- Email: [email protected]
Towards a relatively inexpensive, noninvasive, accurate test for colonic motility disorders
GASTROENTEROLOGY
1992;103:36-42
Towards a Relatively Inexpensive, Noninvasive, Accurate Test for Colonic Motility Disorders MICHAEL Gastroenterology Minnesota
CAMILLERI
and ALAN R. ZINSMEISTER
Research Unit and Section of Biostatistics,
Currently available clinical tests of colonic transit, such as the radiopaque marker method, are useful to detect delayed transit but may be less sensitive for rapid transit. The aim of this study was to develop a relatively inexpensive, noninvasive, accurate test of colonic transit using selected scintigraphic observations within the first 24 hours after ingestion of a pH-sensitive, methacrylate-coated, delayed-release capsule containing “‘In-labeled resin pellets. The authors’ previously published coionic transit data on 22 healthy subjects, 9 patients with diarrhea-predominant irritable bowel syndrome, and 7 patients with idiopathic constipation and previously unpublished data on 4 patients with carcinoid diarrhea were analyzed. A logistic discriminant analysis was used to estimate the sensitivity and specificity of selected combinations or simple summaries of transit. Among combined transit summaries, the emptying rate of the proximal colon was significantly different between healthy and constipation groups; the geometric center of isotope in the colon at 4 hours was significantly greater in the diarrhea group than in healthy controls; the geometric center at 24 hours was significantly lower in the constipation group than in the other two groups. From the logistic discriminant analysis, simple summaries of transit also had significant discriminant value; these included the isotopic contents in the ascending, transverse, and descending colon at 4 hours and the counts in the ascending and transverse colon and stool at 24 hours. At 90% sensitivity, the specificity of the transverse colon counts at 4 hours was 79%, which is identical to the specificity of the proximal colon emptying rate, both adjusted for age. Thus, quantitation of isotopic counts in colonic regions on scans taken at 4 and 24 hours provides an accurate summary of colonic transit, with acceptable specificity at a high sensitivity in the detection of motility disorders of the colon. isturbances of colonic transit are commonly encountered in clinical practice and are generally confirmed by means of one of the radiopaque marker
D
Mayo Clinic and Mayo Foundation,
Rochester,
methods.1-3 These are robust, clinically applicable procedures that have gained wide acceptance because they are inexpensive and require little special expertise or equipment. They are particularly useful in the study of constipated patients.4 However, as the marker technique has been adapted to restrict radiation exposure by taking only one abdominal radiograph on the fourth day,3 its sensitivity for detecting rapid transit is reduced because the markers have usually traversed the colonic regions when the radiograph is obtained. Hence, this method may provide only a censored estimate of the actual colonic transit time. Alternative strategies using orocecal intubation and dynamic scintigraphy5v6 to measure colonic transit are applicable in the research arena. To overcome the need for orocecal intubation and to facilitate the study of regional colonic transit of solids, we recently developed a new scintigraphic method that uses a pH-sensitive, methacrylate-coated capsule containing radiolabeled resin pellets.’ Our observations in the unprepared human colons of healthy volunteers support the concept of regional differences in the motor function of the colon: the ascending and transverse colons are regions that temporarily store solid particles, and the descending and rectosigmoid regions function mainly as conduits.7 The development of a noninvasive, relatively inexpensive, and accurate test of colonic transit would be useful in the clinical evaluation of patients with unexplained alterations in bowel habits. We have recently shown that a similar approach using selected observations with simplified data analysis from scans taken 2,4, and 6 hours after ingestion of a radiolabeled meal provides a useful, relatively inexpensive but accurate strategy to identify motility disorders of the stomach and small bowel.’ Thus, our specific aim was to determine whether analysis of data from abdominal scintigraphy using three selected times would discriminate normal from abnormal colonic transit with an accuracy similar to that 0
1992 by the American Gastroenterological 0016~5065/92/$3.00
Association
SCINTIGRAPHIC COLONIC TRANSIT TEST
July 1992
of the more detailed, combined measurements of transit used in research studies. Some of the transit data used in this analysis were collected from other studies performed in our laboratory in healthy subjects and in patients with diarrhea or constipation.9-11 In those previous studies, we showed that combined transit summaries obtained from scintigraphic measurements were at least as accurate as the radiopaque marker method,3 which is currently the “gold standard” for detecting delayed colonic transitUgMoreover, inpatients with diarrhea-predominant irritable bowel syndrome, scintigraphy was more sensitive than the radiopaque marker method for identification of accelerated transit.” Materials and Methods Patients and Volunteers Seven patients (age range, 38-65 years; mean, 48.3 years; 1 man, 6 women) suffered idiopathic constipation with prolonged radiopaque marker transit and normal anorectal and pelvic floor function’; 9 patients (age range, 2555 years; mean, 39 years; 3 men, 7 women) had irritable bowel syndrome with predominant diarrheag; 4 patients (age range, 55-72 years; mean, 65.8 years; 3 men, 1 woman) had carcinoid syndrome, all metastatic disease associated with bronchial (n = 1) or midgut (n = 3) tumors, and diarrhea (stool volume: range, 246-519 g/day; mean, 434 g/ day); and 22 healthy volunteers (age range, 23-57 years; mean, 33.2 years; 10 men, 12 women) were recruited by public advertisements. Clinical data on patients with idiopathic constipation9 and diarrhea-predominant irritable bowel syndrome” have been published elsewhere. Patients were selected on the basis of standard clinical criteria, and no preselection bias was applied except that they were all seen at a tertiary referral center. Radioscintigraphic Camera Imaging
Markers
and
Gamma
Amberlite IR-120 plus resin pellets (size range, 0.5mm; mean, 1.0 mm) were labeled with up to 0.25 mCi of “‘InCl, (Amersham, Arlington Heights, IL) and placed in a single gelatin capsule coated with one layer of methacrylate.’ No colonic preparation was performed before transit studies. The capsule was ingested after an overnight fast, and scans were taken at regular intervals using a largefield-of-view gamma camera with a medium-energy, parallel-hole collimator (GE Starcam, Milwaukee, WI). Anterior and posterior images were acquired with the subjects erect. Standard meals were ingested for breakfast (300 kcal), lunch (540 kcal) 4 hours later, and dinner (575 kcal) 8 hours after breakfast. Scans were obtained every 10 minutes for the first 2 hours after each meal, every 15 minutes for the second 2 hours, and hourly thereafter for up to 15 hours after capsule ingestion. Further images were obtained at 24 hours in all participants and at variable other times according to presenting symptom (e.g., up to 7 days in the constipated group). The breakfast also contained 99”Tc-radiolabeled pellets in a cooked egg to measure gas1.8
37
tric and small bowel transit. The total-body radiation exposure was up to 80 millirads. Data relating to gastric and small bowel transit were not used in the current study but were reported in detail previously.g,‘o
Quantitation of Radioactivity Colonic Regions
in
Data from the 2-minute scans were stored on an on-line computer (GE Starcam) for later analysis. Previously performed barium enemas and later scintigraphic images were used to help identify colonic regions on sequential scans. Using a variable region of interest program, “‘In counts were quantitated in four colonic regions using a window at 245 (k20) keV: ascending, transverse, descending, and rectosigmoid. Counts were corrected for radionuelide decay. Regions of the colon were drawn as in previous studies.g,‘O Stools were collected separately, the time of each bowel movement was recorded, and isotopic content was quantitated in a well counter.
