Table 1.
Participant Demographics.
Fig 1.
Brush sampling of the upper gastrointestinal tract enriches for bacterial abundance and diversity.
(A) Diagram of the human upper gastrointestinal tract indicating regions sampled via biopsy or brush collection method. Anatomic sites were abbreviated with the first and second letter indicating the sampled organ and intra-organ tissue, respectively (ES—squamous esophagus; EB—Barrett’s esophagus; SC—stomach corpus; SA—stomach antrum). (B) Total bacterial versus human DNA recovered from biopsy or brush samples segregated by anatomical site as measured by qPCR and plotted as copy number of bacterial 16S rRNA gene and human 18S rRNA gene. Error bars indicate standard deviation from the mean. (C) Ratio of human 18S rRNA to bacterial 16S rRNA copy number in all biopsy (n = 26) or brush (n = 35) samples. Error bars indicate standard deviation from the mean. (D) Species diversity in biopsy and brush samples as measured by quadratic entropy analysis. The central line within each box represents the median of the data, the whiskers represent the 5th and 95th percentiles and data outside that range are plotted as individual points. Statistical difference between biopsy and brush samples was measured by Wilcoxon rank sum test with continuity correction (p = 0.000594).
Fig 2.
Members of the Firmicutes or Bacteroidetes phyla dominate the upper gastrointestinal tract microbiome.
(A) Phylogenetic relationship of the top 45 OTUs recovered from each of the four sites sampled in individual participants. Respective phyla are noted above main branches of the phylogenetic tree. Numbers in parentheses represent total number of pyrosequencing reads recovered for a given species or genera across all samples followed by the fraction of participants in whom a relative abundance of ≥1.3% of a given species or genera were detected. (B) Species/genera-level profiles of top 45 OTUs detected by 454 sequencing in squamous esophagus, Barrett’s esophagus, stomach corpus and stomach antrum of indicated participants. Data arranged in order of increasing Firmicutes dominance. Individual species/genera are color-coded according to scheme presented in (A). Sequencing reads from brush samples were used when available, otherwise, data from biopsy samples are shown. Species reads were normalized to the total number of reads per corresponding site in a given individual. †Denotes samples collected at a second time point (P2 [t = 4 months]; P7 [t = 2 years]; P9 [t = 3 years]); Hp+ indicates H. pylori-positive individual. Italicized participant IDs denote data from biopsy samples in cases where brush samples were not available for analysis.
Fig 3.
Phylogenetic sample profiles are most similar within individuals rather than across anatomic sites.
(A) Principal component analysis of phylogenetic similarity among samples from each of the four anatomic sites of indicated participants. The number in parentheses corresponds to the percent variance of the data assigned to each indicated principal component. (B) Median KR distance across all samples between individuals (inter) and across sites within individuals (intra). Statistical significance was determined by Wilcoxon rank sum test with continuity correction. (C) Intra-individual KR distance of the three anatomic sites relative to the site indicated at the top of each graph. For B & C, the central line within each box represents the median of the data, the whiskers represent the 5th and 95th percentiles and data outside that range are plotted as individual points.
Fig 4.
Upper gastrointestinal microbiome similarity with replicate sampling.
(A–C) Species/genera-level profiles of microbiota detected by 454 sequencing in squamous esophagus, Barrett’s esophagus, stomach corpus and stomach antrum of individuals P7 at the time of first [t = 0] and second sample collection [t = 4 months] (A), P2 at t = 0 and t = 2 years (B) and P9 at t = 0 and t = 3 years (C). Individual species/genera are presented according to coloring scheme described in Fig 2 (D) Phylogenetic KR distance between (inter) samples from participants P2, P7 and P9 at both time points and within those individuals comparing the 1st and 2nd time points from the indicated anatomic site. The central line within each box represents the median and the whiskers represent the minimum and maximum values.
Fig 5.
Streptococcus to Prevotella species ratio corresponds to phylogenetic distance sample clustering and correlates with Barrett’s esophagus risk factors.
(A) Cluster analysis of KR distances between microbial communities of individual study samples. Pyroseq. Strep:Prev ratio was calculated using relative abundance of mapped reads for all Streptococcus and Prevotella species as determined by pyrosequencing. ddPCR Strep:Prev ratio was calculated using copies/μl of a Streptococcus or Prevotella-specific 16s rRNA gene segment as determined by droplet digital PCR. Samples color-coded based on the majority of calculated Pyroseq. Strep:Prev ratios in a group being <0.5 (blue), 0.5–1.5 (green), 1.5–4.0 (magenta) or >4.0 (red). (B) Boxplots comparing Streptococcus to Prevotella ratio as determined by pyrosequencing and ddPCR. The central line within each box represents the median of the data, the whiskers represent the 5th and 95th percentiles and data outside that range are plotted as individual points. (C) Relationship of Streptococcus to Prevotella ratio (measured by ddPCR) and waist-hip ratio of all male participants segregated by anatomic site. Strength of association between these two variables was determined by Pearson’s correlation test with correlation coefficient squared (r2) values as indicated. (D) Relationship of Streptococcus to Prevotella ratios (measured by ddPCR) and hiatal hernia length in all participants segregated by anatomic site. Strength of association tween these two variables was determined by Pearson’s correlation test with correlation coefficient squared (r2) values as indicated.
Fig 6.
Incidence of cancer and aneuploidy in Seattle Barrett’s Esophagus Research Program cohort.
Kaplan-Meier curves for participants within the cohort for (A) cancer incidence (n = 433, 39 infected) and (B) aneuploidy as measured by DNA content flow cytometry (n = 337, 33 infected). H. pylori infection was assessed by histology of antral biopsies. Note aneuploidy information was not available for all research participants. Statistical significance was determined using the Log-rank test.