Introduction
In recent years, emphasis has been placed on colonoscopic quality, leading to reinvestigation
of basic elements of colonoscopic extubation technique, such as withdrawal time, changing
patient position, operator technique, and bowel preparation [1]
[2]
[3]
[4], that have been shown to impact polyp and adenoma detection rates. Basic technique
can be improved to maximize adenoma detection with minimal additional financial cost.
Routine use of antispasmodics during colonoscopy is controversial. Although the literature
is inconsistent, in some studies, hyoscine butylbromide has been shown to accelerate
colonoscopic intubation, and it may reduce patient procedural discomfort [5]
[6]
[7]. Other antispasmodics that have been investigated include glucagon, dicyclomine,
hyoscyamine, atropine, peppermint oil, and warm water [8]
[9]
[10]
[11]
[12]
[13]
[14]. None of these has been shown to reduce intubation time, and only topical peppermint
oil and warm water have been shown to reduce spasm scores or pain [9]
[14]. Until recently, very little investigation has been done of use of antispasmodics
during the extubation or withdrawal phase of colonoscopy; however, the rationale for
use of antispasmodics during extubation is not unreasonable. By reducing smooth muscle
tone, haustral folds can be flattened, allowing better visualization of blind spots
behind them, and peristaltic waves may be reduced, giving a still colonic surface
on which to detect lesions. Indeed, studies using unblinding at CT colonography (CTC)
suggest that most polyps missed by optical colonoscopy are on the back of folds in
these blind spots [15]. A recent meta-analysis reported a relative risk for adenoma detection of 1.09 (95 %
confidence interval 0.91 – 1.31) and for polyps of 1.13 (95 % CI 0.92 – 1.38), numerically
but non-significantly, in favor of hyoscine [16], and in a very large cohort study (n = 31088), the proportion of patients with at
least one adenoma detected was 50.1 % with hyoscine versus 44.5 % without (relative
increase 12.6 %, P < 0.001) with similar improvements for advanced adenoma detection rates [17]. This pair of estimates suggests that the relative improvement in adenoma detection
rate is likely to be on the order of 9 % to 3 %.
Despite the theoretical advantages of routine use of antispasmodics and expert recommendation,
implementation has been variable worldwide. In a United Kingdom national colonoscopy
audit, for example, only 20 % of colonoscopists used antispasmodics routinely [18]. In the United States and France, hyoscine butylbromide is unlicensed for colonoscopy.
Furthermore, use of antispasmodics is not recommended in the US multi-society taskforce
guidelines on quality in colonoscopic technique [19]; however, the converse is true in Japan, where usage is routine.
Hyoscine butylbromide is the antispasmodic most commonly used during colonoscopy.
Data from the CTC and barium enema literature suggest that it is superior to both
glucagon and placebo in improving colonic distension and diagnostic quality [20]
[21]
[22]. Assessing the effect of hyoscine butylbromide on mucosal visualization during colonoscopic
extubation is technically very challenging. However, recently developed, customized
CTC software is able to calculate the amount of colonic surface seen during simulated
colonoscope withdrawal, corrected for the field of view of a modern colonoscope, as
we have previously reported [23].
The purpose of our study was to approximate the changes in percentage surface visualization,
numbers and sizes of missed areas, and changes in colonic length that are likely to
be encountered with intravenous (IV) hyoscine butylbromide when colonoscopy is simulated
by CTC in patients suspected of having colorectal cancer. Given that this is a simulation
of colonoscopic withdrawal, the study results should be viewed cautiously; however,
they may give some insight into the mechanism by which hyoscine appears to improve
adenoma detection in clinical studies.
Patients and methods
The protocol for the original CTC study was approved by our Local Regional Ethics
Committee and all patients gave written informed consent. Specific permission for
this additional analysis was sought from and granted by the same Ethics Committee.
CTC dataset selection and spasmolytic administration
CTC datasets were collated from a previous randomized trial assessing the effect of
hyoscine butylbromide on colonic distension at CTC in 136 patients. All patients were
clinically suspected of having colorectal cancer. The methods used in the previous
CTC study have been reported in detail elsewhere and are described briefly below.
