The use of super-selective mesenteric embolisation as a first-line management of acute lower gastrointestinal bleeding.

INTRODUCTION: In this study, we aim to assess the efficacy and safety of digital subtraction angiography (DSA) and super-selective mesenteric artery embolisation in managing lower gastrointestinal bleeding (LGIB) at a multi-centre health service in Melbourne (Australia).
METHOD: A retrospective case series of patients with LGIB treated with superselective embolisation in our area health service. Patients with confirmed active LGIB, on either radionuclide scintigraphy (RS) or contrast-enhanced multi-detector CT angiography (CE-MDCT), were referred for DSA, and subsequently endovascular intervention. Data collected included patient characteristics; screening modality; bleeding territory; embolisation technique; technical and clinical success; short to mediumterm complications and mortality up to 30 days; and the need for surgery related to procedural failure or complications.
RESULTS: There were 55 hospital admissions with acute unstable lower gastrointestinal bleeding that were demonstrable on CE-MDCT or RS over a 30-month period (from 1 January 2014 to 30 June 2016). Of these, eighteen patients were embolised. Immediate haemostasis was achieved in all embolised cases. Eight patients (44%) had clinical re-bleeding postembolisation and warranted repeated imaging. However, only one case (5.6%) had active bleeding identified and was re-embolised. There was no documented case of bowel ischemia or ischemic-stricture and none progressed on to surgery. 30 day mortality was zero.
CONCLUSION: Super-selective mesenteric embolisation is a viable, safe and effective first line management for localised LGIB. Our results overall compare favourably with the published experiences of other institutions. It is now accepted first-line practice at our institution to manage localised LGIB with embolisation.

Introduction

Lower gastrointestinal bleeding (LGIB) is defined as bleeding from a source distal to the ligament of Treitz and represents 20–24% of all cases of gastrointestinal bleeding [1], [2]. The incidence of LGIB ranges from 20.5 to 27 cases per 100,000 in adults in the USA [2], [3], and increases markedly with age, reflecting underlying diseases such as diverticular disease, angiodysplasia, colitis and neoplasia [2], [3], [4].

Endoscopic and angiographic interventions have reduced the need for surgery in most cases of severe LGIB. Surgery would involve either directed segmental colectomy; blind segmental colectomy; or subtotal colectomy. It is usually reserved for unstable patients who have failed conservative, endoscopic or endovascular management. There is an increased mortality rate associated with blind segmental and subtotal colectomy, particularly in elderly patients with medical co-morbidities [3].

Western Health, a multi-centre institution servicing a large metropolitan area in Melbourne (Australia), has embraced super-selective embolisation of LGIB as an important treatment modality. It can obviate the need for surgery where endoscopy would otherwise be unsuccessful. Unstable patients who thus present with severe LGIB undergo screening with either contrast-enhanced Multi-Detector Computed Tomography (CE-MDCT) or Radionuclide Scintigraphy (RS). Following identification and localisation of active bleeding in a vascular territory, the patient is transferred to the digital subtraction angiography (DSA) suite for further investigation and, where feasible intervention.

The aim of this study is to review our institution's experience and compare it to the existing body of evidence on arterial embolisation of LGIB.

Methods

This study is a case series of patients presenting with LGIB treated with super-selective embolisation reported in line with the PROCESS criteria [5]. A search was performed through our DSA suite logbook. Our inclusion criteria consisted of all patients with confirmed LGIB (on either RS or CE-MDCT) and referred for angiography. All patients found to have active bleeding subsequently underwent endovascular treatment. Those patients who had bleeding from sites other than the lower gastrointestinal tract were excluded. The catchment for this institution consists of the western metropolitan region of Melbourne, Australia. There are three campuses that service this area however patients with LGIB requiring immediate treatment are transferred to the main campus where interventional radiological services are concentrated.

Following patient identification, a review of their medical records was performed. Data included patient characteristics such as age, sex and co-morbidities; screening modality (RS and/or CE-MDCT); bleeding territory; embolisation technique; technical success; clinical success; short to medium-term complications including infarction, ischemia-related stricture and mortality up to 30 days; and need for surgery related to procedural failure or complications.

