The study, led by Edoardo Mandalà and Martina Coppini, enrolled twelve patients at the Oral Medicine Unit of the Paolo Giaccone Policlinico University Hospital in Palermo, Italy. Seven had confirmed oral squamous cell carcinoma, while five served as cancer-free controls, including two healthy individuals and two with oral leukoplakia, a potentially precancerous condition. After one control patient was identified as a biological outlier because their microbial profile clustered suspiciously close to the cancer group, eleven subjects remained in the final analysis. From each participant, the team collected paired samples: unstimulated morning saliva and a fresh fragment of tissue taken during incisional biopsy, frozen immediately at minus 80 degrees Celsius.

The technical core of the study was next-generation sequencing of the 16S ribosomal RNA gene, the workhorse marker of microbiome science. This gene contains hypervariable regions that act like molecular barcodes, allowing researchers to identify which bacteria are present in a sample. The team targeted the V1, V2, and V3 hypervariable regions, amplified them by polymerase chain reaction, and sequenced the libraries on an Illumina MiSeq platform. After taxonomic classification using the Illumina BaseSpace 16S Metagenomics app, more than 1.2 million species-classified reads were retained across the 22 analyzed samples, averaging roughly 55,000 classified reads per sample. That depth allowed the researchers to identify 1,711 unique bacterial species distributed across 737 genera, providing an unusually fine-grained census of the oral ecosystem in health and disease.

The first surprise was that the total number of bacterial species did not differ dramatically between cancer patients and controls. Alpha-diversity analysis, which measures species richness and evenness within a sample, showed no significant differences in saliva and only a trend toward higher diversity in tumor biopsies. But when the researchers examined beta diversity, which compares the overall structure of microbial communities between samples, a different story emerged. In saliva, the communities of cancer patients and controls looked largely similar, with no distinct clustering. In biopsy samples, however, the two groups separated noticeably along the primary axes of a principal coordinate analysis based on Bray-Curtis dissimilarity, with a statistical trend approaching significance. The dysbiotic shifts associated with cancer, in other words, are far more pronounced inside the tissue microenvironment than in the saliva bathing it.

At the genus level, the core microbiome of the cohort was dominated by Streptococcus, which accounted for nearly a quarter of all sequencing reads, followed by Prevotella and Neisseria. But the distribution of these genera across the four study groups revealed a striking reorganization. In control saliva, Streptococcus reigned at 28 percent abundance, while in cancer patients’ saliva it dropped to 17 percent, nearly matched by an expansion of Neisseria to 17 percent from just 4.4 percent in controls. In tissue, the contrast was even sharper. Control biopsies were heavily dominated by Streptococcus at 37 percent, whereas cancer biopsies showed Streptococcus reduced to 22 percent alongside a marked rise of Fusobacterium, which climbed from 1.7 percent in healthy tissue to 7.4 percent in tumors.

Resolving these genus-level shifts to the species level produced the study’s most consequential findings. In tumor tissue, Fusobacterium nucleatum, an anaerobic bacterium already infamous for its role in colorectal cancer, exhibited a 29-fold increase in abundance, rising from a negligible 0.2 percent in control tissues to 4.9 percent in tumors, a difference that reached nominal statistical significance. This pathogen is no passive bystander. Previous research has shown that it stimulates cancer cell proliferation through the E-cadherin and beta-catenin signaling pathway and promotes immune evasion by binding the TIGIT inhibitory receptor on host immune cells. Meanwhile, health-associated commensals were systematically eradicated: Streptococcus salivarius, the most abundant species in control biopsies at 11.5 percent, plummeted to 0.3 percent in tumors, while Streptococcus oralis fell from 8 percent to 1.4 percent. These streptococci normally maintain oral homeostasis by producing bacteriocins and hydrogen peroxide, so their loss represents a functional collapse of colonization resistance, potentially opening the door for opportunistic anaerobes to invade.

The salivary signature told a complementary but distinct story. In cancer patients’ saliva, Neisseria macacae surged tenfold, from 0.9 percent in controls to 9.8 percent, becoming the single most abundant species in the cancer salivary microbiome. Concurrently, Haemophilus parainfluenzae and Neisseria perflava rose, while the commensal Streptococcus salivarius collapsed from 9.4 percent to 1.6 percent. This finding is particularly intriguing because it suggests that saliva, far from being a perfect mirror of the tumor, reflects a broader ecosystem shift that could be exploited for non-invasive screening. The researchers propose that the imbalance between Streptococcus and Neisseria in saliva might serve as a candidate biomarker, distinct from the intratumoral pathogen load that requires a biopsy to measure.

