The were the only multi-centre studies including two
searches of PubMed and EMBASE yielded a total of 79 studies. Two additional
studies were identified after screening of references, generating a total of 81
studies. After correcting for duplicates, 73 studies remained. Of these, 53
were excluded because after reading their titles and abstracts they clearly failed
to meet the eligibility criteria or did not answer the research question. The full-text
of the remaining 20 studies was assessed in more detail. Fourteen of these
studies did not meet the pre-defined inclusion criteria due to various reasons
(wrong population (n = 3), addition of technique to conventional control (n =
10), not clear randomization process (n = 1)). Six studies met the eligibility
criteria and were included in this systematic review 17-20, 25, 26. An
overview of the study selection is provided in the flow diagram of Figure 1.
An overview of the study characteristics
can be found in Tables 1 and 2. A
more detailed summary of the characteristics is given this chapter and has been
attached in Appendix III.
3.2.1 Setting and Methods
All six included
studies were RCTs assessing the effect of chromoendoscopy on the detection of
colorectal neoplasms compared to conventional colonoscopy. Two trials, Pohl et al.18 and Kiesslich et al.26, were
from Germany and carried out by one of five or one of two experienced
endoscopists respectively. Hashimoto et
al.17 and Park et al.25 were
the only trials from Asia (Japan and Korea) and a total of ten experienced
endoscopists divided over the two trials performed the procedures. One trial
was from five centres in Canada, USA and Israel and colonoscopies were carried
out by one of the eight trained endoscopists (Stoffel et al.19).
Another trial was from the UK and one of two experienced endoscopists performed
the procedures (Hurlstone et al.20). Pohl et al.18 and Stoffel et al.19 were
the only multi-centre studies including two and five centres respectively (Table 1).
Three out of six
trials had a different study design (Hashimoto
et al.17; Park et al.25; Stoffel et al.19). In
these three trials two consecutive colonoscopies were performed on each
participant. During the first intubation the colon was examined with
conventional WLC in both study arms. The second intubation was either
chromocolonoscopy in the intervention arm or repeated conventional WLC in the
control arm. As these trials essentially measured the additional number of
neoplasms detected, subgroup analysis differentiating between single and double
intubation was carried out. The remaining three trials intubated participants
once (Pohl et al.18; Hurlstone et al.20; Kiesslich et al.26).
A total of 1913
participants (mean age: 56.3 ± 8.3) divided
over six studies were included in this review (Table 1). These participants varied from consecutive patients
undergoing routine colonoscopies in three trials (Pohl et al.18; Park et al.25; Hurlstone et al.20), to
patients undergoing follow-up screening for alarming abdominal symptoms or
suspicion of malignancies (Hashimoto et
al.17), patients with longstanding ulcerative
colitis (Kiesslich et al.26) and patients with a previous personal
or familial history of CRC (Stoffel et
al.19). The population of the three last-mentioned studies was
considered as high-risk and as a consequence subgroup analysis was performed.
All studies investigated the effect of chromocolonoscopy on
neoplasm detection. Four studies performed pancolonic dye-spraying using an
endoscopic catheter (Pohl et al.18; Stoffel et al.19; Hurlstone et al.20; Kiesslich et al.26),
whereas the two other studies only applied dye to the ascending colon (Park et al.25)
and descending colon (Hashimoto et al.17). Indigo carmine was used in all studies except for Kiesslich et al.26,
where methylene blue was used as the dye. Conventional WLC was used as the
comparator in all studies (Table 2).
assessed the primary outcome measure of interest: the number of detected colorectal
neoplasms per patient. Data regarding secondary outcome measures was available
in most of the studies. All studies reported extubation times, three out of six
studies reported the number of adenocarcinomas and four out of six studies
reported the number of hyperplastic lesions. The occurrence of adverse events
was only reported in two studies (Table
Risk of bias within studies
Methodological quality of the included studies was assessed for six
different criteria. Judgement of bias was scored as low risk, high risk or
unclear risk and results are presented in Figure
2 and Figure 3. Support for these judgements has been attached in Appendix
III. Overall risk of bias
was graded as low/unclear.
Hashimoto et al.17was
the only study that did not provide details on the randomization process and concealment
of allocation. The other studies, who provided these details, randomized participants using
standard randomization lists (Pohl et al.18) stratified randomization (Stoffel et al.19) and
computer-generated randomization lists (Park
et al.25; Hurlstone et al.20; Kiesslich et al.26). In
these studies allocation remained concealed until the intervention started.
Due to the
nature of this intervention it was impossible to blind the colonoscopists.
Blinding of participants was irrelevant because they underwent the procedure
and could not have affected the outcome measures. In an attempt to reduce, or
at least minimize, bias two studies (Pohl
et al.18; Hurlstone et al.20) tried
to correct for longer extubation times in the chromocolonoscopy arm by using a
saline solution to spray the colon in the control arm.
3.3 Incomplete outcome data
except Stoffel et al.19 and Hurlstone et al.20
documented reasons for secondary dropouts (withdrawals after randomization).
Dropout rates were comparable across study arms (Table 1).
3.3.4 Selective reporting
reported pre-specified outcomes, but differentiation between type of lesions
was not clearly documented in Hashimoto
Furthermore, no study reported the occurrence of adverse events.
3.3.5 Other bias
Pohl et al.18 and Hashimoto et al.17
reported statistically significant increases in extubation times in the
chromocolonoscopy arm, whereas the other studies either controlled for possible
increases in extubation times or showed no statistical differences.
3.4 Results of individual
individual studies have been extracted based on predefined outcome measures of
interest and are presented in Table 2.
The primary outcome measure, mean number of detected colorectal neoplasms per
patient, was reported in all studies and ranged from 0.19-0.95 (SD: 2.00) in
the chromocolonoscopy group to 0.06-0.66 (SD: 2.00) in the WLC group. With
regard to secondary outcome measures, data on number of adenocarcinomas was
unavailable in three studies (Hashimoto et al.17; Park et al.25; Stoffel et al.19) and
number of hyperplastic lesions in two studies (Hashimoto et al.17; Kiesslich et al.26). Studies
that did measure these secondary outcomes, reported similar results across
study arms for adenocarcinoma detection (Pohl
et al.18; Hurlstone et al.20)
except for Kiesslich et al.26 who
found three adenocarcinomas in 84 patients in the chromocolonoscopy and only
one adenocarcinoma in 81 patients in the WLC arm.
and adverse events have not been included in the meta analyses due to the
nature of these outcome measures (extubation) and availability of data (adverse
events) . Only two studies (Park et al.25; Stoffel et al.19) reported
on occurrences of adverse events (n= 0 in both studies) and the remaining four studies
did not disclose any information on procedure-related adverse events. Extubation
times differed significantly across studies, but in all studies
chromocolonoscopy took longer to complete than conventional WLC. Hashimoto et al.17 and Park et al.25, who
only inspected segments of the colon, revealed shorter extubation times as a
logical consequence (10.9 vs. 6.8 and
1.9 vs. 1.8 minutes, respectively). Overall
differences in mean extubation time varied from 0.1 minutes to 9.6 minutes
between the two study arms.
3.5 Syntheses of results
difference in favour of chromocolonoscopy was observed for almost all measured
outcomes. The mean number of detected colorectal neoplasms per patient (primary
outcome) was greater for all studies in the chromocolonoscopy arm except for Hashimoto et al.17, in
which more neoplasms were detected in the conventional arm (Table 2). Combining the studies resulted in a great statistical significance
(Mean difference (MD) 0.29; 95%CI 0.12-0.46; P=0.0007; 6 studies; 1913 participants) with no significant
heterogeneity (I2 = 0%, P=0.52)(Figure 4, Analysis 1.1). Subgroup
analysis differentiating between study designs using single or double
intubation had effect on effect estimates. The increased detection of
colorectal neoplasms in the double intubation group was not statistically
significant (MD 0.24; 95%CI -0.04-0.53; P=0.09)(Figure 4, Analysis 1.1.1),
but remained significant in the single intubation group (MD 0.31; 95%CI 0.11-0.52; P=0.003)(Figure 4, Analysis 1.1.2). A subgroup
analysis considering the effects of high or low-risk patients on the primary
outcome ruled out differences as increase in yield was comparable between
groups (Figure 7, Analysis 4.1).
adenocarcinoma detection with chromocolonoscopy could be of great clinical use.
Unfortunately, chromocolonoscopy did not affect the detection rate of
adenocarcinomas (MD 0.00; 95%CI -0.10-0.11; P=0.95;
3 studies; 1433 participants)(Figure 5, Analysis 2.1). Furthermore,
hyperplastic lesions were clearly detected more in the chromocolonoscopy arm (MD
0.49; 95%CI 0.30-0.67; P<0.00001; 4 studies; 1618 participants)(Figure 6, Analysis 3.1). However, substantial heterogeneity was detected (I2 = 60%) and differentiation between intubation strategy revealed that this statistical significant yield was only maintained in the single intubation group (Figure 6, Analysis 3.1.1). Finally, a statistically significant increase in the number of patients with at least one colorectal neoplasm was found in the chromocolonoscopy arm (Odds ratio (OR) 1.89; 95%CI 1.53-2.32; P<0.00001; 5 studies; 1783 participants)(Figure 8, Analysis 5.1) and this increase was maintained following a subgroup analysis of both study designs (Figure 8, Analysis 5.1.1 and 5.1.2).