Systematic Review

The Effects of Proton Pump Inhibitors on the Microbiome of the Digestive Tract

Evidence From Next-generation Sequencing Studies

Lukas Macke; Christian Schulz; Leandra Koletzko; Peter Malfertheiner


Aliment Pharmacol Ther. 2020;51(5):505-526. 

In This Article


Based on the available literature, PPIs are associated with moderate perturbations of the microbiota of the upper and lower gastrointestinal tract. In the majority of studies, these changes were not related to richness and diversity measures, but rather to altered abundance of specific taxa. Studies on upper gastrointestinal microbiota are limited by small sample size and heterogenous design, but support early observations that PPIs promote oesophageal, gastric and duodenal overgrowth of orally derived microbiota. In faecal samples, PPI-associated alterations of microbiota profiles have been well characterized in both interventional studies and large observational cohorts: PPI use is associated with increases in multiple taxa from the orders Bacillales (eg, Staphylococcaceae) and Lactobacillales (eg, Enterococcaceae, Lactobacillaceae and Streptococcaceae), the order Actinomycetales (eg, Actinomycetaceae and Micrococcaceae), families Pasteurellaceae and Enterobacteriaceae and genus Veillonella. Taxa decreased by PPI use include the families Bifidobacteriaceae, Ruminococcaceae and Lachnospiraceae and class Mollicutes. These data support the conclusion that PPI-induced hypochlorhydria allows upper gastrointestinal microbiota to colonize more distal parts of the digestive tract.

Limitations of the Reviewed Studies

This systematic review demonstrates significant heterogeneity among the study results. This may reflect the nature of the gut microbiome as a dynamic and sensitive equilibrium, but also differences in study design, study cohorts and methodology:

One issue is the choice of PPI drug and the duration of use. The experimental trials in this review characterized the effects of defined, short-term PPI challenges of few weeks, while some of the retrospective studies investigated long-term PPI-users. For the vast majority of individuals in the observational cohorts, duration of PPI intake was not documented, and criteria for the classification of PPI users vs non-users were not consistent between the studies. For example, a majority of PPI users in the TwinsUK cohort, defined as ever having taken PPIs in the median 3 years before faecal sampling, would have been categorized as non-PPI users by Imhann et al who assessed PPI intake at the time of stool sampling.[87,88] Furthermore, most studies did not differentiate between the different types and dosages of PPIs.

Secondly, PPI intake is highly correlated with age, body-mass-index, diet, comorbidities and concurrent medication. Each of these factors represents a relevant confounder for microbiome analyses.[37,89,90] Most of the observational studies reviewed here controlled for possible confounding variables, but even rigorous data adjustment cannot eliminate the risk of unaccounted bias. Furthermore, the majority of the available studies are limited by small numbers of patients. In 12 of the 23 reported study cohorts, less than 20 individuals were exposed to PPIs.[68,71–73,75–78,80,82,86] In addition to the small sample sizes, the high interindividual and temporal variations in the gut microbiome do not allow firm conclusions about PPI-induced microbiome alterations.

At the methodological level, 16S DNA sequencing is limited by the fact that it cannot differentiate viable from non-viable bacteria. A significant portion of the taxa identified by sequencing may therefore not be metabolically active. Only one study addressed this issue by sequencing reverse-transcribed rRNA,[75] while the more feasible approach—the use of propidium monoazide, a DNA intercalator selectively penetrating into cells with compromised membrane integrity—has not yet been used in this context. Propidium monoazide renders DNA insoluble so it gets lost during subsequent DNA extraction for sequencing analysis.[91]

Functional Relevance of PPI-associated Dysbiosis in Health and Disease

The current evidence for causality of PPIs in the majority of associated adverse events is insufficient.[3,7,13] To comprehend the role of PPI-related dysbiosis in disease, it will be crucial to understand the functional impact of PPI-induced acid inhibition on microbial metabolic pathways. In metagenomic analyses across the Kyoto Encyclopedia of Genes and Genomes (KEGG), Katagi et al found increased proportions of genes related to apoptosis, retinoic acid inducible gene-I-like receptor and phosphatidylinositol signalling, selenocompound metabolism, ether lipid and lipoic acid metabolism, and ubiquitin system in the microbiota of PPI users.[85] Seto et al and Otsuka et al, on the contrary, identified no significant PPI-induced KEGG pathway changes.[76,78] Further studies using metagenomic, metabolomic and metatranscriptomic approaches are required to understand the effects of PPIs on the functional capacity of the gut microbiome. Two examples of diseases that have been linked to PPI use will be discussed in the following.

