Pancreatic arteriovenous malformations as a manifestation of hereditary haemorrhagic telangiectasia (Rendu-Osler-Weber disease): a systematic review

Abstract

Objectives

Limited data are available on pancreatic arteriovenous malformations (AVMs) in patients with hereditary haemorrhagic telangiectasia (HHT), also known as Rendu-Osler-Weber disease. This systematic review aimed to identify all reported cases of pancreatic AVMs in patients with HHT and summarise patient demographics, clinical presentation, diagnostic modalities, treatment strategies and follow-up.

Design

Systematic review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Data sources

Databases PubMed, Embase, Cochrane Library and Web of Science were systematically screened up to June 2025.

Eligibility criteria

All English-language, full-text studies reporting pancreatic AVMs in patients with HHT. Only patients with definite diagnoses of HHT and pancreatic AVM were included.

Data extraction and synthesis

All identified articles were independently screened by two reviewers. Data regarding patient demographics, clinical presentation, diagnostic modalities, treatment strategies and follow-up were collected. The mean AVM size was calculated when individual AVM size and number of AVMs were reported.

Results

Of 429 screened records, 46 articles underwent full-text assessment. Nineteen patients from 13 studies met the inclusion criteria. Most patients were female (58%), with a mean age of 52 years, presenting asymptomatically (50%) without laboratory findings (73%). CT was the most frequently used diagnostic modality (56%), and the mean AVM size was 7.1 mm. Most patients were treated conservatively (69%). Prognosis was frequently unavailable (79%). Differential diagnoses consisted mostly of neuroendocrine tumours (75%).

Conclusions

Pancreatic AVMs in patients with HHT appear to be smaller and are more frequently asymptomatic compared with the general population. Severe complications, such as portal hypertension and gastrointestinal bleeding, were less frequent. Conservative management was most common. Additional studies are needed to better characterise prognosis and treatment outcomes, particularly in symptomatic patients.

PROSPERO registration number

CRD420251077109.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Hereditary haemorrhagic telangiectasia (HHT) is a rare autosomal dominant vascular disorder associated with arteriovenous malformations (AVMs) in multiple organs, including the pancreas.

• Data on pancreatic AVMs specifically in patients with HHT remain limited to isolated case reports.

WHAT THIS STUDY ADDS

• This is the first systematic review to characterise pancreatic AVMs in HHT, identifying 19 patients across 13 studies.

• Compared with general pancreatic AVMs, HHT-associated pancreatic AVMs appear to be significantly smaller (7.1 mm vs 32.1 mm), more frequently asymptomatic (50% vs 5%) and less likely to require intervention (31% vs 71%).

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Based on the currently available evidence, pancreatic AVMs in HHT appear to follow a more indolent course, suggesting that conservative management with surveillance may be considered in asymptomatic cases.

• Larger prospective studies are needed to establish evidence-based treatment guidelines for this rare manifestation of HHT.

