The database searches retrieved 232 titles and abstracts, and a manual search of relevant bibliographies identified one additional record. After the removal of duplicates, 182 unique titles remained. These titles and abstracts were screened for eligibility; 128 records were excluded, and 54 full-text articles were reviewed. In the analysis, 12 articles were included. One of the articles reported two interventions [25
], and one article reported two probiotic study arms and one shared placebo arm [26
]. Thus, the analysis ultimately included 14 comparisons. A flow diagram of the identification and selection of studies is shown in Figure 1
A total of 951 subjects (mean: 68, range: 8–249/comparison) were analyzed in the 14 comparisons that were published in the 12 included articles. The subjects were healthy adults, including elderly persons. In most studies, both genders were evenly distributed in the analyzed population (Table 1
). Daily probiotic doses ranged from 0.05 × 109
to 46 × 109
colony-forming units (CFU) (mean 5.8 × 109
CFU). Treatment durations ranged from 1 to 12 weeks (mean six weeks) (Table 1
A total of eight probiotic or synbiotic products were studied, containing between one and six strains (Table 1
). Ten were single-strain products—L. gasseri
LG21, B. bifidum
YIT 10347, Bifidobacterium animalis
HN019, and Lactobacillus reuteri
DSM 17938—whereas the four remaining products were multi-strain products, containing various strains in species of B. bifidum
, B. lactis
, Bifidobacterium longum
, Lactobacillus casei
, Lactobacillus plantarum
, Lactobacillus rhamnosus
, and Lactobacillus acidophilus
. Four study products also contained other ingredients, such as antioxidants and prebiotics. In the included studies, the probiotics were administered in various formats: fermented dairy (seven comparisons), pill-like (four comparisons), powder (two comparisons), and olive oil (one comparison) (Table 1
Of the 13 included studies, six were randomized and seven performed blinding of the patients; various study designs were used, including parallel groups (six studies), before–after comparisons (five studies), and crossover designs (two studies) (Table 2
). After qualitative rating of the study design, per the Jadad scale, five randomized controlled trials (RCTs) with a parallel-group design were defined as high-quality, two RCTs with a parallel-group or crossover design were medium-quality, and the six remaining studies were low-quality (Table 2
). Although it is not part of the Jadad score, reporting on compliance is an important marker of quality. Nearly half of the comparisons (n
= 6) did not report compliance with the product (Table 3
As shown in Table 3
, of the 14 comparisons, 3-Gomi et al. (2015) study A and Waller et al. (2011) studies A and B reported significantly reduced (acid) regurgitation, and three comparisons [30
] did not report any improvement. With regards to reflux or heartburn, two comparisons noted a significant improvement [22
]. Trends for improvement were observed by [31
] with the modified GSRS and by [21
] with the original GSRS, but not with the FSSG questionnaire. Five comparisons reported no improvement: [25
] Studies A and B and [21
]. The remaining comparisons did not assess or report reflux syndrome symptoms.
Dyspepsia-related symptoms improved or declined in five comparisons: Gomi et al. (2015) study A and [21
]. One study reported increased symptoms for dyspepsia [22
], and three studies found no difference [31
Five comparisons recorded a significant reduction in pain (abdominal or epigastric): Gomi et al. (2015) Study A, Waller et al. (2011) Studies A and B, and [27
]. Two studies saw trends in reduced pain in the upper-GI region [30
], and one reported no effect [32
]. Three comparisons reported significantly less nausea: Waller et al. (2011) Studies A and B and [28
]. Improvements in gas-related upper-GI symptom severity were observed in four comparisons: Gomi et al. (2015) Study A, Ianiro et al. (2013), and Waller et al. (2011) Studies A and B. One study noted an improvement in the prevalence of flatus [31
Overall, of the 13 selected studies, 11 comparisons (79%) reported probiotic benefits on the symptoms of GERD, whereas no benefit was seen in the three remaining comparisons [32
]. Of the 11 former comparisons, five (45%) reported benefits for reflux symptoms, versus five (45%) for dyspepsia symptoms and nine (81%) for other upper-GI symptoms, such as nausea, abdominal pain, and gas-related symptoms (belching, gurgling, burping). Of the five high-quality RCTs, two comparisons in Waller et al. (2011) showed efficacy (40%) with regard to reflux symptoms and other upper-GI symptoms, primarily gas-related symptoms; one study (20%) [30
] noted improvements in dyspepsia-related symptoms.
