted in the pediatric population and eight studies included both adults and children. Mixed studies (including children and adults, n = 8) were classified under the corresponding population group with a larger sample size (pediatrics (n = 4), adults (n = 4)) ( Figure 1).

Studies Conducted in Children with CD
We found 15 studies that evaluated psychiatric disorders in children or young adults with CD, 11 of which were conducted in clinical-based settings and four were conducted in community-based settings (Table 1). Studies were published between 1997 and 2018 [30][31][32][33][34][35][36][37][38][39][40][41][42][43][44]. Most studies (n = 12) were cross-sectional, although one of them included a subsequent longitudinal phase [41]. Three studies used a population-based cohort design and were conducted in Sweden using the same data source for patients with CD [31,34,44]. Finally, one study used a cohort design [38]. With the exception of this later study which was conducted in several countries [38], the remaining studies were conducted in European countries or Turkey.
According to a population-based cohort study, children with CD have a 70% increased likelihood of presenting a psychiatric disorder with intellectual disability being the most likely disorder (HR 1.7, 95% CI 1.4 to 2.1) [44]. A summary of results of studies evaluating the association between CD and the occurrence or presence of psychiatric disorders is presented in Table 2. Regarding specific conditions, cohort studies have shown that CD is associated with an increased likelihood of occurrence of depression (HR = 1.8, 95% CI 1.6 to 2.2) [34] or mood disorders (HR 1.2, 95% CI 1.0-1.4) [44], although this latter result did not reach statistical significance. In contrast, most cross-sectional studies have found that the point prevalence of depression or the severity of depressive symptoms did not differ in children with CD as compared with controls [33,[35][36][37]. Pynnonen et al. [32], using a cross sectional study, found no differences between patients with CD and controls in the point prevalence of major depressive disorder, but the lifetime prevalence of major depressive disorder was significantly increased in patients with CD (31% vs. 7%; OR = 6.06, 95% CI 1. 18-31.23). Although a population-based study found an increased likelihood of occurrence of anxiety disorder in patients with CD as compared with controls (HR 1.2, 95% CI 1.0 to 1.4, p <0.05) [44], cross-sectional studies have not shown differences between patients with CD and controls in the prevalence or severity of symptom of anxiety [32,35,36]. In children, no association has been found between CD and the occurrence of bipolar disorder [34].
The association of CD with psychotic disorders in children has been scarcely investigated, showing no association with the occurrence of schizophrenia [31] or psychotic disorder [44]; an association has been reported between CD and non-schizophrenic non-affective psychosis (HR 1.61, 95% CI 1.19-2.20) [31].
A population-based study found a significant association between CD and the occurrence of an eating disorder (HR 1.4, 95% CI 1.1 to 1.8) [44], and the presence of the disorder seems to have a negative impact on some dimensions of quality of life (namely, ill-being and joy-in-life) [39]. A population-based cohort found an excess likelihood of occurrence of an autism spectrum disorder in patients with CD as compared to controls [44]; however, a cross-sectional study did not find an association between both disorders [30]. A slight, but significant, increase in the likelihood of occurrence of attention deficit and hyperactive disorder (ADHD) in patients with CD has been reported [44].
Several factors have been suggested to contribute to depressive symptomatology in the pediatric population including the presence of parental depressive disorders, low parental educational level, divorce of the parents, presence of functional comorbid conditions and female gender [32,33,43]. Older age, higher body mass index and history of dietary restrictions were linked to higher risk of eating disorders [39,40].

Studies Conducted in Adults with CD
We found 32 studies that evaluated psychiatric comorbidities in adult patients with CD or NCGS, 18 of which were conducted in clinical-based settings and nine were conducted in community-based settings (Table 3). Studies were published between 1982 and 2018 . More than half of these studies were of cross-sectional design [45,48,51,[53][54][55][56][57]59,60,63,67,68,71,72,[74][75][76] and four of them were representative of the general population [47,70,73,76].  Table 2. Summary of outcomes evaluating the association between gluten-related disorders and psychiatric disorders in children and young adults.

