Anatomy, physiology and aetiology

Three main aortic branches provide the arterial blood supply to the gastrointestinal tract: the coeliac artery (CA), superior mesenteric artery (SMA) and inferior mesenteric artery (IMA). The general gastrointestinal vascular anatomy dictates that the CA supplies stomach, liver, part of the pancreas, and the proximal part of the duodenum. The SMA supplies the distal part of the duodenum, the entire small bowel and the proximal colon, and the IMA is relatively small and supplies the distal colon; see figure 1. Branches of these main gastrointestinal arteries enter the serosa of the gut on the mesenteric side to form a serosal vascular plexus around the intestine. A submucosal plexus is formed after penetration of the intestinal wall. From here, arterioles penetrate the muscularis mucosa to the superficial mucosal layers. At the mucosal tip they branch into an intense capillary network of capillaries and venules. In a fasting state under basal conditions, approximately 20% of the cardiac output flows through the gastrointestinal arteries. The CA receives 800 ml blood/min, dividing this flow into 350 ml/min to both the splenic and hepatic arteries. The remaining 100 ml/min supplies the stomach, duodenum and part of the pancreas. Postprandially, the flow in the CA increases by 30% to 1100 ml/min. The SMA, which has a similar diameter to that of the CA, has a basal flow of 500 ml/min, which increases by more than 150% after a meal, with volumes up to 1400 ml/min. The IMA is relatively small in diameter as compared to the CA and SMA; see figure 1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Overview of the blood circulation to and from the digestive tract. The blood is supplied to the digestive tract through the coeliac artery, superior mesenteric artery and inferior mesenteric artery. There are some important anastomoses of the mesenteric arteries, which can compensate the blood supply in case of stenosis or occlusion of one of the arteries. (1) Collateral circulation from the mammary artery to the coeliac artery, (2) an anastomosis between the superior pancreaticoduodenal artery from the coeliac artery, and the inferior pancreaticoduodenal artery from the superior mesenteric artery (anastomosis of Bühler), (3) an anastomosis of the superior mesenteric artery and inferior mesenteric artery (arc of Riolan or meandering mesenteric artery) and (4) an anastomosis between the superior rectal artery of the inferior mesenteric artery and the rectal arteries from the internal iliac artery.

The gastrointestinal arterial circulation is provided with an abundant collateral network. It is generally assumed that this collateral circulation of the arterial mesenteric tract prevents clinically relevant gastrointestinal ischaemia in the majority of cases with single- or multi-vessel gastrointestinal arterial stenosis.

The aetiology of the majority of stenoses of the gastrointestinal arteries can be divided into concentric and eccentric diseases; see box 1. The coeliac artery compression syndrome (CACS) is the primary cause of gastrointestinal arterial stenosis in younger patients. In elderly patients, atherosclerotic disease is the major cause of gastrointestinal arterial stenosis. Non-concentric intra-arterial disease due to arterial thrombosis or dissection causes significant stenoses in a minority of cases.

Epidemiology of gastrointestinal arterial stenosis

Autopsy studies have provided data about the prevalence of, mostly, asymptomatic gastrointestinal arterial stenosis. Critical narrowing, defined as a luminal occlusion of >50%, of either the CA, SMA or IMA was frequently found, with a prevalence ranging from 11 to 21% in two autopsy series with in total 313 subjects included, with age of death ranging from 19 to 97 years.9 10 A more recent Finnish autopsy study, including 120 subjects, reported a 29% prevalence of stenosis of either the CA or SMA, with the CA being the most common affected site.11 In 15% of all cases at least two gastrointestinal arteries were affected. The latter study also showed a strong correlation between gastrointestinal arterial stenosis and ageing: in subjects younger than 40 years the prevalence of gastrointestinal arterial stenosis was 6%, rising to 14% between the age 40 and 59, and to 67% in subjects aged 80 years or more.11 A large study reviewing abdominal angiographies of 713 patients showed stenosis or occlusion of at least one of the gastrointestinal arteries in 17% of cases.12 Recent studies using modern imaging modalities of the gastrointestinal arteries also show a high prevalence of, often asymptomatic, gastrointestinal arterial stenosis. In a prospective series of asymptomatic patients undergoing gastrointestinal angiography for chemoembolisation of hepatic tumours, 7% of patients were found to have a significant stenosis of the CA.13 In a recent study in a large cohort of asymptomatic elderly Americans with a mean age of 77 years, the prevalence of stenosis in the CA and SMA was evaluated using abdominal duplex ultrasound measurements. In 18% of these subjects, a stenosis of the CA and/or SMA was found. Eighty-six per cent of these stenoses were isolated CA stenoses, 2% isolated CA occlusions, 5% isolated SMA stenoses, and 7% combined CA and SMA stenoses.14 In another report using abdominal duplex ultrasound measurements in asymptomatic subjects, a 3% prevalence of CA stenoses was reported in subjects <65 years of age, rising to 18% in subjects >65 years of age. Among patients with an isolated stenosis, 81% of lesions were present in the CA, and 19% in the SMA.15 Taken together, the latter studies suggest that the prevalence of asymptomatic gastrointestinal arterial stenosis in the general population ranges between 6 to 29% and increases with age.