Three comparisons reported benefits for regurgitation, using frequency score as the endpoint, in which two single-strain probiotics were used: B. bifidum
YIT 10347 in Study A in [25
] and B. lactis
HN019 for two comparisons in [26
]. Gomi et al. (2015) in Study A reported a lower frequency of regurgitation. Waller et al. (2011) observed similar efficacy in reducing the frequency score for regurgitation at a high dose (17.2 × 109
CFU) by 12.6, and by 9.0 at the lower dose (1.8 × 109
In this review, four comparisons were performed with B. bifidum
YIT 10347 with the same intervention regimen, but only one showed an effect on both regurgitation and dyspepsia—Study A in [25
]—one saw improvements in acid-related dyspepsia [21
], and two reported positive effects on gas-related symptoms [25
Most studies (n
= 9) recorded adverse events (AEs) (Table 3
), but none were associated with the probiotic intervention. No serious adverse events were reported.
In this systematic review, 13 prospective clinical studies, comprising 14 comparisons, were reviewed to determine the potential of probiotics to alleviate upper-GI symptoms in GERD in the general adult population. The mechanism of action of probiotics has focused primarily on the lower digestive tract, and the activities of probiotics in the upper-GI tract remain largely unknown [15
Nevertheless, probiotics of the genera Lactobacillus
are associated with modulations in the immune response and antagonistic activity toward potential pathogens through the production of short-chain fatty acids, such as lactic acid. Further, probiotics accelerate gastric emptying by interacting with stomach mucosal receptors, which are suspected of triggering transient lower esophageal sphincter relaxation, one of the pathophysiological mechanisms of GERD [35
]. In addition, probiotics can be beneficial for small intestinal bacterial overgrowth, interfering with immunity or intestinal motility under various conditions [36
]. These properties might be relevant to their effects in GERD, as discussed here.
4.1. Clinical Efficacy and Potential Mechanisms
A majority (79%) of the included comparisons reported probiotic benefits on the symptoms of GERD, such as regurgitation, heartburn, dyspepsia, nausea, abdominal pain, and gas-related symptoms (belching, gurgling, burping). However, the heterogeneity in the outcomes made it impossible to perform a meta-analysis.
Probiotics have positive effects on reflux with regards to the presence of episodes [27
] and frequency scores [22
]. The presence of reflux episodes fell significantly by 40% in 20 pregnant women [27
]. To our knowledge, de Milliano et al. (2012) is the first trial to supplement with multi-strain probiotics, and reported benefits for reflux, particularly in constipated pregnant women. The product in this study contained six probiotic strains from six species, including Bifibacterium
, providing efficacy for a wide range of upper- and lower-GI symptoms, such as abdominal pain and constipation [27
Based on the FSSG, the frequency scores for reflux declined significantly from 6.2 to 4.8 on supplementation with L. gasseri
LG21 for 12 weeks [22
]. Notably, in the same study, pepsinogen (PGI) level was the only stomach-related biomarker that had a significant negative correlation with the reflux symptom score, after the effects of gender and age were adjusted [22
]. PGI was suspected to be involved in the occurrence of symptoms; thus, a higher PGI level indicates accelerated protein digestion in the stomach. This explanation is one basis for the inverse relationship between increased PGI levels and reduced reflux symptoms, particularly in the presence of increased dysmotility-like dyspepsia, from 3.5 to 4.0 on the FSSG [22
In addition to its involvement in protein digestion, other underlying mechanisms of L. gasseri
LG21 in reflux and dyspepsia were examined in two cohorts [29
]. In these two studies, no improvement in regurgitation or heartburn was seen, whereas both reported reduced/lower dyspepsia—i.e., reduced postprandial distress—regardless of the experimental design (self-controlled or placebo-controlled). Moreover, Nakae et al. (2016) observed increased gastric emptying, as evidenced by the increased gastric fluid volume and suppressed gastric acid secretion, based on a higher pH value after treatment. Although Ohtsu et al. (2017) focused only on symptomology, to better differentiate PDS from EPS, they administered questionnaires other than the FSSG in [22
]. Notably, compared with the placebo, postprandial distress syndrome scores declined significantly with the L. gasseri
LG21 intervention (37.5% vs. 17.8%), whereas only a trend of improvement for epigastric burning, with no improvement in epigastric pain, was reported, indicating that L. gasseri
LG21 has greater beneficial effects on PDS symptoms than EPS [30
]. These findings strongly suggest that the underlying mechanisms of L. gasseri
LG21 for improving FD-associated reflux are linked to postprandial effects, involving improved protein indigestion and increased gastric emptying.
