In spite of development of medical intensive care, severe acute pancreatitis (SAP) is still a fatal disease with a high mortality rate. Severe acute pancreatitis is characterized by both systemic organ failure and pancreatic and/or peripancreatic necrosis with infection. Unfortunately, pharmacological therapy including anti-inflammatory drugs, antiproteases, and probiotics could not be clarified to having beneficial effects on the course of SAP as expected (1). Prophylactic antibiotics are also controversial in acute necrotizing pancreatitis (ANP) (2). A new strategy of treatment based on pathophysiology of the disease should be mandatory.

We have clarified pancreatic ischemia with vasospasm precedes necrotic change of the pancreas in the early phase of acute necrotizing pancreatitis (3). Microcirculatory derangement due to ischemia and an increase in intravascular coagulation associated with SAP may easily cause microthrombi formation resulting occlusion of microvessels and necrosis. Therefore, a therapeutic strategy to inhibit the intravascular coagulant activity in the pancreas may be expected to prevent evolution of pancreatic ischemia into necrosis. Nafamostat mesilate is a potent serine-protease inhibitor. In addition, it acts as an anticoagulant as well as antiprotease (4, 5, 6). Regional arterial infusion technique can provide enough amount of nafamostat to the ischemic area of the pancreas. Early initiation of continuous regional arterial infusion (CRAI) of a protease inhibitor and an antibiotic in ANP has been recommended in Japanese guidelines (7). Recently, Piascik et al reported that CRAI of protease inhibitor nafamostat and antibiotic imipenem was effective in preventing complications and reducing mortality rate in SAP based on a randomized controlled trial (8).

Angiographic appearances of arterial irregularities in acute pancreatitis have already been described since 1960’s (9, 10). However, main reason for angiography was examination of the presence of complications, such as a palpable mass, suspicion of abscess or pseudocyst. Nowadays contrast-enhanced computed tomography (CE-CT) enabled us to detect pancreatic abscess, pseudocyst and pancreatic necrosis easily. In addition, pancreatic ischemia is also detected by CE-CT in the early phase of ANP. Unfortunately, relationship between angiographic abnormalities and pancreatic ischemia and/or necrosis on CE-CT findings has never been discussed in the literature.

Angiographic features in the early phase of acute necrotizing pancreatitis were investigated in a few Japanese medical centers using digital subtraction angiography with selective catheterization into the celiac artery and superior mesenteric artery (3, 11, 12). In the early phase of ANP, angiographic appearances with irregularities in various grades were observed. Angiographic abnormalities such as sudden tapering, alternative narrowing and dilatation, angulations and diffuse caliber irregularities of the intrapancreatic branch arteries are characteristic of ANP. Severe ischemic change with diffuse vasospasm and/or obstruction was observed not only in the intra- and peripancreatic arteries, but also in the extrapancreatic arteries such as the superior mesenteric artery and its branches (3). Arterial irregularities with vasospasm were already visible on the initial angiography at 12 hours after the onset. Figure 1 shows a typical appearance of vasospasm on angiography. CE-CT revealed a heterogeneous enhancement in the pancreatic head. Selective angiography via the gastroduodenal artery showed alternative narrowing and dilatation and sudden tapering of the intrapancreatic branch arteries consistent with the poorly perfused area of the pancreas on CE-CT. Improvement of ischemic change with angiographic irregularities of the intrapancreatic branch arteries were also observed on follow-up angiography (fig. 2). These findings strongly suggested that the irregular narrowing of the arteries was not caused by structural or irreversible change due to postinflammatory fibrous adhesion but was caused by vasospasm. The extent of ischemic changes with vasospasm on angiographic appearance was correlated with the extent of the poorly perfused area of the pancreas on CE-CT findings and the severity of acute pancreatitis (fig. 3, 4). Non-occlusive mesenteric ischemia (NOMI) associated with SAP was also investigated based on the angiographic appearance by Hirota et al. (13) Severe vasospastic changes of both the celiac artery and the superior mesenteric artery were observed simultaneously in patients with NOMI accompanied by ANP. Severe ischemic changes in the intraabdominal vessels may aggravate microcirculation not only in the pancreas but also in the abdominal viscera leading to pancreatic necrosis and systemic organ failure.

