The research team, led by Kaoru Furihata of Kochi Medical School, Kochi University, together with colleagues at the Kanagawa Cancer Center, Tokyo Metropolitan Tama Medical Center, and Chikamori Hospital, enrolled patients with resectable pancreatic cancer who were scheduled for surgery, some after a course of neoadjuvant chemotherapy with gemcitabine and S-1, a regimen conditionally recommended in Japan for resectable disease. The critical technical innovation was the source of the tissue: rather than relying on specimens obtained during the operation itself, the investigators used samples collected by endoscopic ultrasound-guided fine-needle aspiration biopsy, or EUS-FNAB, a minimally invasive procedure that is already standard practice in the diagnostic workup of pancreatic masses. This means the prognostic information is available at the very beginning of the clinical pathway, when treatment decisions still lie ahead rather than behind.

Measuring protein expression in tiny biopsy cylinders is a genuine technical challenge. Needle aspirates contain limited cellular material, often admixed with blood, necrotic debris, and desmoplastic stroma, the dense scar-like tissue that characterizes pancreatic tumors. The team addressed this by developing novel monoclonal antibodies against PODXL and ITGB1 and pairing them with an automated detection system designed to quantify staining reproducibly across specimens. Automated image-based quantification reduces the subjectivity that has historically plagued immunohistochemistry, where pathologists grading staining intensity by eye can disagree. By standardizing both the antibody reagents and the scoring pipeline, the researchers aimed to produce a biomarker that could, in principle, be reproduced across institutions rather than confined to a single expert laboratory.

The biology underlying the marker choice is rooted in cancer metastasis. PODXL is a transmembrane sialomucin of the CD34 family, best known for maintaining the filtration slit diaphragm of kidney podocytes, but in cancer it has been implicated in cell adhesion changes, invasiveness, and epithelial-mesenchymal transition, the program by which epithelial tumor cells acquire motile, invasive properties. ITGB1, the beta-1 subunit of multiple integrin receptors, mediates attachment of cells to collagen, laminin, and fibronectin in the extracellular matrix, and is a well-established driver of tumor cell migration, survival signaling, and resistance to chemotherapy. The hypothesis was straightforward: tumors in which both molecules are abundant may be particularly adept at detaching, invading, and colonizing distant sites, and therefore carry an intrinsically worse prognosis that is detectable even in a preoperative sample.

The results are striking in their magnitude. Among the primary cohort of 55 patients with resectable pancreatic ductal adenocarcinoma, 24 patients, or 43.6 percent, showed high expression of both PODXL and ITGB1, a group the investigators designated Both-high. The remaining 31 patients formed the comparison group. Median overall survival in the Both-high group was 1.99 years, whereas median survival had not been reached in the Others group at the time of analysis, with follow-up extending beyond 3.7 years. The hazard ratio was 3.77, with a 95 percent confidence interval of 1.74 to 8.18 and a log-rank p value of 0.0003, meaning the Both-high group faced nearly four times the risk of death at any given time point.

The association held up under scrutiny. In the subgroup of 37 patients who received neoadjuvant chemotherapy with gemcitabine and S-1 before surgery, the Both-high group again fared far worse, with a median overall survival of 2.59 years versus a median not yet reached in the others, a hazard ratio of 5.56, a 95 percent confidence interval of 2.03 to 15.21, and a log-rank p value of 0.0002. In multivariate analysis adjusting for other clinical variables, Both-high status remained an independent prognostic factor, with a hazard ratio of 4.41, a 95 percent confidence interval of 1.91 to 10.17, and a p value of 0.0005. Classification and regression tree analysis, a statistical technique that recursively splits patients into risk groups, identified Both-high status as the primary explanatory factor for prognostic discrimination, suggesting the marker dominates over conventional clinical variables in separating good from poor outcomes.