[21]
The first 20 patients received 20 mg hyoscine butylbromide (Buscopan; Boehringer Ingelheim,
Bracknell, England) IV given just before colonic gas insufflation (see below). Thereafter
patients were randomised to receive either no antispasmodic or a slow bolus of 20 mg
(given as before) or 40 mg of hyoscine butylbromide. If patients were randomised to
40 mg, 20 mg was given before the colonic gas insufflation in the prone position (performed
first) and then an additional 20 mg was given before the supine scan (performed immediately
after the prone scan). Therefore, only supine datasets reflect the full 40-mg dosage
for those randomised to 40 mg. Contraindications to antispasmodic were recently symptomatic
ischemic heart disease and a history of closed-angle glaucoma. Three patients originally
randomized to the antispasmodic group had contraindications and were assigned to the
no antispasmodic group. Minimization was used to balance the groups as accrual progressed.
In total, 40 patients received no antispasmodic, 66 received 20 mg, and 30 received
40 mg hyoscine butylbromide. Due to data loss, data corruption, or inadvertent destruction,
only 86 of the original 136 datasets were available for the current study, 33 without
antispasmodic, 35 assigned to 20 mg, and 18 assigned to 40-mg hyoscine butylbromide.
CTC protocol
All patients underwent colonic insufflation of carbon dioxide via manual compression
of a previously filled enema bag, until approximately 2500 mL had been introduced
or to the limit of patient tolerance. Gas was introduced via either a thin rectal
tube or a rectal balloon catheter, with allocation via a separate randomization sequence.
Patients were scanned initially in the prone position. After this, those who were
allocated to 40 mg hyoscine had further administration as described above. All patients
then had additional gas introduced via the rectal catheter to account for carbon dioxide
absorption during the prone scan. Scout scans were used before data acquisition in
each position to determine if distension was adequate.
CTC dataset analysis
The 86 datasets were analysed with customized proprietary CTC software (V3 D colon;
Viatronix, Stonybook, NY) by a gastroenterologist who had received formal training
in CTC examination (over 40 endoscopically validated datasets) and had experience
with the software functionality. The technique has been described previously [23]. In brief, the software automatically segments gas-filled colon from surrounding
organs and calculates a centerline within the segmented lumen from the anal verge
to the cecum, facilitating automated 3-dimensional (3 D) endoluminal navigation ([Fig. 1 a]). The user is able to check the segmentation via a 3 D colon map and adjust if necessary.
The software tracks the amount of colonic surface within the field of view during
automated endoluminal navigation and calculates the amount of visualized colonic surface
area (expressed as a percentage of the whole colonic surface), and the number, size,
and distance from the anal verge of all missed areas (i. e., areas of unseen colonic
surface) [24]. The visualized surface is “painted” green by the software, thus indicating missed
areas to the observer ([Fig. 1 b]). Total colonic length is also reported automatically, calculated via the centreline.
Fig. 1 a Overview of the colon with centerline for navigation (green) automatically drawn
by the CT colonography software. b Overview after unidirectional 3D-endoluminal flythrough facing the cecum, where the
software had “painted” visualized areas green. Unseen areas remain beige.
Customization of the software for the purposes of the current study allowed the observer
to vary the field of view of the endoluminal “camera” from 0 to 180 degrees, and to
download missed area data onto a computer spreadsheet.
The observer then performed an automated flythrough without detours along the software-determined
centerline, with the virtual camera facing the cecum, equivalent to the view during
optical colonoscope withdrawal. The field of view was set to 140 degrees for the current
study, specifically to simulate a standard, optical colonoscope. The observer performed
the flythrough on both prone and supine series from all 86 CTC datasets, recording
percentage mucosal visualization, number of missed areas [total, intermediate sized
(300 – 1000 mm2), and large (> 1000 mm2)], and total colonic length.
Statistical analysis
The primary outcome measure was the percentage surface visualization achieved in patients
receiving either no antispasmodic or antispasmodic. The supine position was chosen
for analysis because it was felt to give the best distension overall, particularly
in the transverse colon, and of the two available options (supine or prone), was felt
to best model usual patient position during colonoscopy. Age, percentage colonic surface
visualization, missed areas, and colonic length were compared between the two groups
using unpaired, 2-tailed, t-tests. Sex ratios and use of a rectal balloon catheter
were compared with Fisher’s exact test. Further data analyses were performed for a
split by dose of antispasmodic. Probability values were considered significant at
the 5 % level and P values ≥ 0.05 but < 0.1 were considered to indicate a statistical trend.
