Clinical Gastroenterology and Hepatology 2014;12:854–862
PANCREAS, BILIARY TRACT, AND LIVER
Portosplenomesenteric Venous Thrombosis in Patients With AcutePancreatitis Is Associated With Pancreatic Necrosis and UsuallyHas a Benign Course
Jeffrey Easler,*,a Venkata Muddana,*,a Alessandro Furlan,‡ Anil Dasyam,‡ Kishore Vipperla,*Adam Slivka,* David C. Whitcomb,* Georgios I. Papachristou,*,b and Dhiraj Yadav*,b
*Division of Gastroenterology and Hepatology and ‡Department of Radiology, University of Pittsburgh Medical Center,Pittsburgh, Pennsylvania
BACKGROUND & AIMS:
Although there are some data on prevalence of portosplenomesenteric venous thrombosis(PSMVT) in patients with acute pancreatitis (AP), the progression of PSMVT in patients whohave and have not received anticoagulants has not been studied systematically. We evaluatedthe prevalence and natural history of PSMVT in a well-defined cohort of individuals with AP.
METHODS:
In a retrospective study, we analyzed data from the University of PittsburghMedical Center on 162patientswith a sentinel attackofAP from2003–2010.Datawere collectedonpatientdemographics,clinical presentation, etiology, clinical course, and outcomes. One hundred twenty-two patientsunderwent contrast-enhanced computed tomography; the scans were reviewed to identifythromboses and/or narrowing of splanchnic veins (splenic, superior mesenteric, and portal).
RESULTS:
PSMVT was detected in 22 patients overall (14%; 18% among patients who underwentcontrast-enhanced computed tomography). Median time to detection of PSMVT was 17 days(interquartile range, 11–40 days). PSMVT formed most frequently in the splenic vein (19 of 22,86%), followed by portal (8 of 22, 36%) and superior mesenteric (6/22, 27%) veins. Devel-opment of PSMVT was associated with presence (21 of 22, 95%), location, and extent ofpancreatic necrosis. Fifty-three percent of patients (21 of 40) with necrosis developed PSMVT.Anticoagulants were administered infrequently (6 of 22, 27%) and always for indicationsunrelated to PSMVT. Most patients with PSMVT developed collateral veins (19 of 22, 86%), and27% (6 of 22) were found to have varices during endoscopic evaluation, but clot resolution wasinfrequent (2 of 22, 9%). No patient developed complications directly related to PSMVT.
CONCLUSIONS:
PSMVT develops in about half of patients with necrotizing AP and is rare in the absence ofnecrosis. Despite infrequent administration of anticoagulants, complications directly related toPSMVT are rare.
Acute pancreatitis (AP) is the leading gastrointes-tinal cause of hospital admissions in the United
States and continues to rise in incidence.1 Most APpatients have a mild course with complete recovery daysafter disease onset.2,3 However, a variable fractiondevelop severe disease. Complications of severe AP canbe broadly categorized as systemic (organ failure) orlocal (pancreatic necrosis and/or fluid collections),4–6
which are associated with prolonged hospital course,significant morbidity, and mortality.7
Peripancreatic vascular involvement in AP is oftendetected incidentally on imaging studies performed forevaluation of symptoms or complications. Although thepathogenesis is unclear, it is believed that local extensionof inflammatory process can involve vessels in the
vicinity of the pancreas, which, along with the action ofproteolytic enzymes, weakens vessel walls and causesstasis of blood flow.8,9 Venous involvement is manifestedby splanchnic (portal/splenic/mesenteric) venous
May 2014 Portosplenomesenteric Venous Thrombosis in AP 855
thrombosis (PSMVT) and arterial involvement by pseu-doaneurysms from direct erosion of the vessel wall.Complications resulting from vascular involvementinclude bleeding, organ ischemia, or necrosis.8,9
Pancreatitis is an uncommon cause of PSMVT. In 832patients treated at the Mayo Clinic, only 13% of caseswere related to pancreatitis.10 A fairly large body ofliterature exists on PSMVT in chronic pancreatitis (CP).Most studies have evaluated the natural history ofsplenic vein (SV) thrombosis in CP,11–16 with few directlymeasuring its prevalence.11,15–17 Although no formalrecommendations exist, anticoagulation is generally notused for PSMVT in CP patients.