Introduction

Hereditary haemorrhagic telangiectasia (HHT) is a rare autosomal dominant vascular disorder, also known as Rendu-Osler-Weber disease. This disease is characterised by arteriovenous malformations (AVMs) and telangiectasias, with an estimated incidence of 10–20 per 100 000 individuals.¹ Early descriptions focused on clinical manifestations, while Halpern et al first demonstrated visceral AVMs using angiography.² Diagnosis is based on the Curaçao criteria, which include symptoms of (1) spontaneous and recurrent epistaxis, (2) characteristic mucocutaneous telangiectasias (lips, oral cavity, face and fingers), (3) visceral (brain, lung, liver, bowel) AVMs and (4) diagnosis of HHT in a first-degree relative using these same criteria.³ A definite diagnosis requires fulfilment of at least three of these criteria. Visceral AVMs may occur in multiple organs, most commonly the lungs, brain, liver and gastrointestinal (GI) tract, including the pancreas. Associated complications depend on the affected organ and may include haemorrhage, hypoxaemia, stroke, seizures and/or portal hypertension.¹ The underlying pathophysiology involves a loss of regulatory sphincteric control at the arteriolar–capillary junction that may permit unregulated arterial flow into capillaries and venules. This leads to the formation of arteriovenous shunts and progressive dilatation, elongation and tortuosity of the involved vessels.⁴ The causative mutations affect genes involved in endothelial cell growth and vascular integrity, such as endoglin (ENG), activin A receptor-like type 1 (ACVRL1), SMAD family member 4 (SMAD4), growth differentiation factor 2 (GDF2) and RAS p21 protein activator 1 (RASA1).⁵ Pancreatic AVMs are a rare but possible, manifestation of HHT. In image-based cohort studies of asymptomatic patients with HHT, pancreatic involvement has been reported in approximately 16–30% of cases.^6–8 However, they can also occur independently of HHT, with an estimated incidence of 0.9%.⁹ Most pancreatic AVMs (90%) are congenital, with only 10–30% associated with HHT.¹⁰ Other causes include chronic pancreatitis, pancreatic tumours or pancreatic transplantation surgery.¹¹ Complications of pancreatic AVMs can be severe and include GI bleeding or treatment-resistant portal hypertension.¹¹ Hyperkinetic circulation due to the shunting of a pancreatic AVM can lead to portal hypertension, which can be persistent even after surgical removal of the AVM.¹⁰ There are several treatment modalities for pancreatic AVMs, ranging from conservative treatment to radiological embolisation of the feeding artery, surgical occlusion of the feeding artery by clipping or ligation, or (partial) pancreatectomy.¹⁰ Most patients with general pancreatic AVMs receive surgical treatment, which consists of a pylorus-preserving pancreaticoduodenectomy for head lesions, distal pancreatectomy for body and tail lesions and total pancreatectomy for lesions of the entire pancreas. Surgical resection is mostly opted for in symptomatic patients, given the serious consequences of GI bleeding and portal hypertension. However, due to the possible complication of therapy-resistant portal hypertension, some authors suggest early surgical resection of larger pancreatic AVMs.¹⁰ Nevertheless, these surgeries are associated with complications and consequences including diabetes mellitus and pancreatic exocrine insufficiency.¹² Embolisation represents another treatment option; however, a recurrence rate of 37% has been reported,¹² due to growth of new collateral vessels. Other complications are haemorrhage, the development of portal hypertension and necrosis if multiple vessels are embolised.¹⁰ Differential diagnosis of a pancreatic AVM consists of hypervascular neoplasms, such as neuroendocrine tumours or other rare carcinomas.

Literature focusing on the presentation and treatment of pancreatic AVMs in patients with HHT is limited due to the low prevalence of this disease. Nonetheless, pancreatic AVMs in this patient population can lead to potentially serious complications, which may be prevented with appropriate treatment. Moreover, earlier research showed pancreatic AVMs may represent the sole visceral manifestation of HHT.⁸ Therefore, it is important to gain more evidence and knowledge regarding the diagnosis, treatment and prognosis of these pancreatic manifestations of HHT. The aim of this systematic review of the existing literature is to identify all reported cases of pancreatic AVM in HHT and summarise data on patient demographics, clinical presentation, imaging, treatment and outcomes. Improved understanding of these pancreatic abnormalities may raise awareness among clinicians, optimise diagnostic and therapeutic approaches, and ultimately improve patient outcomes and quality of life.

Methods

Study design

This systematic review was conducted using a search strategy and with the help of professional librarians. The following databases were searched: PubMed, Embase, Cochrane Library and Web of Science. Articles published before 9 June 2025 were considered eligible. No restrictions were applied regarding the start date of publication. The full search strategy can be found in online supplemental table 1.

All identified articles were independently screened by two reviewers. Disagreements were resolved through discussion and consensus. All selected articles from the screening were read in full. The articles that met the inclusion criteria were included in the systematic review—those that underwent full-text screening but were excluded are listed in online supplemental table 2, along with reasons for their exclusion. Additional articles were sought by scanning the references of the included articles to identify potential articles that were missed in the search (snowball strategy). A flowchart of the inclusion process is presented in figure 1. This study is reported in accordance with the PRISMA guidelines (a completed PRISMA checklist presented in the online supplemental materials). The study was registered with PROSPERO (CRD420251077109). A full written protocol was not prepared.

