Clinical utility of the serum CA 19-9 test for diagnosing pancreatic carcinoma in symptomatic patients: a prospective study

Pancreas. 1992;7(4):497-502. doi: 10.1097/00006676-199207000-00012.


The diagnostic accuracy of the serum CA 19-9 determination was prospectively evaluated in patients selected for the presence of signs or symptoms highly suggestive for pancreatic cancer. Of 110 patients included in the study, 54 had a final diagnosis of pancreatic adenocarcinoma (49% prevalence). CA 19-9 values were higher than 40 U/ml in 45 patients with pancreatic carcinoma and in 18 of the 56 patients with other final diagnosis (sensitivity, 0.83; specificity, 0.68; positive predictive value [PPV], 0.71; negative predictive value [NPV], 0.81). The serum CA 19-9 determination was not capable of shortening the diagnostic workup of patients with strong clinical suspicion of pancreatic cancer since adequate imaging of the pancreas was required to confirm or exclude the diagnosis. However, values above 120 U/ml were strongly suggestive for pancreatic carcinoma in the overall population (PPV, 0.85) and they were diagnostic (PPV, 1.0) in the anicteric portion. Combined with pancreatic imaging, the CA 19-9 was an excellent confirmatory test; a normal value in a patient with negative imaging ruled out pancreatic carcinoma as the cause of symptoms (NPV, 1.0), whereas a pathological result in the presence of positive or equivocal pancreatic radiology was highly suggestive for the presence of the disease (PPV, 0.93).

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adenocarcinoma / blood*
  • Adenocarcinoma / diagnosis*
  • Adenocarcinoma / epidemiology
  • Antigens, Tumor-Associated, Carbohydrate / blood*
  • Biomarkers, Tumor / blood
  • Humans
  • Pancreatic Neoplasms / blood*
  • Pancreatic Neoplasms / diagnosis*
  • Pancreatic Neoplasms / epidemiology
  • Predictive Value of Tests
  • Prevalence
  • Prospective Studies


  • Antigens, Tumor-Associated, Carbohydrate
  • Biomarkers, Tumor