Kenneth Nichols PhD, Presenter: Royalties, Syntermed, Inc

Zaman Javaheri MD, Abstract Co-Author: Nothing to Disclose

Christopher J. Palestro MD, Abstract Co-Author: Research grant, BioMed Valley Discoveries, Inc

It has been proposed that gastric emptying (GE) time-activity curves can be fit to a biphasic mathematical model of the form 1-(1-e-A1*t)A0 (J Nucl Med 1985;26:1319-22). This investigation was undertaken to determine whether fitting GE data aids in shortening data collection from 4 hrs. to 3 hrs., while still providing accurate detection of gastroparesis.

GE data collected using a standardized protocol at 0, 1, 2, 3 and 4 hrs. for 250 pts. (65 men & 185 women; mean age 54±20 yrs) were analyzed, with gastroparesis defined as retention >10% at 4 hrs. (J Nucl Med Technol. 2008;36:44-54). Attenuation-corrected conjugate view counts were determined within manually drawn regions over the stomach. In fitting these 0-3 hr. data using a gradient-expansion algorithm, Poisson noise was modeled, and χ2 convergence tolerance of 10-5 was imposed. Per cent of cases that were biphasic (as evinced by convergence of the model), standard error (SEE) of fit to 0-3 hr. data, and % error of predicted (% pred) versus actual retention at 4 hrs. were calculated. Per cent retention values at 3 hrs. (p3) that maximized ROC area versus gastroparesis were determined for biphasic and non-biphasic cases.  

38/250 (15%) pts. had gastroparesis. Significantly more normal than gastroparesis cases had biphasic GE (86% versus 42%, p<0.0001), for which the model converged & had small SEE of fit (3%±3%, p<0.0001) & small % pred (3%±3%, p=0.0002). p3 was significantly smaller for biphasic than non-biphasic cases (10%±9% versus 29%±30%, p<0.0001). Optimal p3 values were 8% for biphasic cases (ROC area = 89%±5%) and 27% for non-biphasic cases (ROC area = 99.5%±0.9%). By these criteria, sensitivity of abnormal retention at 3 hrs. for detecting gastroparesis was 80%, specificity was 96%, & ROC area = 88%±4%, with “good agreement” with (κ = 0.76) and negligible difference from 4 hr. results (McNemar’s Δ = 0.4%, p=1.0).

While the majority of pts. with normal GE demonstrate biphasic curves, many pts. with gastroparesis do not. By using different criteria for abnormal GE at 3 hrs. for pts. with biphasic versus non-biphasic emptying, high test specificity was achieved.

By applying different criteria for abnormal GE at 3 hrs. for pts. with biphasic versus non-biphasic emptying, it is feasible to shorten the test to 3 hrs. in many cases.  

Nichols, K, Javaheri, Z, Palestro, C, Optimizing Evaluation of Gastric Emptying by Curve Fitting.  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL. http://archive.rsna.org/2010/9012035.html