Emerging Screening Modalities in Pancreatic and Gastric Cancers
Presenter:
Rachael A. Safyan, MD; University of Washington; Fred Hutchinson Cancer Center
Conference:
Best of ASCO GI 2026
Early detection remains one of the greatest unmet needs in gastrointestinal oncology, particularly in pancreatic cancer, where most cases present with advanced disease. At Best of ASCO GI 2026, Dr Rachael Safyan from The University of Washington and Fred Hutchinson Cancer Center reviewed the current state of screening for pancreatic and gastric cancers and discussed emerging technologies that may enable earlier detection in the future. Her presentation highlighted the limitations of current screening strategies and explored promising approaches including blood-based biomarkers and artificial intelligence–assisted imaging. She stressed that while technological innovation is advancing rapidly, screening strategies must remain risk-adapted and evidence-based, particularly in cancers with relatively low incidence in the general population.
Pancreatic Cancer: The Case for Earlier Detection
Dr Safyan emphasized that pancreatic cancer continues to carry one of the worst prognoses among solid tumors, noting that, in the United States, an estimated 67,440 individuals will be diagnosed with pancreatic cancer in 2025, with nearly 52,000 deaths, making it the third leading cause of cancer-related mortality. Importantly, despite therapeutic advances, survival remains highly stage dependent. Five-year survival is approximately 13.3% overall, but outcomes improve dramatically when disease is detected early. As Dr Safyan put it: “Patients with stage 1A pancreatic cancer have a five-year survival of about 84%, which really underscores the potential of improving outcomes through early detection.” Widespread screening, however, is not feasible because the lifetime risk of pancreatic cancer in the general population is only about 1.7%. For this reason, the U.S. Preventive Services Task Force recommends against screening average-risk individuals. Instead, current guidelines focus on identifying high-risk populations, including individuals with:
Germline mutations associated with pancreatic cancer
Strong family history of pancreatic cancer
These patients, Dr Safyan noted, have an estimated lifetime risk exceeding 5%, making surveillance strategies more clinically justified.
“Patients with stage 1A pancreatic cancer have a five-year survival of about 84%, underscoring the importance of early detection.”
Current Screening Strategies
For high-risk individuals, Dr Safyan cited the current standard screening strategy, which consists of annual MRI/MRCP and/or endoscopic ultrasound (EUS). Each of these modalities offers complementary strengths, specifically, MRI is particularly sensitive for detecting cystic lesions and ductal abnormalities whereas EUS is more sensitive for identifying small solid lesions. Because of these differences, many surveillance programs alternate between MRI and EUS on a yearly basis.
Dr Safyan noted that although these approaches can detect earlier-stage disease, screening programs face several logistical and biological challenges. These include difficulty accessing pancreatic tissue, risks associated with invasive procedures, cost considerations, and the psychological impact of false-positive findings. “There are challenges associated with the pancreas itself, including difficulty accessing tissue and the risk of additional procedures for lesions that may ultimately prove benign,” Dr Safyan noted.
Evidence Supporting Surveillance Programs
Dr Safyan also noted data from the Cancer of the Pancreas Screening (CAPS) consortium, which provide important insights into the potential benefits of surveillance. Across more than 1,700 high-risk individuals, 26 pancreatic cancers were identified, with 19 cases detected during surveillance and 7 diagnosed outside surveillance programs. Importantly, cancers detected through surveillance were far more likely to be diagnosed at earlier stages, which translated into improved survival outcomes compared with those detected outside surveillance. Another large retrospective analysis of more than 2,500 high-risk individuals across 16 surveillance programs which Dr Safyan reviewed revealed an additional challenge: pancreatic cancers may arise rapidly even when imaging studies are negative. Nearly half of patients in this study who developed pancreatic cancer had no detectable lesion on prior imaging, suggesting that tumor progression may occur between annual screening examinations. These findings underscore the need for more sensitive screening tools capable of detecting cancer earlier in its biological evolution.
Emerging Biomarker Approaches for Pancreatic Cancer
Dr Safyan cited several liquid biopsy strategies which are being investigated to address these limitations. These include circulating tumor DNA (ctDNA), protein biomarker panels, microRNA signatures, exosome-based diagnostics, and multi-cancer early detection assays. Dr Safyan emphasized, however, that ctDNA currently has limited sensitivity for early pancreatic cancer detection. “There is currently no utility for ctDNA outside of a clinical trial, because the percentage of stage I patients with detectable ctDNA remains low,” she said. Despite these limitations, biomarker research continues to advance. One promising approach Dr Safyan noted involves a four-protein biomarker panel combining CA 19-9, thrombospondin-2 (THBS2), aminopeptidase N (ANPEP), and polymeric immunoglobulin receptor (PIGR). In recent studies, this panel demonstrated a 91.9% sensitivity for pancreatic cancer detection, an 87.5% sensitivity for stage I/II disease, and 95% specificity. These findings suggest that multi-marker blood panels may eventually enhance early detection, particularly when combined with clinical risk factors.
