Pancreatic Cancer — Microbiome Signature

> Clinical disclaimer: This signature page synthesizes peer-reviewed evidence for practitioner education. It does not constitute medical advice. All interventions require individualized clinical assessment. Discuss changes with a qualified healthcare provider. Many findings described here are from discovery-phase studies with limited sample sizes; diagnostic and prognostic claims require prospective validation before clinical application.

Overview

Pancreatic cancer is the fifth leading cause of cancer death in Western nations. Pancreatic ductal adenocarcinoma (PDAC) accounts for >90% of cases, with five-year survival of approximately 12%. The microbiome signature framework reveals pancreatic cancer as a convergence disease where metallomic disruption, oral-gut-tumor microbiome translocation, and mycobiome dysbiosis create an ecology that promotes carcinogenesis, evades detection, and drives therapeutic resistance.

The signature spans multiple biological kingdoms — bacteria, fungi, viruses, and phages — and multiple body compartments — oral cavity, gut, bile duct, and tumor tissue itself. The intratumoral microbiome directly mediates chemotherapy resistance through bacterial cytidine deaminase (CDD) metabolism of gemcitabine. This is not merely a biomarker story; the microbiome is a functional participant in disease progression.

This signature is built from 22 peer-reviewed papers spanning urine metallomics, tumor microbiome sequencing, a JAMA Oncology oral microbiome prospective study, Mendelian randomization, mycobiome profiling, metabolomics, and intervention trials.

Metallomic Signature

The landmark urine metallomics study by Schilling et al. (2020) demonstrated that a combined panel of Ca, Mg, Zn, and Cu achieves AUC 0.99 (sensitivity 95.2%, specificity 97.8%) for PDAC detection ^[1].

NOTE: This is a discovery study (n=67). The AUC 0.99 result requires prospective validation in larger, independent cohorts before any diagnostic claims can be made. Discovery-phase biomarker studies routinely show performance degradation upon external validation.

Metal	Direction	Key Evidence
copper	Elevated (urine, serum)	ATP7A overexpression in PDAC; Cu elevated across cancer types as near-universal biomarker [2]
zinc	Elevated (urine), depleted (tissue)	Disrupted ZnT/ZIP transporters (ZIP3, SLC30A); Zn isotope fractionation as novel biomarker dimension (median delta-66/64-Zn = -0.15 per mille vs +0.02 controls, p=0.002) [1]
Ca	Decreased (urine)	S100 protein dysregulation; AUC 0.796 individually [1]
Mg	Decreased (urine)	Disrupted cell proliferation; AUC 0.783 individually [1]
cadmium	Elevated	Cd increase confirmed in pancreatic cancer tissue [2]
Arsenic	Exposure risk	As exposure linked to pancreatic carcinogenesis; microbiome required for As detoxification
Se	Depleted	Impaired selenoprotein antioxidant defense [2]

Key finding: The healthy Zn-to-Cu concentration correlation (r2=0.66) is completely abolished in PDAC (r2=0.0002), indicating fundamental disruption of metal homeostasis ^[1].

Oral Microbiome Connection

The JAMA Oncology study by Meng et al. (2025) — a nested case-control within 122,000 individuals (445 PC cases, median 8.8-year follow-up) — established the oral microbiome as a prospective predictor of pancreatic cancer ^[3].

Finding	Detail
Microbial Risk Score (MRS)	27 bacterial and fungal species combined
Risk magnitude	3.44-fold increased PC risk per 1-SD increase (95% CI 2.63-4.51)
Key pathogens	P. gingivalis, E. nodatum, P. micra (red/orange complex periodontal pathogens)
Fungal component	Candida tropicalis included in MRS
Translocation mechanism	Hematogenous or biliary translocation of oral pathobionts and their inflammatory mediators
Follow-up	Median 8.8 years — oral signature predates diagnosis by nearly a decade

This is the strongest epidemiological evidence to date linking the oral microbiome to pancreatic cancer risk, with a prospective design that addresses reverse causation.

Tumor Microbiome

PDAC tumors harbor intratumoral bacteria, confirmed by 16S rRNA FISH and LPS immunohistochemistry ^[4]:

Feature	Detail
Dominant class	Gammaproteobacteria (Pseudomonas predominant)
Subtype variation	Basal-like PDAC enriched in Acinetobacter, Pseudomonas, Sphingopyxis — predicting significantly worse survival
Chemoresistance mechanism	Bacterial CDD (cytidine deaminase) metabolizes gemcitabine into its inactive form (dFdU), directly mediating chemotherapy resistance
Therapeutic implication	Antibiotic co-administration may restore gemcitabine sensitivity

The tumor microbiome is not a passenger — it is a functional participant in treatment failure. Bacterial CDD activity represents a direct, targetable mechanism linking intratumoral microbiota to clinical outcomes.