Data and Statistical
Analysis
Combined transit summaries. Two main parameters were used: first, the rate of emptying of the proximal colon, defined by the combined ascending and transverse regions. This summary used the data from the frequent scanning sequence to estimate the rate of emptying over 24 hours using the linear and quadratic trends in the proportion of counts in the proximal colon at 0, 4, 12, and 24 hours. These proportions were first transformed using an arcsin square root transformation. Second, the geometric center6 or weighted average of counts in the colon was used to summarize overall transit in the colon. For this analysis, the following colonic regions were designated by the numbers 1-5, respectively: ascending, transverse, descending, rectosigmoid, and stool. Thus, a low geometric center implies that most radiolabel is closer to the cecum; a high geometric center implies that most radiolabel is closer to the stool. Measurements of the proximal colonic emptying rate and the geometric center at 4, 12, and 24 hours were used to provide evidence for colonic transit disturbances in the proximal colon and left co1on.g,‘o Simple summaries. The isotopic content of the four colonic regions at 4, 12, and 24 hours was calculated. These proportion counts were transformed using an arcsin square root transformation. Statistical analysis. Data for the three groupspatients with diarrhea, patients with constipation, and healthy controls-were summarized and compared by nonparametric analysis of variance (Kruskal-Wallis test) and multiple group comparison by Dunnett’s method based on the rank-transformed data. Logistic discriminant analysis’3,‘4 was used to identify a subset of the transit summary variables that best discriminated between volunteers (n = 22) and patients (n = 20). This analysis assumes that the probability of having altered colonic transit (i.e., being a patient) can be written as a logistic function of a set of patient characteristics (e.g., age, proportionate counts in a specific region at fixed times, emptying rate from proximal colon). A forward stepwise method for the logistic dis-
38
CAMILLERI AND ZINSMEISTER
GASTROENTEROLOGY Vol. 103, No. 1
criminant analysis used a simultaneous test (at a = 0.10) to evaluate all variables not in the model at each stepsi This “residual x2”test was used to avoid adding variables to the model simply as a result of considering many possible candidates for inclusion. This analysis was conducted with and without age and sex included in the logistic models because the groups were not matched for these variables. The sensitivity and specificity of several discriminant models were estimated using cutoff scores from the logistic analysis. These were summarized by plotting the corresponding receiver operating characteristic (ROC) curves.” A similar logistic discriminant analysis was used to try to discriminate the group with diarrhea from the group with constipation. Results Transit Data The proximal colonic emptying rate was significantly accelerated (P < 0.05) in patients with diarrhea (2.4% 4 0.5% per hour, mean 3~ SEM) and retarded (P < 0.05) in patients with constipation (0.13% + 0.2% per hour) compared with healthy controls (1.4% + 0.4% per hour). The geometric center at 4 hours was significantly greater in the diarrhea group than in controls (P < 0.001) and at 24 hours was significantly smaller in the constipation group than in healthy controls (P < 0.05). These data are summarized in Table 1. Univariate comparisons of the proportion of counts in each colonic region and stool at 4, 12, and 24 hours showed that several parameters were individually different in health and disease groups (Table 2). In particular, highly significant differences were seen in the proportion of counts in ascending, transverse, and descending colon at 4 hours and counts in ascending and transverse colon and stool at 24 hours. Comparisons of diarrhea and constipation groups with healthy subjects using Dunnett’s test showed that counts in ascending and transverse colon at 4 hours and counts in stool were significantly different in both disease groups from those in healthy subjects at 24 hours. Moreover, ascending and transverse colon counts at 24 hours in the constipated group and descending colon and stool counts at 4 hours in the diarrhea group were different from those in healthy subjects.
Table
1. Geometric
Center of Colonic Isotope Content 4 Hour
Healthy subjects Constipation Diarrhea
1.14 f 0.07 1.03 + 0.21" 2.30 f 0.34"
NOTE. Data represent means + SEM. “P < 0.05 vs. healthy subjects. bP < 0.05 vs. healthy subjects.
12 Hour 1.87 + 0.23 1.53 * 0.16 2.36 + 0.35
24 Hour 2.83 + 0.25 1.77 IL0.17b 3.68 e 0.34
Table 2. Summary of Univariate Comparisons Between Healthy Controls and Two disease Groups Variable
Kruskal-Wallis
Age AC,4h AC, 12 h AC, 24 h TC, 4 h TC, 12 h TC, 24 h DC, 4 h DC, 12 h DC, 24 h RS, 4 h RS, 12 h RS, 24 h Stool, 4 h Stool, 12 h Stool, 24 h GC, 4 h GC, 12 h GC, 24 h Proximal colonic emptying rate
test
Dunnett’s test’
P < 0.005 P < 0.001 NS P < 0.01 P < 0.0001 NS P < 0.05 P < 0.005 NS NS NS NS NS P < 0.05 NS P < 0.001 P < 0.005 NS P < 0.005
D only C and D C only C and D C only D only
D only CandD C and D C only
P < 0.01
C only
AC, ascending colon; TC, transverse colon; DC, descending colon; RS, rectosigmoid; GC, geometric center; C, constipation group; D, diarrhea group. “Dunnett’s test vs. healthy controls at a = 0.05.
Table 3 shows the unadjusted (for age and sex) x2 statistic for discriminating between the two disease groups. Significant values are obtained for the proximal colonic emptying rate, geometric center at 4 and 24 hours, and proportion of counts in the ascending and transverse colon at 4 and 24 hours, although the residual x2 test did not show significance (P > 0.20). Estimation of Sensitivity of Transit Parameters
and Specificity
By analysis of variance, we detected significant difference in age between the three groups studied, and in the logistic discriminant analysis age was
Table 3. Summary of Univariate Comparisons Between Constipation
and Diarrhea
Variable
f
AC, 4 h AC, 12 h AC, 24 h TC, 4 h TC, 12 h TC, 24 h GC,4h GC, 12 h GC, 24 h Proximal colonic emptying rate AC, ascending
Groups
colon; TC, transverse
6.12 3.67 6.93 4.18 0.34 6.22 4.32 2.61 9.23 9.60
P 0.01 0.06 to.01 0.04 0.56 0.01 0.04 0.11
colon; GC, geometric center.
SCINTIGRAPHIC
July 1992
significant variable in discriminating between patients and healthy subjects. Thus, in summarizing the sensitivity and specificity of the transit parameters, age and sex were incorporated in the logistic models. For example, we first constructed the ROC curves for age and sex and for age, sex, and proportion of counts in the transverse colon at 4 hours. Figure I shows the ROC curves in the three groups for age and sex and for age, sex, and proportion of counts in the transverse colon at 4 hours. Although age was identified as a useful discriminator, these curves clearly show the lack of specificity of age and sex alone. Thus, at 90% sensitivity the specificity for age and sex alone was &I%, whereas the addition of proportion of transverse colon counts at 4 hours increased the specificity to 80%. Figure 2 shows the ROC curves for all patients vs. healthy controls summarizing the sensitivity and specificity of proximal colonic emptying rate over 24 hours and the proportion of counts in the transverse colon at 4 hours in the three groups. The close approximation of the two curves reflects their similar accuracies; thus, at 90% sensitivity each has a specificity of 79%, and at 95% sensitivity the specificity is matched at 72%. Similar summaries for patients vs. controls compared the combination of geometric centers at 4, 12, and 24 hours and the proportion of counts in the transverse colon at 4 hours (Figure 3), both models incorporating age and sex. The sensitivity and specificity to discriminate between the diarrhea and constipation groups were greater for the more detailed proximal colonic emptying rate than for the proportion of counts in transverse colon at 4 hours (Figure 4); however, these differences were not significant. The later indices of transit, such as geometric
1.0 -
a
09
I
-
E >
0.6
-
:s
0.5
-
*s
0.4
-
0.3
-
0.2
-
-
0.1
-
-Proximal
-
.z .z?
