Introduction
Missed adenomas during screening colonoscopy remain a major concern and a leading
cause of interval cancer [1], making the adenoma detection rate (ADR) an important indicator of quality performance
for the endoscopist [2]. Polyps that arise in concealed locations can be difficult to detect with 140˚ to
170˚ forward-viewing conventional colonoscopy (CC) and may account for a proportion
of missed adenomas [3]. A number of colonoscopic devices that improve detection of these polyps have recently
been introduced [4]
[5].
The Aer-O-Scope™ Colonoscope System (GI View Ltd., Ramat-Gan, Israel) (AOS) is a disposable
gas-propelled colonoscope with a visualization system that utilizes a novel optical
concept designed to match the colon anatomy: a circumferential omni-directional 360°
panoramic viewer (OMNI view) projects a 360˚ view of the lumen [6]. An additional conventional forward view (57˚ field of view) is provided. The recent
addition of joystick-controlled steering for the colonoscope allows for simpler navigation
and centering control during withdrawal scan. Pathologies are visualized in both views
which are complementary, resulting in a substantially larger visualized area of the
colonic mucosa. These combined views are projected onto a single, user-friendly screen
with the OMNI view surrounding the forward view ([Fig. 1 a] and [Fig. 1 b]).
Fig. 1 Beads visualized by Aer-O-Scope™ Colonoscope. a Bead in panoramic (OMNI) view. b Bead in forward facing view.
The visualization properties of the AOS have been enhanced recently: New LEDs with
improved orientation provide greater coverage and higher light uniformity, gamma control
improves visibility in darker areas of the colon, a new front lens eliminates front-view
distortion, and full digital video pass from the optical head to the screen removes
analogue noise.
In this comparison to forward-viewing CC, we aimed to evaluate the performance of
the AOS in identifying simulated polyp-like lesions in an in vivo swine model, prior
to its introduction for clinical studies.
Patients and methods
Swine model
This was a prospective live swine model study performed at the Lahav Animal Research
Institute, Lahav, Israel. Twelve female swine (Sus scrofa domestica-large white),
aged 3.5 to 4.5 months and weighing 65 to 80 kg were studied. Animal care, facilities,
and activities were approved and monitored according to ISO9001 standards for quality
and service. The Tel Aviv University Animal Ethics Committee approved the study. Bowels
were cleansed using Soffodex and fleet enemas. Subsequently, simulated colon polyps
were surgically introduced under general anesthesia. Because the swine colon is tortuous,
the colon was ligated with thread just beyond the first sharp colonic curve (100 – 120 cm
from the anus) to facilitate colonoscope passage. Next, six to 10 colored beads measuring
2.5 to 10 mm in diameter were then introduced per animal in a preset random order
distal to the ligation (total of 91 beads placed in 12 swine). To better simulate
polyps, half the beads (46 /91) were pushed from outside the bowel wall inward and
tied at their base, resulting in mucosal coating of the bead (“pseudopolyps”) ([Fig. 2]). All other beads were surgically sutured to the colonic mucosa through a small
incision in the colonic wall. The endoscopists were blinded to all implanted beads.
Fig. 2 Schematic of simulated polyp implantation in wall of swine colon.
Endoscopist AOS training
Two gastroenterologists (NG & SF) highly experienced in colonoscopy ( > 5 years) attended
a 1-hour lecture and performed three AOS procedures on swine with implanted beads
(3 hours).
Study protocol
Each of the 12 swine underwent a total of three AOS and two CC (Olympus CF 140 L SD)
procedures (study total 60; 36 AOS and 24 CC). Lesions were counted and recorded during
withdrawal phase.
Several measures were taken to ensure that endoscopists were blinded to which pig
was being examined at any given time, thus preventing them from recalling previously
viewed pathologies within the colon of a specific animal. Procedures were randomly
assigned to the two endoscopists, procedures were performed in two separate rooms
on two different swine simultaneously, and the sequence of procedures (AOS or CC)
in each swine was random. The one exception to the random sequence was the condition
that six swine would begin the study with an AOS while six would begin with CC. This
exception was made to prevent bias and to neutralize the "second pass" effect of increased
polyp detection in tandem colonoscopy [7]. Swine, physicians, and colonoscope towers also were randomly alternated between
procedure rooms. To facilitate that, physicians had to leave rooms post procedures.
