Toward an expanded role for video capsule endoscopy in patients with gastrointestinal bleeding: A review of the literature

Abstract

The role of small bowel capsule endoscopy (CE) in suspected small bowel bleeding is well established and current European and other international guidelines position it as the first line test after negative bidirectional endoscopies. In suspected mid-lower gastrointestinal bleeding (MLGIB) the diagnostic yield of colonoscopy is poor and may cause a delay in detecting small bowel disease. A crucial aspect of small bowel capsule endoscope performance is the timing of the procedure, the interval between the bleeding episode and capsule ingestion is inversely related to the diagnostic yield as confirmed by a recent meta-analysis. Currently the ongoing advances achieved by video CE in particular to evaluate both small bowel and colon in a single test using double headed capsules, raises questions regarding the position of pan intestinal capsule in the current algorithm to investigate patients presenting with suspected MLGIB ahead of colonoscopy. Early evidence suggests pan intestinal capsule could fit well as a diagnostic “filter” test in this cohort of patients, thereby reserving invasive conventional colonoscopy or device assisted enteroscopy as therapeutics options only. The recent development of magnetically controlled CE and a blood sensing capsule push the boundaries of CE even further in patients presenting with suspected gastrointestinal bleeding. This review will discuss the current available evidence and future directions of CE in suspected MLGIB.

Key Words: Pan-intestinal capsule endoscopy; Suspected mid-lower gastrointestinal bleeding; Small bowel; Colon; Intestinal preparation; Artificial intelligence

Core Tip: This review comprehensively discusses the emerging role of pan intestinal capsule in patients with gastrointestinal (GI) bleeding. With recent technical advancements capsule endoscopy have enabled to potentially assess the entire GI tract mucosa in a single, minimally invasive procedure. This will contribute to optimize the diagnostic work up of GI bleeding patients by reducing the need for multiple investigations and time to diagnosis.

INTRODUCTION

Capsule endoscopy (CE) is a minimally invasive wireless device capable of acquiring high quality video of the entire digestive tract mucosa. This makes CE particularly useful for examining areas that are difficult to reach with conventional endoscopy. The gastrointestinal (GI) tract is approximately 9 meter long from mouth to anus with the small intestine making up about two-thirds of the entire GI tract. The small bowel is subject to diseases including but not limited to bleeding, ulcers, malabsorption, inflammation, strictures, polyps, and cancer etc. So far, the most common application of CE has been in identifying a source of bleeding within small bowel. Obscure GI bleeding was traditionally defined as bleeding of unknown origin that persists or recurs after conventional upper endoscopy and ileocolonoscopy[1]. Recently, newer guidelines have proposed the term should be reclassified as suspected small bowel bleeding (SSBB) once bidirectional endoscopies are negative[2].

Single-camera small bowel video CE (SBCE) was introduced in 2001, enabling examination of the small bowel mucosa. This device allowed for minimally invasive examination of the small intestine in patients presenting with SSBB. SBCE allows for non-invasive evaluation of patients with SSBB with a high diagnostic yield and is helpful in selecting the insertion route for balloon assisted enteroscopy[3,4]. The effectiveness of SBCE is highly time sensitive particularly in patients with overt bleeding. A recent systematic review with meta-analysis has found that the diagnostic yield of SBCE remains high, 80.35% when performed within 48 hours from the last bleeding episodes as compared to more than 72 hours after presentation[5].

The role of SBCE is well established in investigation of SSBB after negative bidirectional endoscopies. However recently, the use of pan-intestinal CE (PCE) has been evaluated for suspected mid to lower GI bleeding after negative gastroscopy[6]. The mid-lower GI bleeding (MLGIB) refers to bleeding that originates distal to the ligaments of Treitz and includes sources within the small intestine as well as colon and rectum. The current European and other international guidelines position CE as the first line test only after negative bidirectional endoscopies. However, the diagnostic and therapeutic yields of colonoscopy in patient presenting with iron deficiency anemia with or without overt bleeding is quite low[7]. PCE may appear as a possible gamechanger in the setting of suspected mid to lower GI bleeding when a gastroscopy is negative, as a means of non-invasive and adequately timed pan-intestinal evaluation. This approach may also guide subsequent management, depending on the type and location of any potential haemorrhagic lesions (PHLs) avoiding further unnecessary invasive investigations including colonoscopy[6]. As clinicians we are expected to make a correct diagnosis in a timely and least invasive manner, PCE may well fit these principles in a single test to investigate patients with suspected MLGIB. In this review we will discuss recent evidence addressing the role of PCE in suspected MLGIB.

