ABSTRACT

   Metabolites and their interactions with microbiota may be involved in
   Helicobacter pylori-associated gastric lesion development. This study
   aimed to explore metabolite alterations upon H. pylori eradication and
   possible roles of microbiota-metabolite interactions in progression of
   precancerous lesions. Targeted metabolomics assays and 16S rRNA gene
   sequencing were conducted to investigate metabolic and microbial
   alterations of paired gastric biopsy specimens in 58 subjects with
   successful and 57 subjects with failed anti-H. pylori treatment.
   Integrative analyses were performed by combining the metabolomics and
   microbiome profiles from the same intervention participants. A total of
   81 metabolites were significantly altered after successful eradication
   compared to failed treatment, including acylcarnitines, ceramides,
   triacylglycerol, cholesterol esters, fatty acid, sphingolipids,
   glycerophospholipids, and glycosylceramides, with P values of <0.05 for
   all. The differential metabolites showed significant correlations with
   microbiota in baseline biopsy specimens, such as negative correlations
   between Helicobacter and glycerophospholipids, glycosylceramide, and
   triacylglycerol (P < 0.05 for all), which were altered by eradication.
   The characteristic negative correlations between glycosylceramides and
   Fusobacterium, Streptococcus, and Gemella in H. pylori-positive
   baseline biopsy specimens were further noticed in active gastritis and
   intestinal metaplasia (P < 0.05 for all). A panel including
   differential metabolites, genera, and their interactions may help to
   discriminate high-risk subjects who progressed from mild to advanced
   precancerous lesions in short-term and long-term follow-up periods with
   areas under the curve (AUC) of 0.914 and 0.801, respectively.
   Therefore, our findings provide new insights into the metabolites and
   microbiota interactions in H. pylori-associated gastric lesion
   progression.

   IMPORTANCE In this study, a panel was established including
   differential metabolites, genera, and their interactions, which may
   help to discriminate high-risk subjects for progression from mild
   lesions to advanced precancerous lesions in short-term and long-term
   follow-up.

   KEYWORDS: Helicobacter pylori, gastric metabolites, gastric microbiota,
   interactions, precancerous lesions

INTRODUCTION

   Helicobacter pylori is an important risk factor for various gastric
   disorders, including chronic gastritis, glandular atrophy, intestinal
   metaplasia (IM), epithelial dysplasia, and even gastric cancer (GC)
   ([54]1). Eradication treatment has shown great benefits for GC
   prevention and regression of precancerous lesions ([55]2[56]–[57]4).
   Pathogenic mechanisms of gastric carcinogenesis involve complex
   interactions among H. pylori, other gastric microbiota, and associated
   metabolites, although systematic studies are still needed.

   The advances in sequencing technology have revealed altered microbial
   diversity and different bacterial interactions in GC and precancerous
   lesions ([58]5, [59]6). Our previous intervention study confirmed that
   H. pylori may induce gastric microbial dysbiosis, which can be restored
   by successful eradication ([60]7). Our subsequent follow-up study
   further showed that the differential bacteria in the
   progression-to-dysplasia/GC subjects were enriched in protein and
   adipose metabolism pathways by microbial functional-capacity prediction
   ([61]8). However, the exact microbiota-metabolite interactions still
   need integrative microbiome and metabolomics confirmation.

   Recent studies have reported that interplays between the gut microbiota
   and metabolites may regulate inflammation and the immune system in
   colorectal carcinogenesis, such as between Bifidobacterium,
   Lactobacillus, and short-chain fatty acids ([62]9[63]–[64]11). An
   integrative microbiome and metabolomics study in GC tissues found
   significant correlations between Helicobacter, Lactobacillus, and
   differential metabolites ([65]12). Nevertheless, the
   microbiota-metabolite interactions in the early stage of gastric lesion
   progression during H. pylori infection, which are important for a
   deeper understanding of GC etiology and prevention, still remain
   unclear.

