The effects of probiotics on gastrointestinal symptoms and microbiota in patients with celiac disease: a systematic review and meta-analysis on clinical trials

Gluten-free diet (GFD) is the most effective method to manage celiac disease (CD). Many patients do not reach the complete symptom alleviation, even by strict GFD. Recent studies have reported inconsistent results regarding the beneficial benefits of taking probiotics. Therefore, we aimed to evaluate the effects of probiotics on gastrointestinal (GI) symptoms and the possible underlying causes in CD and celiac disease autoimmunity (CDA) patients. Databases, including PubMed, Scopus, Embase, Web of Science and Google Scholar, were searched for clinical trials published until July 2022 about assessing the effects of probiotics or synbiotics on CD or CDA patients. We collected data on GI symptoms, CD markers, inflammatory and immune responses, adverse events, and gut microbiota. A random effect meta-analysis was used to estimate the pooled standardized mean difference (SMD) and confidence interval (CI). We screened 7234 articles, of which 14 were included in the qualitative analysis and 5 in the quantitative analysis. Probiotics might alleviate GI symptoms, especially in the highly symptomatic patients, and improve immune response in CD and CDA patients. Results of the meta-analysis showed that probiotics increased the abundance of Bifidobacterium (SMD: 0.72, 95%CI (0.13, 1.30) and Lactobacillus (SMD: 0.49, 95%CI (0.18, 0.80) as compared with placebo. Probiotics did not increase the adverse events compared to the placebo. Probiotics might alleviate GI symptoms and immune response and improve dysbiosis in CD and CDA patients. However, high-quality clinical trials are needed to increase the level of evidence. Also, the most suitable combination of probiotics is yet to find.


Introduction
Celiac disease (CD) is a gastrointestinal disease affecting approximately 1% of the global population [1].It is substantially more common in children [1].CD is an autoimmune disorder in which gluten-derived peptides induce small intestine inflammation, villous atrophy, and crypt hyperplasia [2].Uncontrolled CD could lead to the disrupted calcium absorption, osteoporosis, ulcerative jejunoileitis and possible complications of intestinal ulceration, including hemorrhage, perforation, or obstruction and in the most extreme cases malignancy [3].
Gluten is mainly composed of glutenin and prolamins, which contain epitopes presented by human leukocyte antigen-DQ2 (HLA-DQ2) or HLA-DQ8, inducing a CD4 + T-lymphocyte-mediated immune response.Gliadin is a prolamin which interacts with intestinal epithelial cells in patients susceptible to CD and disassembles interenterocyte tight junctions (TJ), leading to increased gut permeability [3].
Until now, a lifelong strict gluten-free diet (GFD) is the most effective method to manage celiac disease [4].However, even products specified for these patients may contain a bit of gluten and a rigorous GFD is not still enough to completely alleviate the symptoms in many patients who have refractory type.Meanwhile, GFD imposes significant limitations on one's diet and has been reported to adversely affect body composition and increase the risk of obesity in children [5].In addition to GFD, the adjuvant therapies are needed.
The gastrointestinal tract contains a complex community of bacteria called microbiota and produces thousands of metabolites as intermediate signals.Changes in microbiota composition are associated with various chronic inflammatory diseases, as it plays a crucial role in the mucosal, innate, and adaptive immune functions.Some bacterial species in gut microbiota, especially Bifidobacterium and Lactobacillus, produce endopeptidases, digesting gluten epitopes and reducing their immunogenicity.Gut dysbiosis could disrupt intestinal TJ, leading to increased intestinal permeability and altering anti-inflammatory and pro-inflammatory balance contributing to the pathogenesis of CD by the upregulation of T-helper type 1 (Th1) [6].The gut microbiota of celiac patients shows a certain dysbiosis compared to healthy individuals [7].Evidence showed that the abundance of Lactobacillus, Bifidobacterium, Clostridium histolyticum, Clostridium lituseburense, and Faecalibacterium prausnitzii decreased, while the frequency of Bacteroides, Escherichia coli and ratio of Gram(−) to Gram( +) bacteria increased in the fecal microbiota of CD patients [8][9][10].
According to in vitro studies, probiotics containing Lactobacillus or Bifidobacterium alleviate the effects of gluten and gluten-derived peptides on the intestinal cell cultures [11].In addition, Bifidobacterium counteracts increased intestinal permeability [12], and Lactobacilli can detoxify gliadin and immunogenic products of Pseudomonas aeruginosa proteases which reduce the immunogenicity of gluten [9].So, probiotics could be an effective adjuvant therapy for CD.Some human studies reported the beneficial benefits of probiotic administration in CD patients.However, others did not report any benefits, and the latest meta-analysis was inconclusive [13].Therefore, this systematic review and meta-analysis aimed to evaluate the effect of probiotic administration on laboratory and clinical findings and fecal microbiota composition in CD and celiac disease autoimmunity (CDA) patients.

