INTRODUCTION
The severe acute respiratory distress syndrome coronavirus 2 (SARS-CoV-2) emerged as the cause of coronavirus disease 2019 (COVID-19) in December 2019 and was declared a global pandemic by the World Health Organization (1). The number of global infections continue to rise with an excess of 2,5 million infections and over 75 000 deaths being recorded in South Africa by August 2021 (2). Whilst COVID-19 predominantly manifests with respiratory symptoms (3), gastrointestinal (GI) symptoms are seen in up to half of the patients presenting to hospital (4). The most common gastrointestinal symptoms are anorexia, diarrhoea, nausea, vomiting and abdominal pain (4–6). Furthermore, hepatic involvement evidenced by elevated transaminases has been documented in up to 51% of infected patients (7). Angiotensin converting enzyme 2 (ACE2) receptors have been identified in gastric, duodenal and rectal glandular cells but are much less frequently observed in oesophageal mucosa (8). These receptors likely facilitate viral cellular entry and the subsequent pathogenic sequelae.
Most UGIB occurr during hospitalization and is not the primary reason for presentation amongst COVID-19 patients (9). The prevalence of UGIB amongst COVID-19 patients varies between centers with rates between 0.1% and 1.8% being documented (10,11). This variation in prevalence is likely influenced by differing definitions of UGIB (clinical, anatomical and endoscopic), differing ulcer prophylaxis practices, the presence of underlying comorbidities and the severity of illness from COVID-19. Furthermore, the true number of patients with GI bleeding is likely underrepresented due to changes in endoscopic departmental protocols and increased thresholds to perform endoscopy with more conservative therapy being practiced during the pandemic (12,13). Most regional guidelines advocate for the stratification of bleeding as variceal or non-variceal and the subsequent performance of upper endoscopy within 24 hours of non-variceal upper GI bleeding (14–16). The need for and timing of endoscopy, however, needs to be individualized by a risk/benefit analysis and can be guided by utilizing risk scores such as the clinical Rockall score or the Glasgow-Blatchford score (GBS)(17). The South African Gastroenterology Society (SAGES) proposed a guideline on safe endoscopy during the pandemic with regards to pre and post procedure practices, risk stratification of procedures and the level of personal protective equipment (PPE) needed to ensure safe endoscopy (18). Given the aerosol generating nature of upper endoscopy, critical shortages of PPE in some institutions need to be taken into account in the decision-making process. In addition, patient safety with the risk of peri-procedural respiratory decompensation secondary to anaesthesia, aspiration and the procedure itself, necessitates the need for careful patient selection. A deterioration in cardiorespiratory function both peri- and post-procedure may necessitate endotracheal intubation and mechanical ventilation. This may place further constraints on an already limited number of intensive care beds and ventilatory supportive equipment.
CASE SERIES
Between the period of the 1st March 2020 and the 31st October 2020, 5 COVID-19 patients (confirmed by a positive reverse transcription polymerase chain reaction test) had a gastroduodenoscopy performed for suspected UGIB. Their clinical characteristics and endoscopic findings are depicted in tables 1 and 2 respectively. Data collection and publication of the data had Institutional Board approval. Only 1 patient was admitted with a presenting complaint of passing melaena stool (case 2). All other patients were admitted on the basis of respiratory symptoms and subsequently developed UGIB during their hospitalization. No deaths were recorded during hospitalization and no repeat endoscopic interventions were needed. The high GBS noted in case 4 was driven by the presence of end stage renal disease.
