Haematological risk factors predicting clinical success in transarterial embolisation for acute gastrointestinal bleeding

Abstract

Objectives:

Evaluate clinical outcomes in transarterial embolisation (TAE) for acute gastrointestinal bleeding (GIB) and determine risk factors for 30-day reintervention for rebleeding and mortality.

Methods:

TAE cases were retrospectively reviewed between March 2010 and September 2020 at our tertiary centre. Technical success (angiographic haemostasis following embolisation) was measured. Uni- and multivariate logistic regression analysis were performed to identify risk factors for clinical success (absence of 30-day reintervention or mortality) following embolisation for active GIB or empirical embolisation for suspected bleeding.

Results:

TAE was conducted in 139 patients (92 (66.2%) male; median age:73, range: 20–95 years) for acute upper GIB (n = 88) and lower GIB (n = 51). TAE was technically successful in 85/90 (94.4%) and clinically successful in 99/139 (71.2%); with 12 (8.6%) reintervention cases for rebleeding (median interval 2 days) and 31 (22.3%) cases of mortality (median interval 6 days). Reintervention for rebleeding was associated with haemoglobin drop > 40 g l⁻¹ from baseline based on univariate analysis (p = 0.047). 30-day mortality was associated with pre-intervention platelet count < 150×10⁹ l⁻¹ (p < 0.001, OR 7.35, 95% CI 3.05–17.71) and INR > 1.4 (p < 0.001, OR 4.75, 95% CI 2.03–11.09) on multivariate logistic regression analysis. No associations were found for patient age, gender, antiplatelet/anticoagulation prior to TAE, or when comparing upper and lower GIB with 30-day mortality.

Conclusion:

TAE had excellent technical success for GIB with relatively high (1-in-5) 30-day mortality. INR > 1.4 and platelet count < 150×10⁹ l⁻¹ were individually associated with TAE 30-day mortality, and pre-TAE > 40 g l⁻¹ haemoglobin decline with rebleeding requiring reintervention.

Advances in knowledge:

Recognition and timely reversal of haematological risk factors may improve TAE periprocedural clinical outcomes.

Introduction

Acute gastrointestinal bleeding (GIB) is a life-threatening presentation with high complication and mortality rates in the United Kingdom, ¹ and a commonly occurring medical emergency. ² GIB is subdivided into upper and lower bleeds depending on the arterial territories involved. Management of both upper and lower GIB involves several options, including endoscopy (allowing endoscopic evaluation for luminal evidence of haemorrhage, with haemostatic interventions used in such cases), surgical artery ligation or bowel resection, local vasoconstrictive therapy and endovascular approaches. ³ The choice of treatment usually depends on factors such as haemodynamic stability and availability of services. Endoscopic failure, rebleeding, or patients unfit for endoscopy or surgery are criteria for potential angiographic endovascular procedures for upper GIB. ⁴ In contrast, endovascular approaches are now considered the main primary treatment method for lower GIB. ⁵ CT angiography (CTA) is performed initially in each case as a screening measure, as it is sensitive and specific for active arterial bleeding. ^6,7 Patients can then proceed to a catheter angiography where both diagnostic and therapeutic functions can be deployed, the latter through transarterial embolisation (TAE). Being an emergency procedure, TAE is commonly performed in patients who present acutely with complications of GIB including severe anaemia, coagulopathy and hypovolemic shock, and this may be compounded by a background of multiple comorbidities. Importantly, current national guidelines have specifically outlined the role of interventional radiology in the management of non-variceal upper GIB, by advising that TAE should be offered to unstable patients who rebleed after initial endoscopic treatment, or when endoscopic therapy is technically difficult and the endoscopist cannot achieve or secure haemostasis. ⁸ Catheter angiography confers a significant benefit as it is able to localise a specific point of bleeding and is preferable to surgery, due to the high post-operative mortality in this group of patients, most of whom are extremely unwell at the time of intervention.

