Serological tests for pancreatic cancer are little used, partly because such assays have proved insufficiently specific for screening. However, retrospective studies have reported results that compare well with commonly used scanning techniques. In this prospective study we assessed a new type of combined lectin/antibody enzyme-linked mucin assay, CAM 17·1, in a routine clinical setting.

Clinicians at a 1200-bed teaching hospital were encouraged to request the CAM 17·1 assay for any patient whose differential diagnosis included pancreatic cancer. Serum samples from 250 patients were tested during an 18-month period. Patients were followed up for at least 8 months. 75 patients who did not have symptoms of pancreatic cancer and had alternative diagnoses were also studied as a control group.

Of the 250 patients, 36 had pancreatic cancer, as defined by histological and imaging criteria, and eight of these patients had a resectable tumour. The sensitivity and specificity of the CAM 17·1 assay were 86% and 91%, respectively, in all patients, 85% and 81% in those who presented with jaundice, and 89% and 94% in patients who did not have jaundice. The sensitivity of the assay compared well with that of ultrasound scanning (59%) and computed tomography (83%) in these patients. Use of the CAM 17·1 assay in combination with ultrasonography allowed identification of 94% of patients with pancreatic tumours and all of those with resectable tumours. CAM 17·1 binding activity did not correlate with tumour size.

Our study confirms the usefulness of the CAM 17·1 tumour-marker assay for the diagnosis of pancreatic cancer. Serological mucin assays should be used more widely in combination with ultrasonography in the investigation of non-jaundiced patients with unexplained abdominal pain or weight loss.