Materials and methods
We reported the study selection process following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2009 Checklist).In addition, we have registered the protocol for this systematic review on PROSPERO (CRD42022300447).
The reference lists of all relevant studies were checked for any potentially eligible items which had not been included.First, search results were all added to EndNote X9.3.3 reference manager software, and then, duplicates were removed from the list by "find duplicates" of the EndNote and manually.

Study selection
All eligible clinical trials assessing the effects of probiotics, prebiotics, or synbiotics on CD or CDA patients and

Data extraction
Two independent researchers extracted the data into the "Data Extraction Form" generated by Microsoft Excel as follows: First author's name, publication year, country, study design, run-in period, treatment duration, follow-up duration, sample size, age, sex, study population, diagnosis method, disease duration, prior gluten-free diet (GFD) duration, strain and dose of probiotic, frequency of probiotic administration, and main findings, including gastrointestinal (GI) symptom scores, quality of life (QoL) scores, adverse events, inflammatory markers, serologic markers of CD, fecal microbiota composition, leukocyte composition, urine and stool gluten immunogenic peptides (GIP), intestinal permeability, duodenal biopsy findings, fecal IgA, and urinary D-lactate.

Quality assessment
Two independent team members independently assessed the quality of included studies by the Cochrane risk of bias assessment tool (RoB 2) [14].RoB 2 consists of seven domains as follows: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessments, incomplete outcome data, selective reporting, and other biases.Each domain was assessed and reported as high risk of bias, unclear, or low risk of bias, considering its impact on outcomes.Any disagreement was addressed through discussion with a third party.

Statistical analysis
The I 2 and Cochran's Q tests were used to assess the heterogeneity across studies.If the P value of heterogeneity was less than 0.1, a random-effect model was used; otherwise, a fixed model was employed to pool the effect sizes.Standardized mean differences (SMD) and risk ratios within studies were extracted and pooled for meta-analysis.Meta-analysis was performed only if an outcome had at least two records with identical measurements.Sub-group (sensitivity) analysis was performed according to the type of probiotic supplementation and being on GFD.Egger's test for overall intestinal bacterial flora change level was used to determine publication bias.If publication bias was seen, trim fill analysis was performed to impute the missing data.STATA version 17 (Stata Corporation, College Station, Texas, USA) was used to analyze the data.
In terms of following GFD before enrolling in the studies, three out of all CD [16,17,20] and CDA trials [25][26][27] included the participants who were not on GFD prior to the beginning of the study.In other studies, the duration of GFD before the enrollment varies from three months to fifteen years.In addition, three studies [6,15,21] included patients with refractory CD.Contrary to other studies, in Smecuol et al. 's, participants were still on GCD during the study [20].

Qualitative findings
Qualitative results of this systematic review are summarized in four domains as follows.