0.6
-
.%
0.5
-
s v)
0.4
-
0.3
-
-
-
Health vs. 2 disease
0
0.2
-
0.1
-
-Age. -
I
o0
0.1
groups
groups
TC, colon emptying
rate
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
l-specificity Figure 2. ROC curve comparing the sensitivity and specificity of the proximal colonic emptying rate and counts in the transverse colon at 4 hours, both adjusted for age and sex, in healthy subjects and the two disease groups combined. Note that at 90% sensitivity, the specificity is identical at 79%.
center at 24 hours (which is significantly different in the constipation group; Table 11, showed poor accuracy similar to that of the combination of proportion of counts in ascending and transverse colon at 24 hours (Figure 5). The apparent loss of ability to discriminate between patients and healthy subjects in the &hour data is discussed below. Discussion This study shows that after ingestion of a delayed-release capsule containing radiolabeled pellets, scans taken at 4 and 24 hours provide a high degree of sensitivity for identifying disordered coionic transit. The specificities provided by the results
09
Health vs. 2 disease
39
_
0.9 0.7
TEST
I
0.7
r
1.0 -
-
TRANSIT
-
0.6
1.0 -
0.6
COLONIC
1
-
0.6
-
0.7
-
.g .g
0.6
-
.-z!
0.5
-
$j
0.4
-
0.3
-
0.2
-
0.1
-
Health vs. 2 disease
sex. TC4
-
TC4
-
Combination
0.5
0.6
groups
GC,,
GC,a. GC,,
Age. sex
I
I
I
I
I
I
I
I
I
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
o0
0.1
0.2
0.3
0.4
0.7
0.6
0.9
1.0
l-specificity
l-specificity Figure 1. ROC curve comparing the sensitivity and specificity of age and sex with and without proportion of counts in the transverse colon at 4 hours in the two disease groups combined and in healthy subjects. Note the more rapid reduction in specificity with increasing sensitivity for age and sex alone.
Figure 3. ROC curve showing that in healthy subjects and the two disease groups combined the sensitivity and specificity of the linear combination of geometric center of colonic counts at 4, 12, and 24 hours with counts in the transverse colon at 4 hours were almost identical. Both transit summaries were adjusted for age and sex.
40 CAMILLERI AND ZINSMEISTER
0’0
’
0.1
GASTROENTEROLOGY
I
I
I
I
I
I
I
I
I
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
l-specificity Figure 4. ROC curve showing a comparison in the two patient groups only of proximal colonic emptying rate and proportion of counts in the transverse colon at 4 hours. Although the former summary is clearly more accurate, the latter is also of significant discriminant value (see Table 3).
from these observations are comparable with those of combined analyses, such as the rate of emptying of the proximal (ascending and transverse) colon, which require frequent scanning with attendant increments in camera use, technician time, and hence overall costs. Our data are particularly relevant because the proposed scanning sequence is applicable to conditions associated with either accelerated or delayed colonic transit. In this regard, our method is unique among noninvasive methods; in contrast, the widely used radiopaque marker method performed with a single radiograph taken on the fourth day3 provides a clinically useful investigation for patients with delayed colonic transit but has never been validated for rapid colonic transit. In the evaluation of rapid transit in the irritable bowel syndrome, the radiopaque marker method3 with a single radiograph on the fourth day is less sensitive than the scintigraphic delayed-release capsule method.” Because radiopaque markers may already have traversed one or more regions of the colon by the fourth day, when the radiograph is taken, the latter can only provide a censored estimate of transit time. Clearly, increasing the frequency of abdominal radiographs would increase the sensitivity of the radiopaque marker method for evaluation of rapid colonic transit. However, that could only be achieved with an increment in the radiation exposure. The scintigraphic delayed-release capsule method using 0.25 mCi of ‘1’InC13 results in the same total-body radiation exposure as a single abdominal radiograph (80 millirads). In fact, the same accuracy in counting isotopic content in the different colonic regions in humans could be achieved using as little as 0.1 mCi while the gamma camera acquisition time is increased to 5 minutes instead of the current 2
Vol.
103,No.1
minutes. Such a strategy would not appreciably alter the cost of the test but would be advantageous by reducing the radiation dose to which internal organs are exposed and would make the scintigraphic method more acceptable in adolescents and young adults. We have also calculated that the organ dose per exposure in adolescents would be identical to that given to adults if the dose of isotope used was prorated according to body weight, assuming that the adult dose (e.g., 0.25 mCi) is administered to a X)-kg adult (Table 4). The observation of the high discriminant value of ascending and transverse colon counts at 4 and 24 hours in differentiating healthy subjects from those in disease groups confirms the importance of the proximal colon in the overall motor function of the colon, as suggested in our previous studies.7~g~‘0Initial analysis of our data indicates statistically significant differences in the ages of patients in the diarrhea group and healthy controls; however, this effect is coincidental and of little influence on the specificity of transit data, as shown by Figure 1. Moreover, because patients with the irritable bowel syndrome have an age range similar to that of controls, we believe that the observed effect of age was attributable to the fact that carcinoid diarrhea patients usually present in the sixth to eighth decades of life, as in our study group. The costs involved in this colonic transit test are comparable to those of a scintigraphic gastric-emptying test if the number of scans taken is restricted. Our data suggest that a limited number of scans using simple or combined summaries of transit provides an accurate estimate of transit that is applicable for a wide spectrum of colonic motility disorders. 1.0 0.9 0.8 0.7 .z .z?
0.6
‘ji
0.5
*5
0.4 0.3
Health vs. 2 disease
groups
-=24
0.2
-
0.1
Combination
AC24, T&4
v-
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
l-specificity Figure 5. ROC curve showing that in healthy subjects and the two disease groups combined there is similar “poor” sensitivity and specificity of the geometric center at 24 hours and of the linear combination of the proportion of counts in the ascending and transverse colon at 24 hours. Note the apparent lack of sensitivity of the Z&hour data in discriminating the healthy subjects from those in the two disease groups.
July 1992
SCINTIGRAPHICCOLONIC TRANSIT TEST
Table 4. Radiation Exposure
Participant
Adult, 70 kg Adolescent, 50 kg H,, Effective
ofParticipants
41
[Adults and Adolescents)
Isotope form
Dose (mCi)
UPPer
LCMer
Total
Bone
Stomach
Small
large
large
Ovary
Testis
body
marrow
H,
“‘In pellets
0.25 0.18
0.24 0.13
0.46 0.36
1.10 0.88
1.90 1.69
0.40 0.32
0.03 0.03
0.06 0.04
0.09 0.05
0.36 0.29
“‘In pellets
dose equivalent.