The mechanism of vasospasm in SAP has not yet been elucidated. Hypovolemia and hypotension (14), and sympathetic stimulation (15) causes vasospasm. Local inflammation and activation of complement system (16) may be other possible causes of vasospasm. The activation of the complement system that is observed in SAP may be involved in vasospasm. The complement system-derived immune system is considered to contribute to the pathogenesis of cerebral vasospasm (17). Recently, endothelin-1 (ET-1) has been noticed as a candidate for mediating the vascular changes of SAP. Infusion of endothelin changed edematous pancreatitis into hemorrhagic necrotizing pancreatitis in an animal model (18). Continuous infusion of ET-1 produced prolonged arterial narrowing (19). ET-1 receptor antagonist prevented pancreatic blood flow from decreasing in severe experimental acute pancreatitis (20). Thrombin has also been considered to play an important role in arterial vasospasm (21, 22). Thrombin suppresses endothelial NO synthase and upregulates endothelin-converting enzyme-1 (23). An imbalance between vasodilator NO and vasoconstrictor ET-1 may be involved in the regional vasospasm of the arteries in the inflamed pancreas. Production of ET-1 by HUVECs was significantly enhanced by stimulation of thrombin and that a close relationship may exist between ET-1 and vasospasm in acute pancreatitis (24).

Hypovolemia and/or hypovolemic shock, activation of thrombin, and coagulant system are commonly observed in SAP, and they are concomitant with vasospasm and pancreatic ischemia. Microcirculatory derangement due to the formation of microthrombosis in the pancreatic vessels may thus easily occur and pancreatic ischemia may change to be necrotic.

A synthetic protease inhibitor nafamostat mesilate has a broad inhibitory action on pancreatic enzymes, the coagulation system, the complement system and the production of proinflammatory cytokines. Nafamostat inhibits trypsin and phospholipase A2, and it also inhibits the classical and alternative pathway of complement system and the activation of thrombin, kalliklein, plasmin, factor VIIa and factor Xa. Nafamostat is not only a protease inhibitor but also a potent anticoagulant that is widely used for hemodialysis and the treatment of disseminated intravascular coagulation. Nafamostat suppresses activation of thrombin and thereby inhibits the production of ET-1, thus preventing vasospasm. Nafamostat may also protect the pancreas against necrotic change by inhibiting increase in coagulability and microthrombi formation.

However, protease inhibitors could not show beneficial effects in human acute pancreatitis as expected. Unfortunately, the protease inhibitors do not easily reach the pancreas when administered intravenously because of their pharmacokinetic characteristics (25) and impaired microcirculation. On the other hand, administration of a protease inhibitor through the catheter placed at one of the arteries supplying the inflamed area of the pancreas dramatically increases the tissue concentration of a protease inhibitor (26, 27). Mikami et al. clarified the efficacy of continuous regional arterial infusion of a protease inhibitor nafamostat mesilate on severe acute pancreatitis in rats (27). The tip of needle was placed at the celiac artery in rats. CRAI rats had six times higher concentrations of nafamostat in the pancreas than those infused intravenously. CRAI significantly reduced the levels of serum IL-6. The mortality rate was significantly reduced after CRAI compared with the intravenous infusion of nafamostat.

The concentration of antibiotics was also markedly increased in the pancreatic tissue when administered intraarterially in a experimental acute pancreatitis models, and protective effect intraarterial infusion of an antibiotic against pancreatic infection was demonstrated (28, 29).