Patient characteristics
| Case 1 | Case 2 | Case 3 | Case 4 | Case 5 | |
|---|---|---|---|---|---|
| Age | 70 | 74 | 71 | 56 | 53 |
| Gender | Male | Male | Female | Male | Male |
| Comorbidities | Hypertension, renal mass | Non-cirrhotic portal hypertension, prostate cancer | Peripheral vascular disease, new onset diabetes | HIV, hypertension, end stage renal disease (prior renal transplant with poor graft function), previous DVT, new onset diabetes | Epilepsy, cirrhosis |
| Chest X-ray | Normal | Normal | Infiltrates | Infiltrates | Infiltrates |
| Renal dysfunction | Yes | No | No | Yes | No |
| Anti-coagulation | LMWH 80u daily | No | LMWH 60u bd | Warfarin (prior to admission) | LMWH (dose N/A) |
| Corticosteroids | Hydrocortisone 50mg qid | Nil | Yes (dose N/A) | Hydrocortisone 100mg tds | Prior prednisone use 30mg daily |
| Oxygen requirements | Rebreather mask | Nil | Mechanical ventilation | Rebreather mask | Rebreather mask |
| Glasgow-Blatchford score | 11 | 7 | 2 | 12 | 10 |
(LMWH - low molecular weight heparin, N/A - not available)
Endoscopic findings and interventions
| Case 1 | Case 2 | Case 3 | Case 4 | Case 5 | |
|---|---|---|---|---|---|
| Endoscopic findings | Large duodenal ulcer – Forrest IIb | Large oesophageal varices with high risk stigmata, portal hypertensive gastropathy | Oesophageal candidiasis, oesophagitis, erythematous pangastritis, 2 small gastric angiodysplastic lesions | Oesophageal candidiasis | Small oesophageal varices without high risk stigmata, erythematous pangastritis, portal hypertensive gastropathy, duodenitis |
| Endoscopic intervention | Adrenaline injection, clot dislodged with an underlying Forrest IIc ulcer | Endoscopic variceal ligation | Nil | Nil | Nil |
DISCUSSION
Although the exact mechanisms underlying gastrointestinal mucosal injury in COVID-19 patients remain unknown, a variety of mechanisms have been postulated. A meta-analysis has shown the presence of viral RNA in 48.1% of stool samples from COVID-19 patients, with the majority being isolated after viral loss in respiratory samples. Furthermore, viral RNA levels were higher and more commonly found in the stool of patients with symptomatic diarrhoea (5). Whether this translates into the potential for infectivity via stool remains unknown but has been proposed (19). Pathogenetically, viral entry into cells has been shown to occur via ACE2 receptors. Viral binding to ACE2 receptors in the GI tract may lead to its downregulation with a subsequent decrease in tryptophan and ultimately lower levels of niacin in mouse models. This may result in alterations in GI microbiota which may be contributory to the development of diarrhoea (20). It is further speculated that direct tissue damage and an increased number of proinflammatory mediators in ACE2 cells may contribute to mucosal damage and subsequent bleeding (20). Not all patients with GI symptoms, however, have detectable viral RNA in stool thus leading to the concept that direct viral effects are not the only contributing mechanisms to GI symptomatology (5). A bidirectional effect of bacterial translocation and cytokine mediators between the gut-lung axis and gut-liver axis has been proposed (20). In addition, the role of antibiotic associated diarrhoea as well as adverse effects from pharmacological agents like chloroquine phosphate, corticosteroids, heparin and its derivatives, remdesivir and lopinavir may be contributory (20).
Despite the often severe underlying inflammatory state, conservative management maybe a reasonable approach with endoscopy being reserved for those who fail conservative treatment. A case series by Cavaliere's group showed resolution of GI bleeding in all 6 patients managed conservatively with an intravenous proton pump inhibitor (PPI), careful monitoring and blood transfusion as required (21). In this series, 5 patients required supplementary oxygen and 1 required mechanical ventilation. It was not noted whether bleeding was present at admission or developed during hospitalization. In a matched case control study by Martin's group, COVID-19 patients with GI bleeding were matched to COVID-19 patients without bleeding (9). Of the 31 patients with UGIB, 10 patients with ongoing haemodynamic instability or anaemia underwent endoscopy with gastric or duodenal ulcers being the most common finding. In these patients, 4 required endoscopic interventions. Transfusion requirements amongst those undergoing endoscopy and those treated conservatively were not statistically significant. Despite a delay to endoscopy by 2.4 days, no deaths were attributed to GI bleeding. The only predictor of an UGIB was a previous history of GI bleeding, whereas the concomitant use of anti-coagulation trended towards an increased bleeding rate but did not reach statistical significance (9). Similar findings were noted in New York where the use of anti-coagulation, anti-platelet agents and corticosteroids did not increase the risk of GI bleeding. In this study GI bleeding increased mortality in COVID-19 patients (22). In a multicenter Italian study, 23 COVID-19 patients with UGIB were divided into those undergoing endoscopy within 24 hours and those in which either no endoscopy or endoscopy beyond 24 hours was performed (11). The GBS and mortality rates were similar in both groups. In this study 65% of all patients had two or more comorbidities and the majority were on anticoagulation at the time of bleeding. The commonest aetiology for bleeding was peptic ulcers followed by gastritis.