However, despite being an established interventional procedure, there is a paucity of evidence on the predictors governing TAE therapeutic success; namely, reintervention for rebleeding and the 30-day mortality rate. Small-scale studies have identified specific pre-embolisation risk factors for rebleeding following GIB to include a low platelet count ⁹ and coagulopathy ¹⁰ ; although these findings are yet to be validated by larger studies. Patient related factors such as age, aetiology of bleed and comorbidities have proven important predictors of outcome in some studies. ^11,12 Others have evaluated angiographic-related factors, demonstrating an association between negative angiographic findings and lower rebleeding rates, ¹³ and a higher rebleeding rate when embolisation was delayed, ¹⁴ and targeted over prophylactic embolisation leading to angiographic success. ^15,16 Previous work examining the effects of treatment-related factors indicates that greater red blood cell transfusion requirements were associated with a higher rate of rebleeding. ^14,16,17

The aim of our study was to evaluate the clinical outcomes of TAE used for acute upper and lower GIB, in order to identify and stratify potential predictors of adverse procedural outcomes, such as complication rates, technical and clinical success (reintervention for rebleeding), and the procedural 30-day mortality. This would provide a prognostic insight into any potentially reversible risk factors that could be addressed pre- or intraprocedurally with TAE.

Methods and materials

TAE intervention cases for GIB were retrospectively reviewed between March 2010 and September 2020 at our single tertiary centre; identified from a system search of our institutional image-reporting software CRIS (by Healthcare Solutions Software, UK) database. TAE procedural reports were primarily reviewed to include interventions proceeding to GIB embolisation therapy; pure gastrointestinal angiographic studies (no embolisation) were not analysed for procedural outcomes in this study. Technical success, clinical success, type of embolic agent and procedural complications including reintervention for rebleeding were collected for each patient. Technical success was defined as the successful deployment of an embolic agent with satisfactory intraprocedural haemostasis.Catheter angiography demonstrating evidence of fluoroscopic contrast extravasation was defined as an active GIB and embolisation was attempted in all cases. If fluoroscopic contrast extravasation was not demonstrated during angiographic but there was a high suspicion or risk for active bleeding, empirical embolisation was performed; 35/83 in upper GIB and 6/56 lower GIB cases. Clinical success was defined as the absence of post-procedural bleeding requiring a secondary reintervention (TAE or endoscopic) and/or mortality within 30 days after the primary TAE procedure; whether it was embolisation for active bleeding or empirical treatment.

All TAE for acute GIB were performed by a vascular interventional radiology (VIR) consultant with a high level of experience in the procedure. The underlying diagnosis causing the GIB in each case was determined on the basis of initial endoscopy findings (if performed), CT mesenteric angiography (CTA), clinical history, and patient medical or surgical history.

A number of pre-embolisation patient characteristics were determined prior to TAE. Pre-embolisation haemoglobin, platelet count and international normalised ratio (INR) data were acquired from Integrated Clinical Environment (ICE) Care Coordination Software (Conquest Laboratory, UK), and iLAB Operations Software (Agilent Technologies, USA); detailed in Table 1. Haemoglobin baseline for patients was calculated using the most recent haemoglobin result available within 1 year of TAE prior to hospital admission for GIB. Change in patient haemoglobin from baseline calculated and a haemoglobin drop > 40 g l⁻¹ was categorised as a binary variable for analysis; average taken from the Bleeding Academic Research Consortium (BARC) standardised definition for Type 3a cardiovascular bleeding (haemoglobin drop 30–50 g l⁻¹). ¹⁸

Table 1.

Characteristics of patients presenting with acute GIB

Patient characteristics	n = 139
Male	92 (66.2%)
Age in years, Median (range)	73 (20 – 95)
Pre-embolisation blood tests
Hb (g l−1), Mean (range)	80 (18 – 139)
Baseline Hb (g l−1), Mean (range)	118 (80 – 164)
Platelets (x109 l−1), Median (range)	239 (24 – 956)
INR, Median (range)	1.4 (0.9–3.4)
Significant comorbidities (Table 2)
Cardiovascular	45 (32.4%)
Gastrointestinal	33 (23.7%)
Major thoraco-abdominal surgery	26 (18.7%)
Diabetes mellitus	24 (17.2%)
Neurological	20 (14.4%)
Respiratory	18 (12.9%)
Chronic kidney disease stage ≥ 3	14 (10.1%)
Haematological malignancy	7 (5.0%)
Underlying diagnosis
Gastro-duodenitis / peptic ulcer	41 (29.4%)
Abdominal malignancy	19 (13.7%)
Anticoagulation or antiplatelet	15 (10.8%)
Pancreatic pseudoaneurysm	15 (10.8%)
Post-surgical haemorrhage	10 (7.2%)
Angiodysplasia / AV malformation	10 (7.2%)
Diverticular disease	9 (6.5%)
Inflammatory bowel disease	2 (1.4%)
Hepatic coagulopathy	1 (0.7%)
Ehlers Danlos syndrome	1 (0.7%)
Unknown	16 (11.5%)
Pre-intervention antiplatelet and anticoagulant agents
Antiplatelet therapy	34 (24.5%)
Aspirin	27 (19.4%)
Clopidogrel	9 (6.5%)
Anticoagulant therapy	23 (16.5%)
Warfarin	10 (7.2%)
Low molecular weight heparin	9 (6.5%)
Apixaban	3 (2.2%)
Dabigatran	1 (0.7%)
None	83 (59.7%)