There are few data on the prevalence17–19 and naturalcourse of PSMVT19 in AP. In contrast to CP where thetiming of PSMVT development is often uncertain, theonset of pancreatic injury in AP is known, and thrombusformation is acute/subacute. Therefore, systematicstudies are needed to better understand the risk anddeterminants of PSMVT and the utility of anticoagulationin AP. In this study of a well-phenotyped cohort of APpatients,4,20 we found the prevalence of PSMVT to beclosely linked to pancreatic necrosis. We report the riskof PSMVT and compare the clinical course of patientswith necrosis who did/did not develop PSMVT. Finally,we report on the use of anticoagulation and complica-tions in patients with PSMVT and the prevalence ofarterial pseudoaneurysms in our cohort.
Methods
Patient Cohort
Severe acute pancreatitis study (SAPS) at the Uni-versity of Pittsburgh Medical Center prospectivelyenrolled 256 AP patients from 2 tertiary care hospitals(University of Pittsburgh Medical Center-Presbyterian,University of Pittsburgh Medical Center-Shadyside) inthree 1-year periods between June 2003 and April 2010after obtaining informed consent and by using stan-dardized criteria.4,20 The study protocol was approvedby the Institutional Review Board of the University ofPittsburgh. For the present study, we included SAPSpatients who were admitted with their first AP attackand excluded patients with a history of AP or CP.
Data Collection and Disease Severity
As previously described,4,20 data collection in theSAPS study included detailed information on de-mographic, clinical, laboratory, and radiographic pa-rameters. Organ failure was defined by the presence ofshock (systolic blood pressure <90 mm Hg), pulmonaryinsufficiency (arterial PO2 <60 mm Hg at room air or theneed for mechanical ventilation), or renal failure (serumcreatinine level >2 mg/dL after rehydration or hemo-dialysis).4,20 Treatment and interventions (endoscopic,
radiologic, or surgical) for either pancreatic necrosis orperipancreatic fluid collections (PFCs) were noted.
Although information on the clinical course duringthe index admission was collected prospectively, infor-mation on specific clinical (eg, use of anticoagulation,development of complications including gastrointestinalbleeding, interventions, etc) and radiologic data onsubsequent admissions or during outpatient care asapplicable was collected retrospectively. Clinical careincluding laboratory testing, performance of radiologytests and endoscopy during index admission, and follow-up was based on patients’ clinical course.
Evaluation of Portosplenomesenteric VenousThrombosis and Venous Narrowing
All computed tomography (CT) scan reports werereviewed for presence of necrosis, PFC, PSMVT, or nar-rowing. CT scans for patients with positive findings werereviewed by 2 expert radiologists with specific interestin abdominal imaging. The radiologists were blinded tothe clinical history and course of the patients. Wheneveravailable, imaging studies from the referring hospitalswere also reviewed.
CT evaluation was performed on multidetectorhelical CT scanners with 4–64 detector rows (GeneralElectric Medical Systems, Waukesha, WI). Contiguous5-mm–thick or 2.5-mm–thick axial sections were dis-played from the diaphragm to the symphysis pubica.Patients undergoing a contrast-enhanced CT (CECT)received non-ionic intravenous contrast material(ioversol [Optiray 350; Mallinckrodt Imaging, Dublin,Ireland] or Isovue-370 [iopamidol injection 76%;Bracco Diagnostics Inc, Princeton, NJ]) that wasadministered at 3 or 5 mL/s at a volume of 125–150mL. All patients were imaged in the portal venousdominant phase, and the majority, but not all, hadmultiphasic scans with additional imaging in the unen-hanced and hepatic arterial dominant phase. Arterialand portal venous phases of intravenous contrast weretimed by using bolus tracking software (SmartPrepsoftware; GE Medical Systems).