Takeda et al. reported a new trial of CRAI therapy for ANP in 1996 (30). CRAI of both a protease inhibitor and an antibiotic reduced not only the mortality rate but also the frequency of infected pancreatic necrosis. A nationwide survey of CRAI therapy in Japan also reported that CRAI of both the protease inhibitor and the antibiotic was superior to CRAI of the protease inhibitor alone (31). There was no significant difference in the mortality rate between patients who received the protease inhibitor via CRAI and the antibiotic intravenously, and patients who received both the protease inhibitor and antibiotic via CRAI. However, the frequency of infected pancreatic necrosis was significantly lower than in patients with involving both drugs that in patients with CRAI involving protease inhibitor alone. Imaizumi et al. reported efficacy of CRAI therapy in patients admitted to an intensive care unit (32). Cumulative survival rate at 1, 6 and 12 months were 77.9%, 48.9% and 48.9% in the non-CRAI group compared with 100%, 100% and 87.1% in the CRAI group. The rate of surgery was 32% in the non-CRAI group and 9% in CRAI group. Yasuda et al. also demonstrated that in the CRAI group, the incidence of infection, the frequency of surgery, and the mortality rate were lower than those in the non-CRAI group: 34% versus 51%, 27% versus 63%, and 37% versus 54% (33). Shirai et al. conducted a prospective randomized controlled trial in patients with SAP, comparing one group who were given meropenem and the another group who received imipenem using CRAI (34). There were no significant differences in the frequency of infected pancreatic necrosis, surgical treatment, and the mortality rate. CRAI of meropenem combined with nafamostat was effective as well as imipenem. Ino et al. evaluated the efficacy of CRAI of another protease inhibitor gabexate mesilate in patients with SAP (35). The duration of abdominal pain in the CRAI group was 1.9 +- 0.29 d, whereas that in the non-CRAI group was 4.3 +- 0.5. The duration of SIRS in the CRAI group was 2.2 +- 0.22 d, whereas that in the non-CRAI group was 3.2 +- 0.28. The average length of hospitalization was significantly shorter in the CRAI group than that in the non-CRAI group. As above-mentioned, many clinical trials on the efficacy of CRAI therapy have been reported, but unfortunately, there had been no published randomized clinical trials so far. However, recently, results of a randomized controlled study of CRAI therapy in patients with SAP were reported from Poland (8). 78 patients with SAP were included in that study. The CRAI patients were treated continuously with nafamostat mesilate 240 mg/d and imipenem 1 g/d for 5days via one of the arteries perfusing the pancreas. The non-CRAI patients received imipenem 1 g/d intravenously. The additional antibiotics were applied in 8 of CRAI patients and in 18 of non-CRAI patients (p = 0.02). Urgent surgical intervention was necessary in 10.3% of CRAI patients and in 33.3% of non-CRAI patients (p = 0.01). The mortality rate was 5.1% in CRAI and 23.1% in non-CRAI group (p = 0.02). These results demonstrated that CRAI of a protease inhibitor and an antibiotic was effective in preventing complications and in reducing mortality rate in SAP.

Anai et al. evaluated the relationship between the therapeutic effect of CRAI therapy and drug distribution on CT-angiography (36). In patients having a good drug distribution on CT angiography, there was a rapid decrease of the APACHE II score and the CT severity index by Balthazar. The CT severity index was 5-10 (mean 7.9) before CRAI therapy, but it decreased to 0-3 (mean 0.6). Hirota et al. also reported a case with NOMI accompanied by SAP (13). In that case, severe vasospasm occurred in both the celiac artery, the superior mesenteric artery and their branches simultaneously. The patient received CRAI therapy with nafamostat solely via the celiac artery. The pancreas was spared from diffuse necrosis in contrast to the diffuse intestinal necrosis, which occurred due to mesenteric vasospasm.

