Pharmacological modulation of human gastric volumes demonstrated noninvasively using SPECT imaging


Michael Camilleri MD Mayo Clinic, CharIton 7-154, 200 First St SW, Rochester, MN 55905, USA.Tel.: + 1 507 2662305; fax: + 1 507 2555720; e-mail:


Three-dimensional single-photon emission computed tomography (SPECT) imaging allows noninvasive measurement of human postprandial gastric accommodation. The aim of this study was to determine whether 99mTCO4-SPECT demonstrates effects on pre- and postprandial gastric volumes of intravenous (i.v.) erythromycin lactobionate and sublingual isosorbide dinitrate, as predicted from previous literature. Twenty volunteers received no medication (controls), while 12 were randomized to either i.v. erythromycin 2 mg kg−1 over 20 min, or 10 mg sublingual isosorbide. After a 10-min preprandial SPECT measurement, a standard 300-mL, 300-kcal liquid meal was ingested, followed by a 20-min postprandial measurement. Gastric images were reconstructed from transaxial images and total volume was measured using the Analyseð software system. Fasting gastric volume was greater with isosorbide [223 ± 14 (SE) mL vs. 174 ± 9 mL, control; P < 0.05], and postprandial volume was lower with erythromycin [393 ± 27 mL vs. 582 ± 17 mL, control; P < 0.05]. The ratio of postprandial over fasting volume and mean difference between pre- and postprandial volumes were significantly lower in both drug groups compared to controls. We conclude that 99mTCO4-SPECT imaging is able to semiquantitatively demonstrate pharmacological modulation of fasting gastric volume and postprandial accommodation in humans.


The gastric accommodation response to a meal is a vagally mediated reflex.1–3 Physiologically, it allows the ingestion of large volumes of food without unduly increasing gastric intraluminal pressure or emptying rate,4,5 or inducing the vomiting reflex or postprandial symptoms.6–8 Impaired gastric accommodation is a common phenomenon in conditions such as functional dyspepsia,9–12 postfundoplication dyspepsia, rumination syndrome,13 and diabetic vagal neuropathy,14–16 or following vagotomy or gastric surgery.17,18 As such, impaired accommodation may, at least partly, contribute to the development of postprandial symptoms of bloating, epigastric pain, nausea, vomiting and early satiety in such patients.19,20

We have previously shown that three-dimensional single photon emission computed tomography (SPECT) is a potentially useful noninvasive method to measure the postprandial gastric accommodation response.21 Such a noninvasive tool for measuring gastric accommodation has the potential to facilitate pharmacodynamic studies. In addition, it may enhance the rational choice of therapy for dyspepsia associated with impaired accommodation. A comparison of simultaneous barostatic balloon measurements and SPECT measurements of gastric accommodation showed significant correlations. Thus, proximal gastric volumes estimated as absolute volume measurements (R=0.82, P < 0.01) and as the ratio of postprandial to fasting volumes (R=0.74, P < 0.01) by the two methods were significantly correlated.22 As part of a series of studies intended to validate the measurement of gastric accommodation using SPECT, our goal was to determine whether SPECT is able to demonstrate effects of pharmacological perturbations of gastric motor function, as predicted from previous literature. Thus, we chose to evaluate the effects of intravenous (i.v.) erythromycin and sublingual isosorbide dinitrate.

The effects of erythromycin on gastrointestinal motility have been extensively studied. Erythromycin induces migrating motor complexes,23 increases postprandial antral motility24,25 and enhances gastric emptying,26 possibly via a motilin receptor agonist pathway.27 Using an intragastric barostat balloon, Bruley des Varannes and colleagues28 demonstrated that erythromycin increases fasting and postprandial tone in the human proximal stomach. These actions account for its usefulness in postvagotomy,29 diabetic29,30 and idiopathic gastroparesis.31–34

