Stage-dependent and heterogeneous tumor association of Fusobacterium nucleatum in Romanian patients with colon adenocarcinoma

Abstract

BACKGROUND

Colorectal cancer remains a major cause of cancer-related mortality worldwide, with a particularly high burden in Eastern Europe. Increasing evidence implicates the gut microbiome, especially Fusobacterium nucleatum (F. nucleatum), in colorectal carcinogenesis; however, tumor-associated microbial patterns are heterogeneous and population-specific. To date, colorectal cancer microbiome data from Romanian patients are lacking. We hypothesized that F. nucleatum tumor association in Romanian patients is heterogeneous and varies according to tumor-related biological context rather than showing uniform enrichment.

AIM

To investigate tumor-associated distribution patterns of F. nucleatum in Romanian patients with colon adenocarcinoma.

METHODS

This prospective observational pilot study included 15 patients undergoing curative-intent surgery for colon adenocarcinoma at a tertiary referral center. Paired tumor and adjacent non-tumoral colonic tissues were analyzed using quantitative real-time polymerase chain reaction for F. nucleatum DNA (FS17 assay) and microRNA-21 (miR-21) expression. Molecular data were integrated with tumor stage, nodal status, inflammatory markers, and clinicopathological variables using paired comparisons and exploratory statistical analyses.

RESULTS

F. nucleatum detection showed marked interindividual variability across paired samples. Tumor-predominant detection was observed in slightly more than half of the cases, without a significant overall difference between tumor and adjacent tissue (P = 0.82). FS17 tumor Ct values were significantly lower in T2 tumors compared with T3-T4 tumors (24.63 ± 1.45 vs 28.53 ± 3.85, P = 0.01), indicating higher bacterial signal in earlier-stage disease. No statistically significant associations were observed with nodal status, demographic variables, or surgical characteristics. miR-21 expression was increased in tumor tissue, but did not correlate with F. nucleatum detection.

CONCLUSION

Tumor association of F. nucleatum in colon adenocarcinoma is heterogeneous and stage-dependent rather than uniform. This pilot study provides the first paired tissue-based microbiome data from Romanian patients and establishes a foundation for larger, longitudinal investigations.

Key Words: Colorectal cancer; Fusobacterium nucleatum; Gut microbiome; Tumor heterogeneity; MicroRNA-21; Romanian population

Core Tip: This pilot study presents the first paired tumor-adjacent tissue analysis of Fusobacterium nucleatum in Romanian patients with colon adenocarcinoma. Using quantitative real-time polymerase chain reaction, we show that tumor-associated Fusobacterium nucleatum signals are heterogeneous and stage-dependent rather than uniformly enriched. Higher bacterial signal was observed in earlier T-stage tumors, while no consistent associations were identified with demographic, surgical, or microRNA-21 expression profiles. These findings provide initial population-specific insight into tumor-microbe interactions and establish a foundation for larger, longitudinal studies evaluating the clinical relevance of colorectal cancer-associated microbiota.

INTRODUCTION

Colorectal cancer (CRC) remains a major global health burden and is among the most frequently diagnosed malignancies worldwide, continuing to rank as a leading cause of cancer-related mortality despite advances in screening, molecular profiling, and therapy[1]. The steady rise in CRC incidence, together with persistent challenges related to its metastatic spread, recurrence, and therapeutic resistance, underscores the need for a more nuanced understanding of CRC biology that extends beyond classical genetic alterations[2-4].

The burden is particularly evident in Eastern Europe, where CRC mortality rates remain disproportionately high[5,6]. In Romania, CRC represents a substantial proportion of newly diagnosed cancers, with a significant number of patients presenting at advanced stages, reflecting persistent gaps in early detection and risk stratification. In parallel, the increasing incidence of early-onset CRC further emphasizes the contribution of environmental and lifestyle-related exposures that act in concert with host susceptibility to accelerate colorectal carcinogenesis[7,8].

Within this evolving framework, CRC is increasingly viewed as a disease shaped by complex, context-dependent interactions between host genetics, environmental factors, and the gut microbiome. Rather than functioning as a passive bystander, the intestinal microbiota can actively modulate epithelial integrity, immune surveillance, and oncogenic signaling through its metabolic and inflammatory outputs[9-11]. Among CRC-associated microbes, Fusobacterium nucleatum (F. nucleatum) has consistently emerged as a tumor-associated bacterium implicated in immune modulation, metastatic behavior, and resistance to therapy[12-15]. Importantly, accumulating evidence indicates that F. nucleatum tumor association is heterogeneous, varies across patients and tumor contexts, and is not uniformly observed across all CRCs[16,17].