The authors were careful to characterize the incremental value of adding ITGB1 to PODXL as modest, supported by exploratory analyses using decision curve analysis and time-dependent area under the curve metrics reported in the supplementary material. Decision curve analysis evaluates whether a model improves clinical decision-making across a range of risk thresholds, while time-dependent AUC measures discrimination as survival time accrues. This kind of methodological honesty matters, because biomarker studies in oncology have a long history of promising signals that dissolve under independent validation. By quantifying exactly how much the two-marker combination adds over PODXL alone, the study gives future investigators a realistic benchmark rather than an inflated claim.

Equally important are the study’s stated limitations, which the authors articulate with unusual clarity. Because the analysis included only patients who ultimately reached surgical resection, the biomarkers should be understood as prognostic stratifiers, not as definitive predictive markers of neoadjuvant treatment efficacy or of loss of surgical opportunity. In other words, the test identifies which resected patients will do worse, but it has not yet been shown to forecast which patients will progress during chemotherapy and forfeit surgery altogether, the very scenario that motivates much of the interest in preoperative risk stratification. A small exploratory cohort of seven patients with borderline resectable disease was analyzed separately and cannot support firm conclusions. The authors explicitly call for future prospective interventional trials that include non-resected patients to determine whether biomarker-guided modification of treatment actually improves outcomes, a step that separates marker discovery from marker-informed clinical practice.

The clinical logic of the finding is nonetheless compelling. Neoadjuvant chemotherapy with gemcitabine and S-1 is conditionally recommended for resectable pancreatic cancer in Japan, yet a subset of patients loses the opportunity for surgery as their tumors progress during preoperative treatment. If a biopsy-based test performed at diagnosis can flag the patients most likely to follow an aggressive course, oncologists could, in future trials, test whether those patients need intensified regimens, alternative drug combinations such as gemcitabine plus nab-paclitaxel, earlier escalation to other modalities, or closer surveillance. Conversely, patients in the favorable group might be spared unnecessary treatment intensification. The fact that the required tissue is already obtained through EUS-FNAB in nearly every diagnostic pathway for suspected pancreatic cancer means the barrier to implementation is analytical rather than procedural: the biopsy is taken anyway, and the question is only what additional information can be extracted from it.

The study, conducted under the Declaration of Helsinki with institutional review board approval at all participating centers and written informed consent from every patient, was supported by Grants-in-Aid for Scientific Research (KAKENHI) and published open access. For a disease in which the five-year survival rate has stubbornly remained in the single digits despite decades of effort, any tool that sharpens prognostic precision before treatment begins deserves attention. The road from a 55-patient cohort to routine clinical use is long, and independent validation in larger and more diverse populations will be essential. But the concept that a two-protein stain on a routine needle biopsy could sort pancreatic cancer patients into meaningfully different risk categories before therapy starts represents exactly the kind of practical, testable advance that the field has been waiting for, and it sets the stage for the interventional trials that will determine whether this biomarker can change not just what clinicians predict, but what patients experience.

Subject of Research: Preoperative prognostic stratification of resectable pancreatic cancer using PODXL and ITGB1 expression in EUS-FNAB biopsy specimens

Article Title: Preoperative prognostic stratification of high-risk resectable pancreatic cancer: combined Podxl/itgb1 expression in EUS-FNAB specimens as a prognostic biomarker

Article References: Furihata, K., Sakaguchi, M., Ueno, M., Kobayashi, S., Yokose, T., Nakazono, A., Inoue, D., Hayashi, T., Okawa, Y., Furihata, M., Kurabayashi, A., Uchida, K., & Taniuchi, K. (2026). Preoperative prognostic stratification of high-risk resectable pancreatic cancer: combined Podxl/itgb1 expression in EUS-FNAB specimens as a prognostic biomarker. BMC Cancer. https://doi.org/10.1186/s12885-026-17094-9

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17094-9

Keywords: pancreatic cancer, PODXL, ITGB1, EUS-FNAB, prognostic biomarker, neoadjuvant chemotherapy, gemcitabine and S-1, risk stratification, immunohistochemistry, overall survival, pancreatic ductal adenocarcinoma, biomarker