   In the present study, we compared the metabolomics profiles of paired
   gastric biopsy specimens at baseline and follow-up time points from
   subjects with successful or failed anti-H. pylori treatment based on a
   prospective population-based cohort. The differential metabolites by H.
   pylori eradication were integrated with the differential microbiota in
   the same intervention subjects. This study provided us a unique
   opportunity to unravel the possible interactions between the gastric
   microbiota and metabolites in the early stage of precancerous lesion
   progression.

RESULTS

   The baseline characteristics of successful eradication and failed
   treatment groups are presented in Table S1 in the supplemental
   material. No significant differences in baseline age,
   delta-over-baseline (DOB) value in [^13C]urea breath test (^13C-UBT),
   body mass index (BMI), smoking habits, alcohol consumption, or presence
   of active gastritis or gastric lesions were found between successful H.
   pylori eradication and failed treatment groups (P > 0.05 for all).
   There was a higher frequency of male subjects in the failed treatment
   group than in the successful eradication group (70.2% versus 51.7%;
   P = 0.043).

Alterations of gastric metabolites by H. pylori eradication.

   A total of 267 metabolites were quantified using the Biocrates P500
   platform in 230 gastric biopsy specimens, including 58 pairs before and
   after successful eradication and 57 pairs before and after failed
   treatment. Orthogonal partial least-squares discrimination analysis
   (OPLS-DA) found significant differences in gastric metabolomics
   profiles before and after successful eradication (R^2 = 0.84, Q^2 =
   0.68, and P = 0.038 [[66]Fig. 1A and [67]B]). However, in the failed
   treatment group, gastric metabolomics profiles were not significantly
   changed by medical therapy (R^2 = 0.71, Q^2 = 0.53, and P = 0.167
   [[68]Fig. 1C and [69]D]).

FIG 1.

   [70]FIG 1
   [71]Open in a new tab

   Metabolomics profiles in gastric biopsy specimens before and after
   anti-H. pylori treatment. Orthogonal partial least-squares
   discrimination analysis (OPLS-DA) found significant differences in
   gastric metabolomics profiles before and after successful H. pylori
   eradication (A); however, no significant changes were found for
   participants who failed to clear H. pylori by medical therapy (C). The
   OPLS-DA models in panels A and C were validated in panels B and D,
   respectively.

   A total of 81 metabolites showed significant alterations after
   successful eradication compared to failed treatment, with fold changes
   of >1.5 and P values of <0.05 adjusted for multiple comparison by the
   false-discovery rate (FDR) (Table S2). The differential metabolites
   include 4 acylcarnitines (Cx), 3 ceramides (Cer), 3 cholesterol esters
   (CE), 1 fatty acid (FA), 10 sphingolipids (SM), 6 triacylglycerols
   (TG), 44 glycerophospholipids, and 10 glycosylceramides. The
   glycerophospholipids and glycosylceramides can be further subdivided as
   1 lyso-phosphatidylcholine (lysoPC), 43 phosphatidylcholines (PC), 5
   hexosylceramides (HexCer), 3 dihexosylceramides (Hex2Cer), and 2
   trihexosylceramides (Hex3Cer).

   We investigated the potentially relevant influence factors for gastric
   metabolite alterations after H. pylori eradication. To represent the
   overall metabolic status in each gastric biopsy specimen, we calculated
   a comprehensive metabolic index using the 81 differential metabolites.
   Potentially relevant factors include DOB value in ^13C-UBT representing
   H. pylori infection status, microbial Shannon and Richness indexes
   representing gastric microbial diversity, and gastric juice pH value.
   [72]Figure 2A shows significant increases in microbial diversity
   indexes (P < 0.001 for both) and decreases in pH values (P = 0.001) as
   well as in metabolic indexes (P < 0.001) accompanying the dramatically
   decreasing trend of DOB values (P < 0.001) after successful
   eradication. However, except for the fact that pH values were increased
   (P = 0.020) in the failed treatment group, the factors showed no
   significant alterations (P < 0.05 for all).

FIG 2.