PAN INTESTINAL CAPSULE ENDOSCOPY USE IN PATIENTS WITH GASTROINTESTINAL BLEEDING: AVAILABLE EVIDENCE

As mentioned before, the role of SBCE in SSBB is well established, however role of PCE in patients with MLGIB remains limited. To date only three studies have been published, two prospective and one retrospective study. These studies will be reviewed in detail below (Table 1).

Table 1 Pan-intestinal capsule endoscopy for suspected mid-lower gastrointestinal bleeding the evidence, n (%).

Ref.	Year	Study design	Number of patients	Pan-intestinal capsule diagnostic yield (%)	Potentially saved colonoscopies
Mussetto et al[6]	2021	Prospective	12	83	6 (50)
Carretero et al[8]	2021	Retrospective	67	61	46 (69)
Rosa et al[10]	2024	Prospective	100	46	72 (72)

Carretero et al[8] performed a retrospective observational study in 2021 which included 67 high risk patients with suspected GI bleeding who had pan-endoscopic capsule procedures. Of which 46/67 had occult bleeding and 21/67 had overt bleeding. These patients ingested PCE using PillCam Colon 2 system. PillCam Colon 2 has two cameras, each one with an angle view of 172 degree. It weighs approximately 3.4 g and dimensions includes 11.6 mm by 31.5 mm. It has an adaptive frame rate (AFR) that becomes active once the capsule automatically detects its location in the small bowel. The AFR ranges from four frames per second (when the capsule is stationary or circulates at a very low speed) to 35 frames per second when capsule is in motion. Once the capsule is activated, it starts blinking at four frames/second and three minutes later, the frame rate decreases to 14 frames/minute to save battery life. For PCE purpose, AFR was manually activated in this study. Therefore, the whole gut can be explored at 4-35 frames/second, offering the reader images from mouth to the anus. The bowel preparation involved a split-dose regimen of 4 L of polyethylene glycol (PEG) and prokinetic administration with sodium phosphate compound as a booster. The day before examination, all patients had two liters of PEG from 19:00 to 21:00; then, the day of examination the patient took two additional liters of PEG before capsule ingestion. Additional laxatives (boosters) were required to let the capsule reach the anus and keep the colon clean throughout the exploration. The 30 mL of sodium phosphate for the first boost and 15 mL of sodium phosphate for the second boost was used. The first boost was given once the capsule had entered the small bowel; the second boost was given three hours later, if the capsule had not yet been excreted. PCE identified colonic findings in 21/67 (31%) which were considered a bleeding source. However, only 14/67 (21%) required therapeutic colonoscopy. In 28/67 (42%) patients, small bowel findings were considered as the bleeding source. This would suggest that performing a PCE would have avoided 79% of colonoscopies. The complete procedure rate in this study was only 69%, which is lower than previously reported[8]. A lower proportion of complete studies may have been explained by a higher number (70%) being inpatients at the time of their PCE. Our group has previously shown that inpatient status negatively impacts both colon and pan-intestinal capsule completion[9]. Nevertheless, capsule was able to identify potential bleeding source in 70% of these incomplete studies. This study had multiple limitations including retrospective nature, poorly defined inclusion or exclusion criteria and no comparison with standard diagnostic procedure. However, despite these this study highlighted potential role of PCE in IG bleeding.