Gut Microbiome

Enriched Taxa

Taxon	Evidence	Pathogenic Mechanism
fusobacterium	Enriched in PDAC gut and tumor	Pro-inflammatory; oral-gut translocation; promotes NF-kB activation [5]
porphyromonas	Key MRS component (Meng 2025)	Periodontal pathogen; hematogenous translocation to pancreas [3]
streptococcus	MR risk-increasing (OR 1.712)	Causal association via MR [6]
Odoribacter	MR risk-increasing (OR 1.899)	Strongest MR risk signal [6]
Ruminiclostridium9	MR risk-increasing (OR 1.976)	Causal association [6]
Gammaproteobacteria	Intratumoral dominant	Pseudomonas predominant; CDD-mediated gemcitabine resistance [4]

Depleted Taxa

Taxon	Normal Function	Evidence
faecalibacterium prausnitzii	Primary butyrate producer; anti-inflammatory	Depleted; responder-enriched phages target Faecalibacterium [7]
roseburia	Butyrate/propionate production	Depleted; phages targeting Roseburia enriched in immunotherapy responders [7]
Romboutsia	Gut homeostasis	MR-confirmed protective (OR 0.87) across multiple sensitivity analyses [8]
Senegalimassilia	Gut homeostasis	MR-confirmed protective (OR 0.635) [6]
SCFA producers (general)	Barrier integrity; Treg induction; anti-inflammatory	Global SCFA producer depletion drives chronic low-grade inflammation [5]

Mycobiome

Oral and gut fungal communities are markedly altered in PDAC, representing a critical and often overlooked dimension of the signature:

Finding	Detail	Source
aspergillus as salivary biomarker	AUC 0.983 for PDAC detection	[9]
Cladosporium	AUC 0.969 for PDAC detection	[9]
Oral fungal diversity explosion	5,022 vs 830 OTUs in PDAC vs controls (with decreased Shannon diversity)	[9]
**[[candida-albicans	Candida]] in pancreatitis**	Dominates fecal mycobiome at 61% in acute pancreatitis (PC precursor)	[10]
Aspergillus-WBC correlation	Suggests fungal-driven inflammatory amplification	[10]

The mycobiome connects to the metallomic signature through fungal iron dependence: metal-driven shifts in the bacterial microbiome create ecological niches that fungi exploit, while fungal iron acquisition (siderophores) further disrupts metal ecology.

Virome

The gut virome adds a third biological kingdom to the signature:

Finding	Detail	Source
Virome predicts immunotherapy response	AUC 0.768 (outperforms bacterium-only AUC 0.664)	[7]
Responder-enriched phages	Target Faecalibacterium and Roseburia (SCFA producers)	[7]
Non-responder phages	Target Clostridium and Bacteroides	[7]
Phage-based therapeutics	Phage-derived peptides show selective PDAC targeting	[11]

Metabolomics

Metabolite Class	Direction	Key Evidence
SCFAs (butyrate, propionate)	Depleted	SCFA producer depletion → chronic inflammation → carcinogenic environment [5]
BCAAs (Leu, Ile, Val)	Elevated in tumor	Sustain PDAC growth via BCAT2/BCKDHA-driven lipogenesis [12]
Deoxycholic acid	Elevated	Promotes DNA damage via EGFR ligand amphiregulin [5]
Mannitol	Protective (MR)	OR 0.97 per unit increase — causal protective metabolite [8]
Methionine	Protective (MR)	OR 0.97 — causal protective metabolite [8]
Carnitine	Risk-increasing (MR)	Causal risk metabolite [8]
Serum 4-metabolite panel	Diagnostic	AUC 0.93 (xylitol, 1,5-AG, histidine, inositol) — outperforms CA19-9 in early-stage [13]

Ecological Features

1. Tumor microbiome subtypes: Basal-like PDAC harbors a distinct intratumoral microbiome (Acinetobacter, Pseudomonas, Sphingopyxis) that predicts worse survival. The tumor microbiome is not random colonization — it reflects selection by the tumor microenvironment ^[4].

2. Gemcitabine resistance via bacterial CDD: Intratumoral Gammaproteobacteria express cytidine deaminase that converts gemcitabine to its inactive metabolite (dFdU). This is a direct, mechanistic link between the microbiome and treatment failure — not a correlation ^[4].

3. Oral-pancreatic translocation: Periodontal pathogens (P. gingivalis, Fusobacterium) translocate to the pancreas via hematogenous or biliary routes. The oral MRS predates diagnosis by a median of 8.8 years, suggesting this translocation is an early event in carcinogenesis ^[3].

4. Chronic low-grade inflammation: LPS from Gram-negative bacteria activates NF-kB and MAPK signaling. SCFA depletion removes anti-inflammatory brake. Obesity and T2D — both PC risk factors — converge on this inflammatory dysbiosis ^[5].

5. Bile acid dysmetabolism: Altered bacterial bile acid deconjugation produces excess deoxycholic acid, which promotes DNA damage. This connects the gut microbiome to pancreatic carcinogenesis through the biliary-pancreatic anatomical axis.

