Probiotic Interventions in Coeliac Disease: A Systematic Review with a Focus on Cardiovascular Risk

: People with coeliac disease (CD) have a higher risk of developing cardiovascular disease (CVD), potentially due to inflammation. Probiotics can influence CVD risk through several mechanisms including modifying inflammation. We performed a systematic review of probiotic interventions in people with CD. In total, 4 databases were systematically searched for studies published up to March 2023. All outcomes, inclusive of any cardiovascular risk factors, were collated and reported. We screened 8084 articles and 11 publications reporting on 7 RCTs and 2 non-RCTs met the inclusion criteria for qualitative analysis. In total, 1 RCT and both non-RCTs were considered to have a high risk of bias. There was large heterogeneity between the studies and adherence to a gluten-free diet was only measured in two studies. No specific outcomes related to cardiovascular risk were reported. Two studies reported a significant reduction on serum TNF-α in children over time after probiotic supplementation. One study reported no significant change in intestinal permeability over a 3-week intervention. Currently there is insufficient evidence to advocate a positive impact of probiotics on inflammation in CD, due, in part, to the limited data on adherence to the gluten-free diet and active disease.


Introduction
Coeliac disease (CD) is a chronic inflammatory bowel condition, triggered by an autoimmune response which is characterized by a permanent sensitivity to gluten in genetically predisposed people [1].The small intestines are primarily affected at the onset and duration of the disease, however, there are broad clinical manifestations usually presenting as both intestinal and extraintestinal symptoms.Gluten is the term used for storage proteins derived from cereal grains barley, rye, and wheat.The prevalence of CD is steadily increasing with approximately 1% of people affected in the world's population [2].With no cure, lifelong adherence to a gluten-free diet (GFD) (a combination of naturally gluten-free and processed foods comprising less than 20 ppm gluten) is recognized as the only effective treatment for CD [3].Adhering to a GFD can be very challenging; it requires knowledge, skills and modified behaviours to undertake the substantial changes to dietary habits [4].
The gut microbiota, which is a complex community of microorganisms, within people with CD differs compared with controls, whether this CD-associated dysbiosis is the consequence of the inflammation or a concurring causative factor remains to be established [5].Intestinal dysbiosis persists on the GFD, which itself can contribute due to the reduced fibre content [6,7].Dysbiosis has been associated with cardiovascular disease (CVD), through several mechanisms including microbiota-derived metabolites [8,9].People with CD are more likely to develop atherosclerosis, likely as a result of systemic inflammation and low high-density lipoprotein levels that may contribute to atherosclerosis [10,11].It has been reported that a third of patients with CD adhering to a GFD had concurrent non-alcoholic fatty liver disease (NAFLD), accounting for a three-fold increased risk compared with the general population [12].Conroy et al. (2023) report, analysis from UK Biobank data, people with CD have a higher risk of developing cardiovascular disease than did people with no coeliac disease [13].
Studies have suggested a GFD itself may also contribute to metabolic syndrome and hyperlipidaemia observed in people with CD [14][15][16][17][18], with significantly more saturated fat in GF compared with gluten containing counter parts [19].Based on findings from a large epidemiological study, Lebwohl et al. (2017) suggest the avoidance of gluten may result in reduced consumption of beneficial whole grains, which may affect cardiovascular risk [20].A systematic review of studies up to 2021, indicated there was insufficient evidence to link a GFD with CVD, however, there was considerable heterogeneity within the studies [21].
The gut microbiota has an essential role in regulating lipid absorption in the human intestine [9].Two studies reported participants consuming Lactobacillus plantarum ECGC 13110402 (LP LDL ® ) supplement, exhibited reductions in their LDL-cholesterol and increased HDL Cholesterol [22,23].Derosa et al. (2020) reported significant reductions were seen in blood pressure and level of LDL cholesterol after three months of LP LDL ® [24].They concluded that major risk factors for cardiovascular events such as high blood pressure and hypercholesterolemia could be managed with a probiotic [24].
Previous literature reviews on probiotics in coeliac disease have reported their potential role in inflammation and gastrointestinal symptoms [25,26].We performed a systematic review of the literature to evaluate the effects of probiotic interventions in people with CD to collate knowledge from all outcomes inclusive of any cardiovascular risk factors.