Author (Year) Design Sample Size and Demographic Characteristics Summary of Outcomes Associated Factors with Psychiatric Comorbidities and Other Relevant Information
Autism spectrum disorders Pavone (1997) [30] Cross-sectional CD, n = 120 (mean age 9.6 years, 48% females) Recently-diagnosed CD, n = 27 CD on strict GFD, n = 70 GFD non-adherent CD, n = 23 Controls, n = 20 (mean age 9.6 years, 48% females) -Autism diagnosis: none of the recently-diagnosed CD -Language delay: Two subjects in GFD-compliant, one subject in the non-adherent group -Differences were not statistically significant compared to controls NR Schizophrenia Spectrum

NR
Simsek (2015) [37] Phase 1: Cross-sectional Phase 2: Case-series CD, n = 25 (mean age 11.8 years, 72% females) Controls, n = 25 (mean age 12.2 years, 64%) -At the time of diagnosis (CD vs. controls): CDI scores: 9 vs. 6, p = NS -6 months following GFD initiation: CDI scores in CD: 9 before diet vs. 9.5 after diet, p = NS -Total scores of HRQOL were significantly lower in CD patients (p <0.05) Smith (2017)  -Lifetime prevalence of EDs: 5.3% of girls with CD: anorexia nervosa (n = 1), bulimia nervosa (n = 4), and EDs not otherwise specified (n = 6); 3.9% suffered from current ED -Criteria for lifetime subclinical EDs: 21 girls (10.2%) with CD -Higher BMI and self-directedness were predictors of greater risk of ED -Higher ill-being and lower joy in life were reported by patients with CD with ED compared with patients without EDs, even when controlling for age and depression levels -No differences between patients (with CD) with and without EDs in coping strategies were found -Higher BMI and lower self-directedness were linked to higher risk of ED in CD Babio (2018) [40] Cross-sectional CD, n = 98 (mean age 15 years, 60% females) Controls, n = 98 (mean age 15 years, 60% females)      A summary of results of studies evaluating the association between CD and the occurrence or presence of psychiatric disorders is presented in Table 4.
The prevalence rates of depression or depressive symptomatology were significantly higher in patients with CD compared to controls in the majority of the published studies except for two [56,60]. Nevertheless, significant variability in the point-prevalence of depression or depressive symptomatology exists, ranging from 14% to 68.7% [53,56,57,59,60,63]. In a meta-analysis conducted by Smith et al. [61], depression was shown to be more common and severe in CD than in healthy adults, but not compared to patients with other medical conditions. Comorbid illnesses, including type I diabetes mellitus or subclinical thyroid disease, and stress were associated with the presence of depressive symptomatology in CD [57,60]. Increased severity of gastrointestinal symptoms in CD was linked to worsened depressive symptoms [75] which, in turn, led to poorer QOL compared to controls [63]. Although gluten-free diet (GFD) did not lead to any improvement in depressive symptoms in two longitudinal studies [52,55], a meta-analysis conducted by Sainsbury et al. [66] found a moderate association between poor adherence to GFD and greater depressive symptoms. With respect to post-partum depression, it was assessed in a single study in which it turned out to be significantly more prevalent in women with CD compared to controls (41% vs. 11%, p < 0.01) [65]. Table 4. Summary of outcomes evaluating the association between gluten-related disorders and psychiatric disorders in adults.

Author (Year) Design Sample Size and Demographic Characteristics Summary of Outcomes Associated Factors with Psychiatric Comorbidities and other Relevant Information
Attention-Deficit/Hyperactivity Disorder Zelnik (2004) [45] Cross-sectional CD, n = 111 (mean age 20.     -A significant association between CD and dementia among the age group 60-69 was found, which was not present in the younger or older age groups -Increased risk of dementia was found in the first year following CD diagnosis Various psychiatric conditions Fera (2003) [74] Cross-sectional CD, n = 100 (mean age 40 years, 75% females) DM, n = 100 (mean age 53 years, 74% females) HC, n = 100 (mean age 41 years, 68% females) -CD, prevalence of OCD 28%, depressive disorder/dysthymia 19% -DM, prevalence of OCD 0%, depressive disorder/dysthymia 10% HC, anxiety and depression in 10% of subjects -QOL was poorer in both CD and diabetic patients than in healthy controls and significantly correlated with anxiety Sainsbury (2013) [75] Study 1: cross-sectional n = 390 (mean age 44 years, 82.8% females) -Severe gastrointestinal symptoms at CD diagnosis were associated with: increased depression (r = 0.28, p <0.001), anxiety (r = 0.29, p <0.001), stress (r = 0.28, p <0.001), eating disorder (r = 0.15, p <0.01), and emotion-oriented coping (r = 0.17, p <0.01) -Poorer QOL was significantly associated with a greater number and longer duration of CD symptoms prior to diagnosis -Higher number of symptoms was associated with poorer QOL -There were no gender differences in QOL, although females reported a greater number of symptoms -More severe gastrointestinal symptoms at diagnosis were also associated with increased psychological manifestations  In two studies conducted by the same research group, the prevalence of state anxiety in patients with CD was substantially higher than in controls (62.5% vs. 31.3%, and 71.4% vs. 23.7%), although the difference was statistically significant in only one study [53,55]. Generalized anxiety disorder diagnosis in CD was not shown to be prevalent in CD compared to controls [57] and the overall prevalence of anxiety was not significantly higher compared to healthy adults in the meta-analysis conducted by Smith et al. [61]. The prevalence of social phobia in CD reached 70% in one cross-sectional study [59]; however, its lifetime prevalence in another study was only 8.3% [57]. Bipolar disorder and panic disorders were significantly more prevalent in patients with CD [57,58,63].