   [73]FIG 2
   [74]Open in a new tab

   Potentially relevant influence factors and pathway enrichment analyses
   for gastric metabolite alterations after H. pylori eradication. (A)
   Paired plots showed significant decreasing trends of DOB values
   (P < 0.001), pH values (P = 0.001), and metabolic indexes (P < 0.001)
   and significant increasing trends of microbial diversity indexes
   (P < 0.001 for both) after successful H. pylori eradication. However,
   in the failed treatment group, except that pH values were increased
   significantly (P = 0.020), the factors showed no significant
   alterations (P > 0.05 for all). a, metabolic index was calculated using
   the 81 differential metabolites after H. pylori eradication to
   represent the overall metabolic status in each gastric biopsy specimen
   by the logistic regression equation in Materials and Methods. DOB,
   delta over baseline. (B) A total of 26 pathways were enriched by both
   the differential metabolites and microbiota after successful
   eradication. Among them, 5 pathways were significantly changed by
   eradication in both metabolomics and microbiome analyses, including
   pathways for leishmaniasis, adipocytokine signaling, lipid and
   atherosclerosis, choline metabolism in cancer, and cholesterol
   metabolism (P values <0.05 for all). *, P values < 0.05; **, P values <
   0.01.

   The putative functions of the 81 differential metabolites after H.
   pylori eradication were found to be enriched in 29 pathways using the
   Kyoto Encyclopedia of Genes and Genomes (KEGG) database. We compared
   the 29 metabolic pathways with the previously predicted microbiota
   functional-capacity changes ([75]7) in the same intervention
   participants. A total of 26 pathways were enriched both by the
   differential metabolites and microbiota, covering 5 main classes, such
   as cellular processes, environmental information processing, human
   diseases, metabolism, and organismal systems ([76]Fig. 2B). Among them,
   5 pathways were significantly changed after eradication in both
   metabolomics and microbiome analyses, including leishmaniasis,
   adipocytokine signaling, lipid and atherosclerosis, choline metabolism
   in cancer, and cholesterol metabolism pathways (P < 0.05 for all).

Interactions between differential metabolites and microbiota before and after
H. pylori eradication.

   In our microbiome analysis, 65 bacterial taxa were significantly
   changed after successful H. pylori eradication ([77]7). We used
   Spearman’s correlation analysis to assess the potential interplays
   between the 81 differential metabolites and the 65 differential taxa in
   the 91 participants with both microbiome and metabolomics results. We
   found 71 negative (−0.29 < r < −0.21; P < 0.05 for all) and 177
   positive (0.21< r < 0.28; P < 0.05 for all) significant correlations
   between 34 metabolites and 58 taxa in H. pylori-positive baseline
   biopsy specimens ([78]Fig. 3A). After successful eradication, stronger
   correlations were found between 30 metabolites and 41 taxa, including
   49 negative (−0.44 < r < −0.34; P < 0.05 for all) and 44 positive (0.34
   < r < 0.52; P < 0.05 for all) correlations ([79]Fig. 3B). After failed
   treatment ([80]Fig. 3C), 20 metabolites and 63 taxa were significantly
   correlated, with 163 negative associations (−0.52 < r <−0.26; P < 0.05
   for all) and 60 positive associations (0.26 < r < 0.48; P < 0.05 for
   all).

FIG 3.

   [81]FIG 3
   [82]Open in a new tab

   Interactions between differential metabolites and the microbiota before
   and after anti-H. pylori treatment. The heat maps show correlations
   between 81 differential metabolites and 65 previously identified
   differential bacterial taxa in biopsy specimens of subjects at H.
   pylori-positive baseline (A), after successful eradication (B), or
   after failed treatment (C). *, P values <0.05. The correlation networks
   showed the significant associations between differential metabolites
   and previously identified differential genera in biopsy specimens of
   subjects at H. pylori-positive baseline (D), after successful
   eradication (E), or after failed treatment (F). Red lines represent
   positive genus-metabolite correlations. Blue lines represent negative
   genus-metabolite correlations. Green nodes represent differential
   metabolites. Orange nodes represent gastric microbiota. Node radii are
   based on the number of significant genus-metabolite correlations. Cx,
   acylcarnitines; Cer, ceramides; CE, cholesterol esters; FA, fatty
   acids; HexCer, hexosylceramides; Hex2Cer, dihexosylceramides; Hex3Cer,
   trihexosylceramides; PC, phosphatidylcholines; lysoPC,
   lyso-phosphatidylcholine; SM, sphingomyelins; TG, triacylglycerols.