Although our scintigraphic colonic transit test is more costly than the radiopaque marker method, it appears to be advantageous in the assessment of accelerated colonic transit and provides the opportunity for more detailed assessment of transit without increasing radiation exposure. Development of this relatively inexpensive but accurate scintigraphic method for detecting accelerated or delayed colonic transit has important practical implications in gastroenterology clinics, which continue to be attended by large numbers of patients with functional gastrointestinal disease.16*17 Accurate questionnaire methods” allow categorization of patients into different subsets, e.g., diarrhea-predominant and constipation-predominant, but by their very nature are subjective assessments and on their own cannot be used to determine objectively the severity of illness and hence stratify patients for treatment. After the exclusion of mucosal disease or strictures by contrast colon x-ray or colonoscopy and of endocrine or metabolic problems such as hypothyroidism, our noninvasive method would provide a useful investigation to determine the severity of the transit disturbance. Such objective measurements of accelerated or delayed transit may also facilitate the choice of therapy. Thus, constipated patients with only modest prolongation of colonic transit may be given a fiber supplement and an osmotic laxative; in contrast, more significant prolongation in transit may warrant addition of a prokinetic agent or, in severe intractable cases, consideration of subtotal colectomy if dynamic studies of defecation are normal.” The relatively few scans proposed by the current method might preclude an accurate determination of the region(s) of abnormal transit in the colon. However, until medications or operative approaches that selectively correct regional abnormalities are developed. this deficiency is of little practical importance. The logistic discriminant analysis relies on large sample approximations to identify and estimate the model parameters and make inferences of their statistical significance. These results should be updated with additional subjects (controls and patients) spanning the spectrum of colonic transit.
The observations in these studies thus need to be confirmed in a larger group of patients, particularly in those with less severe colonic motility disorders than were observed in our patients with idiopathic constipation and carcinoid diarrhea. The data-analysis models developed in this preliminary work are being evaluated prospectively in a larger number of patients with functional and motility disorders of the gastrointestinal tract. The apparently greater accuracy of early (J-hour; Figures 2 and 3) than of later (&hour; Figure 5) transit parameters in this study probably reflects three facts: [a) marked difference in transit was seen in the diarrhea group, particularly the patients with carcinoid diarrhea, which weighted the J-hour data; (b) the data of the two disease groups are at the opposite ends of the transit spectrum relative to health, so that when combined as a single disease group for the ROC curves they tend to neutralize each other (Figure 5); and (c) we took an arbitrary cut-off of 24 hours for the “late” scan. Indeed, we anticipated that the %-hour data would have less discriminant value in the detection of the constipated group in view of the relatively prolonged residence time of solid residue in the proximal colon in health.7 Nevertheless, the .%-hour data have clear discriminant value to separate each of the two disease groups from healthy subjects (Table 1).It is conceivable that later scans, such as a @-hour image, may be necessary to enhance the accuracy of the scanning protocol in less severe motility disorders than those of the groups studied here. However, this increase in the number of later scans can be achieved with no further radiation exposure and minimal increments in the cost of the study. In conclusion, this study confirms the accuracy of our noninvasive, relatively inexpensive strategy using the delayed-release capsule method. Quantitation of isotope in the four colonic regions at 4 and 24 hours accurately detects accelerated or delayed transit in motility disorders of the colon. References 1. Hinton
studying
JM. Lennard-Jones JE, Young AC. A new method for gut transit times using radiopaque markers. Gut
1969;10:842-847.
42
CAMILLERI AND ZINSMEISTER
2. Arhan P, Devroede G, Jehannu B, Lanza M, Faverdin C, Dornit C, Persoz B, Titreault L, Perey B, Pellerin B. Segmental colonic transit time. Dis Colon Rectum 1981;24:625-629. 3. Metcalf AM, Phillips SF, Zinsmeister AR, MacCarty RL, Beart RW, Wolff BG. Simplified assessment of segmental colonic transit. Gastroenterology 1987;92:40-47. 4. Chaussade S, Khyari A, Roche H, Garret M, Gaudric M, Couturier D, Guerre J. Determination of total and segmental coionic transit time in constipated patients. Dig Dis Sci 1989;34:1169-1172. 5. Krevsky B, Malmud LS, D’Ercole F, Maurer AH, Fisher RS. Colonic transit scintigraphy. A physiologic approach to the quantitative measurement of colonic transit in humans. Gastroenterology 1986;91:1102-1112. 6. Kamm MA, Lennard-Jones JE, Thompson DG, Sobnack R, Garvie NW, Granowska M. Dynamic scanning defines a coionic defect in severe idiopathic constipation, Gut 1988;29: 1085-1092. 7. Proano M, Camilleri M, Phillips SF, Brown ML, Thomforde GM. Transit of solids through the human colon: regional quantification in the unprepared bowel. Am J Physiol 1990;258:G856-G862. 8. Camilleri M, Zinsmeister AR, Greydanus MP, Brown ML, Proano M. Towards a less costly but accurate test of gastric emptying and small bowel transit. Dig Dis Sci 1991;36:609615. 9. Stivland T, Camilleri M, Vassallo M, Proano M, Rath D, Brown
M, Thomforde G, Pemberton J, Phillips S. Scintigraphic measurement of regional gut transit in idiopathic constipation. Gastroenterology 1991;101:107-115. 10. Vassallo M, Camilleri M, Phillips SF, Brown ML, Chapman NJ, Thomforde GM. Transit through the proximal colon influences stool weight in irritable bowel syndrome. Gastroenterology 1992;102:102-108. 11. Camilleri M, Thomforde GM, Vassallo MJ, Kvols LK, Powers SP. Gastrointestinal transit, colonic capacitance and possible mediators in carcinoid diarrhea. Gut 1991;32:A1215.
GASTROENTEROLOGY Vol. 103, No. 1
12. Vassallo M, Camilleri M, Phillips SF, Chapman NJ, Thomforde GM. Colonic transit in diarrhea-predominant irritable bowel syndrome BBS): comparison by scintigraphy and radiopaque markers (abstr). Am J Gastroenterol 1990;85:1286. 13. SAS users guide: statistics. Cary, NC: SAS Institute, 1982:1537, 297-308. 14. SUGI supplemental library user’s guide. Cary, NC: SAS Institute, 1986:269-293. 15. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8:283-298. 16. Switz DM. What the gastroenterologist does all day. Gastroenterology 1976;70:1048-1050. 17. Lennard-Jones JE. Functional gastrointestinal disorders. N Engl J Med 1983;308:431-435. 18. Talley NJ, Phillips SF, Wiltgen CM, Zinsmeister AR, Melton LJ III. Assessment of functional gastrointestinal disease: the bowel disease questionnaire. Mayo Clin Proc 1990;65:14561479. 19. Pemberton JH, Rath DM, Ilstrup DM. Evaluation, management strategies and outcome of patients with severe chronic constipation (abstr). Gastroenterology lOO:A481, 1991.
Received May 31, 1991. Accepted December 10, 1991. Address requests for reprints to: Michael Camilleri, M.D., Gastroenterology Research Unit, Mayo Clinic, Rochester, Minnesota 55905. Supported in part by Digestive Disease Center grant DK34988 and by the General Clinical Research Center grant RR00585 from the National Institutes of Health. Reported in part at the British Society of Gastroenterology Meeting, London, England, September 1991, and published in abstract form Gut 1991;32:A1248. The authors thank Drs. M. L. Brown, L. A. Forstrom, M. Proano, T. A. Stivland, and M. J. Vassallo and G. M. Thomforde for technical support and Cindy Stanislav for typing and preparing the manuscript.