Figure 1. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart. HHT, hereditary haemorrhagic telangiectasia; P-AVM, pancreatic arteriovenous malformation.

Eligibility criteria

Studies eligible for inclusion in this review were case–control studies, cohort studies, case series and case reports. Only English-written, full-text and human-based studies were included. Additional inclusion criteria required patients to have both a confirmed diagnosis of HHT and a pancreatic AVM. The criterion for HHT was either a known diagnosis of HHT based on medical records or a new, definite diagnosis of HHT based on the above-mentioned Curaçao criteria.

The criterion for a pancreatic AVM was an image-based diagnosis of AVM done by a certified radiologist, based on typical imaging findings on CT imaging, angiography, MRI or ultrasound. On contrast-enhanced CT, nodular enhancement is typically observed, accompanied by early portal venous opacification.¹¹ On angiography, a racemose vascular network can be visualised with dilated and tortuous feeding arteries and early venous filling.¹⁰ On MRI, the pancreatic AVM is visible as a circular lesion with signal void in the T1 and T2 images.¹³ On ultrasound, a hypoechoic or anechoic lesion with a mosaic appearance may be observed, accompanied by a pulsatile portal venous waveform on Doppler examination.¹¹ There were no age limitations on the inclusion of patients. Studies could be included also if the study cohort investigated the broader context of AVMs or HHT, but individual data regarding a pancreatic localisation of the AVM in patients with HHT could be extrapolated from the study data.

Patients were excluded if the pancreatic vascular abnormality turned out to be a diagnosis different from AVM. Furthermore, patients were excluded if the pancreatic AVMs had a different origin than HHT, such as congenital origin or iatrogenic origin due to pancreatic surgery or secondary to trauma, pancreatitis or tumour. An overview of the selected articles and their main results can be found in table 1.

Table 1. Overview of the included articles and their main findings.