Technical success was defined as cessation of contrast extravasation as demonstrated on DSA after deployment of embolising agent. Clinical success was defined as normalisation of vital signs; no further need for fluid resuscitation; transfusion requirement of less than two units of packed red blood cells; and no further radiologically demonstrated LGIB and subsequent further intervention for ongoing LGIB. All cases achieved 30-day follow-up and those cases performed up to March 2016 achieved 6-month follow-up.

Results

During the period of 1 January 2014 to 30 June 2016, there were 414 patients admitted with the diagnosis of acute lower gastrointestinal bleeding across all three campuses as obtained via the Diagnosis Related Group (DRG) coding from the institution's health information service. 62 of them were haemodynamically unstable on admission and underwent CE-MDCT or RS, of which 55 demonstrated LGIB. All 55 of these cases progressed onto DSA without any significant delay with the intention of endovascular intervention.

One case was excluded on grounds of pan-gastrointestinal bleeding in a patient who was diagnosed with hemophagocytic syndrome consisting of pancytopenia and disseminated intravascular coagulation. Micro-coil embolisation was employed with success at controlling a significant bleeding point in the terminal ileum. However, this patient eventually died from hepatic failure.

Out of the remaining 54 cases of LGIB investigated with DSA, 18 cases (33%) had endovascular intervention. These 18 cases had a male to female ratio of 2:1. The median age was 74.50 years (range 59–92). There was no active bleeding demonstrated on DSA in the remaining 36 patients, hence, no endovascular intervention was performed. Ten patients (56%) were on anticoagulation/antiplatelet medications prior to admission: six patients were on aspirin alone; three on aspirin plus clopidogrel; and one on warfarin. The indication for warfarin was for an in-situ mechanical heart valve. Patients on clopidogrel were prescribed for recent insertion of coronary artery stents while patients were on aspirin for prophylactic treatment of cardiovascular accidents (CVA).

6 of the 10 patients on anticoagulation/antiplatelet medications were part of the group that underwent endovascular intervention while the remaining 4 patients did not. Given the risk of cardiovascular and cerebrovascular events, patients on warfarin were bridged with low-molecular-weight heparin (LMWH) prior to any intervention. Patients on both aspirin and clopidogrel had clopidogrel withheld on admission but remained on aspirin, while patients on aspirin only remained on it prior to any intervention. The effect of anticoagulation/antiplatelet medications on LGIB was not measured as an endpoint in this study. Further studies looking into its association with LGIB can be considered in the future. Patient demographics and comorbidities are summarised in Table 1.

Table 1

Demography.

Demography	Number
Gender
• Male	12 (67%)
• Female	6 (33%)
Median Age	74.50 (Range 59–92)
Co-morbidities
• Ischaemic Heart Disease	7 (39%)
• Atrial Fibrillation	3 (17%)
• Hypertension	12 (67%)
• Diabetes	6 (33%)
Anticoagulation
• Aspirin	6 (33%)
• Aspirin + Clopidogrel	3 (17%)
• Warfarin	1 (6%)

All but two of cases underwent embolisation within 24 h of admission. Of the two exceptions, one was initially diagnosed with peri-prostatic abscess causing secondary rectal bleeding. This patient's bleeding gradually worsened to the point of requiring active intervention. The other case involved embolisation of colonic bleeding from an end colostomy more than 24 h following a Hartmann's procedure for perforated sigmoid colon.

Presumptive causes of bleeding were diverticular disease [9], angiodysplasia [2], inflammatory [2], iatrogenic [2], haemorrhoidal [1], infective [1] and unknown [1] (Fig. 1). The presumptive causes of bleeding were derived from the results of the RS or CE-MDCT in addition to the patient's history and past investigations such as previous colonoscopies and radiological scans. Cases of note were: bleeding from a rectal ulcer due to CMV proctitis and bleeding from a point in the terminal ileum of which the cause was unascertainable from the records.

Fig. 1

Presumptive aetiology of LGIB.

CE-MDCT was performed in eleven cases while five had radionuclide scans, and two required both modalities to identify the source of bleeding (Fig. 2).

Fig. 2

Diagnostic imaging modality performed.