The main portal vein (PV) and its branches, the su-perior mesenteric vein (SMV) and the SV were evaluatedfor patency, thrombosis, and narrowing. Vessels weredefined as patent if the entire lumen was filled withcontrast on enhanced images and there was no luminalnarrowing. Thrombus was defined as a filling defectwithin the lumen of the vessel seen on contrast-enhanced images. Thrombi were classified as occlusiveand non-occlusive, depending on the absence or pres-ence of patent contrast-opacified lumen adjacent to theexisting thrombus. Non-occlusive thrombi were furtherdivided into those occupying more or less than 50% ofthe lumen on axial images. Narrowing of a vein wasdefined as >50% decrease in caliber of the lumen. In thepresence of both thrombus and narrowing, only the
856 Easler et al Clinical Gastroenterology and Hepatology Vol. 12, No. 5
presence of thrombus was recorded. Cavernous trans-formation of PV was defined as an abundance of collat-eral veins in the hepatic hilum.21 Disproportionateincrease in caliber and number of vessels in any locationwas considered as evidence for collaterals.
Assessment was made for sequelae/complications ofvenous thrombosis including formation of collaterals, as-cites, mesenteric edema, splenic infarcts (defined aswedge-shaped non-enhancing areas in spleen), and bowelischemia (findings of bowel wall thickening, ileus,pneumatosis, portomesenteric venous gas, pneumo-peritoneum). Findings of esophageal or gastric varices onesophagogastroduodenoscopy (EGD)/transnasal endos-copy when mentioned were noted. Although an arterialphase of imaging was not obtained in all patients, the su-perior mesenteric, splenic, and common hepatic arteries(andwhere possible, their major branches) were assessedfor patency, thrombosis, and pseudoaneurysm formation.
Local Complications
We recorded the presence of necrosis and PFCs.When pancreatic necrosis was present, its location (head,body, tail, diffuse) and amount (<30%, 30%–49%,
Table 1. Demographics, Etiology, and Disease Severity in Stud
VariableAll patients(n ¼ 162)
PSMVT(n ¼ 22)A
PS
Age, y, mean (SD) 55 (20) 54 (14)Male, n (%) 84 (52) 14 (64)White, n (%) 140 (86) 20 (91)Body mass index, kg/m2, median (IQR) 30 (25–33) 30 (26–33)Etiology, n (%)a
Transfers,a n (%) 95 (59) 18 (82)Intensive care unit admission, n (%) 58 (36) 13 (59)SIRS (score �2),a n (%) 71 (45) 11 (58)BISAP (�3), n (%) 20 (12) 4 (18)APACHE (�8),a n (%)
Ranson’s score (�3),a n (%) 63 (42) 14 (78)Organ failure, n (%) 49 (31) 13 (59)
Single organ 22 (14) 4 (18)Multiorgan 27 (17) 9 (41)
Local complications, n (%)None 86 (53) 0 (0)PFCs 36 (22) 1 (5)Necrosis � PFCs 40 (25) 21 (95)
Mortality, n (%) 10 (6) 1 (5)Length of stay,b median (IQR) 9 (5–17) 31 (17–49)
NOTE. Other etiologies: post-ERCP, 18; hypertriglyceridemia, 5; medications, 9;APACHE, Acute Physiology, Age, and Chronic Health Evaluation; BISAP, Bedsidesyndrome.aMissing information; percentages were rounded to whole numbers and are basebShown for patients surviving index admission.
>50%) was noted. Because most patients with glandnecrosis also have associated peripancreatic necrosis, adistinction between pancreatic plus peripancreatic andpure pancreatic necrosis was not made. Presence ofisolated peripancreatic necrosis was noted, but thesepatients were included in the necrosis group for all an-alyses. Because most patients with necrosis also havePFCs independent of walled-off necrosis, presence ofnecrosis with or without fluid collections was used asone group for analyses.
Local complications were therefore classified into 3groups: no necrosis or PFCs, necrosis with/without PFCs,and PFCs only.
Treatment and Outcomes ofPortosplenomesenteric Venous Thrombosis
We recorded information on the time between theonset of AP to the detection of thrombosis, time tothrombus resolution, whether anticoagulation was used,development of collaterals, resolution of thrombus, anddevelopment of complications related to PSMVT (esoph-ageal and/or gastric varices, gastrointestinal hemorrhage,splenic infarction, and mesenteric ischemia).
hereditary, 2; malignancy/neoplasm, 6; hypercalcemia, 1.Index of Severity in Acute Pancreatitis; SIRS, systemic inflammatory response
d on effective numbers.