AV, Arteriovenous; GIB, gastrointestinal bleeding;Hb, Haemoglobin; INR, international normalised ratio.

Pre-embolisation antiplatelet and anticoagulant therapies were retrieved using ICE software, and cross-referenced with TAE procedural report documentation and the Electronic Prescribing and Medicines Administration System (Medchart, UK); Table 1. Patient age, comorbidities and mortality data were also obtained from ICE Software, and patient comorbidities deemed moderate to severe enough to cause significant morbidity were grouped into bodily systems (Table 2).

Table 2.

Comorbidities of patients undergoing TAE for GIB

System category	Comorbidities
Cardiovascular (32.4%)	Ischaemic heart disease
	Congestive cardiac failure
	Atrial fibrillation
	Peripheral vascular disease
	Valvulopathy
	Previous cardiac surgery (any)
Gastrointestinal (23.7%)	Inflammatory bowel disease
	Hepatic failure
	Upper and lower GI malignancy
	Peptic ulceration
	Oesophageal dysmotility
	Chronic pancreatitis
	Previous abdominal surgery (any)
Respiratory (12.9%)	Asthma (severe)
	COPD
	Lung and laryngeal malignancy
	Pneumonia (recurrent)
	Pulmonary embolism
	Obstructive sleep apnoea
	Interstitial pulmonary fibrosis
Neuromuscular (14.4%)	Motor neuron disease
	Multiple sclerosis
	Parkinson’s disease
	Hypoxic brain injury
	Dementia
	Brain tumours
	Epilepsy
	Cerebrovascular accident
Metabolic (30.2%)	Hypothyroidism
	Diabetes mellitus (Type 1 or 2)
	Chronic kidney disease

COPD, chronic obstructive pulmonary disease; GIB, gastrointestinal bleeding; TAE, transarterial embolisation.

We used uni- and multivariable logistic regression to identify predictors for 30-day mortality and presented this as crude and adjusted odds ratios [ORs]. Only univariable analysis was performed to identify predictors for 30-day reintervention, as the number of reintervention cases was deemed too low to perform an accurate multivariable logistic regression. In the regression model, age in years was treated as continuous variables, whereas haemoglobin drop, INR and platelet count were categorised into binary variables using the clinically important threshold values of > 40 g l⁻¹, > 1.4 and < 150×10⁹ l⁻¹ respectively. Patient age, pre-intervention antiplatelet/anticoagulant, > 40 g l⁻¹ haemoglobin drop, platelet count < 150 x 10⁹  l⁻¹ and INR > 1.4 were all controlled for; based on the biological plausibility in attributing to reintervention for rebleeding and 30-day mortality.

All analyses were conducted using Stata (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC.). p values < 0.05 were considered statistically significant.

The study was designed to evaluate institutional clinical service outcomes, with a focus on reintervention rate as a delayed complication and overall 30-day mortality, and therefore did not require any formal ethical approval in accordance within our National Research Ethics Service (NRES) guidelines. Local clinical audit committee approval was obtained.