Despite the rapid advances in CRC microbiome research across Western Europe, North America, and East Asia, CRC-specific microbiome data from the Romanian population are currently lacking. As a result, population-specific patterns of tumor-microbe interaction remain poorly defined. While indirect insights can be drawn from metabolic and population-based cohorts, such data do not capture tumor-centered microbial dynamics[16,18]. In this regard, paired analyses of tumor and adjacent non-tumoral tissues represent a critical methodological approach for minimizing interindividual variability and for disentangling tumor-associated microbial signals from background microbiome variation[18].

Accordingly, this study aimed to characterize the distribution of F. nucleatum in paired tumor and adjacent non-tumoral colonic tissues from Romanian patients with colon adenocarcinoma and to explore its heterogeneous and context-dependent associations with tumor stage, nodal status, host molecular features [microRNA-21 (miR-21)], systemic inflammatory markers, and clinicopathological characteristics using qRT-PCR-based analyses.

MATERIALS AND METHODS

Study design and patient cohort

This pilot, prospective, observational study included 15 Romanian patients diagnosed with colon adenocarcinoma who underwent curative-intent surgical treatment at Colțea Clinical Hospital, Bucharest, Romania. Patient recruitment was performed preoperatively, and all participants provided written informed consent for the use of biological material for research purposes. The study protocol was approved by the hospital’s institutional authorities and conducted in accordance with the Declaration of Helsinki.

Eligibility criteria included histologically confirmed colon adenocarcinoma and availability of paired tumoral and adjacent non-tumoral colonic tissue. Clinical, demographic, paraclinical, and anatomopathological data, including age, sex, tumor localization, tumor-node-metastasis staging, inflammatory markers, and surgical characteristics, were prospectively collected for each patient.

Tissue sampling and biobanking

Immediately after surgical resection, paired samples of tumor tissue and macroscopically normal adjacent colonic tissue were collected. Tissue fragments (approximately 1 cm³) were evaluated and processed in the Department of Pathology, then aliquoted into labeled cryovials to ensure full traceability. Samples were snap-frozen and stored at −80 °C in the biorepository of Colțea Clinical Hospital until molecular processing.

Sample transport and laboratory processing

All biological samples were transported under controlled frozen conditions (−80 °C cold chain) to the School of Medicine, National and Kapodistrian University of Athens, Greece, where molecular analyses were performed. Upon arrival, samples were inspected for integrity, verified against accompanying documentation, and incorporated into the institutional laboratory biobank.

RNA and DNA extraction

Total RNA and genomic DNA were extracted from all paired tumor and adjacent tissue samples using standardized laboratory protocols. Controlled thawing and additional tissue fragmentation were performed when necessary to optimize extraction yield. RNA and DNA concentration and purity were assessed spectrophotometrically, and all extracted nucleic acids were recorded in an internal laboratory database, including sample code, tissue type, concentration, and extraction date.

cDNA synthesis and qRT-PCR for miR-21

Initial experiments for miR-21 expression analysis were conducted using a Solis reverse transcription kit for calibration purposes. Subsequently, all analyses were repeated using the miRCURY LNA RT Kit and miRCURY LNA miRNA PCR Assays (QIAGEN, Hilden, Germany), which provided improved sensitivity and reproducibility. miR-21 expression levels were quantified in paired tumor and adjacent tissue samples, with U6 small nuclear RNA used as an internal reference control.

Detection of F. nucleatum by qRT-PCR

Genomic DNA extracted from paired samples was used for the detection of F. nucleatum using an EvaGreen-based qRT-PCR approach. Several primer pairs described in the literature were initially tested. Based on specificity, amplification efficiency, melting curve analysis, and reproducibility, the FS17 primer set was selected for downstream analyses. Given the pilot design of the study and the limited sample size, F. nucleatum detection was analyzed using cycle threshold (Ct) values as a relative signal intensity metric to enable paired tumor-adjacent tissue comparisons within individual patients, rather than absolute bacterial quantification. This approach was adopted in accordance with the study design and the methodological framework agreed upon by the project coordinator for this initial population-specific investigation.