The exact prevalence of symptomatic gastrointestinal arterial disease is currently unknown. This is due to a variety of reasons including lack of symptom specificity, limited awareness of the spectrum of disease and the persistence of old theorems on symptoms and the prerequisite of multi-vessel disease. Furthermore, the former need for invasive testing, the lack of functional tests, and the invasive character of current therapies also interfere in this process. Nevertheless, new data will provide actual insights into the prevalence of symptomatic gastrointestinal arterial disease.

Coeliac artery compression syndrome as a cause of CGI

The main cause of single artery stenosis is extrinsic compression due to CACS, also known as median arcuate ligament syndrome, or Dunbar's syndrome. This condition was first described by Lipshutz in 1917, followed decades later by Harjola (in 1963) and Dunbar et al (in 1965).16–18 The median arcuate ligament usually passes over the aorta at the level of the first lumbar vertebral body, superior to the origin of the coeliac artery (see figure 2A). The actual cause of clinical symptoms in the CACS is still discussed. Currently, two main theories are used to explain the pathogenesis of symptoms. According to the first hypothesis, an anomalous fibrous diaphragmatic band compresses the coeliac axis in patients with a relatively low insertion of the diaphragm, especially during expiration with partial relief during inspiration. This extrinsic compression limits the inflow of blood, causing mucosal ischaemia. According to the second theory, the pain results from stimulation of the coeliac plexus either by mesenteric vasoconstriction or local irritation.19 CACS is typically diagnosed in women under the age of 50 years. A significant compression of the CA by the median arcuate ligament of the diaphragm can give rise to symptoms of CGI (see figure 2B). In a recent prospective study concerning 43 CACS patients with a mean age of 38 years (range 14–73), the most predominant symptoms were postprandial pain (present in 80% of patients), weight loss (77%), and exercise-related pain (40%). An abdominal bruit on physical examination was found in only 23% of patients. A history of smoking was reported by 63% of patients, other cardiovascular risk factors were present in only 6%.8

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Coeliac artery compression syndrome. (A) A diagrammatic representation of the compression of the coeliac artery by the median arcuate ligament (a) or by the coeliac plexus (b). This is an adapted figure from ‘The clinical anatomy of coeliac artery compression syndrome’ by Loukas et al.19 (B) Angiography of the coeliac artery compression syndrome. Magnetic resonance angiography (a) shows a subtotal occlusion of the coeliac artery as indicated by the white arrow. In the digital substraction arteriography performed in the same patient, the compression of the coeliac artery is shown to be dependent on breathing, with a subtotal occlusion seen at expiration (b) (closed arrow) and a 60% stenosis seen at inspiration (c) (closed arrow). Also visible is the marked post-stenotic dilation of the coeliac artery (a and b) (open arrow).