Results
Records identified through initial literature search were 9672 in total.This number was reduced to 8084 after duplicates were removed.As shown in the flow chart (Figure 1) we reviewed thirty-one full-text studies.Upon further review, we excluded twenty studies, and the remaining eleven publications met the inclusion criteria and were eligible for qualitative synthesis which is summarised in Table 1.

Characteristics of Studies
All studies included were in English and did not require translation.We performed
Four studies reported only faecal analysis from their participants [35,37,39,41], whilst one reported blood analyses only [29].Five studies conducted both faecal and blood analyses [27,28,30,34,36].One study reported a duodenal biopsy analysis only [32].The probiotic strains and dosages were the same for the three publications of the Slovenia cohort as well as the two Argentina studies but were different in all other studies.Six studies monitored adverse events [27][28][29][30]35,36] comparing probiotics to placebo no differences in the number of adverse events were reported.All except one RCT used either a placebo capsule or sachet with the one RCT [37] did not use a placebo for their controlled study.Every study excluded participants who were on any medications including antibiotics and immunosuppressants as they could have had effects on the outcome.Studies also excluded participants who had severe malnutrition, organic disorders, or suspicion of cancer.
Most reports, except four [27,28,32,37], participants had been diagnosed of coeliac disease by serologic testing and had been on a GFD for over 6 months before starting in the study.Persistent symptoms in participants were reported in three studies while on a GFD [30,35,36].Francavilla et al., reported participants had an additional diagnosis of irritable bowel syndrome [35].The studies undertaken by Ali et al. and Olivares et al. included newly diagnosed children who started a GFD for the study [28,37] and one study included newly diagnosed adults who were on a GFD with no other comorbidities [27].Only two studies monitored the adherence to a GFD [28,30] even though for all other studies it was a pre-requisite to be on GFD in order to be included in the study.Olivares et al. used the 72 h food diary whilst Harnett et al. used the 3-day food diary.

Risk of Bias Assessment
Five RCTs were considered to have a low risk of bias, three with some concerns and one with high risk of bias; details for each domain can be found within Table 1.Full details of subcategories within Supplementary Materials in Tables S1 and S2.Both non-RCTs were considered to have a high risk of bias; details for each domain can be found in Table 2.The domain 'measurement of outcomes' was analysed for two outcomes: For gastrointestinal symptoms, four reports were judged as LOW risk of bias [27,28,35,36] whilst one had a HIGH risk of bias [37].For inflammatory markers, three studies measured this outcome and were judged as LOW risk of bias [27,[29][30][31][32][33].[33] High risk of bias Green: Low risk of bias; Orange: Some concerns; Red: High risk of bias.Full details of subcategories within supplementary information.

Impact of Probiotic Intervention on Outcomes
The Effects of Probiotics on Gut Microbiota: Eight studies conducted faecal analysis but only four measured the microbiota composition and reported data for absolute abundance inclusive of bifidobacterial and lactobacilli species [28,30,31,35].One of the four studies reported significant increases in bifidobacteria species after probiotic intervention when compared to placebo and no significant increase was observed in the lactobacilli species.Only two studies out of the four had lactobacilli as part of their formulations, however, the counts of the lactobacilli species were not high.
The Effect of Probiotics on Cardiovascular Risk Factors: None of the studies reported outcomes on cardiovascular risk factors (Tables 3 and 4).Inflammatory markers: Two studies assessed the effects of probiotics on serum TNF-α in children (Tables 3 and 4).In the Klemenak et al. study, participants had been on a GFD for at least 6 months, they reported a significant reduction over time [29].Whereas patients in Olivares et al. study patients started the GFD simultaneously with probiotics, a potential trend towards a reduction in TNF-α was observed [28].One study reported intestinal permeability with the use of lactulose/mannitol ratio intestinal permeability test [27].No change in intestinal permeability was observed over the 3-week intervention.
The Effect of Probiotics on the Quality of Life (QoL): Only two studies reported outcomes for QoL (Tables 3 and 4).Francavilla et al. used the validated irritable bowel syndrome-QoL questionnaire tool (IBS-QoL) and Harnett et al. used the validated coeliac disease questionnaire tool (CDQ).In both studies, QoL scores after treatment did not differ between probiotic and placebo treatments [30,35].

Biochemical and microbial data:
Slight reduction in serum TNF-α for probiotic group (p = 0.067); but difference between both groups was not significant (p = 0.085).No significant differences found in both groups for serum IL-

Biochemical and microbial data:
In the probiotic and placebo group, TNF-α had a similar level as the healthy children.In CD patients, TNF-α had positive correlation with Verrucomicrobia and negative one with Parcubacteria.There was a high statistical significance between TNF-α and unclassified Bacteria group and positive correlation with TNF-alpha and unclassified Archaea.