Statistical analysis

Statistical analyses were conducted within an exploratory framework, given the observational, non-randomized design of the study and the limited sample size (n = 15 patients). All analyses were performed using R statistical software.

RESULTS

Differential presence of F. nucleatum in tumor vs adjacent tissue

Quantitative real-time polymerase chain reaction analysis using the FS17 primer set revealed a heterogeneous, patient-specific distribution of F. nucleatum across paired tumor and adjacent non-tumoral colonic tissues. In slightly more than half of the cases, FS17 detection was higher in tumor tissue than in the corresponding adjacent mucosa, as reflected by lower Ct values. However, a substantial proportion of patients exhibited comparable or lower FS17 detection in tumor tissue, underscoring marked interindividual variability rather than uniform tumor enrichment (Figure 1A).

Figure 1 Scatter plot. A: FS17 detection in tumor vs adjacent colonic tissue. Scatter plot of paired quantitative real-time polymerase chain reaction cycle threshold (Ct) values for Fusobacterium nucleatum (F. nucleatum) detected with the FS17 primer set in tumor and adjacent non-tumoral colonic tissues. Lower Ct values indicate higher bacterial detection. Marked interindividual variability was observed, with higher tumor-associated FS17 detection in slightly more than half of the cases; B: Correlation between microRNA-21 (miR-21) expression and FS17 tumor Ct values. Scatter plot illustrating the relationship between miR-21 Ct values and FS17-derived F. nucleatum Ct values in tumor tissue. Pearson correlation analysis demonstrated a weak and non-significant association (r = −0.12, P = 0.66), indicating no direct relationship between miR-21 expression and tumor-associated bacterial detection; C: Correlation between miR-21 expression and FS17 Ct values in adjacent tissue. Scatter plot showing the association between miR-21 Ct values and FS17-derived F. nucleatum Ct values in adjacent non-tumoral colonic tissue. No significant correlation was observed (r = −0.09, P = 0.74), suggesting independent variation of miR-21 expression and bacterial detection in non-tumoral tissue. miR-21: MicroRNA-21.

This variability justified dichotomizing FS17 distribution into tumor-predominant vs adjacent-predominant patterns and underscores the context-dependent nature of microbial colonization in CRC.

Baseline demographic, clinical, and surgical characteristics

Baseline demographic, clinicopathological, and surgical characteristics of the study cohort are summarized in Table 1. The study included 15 patients with colon adenocarcinoma, with a median age of 67 (interquartile range [IQR]: 65-73) years. Most patients were male (73%), and 40% reported a history of smoking. Tumors were predominantly located in the right colon (53%) or sigmoid colon (40%). Preoperative tumor staging indicated that T3 tumors accounted for the largest proportion of cases (53%), while nodal involvement (N1-N2) was present in 73% of patients. With respect to surgical management, laparoscopic resection was performed in 67% of cases, hemicolectomy was the most common procedure (60%), and manual anastomosis was used in 73% of patients. All interventions were elective, with a median operative duration of 180 (IQR: 160-200) min.

Table 1 Baseline demographic, clinicopathological, and surgical characteristics, n (%).

Variable	Value
Age, median (IQR), years	67 (65-73)
Sex	Female 4 (27); male 11 (73)
Smoking status	Former 6 (40); never 9 (60)
Tumor localization	Right colon 8 (53); left colon 1 (7); sigmoid 6 (40)
Preoperative T stage	T2 6 (40); T3 8 (53); T4 1 (7)
Preoperative N stage	N0 4 (27); N1 8 (53); N2 3 (20)
Surgical approach	Laparoscopic 10 (67); open 5 (33)
Type of resection	Hemicolectomy 9 (60); segmental 6 (40)
Anastomosis type	Manual 11 (73); mechanical 4 (27)
Operative duration, median (IQR), min	180 (160-200)

IQR: Interquartile range.

Baseline laboratory and molecular parameters

Baseline laboratory and inflammatory parameters are summarized in Table 2, indicating overall low-to-moderate systemic inflammatory activity across the cohort. Median hemoglobin values were within the lower normal range, and white blood cell count, C-reactive protein (CRP), fibrinogen, and albumin levels did not indicate pronounced systemic inflammation at baseline, supporting clinical stability at the time of surgery.

Table 2 Baseline laboratory and inflammatory parameters.

