Functional abdominal pain and symptoms of irritable bowel syndrome (IBS) are common in adolescents and frequently result in use of health care resources (1). Disorders of gastrointestinal transit and accommodation contribute to the development of dyspepsia and abdominal pain in children (2,3). Other studies of children with functional abdominal pain showed hypersensitivity of hollow organs such as the stomach (4). Disturbances of gastric function, as well as age and sex, account for 50% of the variance in the symptoms of adult patients with dyspepsia (5).
To date, most of the literature investigating gastric accommodation has used the barostat balloon (as per oral intubated measurement of tone) or 99mTc-single photon emission computed tomography (SPECT), which measures gastric volume (GV). SPECT has some disadvantages that reduce its applicability in children or adolescents: exposure to ionizing radiation, the need for an intravenous injection, and the need for specialized equipment. The radiation exposure limits its application in females who are known to be pregnant and in children, or when repeated measurements are needed such as in therapeutic studies. As a result, there are no normal SPECT-based data documenting the normal GV during fasting and after a meal in children.
Alternative noninvasive approaches that avoid ionizing radiation include 3-dimensional ultrasound (3D-US) and magnetic resonance imaging (MRI) (6–10), which is used at a few specialized centers; the 2 methods require further validation (11). Using 3D-US, Olafsdottir et al (12,13) have shown impaired proximal stomach accommodation and intragastric maldistribution of a liquid meal in children with functional abdominal pain.
Ultrasonography has several potential advantages, including competitive pricing, point of care capability, and lower likelihood to induce stress. Operator dependence and technical challenges associated with body habitus and abdominal pannus may compromise the ability of ultrasound to measure GV. Therefore, there is a need for formal validation of the 3D-US technique and to obtain data from healthy adolescent controls to be used in assessing the GV of symptomatic children and adolescents. The availability of a noninvasive method to measure GV without radiation exposure may usher in a novel clinical evaluation for children and adolescents with abdominal pain and dyspepsia, may allow for longitudinal studies, may help explore the effects of medications, and may facilitate study of the mechanisms of patients' symptoms.
The study hypothesis was that, in healthy participants, 3D-US is accurate, reproducible, and can be applied to measure fasting and postprandial (PP) GV in adolescents. The study aims were to compare fasting and PP GVs measured simultaneously by 99mTc-SPECT (standard test) and 3D-US in adults; to assess the performance characteristics of 3D-US measurement of GV, interindividual coefficient of variation (COV) and intraindividual COV for 2 3D-US estimates of fasting and PP GV; and to develop normative data for fasting and PP GV in 24 asymptomatic adolescents.
PATIENTS AND METHODS
Twelve healthy adults underwent SPECT and 3D-US in random order; 1 week after the first ultrasound examination, a second ultrasound examination was conducted in all of the participants. In the next phase of the study, 24 adolescents (12 males, 12 females; 12 ages 13–15 years, 12 ages 16–17 years) underwent a single ultrasound determination of fasting and PP GV.
The protocol was reviewed and approved by the Mayo Clinic institutional review board, and signed informed consent was obtained. The study enrolled 12 healthy adults and 24 healthy adolescents. Adult participants were 18 to 65 years old and adolescents were 13 to 17 years old. The main eligibility criteria were body mass index (BMI) of 18 to 32 kg/m2, negative urine pregnancy test for women of childbearing potential, absence of gastrointestinal symptoms on an abridged version of the validated Bowel Disease Questionnaire (14), and lack of medications, prior surgeries, or illnesses that could interfere with the conduct or interpretation of the study.
SPECT Procedure to Measure Fasting and Postprandial Gastric Volume
SPECT exam was conducted as described in several studies from our laboratory in the last decade (15). We used a noninvasive method to measure GV during fasting and 30 minutes after 300 mL of Ensure (316 kcal) using SPECT. Subjects were placed in a supine position on the imaging table with the detectors over the upper abdomen. Participants received an intravenous injection of 99mTc-sodium pertechnetate, which is taken up by the parietal and nonparietal cells of the gastric mucosa, allowing visualization of the stomach wall. At 10 minutes after the intravenous injection of the radioactive marker, tomographic images of the gastric wall were obtained throughout the long axis of the stomach using a dual-head gamma camera (Multispect II, Siemens, Malvern, PA) that rotates around the body. This allows assessment of the radiolabeled circumference of the gastric wall rather than the intragastric content.
