Performance characteristics of scintigraphic transit measurements for studies of experimental therapies

Summary

Background : The intra- and inter-individual reproducibility of gastrointestinal and colonic transit tests require full characterization.

Aims : (i) To characterize the normal values and reproducibility effects of age and gender on the scintigraphic transit of solids in health. (ii) To compare scintigraphic and radio-opaque marker measurements of colonic transit. (iii) To estimate demonstrable effect sizes for different transit end-points based on observed variations.

Methods : A scintigraphic gastrointestinal and colonic transit study and the mean colonic transit time were measured using radio-opaque markers in 37 healthy volunteers; 21 subjects had a repeat scintigraphic test 3 weeks later.

Results : Gastric emptying at 4 h was highly reproducible (coefficient of variation, 4%) on repeat testing. The colonic measurement varied by more than 1 geometric centre unit in 37% of subjects at 24 h and in 26% of subjects at 48 h. There were no age- or gender-related differences in transit. Effect sizes demonstrable with 14 subjects per group were in the range previously shown to be clinically relevant: 25% change in gastric emptying at 4 h; 1.5 geometric centre unit change in colonic transit at 48 h.

Conclusions : These data demonstrate the reproducibility and performance to be expected of transit measurements and are essential for designing studies in experimental therapeutics.

Introduction

Several studies have reported considerable variability in the rates of gastrointestinal transit in asymptomatic, healthy subjects. The scintigraphic measurement of gastric emptying, first introduced in 1966,¹ is widely accepted by clinicians and researchers. Scintigraphy was introduced for the quantitative evaluation of colonic transit in 1986, with initial measurements performed by injecting isotope through an oro-caecal tube.² At different centres, there are considerable variations in the execution, performance and reporting of tests. Radiological assessment of colonic transit using radio-opaque markers is widely used, because it is simple and relatively inexpensive. Colonic transit investigations, with radioisotopes delivered either with a meal or via a delayed-release capsule, have been extensively used to characterize transit and motility in patients with functional bowel syndromes and to evaluate the effects of novel therapies.^3–11 Test reproducibility has not been thoroughly evaluated in healthy subjects or in disease states. Degen and Phillips performed such studies using ^99mTc-resin pellets to evaluate gastric and small bowel transit and ¹¹¹In-resin pellets to evaluate colonic transit.¹² Resin pellets are no longer used because of the need for cumbersome regulatory reporting. Thus, we developed and validated a process of radiolabelling activated charcoal particles with ¹¹¹In to measure colonic transit.¹³

Our overall aim is to develop a reproducible technique to evaluate the effect of experimental therapies on gastrointestinal motor function. As a prerequisite for future studies, the specific aims of the present study were: (i) to characterize the normal values and reproducibility of the gastrointestinal and colonic transit of solids in health; (ii) to assess the results across the age range 18–60 years and in both genders; (iii) to compare data obtained with the scintigraphic method with the results obtained with astandardized radio-opaque marker technique for the measurement of colonic transit in health; and (iv) to use the coefficients of variation observed to estimate the sample sizes necessary to demonstrate specified effect sizes for gastric, small bowel and colonic transit end-points in future studies of experimental therapies.

Methods

Participants

Healthy volunteers were recruited through public advertisement. Current gastrointestinal symptoms and a previous history of gastrointestinal disease or abdominal surgery, other than appendectomy, tubal ligation or hysterectomy, were exclusionary criteria. Participants were screened with an abridged gastrointestinal symptom questionnaire. Subjects with any acute symptom or currently using medications known to change gastrointestinal motility were not included in the study. Women with a pregnancy within the last year were also excluded. Thirty-seven healthy volunteers entered the study. Written informed consent was obtained from each subject after discussion of the procedure. Females participating in the study had to demonstrate a negative β-chorionic gonadotropin pregnancy test within 48 h before the first scintigraphic measurement.

Subjects were asked to maintain their usual dietary habits during the week prior to the start of the procedure and in the period between repeated measurements. They were advised to avoid unusually heavy physical activity during the 48-h course of the transit measurements.

Experimental design

Figure 1 summarizes the experimental design. All 37 participants simultaneously underwent scintigraphic and radio-opaque marker procedures for gastrointestinal and colonic transit. Subjects ingested the capsule containing radio-opaque markers at 06.00 h (day 1), immediately before starting the scintigraphic study, and at the same time on days 2 and 3. An abdominal radiograph was taken at 06.00 h on day 4, that is 72 h after the first markers had been ingested. The scintigraphic study lasted 48 h.