Further supporting the notion of conservative treatment, a small case series of 3 patients by Gadiparthi showed resolution of bleeding in all patients (23). In this series 2 patients were noted to have GI bleeding on admission, but both had a recent admission within the last 2 weeks. The third patient developed fresh rectal bleeding which was possibly related to the use of a fecal management system. Of note, one patient developed a recurrent episode of GI bleeding subsequent to intravenous heparin infusion for the treatment of a pulmonary embolism. This recurrent bleeding episode was managed conservatively with subsequent resolution. Mortality in this series was related to worsening respiratory failure and not GI bleeding.
The prophylactic use of PPIs has been advocated for high-risk ICU patients in national and international guidelines (24,25). Data from New York however, failed to show a protective effect from GI bleeding with the use of both PPIs and histamine receptor antagonists (H2RA) (22). This may allude to the notion that GI bleeding amongst COVID-19 patients may be a result of mucosal injury from a variety of mechanisms independent of acid secretion. Furthermore, it has been suggested that PPIs have some anti-viral activity and may reduce the risk of severe disease with SARS-CoV-2 (26). The therapeutic role of famotidine (a H2RA) has been evaluated in small studies with mixed results. An observational study has shown a lower level of inflammatory markers, decreased intubation rates and decreased mortality with the use of famotidine in COVID-19 patients (27). In this study although the famotidine group was younger, the benefit persisted even after adjusting for age. In comparison, there was no association between famotidine or PPI use and severity of illness in Hong Kong (28). Furthermore, a meta-analysis of 36 635 patients showed a non-significant reduction in the progression to severe disease, mechanical ventilation and death with the use of famotidine in COVID-19 patients (29).
Amongst patients with cirrhosis, in a non-pandemic setting, current guidance from the BAVENO VI guidelines recommend that endoscopy be performed within 12 hours of UGIB. In addition, the BAVENO VI guideline recommends the use of both endoscopic treatment and vasoactive agents in managing variceal bleeds (30). A more conservative approach in COVID-19 patients may, however, be feasible. In an Indian study, all patients (23 patients) with chronic liver disease and GI bleeding were managed conservatively with the use of splanchnic vasoconstrictors (somatostatin or terlipressin), proton pump inhibitors and a restrictive transfusion policy. Haemostasis was achieved in all patients within 24 hours and none required emergency endoscopy. One patient had ongoing intermittent bleeding and was found to have gastric antral vascular ectasia. No mortality was attributed to GI bleeding. Of note, none of these patients presented with shock at the time of bleeding and only one patient had a Child Pugh C score (10).
CONCLUSION
The COVID-19 pandemic poses unique challenges to the attending gastroenterologist and has demanded careful evaluation of the use of endoscopy as a therapeutic modality. The decision to perform upper GI endoscopy in COVID-19 patients requires meticulous risk stratification, with particular emphasis on patient safety, operator safety and the preservation of resources. All patients should be evaluated to identify risk factors for variceal bleeds. Despite prior guidance to perform endoscopy within 24 hours in patients with non-variceal bleeds, a conservative approach in the era of COVID-19 may be a reasonable alternative. Current data suggests that conservative therapy may be appropriate with careful re-evaluation of the need for endoscopy after 24 hours. This approach should be individualized with consideration given to haemodynamic status, respiratory status, comorbidities and other risk factors for mortality. Clinical risk scores, such as GBS, should be utilized to further aid in the decision making process with regards to both the need for and the urgency of endoscopy. There is currently insufficient evidence to advocate for the widespread use of prophylactic PPI or H2RA therapy in patients with COVID-19. While further data is needed to evaluate the role of acid suppressive therapy as both a prophylactic and therapeutic modality in COVID-19, its prophylactic use should be considered in critically ill patients. This is particularly relevant in resource constrained settings where high risk patients may not gain access to endoscopy.