Three studies assessed the prevalence and risk of eating disorders in CD. Their prevalence was significantly higher in adults with CD compared to healthy controls as demonstrated via the elevated scores on the different assessment scales employed in two cross-sectional studies [67,68]. Elevated Eating Attitudes Test scores were seen in around 16% of patients with CD in both studies, whereas elevated Binge Eating Scale scores were only elevated in one study with 19.7% of adults with CD reporting high scores [68]. Moreover, severe gastrointestinal symptoms were linked to greater risk of eating disorders [75]. In the register-based cohort and case-control study conducted by Marild et al. [69], the likelihood of developing anorexia nervosa was significantly higher in women with CD (HR = 1.46, 95% CI: 1.08-1.98) and the likelihood was highest in women with normal mucosa and positive serology (HR = 2.45, 95% CI: 1. 1-5.45).
While patients with CD were less likely to experience alcohol-related disorders [72], their risk of developing dementia was significantly higher as shown in a population-based cohort study [70,73]. The likelihood of developing poor sleep in CD based on the use of hypnotics was significantly elevated compared to controls HR = 1.36, 95% CI: 1.3-1.41) in the population-based case-control study conducted by Marild et al. [70]. On the other hand, sleep difficulty as measured with the Patient Health Questionnaire did not differ significantly between adults with CD and controls (37.3% vs. 27.4%, p = 0.15) in a population-based cross-sectional study [76].
While gender did not seem to affect the prevalence rates of ADHD in CD patients [45], males with CD were less likely to experience poor sleep problems [70] or subsequent anorexia nervosa [69] and they tended to score higher on the different Psychological General Well-being Index domains [71]. Conflicting results concerning the effect of the CD onset time on psychological symptomatology were obtained; on one hand, earlier onset of CD symptoms was linked to higher prevalence of major depressive disorder in one study [57] and on the other hand, depressive symptomatology scores did not differentiate between childhood or adulthood diagnosis of CD [54]. Finally, severe gastrointestinal symptomatology significantly correlated with increased psychological manifestations [68,75].
The risk of schizophrenia in patients with CD was assessed in three studies [47][48][49]. While one population-based case-control study showed no increased risk of schizophrenia in CD (OR = 0.75, 95% CI: 0.4-1.4) [47], its risk was shown to be significantly elevated in a population-based cohort study (Incidence rate ratio = 2.11, 95% CI: 1.1-3.6) and a cross-sectional study (adjusted incidence rate = 3.6, 95% CI: 1.2-10.6) [48,49]. In a meta-analysis including four studies, an increased risk of schizophrenia among patients with CD was found (OR = 2.03, 95% CI: 1.45-2.86) [50]. With respect to autistic spectrum disorders, its risk in a population-based cohort of CD appeared to be increased with the highest risk being present in patients with normal mucosa and positive serologic findings (HR = 3.09, 95% CI: 1.99-4.8) [46].
ADHD was assessed in one cross-sectional study that reported an increased prevalence of this disorder in adults with CD compared to controls (20.7% vs. 10.5%, p <0.01) [45]. The overall psychological status in adults with CD was evaluated in one study whereby no difference in the total Psychological General Well-Being Index was found between CD and controls [71].
In the two studies that evaluated the effects of gluten ingestion in adults with NCGS [62,64], significant worsening of depressive symptomatology [64] and increase in the depression subscale scores of Spielberger State Trait Personality Inventory [62] were reported.

Discussion
Our current review of the literature revealed the existence of an association between CD and other gluten-related disorders with psychiatric disorders across different age groups. CD is primarily an autoimmune disorder that is characterized by villous atrophy of the intestinal mucosa along with intraepithelial lymphocytosis and crypt hyperplasia [77]. Nevertheless, a major shift in clinical presentation with extraintestinal manifestations becoming more prevalent than classical gastrointestinal symptoms has been suggested [78]. The reviewed data demonstrate that a wide range of psychiatric disorders have been investigated in CD and NCGS including autism spectrum disorders, schizophrenia, attention-deficit disorder, depression and mood disorders, anxiety disorders, bipolar disorder, feeding and eating disorders, sleep disorders, substance-related and addictive disorders and neurocognitive disorders.