Figure 1. Transit measurements. Experimental design. X-ray, radiograph.

Three weeks after the initial evaluation, 21 of the 37 volunteers repeated the scintigraphic transit test using the same procedure. The remaining 16 subjects were unable to attend for repeat transit testing.

Scintigraphy

After an overnight fast, subjects were given a methacrylate-coated, delayed-release gelatin capsule with an 85 mL glass of water. The gelatin capsule contained 5 mg of dried activated charcoal (Eli Lilly, Indianapolis, IN, USA) with adsorbed isotope. ¹¹¹InCl₃ (0.1 mCi) had previously been mixed with charcoal in a slurry. On drying, the charcoal with adsorbed isotope was packed into a gelatin capsule, which was then immersed in a methacrylate (Eudragit S100; Rohm Pharmaceutical, Malden, MA, USA) solution as in previous studies.¹³ The methacrylate coating dissolves in the alkaline pH of the terminal ileum. Markers placed on the anterior superior iliac crest facilitated identification of the small bowel, in order to ascertain that the capsule had emptied from the stomach before feeding the radiolabelled meal. Subjects then received a standard breakfast consisting of two scrambled eggs in two slices of buttered bread and 1240 mL of skimmed milk (296 kcal total; 32% protein, 35% fat, 33% carbohydrate). ^99mTc-sulphur colloid (1.0 mCi) had been added to the two eggs during the cooking process. Another standardized 535-kcal meal was provided 4 h after ingestion of the radiolabelled egg. Subjects were free to follow their usual dietary habits from 8 h onwards.

Gamma camera imaging and data analysis

Anterior and posterior images were obtained with the subject standing using a large field-of-view gamma camera with a medium-energy, parallel-hole collimator (GE Starcam, Milwaukee, WI, USA). Images were acquired at 0, 1, 2, 4, 6, 24 and 48 h after radioactive meal ingestion. ^99mTc counts were quantified within a 140 keV (± 20%) window. ¹¹¹In quantification used a 247 keV (± 20%) window. Geometric means between anterior and posterior counts were used to correct for tissue attenuation, and counts were corrected for isotope decay.

The analysis of the data was performed as in previous studies using a variable region of interest program. Gastric and four different colonic regions of interest were drawn to quantify isotope counts. The ascending (AC), transverse (TC), descending (DC) and combined rectosigmoid (RS) regions were analysed separately. The primary summaries for transit profiles were the percentage of radioisotope emptied from the stomach (gastric emptying) at 1, 2 and 4 h, the gastric half-emptying time from linear interpolation of the gastric residuals, the percentage of colonic filling at 6 h and the colonic geometric centre at 4, 24 and 48 h.

The geometric centre is the weighted average of the counts obtained from the different colonic regions. To obtain the geometric centre, the proportion of counts ineach region is multiplied by the region's weighting factor, which is one for AC, two for TC, three for DC and four for RS. The isotope count in the stool was inferred as 100% of the total count minus the isotope count remaining in the colon, as previous studies have shown that this approximates to the actual count in the stool. The weighting factor for the stool is five. Thus, at any time, geometric centre=(%AC × 1 + %TC × 2 + %DC × 3 + %RS × 4 + %stool × 5)/100. All of these summaries have been used extensively in past literature, and have been validated against more detailed transit profiles using receiver operating characteristic curves.¹⁴ Colonic filling at 6 h has been proposed, but is a relatively poor surrogate of the small bowel transit time.¹⁵

Radio-opaque markers

Radio-opaque markers were used to calculate the mean colonic transit time, following the method proposed by Metcalf et al.¹⁶ On each of three consecutive days, starting from the first morning of scintigraphic testing, all 37 subjects received a capsule containing 24 radio-opaque markers (Sitzmarks, Konsyl Pharmaceuticals, Fort Worth, TX, USA). A plain abdominal radiograph was obtained on day 4, and the number of markers in each colonic region was counted. If clear outlines of the bowel were not visible, markers situated to the right of the vertebral spinous processes were assigned to the right colon; those to the left of the spinous processes and above a line drawn from the fifth lumbar vertebra to the left anterior superior iliac crest were assigned to the left colon. Markers located inferior to the line from the pelvic brim on the right to the superior iliac crest on the left were assigned to the sigmoid colon and rectum.

