Implications of assessing gastric function at 2.5 hours

Objectives The standard scintigraphic protocol for assessing gastroparesis requires 4 hr. data collection (JNM Tech 2008;36:44). It has been suggested that gastroparesis & abnormally rapid gastric emptying can be assessed by quantifying gastric retention (GR) at 2.5 hrs (GR2.5), as GR2.5>40% & GR2.5<15%, respectively (RSNA 2010, p626). This investigation compared GR data to a curve-fitting approach for shortening data collection time.

Methods Data were collected for 389 pts., 98 men & 291 women (age 55±19 yrs.) using standardized meal & imaging at 0, 1, 2, 3 & 4 hrs. Conjugate-view counts were obtained from manually drawn anterior & posterior stomach regions. GR2.5 values were obtained by interpolating 2 & 3 hr. GR values. For curve fitting, counts from 0-3 hrs. were fit to a biphasic model (JNM 1985;26:1319), using Poisson noise modeling & gradient-expansion algorithms with imposed convergence tolerance of 10^-5, in which case extrapolated GR>10% at 4 hrs. predicted gastroparesis. When data could not be fit to a biphasic model, GR>26% at 3 hrs. was used to predict gastroparesis based on ROC area optimization. Results were compared to predicted gastroparesis based on GR2.5 >40% versus GR>10% at 4 hrs. Actual prevalence of abnormally rapid gastric emptying, (GR<50% for females & GR<30% for males at 1 hr.; Am J Gastroenterol 2000;95:1456), were compared to prevalence based on GR2.5<15%.

Results 84/389 (22%) pts. had gastroparesis. 56/389 (14%) had rapid gastric emptying. Significantly more normal than gastroparesis cases exhibited biphasic curves (267/305 (88%) vs. 33/84 (39%), p<0.0001). Curve fitting criteria were more sensitive (87% vs. 80%) & accurate (94% vs. 92%) than GR2.5 for predicting gastroparesis. Agreement with actual gastroparesis was higher for curve-fitting than GR2.5 (Κ=0.81 for “very good agreement” vs. Κ=0.75 for “good agreement,” p=0.02). GR2.5 overestimated rapid emptying prevalence vs. actual (31% vs. 14%, p<0.0001).

Conclusions Curve-fitting for data collected up to 3 hrs. is more reliable than gastric retention at 2.5 hrs. for identifying abnormal gastric emptying