Using image processing libraries (AVW 3.0, Biomedical Imaging Resource, Mayo Foundation, Rochester, MN), a 3-dimensional rendering of the stomach was obtained and its volume calculated in milliliters. We previously have validated the method in vitro and in vivo (15). In healthy volunteers, simultaneous measurements of PP GV changes with SPECT and the barostat balloon device (the gold standard to measure GV) were strongly correlated (15). There is high intraobserver reproducibility of measurements of GV with this technique. Intraobserver COV (at average 9 months) is ∼13% for SPECT measurements (16).
A commercially available ultrasound scanner (Voluson E8, GE Healthcare Ultrasound, Wauwatosa, WI) was used with an external transducer (1.4–5.8 MHz, model RAB 2-5-D) capable of automated volume acquisition through an elevational sweep angle of up to 85°. Data were stored on the Voluson E8 local hard drive. Data processing and volume estimation were calculated using the onboard proprietary software.
For 3D data acquisition, subjects were scanned immediately before (t = −5 minutes) and after 300 mL of Ensure (Ross Laboratories, Abbott Park, IL) drink ingestion (t = 0), followed by further imaging at t = 10, 20, and 30 minutes. Ultrasound datasets were acquired with the transducer held stationary in the position optimal for each individual to allow complete stomach visualization using appropriate breath control and patient positioning. This was necessary because the axis, shape, and size of the stomach vary between individuals and at different times in the PP period. Subjects were instructed not to move and the stomach was scanned via automated sweeps of approximately 5 to 10 seconds. When gastric contractions were observed during acquisition, data were discarded and the acquisitions repeated. All data acquisition and analysis were performed by 1 investigator (D.D.M.). Examples of gastric images by 3D-US and SPECT for the same subject immediately after ingestion of the 300 mL Ensure meal are shown in Figure 1.
Primary and Secondary Endpoints
Primary endpoints were: fasting, PP, and PP change in GV (delta GV) measured by 3D-US in adults and adolescents. Secondary endpoints were intra- and interindividual COVs in adults for fasting and PP GV and accuracy of 3D-US in measurement of fasting and PP GV, relative to simultaneous measurements by SPECT, using a Bland–Altman plot.
Sample Size Assessment
The primary emphasis of this study was estimation and assessment of the agreement of 3D-US with SPECT. From a previous study, the intraindividual COVs for SPECT imaging of 12 overweight and 11 obese adults were 26% fasting, 13% PP, and 19% for delta GV (17). The sample size used in the study was powered to detect clinically important differences, 50 mL in delta (PP fasting) GV, between 3D-US and SPECT and between the 2 measurements by 3D-US.
We compared fasting and PP GV measured by 3D-US and SPECT, calculated interindividual COV and intraindividual COV for GV by 3D-US, and assessed accuracy of 2 replicates of 3D-US using Bland-Altman plot. Intrasubject COV% (100×standard deviation [SD] of differences/grand mean of differences) were estimated and a Bland–Altman plot examined to check for bias associated with different GV. The distribution of 3D-US volumes in the 24 adolescents was summarized, and estimates of 25th and 75th percentiles were generated.
Study Subjects and Demographics
A total of 36 healthy subjects completed the study protocols. US images of 1 adult participant were not available for volume estimation (because of loss of data that could not be retrieved). The final data analysis included 35 participants, 11 adults (6 males and 5 females) and 24 adolescents (12 of each sex). The mean (±SD) age for the adult group was 34.2 ± 13 years, with mean BMI of 25 ± 3.65 kg/m2. The mean age for the adolescent group was 15.3 ± 1.37 years, with mean BMI of 21.3 ± 2.7 kg/m2. The studies were completed without adverse events and were well tolerated by all of the participants.