Nitric oxide is thought to be the putative mediator involved in the gastric accommodation response.35–37 As such, exogenous nitric oxide donors may be potentially useful in conditions with impaired accommodation. However, the effects of nitric oxide donors on proximal gastric motor function have been controversial. Thumshirn and colleagues38 have shown, using an intragastric barostatic balloon, that intravenous nitroglycerin enhances pre- and post-prandial proximal gastric volumes. Other studies, using abdominal ultrasound, have shown that sublingual nitroglycerine has no effect on proximal gastric accommodation in diabetic patients,39 while enhancing it in patients with functional dyspepsia.40

Our hypothesis was that SPECT imaging of the gastric volume identifies changes induced by pharmacological perturbations designed to contract or relax the human stomach. We anticipated that erythromycin would contract, and isosorbide expand, postprandial gastric volume. The specific aim of this study was to demonstrate the predictable effects of intravenous erythromycin lactobionate and sublingual isosorbide dinitrate on pre- and postprandial gastric volumes using SPECT imaging. Documentation of these pharmacological modulations of gastric volumes with SPECT would suggest that SPECT can also be used in pharmacodynamic studies and potentially to assess changes in gastric accommodation in disease states.


Healthy volunteers

Thirty-two healthy volunteers (14 men and 18 women; median age 35 years, range 25 –50 years) were recruited for this study: six in each drug group and 20 in the control group. The control group comprised 10 healthy human controls who were previously evaluated in a separate study using identical methodology. The results of those participants were reported elsewhere.21

An abridged bowel disease questionnaire was used to screen volunteers to ensure that they had no gastrointestinal symptoms. The following inclusion criteria were applied: (i) no previous abdominal surgery except appendectomy; (ii) no intake of medications other than stable doses of contraceptive pill, L-thyroxine or oestrogen replacement therapy; and (iii) no intake of medications in the previous week that could alter gastrointestinal motor function. All participants were fasting and refrained from smoking for at least 12 h prior to the study. Females of childbearing potential underwent a pregnancy test within 48 h of each study. The protocol was approved by the Institutional Review Board and Radiation Safety Committee of the Mayo Clinic. All subjects gave written informed consent.

Imagng of gastric mucosa

99mTc-pertechnetate is preferentially taken up by parietal (oxyntic) cells and nonparietal (mucous) cells of the stomach and provides a means to detect the gastric mucosa. Uptake of99mTc-pertechnetate is found in all parts of the stomach, with regional differences in radionuclide incorporation due to variations in thickness or surface area of the gastric mucosa.41–43 This property has been widely used to diagnose ectopic gastric mucosa in patients with suspected Meckel’s diverticulum and retained antral mucosa using radionuclide imaging.41,43–46 We have previously shown that 10–20 mCi of intravenous99mTc-pertechnetate is a sufficient dose for visualization and quantification of gastric volumes on tomographic images.21 Radiation exposure was within permissible ranges for research and clinical studies (Table 1).

Table 1.   Radiation exposures from spect imaging for gastric accommodation Thumbnail image of

SPECT imaging

Tomographic studies were acquired on a large field of view dual-head gamma camera system (Helix SPECT System™, Elscint Ltd, Haifa, Israel) equipped with low energy, high-resolution collimators. All subjects (controls and those given medications) were positioned supine on the imaging table with the detectors over the upper and mid-abdomen to ensure imaging of the stomach and small bowel. Ten min after the intravenous injection of 20 mCi99mTc-pertechnetate, dynamic tomographic acquisition was performed using the multiorbit mode of the system. Briefly, in this mode, the system performed three complete 360° orbits at 10 min per orbit. For each orbit, images were acquired into a 64 × 64 matrix every 6°, at 10 s per image. Images were then reconstructed using filtered back-projection (Ramp-Butterworth filter, order 10, cut-off 0.45 Nyquist) to produce transaxial images of the stomach.