Statistical analysis

Power exponential analysis¹⁷ was used to estimate t_1/2 and t_5% (lag time). End-point summaries for regional transit, calculated from scintigraphic scans, were expressed as the mean ± S.E.M. Bland–Altman plots¹⁸ were used to assess the agreement between paired scintigraphic observations in the group of 21 subjects. The difference between the results in the two repeated measurements was plotted against the mean values ofthe repeated results, and evaluated against the mean ± s.d. for the results. Colonic transit results by scintigraphy (geometric centre at 24 and 48 h) and by the radio-opaque marker technique (mean colonic transit time) were compared using Bland–Altman plots and residuals from the line of fit. The residuals from the line of fit on Bland–Altman plots also provided a measure of the variability, in geometric centre units, for a given mean colonic transit time in the 37 subjects.

The Wilcoxon rank-sum test and Kruskal–Wallis test were used to test differences in transit results with respect to the gender and age (subdivided into decades) of the subjects.

Based on the inter-individual variance in the data from 37 healthy volunteers in the first test, we also calculated the appropriate sample size for future two-group comparison studies to detect a priori defined clinically relevant differences between end-points for gastrointestinal transit^{3, 4, 14, 16, 19, 20} with a power of 80%, assuming α=0.05. For example, for gastric emptying, the clinically relevant difference in the gastric residual was 25% at 4 h, based on previous results showing a median of 2% in health (interquartile (IQ) range, 0–40%) and 60% in patients with gastroparesis (IQ range, 41–68%).¹⁵ For colonic transit, 1.5 colonic regions at 48 h was deemed to be clinically relevant, based on previous results in patients with idiopathic constipation,³ diarrhoea-predominant irritable bowel syndrome⁴ and carcinoid diarrhoea.²⁰ A sample size of at least 20 subjects was adequate, with a power of 80% and α=0.05. Based on the desire to detect differences in transit end-points between health and disease,^{3, 4, 14, 16, 19, 20} we calculated that a sample size of at least 20 would be required to characterize intra-individual variation, for the purpose of future planning of cross-over pharmacotherapeutic studies. Our sample size was not planned to study the effects of age and gender on transit end-points. However, in the study by Degen and Phillips,¹² which focused primarily on age- and gender-related differences in gastrointestinal transit by scintigraphy, the overall sample size was slightly smaller than ours (32 vs. 37 subjects), with a higher proportion of females relative tomales (12/32 females in the study of Degen and Phillips;¹² 10/37 females in the present study).

Results

Reproducibility of scintigraphic transit measurements

Bland–Altman plots showed high test reproducibility (Figure 2). Differences between the repeat measurements were within ± 10% for 70% of the data for gastric emptying at 1 h and gastric emptying at 2 h. Gastric emptying at 4 h appeared to offer the best reproducibility, with only 14% of participants showing differences of more than 10% on repeat measurements. The calculated stomach half-emptying time t_1/2 showed a higher variance, with 60% of points differing by more than 10 min and 34% of points differing by more than 25 min on repeat measurements. Variance in the percentage of colonic filling at 6 h was greater than 10% in 45% of participants. Colonic measurements varied by more than 1 geometric centre unit in 37% of subjects at 24 h and 26% of subjects at 48 h. The coefficients of variation for all end-points are shown in Table 1.

Figure 2. Bland–Altman plots to assess the reproducibility of gastric emptying (GE) at 1 h (a), 2 h (b) and 4 h (c), the gastric half-emptying time ( t_1/2 ) (d), colonic filling (CF) at 6 h (e) and the colon geometric centre (GC) at 24 h (f) and 48 h (g). Note the best reproducibility is demonstrated in plots (c) and (g).

Table 1.  Details of gastrointestinal and colonic transit end-points (mean ± S.E.M.) with the coefficient of variation between subjects ( C_var intersub) and between repeated measurements ( C_var intrasub)

