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


Twelve independent observational with 4277 participants (708 using PPIs) and 11 interventional study cohorts with 180 participants (all using PPIs) were included in the review. A summary of the study cohorts is provided in the Table S1.

Oral Cavity Microbiota

Three studies with 171 participants (87 PPI users) assessed the effect of PPIs on the microbiome of the oral cavity (Table 2 and Table 3). The microbial communities in the saliva of 18 subjects on long-term PPI therapy for functional dyspepsia or gastro-oesophageal reflux disease and 27 healthy non-PPI users did not differ in species richness and in bacterial abundance at the genus level. However, the PPI group exhibited greater phylogenetic diversity of the salivary microbiota based on unweighted UniFrac distances, while weighted UniFrac distances—which take relative bacterial abundance into account—did not differ significantly.[68,69] In a cross-sectional cohort study of 116 children undergoing bronchoscopy and upper gastrointestinal endoscopy for the evaluation of chronic cough, PPI use within the past 48 hours (59 children) was not associated with altered oropharyngeal microbial diversity. At the genus level, however, PPI use was associated with increased prevalence of Butyrivibrio and increased abundance of 10 taxa including Streptococcus, Yersinia, Janthinobacterium and Zoogloea in oropharyngeal swabs.[70] A prospective investigation in 10 healthy adults showed that a 4-week course of esomeprazole (20 mg once daily) decreased species diversity of saliva microbiota, while not significantly affecting species richness and community diversity of saliva or periodontal pocket microbiota. At the genus level, PPIs reduced Veillonella, Neisseria and Haemophilus abundances in the saliva and increased Fusobacterium and Leptotrichia in the periodontal pocket[71] (Table 2 and Table 3).

Upper Gastrointestinal Microbiota

One prospective study with eight participants and four observational studies with 280 participants (108 PPI users) addressed the association of PPIs with changes in the oesophageal, gastric and duodenal microbiome (Table 2 and Table 3).

In eight patients with oesophagitis, Barrett's oesophagus and controls with heartburn and normal-appearing oesophageal mucosa, an 8-week course of lansoprazole (30 mg twice daily) resulted in distinct clustering of microbiota according to phylogenetic distance, significant increases of the families Clostridiaceae, Lachnospiraceae, Microccocaceae, Actinomycetaceae in oesophageal biopsies and Erysipelotrichaceae in gastric fluid samples, as well as decreases of Comamonadaceae in oesophageal biopsies and Moraxellaceae, Flavobacteriaceae, Comamonadaceae and Methylobacteriaceae in gastric fluid.[72]

In the paediatric cohort described above, PPI use was not associated with altered species diversity in gastric fluids, but with increased prevalence of Propionibacterium, Flavobacterium, Corynebacterium, Bacteroides, Zoogloea and Janthinobacterium and increased abundances of Streptococcus, Coprococcus, Brevibacterium, Blautia and Granulicatella.[70] In the previously described cohort of 18 long-term PPI users for gastro-oesophageal reflux disease or functional dyspepsia and 27 healthy non-PPI users, the groups showed no differences in species richness or bacterial abundance at the genus level. But similar to the salivary samples, the PPI group exhibited greater phylogenetic diversity of the gastric fluid microbiome, based on unweighted UniFrac distances, while weighted UniFrac distances did not differ significantly.[68] In a cohort of 24 adults undergoing upper gastrointestinal endoscopy for evaluation of dyspepsia, long-term intake of PPIs in 12 patients was not associated with alterations in species richness or diversity indices, but with increased abundance of the phylum Firmicutes, family Streptococcaceae and genus Streptococcus, independently of H pylori status.[73]

Due to the stability of the DNA molecule even in the acidic environment of the stomach, 16S DNA sequencing does not provide information on whether the bacteria are still viable and metabolically active. Sequencing of 16S rRNA after reverse transcription to DNA is one approach to overcome this limitation and to allow for the characterization of the active bacterial assemblages.[74] Using this technique, Parsons et al studied the gastric microbiota of 75 individuals with H pylori-induced atrophic gastritis, autoimmune atrophic gastritis and PPI use, conditions associated with similar levels of hypochlorhydria, but different risks and types of gastric cancer—specifically neuroendocrine and adenocarcinoma. Comparing the 19 PPI users from this cohort with 20 healthy controls, a significant reduction in bacterial richness and diversity was detected in gastric biopsies of the PPI users, along with significant decreases in the genera Actinobacillus and Tannerella[75] (Table 2 and Table 3).

