Capsule Endoscopy in the Detection of Non-Small Bowel Lesions Missed by Endoscopy

Background and aims: Capsule endoscopy (CE) is indicated in cases with obscure gastrointestinal bleeding (OGIB). However, lesions detected by CE are frequently within the reach of conventional upper or lower GI endoscopy. We evaluated the accuracy of CE in the study of OGIB, examining the incidence of CE-detected non-small-bowel lesions (NSBL) missed by conventional endoscopy and studying its impact on patient management. Methods: We retrospectively analyzed 2010 CE procedures performed in a tertiary-care center (IBD-Unit Referral Centre in Bologna), comparing the ndings on CE to those on prior colonoscopy and upper GI endoscopy performed within 3 months and two weeks of the CE procedure, respectively. We evaluated the impact of CE ndings on patient management. Results: CE revealed abnormal ndings in 1608 out of 2010 patients. Previously missed NSBLs were revealed on CE in 283 cases. Of these, 265 pre-CE endoscopic reports were found to not conform to reporting guidelines. NSBLs on CE led to management changes in 271 patients. Conclusion: This study conrms the utility of CE in patients with OGIB. However, in a considerable number of cases, CE identied lesions missed by conventional endoscopy, suggesting that a second look prior to CE may be appropriate in some patients.


Introduction
The European Society of Gastrointestinal Endoscopy (ESGE), the American Gastroenterological Association (AGA), and the American College of Gastroenterology (ACG) recommend that rst-line evaluation of the small bowel (SB) be performed using small bowel capsule endoscopy (CE) following upper and lower GI endoscopy, though in some scenarios a second-look endoscopy may be of use [1][2][3].
In addition, recent guidelines on the management of iron de ciency anemia (IDA) from the British Society of Gastroenterology (BSG) [4] suggest that examination of the small bowel be performed in cases involving symptoms indicating SB disease or with unsatisfactory response to iron replacement therapy. Evaluation of the small bowel can also be performed by cross-sectional imaging or enteroscopy, yet CE has a relatively high diagnostic yield and the bene t of being a minimally invasive endoscopic technique.
Frequently, lesions detected on CE are located in areas of the GI tract that are within the reach of a conventional endoscope. This suggests that in some cases a second-look upper or lower GI endoscopy may be indicated. Moreover, CE has been reported to reveal non-small-bowel lesions (NSBLs) missed upon prior conventional endoscopy with a detection rate ranging from 3.5% to over 30% [5][6].
Investigation by CE reportedly leads to the identi cation of a culprit lesion in the small bowel in roughly two-thirds of patients with obscure gastrointestinal bleeding (OGIB) [6][7][8][9][10][11][12][13]. The entire small bowel may be evaluated by CE in up to 90% of patients, and CE has a diagnostic yield of 38%-83% in cases of suspected small bowel bleeding [14]. The positive and negative predictive values of CE in the evaluation of GI bleeding are high (94-97% and 83-100%, respectively) [15][16], and ndings on CE lead to changes in patient management in 37-87% of cases [16][17]. The primary limits of CE include low speci city and a false-negative rate of 10-36%, along with lack of visualization of the major duodenal papilla in a substantial number of patients, potentially leading to lower accuracy in the identi cation of duodenal lesions [18][19][20][21][22].
The current study evaluated the diagnostic yield of CE in the investigation of OGIB and analyzed the incidence of CE-detected NSBLs that were missed at conventional endoscopy and its impact on patient management in a monocenter case series.

Materials And Methods
This single-center retrospective study analyzed data from patients undergoing CE in the IBD-Unit Referral Center in Bologna between February 2003 and February 2020. Procedures in which CE failed to reach the cecum before battery-life expiration were considered as incomplete and were excluded from the study.
We analyzed patient demographics, indications for CE, procedural data such as gastric and small bowel transit time, ndings of the examination, data regarding pre-CE upper and lower endoscopy, diagnosis and management pre-and post-CE, and patient outcome.
NSBLs were de ned as any abnormal CE ndings detected in the stomach, proximal SB, terminal ileum and colon.

CE procedure
After obtaining informed consent, all patients were given standardized instructions. Patients were instructed to stop iron supplementation seven days before the procedure and to start a diet low in ber three days prior to CE, followed by a fasting period starting at midnight before the procedure.
All patients received 4 L of polyethylene glycol (PEG) as bowel preparation, administered as 3 L the evening before the procedure and 1 L the morning of the procedure. The capsule (Pillcam SB2 and 3; Medtronic) was swallowed in the morning.
Patients were allowed to drink liquids after three hours and to consume a light meal after ve hours.
All CE procedures were reviewed by two gastroenterologists experienced in capsule endoscopy (CC, DG). Discrepancies in ndings were discussed in order to reach a consensus on the nal diagnosis. Findings on CE were documented and categorized using standard terminology.
Diagnostic yield was calculated as the number of procedures in which clinically signi cant lesions were identi ed divided by the total number of procedures performed.

De nitions
Endoscopy pre-CE: only colonoscopy and upper GI endoscopy (UGIE) performed within three months and two weeks before the CE procedure, respectively, was considered.
Additional ndings: gastric lesions (GL) and colonic lesions (CL) differing from those identi ed on previous endoscopy.
New ndings: novel GL and CL not detected by previous endoscopy (i.e., patients with negative ndings on previous endoscopy).
Clinical impact: percentage of patients in which CE ndings led to a diagnostic change.
Therapeutic impact: percentage of patients in which CE ndings led to change in treatment.