In the past 50 years, various studies have shown variable results of treatment of CACS, with treatment mainly consisting of cleavage of the ligament.20–25 In small studies varying from 6 to 11 patients, the results of long-term follow-up after treatment were very disappointing, with persistent relief of symptoms in only 0–40% of patients.20 22 25 This led many to the belief that the syndrome was non-existent, or least not due to functional coeliac artery compression, and that symptom improvement in responders was actually due to pain blockade by denervation of the vagal nerve as a result of cleavage of the ligament. Two studies providing data on surgical treatment of CACS in larger cohorts with 20 and 51 patients showed more promising results. The latter two studies showed persistent symptom relief in 68 to 75% of patients.21 23 It should be borne in mind that diagnostic options at the time of these studies were very limited and patients were often diagnosed on the basis of symptoms and an abdominal bruit, in the absence of adequate visualisation techniques and functional tests. More recently, a prospective study applied gastric exercise tonometry as the key investigation for functional assessment of mucosal perfusion in patients with suspicion of CACS. In carefully selected cases with a typical abrupt, expiration-dependent, superior compression of the root of the coeliac trunk, combined with pathological functional test result by tonometry, surgical release of the CA led to persistent relief of symptoms on long-term follow-up in 83% of patients, compared with the 0–75% in earlier studies. The existence of impaired mucosal perfusion in CACS patients was even more underlined by the fact that all successfully treated patients showed improvement or normalisation of gastric exercise tonometry after treatment.8 This was thus the first study to show impaired mucosal perfusion in CACS, also associating symptom and functional improvement. These data are strongly supportive for the true-ischaemia hypothesis for CACS.

Single-vessel atherosclerotic CGI

Atherosclerotic narrowing of the gastrointestinal arteries is quite common. As mentioned earlier, solitary stenoses of the CA or SMA are found in 6–29% of asymptomatic subjects, rising to 69% at age >80 years. Currently, no data exist on the exact incidence of single-vessel CGI. However, one study presented data of a large cohort of CGI patients, and compared symptoms in both single- and multi-vessel disease patients. This showed similar symptoms in both groups, with the exception of weight loss which was significantly more prominent in the multi-vessel disease CGI patients.6 Although no studies have so far focused specifically on outcome of endovascular treatment of single vessel atherosclerotic CGI, some conclusions can be derived from published larger, combined single- and multi-vessel atherosclerotic CGI cases. In patients with an isolated >70% stenosis of either CA or SMA in combination with abnormal tonometry, endovascular therapy was associated with persistent relief/improvement of symptoms in 71 to 75% during a >1 year follow-up.6 7

Presenting symptoms

The clinical symptoms of CGI, whether caused by single- or multi-vessel arterial disease, and whether induced by atherosclerotic or eccentric disease, are similar.6 Postprandial pain, weight loss, exercise-related pain, and an abdominal bruit are presenting symptoms found in, respectively, 85%, 76%, 44%, and 25% of patients with CGI (see table 1).6 8 The postprandial symptoms occur typically within 15–30 min after a meal and last for 60–120 min. Recent studies have shown that only a minority of patients present with the classical ‘abdominal angina’ triad of postprandial pain, weight loss due to fear of eating, and an abdominal bruit.6–8 Other symptoms that are occasionally found include diarrhoea, malabsorption and nausea, often in combination with postprandial pain. During upper endoscopy, oedema and erythema of the gastric mucosa can be observed in 35% and 42% of patients, respectively. However, these findings are seldom discriminative to other pathological conditions. Other endoscopic findings such as atrophy of the duodenal mucosa, and non-Helicobacter pylori/non-NSAID gastric or duodenal ulcers, are observed in only a minority of patients with CGI (figure 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Endoscopic images of patients with chronic gastrointestinal ischaemia. Endoscopic images of the stomach (A–C) and duodenum (D–F) of patients analysed for chronic gastrointestinal ischaemia. In the majority of patients analysed for chronic gastrointestinal ischaemia no specific mucosal abnormalities are detected (A and D). Oedema and erythema of the gastric mucosa is observed in 35% and 42% of the patients with CGI, respectively. This was, however, shown not to be discriminative with other conditions. In 8%, a Helicobacter pylori-negative and NSAID-negative mucosal ulcer is observed in the antrum (B and C). The duodenum seldom shows pathology (D). In two patients with a total occlusion of the coeliac artery and the superior mesenteric artery ulcerations were seen in the antrum en corpus of the stomach (B and C), and whitening, and sometimes bluish colorisation, of the Kerckring folds (E and F).