Study design

Subjects were studied after at least 6 h of fasting. SPECT image acquisition was performed in all participants 10 min following intravenous injection of 10 mCi99mTc-pertechnetate (Fig. 1). Organ-dose exposures to this dose are provided in Table 1 (doses calculated by Mayo Clinic Radiation Control Committee). Twenty volunteers received no medication (controls) while 12 volunteers were randomized to receive either intravenous infusion of erythromycin lactobionate (Erythrocin; Abbott Laboratories, IL, USA), 2 mg kg−1 over 20 min, starting at the initiation of SPECT acquisition, or 10 mg sublingual isosorbide dinitrate (Isordil; Wyeth Pharmaceuticals, PA, USA) given 10 min before starting SPECT acquisition. The tmax. of isosorbide after sublingual administration is less than 10 min; in addition, the plasma elimination half-life of isosorbide after sublingual administration is 45–60 min,51–54 suggesting that high circulating levels were achieved for the entire 30 min of fasting and postprandial measurements.

Figure 1.

 Experimental design.

Volunteers receiving isosorbide were required to be supine from the time of drug administration to minimize the risk of systemic arterial hypotension. After completion of a 10-min preprandial orbit, the supine volunteers drank within 3 min a standard 300-mL, 300-kcal liquid meal [Ensure Plus; Ross Products, Division of Abbott Laboratories, Columbus, OH, USA] with the aid of a straw. After meal ingestion, two 10-min postprandial orbits were completed. Thus, the total duration of imaging was approximately 33 min. The rationale for restricting the measurements of gastric volume to the first 20 min postprandially is based on the vast literature using the barostat balloon to measure gastric accommodation, confirmed in our studies from our laboratory. Thus, the magnitude of the accommodation response is maximal during the first 20 min when it is measured for up to 90 min after a liquid nutrient meal.5–10,13

Data and statistical analysis

For estimation of gastric volume, the transaxial images were transferred via Interfile to a dedicated Unix workstation. Stomach volume measurements were performed using the Analyse™ PC 2.5 software system (Biomedical Imaging Resource, Mayo Foundation, Rochester, MN, USA). To measure the volume of the stomach, it was necessary to identify the stomach in the transaxial SPECT images and separate it from the background noise. This was accomplished using a semiautomated segmentation algorithm (Object Extractor, Analyse™ PC 2.5) which requires the user to identify an appropriate seed point and greyscale threshold.

Because 99mTc pertechnetate is taken up by the gastric mucosa, the ‘fill interior holes’ option was necessary to produce a solid stomach volume as opposed to a hollow shell. Each of the transaxial images was visually inspected to ensure that the segmentation algorithm accurately identified the outline of the stomach. Three-dimensional renderings of the stomach were then produced (Fig. 2), and the user manually removed any extraneous structures such as the upper duodenum or a segment of kidney in close proximity to the stomach, which had not been removed in the segmentation algorithm.

Figure 2.

 Three-dimensional images of fasting (upper panel) and postprandial (lower panel) stomach for one individual in each group, constructed from transaxial SPECT images using Analyse™ software.

This form of correction was required in approximately 10% of the images. All the images were analysed by the same investigator (SSL), who was blinded to the study group of the images being analysed. Total gastric volume was measured during fasting and two postprandial periods (0–10 and 10–20 min). In the previous21 and current study, there were no significant differences in gastric volumes measured during the first and second 10-min postprandial periods. The average (0–10 and 10–20 min) postprandial gastric volume, volume difference between fasting and average postprandial periods, and the ratio of average postprandial over fasting volume were calculated.

Analysis of variance with Dunnett’s multigroup comparison was used to compare the above parameters for each drug vs. controls (no treatment). Within the control group, Student’s t-test was used to examine any gender differences in the parameters, while Pearson’s correlation was used to identify any relationship between each parameter and age or body mass index. The α level for statistical significance was set at 0.05. Values were expressed as mean ± standard error of the mean (SEM) unless otherwise stated. Statistical analysis was undertaken using Sigma Stat (SPSS Inc., Chicago, IL, USA).


Characterization of study subjects

Among the 32 healthy participants, 15 were male and 17 female. Table 2 shows the demographics of the three groups; despite randomization, there was a slight disparity in the gender ratio of the three groups, but there was no difference in age or body mass index between the erythromycin- and isosorbide-treated groups.