Results
Datasets (n = 86) were initially split into those who had received antispasmodic (either
20 mg or 40 mg, n = 53) and those who had not (n = 33) ([Table 1]). There were no significant differences in age, sex ratio, or use of rectal balloon
catheter between the two groups. When antispasmodic was used, there was a highly significant
relative increase in the total colonic surface visualized (2.6 and 3.9 % for prone
and supine datasets, respectively), equivalent to relative reductions in percentage
surface unseen of 18 % and 25 % ([Table 1]).
Table 1
Surface visualization, missed areas, and colonic length with antispasmodic versus
no antispasmodic.
|
Position
|
Hyoscine butylbromide IV
|
P value
|
|
|
No (n = 33)
|
Yes (n = 53)[1]
|
|
|
Age, years
|
|
65.8 ± 11.0
(41 – 89)
|
62.2 ± 12.4
(34 – 85)
|
0.16
|
|
Sex, male/female (%)
|
|
13 (39 %)
|
28 (53 %)
|
0.27[2]
|
|
Balloon catheter (%)
|
|
17 (52 %)
|
29 (55 %)
|
0.83[2]
|
|
% colonic surface visualization
|
Prone
|
87.3 ± 3.9
(77 – 95)
|
89.6 ± 3.6
(78 – 95)
|
0.005
|
|
Supine
|
86.4 ± 4.3
(75 – 95)
|
89.8 ± 3.3
(78 – 96)
|
< 0.001
|
|
Total number missed areas
|
Prone
|
87.9 ± 30.7
(25 – 157)
|
65.6 ± 26.3
(22 – 178)
|
0.002
|
|
Supine
|
89.0 ± 34.8
(28 – 156)
|
68.6 ± 27.7
(22 – 159)
|
0.011
|
|
Missed areas 300 – 1000 mm2
|
Prone
|
24.6 ± 10.2
(6 – 55)
|
20.2 ± 10.8
(6 – 75)
|
0.091
|
|
Supine
|
25.5 ± 12.5
(8 – 54)
|
19.6 ± 10.1
(7 – 62)
|
0.044
|
|
Missed areas> 1000 mm2
|
Prone
|
6.8 ± 4.8
(1 – 20)
|
7.0 ± 5.1
(2 – 33)
|
0.64
|
|
Supine
|
7.2 ± 5.9
(1 – 28)
|
6.6 ± 4.3
(0 – 23)
|
0.73
|
|
Total colonic length, cm
|
Prone
|
168.8 ± 20.3
(133.4 – 222.0)
|
166.9 ± 28.0
(115.0 – 246.7)
|
0.73
|
|
Supine
|
161.3 ± 25.2
(109.9 – 231.2)
|
161.3 ± 26.3
(106.7 – 212.4)
|
1.00
|
Data presented as mean ± standard deviation (range).
1 20 mg or 40 mg hyoscine
2 Fisher’s exact test
The total number of missed areas of any size was reduced significantly by antispasmodic,
from approximately 90 missed areas to just over 65 (28 % relative reduction). When
the missed areas were subclassified by size, there was a significant reduction in
intermediate-size (300 – 1000 mm2) missed areas in the supine position, and a trend toward reduction in the prone position,
from approximately 25 to 20 (20 % relative reduction) missed areas. There were no
significant differences in the numbers of large (> 1000 mm2) missed areas ([Table 1]).
No significant difference in total colonic length was seen with antispasmodic, with
an overall length of between 160 and 170 cm. Colonic lengths ranged from 110 to 247 cm,
with 6 (11 %) and 3 (6 %) colons being longer than 200 cm in the prone and supine
positions, respectively, in the antispasmodic group versus 3 (9 %) and 2 (6 %), respectively,
in the group that received no antispasmodics ([Table 1]).
The supine dataset was then divided into three groups: Those who received no antispasmodic
(n = 33), those who received 20 mg (n = 35), and those who received 40 mg (n = 18)
([Table 2]). Again significant increases were seen in total colonic visualization for both
20 mg and 40 mg compared with no antispasmodic, with effect sizes similar to those
seen in the antispasmodic versus no antispasmodic analysis. There was, however, no
significant difference for any of the parameters tested when the 20-mg and 40-mg groups
were compared. Similarly there were significant reductions or there was a trend toward
significance for reduction in total missed areas and intermediate (300 – 1000 mm2) sized missed areas between the no-antispasmodic group and both the 20-mg and 40-mg
groups, but not between the 20-mg and 40-mg groups ([Table 2]). There were no significant differences in the numbers of large (> 1000 mm2) missed areas or in colonic length among any of the groups.