May 2014 Portosplenomesenteric Venous Thrombosis in AP 857
Statistics
Descriptive data are presented as median and inter-quartile range (IQR) or mean � standard deviation forcontinuous variables. Categorical data are presented asproportions. For bivariate comparisons, patients with noPSMVT or venous narrowing on CECT were comparedwith those who had PSMVT only or those who hadPSMVT and/or narrowing. Bivariate relationships forcategorical variables were assessed by using the c2 orFisher exact test and for continuous variables by usingthe Mann–Whitney test. Analyses were performed byusing SPSS software v19 (SPSS Inc, Chicago, IL). AllP values <.05 were considered significant.
Results
Patient Characteristics
The mean age of 162 patients was 55 years, 84 (52%)were male, 140 (86%) were white, and 95 (59%) weretransfers from other institutions (Table 1). The most
Figure 1. Flow chart show-ing study cohort.
common etiologies were biliary (78, 48%), idiopathic(26, 16%), alcohol (17, 11%), and post–endoscopicretrograde cholangiopancreatography (ERCP) (17, 11%).Overall, intensive care unit care was needed in 58 (36%),and organ failure developed in 49 patients (31%). Duringthe course of AP, pancreatic necrosis with/without PFCswas observed in 40 patients (25%), of whom 2 had onlyperipancreatic necrosis. Pancreatic necrosis was detec-ted on CECT scan in 39 patients and at the time of sur-gery in 1 patient. The median length of stay was 9 days,and overall mortality was 6% (n ¼ 10).
Incidence of Portosplenomesenteric VenousThrombosis and/or Venous Narrowing
CT scan was performed in 87% of patients (141 of162), of whom 87% (122 of 141) had CECT scan(Figure 1). Any PSMVT or narrowing was noted in 35patients (22% overall; 29% among patients with CECT).PSMVT was seen in 22 patients (14% overall; 18%among patients with CECT). SV was the most frequentlocation of PSMVT (isolated in 13 of 22, 59%; with other
858 Easler et al Clinical Gastroenterology and Hepatology Vol. 12, No. 5
veins in 6 of 22, 27%) (Figure 2). Thrombus was occlu-sive in 68% (15 of 22) and subocclusive in 32% ofpatients (7 of 22). PSMVT involved single vein in 68%(15 of 22), 2 veins in 14% (3 of 22), and all 3 veins in18% (4 of 22). Median time to detection of PSMVT was17 days (IQR, 13–40).
Venous narrowing was seen in 18 patients (11%overall; 15% among patients with CECT). SV was themost frequent location (isolated in 11 of 18, 61%; withother veins in 4 of 18, 22%). Narrowing involved singlevein in 67% (12 of 18), 2 veins in 28% (5 of 18), and all 3veins in 6% (1 of 18). Narrowing in the absence ofthrombus formation was noted in 13 patients (8%overall; 11% among those with CECT).
Characteristics of Patients With or WithoutPortosplenomesenteric Venous Thrombosisand/or Venous Narrowing
When compared with patients who had no PSMVT ornarrowing on CECT (n ¼ 87) (Table 1), those withPSMVT (n ¼ 22) were similar in demographic factors andetiology. The prevalence of intensive care unit admission,multiple organ failure, pancreatic necrosis, and length ofstay were significantly higher in patients with PSMVT.The results were similar when comparisons were
Figure 2. Axial CT images demonstrating (A) SV thrombosis(arrow); note surrounding pancreatic necrosis. (B) SV throm-bosis (long arrow) with perigastric varices (short arrow). (C)Marked attenuation of SMV and SV at the junction with PV(arrow). (D) Narrowing of the caliber of SMV (arrow). (E) Threeimages demonstrating extension of partial thrombus in themain PV extending into its branches (arrows).
made between patients with PSMVT and/or narrowing(n ¼ 35) and no PSMVT or narrowing (n ¼ 87) (Table 1).