1992;103:36-42
Towards a Relatively Inexpensive, Noninvasive, Accurate Test for Colonic Motility Disorders MICHAEL Gastroenterology Minnesota
CAMILLERI
and ALAN R. ZINSMEISTER
Research Unit and Section of Biostatistics,
Currently available clinical tests of colonic transit, such as the radiopaque marker method, are useful to detect delayed transit but may be less sensitive for rapid transit. The aim of this study was to develop a relatively inexpensive, noninvasive, accurate test of colonic transit using selected scintigraphic observations within the first 24 hours after ingestion of a pH-sensitive, methacrylate-coated, delayed-release capsule containing “‘In-labeled resin pellets. The authors’ previously published coionic transit data on 22 healthy subjects, 9 patients with diarrhea-predominant irritable bowel syndrome, and 7 patients with idiopathic constipation and previously unpublished data on 4 patients with carcinoid diarrhea were analyzed. A logistic discriminant analysis was used to estimate the sensitivity and specificity of selected combinations or simple summaries of transit. Among combined transit summaries, the emptying rate of the proximal colon was significantly different between healthy and constipation groups; the geometric center of isotope in the colon at 4 hours was significantly greater in the diarrhea group than in healthy controls; the geometric center at 24 hours was significantly lower in the constipation group than in the other two groups. From the logistic discriminant analysis, simple summaries of transit also had significant discriminant value; these included the isotopic contents in the ascending, transverse, and descending colon at 4 hours and the counts in the ascending and transverse colon and stool at 24 hours. At 90% sensitivity, the specificity of the transverse colon counts at 4 hours was 79%, which is identical to the specificity of the proximal colon emptying rate, both adjusted for age. Thus, quantitation of isotopic counts in colonic regions on scans taken at 4 and 24 hours provides an accurate summary of colonic transit, with acceptable specificity at a high sensitivity in the detection of motility disorders of the colon. isturbances of colonic transit are commonly encountered in clinical practice and are generally confirmed by means of one of the radiopaque marker
D
Mayo Clinic and Mayo Foundation,
Rochester,
methods.1-3 These are robust, clinically applicable procedures that have gained wide acceptance because they are inexpensive and require little special expertise or equipment. They are particularly useful in the study of constipated patients.4 However, as the marker technique has been adapted to restrict radiation exposure by taking only one abdominal radiograph on the fourth day,3 its sensitivity for detecting rapid transit is reduced because the markers have usually traversed the colonic regions when the radiograph is obtained. Hence, this method may provide only a censored estimate of the actual colonic transit time. Alternative strategies using orocecal intubation and dynamic scintigraphy5v6 to measure colonic transit are applicable in the research arena. To overcome the need for orocecal intubation and to facilitate the study of regional colonic transit of solids, we recently developed a new scintigraphic method that uses a pH-sensitive, methacrylate-coated capsule containing radiolabeled resin pellets.’ Our observations in the unprepared human colons of healthy volunteers support the concept of regional differences in the motor function of the colon: the ascending and transverse colons are regions that temporarily store solid particles, and the descending and rectosigmoid regions function mainly as conduits.7 The development of a noninvasive, relatively inexpensive, and accurate test of colonic transit would be useful in the clinical evaluation of patients with unexplained alterations in bowel habits. We have recently shown that a similar approach using selected observations with simplified data analysis from scans taken 2,4, and 6 hours after ingestion of a radiolabeled meal provides a useful, relatively inexpensive but accurate strategy to identify motility disorders of the stomach and small bowel.’ Thus, our specific aim was to determine whether analysis of data from abdominal scintigraphy using three selected times would discriminate normal from abnormal colonic transit with an accuracy similar to that 0
1992 by the American Gastroenterological 0016~5065/92/$3.00
Association
SCINTIGRAPHIC COLONIC TRANSIT TEST
July 1992
of the more detailed, combined measurements of transit used in research studies. Some of the transit data used in this analysis were collected from other studies performed in our laboratory in healthy subjects and in patients with diarrhea or constipation.9-11 In those previous studies, we showed that combined transit summaries obtained from scintigraphic measurements were at least as accurate as the radiopaque marker method,3 which is currently the “gold standard” for detecting delayed colonic transitUgMoreover, inpatients with diarrhea-predominant irritable bowel syndrome, scintigraphy was more sensitive than the radiopaque marker method for identification of accelerated transit.” Materials and Methods Patients and Volunteers Seven patients (age range, 38-65 years; mean, 48.3 years; 1 man, 6 women) suffered idiopathic constipation with prolonged radiopaque marker transit and normal anorectal and pelvic floor function’; 9 patients (age range, 2555 years; mean, 39 years; 3 men, 7 women) had irritable bowel syndrome with predominant diarrheag; 4 patients (age range, 55-72 years; mean, 65.8 years; 3 men, 1 woman) had carcinoid syndrome, all metastatic disease associated with bronchial (n = 1) or midgut (n = 3) tumors, and diarrhea (stool volume: range, 246-519 g/day; mean, 434 g/ day); and 22 healthy volunteers (age range, 23-57 years; mean, 33.2 years; 10 men, 12 women) were recruited by public advertisements. Clinical data on patients with idiopathic constipation9 and diarrhea-predominant irritable bowel syndrome” have been published elsewhere. Patients were selected on the basis of standard clinical criteria, and no preselection bias was applied except that they were all seen at a tertiary referral center. Radioscintigraphic Camera Imaging
Markers
and
Gamma
Amberlite IR-120 plus resin pellets (size range, 0.5mm; mean, 1.0 mm) were labeled with up to 0.25 mCi of “‘InCl, (Amersham, Arlington Heights, IL) and placed in a single gelatin capsule coated with one layer of methacrylate.’ No colonic preparation was performed before transit studies. The capsule was ingested after an overnight fast, and scans were taken at regular intervals using a largefield-of-view gamma camera with a medium-energy, parallel-hole collimator (GE Starcam, Milwaukee, WI). Anterior and posterior images were acquired with the subjects erect. Standard meals were ingested for breakfast (300 kcal), lunch (540 kcal) 4 hours later, and dinner (575 kcal) 8 hours after breakfast. Scans were obtained every 10 minutes for the first 2 hours after each meal, every 15 minutes for the second 2 hours, and hourly thereafter for up to 15 hours after capsule ingestion. Further images were obtained at 24 hours in all participants and at variable other times according to presenting symptom (e.g., up to 7 days in the constipated group). The breakfast also contained 99”Tc-radiolabeled pellets in a cooked egg to measure gas1.8
37
tric and small bowel transit. The total-body radiation exposure was up to 80 millirads. Data relating to gastric and small bowel transit were not used in the current study but were reported in detail previously.g,‘o
Quantitation of Radioactivity Colonic Regions
in
Data from the 2-minute scans were stored on an on-line computer (GE Starcam) for later analysis. Previously performed barium enemas and later scintigraphic images were used to help identify colonic regions on sequential scans. Using a variable region of interest program, “‘In counts were quantitated in four colonic regions using a window at 245 (k20) keV: ascending, transverse, descending, and rectosigmoid. Counts were corrected for radionuelide decay. Regions of the colon were drawn as in previous studies.g,‘O Stools were collected separately, the time of each bowel movement was recorded, and isotopic content was quantitated in a well counter.
