Pathology in the gastrointestinal tract causing anaemia- lesion location and the role of capsule endoscopy.

Iron deficiency anaemia (IDA) is a common clinical problem in the adult population. The
prevalence of pan enteric pathology in IDA is unknown. Oesophagogastroduodenoscopy
(OGD) is the gold standard examination, but it can be poorly tolerated and often requires the use of sedation. Meanwhile capsule endoscopy allows us to image the upper gastrointestinal tract (GI) and the small bowel in one sitting and it is well tolerated. The demand for capsule endoscopy is increasing, but evaluation capsule endoscopy videos is time consuming and requires the readers ongoing attention. Preliminary data has shown that the use of artificial intelligence (AI) in capsule endoscopy can significantly reduce reading times. With the growing demand for capsule endoscopy there is a need for biomarkers which can help risk stratify capsule endoscopy referrals. This thesis examines the role of capsule endoscopy in patients with IDA and also evaluated the role of faecal immunochemical test (FIT) for risk stratifying patients for small bowel capsule endoscopy.
As part of this thesis, a multicentre prospective study of 170 patients was carried out which investigated the prevalence of pan enteric pathology in patients with IDA. The diagnostic yield of lesions responsible for blood loss on OGD, capsule endoscopy and colonoscopy were 27.6%, 49.4% and 20% respectively (p <0.0001).
We also compared the use of magnetically controlled capsule endoscopy (MACE) with OGD in the assessment of the upper gastrointestinal tract prospectively in 158 patients with IDA. This study showed that MACE identified significantly more lesions compared with OGD in the upper GI tract 119 vs 57 (95% CI 0.0- 0.08; P <0.001). No adverse events were recorded and there was no capsule retention.
In the setting of IDA and the small bowel, a multicentre prospective study of 128 patients evaluated the role of AI- assisted reading in small bowel (SB) capsule endoscopy. In this study AI- assisted SB reading significantly reduced the reading time [ mean standard read 18.13 min (SD 8.34) vs AI assisted SB reading time 2.27 (SD 2.05) (p<0.0001)]. The per patient diagnostic yield between standard read and AI assisted read was non-inferior [48.4%, 95%CI (40.0- 57.0) vs 53.9%, 95%CI (45.3- 62.3)].
The role of FIT in the detection of small bowel pathology is yet to be defined. In this single centre prospective study the role of FIT in predicting small bowel pathology in 79 patients with IDA was evaluated. We did not find that FIT conferred any additional benefit in the detection of small bowel pathology in patients with IDA, however the diagnostic yield of small bowel capsule endoscopy in patients with IDA is high at 40.2%.
In conclusions this thesis supports the use of MACE in the evaluation of the upper GI tract and small bowel in patients with first presentation and /or refractory IDA. It also supports the role of artificial intelligence in SB capsule endoscopy. There is a pressing need for biomarkers to help triage SB capsule endoscopy referrals, however FIT is not a reliable biomarker for the presence of SB pathology in patients with IDA.