Gastrointestinal symptoms
Francavilla et al. [6] and Smecuol et al. [21], involving 129 patients, studied adult CD patients who were symptomatic despite following GFD at least for two years (refractory CD).Francavilla et al. [6] conducted the only study assessing gastrointestinal symptoms when administering a mixture of Bifidobacterium and Lactobacillus, while Smecuol et al. [21] only administered Bifidobacterium as probiotic.Francavilla et al. [6] reported a significant reduction in irritable bowel syndrome severity scoring system (IBS-SSS) and gastrointestinal symptom rating scale (GSRS) in the probiotic arm in comparison to placebo after a six-week treatment period.Moreover, the number of participants with at least 50% reduction in IBS-SSS was significantly higher in the probiotic arm.In their study, Smecuol et al. [21] did not report any significant differences between probiotic and placebo groups in terms of GSRS and celiac symptom index (CSI) changes.However, in the study, a sub-analysis including only highly symptomatic patients revealed a significantly higher reduction in CSI in the probiotic arm compared to the placebo.Smecuol et al. [20], who conducted the only study on adult CD patients consuming GCD during probiotic treatment, reported no differences between groups in terms of GSRS, diarrhea, distension, gas, and abdominal pain after three weeks of Bifidobacterium treatment.However, indigestion and constipation were significantly alleviated in the probiotic group.
Olivares et al. [17] and Ali et al. [24] prescribed Bifidobacterium in children with CD.Olivares et al. [17] found that height percentile was significantly increased in the probiotic group compared to the placebo group after 12 weeks of treatment.The increased weight did not reach statistical significance.No significant difference was reported regarding diarrhea, constipation, abdominal pain, and vomiting after treatment.Ali et al. [24] reported a significant higher reduction in stool frequency in probiotic arm compared to placebo after four weeks of treatment.peptide-immunoglobulin A (DGP-IgA) were lower in the probiotic group; however, this change did not reach statistical significance [20].Participants were children with CDA in studies conducted by Håkansson et al., and Demiroren et al.Former provided Lactobacillus and maltodextrin for six months, while the latter administered synbiotic containing both Bifidobacteria and Lactobacillus and fructooligosaccharide, and vitamins A, B 1 , B 2 , and B 6 for 20 days [26,27].tTG-IgA and tTG-IgG were similar in the probiotic and placebo groups for Håkansson et al.'s [26].In the contrary, Demiroren et al. reported a significant reduction in tTG-IgA in both groups.Additionally, the reduction was significantly more pronounced in the synbiotic group [27].

Inflammatory and immune responses and intestinal permeability
Three RCTs assessed the effects of probiotics on TNF-α in the pediatric population [17,18,23].Only Bifidobacterium was administered for three months in all three RCTs.Prior GFD was similar for Primec et al. and Klemenak et al. ranging from half to 15 years [18,23], whereas Olivares et al. included participants on GCD prior to study [17].According to Primec et al., TNF-α had a positive correlation with 'Verrucomicrobia' and a negative correlation with 'Parcubacteria' at baseline, and a positive correlation with Firmicutes after the probiotic treatment [18].TNF-α was not significantly different between groups in studies performed by Klemenak et al. and Olivares et al.However, it was reduced in both studies [17,23].This reduction was more prominent, but was not statistically significant in patients who were on GFD for less than a year prior to the study [23].
Four articles studied immunologic response, two on adult CD patients and two on children, of which one studied CDA [16,17,20,26].Pinto-Sa ńchez et al. compared a threeweek course of Bifidobacterium supplementation and one year of GFD prescription.Paneth cell count and expression of human α-defensin-5 (HD-5) significantly reduced independent of GFD.In contrary, a significant reduction in macrophage count was only reported after one year of GFD [16].In another three-week Bifidobacterium trial, macrophage inflammatory protein-1-β (MIP-1β) significantly increased in the probiotic group.Changes in other cytokines, including macrophage chemoattractant protein-1 (MCP-1), were insignificant [20].Twelve weeks of Bifidobacterium significantly decreased mature T cells (CD45 + CD3 +) in CD children.It also increased regulatory T cells (CD45 + CD4 + FoxP3 +); however, it did not reach statistical significance.Cytokines were similar to the placebo group [17].In their study, Håkansson et al. assessed T helper cells after six months of Lactobacillus and maltodextrin.Memory T helper cells and naive T helper cells increased and decreased significantly less than the placebo group, respectively [26].
One RCT, assessing intestinal permeability after three weeks of Bifidobacterium ingestion using a urinary lactulose/mannitol ratio, showed no difference between groups [20].

Gut microbiota
Fecal microbiota was assessed in seven CD studies [6, 15, 17-19, 21, 22] and in one CDA study [25] (Table 3).Four CD studies and one CDA study assessed Lactobacillus and Bifidobacterium abundance, so we conducted a meta-analysis [6,15,17,19,25].In the study conducted by Smecuol et al. [21], symptoms of highly symptomatic patients were significantly lower after a three-week course of Bifidobacterium compared to a placebo.Accordingly, highly symptomatic patients had 6.4 times higher B. longum compared to patients with mild symptoms [21].Bifidobacterium administration significantly increased Lactobacillus abundance to the extent that it resembled healthy controls in one study [19].In a study conducted on CDA, six-month Lactobacillus and maltodextrin supplement significantly increased Lactobacillus genus and Lactobacillaceae family compared to placebo [25].No significant difference in the abundance of Bifidobacterium and Lactobacillus was detected between probiotic and placebo groups in three studies, of which two studied refractory CD and one assessed pediatric population [6,15,17].

