Intestinal Bacteria Fluctuating in Early-Stage Colorectal Cancer Carcinogenesis are Associated with Diet in Healthy Adults.

This hospital-based, cross-sectional study aimed to explore the association between diet and fluctuating intestinal bacteria in early-stage colorectal cancer (CRC) (Atopobium parvulum, Actinomyces odontolyticus, Solobacterium moorei, and Bifidobacterium longum). Healthy participants (n = 212) who underwent total colonoscopy at National Cancer Center Hospital (Tokyo, Japan) were divided into two groups according to the relative abundance of bacteria in their feces: those in the top 25% of relative bacterial abundance as cases and the bottom 25% as controls. The participants were divided into three groups (low, medium, and high) according to their intake of food groups associated with CRC. Multivariable logistic regression analysis was conducted to estimate the association between dietary intake and higher relative abundance of bacteria. Dairy products were inversely associated with a higher relative abundance of A. parvulum, A. odontolyticus, and S. moorei, with odds ratios (high vs. low) and 95% confidence interval as follows: 0.16 (0.06-0.44), 0.25 (0.08-0.82), and 0.29 (0.11-0.78), respectively. Additionally, dietary fiber was inversely associated with a higher relative abundance of S.moorei (0.29 [0.11-0.78]). No association was observed between diet and B.longum. In conclusion, healthy adults with a higher intake of dairy products and fiber had lower odds of having a higher relative abundance of CRC-associated microbiota.

Leveraging explainable AI for gut microbiome-based colorectal cancer classification.

Studies have shown a link between colorectal cancer (CRC) and gut microbiome compositions. In these studies, machine learning is used to infer CRC biomarkers using global explanation methods. While these methods allow the identification of bacteria generally correlated with CRC, they fail to recognize species that are only influential for some individuals. In this study, we investigate the potential of Shapley Additive Explanations (SHAP) for a more personalized CRC biomarker identification. Analyses of five independent datasets show that this method can even separate CRC subjects into subgroups with distinct CRC probabilities and bacterial biomarkers.

Recent advances of machine learning applications in human gut microbiota study: from observational analysis toward causal inference and clinical intervention.

Statistical methods, especially machine learning, learning(ML), are pivotal for the analyses of large data generated by multiomics human gut microbiota study. These analyses lead to the discovery of microbe-disease associations. Furthermore, recent efforts for more data transparency and accessible analytical tools improved data availability and study reproducibility. Our recent accumulated knowledge on microbe-disease associations brings light to the next questions: what is the role of microbes in disease progression and how can we apply our knowledge of microbiome in clinical settings? Here, we introduce recent studies that implemented ML to answer the questions of causal inference and clinical translation.

Association Between Diet and Fusobacterium nucleatum in the Feces of Healthy Adults: A Hospital-based Cross-sectional Study.

Fusobacterium nucleatum is involved in the development and progression of colorectal cancer. Although the gut microbiota is influenced by diet, studies on the association between diet and F. nucleatum are limited. We aimed to evaluate the association between various dietary factors and fecal F. nucleatum in healthy adults without a history of colorectal cancer or precancerous lesions. This was a cross-sectional study. Subjects who underwent total colonoscopy at the National Cancer Center Hospital (Tokyo, Japan) were included. Healthy subjects (n = 212) were divided into two groups according to the presence or absence of F. nucleatum in their feces which was calculated from data of whole-genome shotgun sequencing, with the group with F. nucleatum serving as cases and the group without F. nucleatum serving as controls. Multivariable logistic regression analysis adjusted potential confounders was conducted to estimate the associations between dietary intake and nutrients estimated by a validated food frequency questionnaire and the presence of F. nucleatum in the feces. There was a significant inverse association between dairy products and the presence of fecal F. nucleatum [high vs. low; OR, 0.41; 95% confidence interval, 0.17-0.95; Ptrend, 0.039]. These results may have important implications for colorectal cancer prevention through nutritional intervention. PREVENTION RELEVANCE: F. nucleatum is well known as a colorectal cancer-associated bacterium. Dietary habits alter the composition and function of the intestinal microbiota. A high intake of dairy products in healthy adults may reduce F. nucleatum and prevent colorectal cancer.

Mediation effect of intestinal microbiota on the relationship between fiber intake and colorectal cancer.