Biochemical and microbial data:
The probiotic reduced the Paneth cells (PC) counts without the GFD.However, decreased macrophage counts (p = 0.02) were seen after 1 y GFD as well as further decreased in patients treated with the probiotic only.

Cardiovascular risk factors:
None reported.
The Effect of Probiotics on GI Symptoms: Five studies assessed the impact of probiotics on GI symptoms (Tables 3 and 4).Four studies reported improvements in GI symptoms after comparing probiotic to placebo [27,28,35,37].Smecuol et al. reported improvements in bloating, diarrhoea, abdominal pains, and abdominal distension from self-reported records [27].Olivares et al. also reported improvements in abdominal pains, vomiting, constipation and diarrhoea from self-reported records [28].Ali et al. reported significant improvements in stool frequency [37], though the study had a high risk of bias (Table 1).Francavilla et al. reported no difference in diarrhoea when using the Bristol Stool Form Scale and was considered a low risk of bias (Table 2) [35].Smecuol et al. and Francavilla et al. [27,35] used gastrointestinal symptom rating scale (GSRS) questionnaires and reported a significant reduction in GSRS scores compared with baseline values.Smecuol et al. [36] used the coeliac symptoms index (CSI) while Harnett et al. [30] used CDQ-GI to assess GI symptoms of participants.Probiotics when compared to placebo did not show significant changes in CSI scores.

Discussion
Data on the effect of probiotics in coeliac disease related gastrointestinal symptoms are limited, and data on CVD risks is even less studied.CVD is associated with a chronic state of low-grade inflammation, whereby the microbiota has an impact on maintaining the gut barrier function, critical for reducing the amount of pro-inflammatory bacterial byproducts that can cross into the bloodstream [8].Notably, the two studies that evaluated the effects of probiotics on serum TNF-α reported a reduction in levels [28,29], agrees with a meta-analysis of 12 RCTs in people with diabetes or prediabetes [8].Naseri et al. suggests the modulating effects of probiotics on TNF-α levels (and CRP levels) were more pronounced in patients with heightened inflammation [8].The two studies included European children with CD only, thus, it is clear further studies exploring the impact of probiotics on cardiovascular risk and inflammation are needed.Smecuol et al. [27] reported that a probiotic intervention had no significant impact on intestinal permeability, however, the intervention was only 3 weeks, and it is well known that it may take some time before mucosal integrity returns to baseline [38].Patients with CD continue to have gluten exposures which may account for persistence of histologic inflammation and residual symptoms.If patients are able to completely exclude all gluten, then is remains debated whether the inflammation remains [39].
Four studies reported improvements in GI symptoms after comparing probiotic to placebo [27,28,35,37], although there was a high risk of bias in the outcome measurement for Ali et al., study.Since completing the systematic literature search, Khorzoghi et al. reported improved fatigue scores with probiotic supplementation in a small cohort of adults with CD from Iran [40].A meta-analysis by Seiler et al. [25] concluded probiotics may improve gastrointestinal symptoms, however, the quality of evidence was low or very low; we confirm the recent publications have not been able to overcome this criticism.
A strength of our systematic review is collating all reported outcomes from RCTs and non RCTs relating to the impact of probiotic supplementation in adults and children with coeliac disease.We also bring attention to the reader that three publications were from the same intervention study in Slovenia [29,31,34], thus the body of evidence for probiotics in CD is small.This systematic review was conducted rigorously with pre-defined outcomes and a pre-registered protocol.We highlight that only two studies reported how they determined adherence to a GFD, it is well recognised that adhering to the gluten-free diet is challenging and many patients do not achieve full adherence [4].We recommend all future studies investigating probiotics in CD include an assessment for dietary adherence to the GFD as consumption of gluten would have a substantial impact on study outcomes such as intestinal permeability and inflammatory markers.
Four studies reported the composition of the microbiota with an increase in bifidobacterial evident after the probiotic interventions as would be expected, Mozafarybazargany et al. [26] reported similar findings from pooled analysis from a broad range of studies including people with CD and potential CD.The dysbiosis in treated CD maybe mainly food-induced, but in a small sample of children there is a demonstration of a constant alteration of some species that clustered in CD and not in healthy controls [41].The gut microbiota differs from person to person geographically; thus, the effectiveness of a probiotic may vary for persons in specific geographic areas [42].The review comprised of studies conducted in Europe (Italy, Spain and Slovenia), Australia, South America (Brazil and Argentina) and South Asia (Pakistan) with none from Africa, East Asia and North America; therefore, the results cannot be easily generalized.Knowing the impact of probiotics on the microbial ecosystem is essential; however, it must be emphasized that not all probiotics contain the same bacterial strains and can have different beneficial effects.
To conclude, our systematic review reveals there remains very sparce data on outcomes related to cardiovascular risk, and presently our knowledge of the area relies on data from other disease-based populations.There is emerging evidence of an impact on inflammation and gastrointestinal symptoms in adults and children with CD.