Author, year, country	Number of patients (n=19)	Number of AVMs (n=17)	Age (years)	Gender	Symptoms	Laboratory findings	Image modality	Size AVM (n=15) in mm	Image findings	Treatment	Prognosis	Differential diagnosis
Ogawa et al 2009, Japan13	1	2	41	Female	GI bleeding, oesophageal varices *	Mild anaemia	CT	3.0 mm and 9.0 mm	Two AVMs. One of 3 mm in the tail of the pancreas with early enhancement of the portal vein and/or its branch veins, conglomeration of strong nodular stains in the arterial phase. Feeding artery: anterior sup. pancreaticoduodenal artery and inf. pancreaticoduodenal artery. The second one of 9.0 mm in the head of the pancreas with the same findings and additional portal collateral pathways such as oesophageal varices. Feeding artery: dorsal pancreatic artery.	Clipping	NA	Aneurysm
Siddiki et al 2008, USA1	2	1	55	Male	NA	NA	CT	NA	AVM in the pancreatic head	NA	NA	NA
		1	53	Male	NA	NA	CT	NA	Pancreatic AVM in the tail of the pancreas		NA	NA
Cooper et al 2025, USA26	1	1	35	Female	Asymptomatic	NA	EUS	8×10 mm	10 mm × 8 mm cystic lesion in the pancreatic head with high-intensity Doppler flow of multiple dilated vascular channels	Conservative	NA	NA
Lim et al 2012, Australia15	1	3	70	Female	Intermittent abdominal pain	Various tumour markers (Ca19.9) were within the reference range	EUS	8-10 mm for all three lesions	CT and EUS revealed a small vascular mass at the junction of head and body of the pancreas. MRI showed lesions consistent with pancreatic neoplasm, however, EUS showed three hypoechoic lesions in head, neck and body. The lesions were 8–10 mm in diameter and had a densely vascular pattern on colour Doppler.	Conservative	No changes in size/number of pancreatic AVM up to 2 years of follow-up on CT and EUS.	Neuroendocrine tumour
Lawhon 1980, USA20	1	NA	50	Female	GI bleeding	Persistent anaemia, Hematest-positive stool	Angiography	NA	AVMs present in the ileum and throughout the entire pancreas.	Surgical resection was rejected because of multiplicity of lesions and comorbidity	NA	NA
Chuang et al 1977, USA14	1	NA	56	Male	Massive haematemesis and melena	NA	Angiography	NA	Arteriography: dilated dorsal and transverse pancreatic arteries which led to an extensive racemose vascular network in the body and tail.	Distal pancreatectomy of the tail and body †	Recurrence of intra-abdominal haemorrhage after 2 weeks	NA
Lacout et al 2010, France8	4	7	NA	NA	Asymptomatic	No elevated lipase levels	CT	Mean size: 5.7 mm±1.2 (SD)	NA	Conservative	NA	NA
Wee et al 2014, Singapore16	1	NA	57	Male	Abdominal pain§	NA	EUS	3-10 mm (number of AVMs not reported)	CT scan: incidental lesion in pancreas. CT scan thin cuts: multiple pancreatic enhancing lesions (0.3–1 cm). EUS: anechoic lesions with positive Doppler flow with supply from the superior mesenteric artery.	Conservative	NA	Neuroendocrine tumour
Demirpolat et al 2009, Turkey18	1	1	52	Female	Worsening right-sided heart failure ¶	Anaemia, elevated ALP and gamma-GT	Doppler ultrasound	10 mm	CT late arterial phase: hypervascular lesion with microlobulated contours in the head of the pancreas. Colour Doppler sonography: small cysts in the pancreas of vascular nature with mosaic-like structure, supplied by the dorsal pancreatic artery.	NA	NA	Neuroendocrine tumour
Barral et al 2012, France6	3	NA	NA	NA	Asymptomatic	All patients had normal hepatic (ALAT, ASAT, ALP, gamma-GT and total bilirubin) and pancreatic (lipase, amylase and serum glucose) function tests.	CT	NA	NA	NA	NA	NA
Chang et al 2011, South Korea17	1	NA	37	Female	Chronic fatigue **	Anaemia, elevated ALP and gamma-GT. Mutation in ACVRL1 gene.	NA	NA	NA	Conservative	No symptoms up to 12 months after diagnosis	NA
Gomes et al 1983, USA19	1	1	52	Male	Recurrent GI bleeding, recurrent pancreatitis	NA	Angiography	NA	Large AVM involving the pancreas and duodenum. Supplied by sup. and inf. pancreaticoduodenal arteries, dorsal pancreatic artery and several jejunal branches.	Embolisation of dorsal pancreatic artery and jejunal artery feeders.††	NA	NA
Moore et al 1976, USA21	1	NA	68	Female	NA	NA	Angiography	NA	AVMs of stomach, duodenum, pancreas and liver	Therapeutic trial of diethylstilbestrol and medroxyprogesterone acetate	No substantial haemorrhage up to 4 years after diagnosis	NA

^*Oesophageal varices were considered to be due to liver cirrhosis.

^†This patient also had a liver cirrhosis with varices in the spleen, for which a splenectomy had been recently performed.

^‡Pathology after resection revealed diffuse AVM and slight focal chronic pancreatitis in the body and tail.

^§This patient had a concomitant diagnosis of acute cholecystitis.

^¶This patient had a concomitant diagnosis of a liver AVM.

^**This patient had a concomitant diagnosis of pulmonary arterial hypertension due to multiple AVMs in the lung.

^††The embolisation was complicated by duodenocolic fistula for which pancreaticoduodenectomy, transverse colectomy with primary anastomosis, and partial gastric resection were performed.

ALAT, alanine transaminase; ALP, alkaline phosphatase; ASAT, aspartate transaminase; AVM, arteriovenous malformation; gamma-GT, gamma-glutamyl transpeptidase; GI, gastrointestinal; HHT, hereditary haemorrhagic telangiectasia.

