Sa1355 Improving the Integrated Relaxation Pressure Cutoff Value for the Diagnosis of Achalasia Using a Classification and Regression Tree Model

Abstract

Background: The Integrated Relaxation Pressure (IRP) is a fundamental measurement utilized in Esophageal Pressure Topography (EPT) to assess the pressure dynamics through the esophagogastric junction (EGJ) during swallowing. Data from control subjects suggest that a cutoff of 15 mmHg is a reliable upper limit of normal in the context of propagating peristalsis. Given the interdependence of the IRP and intrabolus pressure during bolus transit, we hypothesized that this value may need to be adjusted when peristalsis is weak, absent or spastic. The aim of this study was to define optimal IRP cutoff values for diagnosing achalasia subtypes using a classification and regression (CART) model. Methods: EPT studies from 522 consecutive patients were studied (10 supine water swallows). Mean IRP values and EPT swallow patterns from each study based on Chicago Classification criteria were used as inputs to train and test a classification and regression tree (CART) model with a MATLABTM program. CART model diagnoses were compared to those of an expert blinded reviewer (JEP). Results: The diagnoses of the 522 patients were: 110 normal, 110 weak peristalsis, 72 frequent failed peristalsis, 28 absent peristalsis, 11 distal esophageal spasm (DES), 11 rapid contractions, 32 hypertensive peristalsis, 21 jackhammer/hypercontractile esophagus, 71 achalasia (type I, II, or III) and 56 EGJ outflow obstruction. The CART model achieved 94% agreement with the expert. The optimal IRP cutoff values varied based on contractile pattern. In the context of absent peristalsis, a cutoff value of ≥ 10 mmHg best distinguished type I achalasia from absent peristalsis. Additionally, a cutoff value of ≥17 mmHg best distinguished type III achalasia from DES while an IRP value of ≥15 was predictive of achalasia when contractile activity was associated with panesophageal pressurization (type II achalasia). The optimal IRP threshold for defining EGJ outflow obstruction was ≥ 15 mmHg. Overall, the accuracy using the CART model for achalasia was 96 % compared to 87% using the standard criterion of a single cutoff value of ≥ 15 mmHg. Conclusion: Applying a CART model, the optimal IRP cutoff value for diagnosing achalasia is modified depending on the associated distal esophageal contractility. The optimal IRP threshold for distinguishing type I achalasia from absent peristalsis is reduced to ≥10 mmHg due to the low intrabolus pressure associated with absent peristalsis. The optimal IRP threshold for distinguishing type III achalasia from DES is increased secondary to the effect of reduced latency on shortening the deglutitive relaxation window to less than 4.5 seconds. Patients with DES may have IRP values that are slightly higher than the standard 15 mmHg threshold, however, the values are usually less than 17 mmHg and associated with a normal nadir EGJ relaxation pressure.