Mussetto et al[6] published a pilot proof-of-concept study the same year in 2021 which tested the feasibility and diagnostic performance of PCE in acutely bleeding patients presenting with melena and negative esophagogastroduodenoscopy. In this study patients were included if they presented with melena (defined as the passage of dark tarry stools confirmed by medical or nursing staff or discovered by rectal examination) and signs of severe bleeding (defined as hemodynamic instability - mean arterial pressure < 65 mmHg and/or heart rate > 110/minute - or drop of haemoglobin level > 2 g/dL/day), or both. All included patients had negative esophagogastroduodenoscopy within 24 hours from admission. Patients who were included in the study to receive PCE were given high volume split dosing regimen of polyethylene-based preparation, similar to that described in the study by Carretero et al[8]. Over the 12 months of the study 23 patients who presented with melena had negative esophagogastroduodenoscopy. Only 12 patients were included in study after 7 patients refused hospital admission and four patients were sent back to the referring hospital where PCE was not available. PCE allowed complete small bowel evaluation in all patients and colonic evaluation in 11 (91.7%) patients. The small bowel and colon cleansing were judged as adequate in 100% and 83.3%, respectively. No capsule retention, no technical or recording problems were reported. PCE found source of bleeding in 10 patients (10/12; 83.3%), with clinically significant findings in small bowel, colon or both in 5 (41.7%), 4 (33.3%) and 1 (8.3%) patient, respectively. In 10 patients with positive PCE, six patients underwent double-balloon enteroscopy and three patients required colonoscopy. Among the two patients with PCE both had negative outpatient colonoscopy and small bowel-computed tomography. None of the patients experienced rebleeding at 30 days and 12 months rebleeding rate was 18.1%. This prospective, proof-of-concept study showed PCE was feasible and a safe procedure when performed in patients admitted with GI bleeding, after a negative esophagogastroduodenoscopy. This study had a very high diagnostic yield (83.3%) possibly related to favorable timing of the PCE (< 72 hours from presentation) and the highly selected patient cohort with overt bleeding. PCE guided further management avoided invasive colonoscopy in 50% of patients. Although there were multiple limitations of this study as it was a proof-of-concept study, this study showed a PCE based approach in patients with suspected MLGIB after negative esophagogastroduodenoscopy is feasible, safe and has a high diagnostic yield. This study brough new innovative insight into the management of suspected MLGIB with PCE.

More recently Rosa et al[10] in 2024 analyzed the role of PCE in patients suspected with MLGIB in a prospective, single center, single blinded study. Unlike the study by Mussetto et al[6] this study included patients both with anemia (occult bleeding) and overt bleeding such as melena. PCE was also followed by same day conventional colonoscopy enabling a direct comparison between the two approaches. This study did not did not provide details on bowel preparation for PCE or the specific capsule used. A total of 100 consecutive patients with suspected MLGIB who presented with iron deficiency anemia (81%) or overt bleeding (19%) were included. Only 16% underwent PCE and colonoscopy within 72 hours, however, 74% had both PCE and colonoscopy within 14 days of presentation. PCE was complete in 76% compared with a 95% completion rate for colonoscopy. Combining the results of PCE and conventional colonoscopy, PHLs were diagnosed in 46% of patients. Overall, small bowel PHLs were detected in 25% and colonic PHLs in 28% of the entire cohort. In 54% of patients, no PHLs were detected by either procedure. PCE detected PHLs in 44% of the entire cohort, corresponding to 95.7% (44/46) of all PHLs. Colonoscopy only detected PHLs in 23% of the entire cohort, corresponding to 50% (23/46) of all patients with PHLs. While these results show a clear overall detection superiority for PCE over colonoscopy, on closer observation of ileocolonic findings only, PCE diagnosed 25/28 (89.3%) as compared to colonoscopy diagnosed 23/28 (82.1%). Interestingly, approximately half of patients in this study had entirely normal examinations (no PHLs). In addition, less than one-third of patients with suspected MLGIB had colorectal PHLs. Therefore, the author’s concluded that in this cohort of 100 patients with suspected MLGIB, a PCE first approach would have saved 72 invasive colonoscopies and only costed 16 more capsules. However, in a real-world clinical setting, 47 colonoscopies would have been saved, considering there were 24 incomplete PCEs. Albeit some may have only required a left colonoscopy or sigmoidoscopy. Overall, this study showed PCE is safe, more effective than conventional colonoscopy in identifying lesions with bleeding potential in both the small bowel and the colon and therefore should be considered as a possible first line examination in patients with suspected MLGIB.

DISCUSSION

As demonstrated by the above three recent studies PCE seems to be feasible, effective, and safe in patients presenting with suspected MLGIB with promising results. PCE offers the unique opportunity to evaluate the entire GI tract in a single non-invasive examination. Thereby fulfilling the philosophy behind this technology when it was first introduced in 2000 which was named the M2A capsule which stands for mouth to the anus. Earlier limitations that had initially prevented the examination of the entire surface of the GI tract have been progressively addressed over time through innovations such as extended battery life, optimization of power consumption and increase in number of cameras with a wider angle of view. Although we have overcome technical barriers however some clinical barrier still persists including the degree of cleanliness, completion rate, and the time needed to process and read PCE images.