Risk factors

The risk factors for atherosclerotic gastrointestinal arterial disease seem to be compatible with the known risk factors for cardiovascular disease. The Framingham Heart Study and other large-scale epidemiological studies have identified major cardiovascular risk factors such as hypertension, dyslipidaemia, glucose intolerance, obesity, hyperhomocysteinaemia and smoking.35–37 These factors also seem to predispose to gastrointestinal arterial disease and CGI. In a cohort study of 168 patients with CGI, the risk factors for gastrointestinal arterial atherosclerosis were compatible with the general established risk factors for arterial atherosclerotic disease. However, the female preponderance (65%) and the high prevalence of hyperhomocysteinaemia (45%) were remarkable. These same findings were recently reported by our group, confirming the female preponderance (66%) and atherosclerotic risk factors.38

Consensus diagnosis and multidisciplinary approach

The diagnosis of CGI is currently based on the outcomes of three main components. This first includes the medical history, symptom assessment, and physical examination of the patient. The second component consists of radiological visualisation of the gastrointestinal arteries, and the third component aims at functional testing of mucosal perfusion. Different diagnostic strategies have been suggested.6 39 40 The combination of gastrointestinal tonometry and duplex ultrasound of the abdominal arteries is currently suggested as the strategy with the highest diagnostic accuracy.39 As a referral centre, we use a standardised checklist for a patient's symptoms and a physical examination, visualisation of the gastrointestinal arteries by means of CT–angiography, and visible-light spectroscopy for detection of mucosal hypoxaemia. The outcomes of the different diagnostic tests are, without exception, discussed in a multidisciplinary team, with an interventional radiologist, a vascular surgeon and a gastroenterologist. The team thus comes to a consensus diagnosis on the presence or absence of CGI, as well as a treatment plan if applicable. The reason for this team approach lies in the multidisciplinary character of diagnosis and treatment of this condition, with the aim to optimise management.

Visualisation of gastrointestinal arteries and blood flow

Abdominal duplex ultrasound scanning, computed tomography angiography (CTA), and magnetic resonance angiography (MRA) are the current non-invasive diagnostic methods for gastrointestinal arterial stenosis. Angiography of the gastrointestinal arteries remains the ‘gold standard’ for diagnosing and staging of gastrointestinal arterial stenosis. The advantages of digital subtraction angiography (DSA) are its high sensitivity and specificity for detection of stenoses in the origin of the abdominal arteries, combined with the good quality of visualisation of the peripheral mesenteric vasculature, the possibility to assess any collateral circulation, and the ability to perform good quality imaging during inspiration and expiration. The latter is of special importance in patients suspected of CACS (figure 2B). Lateral (conventional) arteriography is the primary modality to detect extrinsic compression by the median arcuate ligament compatible with CACS. The syndrome is characterised by a sharp, superior impression of the root of the coeliac artery, which compression typically increases during expiration and decreases during inspiration (figure 2B). The main disadvantages of DSA are its invasive character, the lack of visualisation of surrounding tissues, and potential complications, in particular contrast allergy and renal function impairment.

Abdominal duplex ultrasound has a fairly good sensitivity and specificity for detection of stenoses in the origin of the gastrointestinal arteries. The sensitivity for significant (>50–70%) stenoses of the CA is 75–100%, with a specificity of 88–89%. The sensitivity for significant stenoses of the SMA is 89%, with a specificity of 92–97%.41–44 Duplex ultrasound performed during deep expiration can demonstrate a marked increase in flow velocity at the compressed region of the CA, supporting the diagnosis of CACS.45 A shortcoming of abdominal duplex ultrasound as a screening tool for stenoses in the gastrointestinal arteries is the poor arterial visualisation in 10–20% of patients due to over-projection of abdominal air. Furthermore, a high sensitivity and specificity for detection of arterial stenosis is only obtained in experienced hands.

In recent years CTA and MRA have evolved as promising techniques for assessment of the patency of the gastrointestinal arteries and evaluation of patients suspected of CGI. The main advantages of both imaging techniques are the non-invasive approach and the possibility to visualise the surrounding structures (figure 4). Only a few studies compared CTA or MRA with DSA as the current gold standard in the evaluation of patients suspected for CGI. In a recently published study, 52 patients had both CTA and DSA for visualisation of the gastrointestinal arteries, showing a sensitivity of CTA for abdominal arterial stenosis of 82%, with a specificity of 100%.46 Another study compared MRA and conventional DSA in 65 patients suspected for CGI. In this study the sensitivity of MRA for abdominal arterial stenosis was 100%, with a specificity of 95%.47 The MRA tended to over-rate stenoses in up to 15% of patients, and evaluation of the IMA was only possible in 64% of patients. No studies are yet available comparing CTA and MRA in patients suspected of CGI. However, it seems that the IMA and the smaller peripheral mesenteric vessels are currently better assessed with CTA than with MRA because of the higher spatial and temporal resolution and faster acquisition times of CTA.48 In addition, CTA allows the identification of calcified plaques. The main advantages of MRA over CTA are its lack of radiation exposure and the possibility to perform flow measurements. MRA measurements have shown a consistent relationship between flow in the portal or superior mesenteric vein and flow in the arteries supplying those veins.49 The assessment of flow velocities of the portal and superior mesenteric vein before and after oral caloric stimulation seems another promising diagnostic tool for CGI.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Three-dimensional reconstruction of a computed tomography angiography scan. A sagittal reconstruction of a CT-angiography (panel A) showing a subtotal occlusion of the coeliac artery (CA) with post-stenotic dilation, a total occlusion at the origo of the superior mesenteric artery (SMA) over a distance of 2 cm as indicated by the arrowhead (the artery is missing over a distance of 2 cm). This has been a chronic condition, as in panel B the inferior mesenteric artery (IMA) is hypertrophied with retrograde filling of the branches of the SMA. The blue arrowheads show the meandering artery providing the large and small intestine of blood.