Pancreatic Necrosis, PortosplenomesentericVenous Thrombosis, and Venous Narrowing
Necrosis was present in all but one patient withPSMVT (21 of 22, 95%; in 20 on CECT, 1 at the time ofsurgery) (Table 2). The remaining 19 patients with ne-crosis did not have PSMVT (Table 3). The majority ofpatients with PSMVT had extensive necrosis (>50%gland necrosis in 71% [15 of 21], involvement of multi-ple anatomic regions in 48% [10 of 21]). In contrast,necrosis patients with no PSMVT had more localizeddisease (>50% gland necrosis in 50% [9 of 18],involvement of multiple anatomic regions in 37% [7 of19]). The majority of necrosis patients with no PSMVT(63%, 12 of 19) demonstrated venous narrowing, withSV being the most common vessel affected (10 of 11,91%). In 2 patients, venous narrowing was seen on CECTin the absence of necrosis (one involving SV, the otherinvolving SV and SMV).
Infected necrosis was seen in 16 of 40 patients (40%)with necrosis (48%, 10 of 21 with vs 6 of 19, 32%without PSMVT; P ¼ .35). Infection was documented byCT scan alone (presence of gas) in 1, on tissue cultureand sensitivity in 10, and on both CT scan and tissueculture and sensitivity in 5 patients. During a medianfollow-up period of 12.3 months (IQR, 5.4–34.4; range,1.1–99.7), 68% of patients (27 of 40) with necrosisneeded an intervention. When compared with patientswith no PSMVT, a higher proportion of necrosis patientswith PSMVT needed intervention for local disease con-trol (17 of 21, 81% vs 10 of 19, 53%; P ¼ .10), which islikely reflective of the extent of necrosis rather than theprevalence of infection. Interventions were surgical inmost patients (81%, 22 of 27), with debridement with/without cystgastrostomy being the most common pro-cedure. Endoscopic treatment was performed in 5patients and percutaneous drainage in 1 patient.
Anticoagulation and Complications
Anticoagulation was given to only 6 of 22 patients(27%) with PSMVT, in all for indications other thanPSMVT (deep vein thrombosis 4, stroke 2). Bleedingcomplications were seen in 2 patients while on anti-coagulation (hemorrhage into PFC and bleeding in sur-gical drains 10 days after surgery in one each). Inanother patient with necrosis and no PSMVT, hemor-rhage occurred in a pseudocyst while on anticoagulationfor pulmonary embolism.
Collateral formation was seen in 86% of patients (19of 22) with PSMVT and in 33% of patients (4 of 12) withnecrosis with venous narrowing but no PSMVT. All22 patients with necrosis and PSMVT underwent atleast 1 endoscopic procedure (32 EGDs, 35 transnasal
Table 2. Distribution, Interventions, Complications, and Selected Outcomes in Patients With PSMVT
Patient no.
PSMVTVenous
narrowing
Necrosis
PFC
Anticoagulation Intervention
Location Day seen Location Amount (%) Infected Given Indication Duration Type Timing (days)
1 SV 6 . Head/body <30 No No No . . None .2 SV 70 . Diffuse >50 Yes Yes No DVT 4 daysa Debridement 423 SV 16 . Body/tail >50 Yes No No . . Debridement 44 SV SMV PV 15 . Diffuse >50 No Yes No . . Debridement, cystgastrostomy 805 SV SMV PV 21 . Neck/body <30 No Yes No . . Debridement, cystgastrostomy 606 sv pv 14 . Head <30 Yes Yes No . . Debridement 407 sv 25 . None . No Yes Yes DVT 30 days None .8 sv 2 SV Diffuse >50 No Yes No . . Debridement, cystojejunostomy 1709 pv 8 . Diffuse >50 Yes No No . . None .