Validated Interventions

Probiotic / Microbial

Intervention	Mechanism	Evidence
*Ferrichrome (from L. casei)*	Siderophore-mediated iron chelation; induces p53-mediated apoptosis in PDAC cells including 5-FU-resistant lines; 10 mg/kg reduces xenograft tumor volume	Promising — preclinical; connects iron biology to ferroptosis [14]
Synbiotics (probiotics + inulin)	RCT (90 patients, NCT06199752): CD8+ T cells elevated to 61.5% vs 15.8% placebo; reduced postoperative bacteremia	Validated — Phase II RCT [15]
FMT (from long-term survivors)	Transfers protective microbiome ecology from long-term PDAC survivors; enhances anti-tumor immunity	Promising — preclinical models [16]
Gallium-polyphenol nanoparticles (LGG-loaded)	Reprograms intratumoral microbiota and tumor immune microenvironment	Experimental — preclinical [17]

Dietary

Intervention	Mechanism	Evidence
Dietary fiber	Protective against PC risk; supports SCFA-producing taxa	Validated — meta-analysis confirms dose-response protective association [18]
Quercetin	Inhibits pancreatic cancer stem cell self-renewal; attenuates sonic hedgehog and beta-catenin signaling	Promising — preclinical [19]
Oral hygiene / periodontal care	Reduces oral pathobiont burden (P. gingivalis, Fusobacterium) that translocation to pancreas	Epidemiologically supported — periodontal disease is established PC risk factor; Meng 2025 MRS provides mechanism [3]

STOPs

• STOP: Diagnostic Overclaiming from Discovery-Phase Microbiome Data (Pancreatic Cancer) — The current PDAC microbiome evidence base is almost entirely discovery-phase with no validated prospective diagnostic utility; practitioners who use microbiome signatures to screen for or rule out PDAC risk false reassurance and unnecessary intervention in healthy patients. Evidence: prospective-cohort.

STOP	Why It Matters
Do not overclaim diagnostic utility from discovery-phase biomarkers	The Zn isotope fractionation AUC 0.99 (Schilling 2020, n=67) and Aspergillus salivary AUC 0.983 (Wei 2022) are discovery-phase results. Discovery AUCs routinely degrade 10-20% upon external validation. These are hypothesis-generating, not clinically actionable diagnostic thresholds. Prospective validation in independent, adequately powered cohorts is required before any clinical deployment. Practitioners should NOT present these as validated diagnostic tests to patients.
Caution with antibiotic co-administration for gemcitabine sensitization	While bacterial CDD mediates gemcitabine resistance, broad-spectrum antibiotics would simultaneously destroy protective SCFA producers and potentially worsen overall prognosis. Targeted intratumoral antibiotic strategies require development before clinical application.

Open Questions

• Oral microbiome screening: Can the Meng 2025 MRS (27 species) be reduced to a clinically feasible screening panel? What is the cost-effectiveness in average-risk populations?
• Intratumoral antibiotic targeting: Can narrow-spectrum antibiotics or phage therapy selectively eliminate CDD-producing Gammaproteobacteria without collateral damage?
• Mycobiome as early biomarker: Aspergillus AUC 0.983 in saliva — does this replicate in prospective validation? What is the lead time before diagnosis?
• FMT from long-term survivors: What specific taxa or metabolites from survivor microbiomes drive anti-tumor immunity? Can these be isolated?
• Virome-guided immunotherapy: Can phage profiling guide immunotherapy selection for PDAC patients?
• Metal homeostasis restoration: Does correcting the Zn/Cu ratio imbalance (r2 collapse from 0.66 to 0.0002) alter disease trajectory?

Knowledge Primitives Applied

1. Metals as Selective Pressures — Cu elevation + Zn tissue depletion + Cd/As exposure creates pro-carcinogenic metal ecology
2. Nutritional Immunity as Interpretive Constraint — Zn urinary elevation with tissue depletion reflects disrupted metal trafficking, not simple excess
3. Mis-metallation and Toxic Metal Entry — Cd/As as carcinogenic metals; Zn isotope fractionation reflects metalloprotein dysfunction
4. Microbial Metal Dependencies as Achilles' Heels — Ferrichrome (L. casei siderophore) exploits iron dependency to induce tumor cell death
5. Two-Sided Ecological Engineering — Must suppress Gammaproteobacteria/Fusobacterium AND restore SCFA producers; synbiotics RCT demonstrates this approach
6. Interkingdom Relationships and Functional Shielding — Bacterial-fungal cooperation (Aspergillus, Candida) in tumor ecology; trans-kingdom MRS in oral cavity
7. Estrobolome and Hormone Recirculation — Not primary driver; bile acid dysmetabolism is the relevant hormone-like metabolite axis
8. Siderophore Competition and Iron Ecology — Ferrichrome from L. casei induces ferroptosis in PDAC cells; fungal iron acquisition reshapes mycobiome
9. Oxygen State as Ecological Determinant — Tumor hypoxia selects for anaerobic/microaerophilic intratumoral microbiome composition

Key Sources

• ^[1]
• ^[2]
• ^[3]
• ^[4]
• ^[20]
• ^[14]
• ^[8]
• ^[6]
• ^[9]
• ^[10]
• ^[15]
• ^[7]
• ^[5]
• ^[13]
• ^[12]
• ^[11]
• ^[18]
• ^[19]
• ^[16]
• ^[17]
• ^[21]
• ^[22]