Detection of mucosal ischaemia

Mucosal hypoperfusion, or hypoxaemia, is the first sign of reduced/insufficient blood flow in the gastrointestinal arterial tract. The blood flow within the intestinal wall is unevenly distributed, and may show marked variation between mucosal and serosal sides. The mucosal side, being the metabolically most active, is relatively protected against ischaemia at disadvantage of the serosal layers.50 However, despite this preserved mucosal blood flow, the mucosal villi are most vulnerable to a hypoxaemic state.51 52 This is a consequence of the microvascular situation, with counter-current oxygen exchange occurring over the length of the villus, resulting in hypoxaemia at the tip and normoxaemia at the crypts.

This hypoxaemic state is associated with an anaerobic metabolic state, which eventually leads to intracellular accumulation of CO₂ due to buffering mechanisms. Subsequently, this intracellular CO₂ diffuses into the bowel lumen leading to an increase in intraluminal CO₂. This intraluminal CO₂ can be measured by means of intraluminal tonometry. For this purpose, a balloon-tipped naso-jejunal cathether is placed. After inflation of the semi-permeable balloon, the balloon CO₂ pressure equilibrates with the luminal CO₂ pressure. Repeated aspiration of the gas content of the balloon during fasting and after meals thus provides an insight in the local mucosal CO₂ pressures and can correlate these with symptoms.53 54 The same technique can also be applied in the colon. Currently, gastrointestinal tonometry is the only established functional diagnostic test to detect gastrointestinal mucosal ischaemia. In two large patient cohorts evaluated for possible CGI, gastrointestinal tonometry proved to have a sensitivity and specificity of, respectively, 78–85% and 82–92% for the detection of CGI (figure 5).6 7 Currently, gastrointestinal tonometry is used for diagnosing mucosal ischaemia by two different approaches. The first approach introduced was gastric exercise tonometry which uses sub-maximal exercise as a provocation of gastrointestinal ischaemia, in essence very similar to the concept of exercise testing commonly used for the evaluation of cardiac ischaemia. Because gastric exercise tonometry is quite cumbersome and cannot be performed in up to 20% of patients due to their inability for physical exercise, an alternative tonometric approach was introduced. This method consisted of prolonged (24 h) combined gastric and jejunal tonometry using meals as provocation of gastrointestinal ischaemia.55 This prolonged tonometry test was found to be as accurate as gastric exercise tonometry for detection of gastrointestinal ischaemia with a sensitivity of 77% and a specificity of 94%, and is easier to perform without patient restrictions.55

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Example of 24 h tonometry in healthy subject (A) and patient with chronic gastrointestinal ischaemia (B). 24 h tonometry is performed by means of endoscopically placed nasojejunal balloon catheter. During a period of 24 h the intraluminal Pco₂ is measured in both the stomach and jejunum, both in the fasting condition and with meals. The patient is asked to eat a test meal. The patient is requested to note time and duration of any symptoms in a logbook throughout the test. In panel A, the 24 h reading of the intraluminal Pco₂ in the stomach (blue line) and the jejunum (purple line) of a healthy subject. In panel B, the 24 h Pco₂ reading of a patient with mucosal ischaemia of the jejunum after test meals, with a close association between the pain and a rise in intraluminal Pco₂.