Table 2
Results for varying doses of antispasmodic, supine position.
|
Variable
|
Hyoscine Butylbromide IV
|
P value
|
|
Nil[1]
(n = 31)
|
20 mg[1]
(n = 34)
|
40 mg[1]
(n = 14)
|
Nil vs 20 mg
|
Nil vs 40 mg
|
20 vs 40 mg
|
|
Age, years
|
65.5 ± 10.3
(41 – 81)
|
61.7 ± 13.1
(34 – 85)
|
64.2 ± 9.8
(45 – 83)
|
0.20
|
0.68
|
0.48
|
|
Sex, male (%)
|
12 (39 %)
|
15 (44 %)
|
9 (64 %)
|
0.80[2]
|
0.34[2]
|
0.20[2]
|
|
Balloon catheter (%)
|
16 (52 %)
|
18 (53 %)
|
7 (50 %)
|
1.0[2]
|
1.0[2]
|
1.0[2]
|
|
% Colonic surface visualization
|
86.4 ± 4.3
(75 – 95)
|
89.6 ± 3.6
(78 – 95)
|
90.4 ± 2.6
(84 – 95)
|
0.002
|
< 0.001
|
0.43
|
|
Total number missed areas
|
89.0 ± 34.8
(28 – 156)
|
68.7 ± 29.3
(22 – 159)
|
72.5 ± 24.3
(40 – 114)
|
0.011
|
0.081
|
0.40
|
|
Missed areas 300 – 1000 mm2
|
25.5 ± 12.5
(8 – 54)
|
20.0 ± 11.5
(8 – 62)
|
18.8 ± 6.3
(7 – 26)
|
0.071
|
0.023
|
0.97
|
|
Missed areas > 1000 mm2
|
7.2 ± 5.9
(1 – 28)
|
6.2 ± 4.3
(1 – 23)
|
7.7 ± 4.4
0 – 16
|
0.45
|
0.76
|
0.35
|
|
Total colonic length, cm
|
161.3 ± 25.2
(109.9 – 231.2)
|
158.3 ± 28.3
(106.7 – 212.4)
|
164.9 ± 18.3
(135.1 – 193.0)
|
0.66
|
0.59
|
0.35
|
Data presented as mean ± standard deviation (range).
1 Supine scans unavailable for all patients leading to a reduced number of datasets
for analysis
2 Fisher’s exact test
Discussion
Principal findings
This study suggests that use of hyoscine butylbromide increases the relative percentage
of colonic surface visualized at simulated colonoscope withdrawal by approximately
4 %. There were also significant relative decreases of approximately 20 % in both
the total number of missed areas and the intermediate-sized (300 – 1000 mm2) missed areas with the use of antispasmodic. A reduction in intermediate-sized missed
areas may be important clinically because these areas might harbor a small (6 – 9 mm)
or diminutive (≤ 5 mm) polyp. There was little difference between prone and supine
positioning; however, supine data are preferred to represent the best simulation of
optical colonoscopy as the best distension is achieved, because the transverse colon
can be collapsed in the prone position at CT colonography [23]. Furthermore, prone positioning is rarely used during the withdrawal phase of colonoscopy
[2].
Colorectal length was not significantly altered by antispasmodic. Total colonic length
was 169 cm when prone and 161 cm in supine scans. The proportion of patients with
colons longer than 200 cm – between 6 % and 11 % – is roughly half the reported proportion
of patients who have “difficult” colonoscopies, and may reflect potential looping
problems rather than fixation or diverticulosis [25].
Comparing 20-mg to 40-mg IV hyoscine, there were no significant differences, suggesting
that the lower dose is adequate to optimize visualization and decrease missed areas.
Interestingly, the proportion of missed areas overall that occurred in the rectum
and sigmoid (20 % – 22 %) was much less than might have been predicted from the proportion
of the colonic length examined (37 % – 39 %), suggesting that missed areas are more
common in the proximal colon, an area already known to be at higher risk for cancer
misses and failed cancer prevention after colonoscopy [26]
[27].
