Community-Based Tumor Marker Testing—Are We Ordering Appropriate Tests?

Abstract

Purpose: Carbohydrate antigen 19–9 (CA 19–9) is commonly utilized in the evaluation of patients suspected to have GI related malignancies. We previously found that a large proportion of CA 19–9 tests may be obtained in a manner inconsistent with current recommendations. The present study evaluated the use of concomitant additional tumor markers when CA 19–9 was employed in a diagnostic workup. Methods: We obtained all in-patient test results, demographic data, physician specialty, and reason for ordering for all patients who had a CA 19–9 test performed at a large community-based teaching hospital between May 1, 2002 and November 30, 2004 (30 months). In addition, we performed detailed chart reviews to determine the reason for ordering, differential diagnoses, patient outcomes, other testing related to the reason for ordering the CA 19–9 test, and concomitant ordering of additional tumor markers. Results: Of 224 CA 19–9 tests, 88 (39.3%) were positive (normal cutoff < 37 IU/mol). Of these, 30 (34.1%) were in patients with newly diagnosed, biopsy proven, pancreatic cancer (PC). The remaining 58 (65.9%) had other diseases including pancreatitis, metastatic liver disease, cancer of unknown primary, and jaundice. Other than following patients with PC, the most commonly cited reasons for testing included abdominal pain (37%), abnormal CT scan of the pancreas or suspicion of PC(26.5%), and pancreatitis (20%). Out of 224 inpatients, 154 also had CEA levels ordered [49 (31.8%) were positive]. Only 10% had colon cancer, while the remaining had pancreatic or lung carcinoma or pancreatitis. The average number of concomitant markers was 1.36, in addition to CA 19–9. The most common additional markers were CEA (69%), CA 125 (35%), AFP (21%), and PSA (11%). [figure 1]FigureConclusions: We found that there is a high probability of obtaining other additional tumor markers whenever CA 19–9 is obtained as part of a diagnostic strategy. This may be reflective of clinical uncertainty on the part of the ordering physician. We are using this data to develop a systems-based improvement process to reduce unecessary tumor marker ordering at our institution.