Higher fiber intake has been associated with a lower risk of colorectal cancer (CRC) and has been shown to protect against CRC based on probable evidence. Recent studies revealed a possible mechanism whereby the interaction between intestinal microbiota and fiber intake mediates CRC risk. However, the specific intestinal bacteria and the amount of these bacteria involved in this mechanism are not fully known. Therefore, this single-center study aimed to determine whether specific intestinal bacteria mediated the relationship between fiber intake and CRC risk. We enrolled patients who received colonoscopy at National Cancer Center Hospital. This cross-sectional study included 180 patients with clinically diagnosed CRC and 242 controls. We conducted a causal mediation analysis to assess the natural indirect effect and natural direct effect of specific intestinal bacteria on association between fiber intake and CRC risk. The median age was 64 (interquartile range, 54-70) years, and 58% of the participants were males. We used metagenomics for profiling gut microbiomes. The relative abundance of each species in each sample was calculated. Among the candidate, Fusobacterium nucleatum and Gemella morbillorum had a significant natural indirect effect based on their highest fiber intake compared to the lowest fiber intake, with a risk difference (95% confidence interval, proportion of mediation effect) of -0.06 [-0.09 to -0.03, 23%] and -0.03 [-0.06 to -0.01, 10.5%], respectively. Other bacteria did not display natural indirect effects. In conclusion, Fusobacterium nucleatum and Gemella morbillorum were found to mediate the relationship between fiber intake and CRC risk.

Association between Diet and Fusobacterium nucleatum in the Feces of Healthy Adults: A hospital-based cross-sectional study.

Fusobacterium nucleatum is involved in the development and progression of colorectal cancer. Although the gut microbiota is influenced by diet, studies on the association between diet and F. nucleatum are limited. We aimed to evaluate the association between various dietary factors and fecal F. nucleatum in healthy adults without a history of colorectal cancer or precancerous lesions. This was a cross-sectional study. Subjects who underwent total colonoscopy at the National Cancer Center Hospital (Tokyo, Japan) were included. Healthy subjects (n = 212) were divided into two groups according to the presence or absence of F. nucleatum in their feces which was calculated from data of whole-genome shotgun sequencing, with the group with F. nucleatum serving as cases and the group without F. nucleatum serving as controls. Multivariable logistic regression analysis adjusted potential confounders was conducted to estimate the associations between dietary intake and nutrients estimated by a validated food frequency questionnaire and the presence of F. nucleatum in the feces. There was a significant inverse association between dairy products and the presence of fecal F. nucleatum (High vs. Low, OR 0.41, 95% CI 0.17-0.95, P for trend 0.039). These results may have important implications for colorectal cancer prevention through nutritional intervention.

Surgical Treatment for Colorectal Cancer Partially Restores Gut Microbiome and Metabolome Traits.

Accumulating evidence indicates that the gut microbiome and metabolites are associated with colorectal cancer (CRC). However, the influence of surgery for CRC treatment on the gut microbiome and metabolites and how it relates to CRC risk in postoperative CRC patients remain partially understood. Here, we collected 170 fecal samples from 85 CRC patients pre- and approximately 1 year postsurgery and performed shotgun metagenomic sequencing and capillary electrophoresis-time of flight mass spectrometry-based metabolomics analyses to characterize alterations between pre- and postsurgery. We determined that the relative abundance of 114 species was altered postsurgery (P < 0.005). CRC-associated species, such as Fusobacterium nucleatum, were decreased postsurgery. On the other hand, Clostridium scindens, carcinogenesis-associated deoxycholate (DCA)-producing species, and its biotransformed genes (bai operon) were increased postsurgery. The concentration of 60 fecal metabolites was also altered postsurgery (P < 0.005). Two bile acids, cholate and DCA, were increased postsurgery. We developed methods to estimate postoperative CRC risk based on the gut microbiome and metabolomic compositions using a random forest machine-learning algorithm that classifies large adenoma or early-stage CRC and healthy controls from publicly available data sets. We applied methods to preoperative samples and then compared the estimated CRC risk between the two groups according to the presence of large adenoma or tumors 5 years postsurgery (P < 0.05). Overall, our results show that the gut microbiome and metabolites dynamically change from pre- to postsurgery. In postoperative CRC patients, potential CRC risk derived from gut microbiome and metabolites still remains, which indicates the importance of follow-up assessments. IMPORTANCE The gut microbiome and metabolites are associated with CRC progression and carcinogenesis. Postoperative CRC patients are reported to be at an increased CRC risk; however, how gut microbiome and metabolites are related to CRC risk in postoperative patients remains only partially understood. In this study, we investigated the influence of surgical CRC treatment on the gut microbiome and metabolites. We found that the CRC-associated species Fusobacterium nucleatum was decreased postsurgery, whereas carcinogenesis-associated DCA and its producing species and genes were increased postsurgery. We developed methods to estimate postoperative CRC risk based on the gut microbiome and metabolomic compositions. We applied methods to compare the estimated CRC risk between two groups according to the presence of large adenoma or tumors after 5 years postsurgery. To our knowledge, this study is the first report on differences between pre- and postsurgery using metagenomics and metabolomics data analysis. Our methods might be used for CRC risk assessment in postoperative patients.