Adult Group Fasting and PP GV
Fasting, average 0 to 30 minutes PP, and delta (PP fasting) GVs in adults are summarized in Table 1. Note that the median gastric accommodation (delta GV) volume was greater than 300 mL, which was the meal volume, in the first US measurement and equal to the meal volume during the second 3D-US measurements. During the simultaneous measurements, delta GV was 444 mL (median, interquartile range [IQR] = 422, 535) for 3D-US and 543 mL (median, IQR = 486, 564) for SPECT (P = 0.15).
Fasting and PP GV in Adolescents
Fasting, average 0 to 30 minutes PP, and delta GVs in adolescents are summarized in Table 2. Figure 2 summarizes 3D-US data (median, IQR, and range). The estimated GVs from the adolescent group (median [25th–75th IQR]) were 33 (18–53) mL fasting, 330 (284–357) mL average 0 to 30 minutes PP, and 281 (240–324) mL for delta GV.
Effect of Sex, Age, and BMI on Gastric Accommodation by 3D-Ultrasound
There was no effect of sex on gastric accommodation in the combined group of adolescents and adults (male = 316.5 [246–351] mL; female = 316 [363–427] mL; P = 0.621). There was a weak overall correlation between age and gastric accommodation (r = 0.38; r2 = 0.145; P = 0.024); the correlation between BMI and gastric accommodation was not significant (r = 0.20; r2 = 0.04; P = 0.26).
Comparing US With SPECT for Measuring GV
A box and whiskers plot showing the median and IQR of both fasting and 0 to 30 minutes average PP GVs for SPECT and 3D-US is shown in Figure 3. Despite the difference in actual volumes, the magnitude of increase in volume seen after meal ingestion with SPECT is similar to that seen in 3D-US.
In Figure 4, the Bland-Altman plot compared delta GVs for both techniques. Note that most of the volumes were within 100 mL of difference. This reflects fairly good US estimates of gastric accommodation compared with SPECT.
Reproducibility of 2 Measurements of GVs by 3D-US
In Figure 5, the Bland-Altman plot compares delta GVs on the 2 US measurements. Intraindividual COV for both US measurements were 84% for fasting, 44% for average 0 to 30 minutes PP, and 60% for delta GVs.
In the present study, we demonstrated that 3D-US is feasible, can be conducted safely in adults and adolescents, and provides a fairly accurate estimate of GVs when compared with scintigraphy (SPECT) in healthy adults. 3D-US has been used previously to measure volumes of intraabdominal organs, including the stomach. In vitro, Gilja et al (18) have demonstrated excellent correlation between true and 3D-US estimated volumes of porcine kidney and fluid-filled porcine stomach (18,19). Another study also showed good correlation between MRI and 3D-US in volume estimation of human kidneys with low intraindividual COV for the repeated US volume estimates (20). However, the kidney is a solid organ and a fluid-filled stomach does not have a gastric air bubble. Therefore, we embarked on in vivo validation studies in humans.
Tefera et al (21) reported a close correlation and agreement between true and estimated volumes using 3D-US of a barostat bag filled with soup positioned in the proximal stomach of 6 healthy subjects. They also observed low interobserver variation. Although these studies suggest that 3D-US may allow a precise measurement of gastric accommodation, comparison of this technique with one of the gold standards, SPECT, has hitherto not been performed. To our knowledge, therefore, this is the first study comparing GV estimates by 3D-US to SPECT.
3D-US was able to detect GV accommodation to an Ensure meal in adults and adolescents. The increase in GV after the Ensure meal ingestion was equal to or greater than the meal size, 300 mL, in most participants in both age groups, reflecting true gastric accommodation. The median delta GV using SPECT was 543 mL compared with 444 mL using 3D-US (P = 0.15), reflecting fairly close estimation of gastric accommodation by US compared with SPECT. We showed a gastric accommodation equal to or greater than meal volume. The gastric accommodation volumes estimated by Gilja et al (19) were smaller than meal volume. This may be explained by the lower energy content of the meal used by Gilja et al, which was associated with faster gastric emptying, compared with the 1 kcal/mL nutrient liquid meal in our study.