In order to neutralize polyp-miss due to momentary lack of attention or quick motion
through curves, all procedures were video recorded and later reviewed by two additional
highly experienced endoscopists (ES & AM), who were blinded to swine, pathologies
and the performing endoscopist. These physicians could stop the video, go back and
review sections at their discretion.
To determine the effectiveness of the randomization scheme in preventing recall, miss
rates for successive procedures (first, second, and third passes) for AOS and CC,
online and offline, were compared.
Study endpoints
The study was designed to demonstrate non-inferiority of the AOS relative to CC: the
number of simulated polyps detected by AOS in a given swine was at least the same
as with CC in two out of three repeated procedures. Alternatively, the number detected
in at least six animals would be as good as or better than CC.
Sample size calculation and statistical analysis
Sample size was based on the assumption that the equivalence ratio of AOS to CC (non-inferiority)
by pre-defined criteria was ≥ 80 %. A sample of 12 animals would then provide power
≥ 80 % (type II error < 20 %, type I error < 5 %). This included an estimated dropout
rate (such as swine death unrelated to the procedure) of 10 %.
Tests for calculating simulated polyp detection rates were 2-sided using Pearson Chi-square.
The threshold for a significant P value was 0.05.
Results
[Table 1] lists detection and miss rates for Aer-O-Scope colonscopy vs. conventional colonoscopy
as reported during colonoscopy. Overall, significantly more implanted beads were detected
using AOS [259/273 (94.9 %)] as compared to CC [158 /182 (86.8 %)] (P = 0.002). Interestingly, 20/259 beads detected by AOS (7.3 % of 273 implanted beads)
were visualized exclusively by the forward viewing mode, indicating a complementary
role for this view. The mean agreement rate of AOS with CC for lesion detection or
miss was 88.3 %. Subgroup analysis, stratified by polyp size 1 to 5 mm or ≥ 6 mm,
also demonstrated a statistically significant advantage for AOS in both subgroups.
Table 1
Detection and miss rates for Aer-O-Scope colonoscopy vs. conventional colonoscopy: live endoscopy procedure data.
|
Colonoscope type
|
Total beads implanted in all Procedures
|
Percent of beads visualized (bead no.)
|
Miss rate (%)
|
Pearson χ2
P value
|
|
Overall implanted beads
|
Aer-O-Scope
|
273
|
94.9 (259)
|
5.1
|
0.002
|
|
Conventional
|
182
|
86.8 (158)
|
13.2
|
|
Beads ≥ 6 mm
|
Aer-O-Scope
|
114
|
97.4 (111)
|
2.6
|
0.022
|
|
Conventional
|
76
|
89.5 (68)
|
10.5
|
|
Beads < 6 mm
|
Aer-O-Scope
|
159
|
93.1 (148)
|
6.9
|
0.031
|
|
Conventional
|
106
|
84.9 (90)
|
15.1
|
During post-endoscopy video review, significantly more implanted beads were detected
by AOS [261/273 (95.6 %)] as compared to CC [165/182 (90.8 %)], (P = 0.034) ([Table 2]). The mean agreement rate of AOS with CC for lesion detection or miss was 92.1 %.
Subgroup analysis stratified by size of polyps showed a trend toward an advantage
for AOS, though without statistical significance (P value = 0.055 for beads < 6 mm and 0.365 for beads ≥ 6 mm). There were no statistical
differences between visualization of exposed beads compared to “pseudopolyp”-like
beads by either AOS or CC, neither during live procedures nor offline video review
([Table 3]).
Table 2
Detection and miss rates for Aer-O-Scope colonoscopy vs. conventional colonoscopy: offline video review data.
|
Colonoscope type
|
Total beads implanted in all Procedures
|
Percent of beads visualized (bead no.)
|
Miss rate (%)
|
Pearson χ2
P value
|
|
Overall implanted beads
|
Aer-O-Scope
|
273
|
95.6 (261)
|
4.4
|
0.034
|
|
Conventional
|
182
|
90.7 (165)
|
9.3
|
|
Beads ≥ 6 mm
|
Aer-O-Scope
|
114
|
98.2 (112)
|
1.8
|
0.365
|
|
Conventional
|
76
|
96.1 (73)
|
3.9
|
|
Beads < 6 mm
|
Aer-O-Scope
|
159
|
93.7 (149)
|
6.3
|
0.055
|
|
Conventional
|
106
|
86.8 (92)
|
13.2
|
Table 3
Detection of pseudopolyps and exposed beads.