Variable	Median (IQR)
Hemoglobin (g/dL)	11.92 (11.14-12.40)
White blood cell count (/µL)	7000 (6320-8360)
C-reactive protein (mg/dL)	0.70 (0.30-1.25)
Fibrinogen (mg/dL)	378 (327-421)
Albumin (g/dL)	4.15 (3.93-4.52)

IQR: Interquartile range.

Baseline molecular Ct values are detailed in Table 3, providing an overview of microRNA and bacterial detection signals in paired tissue samples. Median Ct values for miR-21 were higher than those for U6 small nuclear RNA, consistent with the use of U6 as an endogenous reference control rather than a biologically interpreted marker. FS17-derived Ct values for F. nucleatum detection were comparable between tumor and adjacent non-tumoral tissues at the cohort level, with overlapping IQRs. These baseline distributions establish the analytical reference for subsequent paired and stratified analyses.

Table 3 Molecular marker cycle threshold values in tumor and adjacent tissue.

Marker	Median (IQR)
MicroRNA-21	19.11 (17.09-21.90)
U6 small nuclear RNA	16.30 (15.60-17.46)
FS17 tumor tissue	25.44 (24.20-29.22)
FS17 adjacent tissue	25.10 (23.80-27.60)

IQR: Interquartile range.

miR-21 expression and associations

miR-21 expression stratified by sex and smoking status is presented in Table 4. Ct values tended to be slightly lower—reflecting higher expression—in female patients and former smokers; however, these differences were neither statistically nor clinically significant. No meaningful variation in miR-21 expression was observed across demographic subgroups.

Table 4 MicroRNA-21 and FS17 tumor cycle threshold values by sex and smoking status.

Group	mean Ct ± SD	Median (min-max)	P value
miR-21 - female	20.7 ± 2.77	20.9 (17.27-23.72)	0.43
miR-21 - male	19.27 ± 3.21	18.93 (16.32-26.49)
miR-21 - former smokers	20.06 ± 3.71	19.29 (16.32-26.49)	0.70
miR-21 - never smokers	19.38 ± 2.76	19.11 (16.33-23.72)
FS17 tumor - female	26.96 ± 3.99	25.56 (24.09-32.65)	0.99
FS17 tumor - male	26.97 ± 3.69	25.44 (23.10-34.31)
FS17 tumor - former smokers	26.88 ± 4.33	24.96 (23.10-32.65)	0.94
FS17 tumor - never smokers	27.03 ± 3.35	25.96 (24.09-34.31)

Ct: Cycle threshold; miR-21: MicroRNA-21.

Correlation analyses showed no significant association between miR-21 expression and FS17-detected F. nucleatum in either tumor or adjacent tissue. In tumor samples, a very weak and non-significant negative correlation was observed (r = −0.12, P = 0.66; Figure 1B), while adjacent tissue showed a similarly weak association (r = −0.09, P = 0.74; Figure 1C). Consistent with these findings, miR-21 expression did not differ significantly across preoperative T or N stages, with overlapping distributions illustrated in Figures 2A and 3A.

Figure 2 Cycle threshold values according to preoperative T stage. A: MicroRNA-21 (miR-21) cycle threshold (Ct) values according to preoperative T stage. Distribution of miR-21 Ct values stratified by preoperative tumor T stage. Data are presented as individual values with group distributions. Overlapping ranges across stages indicate no significant differences in miR-21 expression according to local tumor invasion; B: FS17 Ct values in adjacent non-tumoral tissue according to preoperative T stage. FS17-derived Ct values for Fusobacterium nucleatum (F. nucleatum) detection in adjacent non-tumoral colonic tissue stratified by preoperative T stage. No consistent stage-dependent pattern is observed, with overlapping distributions across tumor stages; C: FS17 tumor Ct values according to preoperative T stage. Boxplot showing FS17-derived Ct values for F. nucleatum detection in tumor tissue stratified by preoperative T stage. Lower Ct values in T2 tumors indicate higher bacterial detection compared with T3-T4 tumors, demonstrating a stage-dependent tumor association. miR-21: MicroRNA-21.