Data items and risk of bias

The included articles were reviewed for information regarding title and authors, year of publication, number of patients with HHT and pancreatic AVMs, age, gender, symptoms, laboratory finding, imaging, treatment and prognosis. Given that all included studies consisted of case reports and small case series, a formal quantitative risk-of-bias assessment was not performed.

Statistics

Age (in years) and AVM size (in mm) were the only continuous variables and were summarised as means. The mean AVM size was calculated if both the AVM size and number of AVMs were reported. When individual AVM sizes were reported, these values were used directly; when only ranges were available, midpoint estimates were applied; and when two dimensions were reported (eg, length and width), the largest dimension was used. Group-level means were weighted by the number of lesions. The overall mean AVM size was calculated across all 14 lesions.

Patient and public involvement

No patients or members of the public were involved in the design, conduct or reporting of this systematic review. The research question and study design were developed by the research team based on the identification of a gap in the existing literature regarding pancreatic AVMs in HHT. Given the systematic review methodology and the rare nature of the condition studied, direct patient and public involvement was not feasible.

Results

After applying the search strategy, a total of 718 articles were identified. Two articles were identified using the snowball strategy. Following removal of duplicates, 429 articles remained for screening. Titles and abstracts of these articles were screened, after which 46 full-text articles were assessed for eligibility. Of these, 13 studies met the inclusion criteria and were included in the systematic review. From the included studies, individual clinical information could be extracted for 19 patients who met the inclusion and exclusion criteria. The patient characteristics and main results of the included patients are presented in table 2.

Table 2. Patient characteristics.

Patient characteristics	Patients with HHT and pancreatic AVM (n=19)
Age (years)	52 (range 35–70)
Missing	7
Gender
Female	7 (58%)
Male	5 (42%)
Missing	7
Symptoms
Asymptomatic	8 (50%)
Gastrointestinal bleeding	4 (25%)
Abdominal pain	2 (13%)
Other: chronic fatigue and worsening of right-sided heart failure	2 (13%)
Missing	3
Laboratory findings
No laboratory abnormalities	8 (73%)
Anaemia with elevated gamma-GT and ALP	2 (18%)
Anaemia	1 (9%)
Missing	8
Imaging modality for definitive diagnosis
CT	10 (56%)
Angiography	4 (22%)
Endoscopic ultrasound	3 (17%)
Transabdominal Doppler ultrasound	1 (6%)
Missing	1
Location of pancreatic AVM*
Head	3 (38%)
Neck	0 (0%)
Body	0 (0%)
Tail	1 (13%)
Multiple segments	4 (50%)
Missing	11
Size AVM (mm)*	7.1 (range 3.0–10)
Missing†	10
Treatment
Conservative	9 (69%)
Pancreatectomy	1 (8%)
Clipping	1 (8%)
Embolisation‡	1 (8%)
Pharmacological§	1 (8%)
Missing	6
Prognosis
Follow-up without recurrence
1 year	1 (25%)
2 years	1 (25%)
4 years	1 (25%)
Recurrence of symptoms	1 (25%)
Missing	15
Differential diagnosis
Neuroendocrine tumour	3 (75%)
Aneurysm	1 (25%)
Missing	15

Patient characteristics, presented as percentages from the total group of included patients (n=19), excluding the missing patients per category. Age and size of the AVM were presented as a mean with range.

^*Mean size of AVM was calculated if both the number and size of AVMs were available (calculated for 14 AVMs in 8 patients).

^†Reported for total number of patients (n=19).

^‡Due to recurrence of gastrointestinal bleeding, eventually a pancreaticoduodenectomy was performed.

^§This patient was given hormonal treatment (diethylstilbestrol and medroxyprogesterone acetate).

ACVRL1, activin A receptor-like type 1; ALAT, alanine transaminase; ALP, alkaline phosphatase; ASAT, aspartate transaminase; AVM, arteriovenous malformation; AVM, arteriovenous malformation; ENG, endoglin; EUS, endoscopic ultrasound; gamma-GT, gamma-glutamyl transpeptidase; GDF2, growth differentiation factor 2; HHT, hereditary haemorrhagic telangiectasia; HHT, hereditary haemorrhagic telangiectasia; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; RASA1, RAS p21 protein activator 1; SMAD4, SMAD family member 4.