   Of the 65 differential taxa, we further focused on the 18 differential
   genera after H. pylori eradication ([83]7). Significant interactions
   between differential genera and metabolites were visualized by
   correlation network construction. Helicobacter was negatively
   correlated with glycerophospholipids, glycosylceramide, and
   triacylglycerol (−0.24 < r < −0.21; P < 0.05 for all) in baseline
   biopsy specimens ([84]Fig. 3D), while it was positively correlated with
   glycerophospholipids and sphingolipids (0.37 < r < 0.51; P < 0.05 for
   all) after successful eradication ([85]Fig. 3E). Positive correlations
   between triacylglycerols and Prevotella (0.21<r < 0.26; P < 0.05 for
   all) and negative correlations between glycosylceramides and many
   non-Helicobacter genera, such as Fusobacterium, Gemella, and
   Streptococcus (−0.29 < r < −0.20; P < 0.05 for all), were notable in
   baseline biopsy specimens ([86]Fig. 3D) but absent after successful
   eradication ([87]Fig. 3E). After failed treatment ([88]Fig. 3F), more
   negative correlations were identified between differential genera and
   metabolites (−0.52 < r < −0.26; P < 0.05 for all).

Interactions between differential genera and metabolites in gastric lesions.

   Besides anti-H. pylori treatment, some specific genus-metabolite
   interactions are also associated with gastric lesions. In baseline
   biopsy specimens, significant negative correlations of
   glycosylceramides (−0.30 < r < −0.23; P < 0.05 for all) with some
   non-Helicobacter genera, such as Fusobacterium, Gemella, Porphyromonas,
   and Streptococcus ([89]Fig. 4A), were found when active gastritis was
   present, while they were not found when active gastritis was absent
   ([90]Fig. 4B). After anti-H. pylori treatment, we still observed more
   negative correlations of glycosylceramides with these non-Helicobacter
   genera (−0.61 < r < −0.26; P < 0.05 for all) in active gastritis biopsy
   specimens ([91]Fig. 4C) than in biopsy specimens without active
   gastritis ([92]Fig. 4D). Positive correlations of glycosylceramides
   were only found with Helicobacter (0.31 < r < 0.58; P < 0.05 for all)
   after treatment.

FIG 4.

   [93]FIG 4
   [94]Open in a new tab

   Interactions between differential metabolites and genera in various
   gastric lesions. The significant correlations between differential
   metabolites and genera are shown for H. pylori-positive baseline biopsy
   specimens with presence (A) or absence (B) of active gastritis and in
   biopsy specimens after treatment in the presence (C) or absence (D) of
   active gastritis. The significant correlations between differential
   metabolites and genera are also shown for H. pylori-positive baseline
   biopsy specimens with IM (E) or SG/CAG lesions (F) and biopsy specimens
   after treatment with IM (G) or SG/CAG lesions (H).

   We further identified some characteristic genus-metabolite correlations
   in IM compared to mild lesions (superficial gastritis [SG]/chronic
   atrophic gastritis [CAG]). In baseline biopsy specimens, many negative
   correlations were found between differential metabolites, such as
   glycosylceramides, and Fusobacterium, Gemella, Neisseria, and
   Streptococcus (−0.49 < r < −0.28; P < 0.05 for all) in IM lesions
   ([95]Fig. 4E). In contrast, no significant negative correlation was
   found between metabolites and the non-Helicobacter genera in mild
   baseline lesions ([96]Fig. 4F). Similarly, more significant negative
   correlations (−0.63 < r < −0.25; P < 0.05 for all) were found in IM
   lesions after intervention between glycerophospholipids,
   glycosylceramides, and ceramides and various non-Helicobacter genera
   ([97]Fig. 4G) than in SG/CAG lesions ([98]Fig. 4H). Positive
   correlations (0.25 < r < 0.71; P < 0.05 for all) between Prevotella and
   triacylglycerol and between Helicobacter and various differential
   metabolites were also noticed in IM lesions.