Significance of the gut microbiome in multistep colorectal carcinogenesis.

Colorectal cancer (CRC) is highly prevalent worldwide. In 2018, there were over 1.8 million new cases. Most sporadic CRC develop from polypoid adenomas and are preceded by intramucosal carcinoma (stage 0), which can progress into more malignant forms. This developmental process is known as the adenoma-carcinoma sequence. Early detection and endoscopic removal are crucial for CRC management. Accumulating evidence suggests that the gut microbiota is associated with CRC development in humans. Comprehensive characterization of this microbiota is of great importance to assess its potential as a diagnostic marker in the very early stages of CRC. In this review, we summarized recent studies on CRC-associated bacteria and their carcinogenic mechanisms in animal models, human cell lines and human cohorts. High-throughput technologies have facilitated the identification of CRC-associated bacteria in human samples. We have presented our metagenome and metabolome studies on fecal samples collected from a large Japanese cohort that revealed stage-specific phenotypes of the microbiota in CRC. Furthermore, we have discussed the potential carcinogenic mechanisms of the gut microbiota, from which we can infer whether changes in the gut microbiota are a cause or effect in the multi-step process of CRC carcinogenesis.

Influence of gastrectomy for gastric cancer treatment on faecal microbiome and metabolome profiles.

OBJECTIVE: Recent evidence points to the gut microbiome's involvement in postoperative outcomes, including after gastrectomy. Here, we investigated the influence of gastrectomy for gastric cancer on the gut microbiome and metabolome, and how it related to postgastrectomy conditions. DESIGN: We performed shotgun metagenomics sequencing and capillary electrophoresis time-of-flight mass spectrometry-based metabolomics analyses on faecal samples collected from participants with a history of gastrectomy for gastric cancer (n=50) and compared them with control participants (n=56). RESULTS: The gut microbiota in the gastrectomy group showed higher species diversity and richness (p<0.05), together with greater abundance of aerobes, facultative anaerobes and oral microbes. Moreover, bile acids such as genotoxic deoxycholic acid and branched-chain amino acids were differentially abundant between the two groups (linear discriminant analysis (LDA) effect size (LEfSe): p<0.05, q<0.1, LDA>2.0), as were also Kyoto Encyclopedia of Genes and Genomes modules involved in nutrient transport and organic compounds biosynthesis (LEfSe: p<0.05, q<0.1, LDA>2.0). CONCLUSION: Our results reveal alterations of gut microbiota after gastrectomy, suggesting its association with postoperative comorbidities. The multi-omic approach applied in this study could complement the follow-up of patients after gastrectomy.

Author Correction: Metagenomic analysis of colorectal cancer datasets identifies cross-cohort microbial diagnostic signatures and a link with choline degradation.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Metagenomic and metabolomic analyses reveal distinct stage-specific phenotypes of the gut microbiota in colorectal cancer.

In most cases of sporadic colorectal cancers, tumorigenesis is a multistep process, involving genomic alterations in parallel with morphologic changes. In addition, accumulating evidence suggests that the human gut microbiome is linked to the development of colorectal cancer. Here we performed fecal metagenomic and metabolomic studies on samples from a large cohort of 616 participants who underwent colonoscopy to assess taxonomic and functional characteristics of gut microbiota and metabolites. Microbiome and metabolome shifts were apparent in cases of multiple polypoid adenomas and intramucosal carcinomas, in addition to more advanced lesions. We found two distinct patterns of microbiome elevations. First, the relative abundance of Fusobacterium nucleatum spp. was significantly (P < 0.005) elevated continuously from intramucosal carcinoma to more advanced stages. Second, Atopobium parvulum and Actinomyces odontolyticus, which co-occurred in intramucosal carcinomas, were significantly (P < 0.005) increased only in multiple polypoid adenomas and/or intramucosal carcinomas. Metabolome analyses showed that branched-chain amino acids and phenylalanine were significantly (P < 0.005) increased in intramucosal carcinomas and bile acids, including deoxycholate, were significantly (P < 0.005) elevated in multiple polypoid adenomas and/or intramucosal carcinomas. We identified metagenomic and metabolomic markers to discriminate cases of intramucosal carcinoma from the healthy controls. Our large-cohort multi-omics data indicate that shifts in the microbiome and metabolome occur from the very early stages of the development of colorectal cancer, which is of possible etiological and diagnostic importance.