Data and Statistical
Analysis
Combined transit summaries. Two main parameters were used: first, the rate of emptying of the proximal colon, defined by the combined ascending and transverse regions. This summary used the data from the frequent scanning sequence to estimate the rate of emptying over 24 hours using the linear and quadratic trends in the proportion of counts in the proximal colon at 0, 4, 12, and 24 hours. These proportions were first transformed using an arcsin square root transformation. Second, the geometric center6 or weighted average of counts in the colon was used to summarize overall transit in the colon. For this analysis, the following colonic regions were designated by the numbers 1-5, respectively: ascending, transverse, descending, rectosigmoid, and stool. Thus, a low geometric center implies that most radiolabel is closer to the cecum; a high geometric center implies that most radiolabel is closer to the stool. Measurements of the proximal colonic emptying rate and the geometric center at 4, 12, and 24 hours were used to provide evidence for colonic transit disturbances in the proximal colon and left co1on.g,‘o Simple summaries. The isotopic content of the four colonic regions at 4, 12, and 24 hours was calculated. These proportion counts were transformed using an arcsin square root transformation. Statistical analysis. Data for the three groupspatients with diarrhea, patients with constipation, and healthy controls-were summarized and compared by nonparametric analysis of variance (Kruskal-Wallis test) and multiple group comparison by Dunnett’s method based on the rank-transformed data. Logistic discriminant analysis’3,‘4 was used to identify a subset of the transit summary variables that best discriminated between volunteers (n = 22) and patients (n = 20). This analysis assumes that the probability of having altered colonic transit (i.e., being a patient) can be written as a logistic function of a set of patient characteristics (e.g., age, proportionate counts in a specific region at fixed times, emptying rate from proximal colon). A forward stepwise method for the logistic dis-
38
CAMILLERI AND ZINSMEISTER
GASTROENTEROLOGY Vol. 103, No. 1
criminant analysis used a simultaneous test (at a = 0.10) to evaluate all variables not in the model at each stepsi This “residual x2”test was used to avoid adding variables to the model simply as a result of considering many possible candidates for inclusion. This analysis was conducted with and without age and sex included in the logistic models because the groups were not matched for these variables. The sensitivity and specificity of several discriminant models were estimated using cutoff scores from the logistic analysis. These were summarized by plotting the corresponding receiver operating characteristic (ROC) curves.” A similar logistic discriminant analysis was used to try to discriminate the group with diarrhea from the group with constipation. Results Transit Data The proximal colonic emptying rate was significantly accelerated (P < 0.05) in patients with diarrhea (2.4% 4 0.5% per hour, mean 3~ SEM) and retarded (P < 0.05) in patients with constipation (0.13% + 0.2% per hour) compared with healthy controls (1.4% + 0.4% per hour). The geometric center at 4 hours was significantly greater in the diarrhea group than in controls (P < 0.001) and at 24 hours was significantly smaller in the constipation group than in healthy controls (P < 0.05). These data are summarized in Table 1. Univariate comparisons of the proportion of counts in each colonic region and stool at 4, 12, and 24 hours showed that several parameters were individually different in health and disease groups (Table 2). In particular, highly significant differences were seen in the proportion of counts in ascending, transverse, and descending colon at 4 hours and counts in ascending and transverse colon and stool at 24 hours. Comparisons of diarrhea and constipation groups with healthy subjects using Dunnett’s test showed that counts in ascending and transverse colon at 4 hours and counts in stool were significantly different in both disease groups from those in healthy subjects at 24 hours. Moreover, ascending and transverse colon counts at 24 hours in the constipated group and descending colon and stool counts at 4 hours in the diarrhea group were different from those in healthy subjects.
Table
1. Geometric
Center of Colonic Isotope Content 4 Hour
Healthy subjects Constipation Diarrhea
1.14 f 0.07 1.03 + 0.21" 2.30 f 0.34"
NOTE. Data represent means + SEM. “P < 0.05 vs. healthy subjects. bP < 0.05 vs. healthy subjects.
12 Hour 1.87 + 0.23 1.53 * 0.16 2.36 + 0.35
24 Hour 2.83 + 0.25 1.77 IL0.17b 3.68 e 0.34
Table 2. Summary of Univariate Comparisons Between Healthy Controls and Two disease Groups Variable
Kruskal-Wallis
Age AC,4h AC, 12 h AC, 24 h TC, 4 h TC, 12 h TC, 24 h DC, 4 h DC, 12 h DC, 24 h RS, 4 h RS, 12 h RS, 24 h Stool, 4 h Stool, 12 h Stool, 24 h GC, 4 h GC, 12 h GC, 24 h Proximal colonic emptying rate
test
Dunnett’s test’
P < 0.005 P < 0.001 NS P < 0.01 P < 0.0001 NS P < 0.05 P < 0.005 NS NS NS NS NS P < 0.05 NS P < 0.001 P < 0.005 NS P < 0.005
D only C and D C only C and D C only D only
D only CandD C and D C only
P < 0.01
C only
AC, ascending colon; TC, transverse colon; DC, descending colon; RS, rectosigmoid; GC, geometric center; C, constipation group; D, diarrhea group. “Dunnett’s test vs. healthy controls at a = 0.05.
Table 3 shows the unadjusted (for age and sex) x2 statistic for discriminating between the two disease groups. Significant values are obtained for the proximal colonic emptying rate, geometric center at 4 and 24 hours, and proportion of counts in the ascending and transverse colon at 4 and 24 hours, although the residual x2 test did not show significance (P > 0.20). Estimation of Sensitivity of Transit Parameters
and Specificity
By analysis of variance, we detected significant difference in age between the three groups studied, and in the logistic discriminant analysis age was
Table 3. Summary of Univariate Comparisons Between Constipation
and Diarrhea
Variable
f
AC, 4 h AC, 12 h AC, 24 h TC, 4 h TC, 12 h TC, 24 h GC,4h GC, 12 h GC, 24 h Proximal colonic emptying rate AC, ascending
Groups
colon; TC, transverse
6.12 3.67 6.93 4.18 0.34 6.22 4.32 2.61 9.23 9.60
P 0.01 0.06 to.01 0.04 0.56 0.01 0.04 0.11
colon; GC, geometric center.
SCINTIGRAPHIC
July 1992
significant variable in discriminating between patients and healthy subjects. Thus, in summarizing the sensitivity and specificity of the transit parameters, age and sex were incorporated in the logistic models. For example, we first constructed the ROC curves for age and sex and for age, sex, and proportion of counts in the transverse colon at 4 hours. Figure I shows the ROC curves in the three groups for age and sex and for age, sex, and proportion of counts in the transverse colon at 4 hours. Although age was identified as a useful discriminator, these curves clearly show the lack of specificity of age and sex alone. Thus, at 90% sensitivity the specificity for age and sex alone was &I%, whereas the addition of proportion of transverse colon counts at 4 hours increased the specificity to 80%. Figure 2 shows the ROC curves for all patients vs. healthy controls summarizing the sensitivity and specificity of proximal colonic emptying rate over 24 hours and the proportion of counts in the transverse colon at 4 hours in the three groups. The close approximation of the two curves reflects their similar accuracies; thus, at 90% sensitivity each has a specificity of 79%, and at 95% sensitivity the specificity is matched at 72%. Similar summaries for patients vs. controls compared the combination of geometric centers at 4, 12, and 24 hours and the proportion of counts in the transverse colon at 4 hours (Figure 3), both models incorporating age and sex. The sensitivity and specificity to discriminate between the diarrhea and constipation groups were greater for the more detailed proximal colonic emptying rate than for the proportion of counts in transverse colon at 4 hours (Figure 4); however, these differences were not significant. The later indices of transit, such as geometric
1.0 -
a
09
I
-
E >
0.6
-
:s
0.5
-
*s
0.4
-
0.3
-
0.2
-
-
0.1
-
-Proximal
-
.z .z?