Figure 3 Cycle threshold values according to preoperative N stage. A: MicroRNA-21 (miR-21) cycle threshold (Ct) values according to preoperative N stage. miR-21 Ct values stratified by nodal status (N stage). Individual values demonstrate substantial overlap between nodal categories, indicating no significant association between miR-21 expression and lymph node involvement; B: FS17 tumor Ct values according to preoperative nodal status. FS17-derived Ct values for Fusobacterium nucleatum (F. nucleatum) detection in tumor tissue stratified by nodal stage (N0-N2). A progressive increase in Ct values is observed with advancing nodal involvement, although substantial overlap between groups is present; C: FS17 Ct values in adjacent non-tumoral tissue according to nodal status. Boxplot illustrating FS17-derived Ct values for F. nucleatum detection in adjacent non-tumoral tissue stratified by nodal stage. Overlapping distributions indicate no significant association between bacterial detection in adjacent tissue and nodal involvement. miR-21: MicroRNA-21.

FS17 expression in tumor tissue

FS17 tumor Ct values stratified by sex and smoking status are shown in Table 4, demonstrating comparable detection levels across these demographic categories. In contrast, FS17 expression differed significantly according to preoperative tumor stage. As shown in Table 5, T2 tumors exhibited significantly lower mean Ct values compared with T3-T4 tumors (24.63 ± 1.45 vs 28.53 ± 3.85; P = 0.01), indicating higher F. nucleatum detection in earlier-stage locally invasive disease. This stage-dependent pattern is illustrated in Figures 2B and 2C.

Table 5 FS17 tumor cycle threshold values by preoperative tumor and nodal stage.

Stage	mean Ct ± SD	Median (min-max)	P value
T2	24.63 ± 1.45	24.20 (23.10-26.80)	0.01
T3-T4	28.53 ± 3.85	29.03 (24.20-34.31)
N0	24.34 ± 1.19	24.14 (23.10-25.96)	0.20
N1	27.51 ± 3.98	26.06 (23.52-34.31)
N2	29.04 ± 3.61	29.03 (25.44-32.65)

Ct: Cycle threshold.

When analyzed by nodal status, FS17 tumor Ct values increased progressively from N0 to N2 stages (Table 5), corresponding to lower intratumoral bacterial detection with advancing nodal involvement; however, this trend did not reach statistical significance (P = 0.20). Substantial overlap between nodal groups is illustrated in Figures 3B and 3C.

Paired comparison and integrated analysis

Paired analysis revealed no statistically significant difference between FS17 Ct values in tumor and adjacent non-tumoral tissue at the cohort level. Individual paired measurements demonstrated pronounced interindividual variability without a consistent directional shift, reinforcing a heterogeneous, patient-specific pattern of tumor-microbe interaction.

Overall, these results indicate that F. nucleatum distribution in colon adenocarcinoma is heterogeneous and stage-dependent, shaped primarily by tumor-specific biological context rather than demographic or surgical factors alone.

DISCUSSION

This pilot study provides a comprehensive and integrative evaluation of F. nucleatum distribution in paired tumor and adjacent non-tumoral colonic tissues, examined in relation to host molecular markers, systemic inflammation, and clinicopathological characteristics in a Romanian cohort with colon adenocarcinoma. By combining paired quantitative analyses with exploratory statistical modeling, the present study emphasizes the heterogeneous and context-dependent nature of F. nucleatum tumor association rather than a uniform enrichment across CRCs. Importantly, analyses based on matched tumor-adjacent normal tissue pairs revealed that F. nucleatum enrichment was not consistently observed across patients, suggesting that its presence reflects selective colonization of permissive tumor niches rather than a generalized feature of colorectal malignancy. This interpretation aligns with recent genomic and microbiome studies demonstrating that only specific F. nucleatum lineages exhibit preferential tumor enrichment when compared with matched adjacent mucosa. Zepeda-Rivera et al[19] reported that enrichment was largely restricted to the Fna C2 clade, whereas other closely related subspecies were either equally represented or absent across tissue compartments, underscoring pronounced strain-level heterogeneity in tumor association. Such paired-sample observations provide a plausible explanation for discrepancies reported across cohorts and reinforce the notion that differences in strain composition, detection methods, and tumor microenvironmental context critically shape reported associations. Importantly, these observations should be viewed as hypothesis-generating rather than definitive evidence of tumor-specific enrichment patterns. In this exploratory setting, FS17-derived Ct values were therefore interpreted as relative detection signals for paired tissue comparison rather than as absolute measures of bacterial load, consistent with the pilot scope of the study and the staged research strategy. Moreover, mechanistic evidence indicates that the oncogenic and immunomodulatory effects attributed to F. nucleatum are strongly dependent on host molecular features, immune landscape, and microbial interactions, supporting a model in which tumor-normal differentials arise from context-specific microbial adaptation rather than uniform bacterial overrepresentation[20].