Group	n	GE 1 h (%)	GE 2 h (%)	GE 4 h (%)	GE t1/2 (min)	CF 6 h (%)	GC 24 h	GC 48 h
CF, colonic filling; GE, percentage of radioisotope emptying from the stomach; GE t1/2, gastric half-emptying time; GC, geometric centre.
Whole	37	25 ± 2	58 ± 4	90 ± 1	112 ± 6	71 ± 5	2.67 ± 1.09	3.89 ± 0.15
Baseline	21	26 ± 4	54 ± 4	89 ± 1	116 ± 8	77 ± 4	2.81 ± 0.22	3.93 ± 0.22
At 3 weeks	21	25 ± 3	60 ± 3	92 ± 1	104 ± 6	71 ± 5	2.68 ± 0.22	4.00 ± 0.2
Cvar intersub	21	62	29	8.1	30.4	30	37	24
Cvar intrasub	21	20	14	4	14	19	28	14

Effect of age and gender

All 37 subjects (mean age, 39 ± 11 years; 10 females) completed the initial combined scintigraphic and radio-opaque marker transit evaluations, and 21 subjects (mean age, 36 ± 9 years; eight females) repeated thescintigraphic transit measurement 3 weeks later. Results from the first scintigraphic test are shown in Table 1. Within the age range of the participants in the study, none of the end-points (scintigraphic or radio-opaque marker) was significantly influenced by age (Figure 3). There were no significant gender differences with respect to all the scintigraphic and radiological end-points measured (Figure 4).

Figure 3. Effect of age on gastric emptying (GE) at 4 h (a), the gastric half-emptying time t_1/2 (b) and the colon geometric centre (GC) at 24 h (c) and 48 h (d). Differences are not statistically significant.

Figure 4. Box plots showing gender-related differences in gastric emptying (GE) at 4 h (a), the gastric half-emptying time t_1/2 (b) and the colon geometric centre (GC) at 24 h (c) and 48 h (d). Differences are not statistically significant.

Scintigraphy compared to radio-opaque markers to measure colonic transit

Residual plots from the line of fit for scintigraphic end-points vs. the mean colonic transit time (marker count at 72 h) show that the variances in results between the two techniques are smaller in the lower range of the mean colonic transit times (Figure 5). For the geometric centre at 24 h, residuals were within 1 geometric centre unit in 24 of the 37 subjects (65%). For the geometric centre at 48 h, residuals were within 1 geometric centre unit in 27 of the 37 subjects (72%) (Figure 5). Age, gender and body mass index did not significantly influence the degree of agreement between scintigraphy and the mean colonic transit time.

Figure 5. Residual plots from the line of fitting showing the colon geometric centre (GC) at 24 h (a) and 48 h (b) vs. the radio-opaque marker count. In almost 75% of subjects, residuals for colonic transit by scintigraphy are within 1 geometric centre unit across the range of colonic transit measured by the radio-opaque marker method. MCTT, mean colonic transit time.

Effect sizes detectable and sample sizes for future studies

The sample sizes needed to detect clinically meaningful differences in end-points for gastrointestinal transit studies are shown in Table 2. The percentage of gastric emptying at 4 h and the colonic geometric centre at 48 h appear to need the smallest number of subjects todetect a difference of 25% in gastric emptying at 4 h and a difference of 1.5 geometric centre units at 48 h.

Table 2.  Calculation of sample sizes needed for the detection of end-point transit differences (effect size detectable) in studies in healthy volunteers, with 80%, α=0.05, based on our study data. The sample size is the number of subjects per group based on a two-sample t -test (i.e. treatment vs. placebo). The effect size detectable for a sample size of 14 subjects per group is also shown

	GE 1 h (%)	GE 2 h (%)	GE 4 h (%)	CF 6 h (%)	GE t1/2 (min)	GC 24 h	GC 48 h
CF, colonic filling; GE, percentage of radioisotope emptying from the stomach; GE t1/2, gastric half-emptying time; GC, geometric centre.
Effect size detectable (ESD)	25	25	25	25	45	1.1	1.5
Sample size (80%)	13	20	6	41	22	23	14
ESD with 14 subjects per group	24	31	14	46	59	1.7	1.5

Discussion

In the present study, the reproducibility of gastrointestinal and colonic transit measurements varied in different regions. Gastric emptying showed greater reproducibility than small bowel or colonic transit.