0.6
-
.%
0.5
-
s v)
0.4
-
0.3
-
-
-
Health vs. 2 disease
0
0.2
-
0.1
-
-Age. -
I
o0
0.1
groups
groups
TC, colon emptying
rate
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
l-specificity Figure 2. ROC curve comparing the sensitivity and specificity of the proximal colonic emptying rate and counts in the transverse colon at 4 hours, both adjusted for age and sex, in healthy subjects and the two disease groups combined. Note that at 90% sensitivity, the specificity is identical at 79%.
center at 24 hours (which is significantly different in the constipation group; Table 11, showed poor accuracy similar to that of the combination of proportion of counts in ascending and transverse colon at 24 hours (Figure 5). The apparent loss of ability to discriminate between patients and healthy subjects in the &hour data is discussed below. Discussion This study shows that after ingestion of a delayed-release capsule containing radiolabeled pellets, scans taken at 4 and 24 hours provide a high degree of sensitivity for identifying disordered coionic transit. The specificities provided by the results
09
Health vs. 2 disease
39
_
0.9 0.7
TEST
I
0.7
r
1.0 -
-
TRANSIT
-
0.6
1.0 -
0.6
COLONIC
1
-
0.6
-
0.7
-
.g .g
0.6
-
.-z!
0.5
-
$j
0.4
-
0.3
-
0.2
-
0.1
-
Health vs. 2 disease
sex. TC4
-
TC4
-
Combination
0.5
0.6
groups
GC,,
GC,a. GC,,
Age. sex
I
I
I
I
I
I
I
I
I
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
o0
0.1
0.2
0.3
0.4
0.7
0.6
0.9
1.0
l-specificity
l-specificity Figure 1. ROC curve comparing the sensitivity and specificity of age and sex with and without proportion of counts in the transverse colon at 4 hours in the two disease groups combined and in healthy subjects. Note the more rapid reduction in specificity with increasing sensitivity for age and sex alone.
Figure 3. ROC curve showing that in healthy subjects and the two disease groups combined the sensitivity and specificity of the linear combination of geometric center of colonic counts at 4, 12, and 24 hours with counts in the transverse colon at 4 hours were almost identical. Both transit summaries were adjusted for age and sex.
40 CAMILLERI AND ZINSMEISTER
0’0
’
0.1
GASTROENTEROLOGY
I
I
I
I
I
I
I
I
I
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
l-specificity Figure 4. ROC curve showing a comparison in the two patient groups only of proximal colonic emptying rate and proportion of counts in the transverse colon at 4 hours. Although the former summary is clearly more accurate, the latter is also of significant discriminant value (see Table 3).
from these observations are comparable with those of combined analyses, such as the rate of emptying of the proximal (ascending and transverse) colon, which require frequent scanning with attendant increments in camera use, technician time, and hence overall costs. Our data are particularly relevant because the proposed scanning sequence is applicable to conditions associated with either accelerated or delayed colonic transit. In this regard, our method is unique among noninvasive methods; in contrast, the widely used radiopaque marker method performed with a single radiograph taken on the fourth day3 provides a clinically useful investigation for patients with delayed colonic transit but has never been validated for rapid colonic transit. In the evaluation of rapid transit in the irritable bowel syndrome, the radiopaque marker method3 with a single radiograph on the fourth day is less sensitive than the scintigraphic delayed-release capsule method.” Because radiopaque markers may already have traversed one or more regions of the colon by the fourth day, when the radiograph is taken, the latter can only provide a censored estimate of transit time. Clearly, increasing the frequency of abdominal radiographs would increase the sensitivity of the radiopaque marker method for evaluation of rapid colonic transit. However, that could only be achieved with an increment in the radiation exposure. The scintigraphic delayed-release capsule method using 0.25 mCi of ‘1’InC13 results in the same total-body radiation exposure as a single abdominal radiograph (80 millirads). In fact, the same accuracy in counting isotopic content in the different colonic regions in humans could be achieved using as little as 0.1 mCi while the gamma camera acquisition time is increased to 5 minutes instead of the current 2
Vol.
103,No.1
minutes. Such a strategy would not appreciably alter the cost of the test but would be advantageous by reducing the radiation dose to which internal organs are exposed and would make the scintigraphic method more acceptable in adolescents and young adults. We have also calculated that the organ dose per exposure in adolescents would be identical to that given to adults if the dose of isotope used was prorated according to body weight, assuming that the adult dose (e.g., 0.25 mCi) is administered to a X)-kg adult (Table 4). The observation of the high discriminant value of ascending and transverse colon counts at 4 and 24 hours in differentiating healthy subjects from those in disease groups confirms the importance of the proximal colon in the overall motor function of the colon, as suggested in our previous studies.7~g~‘0Initial analysis of our data indicates statistically significant differences in the ages of patients in the diarrhea group and healthy controls; however, this effect is coincidental and of little influence on the specificity of transit data, as shown by Figure 1. Moreover, because patients with the irritable bowel syndrome have an age range similar to that of controls, we believe that the observed effect of age was attributable to the fact that carcinoid diarrhea patients usually present in the sixth to eighth decades of life, as in our study group. The costs involved in this colonic transit test are comparable to those of a scintigraphic gastric-emptying test if the number of scans taken is restricted. Our data suggest that a limited number of scans using simple or combined summaries of transit provides an accurate estimate of transit that is applicable for a wide spectrum of colonic motility disorders. 1.0 0.9 0.8 0.7 .z .z?
0.6
‘ji
0.5
*5
0.4 0.3
Health vs. 2 disease
groups
-=24
0.2
-
0.1
Combination
AC24, T&4
v-
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.6
0.9
1.0
l-specificity Figure 5. ROC curve showing that in healthy subjects and the two disease groups combined there is similar “poor” sensitivity and specificity of the geometric center at 24 hours and of the linear combination of the proportion of counts in the ascending and transverse colon at 24 hours. Note the apparent lack of sensitivity of the Z&hour data in discriminating the healthy subjects from those in the two disease groups.
July 1992
SCINTIGRAPHICCOLONIC TRANSIT TEST
Table 4. Radiation Exposure
Participant
Adult, 70 kg Adolescent, 50 kg H,, Effective
ofParticipants
41
[Adults and Adolescents)
Isotope form
Dose (mCi)
UPPer
LCMer
Total
Bone
Stomach
Small
large
large
Ovary
Testis
body
marrow
H,
“‘In pellets
0.25 0.18
0.24 0.13
0.46 0.36
1.10 0.88
1.90 1.69
0.40 0.32
0.03 0.03
0.06 0.04
0.09 0.05
0.36 0.29
“‘In pellets
dose equivalent.