Theoretically, CRAI therapy should be initiated as soon as possible after diagnosing SAP and ischemic change of the pancreas on CE-CT, because pancreatic necrosis is established four to 10 days after the onset. The effect of the timing of the initiation of CRAI therapy on the outcome was evaluated in patients with ANP (37). The frequency of respiratory failure requiring mechanical ventilation was significantly low in patients with early initiation of CRAI (within 3 days after the onset) compared to that in patients with late initiation of CRAI (more than 4 days after the onset). The mortality rate was also significantly low in patients with early initiation of CRAI therapy. A nationwide survey also demonstrated that the mortality rate was significantly low in patients in whom CRAI therapy was initiated within 48 hours after the onset compared to that in patients in whom CRAI therapy was initiated more than 48 hours after the onset (31). Ishikawa et al. also reported that the mortality rate was significantly low in patients in whom treated with CRAI within 3 days after the onset (38). We investigated the timing of the initiation of CRAI therapy and the morphological changes examined on serial CE-CT (39). Morphological improvement was observed in 84%, 62.5%, and 53% in patients in whom CRAI was initiated within 48 hours, 48-72 hours, and more than 72 hours, respectively.

Goal and aim of the CRAI therapy is to prevent necrotic change of the pancreas and reduce the incidence of infection of the pancreas in SAP. CRAI therapy should be indicated for patients with pancreatic ischemia before the necrotic change of the pancreas has been developed. CRAI of namafmostat is expected to prevent vasospasm and the formation of microthrombi in the ischemic pancreas. CRAI therapy is recommended to be started at least within 72 hours (within 48 hours, if possible) after the onset (31, 37). CE-CT has been a gold standard for the diagnosis of pancreatic necrosis so far. However, in the early phase of SAP, can CE-CT detect pancreatic ischemia or predict pancreatic necrosis accurately? The United Kingdom guidelines for the management of acute pancreatitis recommended late CE-CT examination to confirm pancreatic necrosis, because of low sensitivity of CE-CT to diagnose pancreatic necrosis in the early phase of SAP (40). Pancreatic ischemia or necrosis is often underestimated or sometimes overestimated in the early stage on contrast-enhanced CT examination. In CRAI approach, it is important that the appropriate artery supplying the ischemic area of the pancreas should be selected as a route of administration. A new imaging method to detect pancreatic ischemia and ischemic region more accurately in the early stage should be necessary. If we can predict necrotic area of the pancreas accurately, super-selective angiographic approach is available to deliver adequate amounts of the drugs to ischemic area of the pancreas.

Recently, perfusion computed tomography (perfusion CT) has been proposed as a useful predictive tool of SAP or pancreatic necrosis. Bize et al. demonstrated that pancreatic perfusion measurement using multi-detector row CT (MDCT) with perfusion imaging could help in assessing the severity of acute pancreatitis (41). Tsuji et al. showed that perfusion CT is a useful tool for early detection of ischemic change in the pancreas that leads to pancreatic necrosis (42). The sensitivity and specificity of perfusion CT for predicting pancreatic necrosis was calculated to be 100% and 95.3%, respectively. We also clarified the usefulness of perfusion CT for diagnosis of necrotizing pancreatitis in the early phase of acute pancreatitis (43). Perfusion parameters such as pancreatic blood flow (perfusion), peak enhancement intensity and blood volume of the pancreas were significantly low in the ischemic area of the pancreas developing to necrosis later (fig. 5, 6). Perfusion CT should be a reliable tool for detection of pancreatic ischemia and ischemic region in the early stage of SAP.

Pancreatic ischemia with vasospasm precedes necrotic change of the pancreas in the early phase of acute necrotizing pancreatitis, and increase in coagulability accompanied by severe acute pancreatitis may play a crucial role in the development of pancreatic necrosis. Adequate amounts of a protease inhibitor, which is also a potent anticoagulant, and an antibiotic delivered by continuous regional arterial infusion approach to the ischemic region may prevent necrotic changes in the pancreas and reduce pancreatic infection.