Despite these considerable improvements in diagnostic tools, a diagnosis of CGI remains a challenge in clinical practice. Currently, the main obstacles for a correct and timely diagnosis of CGI are first that symptoms are in most cases only present for a limited period of time after a provocation, and second the limited accessibility and burden of tonometry as the proposed preferred diagnostic tool. Gastrointestinal tonometry is a cumbersome and patient-unfriendly procedure to perform, and proved to be false-negative in up to 23% of CGI patients.6 7 55

Recently, an alternative diagnostic method has been introduced, so-called visible-light spectroscopy (VLS). This relatively new and promising technique allows the direct measurement of mucosal tissue oxygen saturation during endoscopy.56 Briefly, a 1.5 mm probe can be introduced through the working channel of an endoscope, just above the gastric or duodenal mucosa. By analysing the backscattered light the capillary oxygen saturation can be calculated (figure 6). A first study performed by our group showed promising results of application of VLS in patients referred for evaluation of possible CGI.27 57 In this study, cut-off values were determined in a test cohort and subsequently validated in a second cohort of patients. This showed that VLS had a sensitivity and specificity of, respectively, 90% and 60% for correctly diagnosing CGI. This is similar to, or even slightly better, than gastrointestinal tonometry. VLS measurements during upper endoscopy are thus likely to become the preferred functional test method to assess patients clinically suspected for CGI, replacing gastrointestinal tonometry as functional test for mucosal perfusion. This new and easy to use VLS technique may lower the threshold for gastroenterologists in the field to test for this disease entity. Furthermore, VLS during upper endoscopy is considerably more patient-friendly than gastrointestinal tonometry, as it is less invasive, less time-consuming, and can be performed under conscious sedation on an outpatient basis (figure 6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Mucosal oxygen saturation measurement of the duodenal and gastric mucosa with visible-light spectroscopy. A 1.5 mm probe (A) can be introduced through the working channel of an endoscope to be held just above the mucosa (B). The probe transmits visible white light for tissue illumination and a fibre optic bundle that directly transmits back to the monitor for continuous, real-time saturation measurements. During measurement, the endosocopy light source is switched off. The T-Stat reports capillary-weighted oxygen saturation (C). Our group recently validated mucosal oxygen saturation cut-off values for detecting mucosal ischaemia (D).57

Surgical revascularisation

Since the first successful report on surgical treatment for CGI by Shaw and Maynard in 1958,62 many case series have been reported concerning surgical revascularisation for the treatment of CGI. Although many controversies remain on the choice of bypass route, antegrade versus retrograde, choice of conduit, use of vein versus prosthetic graft, and whether one or more vessels should be revascularised in the presence of multi-vessel disease, surgical revascularisation is generally accepted as a successful treatment for CGI. Success rates defined as relief of symptoms vary between 84% and 94%, and surgical revascularisation is associated with an exceptional long-term durability (73–93%) at 5 years, and low recurrence of symptoms (7–14%). The data from the Nationwide Inpatient Sample database showed that surgical treatment of 2128 patients with CGI was associated with a 13% mortality (range 5–26%) and 38% morbidity (range 15–61%).63 These high figures may be explained by several factors. First, the majority of CGI patients are in a debilitated condition due to significant malnutrition and weight loss. This is, among others, caused by the fact that CGI can be a challenging diagnosis, with a diagnostic delay varying between 4 and 48 months (mean 11 months).64 Second, many patients with CGI have serious co-morbidities such as coronary artery disease, symptomatic peripheral vascular disease, renal impairment, hypertension and COPD. A combination of debilitated condition and co-morbidities are known to increase the risk of postoperative mortality and morbidity of major surgery in general.65 Third, individual centres and physicians usually have limited experience with mesenteric revascularisation, as also illustrated by most case series including not more than 40 patients in a total time span of over 10–15 years. Skills and learning curves are thus likely to have influenced the outcome of interventions. The major complications of surgical revascularisation in CGI are acute renal failure (11%), gastrointestinal infarction requiring bowel resection (8%), cardiac arrest (6%), respiratory complications (5%), myocardial infarction (5%) and haemorrhage (3%).63

Endovascular revascularisation

PTA with or without stent placement of gastrointestinal arteries was reported for the first time in 1980.66 The arteries are approached via a trans-femoral or trans-brachial route, and the ostia of the gastrointestinal arteries and collateral circulation can be visualised during selective DSA of the three main gastrointestinal arteries. Total vascular occlusion was initially thought to be a contraindication for endovascular revascularisation due to plaque defragmentation and subsequent distal embolisation, but recent studies have shown its feasibility and safety in revascularisation of occluded gastrointestinal arteries67 68 (figure 7). As the length of the stenosis is an indicator of atherosclerotic plaque burden, the risk of distal embolisation is thought to be increased in patients with long segment (>2 cm) occlusion. In this case, open surgical revascularisation should be considered as first treatment option. Another important exclusion for endovascular treatment is the presence of CACS. Due to the dynamic and non-circular nature of CACS, stents break easily in these patients, often within the first months after placement69–73