Meta-regression
To assess the possible heterogeneity causes seen across study reports on the effect of probiotic supplementation on intestinal bacterial flora, meta-regression was performed for female/male ratio in treatment group, female:/male ratio in the control group and treatment duration; neither of which were the underlying cause of the heterogeneity among studies (the β (SE) for variables, respectively, are as follow: male ratio in treatment group: 0.45 (3.52), female/male ratio in the control group: − 1.64 (2.58), treatment duration: − 0.05 (0.06)).

Discussion
In this systematic review and meta-analysis, we evaluated the effects of probiotic supplementation in CD and CDA patients.The results of this study showed that probiotics might alleviate CD symptoms, especially in the highly symptomatic patients.On the other hand, the meta-analysis of symptoms did not show a significant reduction in diarrhea, GI symptoms, or GSRS.Nonetheless, all studies assessing GI symptoms were not included in the meta-analysis, as questionnaires used in some studies were different and incomparable to others.Additionally, probiotics may have an immunomodulatory effect in both CD and CDA patients and may also improve fecal microbial composition.These could be the potential mediating mechanism of the effect of probiotics on CD.
In two studies, CD patients were not on a GFD while receiving probiotics.They did not report any significant changes in symptoms.Pinto-Sánchez et al.only investigated intestinal immune response, and the other did not report any significant improvement in symptoms [16,20].Consequently, probiotics do not seem very promising as a substitute for GFD.However, studies on probiotics without GFD could assess the sole effect of probiotics.Therefore, it is of benefit to the immunologic and physiologic assessment.
Serum tTG antibodies are associated with the pathologic results in patients as risk of CD [28].However, IgA-tTG and IgG-tTG levels remained unchanged in probiotic and placebo CDA patients after the 6-month consumption of Lactobacillus and maltodextrin [26].On the contrary, another study conducted on CDA with a shorter duration of treatment (20 days) reported a decrease in anti-tTG-IgA after the combination of Bifidobacterium and Lactobacillus and fructooligosaccharide, and vitamins A, B 1 , B 2 , and B 6 administration [27].
There was a significant difference between the two studies reporting TNF-α in the same scale.Both studies included pediatric population.While the study performed by Klemenak et al. had lower TNF-α levels as compared to healthy controls and included symptom-free CD patients who were on GFD, the study performed by Olivares included newly diagnosed patients.This could explain this difference.Both studies detected a reduction in TNF-α after probiotic administration; however, neither reached statistical significance [17,23].
Three weeks of Bifidobacterium supplementation in CD patients on GCD significantly reduced A-HD-5 and PC compared to patients following GFD for one year [16].Duodenal Paneth cells (PC) contribute to intestinal homeostasis and gut microbiota composition.Human α defensin-5 (A-HD-5) is an antimicrobial peptide secreted by PCs, and its expression increases in CD due to PC metaplasia.The effect of probiotic on PC may contribute to its potential underlying mechanism in CD patients.However, there is little certainty as only one low-quality study evaluated it.
After a 6-month Lactobacillus and maltodextrin supplementation, the immune response observed in the synbiotic group was modulated, while the placebo group resembled the one in active CD patients [26].All studies assessing immunologic response with different modalities reported some extent of immunomodulation, although in the study conducted by Smacuol et al., only macrophage inflammatory protein-1β (MIP-1β) significantly increased in patients receiving probiotics and changes in other cytokines, chemokines, and pro-inflammatory status did not reach statistical significance [20].MIP-1β is a potent chemoattractant, and this increase may be due to down-regulation of chemokine receptors after probiotic administration.However, low sample size of the study of Smecuol et al. [20] might be the reason for not reaching statistical significance as opposed to other studies.
Intestinal permeability is another suggested physiopathology in CD.Only one study evaluated intestinal permeability by lactulose/mannitol ratio, which did not reach statistical significance [20].It should be noted that this high-quality study had a limited sample size, and the lactulose/mannitol method lacks adequate specificity and sensitivity [29].Therefore, further studies are needed in this field with larger sample sizes and more accurate measurement methods.
Certain types of dysbiosis have been associated with treated and untreated CD patients [30].The causality of this association is yet to be studied.GFD partially restores normal gut microbiota; however, it might reduce the diversity of Bifidobacterium and Lactobacillus [31].Based on our findings, probiotics increased fecal Bifidobacterium back to a non-significant difference compared to healthy subjects.Thus, the fecal microbiota of CD patients treated with probiotics is similar to healthy subjects regarding Bifidobacterium.Our meta-analysis revealed that probiotics also increase the abundance of Lactobacillus; however, it is still different from healthy controls.
Only one study investigated fecal microbiota after mere Lactobacillus administration, which was combined with maltodextrin and studied CDA.This study reported an increase in Lactobacillus abundance without significant changes in Bifidobacterium [25].In line with our findings, the previous systematic review [13] reported a significant increase in Bifidobacterium abundance.Although, unlike this study, it did not report a significant increase in Lactobacillus abundance.Furthermore, based on our sub-group analysis, the abundance of Bifidobacterium was significantly higher in patients on GFD who received combined (Bifidobacterium and Lactobacillus) supplementation.This emphasizes the synergism of GFD and probiotics on fecal microbiota composition.Meanwhile, a better outcome was reported with combined probiotics.
Regarding the probiotic strains, B. breve and B. infantis were most used, and also, they were included in probiotic supplements of four out of three studies that used singlestrain probiotics.B. bifidum, Bifidobacterium animalis, Bifidobacterium lactis, and B. longum were also used in studies.However, only B. breve, B. infantis, and B. longum were strains which could be detected in human feces [11].Singlestrain studies on B. breve did not assess symptoms, while one out of B. infantis studies yielded improved immunity and another reported alleviated symptoms in highly symptomatic patients [16,21].All Lactobacillus strains used could be detected in intestinal tract except L. paracasei which was only included in the probiotic supplement of one study [11].