10 SMV pv 39 SV Diffuse >50 No Yes Yes DVT 4 daysa Cystgastrostomy 15411 smv 62 SV Diffuse >50 Yes Yes No . . Debridement 6712 SV 13 . Body/tail >50 No Yes No . . Distal pancreatectomy, splenectomy 27813 SV 18 SMV PV Diffuse >50 No Yes No . . Debridement, cystgastrostomy 7014 SV 16 . Body/tail 30–49 No Yes No . . Debridement, cystgastrostomy 6715 SV 1 . Peripancreatic . No Yes No . . None .16 SV SMV PV 50 . Body/tail >50 Yes Yes Yes DVT 30 daysa Debridement 2817 SV pv 1 . Neck/body >50 No Yes No Stroke >10 mo None .18 sv 11 . Diffuse >50 Yes Yes No . . Debridement 3219 sv 55 . Diffuse >50 Yes Yes No . . Cystgastrostomy 5520 SV 20 . Body/tail >50 No Yes No . . Debridement, cystgastrostomy 6821 SV SMV PV 74 . Neck/body <30 Yes Yes No . . Percutaneous drainage, transpapillary drainage 6722 SV 40 SMV PV Diffuse >50 Yes Yes Yes Strokeb >8 mo Debridement 48
NOTE. Thrombosis (in Location column under PSMVT column): Upper case indicates occlusive; lower case indicates non-occlusive.Collaterals developed in all except patients 6, 7, and 9. Clot resolution in patients 18 and 20. Interventions were surgical in all except patient 19 (endoscopic) and patient 21 (percutaneous, endoscopic). Complications:compartment syndrome, patient 3; hemorrhage in pseudocyst, patient 17; bleeding in surgical drains, patient 22.aInferior vena cava filter.bPositive anti-cardiolipin antibody.
May
2014Portosplenom
esentericVenous
Thrombosis
inAP
859
Table 3. Distribution, Interventions, Complications, and Selected Outcomes in Patients With Necrosis Without PSMVT
Patient no. Venous narrowing
Necrosis
PFC
Intervention
Location Amount (%) Infected Type Timing (days)
23 . Diffuse >50 No No Laparotomy (bedside) 224 . Tail <30 No Yes None .25 PV Neck/body 30–49 Yes Yes Debridement, pancreaticojejunostomy 1926 . Diffuse >50 No Yes None .27 . Neck <30 No Yes None .28 . Diffuse >50 No Yes Debridement 1229 SV Neck <30 No Yes None .30 PV SMV SV Diffuse >50 Yes Yes Debridement 3131 PV SV Peripancreatic N/A No Yes None .32 SV Body <30 No Yes None .33 SV Diffuse >50 Yes Yes Transpapillary stent,a debridement 5034 PV, SV Head <30 Yes Yes Debridement 2935 SV Head, body >50 Yes Yes Debridement 5436 . Body <30 No No None .37 SV Diffuse >50 No No None .38 Neck, body <30 Yes Yes Debridement 1239 SV Head, neck, body >50 No Yes Cystgastrostomy 5340 SV Head/body 30–49 No Yes Cystgastrostomy 7441 . Neck, body, tail >50 No Yes None .
NOTE. Hemorrhagic pseudocyst while on anticoagulation for pulmonary embolism, patient 34. Collaterals developed in patients 29, 37–40. Intervention wassurgical in all patients (laparoscopic, patients 39, 40).aEndoscopic stent.
860 Easler et al Clinical Gastroenterology and Hepatology Vol. 12, No. 5
endoscopies, 28 ERCPs; 15 underwent more than 1 typeof procedure) during follow-up for clinical indications(placement of nasojejunal feeding tube, biliary stent, etc).Varices were noted in 6 patients (4 esophageal, 2gastric). Of the 19 patients with necrosis and no PSMVT,15 underwent at least 1 endoscopic procedure (4 EGDs,22 transnasal endoscopies, 12 ERCPs; 8 underwent morethan 1 type of procedure). Varices were not noted inany patient (0 of 15 vs 6 of 22 in those with PSMVT;P ¼ .063).
Clot resolution was infrequent and seen in only 9% ofpatients (2 of 22) with PSMVT (documented on days 125and 349). No patient with PSMVT developed gastroin-testinal bleeding or other complications directly attrib-utable to PSMVT.
Pseudoaneurysms
Pseudoaneurysm related to AP was not detected inany patient who had CECT.
Discussion
In this well-defined cohort of AP patients from a majorU.S. center, we found the prevalence of PSMVT to be 14%overall and 18% among patients who had a CECT scan.Development of PSMVT was strongly associated with thepresence, extent, and location of pancreatic necrosis.Anticoagulation was used infrequently, and when used, itwas always for indications other than PSMVT. No com-plications were noted as a direct result of PSMVT. We
evaluated narrowing of portomesenteric veins. Finally, nopatient developed arterial pseudoaneurysm, indicatingthat this is a rare event in AP.