Gastric Cancer. PPIs can induce bacterial overgrowth of the stomach and long-term PPI use has been associated with increased risk for gastric cancer in observational studies of large patients cohorts.[16–20] Indeed, gastric microbes are believed to contribute to gastric carcinogenesis: Non-H pylori bacterial overgrowth and H pylori were shown to independently and synergistically accelerate the development of atrophic gastritis[92] and in the H pylori INS-GAS mouse model, the gastric commensal microbiota were shown to contribute to carcinogenesis.[93,94] Two human populations with high and low gastric cancer risk in Colombia showed marked differences in gastric microbiota compositions. However, the taxonomic alterations observed in the gastric microbiota of the high-risk cohort of Túquerres—including enrichments of Leptotrichia wadei and a Veillonella sp–do not correspond to the gastric microbiological changes associated with PPIs in the studies reviewed here.[95]

Two interesting studies conducted in independent patient cohorts from Portugal, Mexico and China associated different stages of gastric carcinogenesis[96] with decreasing microbial richness in the stomach. Compared to individuals with chronic gastritis, the cancer-associated gastric microbiome is characterized by overrepresentation of both oral and intestinal commensal bacteria. However, analysis of cancer-specific enrichment and depletion at the taxonomic level yielded conflicting results between the two studies, possibly reflecting different geographic, ethnic and dietary backgrounds of the study populations. Furthermore, comparing the study results with the PPI-associated changes to the gastric microbiota reviewed here, not a single concordantly regulated taxon can be identified.[97,98]

To date, causal connections between gastric bacterial overgrowth, PPI-associated dysbiosis and stomach cancer are highly speculative and require further investigation in larger cohorts and experimental studies. To our current understanding, the putative impact of PPIs on gastric cancer risk is more likely mediated by two factors: PPIs interact with H pylori infection by inducing a shift from antrum-predominant to corpus-predominant gastritis and by accelerating atrophy.[99–104] Secondly, long-term PPI therapy induces hypergastrinemia which is suspected to contribute to gastric carcinogenesis[104–107] (Figure 2).

Figure 2.

Associations of PPI use with alterations of gastrointestinal physiology and microbiota. SIBO, small intestinal bacterial overgrowth

Enteric Infections and Clostridium difficile Infection. Already prior to the introduction of omeprazole, enteric infections have been linked to increased gastric pH.[108,109] The association between PPI use and bacterial infections with Salmonella and Campylobacter was consistently demonstrated in numerous observational studies and a recent meta-analysis.[15] Considering the relatively high ORs and the biological plausibility, given that these pathogens are sensitive to gastric acid, this association is generally considered to be likely causal.

In contrast, the odds for an increased occurrence of C difficile infection during PPI therapy are relatively low[8–12] and PPI-induced hypochlorhydria was shown not to affect C difficile spore integrity or germination.[14] Instead, it has been hypothesized that PPI-associated alterations in the gut microbiome may create a niche that facilitates the germination and disease initiation of C difficile infection. The biological framework provided by the large cohort studies by Imhann et al and Jackson et al and the cross-over trial by Freedberg et al shows partial taxonomic overlap between the microbiota of PPI users and those observed in C difficile infection. These include reduced alpha diversity, increased abundance of the class Gammaproteobacteria, the family Enterobacteriaceae and the genera Veillonella, Enterococcus, Streptococcus and Lactobacillus, along with decrease in the families Ruminococcaceae and Lachnospiraceae and the genus Bifidobacterium.[77,87,88,110] Freedberg et al also identified a significant increase in KEGG pathways corresponding to Staphylococcus aureus infection and to bacterial invasion of epithelial cells, including genes for antibacterial peptides and for maintenance of epithelial integrity. The authors hypothesize that PPIs may play an important role after C difficile sporulation by lowering colonization resistance.[77] However, the current knowledge about the microbial alterations preceding manifest C difficile infection is very limited. Whether C difficile-associated dysbiosis represents a risk factor for C difficile infection or a consequence of intestinal inflammation, is still unclear and warrants further experimental investigation (Figure 2).