Within this framework of biological heterogeneity, a key observation in our cohort was the marked interindividual variability in FS17 expression between tumor and adjacent tissue. Although tumor-predominant FS17 expression was observed in slightly more than half of the cases, paired analyses did not demonstrate an overall statistically significant difference between tumor and adjacent tissue. This variability further supports the view that F. nucleatum colonization characterizes specific patient subsets rather than representing a universal feature of CRC. Consistent with this interpretation, large-scale microbiome studies and tissue-based analyses have shown that F. nucleatum is detected in only a minority of CRC cases and that its abundance varies substantially across individuals and tumor contexts, limiting its applicability as a uniform biomarker[21]. Investigations employing paired tumor and adjacent normal biopsies have similarly demonstrated that Fusobacterium enrichment is frequently context-dependent and often absent in matched normal mucosa, with bacterial activity and biomass differing markedly between patients rather than consistently between tissue compartments[18]. These observations are reinforced by comprehensive reviews indicating that F. nucleatum preferentially accumulates in tumors with specific molecular characteristics, such as microsatellite instability or CpG island methylator phenotype, and may function primarily as a tumor-adaptive or tumor-promoting bacterium rather than a ubiquitous initiator of colorectal carcinogenesis[22]. Taken together, the paired design of the present study strengthens these conclusions by minimizing interindividual microbiome variability and enabling direct within-patient comparisons.

Beyond overall heterogeneity, analysis across tumor stages provided additional insights into the temporal dynamics of F. nucleatum involvement. One of the most robust findings was the significant association between FS17 signal intensity and preoperative T stage, with T2 tumors exhibiting lower Ct values—reflecting higher FS17-detected F. nucleatum signals—compared with T3-T4 tumors. This pattern indicates that F. nucleatum-associated signals may be more prominent in earlier locally invasive tumors rather than increasing progressively with local tumor advancement. Such findings challenge simplified models that associate F. nucleatum abundance exclusively with advanced or aggressive disease and instead support its involvement in early tumor-microbe interactions. Indeed, studies in patients with early CRC have demonstrated that F. nucleatum abundance increases during the earliest stages of carcinogenesis, suggesting a role in tumor initiation or early progression rather than late-stage invasion[23,24]. Mechanistically, preferential detection of F. nucleatum in tumor tissue may be explained by its strong adhesive and invasive properties toward colonic epithelial cells, mediated by the FadA surface adhesin, which binds E-cadherin and activates β-catenin signaling and downstream inflammatory pathways that promote carcinogenesis[25-27]. As tumors progress locally, alterations in tissue architecture, hypoxia, immune infiltration, and therapeutic exposure may reduce bacterial persistence or alter detectability within the tumor microenvironment, potentially accounting for the reduced FS17 signal observed in more advanced T stages. Importantly, FS17 represents a research-use DNA probe designed to detect specific F. nucleatum subspecies and should not be interpreted as a clinical biomarker per se. Rather, subspecies-resolved detection of F. nucleatum, when interpreted alongside tumor stage and molecular context, may contribute to microbiome-informed tumor stratification in research settings, while cautioning against the use of F. nucleatum signals as a universal proxy for advanced disease burden. While the observed differences in FS17-detected signals across T stages are statistically robust within this cohort, they should be interpreted within the methodological constraints of the study, including the limited sample size and the subspecies-specific nature of the FS17 assay, which restrict broader generalization.

