Pancreatic cancer is among the most lethal of all cancers, with fewer than 10% of patients surviving five years. The overwhelming reason for this poor prognosis is late diagnosis — most patients are only identified once the cancer has spread beyond the pancreas or even metastasised to other organs.
Unlike many other cancers, pancreatic cancer produces few distinctive early symptoms. The early stages are often entirely silent, and when symptoms do appear they are typically non-specific — weight loss, back pain, or new-onset diabetes — that might be attributed to many other conditions.
This comprehensive review synthesises evidence on all available strategies for early detection: identifying high-risk populations, imaging approaches, endoscopic techniques, blood biomarkers, and emerging liquid biopsy technologies. Understanding the full landscape is essential for developing practical screening programmes.
Screening the entire population for pancreatic cancer would be impractical and harmful due to the rarity of the disease and the cost of tests. Instead, research focuses on identifying high-risk groups who would benefit most from surveillance.
High-risk individuals include those with inherited genetic syndromes (such as BRCA2 mutations, Lynch syndrome, or familial atypical multiple mole melanoma), those with a family history of pancreatic cancer in multiple first-degree relatives, and patients with long-standing chronic pancreatitis.
New-onset diabetes mellitus (NODM) is a particularly important risk signal: approximately 1% of adults over 50 who develop new diabetes within the previous year will be diagnosed with pancreatic cancer within three years. The ENDPAC score was developed specifically to help clinicians identify which new-diabetes patients are at highest risk.
Multiple imaging modalities are used to detect and characterise pancreatic masses. CT scanning is the most widely used first-line tool, but it misses small lesions. MRI with MRCP (magnetic resonance cholangiopancreatography) provides superior soft tissue contrast and is better at detecting small cysts and ductal abnormalities.
Endoscopic ultrasound (EUS) offers the highest resolution of any pancreatic imaging technique and can detect lesions as small as a few millimetres. EUS-guided fine-needle aspiration (EUS-FNA) allows tissue samples to be taken simultaneously for pathological diagnosis.
Emerging AI-assisted EUS interpretation tools are beginning to automate and standardise the analysis of EUS images, potentially reducing the expertise barrier that limits EUS availability to specialist centres. AI on EUS could make this gold-standard technique more accessible globally.
CA19-9 is the only FDA-approved blood biomarker for pancreatic cancer, but its specificity is too low for screening — it can be elevated in many non-cancerous conditions. Researchers are actively seeking better biomarkers or combinations of markers.
Liquid biopsy approaches — detecting cancer signals in blood without tissue sampling — are among the most exciting developments in the field. These include circulating tumour DNA (ctDNA), circulating tumour cells (CTCs), microRNAs, and cancer-derived exosomes, all of which can potentially be detected years before a tumour becomes visible on imaging.
Multi-omics panels combining multiple biomarker types appear more promising than any single marker alone. Studies combining ctDNA with protein biomarkers or methylation patterns have reported detection of early-stage pancreatic cancer with sensitivities and specificities far exceeding CA19-9 alone.
No single test is sufficient for early pancreatic cancer detection, but the evidence supports a multi-step approach: use risk stratification tools and biomarkers to identify high-risk individuals, then apply intensive surveillance with EUS or MRI to that targeted group.
The review highlights that the field is moving toward AI-integrated multi-modal approaches that combine imaging, genomic, and proteomic data to generate risk scores with sufficient accuracy for clinical use. This is analogous to how cardiovascular risk is now calculated from multiple factors rather than a single measurement.
Translating these advances into population-level benefit will require international collaboration to build large validated cohorts, standardise test protocols, and develop clinical guidelines that can be implemented across diverse healthcare systems. The review calls for coordinated global action to make this happen.