Most of the cross-sectional studies in the pediatric population did not find any significant differences in the point prevalence of depression or anxiety disorders [32,33,[35][36][37], however, these studies had several methodological limitations which mainly included small sample size (ranging between 29 and 42 children with CD), the lack of specialized psychiatric clinical assessment, and the absence of adequate blinding measures to limit assessment bias. On the other hand, two population-based cohort studies including >9000 children each provided evidence for an increased likelihood of occurrence of depression and anxiety disorders in patients with CD [34,44]. In the cohort study conducted by Ludvigsson et al. [34], it was shown that adults and children with CD are at increased risk of being diagnosed with depression but not bipolar disorder later in life (i.e., during adulthood for children diagnosed with CD), whereas in the study conducted by Butwicka et al. [44], CD was identified as a risk factor for mood disorders, anxiety disorders, eating disorders, behavioral disorders, ADHD, ASD, and intellectual disability diagnosed prior to 18 years of age. Although the analyses in the two previous cohort studies were controlled for children's age, stratified analyses to identify the likelihood of occurrence of specific psychiatric disorders across the different age groups are worth evaluation taking into consideration the variation in clinical presentation across the developmental span between 0 and 15 years of age [79].
In adults, the point-prevalence of depression was significantly higher in patients with CD in the majority of published studies. These findings were ascertained by a population-based cohort study in which the HR of depression (in participants ≥16 years at diagnosis) was two folds higher than controls [34] and by a meta-analysis in which depression was shown to be more common and severe in CD than in healthy adults with an overall effect size of 0.97 [61]. A comprehensive review, evaluating the comorbidity of depression and anxiety in CD, concluded that these disorders are common disorders among patients with CD and contribute to a poorer quality of life [24]. Nevertheless, the lack of differences in the prevalence of depression when compared to patients with other physical disorders [61] raises a question about the existence of a specific underlying pathophysiological mechanism in patients with CD or whether depression represents a non-specific disorder affected by physical and psychosocial distress. The association between chronic medical diseases and depression is well-known and many different causes, including both genetic predisposition and environmental factors have been shown to be involved [80][81][82]. This association is frequently bidirectional, as the presence of physical illness often worsens the affective disorder and vice versa [81]. The current information relative to depression in patients with CD does not allow, at the present time, to ascertain the exact relationship and the predisposing factors involved between CD or NCGS and depressive symptomatology.
The association between CD and eating disorders has been investigated in a limited number of studies. Current findings reveal an elevated prevalence of eating disorders in CD among both children and adults with CD [39,40,44,[67][68][69]. These disorders encompassed anorexia nervosa, bulimia nervosa and binge eating. Poor dietary management can occur as a result of physical dissatisfaction, which is not uncommon in patients with CD [83]. Moreover, evidence from the current literature suggests that young adults with chronic illnesses that require dietary modification are at higher risk of developing pathological eating practices [39]. The elevated lifetime comorbidity between depression and eating disorders [84] could be another explanatory mechanism of increased prevalence of eating disorders in CD patients who are more prone to developing depressive symptomatology.
Concerning psychotic disorders, the current evidence provided by solely two population-based cohort studies does not support the presence of an association between these disorders and CD in children [31,44]. However, children and young adults (≥16 years of age) with CD were 1.8-fold more likely to experience non-schizophrenic non-affective psychosis [31]. The authors of the latter cohort study yet did not rule out the presence of a potential association between CD and schizophrenia as the risk of the latter disorder was high despite the low number of individuals with schizophrenia. These findings were similar to another population-based case-control study conducted in adults in which no evidence of an increased risk of schizophrenia in CD was found [47]. In contrast, Benros et al. [49] demonstrated an increased incidence of schizophrenia in patients with prior CD in their population-based cohort study. Furthermore, Eaton et al. [48] showed also 3.8-fold increase in incidence rates of prior CD diagnosis in subjects with schizophrenia. However, in the latter study, data on parents' celiac status were also included in their analysis which might have led to biased findings. A meta-analysis including four studies demonstrated the presence of an increased risk of schizophrenia among patients with CD [50]. We believe that the pooled-effect estimate in the previous meta-analysis could be biased because their pooled analysis on one hand missed the negative findings reported by West et al. [47] and on the other hand included the findings of a study in which the prevalence of CD in patients with schizophrenia was investigated [85]. The objectives and outcome measures of the latter study [85] did not match the principal objective of the meta-analysis whereby the authors investigated the prevalence of autoimmune diseases (including CD) in patients with schizophrenia and not the other way around [85]. The association between CD and gluten-related disorders with schizophrenia has been under investigation for more than five decades but most studies evaluated the prevalence or risk of gluten-related disorders in patients already diagnosed with schizophrenia [86]. Current evidence suggests a two-fold increase in the prevalence of CD in schizophrenia patients [87] and an association between gluten ingestion and exacerbation of schizophrenia symptoms [88]; nonetheless, these findings are highly inconsistent across different clinical, immunological, and epidemiological studies [86] and have not been replicated in patients presenting with CD.