Results

Patient characteristics

During the 10.5-year period (March 2010 and September 2020), TAE was performed for acute GIB in 139 patients (92 male; median age: 73, range: 20–95 years) by 7 individual VIRs. Pre-embolisation mean haemoglobin was 80 g l⁻¹ (range, 18–139), median platelet count was 239 × 10⁹  l⁻¹ (range, 24–956) and median INR was 1.4 (range, 0.9–3.4). The mean baseline haemoglobin was 118 g l⁻¹ (range 80–164). Significant patient comorbidities within the cohort are detailed in Table 2 according to major body systems. Aetiology for acute GIB in our cohort comprised of gastroduodenitis and peptic ulcer disease (29.4%), abdominal malignancy (13.7%), concomitant antiplatelet or anticoagulation (10.8%), pancreatic pseudoaneurysm (10.8%); with the remainder of cases included in Table 1. Pre-procedural concomitant therapy comprised of 34 (24.5%) patients previously receiving antiplatelet and 23 (16.5%) who were anticoagulated prior to TAE; refer to Table 1 for details.

TAE procedure characteristics

Primary intervention with TAE was performed in 139 patients with suspected or confirmed active GIB from either endoscopic of CTA findings. Active GIB was confirmed through catheter angiography in 90 (64.7%) of cases, where therapeutic embolisation was performed in 85 (61.2%) with technical success, whereas in 5 (3.6%) there was intraprocedural failure. No active bleed was demonstrated on angiography in 49 (35.3%), empirical embolisation was performed in each case due to the suspicion or risk of occult GIB. Embolisation techniques included deployment of coils alone in 128 (92.1%), with some patients additionally receiving gelatin sponge (Gelfoam^TM) in 5 (3.6%) or onyx (Embozene^TM) in 1 (0.7%).

TAE was technically successful in 85/90 (94.4%) and clinically successful in 99/139 (71.2%); the absence of reintervention for rebleeding or mortality within 30 days post-procedure. The five causes of procedural failure were due to three cases of persistent vasospasm, and two cases of challenging vessel morphology. Cases of clinical failure within 30 days included 12 (8.6%) reinterventions for rebleeding (median interval of 2 days), and 31 (22.3%) cases of mortality (median interval 6 days). Figure 1 demonstrates a case from our cohort of reintervention for rebleeding after initially achieving successful haemostasis. There were four cases of intraprocedural coil migrations that were not clinically significant, and the procedure was successful in each case. Arterial territories accessed for embolisation were subdivided into upper (coeliac artery axis) and lower (superior and lower mesenteric arteries) territories; see Table 3 for complete reference.

Figure 1.

Coeliac angiography for emergency TAE in a patient presenting with massive haematemesis initially failing endoscopic intervention. Pre-procedural bloods included a haemoglobin drop of 55 g/L from baseline, platelet count of 28 x 10⁹ l⁻¹ and INR 1.7. A. Coeliac angiography demonstrated a large active haemorrhage likely from a left gastric artery branch. The bleed was deemed too peripheral to reach with microcatheter, and a decision to embolisation the main left gastric artery was made and successful haemostasis achieved with multiple 3 mm coils. B. 4 hours following initial TAE procedure, the patient experienced further active upper GIB requiring multiple red cell concentrate transfusions (4 units). Repeat coeliac angiography demonstrated additional feeding vessels from short gastric and gastroepiploic arteries. C. The origin of the left gastric and an aberrant branch were successfully occluded with several coils achieving haemostasis.

Table 3.

Procedural characteristics for upper and lower GIB managed by TAE

Primary intervention: Transarterial embolisation (n = 139)
Active bleed
Therapeutic TAE	85 (61.2%)
Failed procedure	5 (3.6%)
No active bleed
Empirical TAE	49 (35.3%)
TAE interventions
Coils	128 (92.1%)
Coils + gelatin sponge (GelfoamTM)	5 (3.6%)
Onyx (EmbozeneTM)	1 (0.7%)
Angiography only (failed TAE)	5 (3.6%)
Mesenteric axis arterial territories
Upper GIB: Coeliac axis branches	88 (63.3%)
Gastric (left or right)	9
Hepatic	2
Gastroduodenal	73
Superior pancreaticoduodenal	4
Lower GIB: SMA branches	28 (20.1%)
Inferior pancreaticoduodenal	4
Jejunal	9
Ileocolic	8
Ascending colon	7
Lower GIB: IMA branches	23 (16.5%)
Descending colon	4
Sigmoid colon	7
Rectal	12
Complications
30-day mortality	31 (22.3%)
Reintervention (rebleed)	12 (8.6%)
Technical failure	5 (3.6%)
Coil migration	5 (3.6%)
Secondary intervention: Transarterial embolisation (n = 12)
Active bleed
Therapeutic TAE	8 (66.7%)
Endoscopically clipped	1 (8.3%)
No active bleed
Empirical TAE	1 (8.3%)
Angiography only	2 (16.7%)

GIB, gastrointestinal bleeding; IMA, inferior mesentery artery; SMA, superior mesenteric artery; TAE, transarterial embolisation.