Gilja et al also previously estimated GV and calculated the variability in volume measurement in 14 healthy adults using 3D-US (19). The interindividual COV was up to 52% PP compared with a maximum of 27% in our study. The intraindividual COV for PP GV, for 6 repeated US measurements conducted daily for 6 days in 1 subject, was 34.3% compared with 44% in the present study. The slightly smaller variability found by Gilja et al reflects greater familiarity with the individual subject anatomy when performing 6 studies in the same individual (19) rather than 2 studies in 11 individuals in our study.
There were some technical difficulties that contributed to smaller GVs recorded in fasting and PP subjects using 3D-US compared with SPECT. First, the presence of air in the stomach, mainly during fasting, made it difficult to identify the shadowed boundaries of the stomach. Second, the inability to obtain full stomach views, because the 3D-US elevational sweep does not exceed 85°, did not allow full detection of the distal parts of the stomach. This added to a distorted stomach shape, especially when pressure was applied through the US probe to achieve better contact of the probe with the abdominal wall, possibly resulting in smaller estimated GV. Finally, SPECT measures GV including gastric wall thickness (2), while 3D-US measures only intragastric volume. Therefore, it is likely that SPECT overestimates true GV measurements, as shown by Bharucha et al (11) in comparison with MRI.
There was an additional difficulty encountered in obtaining 3D-US images in the adolescent group. The US transducer was not maintained in full contact with the adolescent's abdominal wall, owing to a more lean body habitus. To ensure there was a medium to conduct the ultrasound waves, extra gel was applied to form a “wedge” to fill in the gap between the transducer and the abdominal wall.
The higher average fasting GV seen on the second US in adults compared with the first US measurement in the same adults reflects improved volume estimates as part of a learning curve with a new technique. With experience, the lateral decubitus position seemed to provide an optimal view of the stomach. This change in technique may have contributed to the high intraindividual COV (84%) between 2 fasting 3D-US measurements.
The intraindividual COV (44%) of PP GV between the 2 3D-US measurements is also, in part, a reflection of possibly 2 factors. First, there may be a day-to-day variation in the normal changes in GV in response to a meal. It was previously demonstrated that the intraindividual COV is 13% for PP GV using SPECT (16,17). Second, the change in position during the second 3D-US imaging series to the left lateral decubitus position likely resulted in better gastric image quality and volume estimation.
The high intraindividual COV (84% for fasting, 44% for average 0–30 minutes PP, and 60% for delta GVs) of 2 repeated 3D-US measurements may limit its value in estimating GVs in studies that require repeated measurements. It is necessary that this high variation be minimized for the test to have clinical and research utility; 1 approach in future studies to reduce the variation is to standardize data acquisition, such as participant positioning and 3D-US probe placement. Therefore, further validation studies are needed to overcome the pitfalls encountered in the present study.
Given the coefficient of variation in the estimates of delta or accommodation volume, further validation studies comparing 3D-US with a current standard, such as SPECT, will require 52 participants to identify a mean difference of more than 50 mL, which would be a clinically relevant difference in the estimation of gastric accommodation by the 2 methods.
3D-US provides a fairly accurate estimate of gastric accommodation, when compared with SPECT. It is safe and may be a generally applicable method that can be used in adolescents with upper gastrointestinal symptoms. The variability seen with measurements reflects, in part, a learning curve. We perceive that the performance characteristics of the test will improve with further standardization of this technique.
We thank the nursing staff of the Mayo Clinic Clinical Research Unit and Cindy Stanislav for secretarial support.
1. Hyams JS, Burke G, Davis PM, et al
. Abdominal pain and irritable bowel syndrome in adolescents: a community-based study. J Pediatr 1996; 129:220–226.