The underlying mechanisms behind the association between CD and psychiatric disorders are not well-known. Nevertheless, several potential biological and psychosocial explanations have been suggested: (i) Several psychiatric disorders such as depression, anxiety, and ADHD, among others have been linked to certain nutritional and vitamin deficiencies [89] and it is well-known that patients with CD often suffer from malnutrition prior to diagnosis or as a result of dietary non-compliance [90]; (ii) The immune-mediated processes underlying CD have been postulated as potential causative factors of the different psychiatric manifestations taking into consideration the involvement of chronic immune system activation in the etiology of various psychiatric conditions [91]; (iii) The bidirectional communication between the gastrointestinal tract [92] and the brain may suggest that alterations in the intestinal permeability, which is cardinal manifestation in CD [93], could be eventually involved in the pathophysiology of psychiatric manifestations in patients with CD; (iv) Finally, psychosocial aspects commonly seen in CD could place this population at an increased risk of developing psychiatric disorders, for instance, the introduction of GFD is associated with radical changes in daily life activities, eating habits and lifestyle which could be particularly stressful and difficult to accept [43,94]. In addition, effective adherence to GFD entails greater burden manifested via increased daily expenditure on more expensive products, social isolation and constant fear about dietary mistakes [95].
The studies included in this review provided limited data on potential factors associated with psychiatric comorbidity in patients with CD. Bearing in mind this limitation, none of the demographic factors has been consistently associated with the presence or occurrence of psychiatric comorbidities and the role of ethnicity in this context has not been studied. Regarding clinical factors, only severity of CD symptoms appears to be associated with the presence and/or severity of psychiatric disorders [33,51,52,68,75]. In this regard, the significant positive association between increased severity of gastrointestinal symptoms and worsening of psychiatric manifestations [75] and QOL [63] in CD indirectly demonstrates the importance of adherence to GFD. Nevertheless, few studies have documented the beneficial effects of GFD on psychiatric manifestations in patients with CD [27,66], with the majority of these studies suffering from several methodological flaws limiting our capacity of reaching definitive conclusions supporting the role of GFD in this context.
Only two studies in patients with NCGS supported the association between this relatively new entity and depressive symptomology [62,64]. It has not been until recently that standardized diagnostic criteria for NCGS were established [19], which might explain the limited number of studies investigating psychiatric comorbidities in NCGS. In our current review search, we could not find any study that investigated psychiatric comorbidities in patients with WA.
Limitations of our current review are essentially derived from the limited quality of the majority of the studies that have investigated psychiatric disorders in CD. Most of these studies are of cross-sectional design which does not allow establishing causal relationships and are of small sample size, whereas very few population-based studies have been published.
Evaluating psychiatric comorbidities in different age groups adds strength to our current review since up to the current date, none of the previous reviews had evaluated the evidence of psychiatric disorders in children and adults with CD separately. Interestingly, according to our review, the presence of CD in childhood seems to be associated with an increased risk of developing psychiatric disorders later during adulthood, but not with an increased prevalence of these disorders during childhood.

Conclusions
Our current comprehensive review ascertains the presence of an association between CD and psychiatric disorders with varying grades of evidence from one condition to another. In our view, there is enough evidence supporting an association of CD with depression and, to a lesser extent, with eating disorders. Some studies also point out to an association between CD and panic disorder, autism and ADHD, but the evidence is limited, and these potential associations should be further investigated. Finally, the data regarding the association of CD with schizophrenia or other anxiety disorders is conflicting. Overall, psychiatric symptomatology which could be perceived as part of the atypical manifestations of this chronic condition are linked to significant distortion in quality of life and moderately increased risk of suicide [96] and thus warrants further attention. Therefore, gastroenterologists and other healthcare professionals involved in the management of patients with CD should be aware of the increased risk of psychiatric disorders in these patients. Thus, routine surveillance of potential psychiatric manifestations, especially anxiety and/or depressive symptomatology that seem to be the most common forms of disturbances, should be carried out by the attending physician in order to refer the patient to the mental health services if necessary.

Conflicts of Interest:
The authors declare no conflict of interest.