Table 4 shows crude and adjusted ORs for predictors of 30-day reintervention and mortality.

Table 4.

Multivariate analysis of patient and procedural factors associated with outcomes in TAE performed for acute GIB

	Reintervention for rebleeding (n = 12)			30 day mortality (n = 31)
Variables	Univariate			Univariate			Multivariate
	OR	CI 95%	p	OR	CI 95%	p	OR	CI 95%	p
Age	1.022	0.971–1.078	0.971	1.027	0.989–1.067	0.171	1.027	0.997–1.057	0.082
Gender	1.135	0.264–4.873	0.865	1.178	0.407–3.408	0.762	1.33	0.556–3.168	0.524
Anticoagulation	0.429	0.043–4.249	0.47	0.775	0.167–3.584	0.744	1.185	0.354–3.972	0.783
Antiplatelet	0.175	0.017–1.811	0.144	1.761	0.480–6.466	0.394	1.667	0.692–4.015	0.255
Hb baseline drop > 40 g l −1	4.151	1.021–16.882	0.047*	0.471	0.173–1.285	0.141	0.777	0.347–1.743	0.542
Platelets < 150 x 10 9   l −1	1.048	0.241–4.530	0.953	5.995	2.262–15.889	<0.000*	7.35	3.051–17.705	<0.000*
INR > 1.4	0.601	0.110–3.292	0.557	3.313	1.186–9.257	0.022*	4.747	2.032–11.087	<0.000*
≥ 2 comorbidities	1.025	0.260–4.048	0.972	0.759	0.256–2.246	0.618	1.149	0.517–2.554	0.734
Upper vs lower GI bleed territory	1.1	0.257–4.717	0.9	1.112	0.402–3.080	0.838	1.094	0.467–2.564	0.836

CI, confidence interval; GIB, gastrointestinal bleeding; Hb, Haemoglobin; INR, international normalised ratio; OR, odds ratio; TAE, transarterial embolisation.

^aStatistically significant result (p-value < 0.05).

Multivariate analysis for 30-day mortality following TAE demonstrated a significant association with pre-intervention platelet count < 150×10⁹  l⁻¹ (OR 7.35, p = < 0.001, 95% CI 3.05–17.71) and for the variable INR > 1.4 (OR 4.747, p = < 0.001, 95% CI 2.03–11.09). There was no significant association with the remaining variables including patient age (OR 1.03), gender (OR 1.33), use of anticoagulation (OR 1.19) or antiplatelet (OR 1.67) agent, haemoglobin drop > 40 g l⁻¹ (OR 0.78), ≥ 2 patient comorbidities (1.15), and upper vs lower GIB territories (OR 1.09).

Reintervention

Univariate analysis demonstrated a significant association between reintervention for rebleeding and drop in haemoglobin > 40 g l⁻¹ (OR 4.15, p = 0.047, 95% CI 1.02–16.88). There was no significant association with the remaining variables including patient age (OR 1.02), gender (OR 1.14), use of anticoagulation (OR 0.43) or antiplatelet (OR 0.18) agent, platelet count < 150×10⁹  l⁻¹ (OR 1.05), INR > 1.4 (OR 0.60), ≥ 2 patient comorbidities (OR 1.03), and upper vs lower GIB territories (OR 1.10).

Discussion

Our TAE technical success rate of 94.4% is comparable with published literature (88.6–100%), and our clinical success rate of 71.2% is also inline with other studies (51–92%). ^{9–11,13,15–17,19}

Univariate analysis demonstrated that pre-procedural haemoglobin decline > 40 g l⁻¹ was significantly associated with rebleeding requiring reintervention (p = 0.047). The importance of pre-intervention haemoglobin in acute upper GIB has been emphasised by Hongsakul et al, demonstrating anaemia (haemoglobin level < 80 g l⁻¹) was a predictor of failure to achieve 30-day haemostasis (rebleeding) following TAE. ²⁰ We speculate that looking at haemoglobin level alone is not a true reflection of bleeding severity; our reasoning for using pre-procedural haemoglobin decline as a parameter instead. The basis for this decision is that the average patient within our cohort had baseline anaemia prior to the bleeding (mean haemoglobin of 118 g l⁻¹, range 80–164) and a significant proportion had underlying cardiovascular disease, chronic kidney disease and malignancy; which are all associated with chronic anaemia. We suggest that haemoglobin decline is a more powerful predictor of TAE clinical outcomes in the acute GIB setting, and more research in larger patient cohorts may demonstrate statistical significance of this variable.