Perhaps the most forward-looking result came from an exploratory analysis of Fusobacterium nucleatum sequence variants. In five of seven cancer patients, the tumor niche was dominated by a single 16S sequence variant corresponding to reference NR_117842.1, which maps to the type strain ATCC 25586 of the subspecies nucleatum. This strain is the model organism for Fusobacterium pathogenicity, known to express high levels of the FadA adhesin that binds E-cadherin on epithelial cells and activates beta-catenin signaling. In patients with the highest bacterial burden, this single variant accounted for more than 99 percent of all Fusobacterium reads, suggesting an active selection process driven by the malignant microenvironment rather than passive colonization. One patient instead carried a distinct dominant variant mapping to the subspecies vincentii, indicating that while the specific strain may vary between individuals, the functional requirement for a highly adhesive, virulent phenotype appears conserved. The authors caution that standard 16S sequencing lacks definitive strain-level resolution, so confirming true clonal expansion will require shotgun metagenomics or whole genome sequencing.

When the researchers mapped microbial profiles across the clinical disease continuum, from healthy mucosa through leukoplakia to early and advanced cancer, a staged ecological succession came into view. Fusobacterium remained at baseline levels in healthy tissue and leukoplakia lesions, then surged during early-stage tumors, reaching peaks of nearly 14 percent, before declining in advanced disease. Prevotella, by contrast, established a dominant presence in advanced-stage tumors, stabilizing at abundances above 10 percent. This pattern aligns with the driver-passenger model of tumor microbiology: Fusobacterium may act as an active driver in the initial phases of malignancy, anchoring itself to the nascent tumor through the Fap2 adhesin that binds Gal-GalNAc sugars overexpressed on transformed epithelial cells, while strict anaerobic passengers like Prevotella later colonize the hypoxic, necrotic core of the growing mass.

The authors are careful to frame these findings as hypothesis-generating rather than definitive. With only eleven analyzed subjects, the statistical power is limited, and several trends that appear biologically meaningful did not survive multiple testing corrections. The inclusion of leukoplakia patients in the control group adds biological heterogeneity, the lack of systematic periodontal indices precludes precise assessment of baseline dental health, and the cross-sectional design cannot establish whether microbial shifts cause cancer or merely accompany it. Yet the magnitude of the changes observed, particularly the 29-fold enrichment of Fusobacterium nucleatum in tumors and the near-total eradication of protective streptococci, provides a compelling preliminary signal. If validated in larger, longitudinal cohorts, this work could lay the foundation for a multi-marker diagnostic panel combining salivary microbial ratios with intratumoral pathogen quantification, and it adds momentum to a growing field exploring microbiome-based precision medicine in oncology, where engineered bacteria may one day be deployed to remodel the tumor microenvironment and enhance immunotherapy.

Subject of Research: Oral cancer tumors harbor a distinct dysbiotic bacterial ecosystem dominated by Fusobacterium nucleatum, while saliva shows a separate Neisseria-driven signature, according to a 16S rRNA sequencing pilot study.

Article Title: Analysis of 16S rRNA to evaluate taxonomic differences between salivary and intratumoral microbiota in patients affected by oral squamous cell carcinoma: a single-center pilot study

Article References: Mandalà, E., Coppini, M., Mauceri, R., Pizzo, M., Giammanco, G. M., Campisi, G., & Bonura, C. (2026). Analysis of 16S rRNA to evaluate taxonomic differences between salivary and intratumoral microbiota in patients affected by oral squamous cell carcinoma: a single-center pilot study. Frontiers in Bacteriology, 5, Article 1863255. https://doi.org/10.3389/fbrio.2026.1863255

Image Credits: AI Generated

DOI: 10.3389/fbrio.2026.1863255

Keywords: oral cancer, microbiome, Fusobacterium nucleatum, 16S rRNA sequencing, oral squamous cell carcinoma, tumor microenvironment, salivary biomarkers, dysbiosis, Analysis, rRNA, evaluate, taxonomic