Table 2.   Comparison of demographics using Student’s t-test Thumbnail image of

Characterization of fasting and postprandial gastric volumes in healthy subjects

Within the untreated group of 20 healthy controls, there was no significant gender difference in fasting volumes (169 ± 13 mL in females; 183 ± 11 mL in males) or postprandial volumes (767 ± 28 mL in females; 735 ± 29 mL in males). In addition, there was no significant relationship between any of the above parameters and age or body mass index (data not shown).

Effects of erythromycin and isosorbide on fasting and postprandial gastric volumes

Figure 2 shows representative examples of reconstructed stomach images during fasting and postprandially in the three groups. In the erythromycin group, fasting gastric volume (204 ± 11 mL) was not significantly different from the untreated group (174 ± 9 mL); fasting gastric volume was greater in the isosorbide group (223 ± 14 mL, P < 0.05) than the untreated group (Fig. 3).

Figure 3.

 Effects of erythromycin and isosorbide on the fasting gastric volume. Each box plot shows median value (bar), interquartile range (box), range from 10th to 90th percentile (bar caps) and data outside 10th and 90th percentiles (points). *indicates P < 0.05 vs. controls, comparing mean values by ANOVA with Dunnett’′s multigroup comparison.

The average postprandial volume in the erythromycin group (597 ± 32 mL) was significantly lower than in the untreated group (756 ± 21 mL, P < 0.05). However, there was no significant difference between the postprandial volumes of isosorbide (708 ± 24 mL) and the untreated group (Fig. 4).

Figure 4.

 Effects of erythromycin and isosorbide on average (0–10 min and 10–20 min) postprandial volume. Each box plot shows median value (bar), interquartile range(box), range from 10th to 90th percentile (bar caps) and data outside 10th and 90th percentiles (points). *indicates P < 0.05 vs. controls, comparing mean values by ANOVA with Dunnett’s multigroup comparison.

The ratios of postprandial over fasting volume between fasting and postprandial periods were significantly lower (P < 0.05) in both drug groups (erythromycin 3 ± 0.1; isosorbide 3.2 ± 0. 1) compared to the untreated group (4.5 ± 0.2) (Fig. 5). Similarly, the difference in gastric volume (Fig. 6) was significantly lower (P < 0.05) in both treatment groups (erythromycin 393 ± 27 mL; isosorbide 485 ± 16 mL) compared to the untreated group (582 ± 17 mL).

Figure 5.

 Effects of erythromycin and isosorbide on average ratio of postprandial over fasting volume. Each box plot shows median value (bar), interquartile range(box), range from 10th to 90th percentile (bar caps) and data outside 10th and 90th percentiles (points).*indicates P < 0.05 vs. controls, comparing mean values by ANOVA with Dunnett’s multigroup comparison.

Figure 6.

 Effects of erythromycin and isosorbide on average difference in volume between fasting and postprandial periods. Each box plot shows median value (bar), interquartile range (box), range from 10th to 90th percentile (bar caps) and data outside 10th and 90th percentiles (points). *indicates P < 0.05 vs. controls, comparing mean values by ANOVA with Dunnett’s multigroup comparison.

Adverse effects

None of the volunteers who received erythromycin reported any side-effects, while three subjects in the isosorbide group reported mild headache.


We have previously shown that SPECT-Analyze™ imaging allows in vivo measurement of gastric volume and the meal-induced accommodation response, with low intra- and inter-observer variation.21 In the present study, our results validate this novel technique by demonstration of the pharmacological modulation of gastric volumes.

A potential weakness of the current method is the inability to detect redistribution of content or of volume of proximal and distal stomach in the absence of a change in overall volume. This is particularly important in view of the observations of Troncon et al.17 on intragastric distribution of the meal in dyspepsia. To address this pitfall, we are developing an automated programming method to separate the stomach regions for future studies.

