Early Detection of Pancreatic Cancer

Sushil Kumar Garg; Suresh T. Chari


Curr Opin Gastroenterol. 2020;36(5):456-461. 

In This Article

Finding Early Lesion in an Enriched High-risk Population

Imaging and Artificial Intelligence

The role of imaging for early detection of pancreatic cancer is crucial. Traditional techniques used for noninvasive imaging include CT abdomen with pancreas protocol and MRI abdomen with magnetic resonance cholangiopancreatography (pancreas protocol). CT abdomen with contrast with the pancreatic protocol is the standard-of-care imaging technology for initial evaluation of suspected pancreatic cancer with a sensitivity of 76–96% and a specificity of 67%.[36,37] Normal pancreatic parenchyma appears different from a tumor on a CT scan because of the lack of vascularity and dense desmoplasia in cancer.[38] Patients with pancreatic cancer depending on the location also have main pancreatic ductal dilation, pancreatic duct-cutoff, and pancreatic atrophy. Pancreatic tumors that appear very distinct from surrounding pancreatic parenchyma tend to have a poor prognosis due to aggressive biology and poor outcomes.[39] On the other hand, pancreatic tumors, which are isoattenuating and indistinct borders, are challenging to diagnose but have better survival because of less aggressive biology.[39]

The findings of pancreatic cancer at an early stage can be very subtle and easy to miss. Singh et al.[40] did a study in patients of pancreatic cancer who had high-quality CT scan available 6 months before the diagnosis of pancreatic cancer, 80% patients had a visible finding which was missed potentially due to unrelated indication of initial imaging. In the era of increased cross-sectional imaging and increased caseload for the radiologist, there is a need for artificial intelligence to decrease the missing of pancreatic cancer, increase efficiency,[41] and identification of features to predict future development of pancreatic cancer. Fishman and colleagues were able to differentiate pancreatic cancer from a healthy pancreas with 94.1% sensitivity and 98.5% specificity with use of artificial intelligence.[42] Imaging can also be used to identify the secondary systemic effect of pancreatic cancer for early detection of pancreatic cancer. A study done by Sah et al.[43] showed that 6–18 months before the diagnosis of pancreatic cancer and clinical symptoms, there was evidence of loss of subcutaneous adipose tissue, visceral adipose tissue, and muscle mass on CT imaging and these measurements of different body compartments has been automated.[44]

Endoscopic ultrasound (EUS) is the test of choice to establish the diagnosis in a patient with suspected pancreatic cancer with inconclusive noninvasive imaging. EUS has the highest sensitivity to diagnose early pancreatic cancer. EUS can identify up to 2–3 mm lesions.[45] EUS can detect pancreatic abnormality in high-risk individuals for pancreatic cancer 43% of the time; compared with CT (11%) and MRI (33.3%). EUS with elastography[46] and the use of microbubbles[47] have resulted in improved sensitivity in the detection of pancreatic cancer. The main role of EUS is establishing a tissue diagnosis with fine-needle aspiration (FNA) and differentiate benign versus malignant lesion.[48]