2. Chitkara DK, Camilleri M, Zinsmeister AR, et al
. Gastric sensory and motor dysfunction in adolescents with functional dyspepsia. J Pediatr 2005; 146:500–505.
3. Chitkara DK, Delgado-Aros S, Bredenoord AJ, et al
. Functional dyspepsia, upper gastrointestinal symptoms, and transit in children. J Pediatr 2003; 143:609–613.
4. Di Lorenzo C, Youssef NN, Sigurdsson L, et al
. Visceral hyperalgesia in children with functional abdominal pain. J Pediatrics 2001; 139:838–843.
5. Delgado-Aros S, Camilleri M, Cremonini F, et al
. Contributions of GV and gastric emptying to meal size and postmeal symptoms in functional dyspepsia. Gastroenterology 2004; 127:1685–1694.
6. Gilja OH, Hausken T, Odegaard S, et al
. Ultrasonography and three-dimensional methods of the upper gastrointestinal tract. Eur J Gastroenterol Hepatol 2005; 17:277–282.
7. Gilja OH, Hatlebakk JG, Odegaard S, et al
. Advanced imaging and visualization in gastrointestinal disorders. World J Gastroenterol 2007; 13:1408–1421.
8. De Zwart IM, Mearadji B, Lamb HJ, et al
. Gastric motility: comparison of assessment with real-time MR imaging or barostat measurement initial experience. Radiology 2002; 224:592–597.
9. Hoad CL, Rayment P, Spiller RC, et al
. In vivo imaging of intragastric gelation and its effect on satiety in humans. J Nutr 2004; 134:2293–2300.
10. Camilleri M. New imaging in neurogastroenterology: an overview. Neurogastroenterol Motil 2006; 18:805–812.
11. Bharucha AE, Fidler J, Camilleri M, et al
. Comparison of GV measured by MRI and SPECT. Neurogastroenterol Motil 2007; 19:421–422.
12. Olafsdottir E, Gilja OH, Aslaksen A, et al
. Impaired accommodation of the proximal stomach in children with recurrent abdominal pain. J Pediatr Gastroenterol Nutr 2000; 30:157–163.
13. Olafsdottir E, Gilja OH, Tefera S, et al
. Intragastric maldistribution of a liquid meal in children with recurrent abdominal pain assessed by three-dimensional ultrasonography. Scand J Gastroenterol 2003; 38:819–825.
14. Talley NJ, Phillips SF, Wiltgen CM, et al
. Assessment of functional gastrointestinal disease: the bowel disease questionnaire. Mayo Clin Proc 1990; 65:1456–1479.
15. Bouras EP, Delgado-Aros S, Camilleri M, et al
. SPECT imaging of the stomach: comparison with barostat, and effects of sex, age, body mass index, and fundoplication. Single photon emission computed tomography
. Gut 2002; 51:781–786.
16. De Schepper H, Camilleri M, Cremonini F, et al
. Comparison of GV in response to isocaloric liquid and mixed meals in humans. Neurogastroenterol Motil 2004; 16:567–573.
17. Vazquez Roque MI, Camilleri M, Clark MM, et al
. Alteration of gastric functions and candidate genes associated with weight reduction in response to sibutramine. Clin Gastroenterol Hepatol 2007; 5:829–837.
18. Gilja OH, Thune N, Matre K, et al
. In vitro evaluation of three-dimensional ultrasonography in volume estimation of abdominal organs. Ultrasound Med Biol 1994; 20:157–165.
19. Gilja OH, Detmer PR, Jong JM, et al
. Intragastric distribution and gastric emptying assessed by three-dimensional ultrasonography. Gastroenterology 1997; 113:38–49.
20. Gilja OH, Smievoll AI, Thune N, et al
. In vivo comparison of 3D ultrasonography and magnetic resonance imaging in volume estimation of human kidneys. Ultrasound Med Biol 1995; 21:25–32.
21. Tefera S, Gilja OH, Olafsdottir E, et al
. Intragastric maldistribution of a liquid meal in patients with reflux oesophagitis assessed by three dimensional ultrasonography. Gut 2002; 50:153–158.