Evaluation of previous endoscopic report
Previous UGIE and colonoscopy reports were evaluated to ascertain their conformity to endoscopy reporting guidelines as a surrogate marker of the quality of the endoscopic examination itself.

Statistical analysis
Quantitative data following a normal distribution are presented as mean, standard deviation (SD) and range. Non-normally distributed quantitative data are presented as median and interquartile range. Qualitative data are presented as proportions, and comparison of qualitative data was performed using the Chi-squared test. P values < 0.05 were considered to be statistically signi cant. All statistical analysis was performed using IBM SPSS Statistics, version 20.0 (IBM Corp., Armonk, N.Y., USA).

Ethics
The study protocol was approved by the Comitato Etico Indipendente dell'AOU di Bologna Informed consent was obtained from all patients and all methods were performed in accordance with the relevant guidelines and regulations.

Patient demographics
During the study period, 2271 CE procedures were performed in 2242 patients.
Two CE procedures were excluded from analysis due to technical issues, and CE was considered incomplete in 230 cases (10.27%). As a result, 2010 procedures were included in the analysis; 1201 patients were male (58.75%) and the mean age was 61.5 ± 19.6 years (range 22-94 years).
1685 patients (83.8%) underwent the CE procedure within 10 days following rst level conventional endoscopic examination; 325 patients (16.2%) underwent the procedure between 8 and 28 days following rst level endoscopy. CE was performed the same day as colonoscopy in 32 patients (1.6%). Mean lag time between rst level endoscopy and CE was 7.8 days.
CE detected 17 lesions in the terminal ileum and 128 lesions in the colon that were missed at previous colonoscopy.
Fifteen patients presented lesions in both the stomach or duodenal bulb and the small bowel, and twelve patients presented lesions both in the small bowel and the colon.
Overall, 13.2% of endoscopy reports were found to not conform to reporting guidelines.

Discussion
In keeping with previous reports [27][28], our study corroborates the high diagnostic yield and safety of CE in the examination of the GI tract. CE demonstrated a diagnostic yield of 80% in our study population and identi ed clinically signi cant NSBLs not seen by conventional endoscopy in 16% of cases.
Approximately 49% and 51% of lesions were found in the upper and lower GI tract, respectively.
The difference in NSLBs found on CE with respect to other studies could be due to shorter time between endoscopy and CE as a general protocol adopted by our center. In fact, in our protocol we accepted only colonoscopy and UGIE performed three months and two weeks before the CE procedure, respectively.
Several studies have reported the ability of CE to detect lesions missed on upper and lower GI endoscopy. [5,[22][23][24][26][27][28]. It is unclear why these lesions are missed during initial conventional endoscopy, although possible explanations have been hypothesized. For example, some lesions may have characteristics that make them more di cult to identify, such as size or atypical location. In addition, factors relating to the endoscopic procedures themselves such as quality of the exploration, rate of exam completion or endoscopist experience may play a role [26]. Indeed, digestive endoscopy is an operatordependent procedure, and operator experience is an important factor in the detection of small lesions. Patient intolerance of the endoscopic procedure may also compromise the outcome of the examination. However, time of withdrawal is not always described in the endoscopy reports. In a signi cant portion of patients, endoscopy reports reported that the examination was not completed due to intolerance of the patient or the presence of solid feces, or quality of the endoscopic report is poor. Furthermore, the terminal ileum was not always explored for a number of reasons, including poor bowel cleansing, patient intolerance, or operator dependency of the examination. In addition, excessive air insu ation may also cause NSBLs to be missed as it causes at lesions to be harder to detect, and the hypotensive effects of sedative drugs may conceal angiodysplasias.
As upper and lower GI lesions are typically detectable using traditional endoscopy, in some cases an endoscopic second look may be warranted before proceeding with CE. Previous studies have reported the diagnostic yield of a second look as 35-75% for the upper GI tract and 6% for the lower GI tract [5,[26][27][28]. Although there is currently a lack of a strong recommendation, we believe that the high percentage of NSBLs suggests that a second look endoscopy is of value in patients with unreliable rst level exams, particularly in patients with persistent bleeding. Furthermore, in the event of poor intestinal preparation, patient intolerance, or incomplete examinations, rst level procedures should be repeated before proceeding to CE examination.
One of the main limitations of the current study is its retrospective design, and a prospective follow-up is necessary in order to con rm that the detected lesions are not incidental ndings.
Moreover, in the absence of a clear marker of endoscopy quality, we selected the conformity of the endoscopy reports to reporting guidelines as a surrogate, as the quality of the endoscopic report is directly linked to the quality of the endoscopic examination itself.
In conclusion, our study con rms the importance of capsule endoscopy in the diagnostic work-up of obscure gastrointestinal bleeding. However, the study reveals that in a considerable number of patients, capsule endoscopy identi es lesions within the reach of conventional endoscopic techniques. This suggests that capsule endoscopy may be unnecessary in some cases, in particular when rst level endoscopy is unreliable. The strengthening of rst level endoscopic procedures and the judicious use of conventional endoscopy as a second look could lead to a larger percentage of early diagnoses, with a substantial economic impact.