Limitations and strengths
In this study, we also included patients with suspected CD based on elevated biomarkers, but did not have histopathological diagnosis of CD, known as CDA.Furthermore, we also included studies conducted on synbiotics.However, there was a great heterogeneity in the study populations and interventions, including the prescribed period of GFD before the study, concomitant GFD during the study, the symptoms' intensity at recruitment, dose and species of probiotics, intervention period, and questionnaires used to evaluate symptoms and also only a few studies investigated the effects of synbiotics on CDA and no studies assessed their effect on CD.Additionally, gut microbiota significantly varied in different geographic regions and available studies lacked sufficient geographic distribution.Therefore, findings should be interpreted with caution.Future studies should consider longer follow-up duration, duodenal biopsy findings, and larger sample sizes.

Conclusion
Although some studies reported improvements in GI symptoms after using probiotics, they did not reach statistical significance in the pooled analysis.However, probiotics and synbiotics might modulate immunologic response and might contribute to the restoration of dysbiosis in CD and CDA patients.Studies used different types of probiotics and synbiotics, but more studies are needed to find the most effective one.

Fig. 1
Fig. 1 PRISMA 2020 flow diagram for new systematic reviews which included searches of databases and registers only Smecuol et al., Håkansson et al., and Demiroren et al. measured celiac-specific markers [20, 26, 27].Smecuol et al. 2013 administered probiotics in adult CD patients for three weeks.The serum concentrations of tissue transglutaminase-immunoglobulin A (tTG-IgA) and deamidated gliadin

Fig. 2 Fig. 3
Fig. 2 The pooled effect of probiotic supplementation on intestinal bacterial flora

Fig. 4 Fig. 5
Fig.4 The pooled effect of probiotic supplementation on the intestinal bacterial abundance in leave-one-out analysis (the red line represents the overall SMD without omission, and the blue line represents null effect)

Table 2
Characteristics of included studies investigating the effect of probiotics and synbiotics on celiac

Table 2 (
b no significant difference c significantly higher in probiotic group

Table 3
Significant fecal findings after probiotic or synbiotic intake

Table 4
Quality assessment by the Cochrane risk of bias assessment tool (RoB 2)