Emerging Approaches Aim to Detect Pancreatic Cancer Earlier

Review highlights advances in screening, biomarkers and AI


Avinash Kambadakone
Kambadakone
Shravya Srinivas Rao, MD
Srinivas Rao

In concert with the RadioGraphics monograph, RSNA News shares the latest on the role of imaging in disease screening and early detection.

Pancreatic cancer, or pancreatic ductal adenocarcinoma (PDAC), is among the deadliest cancers, with a five-year survival rate of 13%. It is projected to be the second leading cause of cancer death in the U.S. by 2030, yet it has a low lifetime development rate of only 1% to 1.5%. This raises the question of how a cancer that is so uncommon is so fatal.

The answer is simple: Pancreatic cancer is often detected too late. “Detecting pancreatic cancer at an earlier, potentially resectable stage can substantially improve the opportunity for treatment. Survival is markedly better for localized disease than for advanced-stage disease,” explained Avinash R. Kambadakone, MD, FRCR, chief of abdominal radiology at Massachusetts General Hospital (MGH) and Brigham and Women’s Hospital (BWH), and associate professor of radiology at Harvard Medical School in Boston.

Dr. Kambadakone’s research in pancreatic cancer began with support from the RSNA R&E Foundation, which awarded him a Toshiba America Medical Systems/RSNA Research Seed Grant in 2016. The grant supported his early work using multiparametric dual-energy CT to assess treatment response and determine surgical resectability in patients with pancreatic cancer. That research helped establish the foundation for his ongoing investigations into pancreatic cancer imaging.

He and Shravya Srinivas Rao, MD, an abdominal radiologist at Mass General Brigham, coauthored a RadioGraphics article outlining the current state of pancreatic cancer screening and early detection, as well as emerging innovations shaping the field.

Identifying Patients for Early Detection

One of the reasons pancreatic cancer is not diagnosed early enough is due to its subtle and irregular presentation. “Unlike many other cancers, pancreatic cancer often causes few or no symptoms in its early stages, making it difficult to detect before it becomes advanced,” Dr. Kambadakone said.

While the initial assumption may be to regularly screen everyone, the authors said pancreatic cancer is rare enough that broad population screening would not be efficient or practical. Instead, screening is generally recommended for individuals with specific risk factors that place them at a substantially increased risk of developing the disease.

“High-risk individuals include those with a strong family history of pancreatic cancer, inherited genetic syndromes, hereditary pancreatitis or pathogenic germline mutations,” Dr. Kambadakone said.

Certain precursor lesions such as mucinous pancreatic cysts have the potential to develop into cancer over time, although most do not.  The key is to identify those at higher risk and monitor them with personalized imaging surveillance, ideally as part of a multidisciplinary care program.

Risk factors can also extend to a patient’s lifestyle. Tobacco, alcohol use and red meat consumption can increase the likelihood of developing pancreatic cancer, though that risk is lower than hereditary or genetic predisposition.

Notably, 40% of the patients with pancreatic cancer meet the criteria for diabetes diagnosis. “Studies have shown that individuals over the age of 50 with new-onset diabetes may represent one of the actionable populations for targeted screening, although additional research is needed to determine the best screening strategies” Dr. Srinivas Rao explained.

Once an individual has been identified as high-risk, the next step is imaging-based screening. “Current evidence supports the use of endoscopic US (EUS), MRI and CT as the primary imaging modalities for pancreatic cancer screening and early detection,” Dr. Srinivas Rao said.

However, she warned that detection is not always straightforward because the imaging features of early pancreatic cancer are often subtle and challenging to identify. “There is no single established imaging appearance, but certain features which should raise suspicion include abrupt pancreatic duct cut-off, small isoattenuating lesions, focal changes in enhancement, focal acute pancreatitis and subtle perivascular soft tissue changes,” she said. “Recognizing these findings is important as they should prompt further evaluation.”

PDAC in a 52-year-old patient with chronic pancreatitis.

PDAC in a 52-year-old patient with chronic pancreatitis. (A, B) Axial (A) and coronal (B) contrast-enhanced CT images show a main pancreatic duct stent (arrowhead) without any pancreatic mass. (C, D) Axial (C) and coronal (D) contrast-enhanced CT images 36 months later show interval development of a hypoenhancing mass (arrowhead) in the head/uncinate process. Biopsy demonstrated PDAC, and the patient underwent the Whipple procedure.

https://doi.org/10.1148/rg.250146 ©RSNA 2026

Biomarkers and AI Show Promise

While imaging is a powerful tool for detection, the authors note that there is ongoing research to identify validated biomarkers for early pancreatic cancer detection that can work in concert with imaging-based techniques. Among them is CA19-9, a carbohydrate antigen released into the bloodstream by pancreatic cancer cells, which can be detected by an antibody.

“CA 19-9 is currently the most common serum biomarker elevated in patients with pancreatic cancer and used to monitor disease. However, it lacks sufficient sensitivity and specificity for screening of pancreatic cancer,” Dr. Kambadakone said. “It may be elevated in benign conditions and is non-informative in approximately 5% to 10% of individuals because of Lewis antigen negativity.”

Despite these limitations, Dr. Kambadakone emphasized the potential of biomarkers. “Many other biomarkers are at various stages of validation and show promise in detecting cancers at early stages,” he said. “The road to widescale clinical adoption is long, however. At present, no single biomarker is sufficiently accurate to serve as a standalone screening test for pancreatic cancer.”

Given how subtle the signs are for diagnosing pancreatic cancer, radiologists are leveraging the power of AI in screening. “Recent studies have demonstrated that radiomics-based machine-learning models can detect pancreatic cancer up to a year before clinical diagnosis and, in some studies, from scans originally interpreted as normal,” Dr. Srinivas Rao said.

“The key message is that early detection represents one of our best opportunities to improve outcomes in pancreatic cancer, but it will require more than a single test or imaging technique. As radiologists we have an important role in this transition from detecting established cancer to identifying disease earlier, when intervention can make the greatest difference.”

— AVINASH R. KAMBADAKONE, MD, FRCR

While promising, she notes that this approach requires additional exploration. “Prospective validation, reproducibility and model interpretability remain necessary before widespread clinical implementation,” she said.

Through their review of the current state and prospects of pancreatic cancer screening and early detection, the authors see a promising future ahead.

“The key message is that early detection represents one of our best opportunities to improve outcomes in pancreatic cancer, but it will require more than a single test or imaging technique,” Dr. Kambadakone said. “As radiologists we have an important role in this transition from detecting established cancer to identifying disease earlier, when intervention can make the greatest difference.”

For More Information

Access the RadioGraphics article, “Pancreatic Cancer Screening and Early Detection: Where Are We Now?” and the related commentary, “Pancreatic Cancer Screening and Early Detection: Current Perspectives and Future Opportunity.”

Read previous RSNA News stories on pancreatic cancer imaging: