Accuracy and safety of second-generation PillCam COLON capsule for colorectal polyp detection

Abstract

PillCam COLON capsule endoscopy (CCE) (Given Imaging Ltd, Yoqneam, Israel) is one of the most recent diagnostic, endoscopic technologies designed to explore the colon. CCE is a noninvasive, patient-friendly technique that is able to explore the colon without requiring sedation and air insufflation. The first generation of CCE was released onto the market in 2006 and although it generated great enthusiasm, it showed suboptimal accuracy. Recently, a second-generation system (PillCam COLON 2) (CCE-2) has been developed to increase sensitivity for colorectal polyp detection. In this review, accuracy and safety data for CCE-2 are analyzed.

Introduction

Colorectal polyps are a major cause of morbidity and mortality. Colonoscopy is actually the procedure of choice for the evaluation of colorectal disease and has a relevant role in colorectal cancer screening programs, allowing direct visualization of the colon, the identification and removal of precancerous adenomatous polyps [Ferlay et al. 2007; Ries et al. 2000].

However, there are challenges associated with colonoscopy being perceived as an invasive and potentially painful procedure. Colonoscopy also has a risk of major complications and is associated with limitations related to patients’ fear of discomfort and inconvenience as well as psychological inhibition. For these reasons, colorectal cancer screening uptake is low, especially compared with the high rate of attendance of other cancer screening programs (breast, cervical and prostate) [Lisi et al. 2010; Segnan et al. 2007]. These issues have encouraged research to identify new, minimally invasive, patient-friendly methods to visualize the colon. PillCam COLON capsule endoscopy (CCE) (Given Imaging Ltd, Yoqneam, Israel) is one of the most recent diagnostic, endoscopic technologies designed to explore the colon. CCE is a noninvasive, swallowed ‘colonoscope’ that is able to explore the colon without requiring sedation and air insufflation. CCE has been on the market since 2006 [Eliakim et al. 2006; Spada et al. 2010]. Recently, a second-generation capsule endoscopy system (PillCam COLON 2) (CCE-2) has been developed to increase sensitivity for colorectal polyp detection [Adler and Metzger, 2011; Eliakim et al. 2009; Spada et al. 2011a].

The aim of this review is to analyze accuracy and safety data on CCE-2 from the literature.

PillCam COLON capsule endoscopy system

Compared with the first-generation CCE, the second-generation system includes new developments in the colon capsule (colon capsule 2 – CCE-2), data recorder and software for video processing and viewing (Figure 1). The new CCE-2 is 11.6 mm × 31.5 mm in size and has two imagers with a much wider angle of view that has been increased to 172° for each imager, allowing nearly 360° coverage of the colon. Furthermore, to enhance colon visualization and to save battery energy, the capsule is equipped with an adaptive frame rate. This represents a major progress over the previous version. CCE-2 captures 35 images per second when in motion and 4 images per second when it is virtually stationary. This advanced system for the control of capsule image rate is the result of bidirectional communication between CCE-2 and the new data recorder. Besides storing the images transmitted from the capsule, the data recorder also controls the capsule image rate in real time, analyzing the capsule images. To further save battery energy and to allow automatic identification of the small bowel, CCE-2, instead of going into a ‘sleep’ mode, continues to work at a low rate of only 14 images per minute until small bowel images are detected, then it switches into the adaptive frame rate. The new system is able to recognize when the capsule enters the small bowel. At this point, the data recorder buzzes and vibrates and displays instructions on its liquid crystal diode screen to alert the patient to continue the preparation protocol, assisting and guiding the physician and the patient through the procedure. Real Time Viewer is now incorporated in the new data recorder. Software (Rapid software, Given Imaging Ltd, Yoqneam, Israel) for video processing and viewing has also been implemented. The new software includes a tool for polyp size estimation. One can place the mouse cursor at one end of the polyp and drag the cursor to the other end. The software immediately calculates the distance and shows the dimension of the polyp in millimeters. How precise this system is in polyp size evaluation has not yet been assessed in patients.

Figure 1.

Second-generation colon capsule endoscopy system (CCE-2).

Regimen of preparation

The bowel preparation procedure is specifically designed for CCE [Eliakim et al. 2009; Spada et al. 2011a]. Similarly to colonoscopy, a clean colon is necessary to accurately explore the colonic mucosa, CCE being unable to suck or wash the mucosa. However, in contrast to conventional endoscopy, bowel preparation is also required to promote capsule propulsion, since the colon has only few spontaneous longitudinal contractions per day. For this reason, a unique bowel preparation has been developed. One or two boosters of sodium phosphate (NaP) have been added to the usual preparation of polyethylene glycol solution (PEG) adopted for colonoscopy. The main role of the NaP booster is to accelerate CCE transit through the small and the large bowel within the limited operating time of the CCE battery (i.e.10 h). For CCE-2, the regimen of preparation includes a clear liquid diet and 4 liters of split-dose PEG prior to capsule ingestion. Oral NaP boosters are also administered. In contrast to initial studies in which high doses of NaP were included, low doses of NaP are now included in the regimen of preparation for CCE-2 to reduce the risk of adverse events. In detail, one booster of 30 ml NaP with 1 liter of water is required when the capsule has entered the small bowel, and one booster of 15–25 ml NaP with 0.5 liter of water 3 h later if the capsule has not been egested by that time (Table 1).

Table 1.

PillCam COLON 2 preparation regimen [Eliakim et al. 2009; Spada et al. 2011a].

Day –2	All day	At least 10 glasses of water
	Bedtime	4 Senna tablets (12 mg each)
Day –1	All day	Clear liquid diet
	Evening	2 liters PEG
Exam day	Morning	2 liters PEG
	~10 am	Capsule ingestion*
	1st Boost	30 ml NaP and 1 liter water
	Upon small bowel detection
	2nd Boost	15–25 ml NaP and 0.5 liter water
	3 h after 1st boost
	Suppository	10 mg bisacodyl
	2 h after 2nd boost

^*20 mg domperidone tablet if capsule delayed in stomach >1 h.

NaP, sodium phosphate; PEG, polyethylene glycol solution.

Accuracy and safety

To date, two studies have evaluated CCE-2: an Israeli multicenter trial [Eliakim et al. 2009] and a European multicenter trial [Spada et al. 2011a]. In the five-center Israeli feasibility study, CCE-2 was prospectively compared with conventional colonoscopy as the gold standard for the detection of colorectal polyps and other colonic disease in a cohort of patients scheduled for colonoscopy and having known or suspected colonic disease [Eliakim et al. 2009]. Colonoscopy was independently performed within 10 h after capsule ingestion. A total of 104 patients were enrolled; data from 98 were finally analyzed. Patients were considered to have a significant finding when polyps at least 6 mm in size or masses were detected. Per-patient CCE-2 sensitivity for polyps at least 6 mm in size was 89% [95% confidence interval (CI) 70–97%], and at least 10 mm in size was 88% (95% CI 56–98%), with specificities of 76% (95% CI 72–78%) and 89% (95% CI 86–90%), respectively. The relatively low specificity observed in this series was mainly related to a high number of false-positive results (i.e. capsule detected a polyp that was not confirmed by colonoscopy). In nine patients CCE detected polyps at least 6 mm but less than 10 mm in size, which were not reported by colonoscopy for the same size category. In three of these patients the match was confirmed visually, although colonoscopy indicated a lower size category of less than 6 mm. In 10 patients CCE detected polyps at least 10 mm in size which were not reported in the same way at colonoscopy or not reported at all. In two of these patients, colonoscopy identified the polyps but with a smaller reported size of 6–9 mm. In one patient colonoscopy identified the lesion but it was described as a hemangioma. A further two patients underwent a second colonoscopy; in one case the capsule finding was confirmed, and in the second case the capsule finding was confirmed, although the second colonoscopy indicated a polyp size of 8 mm. The remaining five patients were invited for a second colonoscopy, but this had not been carried out at the time of publication. Colonoscopy combined with biopsy detected adenomatous polyps at least 6 mm in size in 11 patients (11% of patients), which included 5 patients with adenomatous polyps at least 10 mm in size, one of which was cancer. All of these polyps were detected by CCE.

Recently a European, prospective, multicenter trial including eight European sites was published [Spada et al. 2011a]. The CCE-2 was prospectively compared with conventional colonoscopy (as the gold standard) for the detection of colorectal polyps at least 6 mm in size or masses. A total of 117 patients were enrolled. Data from 109 patients were analyzed. Colonoscopy was independently performed within 10 h after capsule ingestion or, at the latest, on the next day. Per-patient CCE-2 sensitivity for polyps at least 6 mm in size was 84% (95% CI 74–95%), and at least 10 mm in size was 88% (95% CI 76–99%) with a specificity of 64% and 95%, respectively. In this series three cancers were detected by CCE-2 and were confirmed at colonoscopy. When restricting analysis to neoplastic lesions, 39 and 30 patients were seen with one or more adenomas at least 6 mm and at least 10 mm in size, respectively. CCE-2 correctly classified 35 and 28 of these patients, corresponding to a detection rate of 90% (95% CI 80–99%) for neoplasia at least 6 mm in size, and 93% (95% CI 84–100%) for neoplasia at least 10 mm in size. Similarly to the Eliakim series, in this study the low specificity for polyps at least 6 mm in size was explained by a substantial rate of false-positive polyps because of size mismatch. The authors reported that 20 (80%) of the 25 false-positive cases at CCE-2 (6–9 mm, 21 cases; ≥10 mm, 4 cases) were because of size mismatching (i.e. <6 or <10 mm polyp at colonoscopy measured as ≥6 or ≥10 mm at CCE-2). No polyp was detected at colonoscopy in the remaining five cases in which CCE-2 detected one or more polyps at least 6 mm in size. When considering colonic cancer, all three cancers detected by conventional colonoscopy were also visualized by CCE-2.

Colon cleansing level and excretion rate

Overall colon cleanliness for CCE was adequate in 78% and 81% of patients in the Israeli and European study, respectively. The natural capsule excretion rate (i.e. complete colonoscopy) was 81% and 85% within 8 h post ingestion in the Israeli and European study, respectively [Eliakim et al. 2009; Spada et al. 2011a]. A total of 88% of the capsules were naturally egested within 10 h [Spada et al. 2011a].

Safety

CCE is a safe procedure. In both studies, no adverse events related to CCE were reported. The adverse events reported were mainly related to bowel preparation. In the Israeli study, 8% of patients reported adverse events related to bowel preparation [Eliakim et al. 2009]. In the European study, eight mild to moderate adverse events were reported (6.8%). Five out of eight of these were related to bowel preparation and included vomiting, nausea, and abdominal pain. Two patients experienced fatigue because of the long capsule procedure. One patient experienced severe abdominal pain during conventional colonoscopy [Spada et al. 2011a].

Discussion

CCE-2 is an evolution of the previous colon capsule generation. The high sensitivity for clinically relevant lesions, in combination with the safety profile and the potential high acceptability, make CCE-2 a valuable option among diagnostic colonic techniques (Figure 2).

Figure 2.

Second-generation colon capsule endoscopy system (CCE-2) images. (a) Normal colonic mucosa; (b) hemorrhoids; (c) 12 mm polyp in the left colon; (d) 15 mm polyp in the left colon. The new software includes a tool for polyp size estimation. One can place the mouse cursor at one end of the polyp and drag the cursor to the other end. The software immediately calculates the distance and shows the dimension of the polyp in millimeters.

The Israeli and European multicenter trials showed comparable, promising sensitivity (84% for polyps ≥ 6 mm; 88–89% for polyps ≥ 10 mm) [Eliakim et al. 2009; Spada et al. 2011a]. This represents a big step forward when considering the findings of a large multicenter series published by Van Gossum and colleagues who reported their experience with the first generation of colon capsule (sensitivity for polyps ≥ 6 mm and ≥ 10 mm was 64% and 60%, respectively) [Van Gossum et al. 2009]. Several explanations might justify such an improvement in sensitivity. The better colon cleansing level observed in CCE-2 studies might be one possible explanation. In two previous studies, Van Gossum and colleagues and Spada and colleagues reported that colon cleanliness has a consistent impact on colon capsule accuracy parameters [Van Gossum et al. 2009; Spada et al. 2011c]. However, when comparing the cleansing level in the study by Van Gossum and colleagues [Van Gossum et al. 2009] (i.e. the first-generation study) with that in the two CCE-2 studies [Eliakim et al. 2009; Spada et al. 2011a] only a slight improvement in the cleansing level is observed (78% versus 81% of adequate cleansing level). This suggests that the impact of improved cleanliness may be marginal and, alone, cannot be responsible for the sensitivity improvement. Technology developments may explain such improvements. Among others, the adaptive frame rate and the wider viewing angle might explain the better performance of the new generation of colon capsule. With the adaptive frame rate, CCE-2 captures 35 images per second when in motion and 4 images per second when it is virtually stationary. This feature allows the enhancement of colon coverage, the higher acquisition rate potentially decreasing the risk of findings being missed in the colon. At the same time, when the capsule is stationary there is no need to acquire many images and the acquisition rate at 4 frames per second serves to conserve the energy of the battery. The other technical development that might explain the better performance of CCE-2 is the viewing angle of the imagers at each end of the capsule, which has increased to 172° (versus 156° in the first generation). When considering the two optical domes, CCE-2 nearly covers all of the mucosal colonic surface. These two changes appear to have produced a substantial improvement in the polyp detection rate of the new-generation capsule.

Although the sensitivity appears to be substantially improved compared with that of the first-generation CCE, specificity remains disappointingly low. The low specificity is explained by a substantial rate of false-positive polyps. This appears quite unusual when dealing with an endoscopic procedure, such as CCE, with virtually 100% specificity for polypoid lesions (irrespectively from histology) being reported with other endoscopic procedures, such as flexible sigmoidoscopy or colonoscopy. There are two possible explanations for the apparently low specificity of CCE-2: suboptimal sensitivity of the reference standard colonoscopy, or an imperfect polyp-matching algorithm. First, colonoscopy is an imperfect reference standard with a suboptimal sensitivity, especially for 6–9 mm polyps. When considering that in the Israeli and European studies [Eliakim et al. 2009; Spada et al. 2011a] post-CCE colonoscopy was performed without unblinding the CCE results, it cannot be excluded that apparently false positives from CCE were in fact false negatives from colonoscopy. To overcome this bias, unblinding of CCE-2 results at colonoscopy should be allowed in future studies. Second, polyp matching was based on polyp size in both studies. Although CCE-2 has a system for estimating the size of polyps, this system has never been proven to be accurate. In addition, polyp measurement during conventional colonoscopy cannot be regarded as accurate. Thus, CCE and colonoscopy might give inaccurate assessments of polyp size, so it is possible that the two techniques actually describe the same polyp, but the inaccuracy of the measurements prevents adequate matching, resulting in untrue ‘false-positive’ CCE results. In the European study, a substantial rate of false-positive results (i.e. 20 out of 25 cases) is related to a misclassification of polyps less than 6 mm in size at colonoscopy as at least 6 mm in size at CCE-2. Therefore, CCE-2 and conventional colonoscopy detected polyps that were differently measured. After unblinding, if these misclassified polyps are considered as true positive results, specificity for any polyp would be as high as 92% [Spada et al. 2011a]. The clinical relevance of this size mismatch is currently unknown, however it is likely to be marginal. In fact, it should be noted that a false-positive result because of a size mismatch is much different than a false-positive result with no polyp finding at post-CCE colonoscopy. Moreover, it should also be emphasized that the new polyp size estimation tool that is now embedded in the software was never validated and studies are needed to assess its accuracy and precision.

CCE has limitations. The regimen of preparation still remains an issue. The inclusion of NaP boosters in CCE may raise some concerns. At colonoscopy, NaP was associated with major complications, such as electrolyte disturbance, acute nephropathy, and kidney failure. The NaP–PEG combination in the preparation regimen for CCE could be a major barrier to CCE. Spada and colleagues evaluated whether the NaP booster could be substituted with a PEG booster [Spada et al. 2011c]. The authors demonstrated the crucial role played by the NaP booster when preparing patients for CCE, and the infeasibility of substituting it with a PEG booster. In particular, the PEG booster resulted in a more than twofold increase in the transit time, from 2 to 5 h, substantially reducing the excretion and completion rate of CCE. When passing from a NaP to a PEG booster, a 75% excretion rate was observed. This reduction in the completion rate (i.e. completeness of colonoscopy) is clinically meaningful for at least two reasons. First, it suggests that one in four patients undergoing CEE would not achieve a full colonic examination. When considering the cecal intubation rate of at least 95% required for screening colonoscopy, it seems unlikely that CCE would be recommended for screening with such a low rate of completion. Second, in contrast to colonoscopy, when the rectosigmoid tract, which is well known to harbor the majority of adenomas and carcinomas, is investigated even in incomplete examinations, an incomplete CCE might result in lack of visualization of the left colon. Therefore, a NaP booster is necessary for bowel preparation for CCE. To reduce the risk of NaP-related adverse events, lower doses (45–55 ml of NaP instead of 75 ml) were used in CCE-2 studies. The low dose of NaP does not hamper the CCE excretion rate, an adequate rate of 81–85% being observed within 8 h. This rate observed in the setting of a clinical trial probably underestimates the rate that might occur in a ‘real-life’ setting. In these trials CCE-2 was compared with conventional colonoscopy, which was performed immediately after capsule excretion or, at the latest, 8–10 h post ingestion. Because of logistical constraints, colonoscopy was performed no later than 8–10 h post capsule ingestion even if the capsule was still active. The battery life of CCE-2 is at least 12 h. Therefore, it could be presumed that a proportion of the 15–19% of patients who had not excreted the capsule would naturally excrete the capsule if it is left in the colon until battery exhaustion. Moreover, in the trials, by definition, CCE was considered complete if the capsule was naturally excreted. In a ‘daily’ setting, CCE might be considered complete if hemorrhoids are visualized, even if the capsule is not excreted within the capsule operating time.

Theoretically, all patients with suspected or known colonic disease, referred for a diagnostic, conventional colonoscopy are potential candidates for CCE examination. However, currently CCE is not an alternative but complementary to conventional colonoscopy when colonoscopy is incomplete, when it is contraindicated, or in patients who are unwilling to undergo the procedure [Spada et al. 2011b].

Finally a general comment seems advisable. To date, only two studies (involving more than 200 patients) have evaluated the performance of CCE-2. This is a clear limitation. However, it should be emphasized that these studies show comparable results in terms of accuracy, cleanliness, excretion rate, and safety, suggesting that they represent the actual CCE-2 performance.

Conclusion

In conclusion, the new generation of colon capsule is a safe, sensitive, noninvasive technique for colon exploration. When coupling this evidence with the feasibility, safety and tolerability of this technique, CCE may be considered an adequate tool to visualize neoplastic lesions arising from the colonic wall. To date, only two studies have evaluated the performance of CCE-2 and further prospective studies are needed to confirm these results