Methods
The systematic review was undertaken and reported in line with the guidelines of preferred reporting items for systematic reviews and meta-analyses [43].A population, intervention, comparison and outcomes (PICO) framework and literature search strategy criteria were developed-including the PICO table.This review has been registered on Prospero, the international prospective register of systematic reviews (CRD42023380433)

Inclusion and Exclusion
We adapted the PICO table strategy for the inclusion and exclusion criteria.Participants/persons, i.e., children and adults diagnosed with coeliac disease using standard procedures including specific serologic tests (anti-tissue transglutaminase, anti-deamidated gliadin peptide, or anti-endomysia antibodies) and confirmed by duodenal biopsy (Marsh score 3 or higher; or equivalent).Interventions included (any dietary intervention using probiotics with a duration of at least 21 days.For comparison, dietary interventions compared to placebo, or any or any other therapy were considered. Outcomes assessed were all reported outcomes, this included quality of life, gastro symptoms, biochemical and microbial data, inflammatory markers and any cardiovascular risk factors.Assessment of stool frequency, stool consistency, bloating, flatulence, abdominal pain prior during and after the dietary intervention.Markers of active disease (total immunological immunoglobin A (IgA) and IgA tissue transglutaminase; IgA endomysial antibodies (EMA), IgG deamidated gliadin peptide (DGP) or IgG tTG).Studies that did not meet the criteria were excluded.

Type of Studies
The systematic review included randomized controlled trials, experimental studies, and non-controlled trials, which were human trials only in this study.There were no restrictions on study size.Observational studies, case reports, and case series were excluded.Duplicate studies and studies without validated diagnoses of CD were also excluded.We conducted a comprehensive search in the following databases: CINAHL, Web of Science, CENTRAL, PubMed, Australian and New Zealand Clinical Trials Register and ClinicalTrials.govfrom inception up to March 2023.

Search Terms
The search strategy was based on Medical Subject Headings (MeSH) and other corresponding words.The terms include "Probiotics", "Bifdobacterium", "Lactobacillus", "Coeliac Disease", "Celiac disease".We used the logical operators "OR" or "AND" to connect the terms above.The search syntax: "Coeliac disease" OR "Celiac disease" AND probiotics OR "probiotic supplementation" was used on CINAHL, CENTRAL and Web of Science Core Selection.The syntax: ("Celiac Disease" [MeSH Terms] OR "Celiac Disease" [Text Word] OR "Coeliac Disease" [Title/Abstract] OR "CD" [All Fields]) AND ("Probiotics" [MeSH Terms] OR "probiotic*" [Text Word] OR "probiotic supplement" [Title/Abstract] OR "Probiotic supplementation" [Title/Abstract]) was used on PubMed Search.The total search results from PubMed Search were 665 results, Web of Science had 6264 results, CENTRAL produced 67 results and CINAHL, 2676 results.

Selection of Studies
Two independent authors (LF and CS), screened titles and abstracts and cleared duplicates as well.Full-text screening was performed by the same reviewers independently.All data were collated in the AI software tool Rayyan and after screening, selected articles were collected into a Rayyan file.For any disagreement case, with the help of an independent third author (YJ), a resolution was reached.
Two independent reviewers (LF and YJ) extracted data and information on authors, publication type (abstract/full text), publication year, geographic region (country), language, follow-up duration, age, gender, ethnicity, method of diagnosis (biopsy/serology/ HLA DQ2/8 status), adherence to a gluten-free diet, additional diagnosis such as IBS, Probiotic strain/combination and type of delivery, dosage, frequency (timing), duration of intervention (length).Comparator population characteristics and dosage, frequency, duration of intervention placebo type, GI symptom scores, QOL scores, adverse events, markers of active disease such as serum TNF-α, intestinal permeability.

