PTU-080 Accuracy of 13C mixed triglyceride breath testing for investigation of exocrine pancreatic insufficiency: an audit: Abstract PTU-080

Abstract

<h3>Introduction</h3> The two main indirect non-invasive clinical investigations for determination of exocrine pancreatic function are faecal and breath tests. The 13C mixed triglyceride (MTG) breath test is simple, inexpensive and readily available and has a reported sensitivity and specificity of 81% and 85%, respectively. <h3>Methods</h3> 13C MTG breath tests performed over a sixteen month period in a single university affiliated hospital were identified from a database. A retrospective review was performed. Each patient had undergone the same test protocol: Two fasting end-expiratory breath samples were taken prior to the patient being given a test meal. The test meal consisted of 100 g of toast coated with butter (0.25 g×patient weight (kg). The butter was laced with labelled MTG (4 mg×patient9s body weight (kg)). Test breath samples were taken every 30 min for 6 h. A cumulative percent 13C-dose recovery (CDPR) of &lt;23% indicated abnormal exocrine pancreatic function. Samples were run using HeliFANas software. All samples were run at least two times and the results of each of each run were compared to assess validity. <h3>Results</h3> In all, twenty eight patients were identified. Patients were made up of two groups: patients with known chronic pancreatitis and those with symptoms suggestive of pancreatic disease. 46% (n=13) of patients were male and the mean age was 57 (±16 years). Analysis revealed an 18% (5/28) variance between test runs 1 and 2 in this population. Concordance between all runs (1–4) were analysed and results are demonstrated below in Abstract 080. <h3>Conclusion</h3> This study raises concerns regarding the reliability of the 13C MTG breath test. In our population, there was an 18% (5/28) variance between two separate breath analysis. Therefore, five patients would have received an inaccurate diagnosis based on the MTG test alone. However, accuracy can be improved by running samples more than once.