Comparison with other studies
The benefit of hyoscine butylbromide in improving adenoma and polyp detection have
been reviewed in a recent meta-analysis, which reported a relative risk for adenoma
detection of 1.09 (95 % confidence interval 0.91 – 1.31) and polyps 1.13 (95 % CI
0.92 – 1.38) numerically but nonsignificantly in favor of hyoscine [16], and in a very large cohort study (n = 31088) from the National Health Service bowel
cancer screening program, the proportion of patients with at least one adenoma detected
was 50.1 % with hyoscine versus 44.5 % without (relative increase 12.6 %, P < 0.001) with similar improvements for advanced adenoma detection rates 27.4 % vs
31.8 %, P < 0.001. These differences persisted after correction for other variables. Therefore,
the 4 % increase in surface visualization seen in our study in the supine position,
most representative of optical colonoscopy, might explain up to half the benefit for
polyp detection seen in clinical studies in which a 9 % to 13 % relative increase
is reported. The residual benefit may be due to having a still surface for polyp detection
Mean colonic length reported here, 161 to 169 cm, is broadly similar to that seen
in other studies assessing length with CTC, but is longer than the length reported
in studies that used barium enema to assess colonic length, 145 – 155 cm [20]
[25]
[28].
One study has investigated the potentially adverse effects of hyoscine on hemodynamics
at colonoscopy by giving larger doses (40 mg) to accentuate responses [6]. The relevance of these findings seems limited if such a large dose is unlikely
to be needed clinically for optimal visualization, as suggested in the current study.
Study limitations
This study has a number of limitations. The original participants, whose data were
analyzed further here, were recruited prospectively from outpatient clinics and had
symptoms suggestive of colorectal cancer. Unfortunately only two-thirds of the dataset
was available for review, making the current study less representative of the original
population and possibly adding unknown bias, although data loss was random and groups
remain well matched. As discussed above, the simulation represents a “straight pull-back”
withdrawal technique where the colonoscopist attempts to keep the tip of the scope
in the center of the lumen and withdraws slowly. In reality, colonoscopists use a
more active withdrawal technique, which may increase the amount of colonic surface
seen beyond that reported here. The data presented, therefore, should not be regarded
as reflecting absolute percentage visualization at optical colonoscopy, but rather,
a guide to likely effect sizes if spasmolysis were employed.
We assumed that increased surface visualization leads to increased polyp detection,
which is logical but unproven; however, recent data from a clinical trial that used
a retrograde viewing auxiliary imaging device to improve surface visualization showed
an overall increase in adenoma detection of 11.0 %, very similar to the predicted
increase of 12.1 % in surface visualization in a previous simulation of such a device
using the current CTC simulation [23]
[29]. Although the simulation gives a quantitative measurement of effect size in terms
of surface visualization and sizes of missed areas, antispasmodic has other effects.
In particular, it reduces peristalsis, which immobilizes the surface visualized, and
could potentially be as or more important than improvements in surface visualization.
Clinical implications
The current study would support the clinical use of hyoscine butylbromide to optimize
colonoscopic visualization, but suggests that the effect size is modest; however,
that may explain up to half the benefit reported for adenoma detection seen in clinical
studies. The minimal cost involved (one 20-mg ampoule Buscopan® for injection, £ 0.29, British National Formulary, 2015) for this modest benefit
may be acceptable.
Overall the lack of change in colonic length seen in this study would argue against
those who have concerns regarding antispasmodics increasing colonic length at colonoscopy.
Excessive colonic length estimated by barium enema is known to predict difficult colonoscopy
[30].
From our study data, it seems unlikely that there is likely to be further benefit
from doses of hyoscine beyond 20 mg. We were not able to assess whether even smaller
doses might give the same clinical benefit.
Conclusions
In this anatomical simulation study of colonoscope withdrawal to examine the mechanism
of benefit on adenoma detection, we found that IV administration of hyoscine butylbromide
increased the percentage of colonic surface visualization by approximately 4 %, with
a relative decrease of 20 % in the number of clinically significant (300 – 1000 mm2) missed areas. This mechanism may explain up to half the improvement in adenoma and
polyp detection seen in clinical studies.