Screening of significant gastric metabolites and their associated genera for
gastric lesion progression.

   To identify significant metabolites and their associated genera in
   advanced gastric lesions, we preliminarily selected 13 and 33
   differential metabolites correlated with the differential genera in
   baseline and follow-up IM biopsy specimens after intervention,
   respectively (Tables S3 and S4). A total of 35 genus-associated
   metabolites (11 overlapping between baseline and follow-up biopsy
   specimens) were compared between the IM and SG/CAG groups by
   multivariate regression. In baseline biopsy specimens (Table S5), 6
   metabolites showed significant differences between the IM and SG/CAG
   groups, including Cer(d18:1/24:1), FA(20:3), PCaeC32:1,
   HexCer(d18:1/24:1), SM(OH)C22:1, TG(18:0_36:3), all P < 0.05. In
   follow-up biopsy specimens (Table S6), Cer(d18:1/22:0),
   Hex2Cer(d18:1/22:0), and Hex3Cer(d18:1/16:0) further showed significant
   differences between IM and mild lesions (P < 0.05 for all). The 9
   differential metabolites were remarkably associated with 14
   differential genera, showing 49 pairs of genus-metabolite correlations
   in IM subjects before and after interventions ([99]Fig. 5A).

FIG 5.

   [100]FIG 5
   [101]Open in a new tab

   Screening of significant gastric metabolites and their associated
   genera for gastric lesion progression. (A) The 9 differential
   metabolites in IM lesions before or after interventions showed 49 pairs
   of significant correlations with 14 differential gastric genera. (B)
   Alterations of 9 metabolites and 14 associated genera were screened by
   LASSO regression, with 7 metabolites and 7 genera showing significant
   associations with risk of lesion progression to IM. (C) These
   significantly altered metabolites and genera were selected by
   multivariate unconditional logistic regression adjusted for age,
   gender, and the effect of anti-H. pylori treatment with
   Cer(d18:1/22:0), Cer(d18:1/24:1), Hex2Cer(d18:1/22:0),
   HexCer(d18:1/24:1) and Fusobacterium, Helicobacter, Neisseria, and
   Streptococcus showing significant associations with risk of lesion
   progression to IM. (D) The selected 4 metabolites, 4 genera, and their
   11 significant interactions according to panel A were further enrolled
   in a multivariate unconditional logistic regression model with 4
   genus-metabolite interactions showing significant associations with
   risk of lesion progression to IM.

   In the 6-month follow-up period after eradication, we identified 21
   high-risk subjects who progressed from SG/CAG to IM and 23 low-risk
   subjects who reversed from IM to mild lesions (8 subjects) or remained
   as showing SG/CAG (15 subjects). We further evaluated the alterations
   of the 9 metabolites and 14 associated genera with the risk of lesion
   progression by least absolute shrinkage and selection operator (LASSO)
   regression. A total of 7 metabolites and 7 genera were significantly
   associated with the risk of progression to IM, including
   Cer(d18:1/22:0), Cer(d18:1/24:1), Hex2Cer(d18:1/22:0),
   Hex3Cer(d18:1/16:0), HexCer(d18:1/24:1), PCae32:1, and SM(OH)C22:1 and
   Bacteroides, Fusobacterium, Helicobacter, Neisseria, Porphyromonas,
   Prevotella, and Streptococcus ([102]Fig. 5B). These significantly
   altered metabolites and genera were selected by multivariate
   unconditional logistic regression adjusted for age, gender, and effect
   of intervention. Finally, 4 metabolites and 4 genera showed significant
   associations with the risk of progression to IM, including
   Cer(d18:1/22:0), Cer(d18:1/24:1), Hex2Cer(d18:1/22:0), and
   HexCer(d18:1/24:1) and Fusobacterium, Helicobacter, Neisseria, and
   Streptococcus ([103]Fig. 5C).