|
AOS live total visualized
|
CC live total visualized
|
AOS review total visualized
|
CC review total visualized
|
Total % missed (live)
|
Total % missed (review)
|
Total % missed (overall)
|
|
Pseudopolyps
|
129/135
|
78/90
|
131/135
|
82/90
|
8.0
|
5.3
|
6.7
|
|
Beads
|
130/138
|
80/92
|
130/138
|
83/92
|
8.7
|
7.4
|
8.0
|
AOS, Aer-O-Scope Colonoscope System; CC, conventional colonoscopy
We wished to ascertain whether the increased number of AOS procedures performed relative
to CC could bias toward better detection, and whether the randomization scheme was
effective in preventing physicians’ recollection of consecutive colonoscopies on the
same swine. As seen in [Table 4], the difference between AOS and CC was already evident in the first passes, and
the detection rates did not improve after several successive procedures as might have
been expected.
Table 4
Detection rates by successive pass
|
Pass #
|
AOS detected/total lesions (%)
|
AOS offline review detected/total lesions (%)
|
CC detected/total lesions (%)
|
CC offline review detected/total lesions (%)
|
|
1
|
86 /91 (95)
|
87 /91 (96)
|
80 /91 (88)
|
87 /91 (96)
|
|
2
|
85 /91 (93)
|
87 /91 (96)
|
78 /91 (86)
|
79 /91 (87)
|
|
3
|
88 /91 (97)
|
88 /91 (97)
|
–
|
–
|
AOS, Aer-O-Scope Colonoscope System; CC, conventional colonoscopy
Complete intubation to the depth of colon ligation was achieved in 60/60 (100 %) procedures.
No adverse events were experienced in any of the procedures. These were not study
endpoints.
Discussion
This comparative study demonstrates the efficacy of AOS, a novel disposable colonoscope,
for simulated polyp detection in a swine model. The advantage of the AOS over CC was
demonstrated overall as well as in lesions of clinically significant size (≥ 6 mm).
Notably, training on the AOS was completed in just half a day, despite differences
in manual technique and image display ([Fig. 1]).
Previous AOS studies have shown images of native swine colon or of ex-vivo sewn beads
obtained with forward and OMNI panoramic views, and safety of use in live swine [6]
[8]. The current study is the first statistically significant, in-vivo comparative blinded
study to test the simulated polyp detection ability of the AOS.
The study design, in which the number of pathologies was given, enabled us to accurately
compare the absolute miss rates of both the AOS and CC. Miss rates in both modalities
were inversely related to polyp size. Interestingly, CC miss rates in this study were
lower than in human tandem colonoscopy studies [9], supporting the adequacy of performance by the operators. Another strength of the
study is the use of four independent physicians.
Some polyps were missed by performing endoscopists in both AOS and CC (more so in
the CC group), despite being visible on video recordings. Despite not being part of
real-world practice, offline video review adds objectivity to the comparison of the
devices and is becoming an accepted quality indicator tool for polyp detection studies
[10], prompting us to include these data. An advantage for AOS was evident both in post-procedural
reports (when findings are typically reported in clinical practice) and in later video
analysis. The offline review also allowed for a larger sample size and physician pool
for statistical analysis.
There are several limitations to this study. Porcine anatomy differs from the human
colon in having no haustral folds. Thus, the reason for improved detection by AOS
relative to CC becomes less clear. In addition, the colonoscope used in this study,
although providing high-quality images, has somewhat less field of view (140°) than
currently available technology (170°). Procedures were repeated in 12 swine, potentially
resulting in physicians remembering specific, previously scoped swine colons. However,
this appears to have been ameliorated by the randomization scheme ([Table 4]).
In conclusion, the Aer-O-Scope gas-propelled colonoscope, featuring a novel 360◦ OMNI-directional view, provided excellent simulated polyp detection rates in this
in vivo swine study and may show an advantage over CC. The rapid training of the physicians
operating this system did not appear to compromise the study results.
Further studies of screening colonoscopy in humans are warranted to validate the clinical
relevance of these animal model findings.