Our findings indicate that pancreatic necrosis and itsassociated inflammatory changes/fluid collections createa milieu that promotes development of PSMVT in AP. Therisk of PSMVT in necrosis patients was about 50%. It wasstriking that all but 1 of 22 patients with PSMVTdemonstrated necrosis. In the absence of pancreatic ne-crosis, PFCs did not increase the risk of PSMVT. Clinicalmarkers of disease severity also showed an associationwith PSMVT. Rather than a causal relationship, theylikely act as surrogates for pancreatic necrosis.
PSMVT prevalence in our study was lower than in arecentmeta analysis,18 similar to a recent United Kingdomstudy,19 and is likely higher than would be seen atnon-tertiary care centers. Our study allows for directdetermination of risk and extrapolation of results to non-tertiary care settings. By using the risk estimate of50% and a prevalence of pancreatic necrosis of5%–10%,5,6,22,23 the prevalence of PSMVT in AP in com-munity or non-tertiary care settings will be about 2%–5%.
The role of anticoagulation in modifying bleedingrisk, thrombotic events, recurrence-free, and overallsurvival has not been systematically studied in PSMVTrelated to AP. Among patients with PSMVT of all causes,these outcomes have been linked to the location, un-derlying cause, and use of anticoagulation.10,24–26
Bleeding is increased 2- to 3-fold in the presence ofgastroesophageal varices (especially large) and use ofanticoagulation,10,24 whereas the risk of local or distantrecurrence is increased in thrombophilic states. The
May 2014 Portosplenomesenteric Venous Thrombosis in AP 861
10-year recurrence-free survival is highest for isolatedSV thrombosis (97%), and overall survival is the highestfor isolated hepatic vein thrombosis (82%). Administra-tion of anticoagulation does not improve the recurrence-free survival but improves the overall survival.10
Direct extrapolation of these results to pancreatitispatients who form a small subset of patients with PSMVTis difficult. Although the risk of complications in thesepatients is real, it is low.14,18 The risk-benefit ratio doesnot favor routine use of anticoagulation in AP patientswith PSMVT because of several reasons. First, SV, whichis the most common site of PSMVT in AP, has the highestrecurrence-free survival.10 Second, the risk of PSMVT inpancreatitis does not appear to be related to an under-lying thrombophilic state.15 Third, SMV thrombus inpancreatitis typically develops close to the portal con-fluence with/without extension into SV or PV. The risk ofbowel infarction from thrombus in this location is lowwhen compared with thrombosis involving the smallerveins that can occlude the venous arcades and vasarecta.25 Fourth, most AP patients with PSMVT havepancreatic necrosis and associated PFCs; anticoagulationcan increase the risk of hemorrhage into PFCs. As notedearlier, bleeding complications were seen in 2 patientswhile on anticoagulation in our study. Therefore, in ourpractice, we do not routinely use anticoagulation in APpatients with PSMVT unless another obvious non-pancreatitis indication is present.
Strengths of our study include a well-defined cohort,blinded review of imaging by expert radiologists, directassessment of PSMVT risk, and extrapolation to AP pa-tients in community settings. In contrast to most priorstudies,18 we performed a comprehensive analysis ofPSMVT and venous narrowing in all splanchnic veins andalso evaluated the risk of arterial pseudoaneurysms. Weonly included patients with a sentinel attack of AP toensure that PSMVT caused by prior AP episodes or in-flammatory/fibrotic changes from coexisting CP does notresult in confounding. Because our study was limited toAP patients, we cannot comment on the relative contri-bution of pancreatic necrosis with its associated PFCs,recurrent inflammatory changes, pseudocysts, or directinvolvement of peripancreatic vessels by fibrotic processin the development of PSMVT in CP.