Although our scintigraphic colonic transit test is more costly than the radiopaque marker method, it appears to be advantageous in the assessment of accelerated colonic transit and provides the opportunity for more detailed assessment of transit without increasing radiation exposure. Development of this relatively inexpensive but accurate scintigraphic method for detecting accelerated or delayed colonic transit has important practical implications in gastroenterology clinics, which continue to be attended by large numbers of patients with functional gastrointestinal disease.16*17 Accurate questionnaire methods” allow categorization of patients into different subsets, e.g., diarrhea-predominant and constipation-predominant, but by their very nature are subjective assessments and on their own cannot be used to determine objectively the severity of illness and hence stratify patients for treatment. After the exclusion of mucosal disease or strictures by contrast colon x-ray or colonoscopy and of endocrine or metabolic problems such as hypothyroidism, our noninvasive method would provide a useful investigation to determine the severity of the transit disturbance. Such objective measurements of accelerated or delayed transit may also facilitate the choice of therapy. Thus, constipated patients with only modest prolongation of colonic transit may be given a fiber supplement and an osmotic laxative; in contrast, more significant prolongation in transit may warrant addition of a prokinetic agent or, in severe intractable cases, consideration of subtotal colectomy if dynamic studies of defecation are normal.” The relatively few scans proposed by the current method might preclude an accurate determination of the region(s) of abnormal transit in the colon. However, until medications or operative approaches that selectively correct regional abnormalities are developed. this deficiency is of little practical importance. The logistic discriminant analysis relies on large sample approximations to identify and estimate the model parameters and make inferences of their statistical significance. These results should be updated with additional subjects (controls and patients) spanning the spectrum of colonic transit.
The observations in these studies thus need to be confirmed in a larger group of patients, particularly in those with less severe colonic motility disorders than were observed in our patients with idiopathic constipation and carcinoid diarrhea. The data-analysis models developed in this preliminary work are being evaluated prospectively in a larger number of patients with functional and motility disorders of the gastrointestinal tract. The apparently greater accuracy of early (J-hour; Figures 2 and 3) than of later (&hour; Figure 5) transit parameters in this study probably reflects three facts: [a) marked difference in transit was seen in the diarrhea group, particularly the patients with carcinoid diarrhea, which weighted the J-hour data; (b) the data of the two disease groups are at the opposite ends of the transit spectrum relative to health, so that when combined as a single disease group for the ROC curves they tend to neutralize each other (Figure 5); and (c) we took an arbitrary cut-off of 24 hours for the “late” scan. Indeed, we anticipated that the %-hour data would have less discriminant value in the detection of the constipated group in view of the relatively prolonged residence time of solid residue in the proximal colon in health.7 Nevertheless, the .%-hour data have clear discriminant value to separate each of the two disease groups from healthy subjects (Table 1).It is conceivable that later scans, such as a @-hour image, may be necessary to enhance the accuracy of the scanning protocol in less severe motility disorders than those of the groups studied here. However, this increase in the number of later scans can be achieved with no further radiation exposure and minimal increments in the cost of the study. In conclusion, this study confirms the accuracy of our noninvasive, relatively inexpensive strategy using the delayed-release capsule method. Quantitation of isotope in the four colonic regions at 4 and 24 hours accurately detects accelerated or delayed transit in motility disorders of the colon. References 1. Hinton
studying
JM. Lennard-Jones JE, Young AC. A new method for gut transit times using radiopaque markers. Gut
1969;10:842-847.
42
CAMILLERI AND ZINSMEISTER
2. Arhan P, Devroede G, Jehannu B, Lanza M, Faverdin C, Dornit C, Persoz B, Titreault L, Perey B, Pellerin B. Segmental colonic transit time. Dis Colon Rectum 1981;24:625-629. 3. Metcalf AM, Phillips SF, Zinsmeister AR, MacCarty RL, Beart RW, Wolff BG. Simplified assessment of segmental colonic transit. Gastroenterology 1987;92:40-47. 4. Chaussade S, Khyari A, Roche H, Garret M, Gaudric M, Couturier D, Guerre J. Determination of total and segmental coionic transit time in constipated patients. Dig Dis Sci 1989;34:1169-1172. 5. Krevsky B, Malmud LS, D’Ercole F, Maurer AH, Fisher RS. Colonic transit scintigraphy. A physiologic approach to the quantitative measurement of colonic transit in humans. Gastroenterology 1986;91:1102-1112. 6. Kamm MA, Lennard-Jones JE, Thompson DG, Sobnack R, Garvie NW, Granowska M. Dynamic scanning defines a coionic defect in severe idiopathic constipation, Gut 1988;29: 1085-1092. 7. Proano M, Camilleri M, Phillips SF, Brown ML, Thomforde GM. Transit of solids through the human colon: regional quantification in the unprepared bowel. Am J Physiol 1990;258:G856-G862. 8. Camilleri M, Zinsmeister AR, Greydanus MP, Brown ML, Proano M. Towards a less costly but accurate test of gastric emptying and small bowel transit. Dig Dis Sci 1991;36:609615. 9. Stivland T, Camilleri M, Vassallo M, Proano M, Rath D, Brown
M, Thomforde G, Pemberton J, Phillips S. Scintigraphic measurement of regional gut transit in idiopathic constipation. Gastroenterology 1991;101:107-115. 10. Vassallo M, Camilleri M, Phillips SF, Brown ML, Chapman NJ, Thomforde GM. Transit through the proximal colon influences stool weight in irritable bowel syndrome. Gastroenterology 1992;102:102-108. 11. Camilleri M, Thomforde GM, Vassallo MJ, Kvols LK, Powers SP. Gastrointestinal transit, colonic capacitance and possible mediators in carcinoid diarrhea. Gut 1991;32:A1215.
GASTROENTEROLOGY Vol. 103, No. 1
12. Vassallo M, Camilleri M, Phillips SF, Chapman NJ, Thomforde GM. Colonic transit in diarrhea-predominant irritable bowel syndrome BBS): comparison by scintigraphy and radiopaque markers (abstr). Am J Gastroenterol 1990;85:1286. 13. SAS users guide: statistics. Cary, NC: SAS Institute, 1982:1537, 297-308. 14. SUGI supplemental library user’s guide. Cary, NC: SAS Institute, 1986:269-293. 15. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8:283-298. 16. Switz DM. What the gastroenterologist does all day. Gastroenterology 1976;70:1048-1050. 17. Lennard-Jones JE. Functional gastrointestinal disorders. N Engl J Med 1983;308:431-435. 18. Talley NJ, Phillips SF, Wiltgen CM, Zinsmeister AR, Melton LJ III. Assessment of functional gastrointestinal disease: the bowel disease questionnaire. Mayo Clin Proc 1990;65:14561479. 19. Pemberton JH, Rath DM, Ilstrup DM. Evaluation, management strategies and outcome of patients with severe chronic constipation (abstr). Gastroenterology lOO:A481, 1991.
Received May 31, 1991. Accepted December 10, 1991. Address requests for reprints to: Michael Camilleri, M.D., Gastroenterology Research Unit, Mayo Clinic, Rochester, Minnesota 55905. Supported in part by Digestive Disease Center grant DK34988 and by the General Clinical Research Center grant RR00585 from the National Institutes of Health. Reported in part at the British Society of Gastroenterology Meeting, London, England, September 1991, and published in abstract form Gut 1991;32:A1248. The authors thank Drs. M. L. Brown, L. A. Forstrom, M. Proano, T. A. Stivland, and M. J. Vassallo and G. M. Thomforde for technical support and Cindy Stanislav for typing and preparing the manuscript.