Artificial Intelligence and Imaging
Dr Safyan noted that AI is also being explored as a tool to improve early detection of pancreatic cancer. She highlighted the PANORAMA study, an international effort evaluating AI-assisted CT analysis for pancreatic cancer detection. The AI system in the study was trained and validated across multiple datasets and then compared directly with radiologist interpretation. She reviewed the results, which demonstrated that the AI model outperformed a pooled group of 68 radiologists in detecting pancreatic cancer on routine CT scans, suggesting that AI may help identify subtle imaging abnormalities earlier in the disease course. While these results are promising, Dr Safyan emphasized that additional validation studies will be necessary before such systems can be widely adopted.
Gastric Cancer: Global Burden and Risk-Based Screening
Dr Safyan then reviewed the current screening paradigm for gastric cancer, which remains a major global health challenge. She noted that globally, gastric cancer is the fifth most common cancer and the third leading cause of cancer death, with nearly 1 million cases diagnosed each year. Incidence varies widely by geography, however, with the highest rates occurring in East Asia, Eastern Europe and South America. Because of this variation, screening approaches differ substantially across regions. Countries with high incidence, such as Japan and South Korea have implemented population-based endoscopic screening programs beginning around age 40. In contrast, the United States adopts a risk-based approach, screening only individuals with increased risk. Dr Safyan cited guidance from the American Gastroenterological Association (AGA), which suggests screening endoscopy to be considered for first-generation immigrants from high-incidence regions, Asian, Hispanic, and Black Americans with elevated risk, individuals with a family history of gastric cancer, and patients with hereditary cancer syndromes.
“Emerging technologies may transform gastric cancer detection, but endoscopy remains the current standard.”
Prevention Through H. Pylori Eradication
One of the most effective prevention strategies for gastric cancer remains eradication of Helicobacter pylori (H. pylori) infection, which is the primary etiologic factor in most non-cardia gastric cancers. H. pylori infection affects approximately 44% of adults worldwide, though only a small proportion of individuals ultimately develop cancer. Nevertheless, the bacterium has been classified as a Group 1 carcinogen by the World Health Organization, and eradication therapy significantly reduces cancer risk. In this regard, Dr Safyan noted results from a meta-analysis of randomized trials which demonstrated that H. pylori eradication reduces gastric cancer incidence, with a number needed to treat of 72 to prevent one case of gastric cancer.
Emerging Non-Invasive Screening Technologies for Gastric Cancer
Dr Safyan cited several non-invasive diagnostic approaches which are currently under investigation for gastric cancer detection. One of the most established methods is serum pepsinogen testing, which is widely used in Japan to identify patients with atrophic gastritis who should undergo endoscopic evaluation. In meta-analyses, she notes that pepsinogen testing demonstrated approximately 69% sensitivity and 88% specificity for detecting atrophic gastritis, a precursor lesion for gastric cancer.
Some of the additional experimental approaches Dr Safyan highlighted include DNA methylation assays (methylomics), cell-free DNA fragmentomics, microRNA panels and protein biomarker signatures (proteomics). One promising microRNA signature she cited, known as the Destinex panel, demonstrated an AUC of 0.948 and 94.7% sensitivity for early-stage gastric cancer detection in validation studies. Another novel technology presented at ASCO GI involved detection of red blood cell DNA (rbcDNA) genomic signatures, which arise from tumor-induced genomic instability in hematopoietic cells. In a multicenter validation study including 865 patients, the assay demonstrated 83% overall sensitivity, 79% sensitivity for stage I gastric cancer and 90% specificity. These findings suggest that blood-based screening approaches could eventually complement endoscopy in high-risk populations.
Artificial Intelligence in Endoscopic Detection
Dr Safyan also reviewed current evidence for AI strategies, which may enhance the performance of endoscopic screening. She noted a recent meta-analysis of 18 studies involving more than 17,000 patients which demonstrated that AI-assisted endoscopy significantly reduced blind spots and miss rates compared with conventional endoscopy. Across studies, AI systems achieved sensitivity and specificity approaching 99% and diagnostic accuracy comparable to expert endoscopists. Importantly, she notes that AI may help bridge the gap between high-volume expert centers and community settings by improving detection accuracy among less experienced endoscopists.
Clinical Takeaways
Dr Safyan concluded by emphasizing that screening strategies must remain tailored to individual risk profiles:
For pancreatic cancer, current surveillance approaches are limited to high-risk individuals, and emerging technologies such as biomarker panels and AI-enhanced imaging may eventually expand early detection capabilities.
For gastric cancer, screening strategies in the United States should remain risk-based, with endoscopy as the current gold standard and H. pylori eradication as a critical preventive strategy.
Speaker Disclosure Information: Dr Safyan reported the following disclosures for this presentation: Consultant or Advisor: Mirati, Agenus, Ipsen, Exelixis, Boston Scientific Research or Grant Funding: Verastem, Replimune, Amgen, Johnson & Johnson, Exelixis, Bristol Myers Squibb, Adagene.