A similarly nuanced pattern emerged when considering nodal involvement. FS17 tumor Ct values showed a progressive increase from N0 to N2 stages, suggesting a trend toward lower FS17-detected F. nucleatum signal with increasing nodal involvement, although this association did not reach statistical significance. Notably, logistic regression analyses in our study revealed complete separation for N2 cases, with all such patients exhibiting tumor-predominant FS17 detection, indicating consistent intratumoral localization despite overall lower expression levels. Together, these findings point to a complex and non-linear relationship between intratumoral F. nucleatum and nodal disease progression. While bacterial presence may be more prominent or biologically active in earlier disease stages, its contribution to lymphatic dissemination appears to be context-dependent rather than strictly proportional to bacterial abundance. This interpretation is consistent with experimental and clinical evidence demonstrating that F. nucleatum actively promotes metastatic behavior through specific host-microbe interactions. In particular, enrichment of F. nucleatum in metastatic CRC tissues and its ability to enhance tumor cell migration via CARD3-mediated autophagy signaling provide a mechanistic basis for its association with nodal and distant spread[28]. Chronic exposure to F. nucleatum has also been shown to promote epithelial-mesenchymal transition, immune suppression, and tumor cell survival, processes central to metastatic competence even when overall bacterial load is modest[29]. Correspondingly, high intratumoral F. nucleatum levels have been linked to poor prognosis, increased metastatic potential, and resistance to chemotherapy, reinforcing its functional involvement in aggressive disease phenotypes rather than a bystander role[20]. Taken together, the coexistence of declining FS17 signal intensity across N stages and consistent tumor predominance in N2 cases suggests a non-linear association with nodal involvement that likely reflects biological heterogeneity combined with sampling constraints, reinforcing the need for validation in larger, molecularly stratified cohorts.

In contrast to tumor-stage and molecular associations, FS17 expression showed no statistically significant relationships with demographic or surgical variables. Neither sex, smoking status, nor operative approach, type of resection, nor anastomotic technique influenced FS17 expression in tumor tissue. These findings suggest that intratumoral F. nucleatum-associated signals are not driven by host demographic characteristics or perioperative factors but are more likely determined by tumor-intrinsic biology. This conclusion is consistent with data from large treatment-naïve cohorts, including the ColoCare Study, which reported no significant associations between F. nucleatum abundance and age, sex, or smoking history, while identifying stronger links with tumor location and molecular features[30]. Similarly, comprehensive reviews emphasize that F. nucleatum is inconsistently associated with basic demographics but is reproducibly linked to tumor-specific characteristics such as histological grade, microsatellite instability/CpG island methylator phenotype (MSI/CIMP) status, immune contexture, and therapeutic response[31]. These tumor-centered associations support a model in which F. nucleatum preferentially colonizes permissive tumor microenvironments shaped by epithelial disruption, hypoxia, immune modulation, and oncogenic signaling rather than systemic patient characteristics[32]. The isolated association observed between FS17 expression in adjacent tissue and anastomotic technique, which was not mirrored in tumor tissue, is therefore unlikely to reflect a biologically meaningful tumor-microbe interaction and may instead represent localized postoperative effects. Collectively, both our data and prior studies position F. nucleatum as a key but context-dependent microbial player in CRC, with its clinical relevance driven primarily by tumor biology rather than demographic or procedural factors[33].

Systemic inflammation provided an additional layer of context for interpreting tumor-microbe interactions. Although systemic inflammatory markers—including white blood cell count, fibrinogen, and CRP—were not significantly associated with FS17 tumor predominance, consistent directional trends were observed, with higher CRP levels tending to associate with tumor-predominant FS17 expression. These findings support a model in which systemic inflammation acts as a permissive or modulatory background rather than a direct driver of bacterial enrichment. Chronic low-grade inflammation can compromise epithelial barrier integrity, reshape immune surveillance, and facilitate microbial persistence within susceptible tumor niches without producing a strong one-to-one correlation at the cohort level. This interpretation is supported by evidence showing that F. nucleatum-high CRC liver metastases are associated with elevated CRP levels, reduced antitumor immune cell densities, and broader metastatic involvement, linking systemic inflammatory burden with F. nucleatum activity and aggressive disease phenotypes[34]. Inflammation-centric CRC studies further demonstrate that CRP reflects a broader cytokine milieu associated with dysbiosis and cancer risk, reinforcing its role as an indicator of a permissive inflammatory terrain rather than a specific determinant of bacterial colonization[35]. Accordingly, the CRP trend observed in our cohort may still be biologically informative, suggesting that patients with heightened systemic inflammation represent a subgroup more conducive to maintaining F. nucleatum-associated tumor signals[36].

