Gut bacteria linked to breast cancer show striking shifts in diversity and function

Scientists have long suspected that the trillions of microbes living in the human gut do far more than aid digestion, and a new computational study now adds weight to the idea that these bacterial communities may be intertwined with one of the most common cancers in the world. In research published in BMC Bioinformatics, a team of Iranian scientists reanalyzed publicly available 16S rRNA gene sequencing data from 804 fecal samples, comparing 386 samples from people with breast cancer against 386…

The scale of the dataset underlying this study deserves particular attention. With 804 fecal samples drawn from the PRJNA658160 archive, the analysis ranks among the larger 16S rRNA-based investigations of the breast cancer gut microbiome published to date. Many earlier studies in this field relied on cohorts of a few dozen participants, which made it difficult to distinguish consistent microbial signals from individual variation. By pooling hundreds of cases and controls, the researchers increased the statistical power available for detecting genus-level differences, although pooling publicly archived data also means the samples were collected under different protocols and conditions that cannot be fully standardized after the fact.

The choice of the 16S rRNA gene V4 region reflects a common compromise in microbiome research. This short hypervariable segment can be sequenced quickly and inexpensively, and it provides enough resolution to classify bacteria at the genus level in most cases. However, it often cannot reliably distinguish closely related species, and it captures only bacteria whose primers match the targeted region. This is one reason the authors framed their pathway-level results as predicted functional potential rather than measured metabolic activity: PICRUSt2 works by matching the observed 16S profiles against reference genomes and inferring which genes and pathways the community is likely to carry, a process that necessarily involves assumptions about how closely related organisms share metabolic capabilities.

The direction of the taxonomic findings aligns intriguingly with hypotheses that have circulated in cancer microbiology for years. Bacteroides, one of the genera enriched in the breast cancer samples in this analysis, is a dominant group in Western-style gut communities and has been implicated in inflammatory and immune-modulating processes in previous work. Fusobacterium, another enriched genus, has attracted considerable attention in colorectal cancer research, where it has been found enriched within tumors themselves, so its elevation in the stool of breast cancer patients adds a cross-cancer dimension to its profile. On the other side, Prevotella and Coprococcus, which were more abundant in healthy controls, are frequently associated with fiber-rich diets and the production of short-chain fatty acids, metabolites that have been linked to anti-inflammatory signaling and immune regulation.

The network analysis component of the study offers a perspective that single-taxon comparisons cannot provide. Rather than asking which bacteria differ in abundance, network methods such as SPIEC-EASI estimate which taxa tend to co-occur or co-exclude across samples, building a web of inferred ecological relationships. In the breast cancer-associated network, Bacteroides emerged as a hub-like taxon, meaning it showed many connections to other members of the community and may occupy a central position in the inferred ecology. Erysipelatoclostridium, described as bridge-like, appears to connect groups of taxa that would otherwise remain separate in the network structure. Such roles are descriptive rather than causal, but they can suggest which organisms might be worth prioritizing in follow-up experiments aimed at understanding community-level dynamics.

The predicted functional results add a metabolic layer to the taxonomic picture. The enrichment of pathways related to carbohydrate degradation, the glyoxylate cycle, and fatty acid beta-oxidation in breast cancer samples suggests a predicted shift toward the breakdown of complex carbohydrates and lipids, while stress-related metabolic pathways hint at an environment where microbial communities face oxidative or other pressures. Conversely, the control-enriched pathways involving menaquinone biosynthesis, biotin biosynthesis, and fermentation point toward vitamin production and fermentative metabolism as features more characteristic of the healthy gut in this dataset. Menaquinones, a class of vitamin K compounds produced by bacteria, and biotin, a B vitamin, are both microbial products that can influence host physiology, though the study cannot establish whether these predicted differences translate into altered metabolite availability in the gut.

The statistical nuances reported in the study merit emphasis for readers interpreting the results. The finding that beta diversity differed significantly between groups, with a PERMANOVA R-squared of 0.013, indicates that group membership explains only a small fraction of the overall variation in predicted pathway profiles. An effect size of this magnitude is typical in human microbiome research, where individual variation is enormous and influenced by diet, medication, age, geography, and countless other factors. More importantly, the authors observed that beta-dispersion was significantly greater in the breast cancer samples, meaning that the cancer group was more internally heterogeneous than the control group. This raises the possibility that some of the apparent group difference reflects increased variability within the cancer cohort rather than a uniform shift in community composition, a distinction that careful microbiome studies increasingly recognize as essential.