Meta-analysis of fecal metagenomes reveals global microbial signatures that are specific for colorectal cancer.

Association studies have linked microbiome alterations with many human diseases. However, they have not always reported consistent results, thereby necessitating cross-study comparisons. Here, a meta-analysis of eight geographically and technically diverse fecal shotgun metagenomic studies of colorectal cancer (CRC, n = 768), which was controlled for several confounders, identified a core set of 29 species significantly enriched in CRC metagenomes (false discovery rate (FDR) < 1 × 10-5). CRC signatures derived from single studies maintained their accuracy in other studies. By training on multiple studies, we improved detection accuracy and disease specificity for CRC. Functional analysis of CRC metagenomes revealed enriched protein and mucin catabolism genes and depleted carbohydrate degradation genes. Moreover, we inferred elevated production of secondary bile acids from CRC metagenomes, suggesting a metabolic link between cancer-associated gut microbes and a fat- and meat-rich diet. Through extensive validations, this meta-analysis firmly establishes globally generalizable, predictive taxonomic and functional microbiome CRC signatures as a basis for future diagnostics.

Metagenomic analysis of colorectal cancer datasets identifies cross-cohort microbial diagnostic signatures and a link with choline degradation.

Several studies have investigated links between the gut microbiome and colorectal cancer (CRC), but questions remain about the replicability of biomarkers across cohorts and populations. We performed a meta-analysis of five publicly available datasets and two new cohorts and validated the findings on two additional cohorts, considering in total 969 fecal metagenomes. Unlike microbiome shifts associated with gastrointestinal syndromes, the gut microbiome in CRC showed reproducibly higher richness than controls (P < 0.01), partially due to expansions of species typically derived from the oral cavity. Meta-analysis of the microbiome functional potential identified gluconeogenesis and the putrefaction and fermentation pathways as being associated with CRC, whereas the stachyose and starch degradation pathways were associated with controls. Predictive microbiome signatures for CRC trained on multiple datasets showed consistently high accuracy in datasets not considered for model training and independent validation cohorts (average area under the curve, 0.84). Pooled analysis of raw metagenomes showed that the choline trimethylamine-lyase gene was overabundant in CRC (P = 0.001), identifying a relationship between microbiome choline metabolism and CRC. The combined analysis of heterogeneous CRC cohorts thus identified reproducible microbiome biomarkers and accurate disease-predictive models that can form the basis for clinical prognostic tests and hypothesis-driven mechanistic studies.

Minor taxa in human skin microbiome contribute to the personal identification.

The human skin microbiome can vary over time, and inter-individual variability of the microbiome is greater than the temporal variability within an individual. The skin microbiome has become a useful tool to identify individuals, and one type of personal identification using the skin microbiome has been reported in a community of less than 20 individuals. However, identification of individuals based on the skin microbiome has shown low accuracy in communities larger than 80 individuals. Here, we developed a new approach for personal identification, which considers that minor taxa are one of the important factors for distinguishing between individuals. We originally established a human skin microbiome for 66 samples from 11 individuals over two years (33 samples each year). Our method could classify individuals with 85% accuracy beyond a one-year sampling period. Moreover, we applied our method to 837 publicly available skin microbiome samples from 89 individuals and succeeded in identifying individuals with 78% accuracy. In short, our results investigate that (i) our new personal identification method worked well with two different communities (our data: 11 individuals; public data: 89 individuals) using the skin microbiome, (ii) defining the personal skin microbiome requires samples from several time points, (iii) inclusion of minor skin taxa strongly contributes to the effectiveness of personal identification.

Large-Scale Prediction of Beneficial Drug Combinations Using Drug Efficacy and Target Profiles.

The identification of beneficial drug combinations is a challenging issue in pharmaceutical and clinical research toward combinatorial drug therapy. In the present study, we developed a novel computational method for large-scale prediction of beneficial drug combinations using drug efficacy and target profiles. We designed an informative descriptor for each drug-drug pair based on multiple drug profiles representing drug-targeted proteins and Anatomical Therapeutic Chemical Classification System codes. Then, we constructed a predictive model by learning a sparsity-induced classifier based on known drug combinations from the Orange Book and KEGG DRUG databases. Our results show that the proposed method outperforms the previous methods in terms of the accuracy of high-confidence predictions, and the extracted features are biologically meaningful. Finally, we performed a comprehensive prediction of novel drug combinations for 2,639 approved drugs, which predicted 142,988 new potentially beneficial drug-drug pairs. We showed several examples of successfully predicted drug combinations for a variety of diseases.