Distal Gastrointestinal Microbiota

The majority of studies addressing the microbiological impact of PPIs were performed on stool samples. Nine interventional trials with 157 participants (Table 4 and Table 5) and seven observational studies with 4042 participants (618 PPI users; Table 6 and Table 7) investigated the faecal microbiological alterations associated with PPI therapy. One prospective investigation studied the effect of PPI withdrawal in 15 patients after long-term use.

Experimental Trials

Seto et al tracked the compositional changes of the faecal microbiome over time in nine healthy individuals during a 28-day course of low dose (20 mg once daily) vs high dose (20 mg twice daily) omeprazole and compared them to microbiota of five treatment-naïve C difficile infected patients. Independently of dose, PPI use resulted in a small but statistically significant reduction in species richness, which was partly reversible after 1 month of recovery. However, the species diversity and the longitudinal ecological distinctiveness remained largely preserved during PPI therapy with no taxonomic overlap with the microbial ecology in C difficile infected patients.[76] In 12 healthy individuals who received 4 weeks of omeprazole (40 mg twice daily) and were then randomized to continuing vs discontinuing PPIs for another 4 weeks, no changes in species diversity, no differential clustering according to phylogenetic distance and—in contrast to Seto et al[76]—no changes in species richness after PPI therapy were observed. However, at the taxonomic level, significant increases in Enterococcaceae, Streptococcaceae, Micrococcaceae and Staphylococcaceae, along with decreases in Clostridiaceae were identified. Interestingly, the alterations induced by 4 vs 8 weeks of PPIs did not significantly differ.[77] In a study comparing 4 weeks of lansoprazole (30 mg once daily, n = 11) vs the more potent potassium-competitive acid blocker vonoprazane (n = 9) in healthy adults, neither of the drugs affected species richness or diversity, but both caused specific clustering of microbiota according to phylogenetic distance. At the taxonomic level, the PPI caused increases in Bacteroides, Carnobacterium, Streptococcus and Oribacterium, while vonoprazane also induced enrichment of the genera Actinomyces, Rothia and Granulicatella—all core members of the oral microbiome—and decreases of the genera Blautia, Coprococcus and Holdemania.[78]

Gut microbiota have been proposed to play an important role in liver disease.[46] Bajaj et al compared the faecal microbiota before and after 2 weeks of omeprazole (40 mg once daily) in 15 patients with compensated cirrhosis and 15 age-matched healthy controls, and they found significant changes in phylogenetic diversity after PPI therapy in both cirrhotic and healthy participants. Specifically, the authors observed a significant reduction in the relative abundances of Lachnospiraceae and Ruminococcaceae in cirrhotic patients and increased abundances of Streptococcaceae in both cases and controls on PPI therapy.[79] In a cohort of 15 patients with decompensated cirrhosis, a 2-week course of omeprazole (40 mg once daily) resulted in significantly increased abundances of Porphyromonadaceae and Streptococcaceae. Conversely, withdrawal of chronic unindicated PPI therapy in 15 patients with decompensated cirrhosis resulted in significantly decreased abundance of in Porphyromonadaceae, Streptococcaceae and Veillonellaceae.[80]

In a cohort of 24 elderly, healthy individuals, a 14-day course of omeprazole (20 mg once daily) did not affect faecal bacterial richness or diversity, but significantly reduced the abundance of Lachnospiraceae, Erysipelotrichaceae and Bifidobacteriaceae, while increasing the abundance of Streptococcaceae.[81] Mishiro et al equally observed increased abundance of faecal Streptococcus levels after 4-weeks of esomeprazole (20 mg once daily).[71]

In a cohort of 12 infants younger than 1 year with gastro-oesophageal reflux disease, in whom the microbiome is still unstable, a 4-week-course of esomeprazole was not associated with alterations in alpha or beta diversity, but resulted in significant alterations of the phyla Firmicutes, Bacteroidetes and Proteobacteria and, at the genus level, decreased relative abundance of Lactobacillus and Stenotrophomonas and an increase of Haemophilus. PPI withdrawal resulted in increasing alpha and beta diversity correlating with patients' age and dietary habits.[82]

Koo et al studied effects of PPIs in a multi-ethnic, healthy Asian population of Chinese, Malaysian and Indian ancestry (n = 34). In this population, similar to previous findings in white Caucasian populations, 7 days of omeprazole (20 mg once daily) did not significantly affect alpha and beta diversity indices, independently of ethnicity and gender. At the taxonomic level, PPIs increased the relative abundances of the class Bacilli, order Lactobacillales, family Streptococcaceae, genera Streptococcus and Veillonella, and species Streptococcus vestibularis and Veillonella dispar, with alterations reverting to baseline after stopping omeprazole for another 7 days[83] (Table 4 and Table 5).