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

Percutaneous transluminal angioplasty with stenting of a total occlusion of the coeliac artery (CA). A pigtail catheter is positioned in the aorta and a diagnostic arteriogram is performed showing total occlusion of the coeliac artery with no contrast filling of the downstream branches (A). A guidewire is passed through the stenosis (A). A balloon expandable stent (arrow) is positioned over the stenosis (B). The stent is deployed (arrowhead) (C), with filling of the downstream branches with a control arteriogram (C).

Endovascular revascularisation is associated with a high primary technical success rate of 90–100%, with symptom relief in 73–99% of patients (tables 2 and 3). However, the relapse rate due to restenosis of the artery or in-stent stenosis is around 28% (range 11–39%), necessitating new endovascular interventions (tables 2 and 3). In earlier studies, endovascular revascularisation was reserved for patients who were not eligible for surgical revascularisation. A recent systematic review including 328 patients from 16 case series, treated with PTA for CGI, showed a PTA-associated morbidity and mortality rate of 9% and 3%, respectively. The study presented from the Nationwide Inpatient Sample database showed a morbidity rate of 20% and a mortality rate of 3.7%.63 The main complications reported after endovascular therapy are related to the access point, and in particular include thrombosis, dissection and haemorrhage. Complications related to the treatment of the vascular stenosis itself are infrequently reported, occurring in 2–16% of patients, the most frequent being dissection and distal embolisation due to plaque defragmentation. Systemic treatment-related complications reported are renal insufficiency, myocardial infarction and anaphylactic reactions.

After stent placement, patients usually require anticoagulant therapy to prevent stent occlusion. At present, the majority of patients receive clopidogrel in combination with aspirin for a period of 1–3 months. There are as yet no studies in gastrointestinal arterial disease which prove that this approach prolongs stent or bypass patency, but such an effect is assumed based on data in other stenotic vascular conditions such as coronary artery disease, renal artery disease and peripheral vascular disease. Recent technical advancements, in particular the introduction of new, partially covered endovascular stents may improve stent patency even further in future.

Comparing endovascular and surgical revascularisation therapy

There are currently no prospective studies that compared surgical and endovascular revascularisation in CGI patients. The incidence of the disease and the diversity in surgical approaches make it unlikely that a direct comparison of both therapy regimens will be performed in the forthcoming years. Most of the knowledge and guidelines thus have to come from case series. Some studies compared cases treated with endovascular revascularisation to historical case series with surgical repair. These comparisons are, however, hampered by differences in patient selection, and improvements in surgical and endovascular techniques over time. For example, endovascular revascularisation was earlier mainly applied in patients with high-risk co-morbidities, or as a ‘bridge to surgery’ in severely debilitated patients. The available retrospective comparative studies showed similar early outcomes with endovascular and surgical revascularisation, but better long-term patency and symptom relief in the surgically treated patients. Symptoms recurred on average in 28% of the endovascular treated patients, against 7–14% in the surgically treated patients. This difference was mainly due to the recurrence of stenosis in the endovascular treated patients. This led to the general conclusion that surgical revascularisation is the first choice treatment in younger patients and in patients without significant co-morbidity. This approach is supported by a recent American study, showing that a choice for surgical or endovascular revascularisation could be made on the basis of a pre-operative risk assessment of mortality (box 2, table 4). In this study, the mortality rose from 0.7% in the absence of risk criteria to 3.1% in the presence of one risk factor, 12.5% with two or three risk factors, and 25% if four or five risk factors were present.74 However, in recent studies, using endovascular intervention as the primary method of revascularisation, the success rate of endovascular treatment, after re-intervention in 28% of patients, was approximately 90% at 3 years follow-up.67 75

Currently, the long-term results and complications of both treatment strategies seem similar. Against this background, the choice of first treatment depends on patient age and condition, underlying cause of CGI, and local experience. These decisions should be made on a patient-to-patient base.