Pre-intervention thrombocytopaenia (platelets < 150×10⁹  l⁻¹) and INR >1.4 were significant risk factors for 30-day mortality following TAE. Coagulopathy has been described in the literature as a risk factor for TAE complications and clinical failure. Chen et al demonstrated increased odds of TAE treatment failure (30-day mortality and/or recurrent bleeding requiring surgery within 7 days) with the presence of coagulopathy. ²¹ Pech et al found that bleeding related to anticoagulation therapy was associated with a significantly higher mortality rate following TAE ²² ; although we found no association with pre-intervention anticoagulation in our study. We believe we are the first to describe platelet count < 150×10⁹  l⁻¹ as a risk factor for 30-day mortality. Several explanations can account for these aforementioned observations. Firstly, as catheter-direct coil embolisation requires a functioning coagulation pathway to achieve thrombosis on the coil fibres and successful vessel occlusion, coagulopathy and thrombocytopaenia likely inhibits these events allowing blood to freely pass the coil. Secondly, coagulopathy and thrombocytopaenia may indirectly reflect significant haemoglobin drop during an active GIB, necessitating red cell concentrate transfusion periprocedurally, which in turn may induce a dilutional effect on platelet count and clotting factors. Several studies have already demonstrated that rebleeding following TAE had a significant association with the quantity of red blood cell concentrate transfused in GIB. ^14,16 Powerski et al identifed haemoglobin level < 5.3 mmol⁻¹ (equivalent to < 95 g l⁻¹) and ≥ 6 units red blood cell concentrate transfusion were negative prognostic indicators for clinical success. ²³ Mensel et al showed decreased need for red blood cell concentrate, fresh frozen plasma (FFP) and coagulation products were predictors for TAE technical success, and post-interventional FFP demand (volume) negatively correlated with clinical success. ¹⁷ It is possible that both platelet and clotting factor deficits may be incompletely rectified pre-procedurally, such as red cell concentrate administration without concurrent correction for thrombocytopaenia and/or coagulopathy. This particularly may be the case when TAE is following an unsuccessful emergent endoscopic procedure for acute GIB, where time and haemodynamic instability are influencing factors. Currently, there is a paucity of evidence to recommend optimal therapeutic platelet count targets in patients with an active GIB, although published standards in haematology literature recommend platelet transfusion to maintain a platelet count of ≥ 50×10⁹  l⁻¹. ²⁴ We suggest that the recognition and timely reversal of platelet and clotting pathway deficiencies through periprocedural transfusion optimisation may improve clinical outcomes for TAE in GIB; specifically, 30-day mortality.

We accept there are several limitations to our study. Firstly, due to a small number of reintervention cases within our cohort a multivariate analysis for this outcome was deemed unsuitable, therefore potential confounders were not accounted for. As a retrospective analysis, we relied on prior TAE procedural documentation accuracy, and also limited by using electronic laboratory data, where patients only having results < 1 year from date of TAE may misrepresent the actual acute haemoglobin decline. TAE was performed by seven different consultant interventional radiologists, with no standardised protocol or method of which embolisation material to use; although only a minority of patients received Gelfoam^TM or Embozene^TM. Furthermore, although our sample size is favourable in comparison to previous studies, it may particularly affect the reintervention analysis as the low event rate increases the probability of type I and II errors.Therefore, a larger study needs to be performed before conclusions can be drawn.

In conclusion, we demonstrate excellent technical and clinical success in TAE for acute GIB. We identified several pre-intervention haematological risk factors for clinical success following TAE in acute GIB. Haemoglobin baseline drop > 40 g l⁻¹ was associated with reintervention with rebleeding. Thrombocytopaenia and coagulopathy were individual risk factors associated with 30-day mortality. Recognition and timely reversal of these haematological deficits may improve post-procedural clinical outcomes in TAE for acute GIB.