Erythromycin reduces the postprandial gastric accommodation response. Thus, the postprandial gastric volume, ratio of postprandial over fasting volume, and volume difference between fasting and postprandial periods were all significantly lower in the erythromycin group relative to the untreated group. This is consistent with the results obtained with an intragastric barostatic balloon by Bruley des Varannes et al.,28 who found increased postprandial proximal gastric tone using a higher dose of erythromycin (i.e. 3 mg kg−1).

There was no significant difference in fasting gastric volumes between erythromycin and untreated groups. Bruley des Varannes and colleagues28 found that erythromycin, at a dose of 1.5 mg kg−1 over 10 min, has a gradual effect on fasting proximal gastric tone with a maximal effect at approximately 5 min after infusion. We infused intravenous erythromycin at the initiation of fasting SPECT acquisition without an equilibration period, and the 10-min duration of the fasting volume acquisition probably included a period when the effective drug level was not yet achieved. In contrast, effective drug levels were achieved postprandially to reduce the accommodation response.

It is also conceivable that the lack of demonstrable effects of erythromycin on fasting gastric volume represents a ‘floor effect’, such that the volume is so low that it cannot be reduced further, or that the noninvasive SPECT technique is not sufficiently sensitive to detect the volume change induced by erythromycin during fasting. Alternatively, erythromycin may interfere with the noncholinergic vagally mediated component of the accommodation reflex or, more likely, directly activate smooth muscle motility receptors, inducing a contractile response that limits the normal reflex accommodation. This is suggested by careful review of the barostatic recordings,28 which show an initial, fairly normal postprandial accommodation, followed by a reduction in gastric volume after a few minutes.

The effects of exogenous nitric oxide donors on the gastric motor function in humans are still poorly understood. In our study, isosorbide significantly enhanced the fasting gastric volume. The relaxation effect of isosorbide, a nitric oxide donor, is consistent with some of the findings of in vitro animal and human studies.35,36,38,47,48

The effects of nitric oxide donors on postprandial proximal stomach volume have been controversial. Thumshirn et al.,38 using an intragastric barostatic balloon, found that intravenous nitroglycerine enhances postprandial proximal gastric volume in healthy subjects. Using ultrasonography in diabetic patients, others have shown that sublingual nitroglycerine has no effect on the proximal stomach but reduces the antral area postprandially.39 Hausken and Berstad have shown, using ultrasonography, that nitroglycerine reduces the pre- and post-prandial antral area in patients with functional dyspepsia.49 It was postulated that nitric oxide donors enhance gastric fundal accommodation, reducing postprandial antral filling. It is possible that nitric oxide donors may have different effects on the proximal and distal gastric volumes, or that ultrasonographic measurement of postprandial gastric volume measurement depends on the distribution and volume of gastric content and, hence, the gastric emptying rate. In rats, inhibition of nitric oxide synthesis inhibits the gastric emptying of liquids.50 Conversely, it is possible that nitric oxide donors may enhance the liquid emptying rate, decreasing the antral gastric volume and resulting in lower measurements of the antral area.

Contrary to our anticipation, we found that isosorbide had no significant effect on the postprandial gastric volume using SPECT. In fact, the mean postprandial volumes were very similar between the isosorbide and untreated groups (Table 2). Thus, the action of isosorbide does not ‘add on’ to the normal postprandial gastric accommodation response, suggesting that, under physiological conditions, there is a natural limit to the extent of relaxation of the stomach with the ingestion of food. Studies of gastric compliance show that the volume response to pressure load eventually comes to a plateau, suggesting that restriction by connective tissue may have limited the active relaxation induced by the nitric oxide donor administered sublingually at the dose used in this study.

Thumshirn et al. found that a meal induced accommodation even after increasing fasting volume with i.v. nitroglycerine.38 The final postprandial gastric volume measured by an intraluminal barostatic balloon was higher than volumes reported in controls receiving no treatment.38 Thus, the dose of nitrate in our study may have been too low to increase the normal meal-induced accommodation; however, this dose was chosen because of its tolerability and potential application in future therapeutic trials. It is, nevertheless, conceivable that the noninvasive SPECT method may not be sensitive enough to detect small increments induced by isosorbide over the normal accommodation response.