Data Synthesis Strategy
Our data synthesis was conducted based on data availability and it comprised studies grouped under all reported outcomes and study details (e.g., number of participants, setting, etc.).Individual study data including specific study outcomes, methods of analysis and results were reported in tables.A narrative synthesis was then carried out by tabulation and narrative summary of populations, interventions, and outcomes.Meta-analysis was not undertaken due to the large heterogeneity between the studies.

Risk of Bias Assessment
Selected articles were reviewed for risk of bias using Cochrane risk of bias in RCTs (ROB-2 tool) and non RCTs (ROBINS-1 tool) [44,45].The Cochrane risk of bias assessment tools were used to assess randomization; concealment and blinding of participants, outcome assessors; deviations from intended interventions, missing outcome data, measurement of outcomes and selection of reported results.The first reviewer independently (LF) assessed the quality of selected articles for all domains of the tool so did a second reviewer (YJ).A table summarising the findings was created to display the quality of each study.

Records identified through database search n 2 Records a ter duplicates removed n 0 4 Records screened n 0 4 Records e cluded n 0 3 Full te t records assessed or eligibility n 31 FullFigure 1 .
Figure 1.Flow Chart of Study Selection.

Table 1 .
Risk of bias for Randomised Controlled Trials.

Table 2 .
Risk of bias for Non-Randomised Controlled Trials.

Table 3 .
Characteristics of studies and reported outcomes for RCTs.
Probiotics: Reduction in serum antibody concentrations (10% for IgA tTG and IgA DGP antibodies) Placebo: Increased antibody serum concentrations (IgA tTG, 7% and IgA DGP, 10%) at the end of the trial.Biochemical and microbial data: Primary endpoint: Non-significant increase of mean lactulose/mannitol ratio from baseline for the probiotic (p = 0.064) and placebo (p = 0.342) and Secondary endpoint: Probiotic: significant reduction in indigestion (p = 0.0035) and constipation (p = 0.0483) symptoms, however, borderline for reflux symptoms (p = 0.0586).Placebo: No significant changes in any syndrome (indigestion, diarrhoea, constipation, abdominal pain).Significant improvement in diarrhoea symptoms Outcome of inflammatory mediators: Probiotic and placebo: No significant changes in Th1 serum cytokines and serum chemokines.Significant increase in high baseline serum concentration of MIP-1β, (p < 0.04), but not in the placebo group.Cardiovascular risk factors: None reported.

Table 3 .
Cont.TNF-α in serum at baseline, significantly higher (p = 0.015) in the probiotic group (14.78 ± 6.43) than placebo group (10.58 ± 3.57).TNF-α in serum at the end of the study, significant decrease (p = 0.020) in the probiotic group (11.97 ± 3.58) from baseline levels.Bacteriodetes 20-40% in CD group, 10-20% in control group.Proteobacteria and Verrucomicrobia high in the placebo group.Statistically, Actinobacteria was low in CD group but increased after probiotic intake.Euryarchaeota, the only Archaea was found predominately in the control group.
10, interferon-y, IL-13, transforming growth factor-β1, IgG1 and IgG4.Placebo: Significant increase in gene copy Bacteroides fragilis group (p = 0.013), Enterobacteriaceae (p = 0.038), non-significant decrease in Bifidobacterium spp.(p = 0.151).Probiotic: No significant differences in Lactobacillus group, Bifidobacterium spp./B. fragilis group and Enterobacteriaceae.Significant decrease in faecal sIgA concentration Cardiovascular risk factors: None reported.TNF-α in serum on f/u, significantly higher in the probiotic group No correlation between positive serologic markers of CD and levels of TNF-α and IL-10 in individual patient with CD.Values for cytokine IL-10 in serum, below assay detection limit (5 pg/mL) and so not analysed.Cardiovascular risk factors: None reported.Cardiovascular risk factors: None reported.
Biochemical and microbial data: Descriptive statistics for the predominant bacteria showed that only Streptomyces spp.(p = 0.058) was different between the two groups at baseline and 12 weeks and Mycoplasma sp.(p = 0.026), 12 weeks only.Bifidobacteria (p = 0.001) reduced significantly, and Escherichia coli (p = 0.005) increased over time.Cardiovascular risk factors: None reported.

Table 4 .
Characteristics of studies and reported outcomes for non-RCTs.