Patient characteristics: age and gender

Age and sex were reported for 12 patients. Among these, five patients were male (42%), and seven patients were female (58%). The mean age was 52 years (range 35–70 years).

Clinical presentation and comorbidity

Clinical symptoms related to pancreatic AVM were reported for 16 patients. Data were missing for three patients. Eight patients (50%) were asymptomatic. If symptomatic, GI bleeding was frequent, with four patients (25%) presenting with haematemesis or melena. Two of these patients had concomitant oesophageal varices.^{13 14}

Two patients (13%) presented with abdominal pain, one presenting as intermittent pain in the right upper quadrant for 18 months.¹⁵ The second patient had a concomitant diagnosis of acute cholecystitis.¹⁶

Another patient presented with chronic fatigue, which was possibly attributed to pulmonary arterio-venous shunting resulting from a concomitant pulmonary AVM.¹⁷ A different comorbidity that was found in the study population was worsening right-sided heart failure, which the authors attributed to concomitant liver involvement of HHT.¹⁸

Laboratory findings

Laboratory results were reported for 11 patients, of which the majority (73%) demonstrated no laboratory abnormalities. Reported normal findings included lipase levels (n=4), hepatic and pancreatic enzymes (including ALAT, ASAT, gamma-GT, ALP, total bilirubin, lipase, amylase, serum glucose, n=3) and tumour markers (n=1). Three patients had abnormal laboratory results, all of whom were anaemic. In two patients (18%), anaemia was accompanied by elevated gamma-GT and ALP.^{17 18} The type of mutation was available for one patient, who carried a mutation in the ACVRL1 gene.¹⁷

Imaging modality and findings

The definite diagnosis of pancreatic AVM was most frequently established using contrast-enhanced CT imaging, which was performed in 10 patients (56%). In four patients (22%), angiography provided the final diagnosis of AVM, and in three patients (17%), an endoscopic ultrasound was conclusive. For one patient (6%), a Doppler ultrasound was used to get a definitive diagnosis,¹⁸ and for one patient, information about the imaging modality was not specified.¹⁷

The anatomical location of the pancreatic AVM was reported for eight patients. Most frequently, multiple segments of the pancreas were affected (50%, n=4). The pancreatic head was affected in three patients (38%), followed by the tail in one patient.¹ For 11 patients, the location of the AVM was missing in the articles.

For eight patients with 14 AVMs, both the size and number of AVMs were available, showing a mean AVM size of 7.1 mm (range 3.0–10.0). Wee and Fernandes reported an AVM size range of 3.0–10.0 mm; however, as the number of lesions was not specified, this source was not included in the quantitative synthesis.¹⁶

In four patients, detailed information regarding the supplying artery of the AVM was provided. In one patient, an AVM in the tail of the pancreas was supplied by the anterior superior pancreaticoduodenal and inferior pancreaticoduodenal artery. This patient had a second AVM in the head of the pancreas with the dorsal pancreatic artery as the feeding artery.¹³ Another patient was also diagnosed with an AVM in the pancreatic head, supplied by this dorsal pancreatic artery.¹⁸

A different study described a patient with an AVM supplied by the superior mesenteric artery.¹⁶ In a fourth patient, a large AVM was found which involved the pancreas and duodenum, with the dorsal pancreatic artery, inferior pancreaticoduodenal artery and several jejunal branches as feeding arteries.¹⁹ Early enhancement in the venous phase of the AVM was described in one patient, which consisted of the portal vein and/or its branches.¹³