Treatment of single-vessel disease CGI

Recent studies have shown that single-vessel CGI is more prevalent than previously thought.6 7 The aim of intervention is to decrease the burden of symptoms instead of decreasing CGI associated mortality and morbidity, as mortality is mainly related to multi vessel CGI.61 At present, only a few studies have provided information on the success rate of intervention in single-vessel CGI patients. The success rates after therapy vary widely from 65% to 83% in larger patient groups after long-term follow-up8 75 The introduction of a functional diagnostic test for mucosal hypoxaemia, that is gastrointestinal tonometry or visible-light spectroscopy measurements, improves the selection of patients who actually have CGI and will therefore be more likely to respond to revascularisation therapy.76 The addition of this functional diagnostic tool seems especially important in patients suspected of single vessel CGI, as only 54% of the patients analysed for CGI with a single artery stenosis were shown to have CGI.6

Surgical treatment of CACS

Patients with CACS require surgical treatment, as the coeliac artery has to be released from its extrinsic compression by the diaphragmatic crurae, medial arcuate ligament, and/or periaortic ganglionic tissue. Treatment consists of trans-section of the median arcuta ligament and the periganglionic tissue from the coeliac trunc all the way on to the aorta. Several investigators reported that intraoperative ultrasound may support identification and extent of the extrinsic compression, as well as monitor the decompressive effect of the trans-section during the operation.77–79 Two studies have published their results of treatment in 4424 and 438 CACS patients, respectively. The investigators performed either open surgical decompression, decompression with dilatation, or decompression with reconstruction. Sustained symptom relief was achieved in respectively 68%24 and 83%8 of patients. In the American study, sustained symptom relief was higher (76%) if CACS decompression was combined with some form of revascularisation, which was either dilatation of surgical revascularisation.24 In the Dutch study, patients eligible for CA release were identified using gastric exercise tonometry, which increased the success rate of the surgical intervention.8 Furthermore, patients that presented with classic symptoms of postprandial pain, aged between 40 and 60 years, and weight loss had a better response to surgery than those without these symptoms, reinforcing the importance of proper patient selection.24

Until recently, release of the coeliac trunk in CACS patients was performed during open laparotomy. Recently, a few groups have studied the use of laparoscopic treatment of CACS. Laparoscopic decompression surgery was shown to be as safe and effective as open surgery with a decrease in the duration of admission.78

Conservative treatment

Some CGI patients are not eligible for endovascular or surgical revascularisation therapy. This can be due to the extent of the arterial disease and/or the presence of co-morbidity. For these patients, conservative treatment can be considered. Currently, no reports have been published to support the use of a pharmacological approach to patients with CGI caused by occlusive gastrointestinal arterial disease. In theory, proton-pump inhibition and dietary modifications, that is smaller, more frequent, and low-caloric meals, both reduce the metabolic demand of the gastric and duodenal mucosa, but both therapies have not been systematically studied in humans. Animal experiments have shown that the gastrointestinal blood flow is influenced by the amount and especially the caloric content of meals.28 29 80 Our own clinical experience with proton-pump inhibition treatment and/or dietary measurements as conservative therapy in these patients varies; however, some patients respond well to this approach. Long-term conservative therapy may also be preferred for some patients with single-vessel disease. Some of these patients respond well to conservative measures, and some refuse revascularisation therapy.

Treatment of chronic NOMI

As described above, there is a distinct group of patients with mucosal ischaemia in the absence of macrovascular pathology. Treatment of these patients primarily aims at any identifiable underlying cause for the chronic hypoperfusion such as heart failure, valve disease, arrythmias, anaemia or vasculitis. There is, however, also a subgroup of patients without these underlying conditions. Two case–cohort studies have been reported in abstract form, which studied treatment of NOMI with vasodilating medication. The first study treated 44 patients with chronic NOMI with either isosorbide dinitrate, ketanserin, nicorandil or doxazosin. Although side effects were common, a success percentage defined as >50% decrease of pain was achieved in 63%.26 The second cohort treated 31 patients with chronic NOMI with isosorbide dinitrate 50 mg od, followed by ketanserin 40 mg od if the nitrate therapy was unsuccessful or had to be stopped due to intolerable side effects. This study reported a success rate, defined as >50% decrease of pain, in 59% of cases.27 Although both reports showed promising data for the treatment of chronic NOMI, these results have to be interpreted with caution, as both studies were not designed as placebo-controlled trials.