Small-bowel capsule endoscopy: A ten-point contemporary review

Table 7 Studies looking at the clinical validity of Suspected Blood Indicator, feature of capsule endoscopy reading software, in small-bowel capsule endoscopy

Ref.	Country	Centre	Objective(s)	Study type	Design	CE type	Outcome(s)	Conclusions
Gross et al[96]	United States	Single centre	Accuracy of SBI to number of blood transfusions	Retrospective	►Gold standard for lesions detected by experienced CE reviewer	M2A; Given® Imaging Ltd.	►Gold standard: 72 pts;	Pts receiving blood transfusions are more likely to have a positive SBI correlating with the localization of active bleeding
							►pts received blood transfusions ranging between 0-16 units;
							►Overall: A total of 17 pts had positive SBI. Active bleeding in 16 pts, who were transfused an average of 8 units before the study;
							►55 pts had a negative SBI and no active bleeding was seen on their capsule studies. In this group, the average number of PRBC transfused was 1 unit. There was one patient who had a false positive SBI with no active bleeding seen in the capsule study review
Liangpunsakul et al[97]	United States	Single centre	Assess accuracy of SBI	Retrospective	►Gold standard for lesions detected by experienced CE reviewer;	M2A; Given® Imaging Ltd.	►Gold standard: 109 lesions;	SBI has good Sens and PPV for actively bleeding SB lesions
							►SBI: 31 potential areas of blood; correctly identified lesions: 28;
					►Significant lesions considered AVMs, ulcers, erosions, active bleeding;		►Overall: SBI (Sens, PPV, accuracy): 25.7%, 90%, 34.8%, respectively;
							►For actively bleeding SB lesions only: SBI (Sens, PPV, accuracy): 81.2%, 81.3%, 83.3%, respectively
					►Reviewing speed: 15fps
D'Halluin et al[98]	France	Multi-centre (7 centres)	Assess Sens/Spec of SBI (in OGIB)	Retrospective	►Gold standard for lesions detected by experienced CE reviewer, SBI tags marked by another investigator;	M2A; Given® Imaging Ltd.	►156 SBCE recordings evaluated: In 83 (normal): either no lesion (n = 71) or P0 lesion (n = 12); in 73 abnormal: P2 (n = 114) and P1 (n = 92) lesions;	►SBI-based detection of SB lesions (with bleeding potential) is of limited clinical value
					►Significant lesions considered Bleeding or having a bleeding potential: high (P2), low (P1), or absent (P0);		►154 red tags analysed: SBI (Sens, Spec, PPV, NPV) for P2 or P1: 37%, 59%, 50%, 46%, respectively
					►Concordance: same time code in frames selected by expert reader and those tagged by SBI;
					►Reviewing speed: NS
Singnorelli et al[99]	Italy	Single centre	Assess Sens/Spec of SBI per lesion, overall, according to red findings (identified by the reader), and per patient	Retrospective	►Gold standard for lesions detected by four experienced CE reviewers;	M2A; Given® Imaging Ltd.	►95 patients; 209 red findings;	►SBI has low Sens/Spec in per-lesion and per-patient SBCE evaluation;
							►Overall Sens: 28%;
					►Outcomes: Sens, Spec and accuracy calculated both per lesion/patient;		►Sens higher for identification of blood (61%) than for nonbleeding “red” findings, e.g., AVMs (26%);
					►Reviewing speed: NS		►Per-patient Sens, Spec: 41%, 70%, respectively	►Complementary/rapid screening tool;
								►Complete review of the recordings is still necessary
Ponferrada et al[100]	Spain	Single centre	Assess accuracy/performance of SBI	Prospective	►Gold standard for lesions detected by experienced CE reviewers	M2A; Given® Imaging Ltd.	►57 consecutive patients;
							►Indications: OGIB (64.9%), CD (14%), malabsorption (14%), suspicion of SB tumour (7.1%);
							►SBI Sens, Spec, PPV, NPV: 58.3%, 75.5%, 38.8%, 87.2%, respectively
Buscaglia et al[101]	United States	Single centre	Assess accuracy/performance of SBI according to CE indications	Retrospective	►Gold standard for lesions detected by experienced CE reviewer;	M2A; Given® Imaging Ltd.	►CE indications: OGIB (n = 112), suspected CD (n = 122), anaemia of unknown origin (n = 53), other (n = 4);	►SBI performance characteristics suboptimal/insufficient to screen for SB lesions with bleeding potential;
					►Significant lesions:AVMs, varices, venous ectasias, red spots, ulcers, erosions, blood, blood clots		►221 lesions with bleeding potential;
							►Overall: SBI (Sens, Spec, PPV, NPV): 56.4%, 33.5%, 24.0%, 67.3%, respectively;
					►Concordant and discordant findings between CE reviewer and SBI;
							►For actively bleeding lesions: SBI (Sens, PPV): 58.3%, 70%, respectively;	►Even in pts with active intestinal bleeding, SBI Sens was only < 60%
					►Reviewing speed: 8-15 fps		►For suspected CD: SBI (Sens, NPV): 64%, 80.4%, respectively;
							►For OGIB: SBI Sens 58.3%;
							►For anaemia: SBI Sens 41.3%;
Park et al[102]	South Korea	Single centre	Investigate whether SBI is affected by background colour and CE velocity	Experimental	►Paper-made phantom SB models in a variety of colours to simulate the background colours observed in CE;	M2A; Given® Imaging Ltd.	►SBI red spots detection rate differed significantly per background colour of SB model, P < 0.001;	►SBI Sens affected by background colour and capsule passage velocity in the models
							►SBI red spots detection rate highest for very pale magenta, burnt sienna, yellow background;
					►Red spots were attached inside them;
					►CE manually passed through models;
					►SBI red spots detection rate was evaluated based on colours of SB models and CE velocities (0.5, 1, 2 cm/s)		►SBI red spots detection rate lowest for dark brown, very pale yellow background

PRBC: Pack red blood cells; fps: Frames per second; SBI: Suspected Blood Indicator; CE: Capsule endoscopy; AVM: Arterio-venous malformations; SB: Small-bowel; Sens: Sensitivity; Spec: Specificity; PPV: Positive predictive value; NPV: Negative predictive value; OGIB: Obscure gastrointestinal bleeding; CD: Crohn’s disease.