Previous studies have reported that both the inter- and intra-individual variability in gut transit in healthy subjects is fairly large.^3, 12, 21 The variability in the present study was lower for some of the end-points considered; this information is important for selecting the optimal conduct of diagnostic or experimentaltherapeutic studies. For example, intra-subject coefficients ofvariation for gastric emptying at 1 h and at 2 h were20% and 14%, respectively. On the other hand, the intra-subject coefficient of variation of gastric emptying at 4 h was 4%. Gastric emptying at 2 h and4 h has already been shown to discriminate between healthy and diseased subjects in previous studies of neuropathic and myopathic motility disorders,¹⁵ and inan international, multicentre study which establishedcontrol standard values of gastric emptying in normal subjects using ⁹⁹Tc-labelled egg substitute.²²

Others have evaluated the reproducibility of the gastric half-emptying time t_1/2 for solids in health. Degen and Phillips performed two estimations and found that the range of variation in t_1/2 was 66 min in women and 183 min in men, but the means and medians of intra-individual differences were not significantly different.¹² Both principles were confirmed by our study. In contrast, Brophy et al. performed the test for the gastric emptying of solids using ^99mTc-sulphur colloid on four different occasions, and found that the intra-individual values of t_1/2 differed by between 15 and 35 min. However, the study was performed in only eight subjects.²¹ Petring and Flachs performed four repeated measurements of the gastric residual at 90 min and identified intra-individual variations of 15–39%.²³

The current study extends these observations to provide estimates of the variability for small bowel and colonic transit. Colonic filling at 6 h, a surrogate of oro-caecal and small bowel transit, shows marked intra-individual variance. We agree with Argenyiet al. that its diagnostic usefulness is limited, particularly when gastric emptying is delayed, making colonic filling at 6 h essentially uninterpretable.²⁴ The estimated detectable effect sizes also suggest that only large effects of therapies would be detectable in typically sized pharmacodynamic studies.

For colonic transit, the geometric centre at 48 h appears to be more reproducible than the geometric centre at 24 h. Approximately one subject in four presents a variation of more than one ‘colonic region’ inthe geometric centre at 24 h on repeat testing. This appears to reflect the natural variation in colonic transit even among healthy individuals. Metcalf et al. noted that the range of normal colonic transit (with radio-opaque markers) in health is 11–68 h.¹⁶ Roberts et al., using ¹¹¹In-diethylene-triaminepenta-acetic acid in the aqueous phase of the meal to measure colonic transit (incontrast to the colonic release used in our studies), determined that the geometric centre at 48 h was the best index to predict delayed colonic transit.⁷ Notghi et al., reviewing data from a large series of patients, observed that transit could be optimally summarized byimages taken at 10 h, 24 h and 48 h.⁸ Published therapeutic trials^9–11 showed significant effects of agents that influenced colonic transit at 24 h and 48 h. Indeed, the colonic geometric centre provides a convenient way to summarize data on scintigraphic transit by several laboratories,^5, 6, 25 and is considered by Notghi et al. to be more appropriate than frame-based curves with radioisotope activity plotted vs. time, or the percentage activity in the colon assessed at definite times.⁸

Variations in transit end-points were not influenced by age; the age range of our participants was comparable to the ranges from other studies. However, we did not study subjects over 65 years of age, and the sample size used was not intended to formally assess the effects ofage and gender. From the published literature, studies have been reported in favour or against a significant or clinically meaningful difference with the phase of the menstrual cycle, gender and age.^12, 26–34 Our sample size was not planned to study the effects of age and gender on transit end-points. However, in the study by Degen and Phillips,¹² which focused primarily on age- and gender-related differences in gastrointestinal transit by scintigraphy, the overall sample size was slightly smaller than ours (32 vs. 37 subjects), with a higher proportion of females relative to males (12/32 females in the study of Degen and Phillips;¹² 10/37 females in the present study).

Based on the data from our group of healthy volunteers, we calculated the sample size necessary for future studies to identify differences in gastrointestinal and colonic transit end-points between two groups, e.g. active treatment and placebo groups. We also estimated the effect size detectable in the same end-points using a sample size of 14 subjects per group, which is often used in experimental therapeutic trials. We concluded that gastric emptying at 4 h and the colonic geometric centre at 48 h were the end-points needing a smaller sample size for a clinically relevant effect size difference to be detected between two groups. Similarly, they were the end-points in which a smaller difference could be detected with a sample size of 14 subjects per group. These sample sizes should be regarded as the minimum needed for each study group in future pharmacodynamic studies in healthy volunteers.

Because the variation may be smaller in patients, the sample sizes needed to demonstrate effects of the same magnitude may be smaller, but this requires further validation in relevant disease states.

Acknowledgements

This study was supported in part by grants R01 DK-54681-04 (MC) and K24 DK-02638-04 (MC) from the National Institutes of Health.