Little is known about gastric microbiota and its function in the pathogenesis of GERD and FD [29
]. In Nakea et al. (2016), FD patients had a clearly different bacterial community compared with healthy controls, in terms of overall community structure and bacterial taxonomic abundance. Prevotella
spp. is the predominant taxa inhabiting the stomach [29
]. In this study, the relative abundance of Prevotella
spp. was lower in FD patients versus healthy controls and significantly negatively correlated with the severity of postprandial distress symptoms scores, implicating Prevotella
spp. in the occurrence of FD symptoms.
infection is associated with symptoms of GERD. However, its eradication is not always associated with an improvement in symptoms [37
]. In view of this, it is unfortunate that the studies have not evaluated the presence or absence of H. pylori
. Although probiotics do not eradicate H. pylori
, they have been shown to reduce its activity [38
]. Information on the carriage of H. pylori
could thus have given information on some of the differences in efficacy.
4.2. Effects of Product Format on Efficacy
Limited evidence is available to compare efficacy between probiotic strains, due to the small number of available trials. However, all three included L. gasseri LG21 studies reported positive effects on various symptoms of GERD. Also, for B. bifidum YIT 10347, four intervention trials indicated positive changes in FSSG and GSRS scores. For other strains, too few studies were available to draw overall conclusions. Very diverse product formats were applied in the selected studies, ranging from fermented foods to various dietary supplement formats.
Although adverse events were reported in three studies [30
], none differed significantly between probiotic and placebo groups. Moreover, the adverse events were assumed not to be product-related effects.
4.4. Study Quality
A potential source of bias for the experimental design was assessed in Table 2
. As demonstrated in [39
], the level of evidence for trials with different experimental designs could be classified as follows (in descending order): high-quality RCTs, low-quality RCTs, prospective cohort studies, and others. All six RCTs were randomized and double-blinded with an identical appearance of placebo and probiotic-containing products. However, especially in food, it might be challenging to manufacture a placebo that is indistinguishable from the probiotic product.
Half of the studies implemented random sequence generation by computer-based randomization programs, block approach, or tables. However, the allocation concealment was not clear in these studies. Adequate allocation concealment is important for decreasing the risk of selection bias in clinical trials [40
]. The odds ratio for the estimated effects of inadequate allocation concealment on treatment can be as high as 41%, having been the only risk of bias until 2008, as reported by RevMan, a program that was used for Cochrane Reviews [41
]. These findings suggest that inadequate allocation concealment is the leading cause of bias. Two RCTs also had an unclear risk of bias due to the lack of a clear description of the reasons for withdrawal or dropouts.
Although it is not part of the Jadad score, reporting on compliance is an important quality marker. Nearly half of the comparisons did not report compliance with the product, including two studies that were judged to be high-quality. Future systematic reviews and meta-analyses should take product compliance into account in their quality evaluation.
Most studies reported positive outcomes for probiotics regarding the symptoms of GERD. However, there was substantial heterogeneity in the outcomes and symptoms. Thus, although the results are encouraging, it is difficult to draw any general conclusions on the effects of probiotics. The heterogeneity in endpoints also made it impossible to quantitively evaluate the results. Further, the quality of the studies is concerning—only 5 of 14 studies were good quality. Nevertheless, despite the diversity in the studied product formats, populations, and experimental designs, the efficacy of the probiotic treatment does not appear to be influenced by the study quality.
Properly designed, randomized, double-blind, placebo-controlled studies with a sufficient number of participants and well-defined endpoints are needed. Studies with a longer duration should also be considered, with an intermediate analysis of the endpoints—for example, through questionnaires—to determine the period in which the benefits can be expected and whether they are long-lasting.