With increased fasting and unaltered postprandial volumes, the ratios of postprandial volume over fasting volume and the changes in volumes between fasting and postprandial periods were significantly lower in the isosorbide group; clearly, the abnormal ratio resulted from the increased fasting volumes.

We had postulated that body mass index influenced gastric volumes or the accommodation response; e.g. it was conceivable that intra-abdominal pressure might increase with body mass index, restricting the gastric volume. Our data suggest that fasting and postprandial gastric volumes, as well as the gastric accommodation response, are unrelated to gender or body mass index. This is in agreement with a previous study by Klatt and colleagues,55 who found that gastric accommodation response to distension is not influenced by age, gender or body mass index. The age range of the participants in our study was too narrow to effectively address the role of age.

Does gastric emptying interfere with SPECT measurement of gastric volume? We believe this is unlikely for several reasons. Firstly, the method images the gastric wall, not the gastric content. Secondly, the measurements during the first and second 10 min postprandially are remarkably constant within the same individual, despite the fact that up to 50% of a small liquid nutrient meal will be emptied during that 20 min, especially after i.v. erythromycin. Hence, unlike other volume measurements that are based on imaging of the intragastric content (e.g. ultrasound, MRI), SPECT imaging appears less vulnerable to the effect of gastric emptying.

To date, most pharmacological studies of the gastric accommodation response have used the barostatic balloon. It is possible that some of the differences in results achieved by studies using the barostat and other noninvasive techniques such as SPECT imaging and ultrasonography may reflect a fundamental difference in the techniques. For instance, the placement of the barostatic balloon and imposition of an intragastric pressure clamp may alter the gastric accommodation response in an unpredictable manner. Samsom et al.56 have provided preliminary evidence to suggest that antral relaxation, as measured by ultrasound, is disturbed by the concomitant placement of a barostatic balloon in the fundus. Another potential explanation for differences between results obtained with barostat and SPECT imaging is the presence of a continuous distending drive during isobaric tone measurements with the barostat. It is conceivable that such a drive is needed to obtain an increase (with isosorbide dinitrate) of an already appropriate postprandial relaxation in healthy subjects.

One major advantage of SPECT is that it allows noninvasive measurement of gastric volumes without introducing any artefacts that may occur with more invasive methods. In the present study, we confirmed our previous concern21 that the method of dividing the stomach into proximal and distal regions is somewhat arbitrary because the incisura of the stomach, the anatomical landmark of division, is often not prominent, especially in the fasting images. Hence, we have restricted our analyses to total gastric volumes. We believe this is the biologically relevant analysis because several studies, including our previous one,21 have documented relaxation of the gastric antrum postprandially. However, in a recent study, we validated SPECT with simultaneous measurement, using a barostatic balloon, of a defined segment of the proximal stomach in healthy individuals. We placed a gamma-emitting marker (111In resin pellets) in the tip of the balloon in order to identify the segment of stomach evaluated by the two techniques. Those studies show a clear and significant correlation of absolute volume changes as well as ratios of volume change postprandially.22


We believe that SPECT-Analyse™ imaging is a useful, noninvasive tool for assessing the gastric accommodation response in humans; it allows in vivo assessment of the effects of pharmacological agents on gastric volumes. Further validations of the ability of the SPECT method to accurately detect a volume load added to the barostatic balloon are ongoing as part of the process to adequately validate this method. In addition, as a result of these initial studies, we have been able to reduce the dose of i.v. pertechnetate to 10 mCi, while maintaining sufficient contrast on imaging of the abdomen to identify and reconstruct the stomach with the segmentation algorithm used.


This study was supported in part by grants RO1-DK54681-03 and K24-DK02638-03 (Dr M. Camilleri) and by General Clinical Research Center grant (no. RR00585) from the National Institutes of Health. The authors thank Mrs Cindy Stanislav for excellent secretarial assistance.