   The 4 metabolites, the 4 genera, and their 11 significant
   genus-metabolite interactions according to [104]Fig. 5A were further
   enrolled in a multivariate logistic regression model to identify
   significant interactions by comparing high-risk and low-risk subjects
   for progression to IM. Four significant genus-metabolite interactions
   were identified, including Cer(d18:1/22:0)-Helicobacter,
   Cer(d18:1/24:1)-Helicobacter, Hex2Cer(d18:1/22:0)-Streptococcus, and
   Hex2Cer(d18:1/22:0)-Neisseria ([105]Fig. 5D).

Improved discrimination of gastric lesion progression by genus-metabolite
interactions.

   Receiver operating characteristic (ROC) curve analysis between
   high-risk and low-risk subjects regarding lesion progression was
   conducted. The discrimination performances were compared among model 1
   (including age and gender), model 2 (including age, gender, 4
   metabolites, and 4 genera) and model 3 (including age, gender, 4
   metabolites, 4 genera, and 4 significant genus-metabolite
   interactions). The area under the curve (AUC) was remarkably increased,
   from 0.651 (P = 0.041) by model 1 to 0.806 (P < 0.001) by model 2.
   Furthermore, model 3 can significantly improve the performance of model
   2 by adding 4 genus-metabolite interactions, with an AUC of 0.914
   (P < 0.001). The discrimination performances of the 3 models were
   verified by random forest with leave-one-out cross-validation and
   compared by the DeLong test, with significant differences between model
   2 and model 1 (P = 0.010), as well as between model 3 and model 2
   (P = 0.033) ([106]Fig. 6A).

FIG 6.

   [107]FIG 6
   [108]Open in a new tab

   Improved discrimination of gastric lesion progression by
   genus-metabolite interactions. ROC curve analysis was conducted to
   compare the discrimination performances of three different models
   between 21 high-risk subjects who progressed from mild gastric lesions
   (SG/CAG) to IM and 23 low-risk subjects who reversed from IM to mild
   lesions or remained with mild lesions in short-term follow-up period
   (A) or between 13 high-risk subjects who progressed to or remained with
   IM/LGIN and 6 low-risk subjects who remained with SG/CAG in the
   long-term follow-up period (B). In particular, the performances of the
   3 models were also compared between 6 high-risk subjects who progressed
   to LGIN and 6 low-risk subjects who remained as showing SG/CAG in the
   long-term follow-up period (C). The discrimination performances of the
   3 models were verified by random forest with leave-one-out
   cross-validation and compared by the DeLong test.

   In the high-risk and low-risk subjects for lesion progression to IM
   after the 6-month follow-up, 19 subjects were further followed once by
   endoscopic screening from 2018 to 2021. Among them, 13 high-risk
   subjects progressed to or remained as showing IM/low-grade
   intraepithelial neoplasia (LGIN) and 6 low-risk subjects remained as
   SG/CAG in the long-term follow-up period. Model 2 with short-term
   metabolite and genus alterations and model 3 with additional
   genus-metabolite interactions still can discriminate the high-risk and
   low-risk subjects for lesion progression in long-term follow-up, with
   AUCs of 0.717 (P = 0.013) and 0.801 (P = 0.001), respectively
   ([109]Fig. 6B). In particular, the 6 high-risk subjects who progressed
   to LGIN in long-term follow-up ([110]Fig. 6C) were well distinguished
   from the 6 low-risk subjects by model 2 (AUC = 0.778 and P = 0.006) and
   model 3 (AUC = 0.905 and P = 0.010).