Differential diagnosis

Initial differential diagnoses were reported for four patients. For one patient (25%), the pancreatic AVM was initially presumed to be an aneurysm based on angiography.¹³ For three patients (75%), the differential diagnosis consisted of a neuroendocrine tumour, due to the hypervascular character of these tumours. In one of these patients, Doppler ultrasound identified the lesion as an AVM, due to the absence of a mass lesion and the presence of early draining into the venous system.¹⁸ In the second patient, the suspicion of a neuroendocrine tumour was raised based on a vascular mass on CT. However, an endoscopic ultrasound (EUS) with colour Doppler revealed an AVM.¹⁵ For the third patient, a neuroendocrine tumour was initially suspected due to the presence of a pancreatic lesion on CT. An EUS eventually revealed a pancreatic AVM.¹⁶

Treatment and prognosis

Treatment strategies were reported for 13 patients. Most patients (69%, n=9) were treated conservatively. For one patient, the research specifically mentioned that surgical intervention was rejected due to the extensiveness of the AVMs and comorbidity of the patient.²⁰ For one of the conservatively treated patients a follow-up of 2 years was mentioned, without change in size or quantity of the AVMs based on CT and EUS imaging.¹⁵ Another conservatively treated patient had no symptoms within a follow-up period of 12 months after diagnosis.¹⁷

Two patients received a radiological intervention, of whom one underwent embolisation that eventually led to recurrence of GI bleeding due to a duodenocolic fistula. Therefore, finally a pancreaticoduodenectomy was performed. The follow-up of this case was not mentioned.¹⁹ The second patient was treated with clipping, of which the outcome was likewise not mentioned.¹³

One patient received a distal pancreatectomy of the pancreatic tail and body and developed recurrence of intra-abdominal haemorrhage. This patient had a concomitant diagnosis of liver cirrhosis.¹⁴

Pharmacological treatment was given to one patient, consisting of hormonal treatment with diethylstilbestrol and medroxyprogesterone acetate as part of a therapeutic trial. A 4-year follow-up showed no substantial haemorrhage.²¹

Discussion

This study represents the first systematic review specifically addressing pancreatic AVMs in patients with HHT. While previous studies have investigated pancreatic AVM in the general population or other visceral AVMs in patients with HHT, only isolated cases of pancreatic involvement in HHT have been reported to date. This research aimed to gather all available evidence regarding pancreatic AVMs in HHT published before June 2025 in order to provide a comprehensive overview for clinicians.

A total of 19 patients with HHT and pancreatic AVMs were identified across 13 studies. Most patients were female with a mean age of 52 years. Notably, a substantial proportion of patients were asymptomatic. If symptomatic, GI bleeding was most frequent. Laboratory findings were mostly normal, and diagnosis was made mainly based on typical imaging findings at CT. AVMs were present in multiple segments of the pancreas, and the mean AVM size was 7.1 mm. Most AVMs were treated conservatively. If treated, an embolisation, clipping, pancreatectomy or pharmacological treatment was used. The prognosis was frequently not provided in the included articles.

In this study, 50% of the patients were asymptomatic, compared with 5% in patients with general AVMs in the study by Onozawa et al,¹² who conducted a systematic review in 2023 in 117 patients with pancreatic AVMs. Consequently, most of their patients were treated (91%), compared with 31% in the current study. This could be attributed to the search strategy of the aforementioned study, which focused on the treatment of pancreatic AVM.¹² However, a separate study by Chou et al reviewed 89 cases of general pancreatic AVM and found a comparable 71% of patients receiving treatment. Another explanation may be that pancreatic AVMs in HHT carry a lower risk, due to differences in size or extensiveness of the AVM. To illustrate, in this study the mean size of the AVMs reported was 7.1 mm (range 3.0–10 mm) in HHT patients, compared with 32.1 mm (range 3.0–97.3 mm) in patients with general pancreatic AVM.¹³

The demographics are largely consistent with earlier studies into pancreatic AVMs in general. To illustrate, the cohort in the study by Onozawa et al consisted mostly of males (87%) with a mean age of 51 years.¹² The location of the AVM seems to be comparable to earlier research, with most of the general pancreatic AVMs appearing in the head of the pancreas (64%).¹²