At the molecular level, miR-21 was consistently overexpressed in tumor tissue, in line with its established oncogenic role in CRC. Independent cohorts have shown that miR-21 is significantly elevated in CRC compared with adjacent tissue and is associated with lymph node involvement, distant metastasis, higher tumor-node-metastasis stage, and reduced survival[37]. Mechanistic studies further support miR-21 as a driver of invasion and metastasis through repression of tumor suppressors such as PDCD4 and regulation of EMT-related pathways, with combined miR-21-target gene signatures improving metastatic risk prediction[38]. Comprehensive reviews highlight miR-21 as a recurrent diagnostic and prognostic candidate across tissue and circulating matrices, while emphasizing that interpretation depends on assay context and mature isoform definition. In contrast, U6 small nuclear RNA was used exclusively as an internal reference control for miRNA quantification and does not represent an active biological regulator; thus, any apparent correlations between miR-21 and U6 reflect normalization behavior rather than biological interaction. Notably, no significant correlations were observed between miR-21 expression and FS17-detected F. nucleatum signals in either tumor or adjacent tissue. This absence of association suggests that miR-21 alone does not determine bacterial enrichment and is biologically plausible given that microbiome-miRNA interactions are multilayered and context-dependent. Indeed, F. nucleatum has been reported to modulate miR-21 through the TLR4/MYD88/NF-κB axis, indicating that any coupling between F. nucleatum and miR-21 depends on bacterial activity, strain context, host signaling competence, and microenvironmental state[39]. Consequently, the lack of a direct FS17-miR-21 correlation in our cohort reinforces the view that intratumoral F. nucleatum persistence is shaped primarily by broader tumor-host-microbiome interactions rather than by individual microRNAs in isolation[31]. Accordingly, the absence of a direct association between FS17-detected F. nucleatum signals and miR-21 expression in this cohort should be interpreted in a contextual manner and does not preclude more complex, environment-dependent regulatory interactions that may emerge in larger or molecularly stratified datasets.

This study has several methodological limitations that should be considered when interpreting the findings. The limited sample size inherent to this exploratory cohort reduces statistical power and constrains broader generalizability. In addition, F. nucleatum detection was performed using the FS17 research-use assay, which targets selected subspecies and does not capture total bacterial burden across tumor compartments. Accordingly, Ct values were interpreted as relative detection signals to enable robust paired comparisons between tumor and adjacent tissue rather than as absolute measures of bacterial load. Although the paired tissue design minimizes interindividual variability, strain-level heterogeneity and tumor microenvironmental differences may still influence bacterial detectability. These limitations are being addressed in a subsequent, larger-scale study designed as a direct continuation of the present investigation, incorporating molecular stratification, absolute bacterial quantification, and expanded microbiome profiling. Finally, the absence of long-term clinical outcome data, including postoperative recurrence, metastatic progression, and survival, reflects the pilot scope and limited duration of the underlying project, which focused on short-term molecular and microbial characterization. Ongoing longitudinal follow-up of the current cohort, together with enrollment of additional patients, will allow future assessment of the prognostic and translational relevance of these findings.

CONCLUSION

In this pilot Romanian cohort, paired qRT-PCR analyses demonstrated that F. nucleatum-associated signals are heterogeneous and context-dependent, with no uniform enrichment across all colon adenocarcinomas. Within-patient comparisons indicated substantial interindividual variability in tumor vs adjacent tissue detection, while exploratory analyses suggested stage-related patterns—particularly higher FS17-detected signals in earlier T-stage tumors—and non-linear relationships with nodal status. At the same time, the lack of consistent associations with demographic or surgical variables and the absence of a direct correlation between miR-21 and FS17 detection support a model in which intratumoral F. nucleatum is shaped primarily by tumor-specific biological context and microenvironmental permissiveness rather than by host baseline characteristics or isolated molecular markers.

Beyond its biological insights, the major contribution of this work is its population-specific relevance. To our knowledge, Romania currently lacks CRC-specific microbiome datasets; therefore, this study provides an important first step toward defining tumor-microbe interactions in Romanian patients with colon cancer using a paired-tissue design that minimizes interindividual variability. Although limited by sample size and exploratory statistical power, the consistency of several biologically plausible trends across clinical and molecular domains supports the value of this approach and highlights the feasibility of integrating microbial detection with clinicopathological and inflammatory profiling in this setting.

These findings lay a foundation for subsequent adequately powered studies in larger Romanian cohorts. Future work will expand the sample size and incorporate broader microbiome characterization together with immune/molecular stratification to validate the observed patterns, clarify the clinical significance of F. nucleatum-associated signals, and determine whether population-specific microbial signatures can contribute to improved risk stratification and mechanistic understanding of colorectal carcinogenesis in Romania.

ACKNOWLEDGEMENTS

We are grateful to Popescu Dan, MD, MCSE, for his assistance with the statistical analysis.