Firstly, in conventional colonoscopy, there is a well-documented correlation between the degree of cleanliness and diagnostic performance[11]. This is even more relevant in PCE, where the capsule is purely passive (i.e., it cannot clean the mucosa or remove residue). Despite large volume bowel preparation and use of prokinetic agents the complete examination rate in the above studies varied from 69%-92%, leaving a portion of the GI mucosa inadequately or not assessed at all. These limitations effect PCE practicality and feasibility in clinical practice. However, these limitations may not be as negative as first perceived. Cleanliness in colonoscopy in the acute setting of bleeding in hospitalized patients may also be a challenge. It is well established that colonoscopy bowel preparation and completion rates are worse in an inpatient setting[11-13].

In addition, the real benefit of a readily accessible, (not dependent on endoscopy units or endoscopists), non-invasive filter test, could be to simply identify if there is in fact active bleeding, (no source was identified in 54% in one study) and where it is happening in the GI tract, there by guiding further targeted intervention. As opposed to a target of detecting and characterizing all PHL’s, while beneficial if it does, would be an added benefit rather than a requirement. While identification of other pathology such as diminutive polyps irrelevant to the bleeding presentation, is simply unnecessary and beyond the remit of a PCE in bleeding. As such our objectives in future studies need to be clearly defined to truly assess its potential efficacy.

Secondly, despite PCE’s clinical feasibility and minimal invasive features, an important limitation to the applicability of PCE in daily practice is the substantial time required to review images acquired from both small bowel and colonic mucosa[14]. This could potentially be overcome through integration of Artificial intelligence (AI) models to reduce review times and obtain accurate diagnosis without missing any lesions. Preliminary reports suggest that AI, in particular deep convolutional neural networks, are able to efficiently identify specific lesions among a large variety, exceeding human performance in visual tasks. AI algorithms are trained on large annotated image datasets to recognize and classify abnormalities such as angiodysplasias, ulcers and erosions, tumor and polyps, active bleeding and inflammatory bowel disease. A study by Ding et al[15] reported that a deep learning system achieved a sensitivity of 99.88% and reduced review time to 6 minutes for full CE studies. AI can also filter redundant or normal frames, drastically reducing the number of images that require review by a human reader. We have recently published our real-world clinical experience with AI assisted small bowel capsule reading for all indications, which had a high diagnostic accuracy for expected pathology, with only atypical findings such as an endoscopic clip being overlooked[16]. Similarly, a recent prospective, multi-center study by Spada et al[14] on the use of AI assisted CE reading in SSBB showed AI assisted reading might provide more accurate and faster detection of clinically relevant small bowel bleeding lesions than standard reading, significantly reducing reading time , from 33.7 minutes to 3.8 minutes while maintaining similar sensitivity and specificity for bleeding lesions. There is no doubt that AI will revolutionize CE in the near future assisting physicians to rapidly identifying pathology. However, we have to keep in mind that AI based systems are developed to support the physician who is still responsible for deciding what are relevant findings and what further management are required.

Taking all these issues into account, one might wonder if pan intestinal CE evaluation is primed to be used as a first line investigation in suspected MLGIB. In support of PCE use, it provides the distinct advantage to evaluate several segments of the GI tract at the same time, in a single examination. It can confirm active bleeding and refine its location in the acute setting, while potentially identifying and classifying the source of occult bleeding in the elective setting, all the while reducing the need for colonoscopy. Access is another important potential advantage of PCE, when timing of the procedure is crucial in patients presenting with GI bleeding and its impact on diagnosis. Furthermore CE, readily accepted by patients when compared to other conventional endoscopy procedures, especially in term of safety and comfort, also has the advantage of being performed comfortably without the need for sedation, which particularly in the acute setting of bleeding, would also be beneficial[17].

CONCLUSION

In conclusion, while not ready for primetime, there is sufficient evidence now to warrant further large-scale studies into this novel approach to suspected MLGIB.