There was an even distribution between the two major vascular territories of the lower gastrointestinal tract (Fig. 3). The CMV rectal ulcer bleed was identified from the middle rectal artery (iliac artery branch), after repeated IMA studies failed to identify a bleeding point.

Fig. 3

Localised Bleeding Sites of all 18 cases by vascular territory.

Micro-coil embolisation was the preferred embolisation agent and appears to be effective in achieving haemostasis with only 2 cases requiring additional Gel-Foam. Immediate haemostasis was achieved in all 18 cases after embolisation. This was demonstrated by no further contrast extravasation on DSA post deployment of embolent (Fig. 4). However, eight cases had clinically significant re-bleeding that warranted repeated imaging within 30 days of initial intervention; three had CT angiogram, three had radionuclide scintigraphy, and two proceeded directly back to DSA; the decision of which modality was based on preference of the treating unit Surgeon and Radiologist on call. A total of 3 of the 8 patients had demonstrable re-bleeding on repeat imaging but only one case required further embolisation (interestingly, a case that was evaluated with DSA). The remaining 5 patients did not demonstrate bleeding on subsequent DSA and no embolisation was performed.

Fig. 4

Super-selective embolisation: a) before and b) after coil deployment.

Importantly, no case of re-bleeding required surgery to control bleeding. There was no 30-day mortality. There was also no documented stricture formation in the intervened segment of the gastrointestinal tract on follow-up colonoscopy, subsequent admission to hospital or outpatient review.

Discussion

Acute LGIB is arbitrarily classified as bleeding of less than 3 days duration and may result in hemodynamic compromise and/or the need for blood transfusion [2]. On the other hand, chronic LGIB is defined as bleeding of greater than 3 days, encompassing both occult and obscure bleeding and usual presents with iron-deficient anaemia [3]. Acute LGIB is most commonly due to diverticulosis (40%), vascular ectasia (30%), various colitis (inflammatory, ischemic, radiation) (20%), colonic neoplasia (14%) and anorectal causes (10%) [6]. In 80–85% of cases of LGIB, bleeding will stop spontaneously and the majority of cases will not require immediate investigation or intervention. Patients with ongoing active bleeding with or without hemodynamic compromise will need diagnostic and possibly therapeutic procedures. The current study represents a single-institution's experience in embolisation for acute LGIB.

Colonoscopy has been advocated as first-line management of LGIB [2]. This has the advantage of being diagnostic with a yield of 89–97% [2] and has an accuracy rate of 72–86% [1], [3]. It also allows for the use of various haemostatic techniques where possible [3], [6]. Ideally the patient should undergo bowel preparation prior to the procedure to facilitate visualisation. However, this is not always possible especially in an unstable patient with significant haemorrhage. In this situation, the alternative is radiologic identification of a bleeding point with a view to subsequent endovascular intervention.

Three techniques are useful to localise LGIB: contrast-enhanced multi-detector CT angiography (CE-MDCT), radionuclide scintigraphy (RS) and digital subtraction angiography (DSA). Angiography has the advantage of being therapeutic as well as diagnostic. It has 100% specificity but has a sensitivity of only 30–47%, requiring a relatively higher rate of bleeding of at least 1.0 ml/min and has a diagnostic yield ranging from 41 to 78% [3], [7]. Scintigraphy has a high sensitivity rate and can detect bleeding rates as low as 0.1 ml/min but is less specific and unsuitable for unstable patients due to a longer study time required as well as reduced diagnostic yield with brisk bleeding [7], [8]. Recently, CE-MDCT is able to detect bleeding rates of 0.3–0.5 ml/min and is highly sensitive and specific [2], [3].

Angiographic intervention for treatment of LGIB has existed since early attempts in the 1970s. During this period there were high rates of bowel infarction from non-selective embolisation. It did not emerge as a viable alternative until the 1990s when the technology had improved. With the development of coaxial microcatheters it became possible to carry out super-selective catheterisation of specific marginal arteries or vasa recta to deliver embolic material in the form of microcoils, gelfoam or polyvinyl alcohol close to the site of bleeding. It was able to reduce the risk of infarction as well as decrease bleeding from collateral vessels [8]. The efficacy of super-selective embolisation has been shown to vary depending on aetiology, with a greater rate of control of bleeding at 30 days in diverticular bleeding than non-diverticular [8]. Following angiographic treatment, the potential complications include bowel infarction, re-bleeding and stricture formation secondary to ischemia. However, it is understood that patients with LGIB can still pass altered blood per-rectum up to 1 week after bleeding has ceased.

