Gastrointestinal and pancreatobiliary cancers can be difficult to detect early, creating a need for practical, noninvasive approaches for early detection and risk assessment.

The gut microbiome has emerged as a potential source of cancer biomarkers, offering practical opportunities for noninvasive testing, as microbial changes can be assessed in faecal samples.

However, using sequencing-based microbiome testing in routine clinical practice remains challenging due to its cost, analytical complexity, and need for specialised data analysis.

Targeted quantitative PCR (qPCR) could offer a simpler way to translate microbiome signals into accessible testing.

Against this backdrop, a research team led by Associate Professor Tadashi Fujii from the Department of Medical Research on Prebiotics and Probiotics at Fujita Health University, Japan, along with fellow researchers Dr. Hideaki Takahashi, Prof. Takumi Tochio, Prof. Yoshiki Hirooka, and Prof. Yohei Doi from the same institute, evaluated six faecal microbial markers and developed disease-specific logistic models for colorectal cancer (CRC) and pancreatic cancer (PC).

Four markers—afb, nan, fsr, and 5ar—were selected for the CRC model, while three—but, fsr, and saa—were selected for the PC model.

Their findings were made available online on June 18, 2026, and published in Volume 90, Issue 9, of the journal Bioscience, Biotechnology, and Biochemistry on September 01, 2026.

The researchers assessed how well the models could distinguish cancer from control samples using leave-one-out cross-validation.

The CRC model produced an area under the curve (AUC) of 0.824, while the PC model produced an AUC of 0.780.

When the fixed models were applied to related clinical groups, the CRC model showed an AUC of 0.716 for colorectal adenoma.

The PC model produced an AUC of 0.804 in an exploratory early PC subgroup and 0.540 when distinguishing pancreatic high-risk individuals without overt cancer from controls.

In a further exploratory comparison, the faecal qPCR score produced an AUC of 0.739 for early PC versus the high-risk group, compared with 0.543 for serum CA19-9.

The difference did not reach conventional statistical significance. “Our findings suggest that a common faecal microbial panel can produce different disease-specific signatures for CRC and PC,” said Dr. Fujii.

“The results also raise the possibility that faecal qPCR could provide complementary information without invasive sampling.” The study also explored whether the PC score changed during treatment in patients who achieved a partial response to chemotherapy.

In this post hoc analysis, the score remained stable in the 1-kestose group but increased in the nonadministration group, whereas CA19-9 decreased in both groups.

Because stool collection is noninvasive, simple, and suitable for repeated or home-based sampling, the researchers suggest that this approach could eventually support screening and surveillance.

For CRC, regular faecal testing might help identify people who may benefit from colonoscopy, including people with precancerous adenomas.

For PC, repeated testing could potentially help clinicians identify high-risk individuals who may need closer assessment with MRI or endoscopic ultrasound.

These applications remain hypothetical and require prospective studies with independent validation.

“Ultimately, we hope this type of testing can become a simple first-step assessment that helps determine which individuals may need further clinical examination and the appropriate timing for follow-up,” said Dr. Fujii.

“Earlier identification of cancer or precancerous changes could help guide timely intervention, reduce unnecessary invasive testing, and potentially improve outcomes and quality of life.”

Overall, the study supports further investigation of targeted faecal qPCR as an exploratory tool for cancer-related risk assessment and surveillance, while larger prospective studies are needed to confirm its accuracy, reproducibility, and clinical usefulness.

Article: Disease-specific gut microbial signatures generate model-derived cancer probability scores through targeted fecal qPCR profiling

Source: Fujita Health University