Observational Trials

Clooney et al assessed the impact of PPIs in 32 individuals with ≥5 years of continuous PPI therapy, along with 29 non-PPI using matched controls, each selected from a population-based database. Consistent with previous studies, the two groups exhibited no significant difference in species richness and diversity, but their overall microbial community composition based on phylogeny was distinct. The authors observed a decrease in the phylum Bacteroidetes, an increase in the phylum Firmicutes and at the species level, Holdemania filiformis were increased and Pseudoflavonifractor capillosus were decreased in PPI users.[84] Similarly, the microbiota of a population of 36 long-term PPI users exhibited no significant difference in alpha diversity indices, compared to 36 age- and sex-matched PPI non-users. However, the microbiota clustered distinctly according to phylogenetic distance and at the taxonomic level, long-term PPI use was associated with an increase of five genera including Streptococcus and Ruminococcus and a decrease of eight genera including Faecalibacterium, SMB53 and Clostridium.[85] Similarly, 18 subjects on long-term PPI therapy for functional dyspepsia or gastro-oesophageal reflux disease exhibited no significant difference in microbial species richness and only mild changes in phylogenetic distance, compared to 27 healthy non-PPI users. In age-matched groups from the two cohorts, altered abundances of Faecalibacterium and Streptococcus were not statistically different.[68] In a cohort of 20 patients with microscopic colitis, concomitant PPI use was associated with decreased abundance of the genera Actinomyces and Faecalibacterium.[86] In an outpatient cohort of 137 cirrhotics and 45 healthy controls, long-term PPI therapy in 59 and 17 patients, respectively, was associated with increased Streptococcaceae and Veillonellaceae and decreased Lachnospiraceae and Ruminococcaceae relative abundance.[80]

To date, the most robust evidence for PPI-associated alterations of the gut microbiome comes from to studies analysing microbiome data of large population-based cohorts.

In three independent Dutch cohorts comprising 1174 participants from the LifeLines-DEEP study, 300 patients with inflammatory bowel disease and a cohort of 341 individuals with irritable bowel syndrome or matched controls, PPI use (n = 211) was not associated with altered overall bacterial composition of the gut at the phylum, class and order level after correction for multiple covariates including age, gender, BMI and antibiotics use. However, in a subsequent meta-analysis of all three cohorts, PPI use was associated with a moderate decrease in species richness and diversity and statistically significant alterations in abundance in 92 bacterial taxa. Specifically, the order Actinomycetales, families Streptococcoceae and Micrococcoceae, genus Rothia and species Lactobacillus salivarius were increased in PPI users in each cohort. By analysing 116 oral microbiome samples from participants in the general population cohort, the authors confirmed a significant shift of the gut microbiome composition towards oral cavity microbiomes in the faeces of PPI users. This study also addressed possible confound due to concomitant medications: In the LifeLines-DEEP cohort, 16 taxa were associated with antibiotics and other commonly used drugs besides PPI. After adjustment for PPI use, six taxa remained associated with certain drugs: Statins, fibrates and laxatives. All 92 taxonomic alterations remained statistically significant after correction for antibiotics and other commonly used drugs.[87]

In the TwinsUK cohort comprising 1827 individuals, PPI use (n = 229) was significantly associated with age, frailty, body mass index and alcohol intake. After adjustment for family and twin structure, indication of drug prescription and the confounders named above, PPI use was not associated with significant changes in multiple indices of alpha diversity. At the taxonomic level, seven species—all from the orders Erysipelotrichales or Clostridiales and one Cyanobacteria—, nine genera—largely members of the family Erysipelotrichaceae—, and five families—most strongly Lachnospiraceae and Ruminococcaceae—, were negatively associated with PPI use. Positive associations were shown for 24 species—particularly from the family Lactobacillaceae and the order Clostridiales—, 24 genera—most significantly Rothia and Streptococcus—, and 10 families—most significantly Streptococcaceae and Micrococcaceae. Some of these findings were replicated in a paired analysis of 70 monozygotic twin pairs discordant for PPI use and in a re-analysis of data from the crossover trial by Freedberg et al confirming an increase in the order Lactobacillales and in particular the family Streptococcaceae to be associated with PPI use. Overall, PPI use was associated with a higher abundance of pharyngeal and environmental commensals in the gut. Among the studies reviewed here, this is the only study to consider host genetics as a variable in microbiome composition[88] (Table 6 and Table 7).