DISCUSSION

   Our previous prospective studies revealed that H. pylori infection can
   induce gastric microbial dysbiosis, which may be involved in
   carcinogenesis ([111]7, [112]8). The present study further profiled the
   microbiota-associated metabolite alterations in gastric mucosa in the
   same group of intervention participants. We found that H. pylori
   eradication can significantly change gastric metabolites and their
   interactions with the microbiota. The interactions between the gastric
   microbiota and metabolites may play a role in the progression of
   precancerous lesions.

   Numerous studies have deeply investigated the molecular characteristics
   of GC tissues by multiomics technologies ([113]13[114]–[115]15).
   However, it is still very difficult to reveal the dynamic mechanisms in
   the natural evolution process of precancerous lesions. Our previous
   anti-H. pylori intervention and follow-up study provides a good
   opportunity to assess the H. pylori-associated metabolite alterations
   in early stage of precancerous development. By comparing the metabolic
   profiles before and after treatment, we found significant changes in
   successful eradication group rather than in failed treatment group.
   Furthermore, 81 metabolites were significantly altered by successful
   eradication. Some significant metabolites in our study, such as fatty
   acid and glycerophospholipids, are consistent with those in a
   GC-associated metabolomics study ([116]12), which suggests that H.
   pylori-associated metabolite alterations in the gastric mucosa may be
   involved in progression of lesions and carcinogenesis.

   Integrative studies of metagenomics and metabolomics have highlighted
   the importance of the microbiota and metabolite interactions in
   gastrointestinal carcinogenesis ([117]11, [118]16, [119]17). Our study
   also investigated possible relevant factors for metabolite alterations
   by H. pylori eradication. Dramatic alterations in gastric microbial
   diversity and metabolic indices were found simultaneously after
   successful eradication. However, such significant changes were not seen
   after failed treatment. These phenomena further imply that the
   metabolite alterations after H. pylori eradication may be associated
   with the comprehensive microbial changes in the gastric mucosa.

   KEGG enrichment analysis can help to better understand the possible
   functions of the differential metabolites by H. pylori eradication. A
   total of 29 metabolic pathways were enriched by the 81 differential
   metabolites, including 26 consistent pathways with the predicted
   functional-capacity changes by the differential microbiota in the same
   intervention participants ([120]7). The 5 pathways significantly
   changed by H. pylori eradication in both metabolomics and microbiome
   analyses include pathways for leishmaniasis, adipocytokine signaling,
   lipid and atherosclerosis, choline metabolism in cancer, and
   cholesterol metabolism. Our results support the importance of
   cholesterol-rich lipid rafts on the cell membrane for bacterial
   pathogen or virulence factor adhesion in the H. pylori infection
   process ([121]18). On the other hand, the significant alterations of
   lipid and cholesterol metabolism pathways after H. pylori eradication
   may also be associated with lipid metabolism reprogramming and
   lipoprotein-mediated cholesterol entry in GC progression ([122]19).
   However, further validation of such mechanisms is needed.

   H. pylori infection and eradication were reported to be associated with
   gut microbiome-metabolome interactions and alterations of serum
   metabolites ([123]20[124]–[125]22). However, the effects of H. pylori
   infection on the interplays between the gastric microbiota and
   metabolites still remain unclear. In the present study, we found many
   significant correlations between the differential microbiota and
   metabolites in H. pylori-positive baseline biopsy specimens, which can
   be changed by eradication. The dominating positive correlations in
   baseline were converted to relatively more negative correlations after
   treatment. Helicobacter was notable for negative correlations with
   glycerophospholipids, glycosylceramides, and triacylglycerols in
   baseline but for positive correlations with glycerophospholipids and
   sphingolipids after treatment.

   In addition to anti-H. pylori treatment, gastric lesions are also
   associated with microbiota-metabolite interactions. The characteristic
   negative correlations between glycosylceramides and Fusobacterium,
   Streptococcus, and Gemella in H. pylori-positive baseline biopsy
   specimens were also noticed in active gastritis and IM lesions compared
   to respective references. Furthermore, positive correlations between