The reduction in alpha diversity observed in the breast cancer samples is consistent with a broader pattern seen across many disease-associated microbiome studies, where reduced diversity is often interpreted as a sign of ecological simplification. A less diverse community may be more vulnerable to perturbation and may provide fewer functional redundancies, though the relationship between diversity and health is not universal and depends heavily on context. The fact that genus-level beta diversity did not show strong global separation between the groups, while alpha diversity and specific taxa did differ, suggests that the dysbiosis associated with breast cancer in this dataset is selective rather than a wholesale restructuring of the community.

The authors’ call for validation through shotgun metagenomics and metabolomics reflects the current trajectory of the field. Shotgun sequencing reads the entire genetic content of the microbial community, allowing species-level resolution and direct gene-based functional inference rather than prediction from 16S profiles. Metabolomics, meanwhile, measures the actual chemical products of microbial metabolism, which is the level at which any influence on cancer biology would ultimately operate. Independent cohort replication is equally important, since microbiome associations are notoriously sensitive to population-specific factors such as dietary patterns, which differ substantially between the populations represented in publicly archived datasets and the broader global population of breast cancer patients.

It is also worth noting what the study design cannot address. Because the data are cross-sectional, comparing patients with cancer against healthy controls at a single time point, the findings cannot determine whether the observed microbial differences contribute to cancer development or are instead consequences of the disease, its treatment, or the altered lifestyle and dietary patterns that often accompany a cancer diagnosis. Chemotherapy, antibiotics, and other medications commonly used in cancer care can profoundly reshape the gut microbiome, and without longitudinal sampling or treatment-naive cohorts, disentangling cause from effect remains impossible. The authors appropriately frame their results as candidate signatures requiring further investigation rather than established biomarkers or causal factors.

Within the broader landscape of microbiome-cancer research, this study contributes a computational, multi-layered characterization that integrates taxonomy, predicted function, and network structure in a single framework. The combination of ANCOM-BC2 for taxonomic differential abundance and LinDA for pathway-level testing represents a methodologically deliberate approach, as both methods were developed to handle the compositional nature of sequencing data and the confounding issues that plague simpler statistical tests. The use of Spearman correlation for pathway networks, in contrast to the SPIEC-EASI approach for taxonomic networks, reflects the different statistical properties of the two data types, though correlation-based networks carry their own caveats regarding indirect associations and the inability to establish directionality.

For clinicians and researchers following the microbiome-cancer field, the practical significance of studies like this one lies in hypothesis generation. The specific genera and pathways identified here, including Bacteroides, Erysipelatoclostridium, Fusobacterium, and the enriched metabolic categories, provide a shortlist of targets that can now be examined in more controlled settings, whether through targeted metabolite measurements, strain-level genomic analysis, or interventional studies in model systems. The open-access availability of the underlying data through public archives also means that other research groups can independently reanalyze the same samples with alternative methods, a form of transparency that strengthens the overall reliability of the field’s cumulative findings.

As the field moves forward, the integration of multi-omics approaches, combining microbial sequencing with host transcriptomics, metabolomics, and clinical metadata, will be essential for moving from association to mechanism. The present study, by carefully documenting both the strengths and the limitations of 16S-based predicted functional profiling, provides a useful template for how such integrative analyses can be conducted and reported, and its findings offer a foundation upon which more definitive investigations of the gut microbiome’s role in breast cancer can be built.

Subject of Research: Gut microbiome

Article Title: Taxonomic, predicted functional, and network-based characterization of the breast cancer-associated gut microbiome

Article References: Nashtahosseini, Z., Navari, M., Sohrabei, S., Aldaghi, T., & Nazari, E. (2026). Taxonomic, predicted functional, and network-based characterization of the breast cancer-associated gut microbiome. BMC Bioinformatics. https://doi.org/10.1186/s12859-026-06713-8

Image Credits: AI Generated

DOI: 10.1186/s12859-026-06713-8

Keywords: breast cancer, gut microbiome, 16S rRNA, PICRUSt2, alpha diversity, dysbiosis, microbiome networks, Bacteroides, Fusobacterium, metagenomics, Taxonomic, predicted