Symptoms such as GI bleeding (25%), abdominal pain (13%) or portal hypertension (0%) appear to be less prevalent than the general pancreatic AVM population with a prevalence for GI bleeding of 42.7%, pain of 50.5%¹² and portal hypertension of 6.7%.²² GI bleeding has been proposed to have three different origins: (1) bleeding into the pancreatic duct and subsequently into the duodenum, (2) an association with concomitant intestinal arteriovenous malformation with direct bleeding, (3) portal hypertension as a consequence of the AVM which leads to oesophageal varices.^{10 20} The latter seems not to appear in HHT patients with pancreatic AVM, while the prevalence in general pancreatic AVM patients has been described up to 26%.¹⁰

Another complication described in general pancreatic AVMs is acute pancreatitis, with a prevalence of 12–17%, and which is particularly associated with AVMs in the pancreatic tail.^{22 23} The underlying pathophysiology proposed is that either ischaemia or bleeding due to the AVM causes pancreatitis.²³ The opposite is also observed, in which a pancreatic AVM is acquired due to the inflammation and auto-digestion associated with pancreatitis.²⁴ However, in the current study no acute pancreatitis as a complication or underlying cause was mentioned for the included HHT patients.

The mutation status of the HHT patients was only available for one patient. Earlier research did not find a significant relationship between the specific mutations and presence of pancreatic involvement in patients with HHT. However, the majority of their cohort with activin type-II-like receptor kinase 1 (ALK1) genetic mutations had pancreatic involvement.⁸

Angiography has been long stated as the gold standard for diagnosing pancreatic AVM, for its ability to depict the feeding arteries for eventual treatment.^{14 25} In some cases in the current study, EUS was used to depict the AVM, showing a hypoechoic or anechoic lesion with positive Doppler flow,^{15 16} and, in one case, multiple dilated vascular channels.²⁶ However, both imaging modalities have an invasive nature, and other studies have designated CT as the optimal imaging modality in depicting pancreatic AVMs.^6–8 Also in the current study, most AVMs were identified using CT.

The included patients had a differential diagnosis of neuroendocrine lesions and aneurysm, which are both described in existing literature. The differential diagnosis of a pancreatic AVM further consists of other hypervascular tumours, such as solid variants of serous cystic neoplasia, acinar cell carcinoma and angiosarcoma. Although these tumours may exhibit opacification resembling that of arteriovenous malformations, the associated draining veins are not dilated and lack early venous filling. Persistent tumour enhancement is commonly observed in the venous phase, whereas AVMs show early wash-out.¹⁸

Onozawa et al recommended conservative treatment with close follow-up for asymptomatic patients, which is in line with the patients in our study who did not exhibit symptoms.¹² Unfortunately, prognosis and follow-up data were insufficiently reported both in the included cases and in the existing literature on pancreatic AVMs, precluding comparison of treatment modalities. In our study, one patient received pharmacological hormone treatment, which was historically believed to prevent bleeding in the GI tract, but later randomised trials failed to demonstrate a significant benefit.²⁷

This study has several limitations. First, as a systematic review primarily based on case reports and small case series, the findings may be subject to publication bias. Moreover, the case reports of patients with pancreatic AVM and HHT diagnosis were somewhat limited and not all clinically relevant information was available, complicating the possibility of drawing firm conclusions.

Nevertheless, this is the first systematic review specifically addressing pancreatic AVMs in patients with HHT, therefore taking the first step into facilitating clinical guidance in this rare, but potentially dangerous manifestation of Rendu-Osler-Weber disease.

In conclusion, pancreatic AVMs are a rare, but potentially dangerous manifestation of HHT disease. Most AVMs appear to be asymptomatic and smaller compared with general pancreatic AVMs. Moreover, severe symptoms and complications such as GI bleeding or portal hypertension were less frequently observed. Therefore, asymptomatic patients were mostly treated conservatively with follow-up of the lesions. In contrast, in symptomatic patients, treatments such as surgical resection or embolisation were performed. However, the available evidence is limited, and additional well-documented case reports and larger studies are needed to improve understanding and support firm treatment recommendations.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplemental information.