Target-Based Drug Repositioning Using Large-Scale Chemical-Protein Interactome Data.

Drug repositioning, or the identification of new indications for known drugs, is a useful strategy for drug discovery. In this study, we developed novel computational methods to predict potential drug targets and new drug indications for systematic drug repositioning using large-scale chemical-protein interactome data. We explored the target space of drugs (including primary targets and off-targets) based on chemical structure similarity and phenotypic effect similarity by making optimal use of millions of compound-protein interactions. On the basis of the target profiles of drugs, we constructed statistical models to predict new drug indications for a wide range of diseases with various molecular features. The proposed method outperformed previous methods in terms of interpretability, applicability, and accuracy. Finally, we conducted a comprehensive prediction of the drug-target-disease association network for 8270 drugs and 1401 diseases and showed biologically meaningful examples of newly predicted drug targets and drug indications. The predictive model is useful to understand the mechanisms of the predicted drug indications.

Systematic drug repositioning for a wide range of diseases with integrative analyses of phenotypic and molecular data.

Drug repositioning, or the application of known drugs to new indications, is a challenging issue in pharmaceutical science. In this study, we developed a new computational method to predict unknown drug indications for systematic drug repositioning in a framework of supervised network inference. We defined a descriptor for each drug-disease pair based on the phenotypic features of drugs (e.g., medicinal effects and side effects) and various molecular features of diseases (e.g., disease-causing genes, diagnostic markers, disease-related pathways, and environmental factors) and constructed a statistical model to predict new drug-disease associations for a wide range of diseases in the International Classification of Diseases. Our results show that the proposed method outperforms previous methods in terms of accuracy and applicability, and its performance does not depend on drug chemical structure similarity. Finally, we performed a comprehensive prediction of a drug-disease association network consisting of 2349 drugs and 858 diseases and described biologically meaningful examples of newly predicted drug indications for several types of cancers and nonhereditary diseases.

Pharmacoepidemiological characterization of drug-induced adverse reaction clusters towards understanding of their mechanisms.

A big challenge in pharmacology is the understanding of the underlying mechanisms that cause drug-induced adverse reactions (ADRs), which are in some cases similar to each other regardless of different drug indications, and are in other cases different regardless of same drug indications. The FDA Adverse Event Reporting System (FAERS) provides a valuable resource for pharmacoepidemiology, the study of the uses and the effects of drugs in large human population. However, FAERS is a spontaneous reporting system that inevitably contains noise that deviates the application of conventional clustering approaches. By performing a biclustering analysis on the FAERS data we identified 163 biclusters of drug-induced adverse reactions, counting for 691 ADRs and 240 drugs in total, where the number of ADR occurrences are consistently high across the associated drugs. Medically similar ADRs are derived from several distinct indications for use in the majority (145/163=88%) of the biclusters, which enabled us to interpret the underlying mechanisms that lead to similar ADRs. Furthermore, we compared the biclusters that contain same drugs but different ADRs, finding the cases where the populations of the patients were different in terms of age, sex, and body weight. We applied a biclustering approach to catalogue the relationship between drugs and adverse reactions from a large FAERS data set, and demonstrated a systematic way to uncover the cases different drug administrations resulted in similar adverse reactions, and the same drug can cause different reactions dependent on the patients' conditions.

Inferring protein domains associated with drug side effects based on drug-target interaction network.

BACKGROUND: Most phenotypic effects of drugs are involved in the interactions between drugs and their target proteins, however, our knowledge about the molecular mechanism of the drug-target interactions is very limited. One of challenging issues in recent pharmaceutical science is to identify the underlying molecular features which govern drug-target interactions. RESULTS: In this paper, we make a systematic analysis of the correlation between drug side effects and protein domains, which we call "pharmacogenomic features," based on the drug-target interaction network. We detect drug side effects and protein domains that appear jointly in known drug-target interactions, which is made possible by using classifiers with sparse models. It is shown that the inferred pharmacogenomic features can be used for predicting potential drug-target interactions. We also discuss advantages and limitations of the pharmacogenomic features, compared with the chemogenomic features that are the associations between drug chemical substructures and protein domains. CONCLUSION: The inferred side effect-domain association network is expected to be useful for estimating common drug side effects for different protein families and characteristic drug side effects for specific protein domains.