Three contemporary small retrospective case series from the modern era of super-selective mesenteric embolisation are described in Table 2.

Table 2

Comparison of outcomes.

Outcomes	Current Study (Australia, 2016) N = 18	Rider et al. [13] (USA, 2009) N = 24	Tan et al. [10] (Singapore, 2008) N = 32	Waugh et al. [11] (Australia, 2004) N = 27
Immediate hemostasis	100%	100%	97%	96%
Rebleeding	16.7%	4.3%	21.9%	29.6%
Repeated embolisation	5.6%	4.3%	3.1%	22%
Ischemia	0%	8.3%	3.1%	14.8%
Progression to surgery	0%	12.5%	28% (4 rebleed; 1 incomplete hemostasis; 1 ischemia; 3 surgeon decision)	7.4%
30d Mortality	0%	0%	9.4%	7.4%
Ischemic stricture	0%	4.2%	Not reported	Not reported

Rider et al. [13] from the Ochsner Clinic (New Orleans, USA) consisted of 24 cases over a 2-year period. They achieved technical success in all cases. One case re-bled and was unsuccessfully re-embolised thus requiring sigmoid colectomy. Two cases developed ischemia and underwent segmental colectomy. There was no mortality. One case developed an ischemic stricture that required subsequent surgical intervention.

Tan et al. [11] reviewed a series of 32 cases of mesenteric embolisation over a period of 82 months at a large teaching hospital in Singapore. Technical success was achieved in 31 cases however clinical success only in 20 cases. 7 cases re-bled with 1 managed with repeat embolisation, 1 hemostased with colonoscopy, 4 progressed to surgery, and the remaining case managed conservatively due to underlying metastatic disease. Three cases underwent bowel resection at the treating surgeons' preference despite no re-bleeding or ischemia and one case had a segmental colectomy for ischemia. It is important to note that only 5 cases underwent surgery for indications related to LGIB and ischemia. There were 3 mortalities in this series however only one case was directly related to intervention. Ischemic stricture was not reported.

The third series by Waugh et al. [12] performed a review of 27 embolisation cases over a period of 63 months at a metropolitan teaching hospital in Melbourne. They achieved technical success in 26 cases, and clinical success in 19 cases with repeat embolisation in 6 cases. Four cases had clinical symptoms of ischemia with 2 mortalities: one due to ischemic gut and the other related to surgical complications associated with resection of the ischemic segment. Two cases progressed to surgery: one for ischemia as stated above and another for ongoing LGIB despite repeated attempts of embolisation. Ischemic stricture was not an endpoint in this series.

We have shown that acute unstable LGIB can be managed effectively using super-selective mesenteric embolisation following radiological localisation. In comparison with the other 3 studies, our study had better outcomes when comparing the proportion of patients who had post embolisation ischemia, the proportion of patients who progressed on to surgery, 30 day mortality and the proportion of patients who developed ischemic stricture post embolisation. However, we had a higher proportion of patients who re-bled post embolisation compared with the Rider et al. [13] study, and a higher proportion of patients who required repeat embolisation compared with the Rider et al. [13] and Tan et al. [11] studies but the rates were lower when compared with the Waugh et al. study.

Based on these positive preliminary results, our institution has adopted a protocol of using CE-MDCT to localise the vascular territory immediately following resuscitation and stabilisation of all acute unstable LGIB. However, this study is limited by factors inherent to retrospective case series type studies; as well as having a small cohort of patients. It is thus underpowered to draw any necessary statistically significant conclusions. Our results suggest that further research is warranted with prospective recruitment of cases and follow up using a defined protocol.

Conclusion

The results of our study show that super-selective mesenteric embolisation continues to be a viable, safe and effective first line management for localised LGIB. Overall, our results also compare favourably with the published experiences of other institutions. It is now accepted first-line practice at our institution to manage localised LGIB with embolisation.

Conflicts of interest

None.