Our study has limitations. All patients did not un-dergo CECT scan during the index visit or systematicallyduring follow-up. Because of this, the actual prevalenceof PSMVT in our study cohort is likely higher than wehave observed. However, performance of imaging studiesis guided by the severity of AP and clinical symptomsrather than to evaluate the presence of PSMVT. More-over, increasing use and sensitivity of cross-sectionalimaging studies over time are likely associated withbetter detection of PSMVT.27 We did not screen all pa-tients who developed PSMVT for thrombophilic states(performed in only 2 patients with PSMVT). Thisapproach is supported by a recent study where theprevalence of thrombophilic states was similar in
alcoholic recurrent acute and CP patients with or withoutextrahepatic portal venous system thrombosis.15 We didnot observe complications directly related to PSMVT.Although this does not definitively exclude occurrencesof complications from PSMVT, it indicates that they areinfrequent to affect clinical management. Finally, ourcohort outlines the clinical course of only a subset of APpatients from our institution who enrolled in the SAPS.
In summary, although a common observation in ourtertiary care cohort of AP patients, PSMVT is likely aninfrequent finding in non-tertiary care settings. PSMVTdevelopment is closely related to the presence, location,and extent of pancreatic necrosis. Complications directlyrelated to PSMVT are infrequent.
References
1. Peery AF, Dellon ES, Lund J, et al. Burden of gastrointestinal
disease in the United States: 2012 update. Gastroenterology2012;143:1179–1187.
2. Banks PA, Freeman ML. Practice guidelines in acute pancrea-titis. Am J Gastroenterol 2006;101:2379–2400.
3. Forsmark CE, Baillie J. AGA Institute technical review on acutepancreatitis. Gastroenterology 2007;132:2022–2044.
4. Papachristou GI, Muddana V, Yadav D, et al. Comparison ofBISAP, Ranson’s, APACHE-II, and CTSI scores in predictingorgan failure, complications, and mortality in acute pancreatitis.Am J Gastroenterol 2010;105:435–441.
5. Saligram S, Lo D, Saul M, et al. Analyses of hospital adminis-trative data that use diagnosis codes overestimate the cases ofacute pancreatitis. Clin Gastroenterol Hepatol 2012;10:805–811.
6. Singh VK, Wu BU, Bollen TL, et al. Early systemic inflammatoryresponse syndrome is associated with severe acute pancrea-titis. Clin Gastroenterol Hepatol 2009;7:1247–1251.
7. Petrov MS, Shanbhag S, Chakraborty M, et al. Organ failure andinfection of pancreatic necrosis as determinants of mortality inpatients with acute pancreatitis. Gastroenterology 2010;139:813–820.
8. Mallick IH, Winslet MC. Vascular complications of pancreatitis.JOP 2004;5:328–337.
9. Mendelson RM, Anderson J, Marshall M, et al. Vascular com-plications of pancreatitis. ANZ J Surg 2005;75:1073–1079.
10. Thatipelli MR, McBane RD, Hodge DO, et al. Survival andrecurrence in patients with splanchnic vein thromboses. ClinGastroenterol Hepatol 2010;8:200–205.
11. Agarwal AK, Kumar RK, Agarwal S, et al. Significance of splenicvein thrombosis in chronic pancreatitis. Am J Surg 2008;196:149–154.
12. Bernades P, Baetz A, Levy P, et al. Splenic and portal venousobstruction in chronic pancreatitis: a prospective longitudinalstudy of a medical-surgical series of 266 patients. Dig Dis Sci1992;37:340–346.
13. Bradley EL 3rd. The natural history of splenic vein thrombosisdue to chronic pancreatitis: indications for surgery. Int J Pan-creatol 1987;2:87–92.
14. Heider TR, Azeem S, Galanko JA, et al. The natural history ofpancreatitis-induced splenic vein thrombosis. Ann Surg 2004;239:876–880.
15. Rebours V, Boudaoud L, Vullierme MP, et al. Extrahepatic portalvenous system thrombosis in recurrent acute and chronicalcoholic pancreatitis is caused by local inflammation and notthrombophilia. Am J Gastroenterol 2012;107:1579–1585.
862 Easler et al Clinical Gastroenterology and Hepatology Vol. 12, No. 5
16. Sakorafas GH, Sarr MG, Farley DR, et al. The significance ofsinistral portal hypertension complicating chronic pancreatitis.Am J Surg 2000;179:129–133.
17. Nordback I, Sisto T. Peripancreatic vascular occlusions as acomplication of pancreatitis. Int Surg 1989;74:36–39.
18. Butler JR, Eckert GJ, Zyromski NJ, et al. Natural history ofpancreatitis-induced splenic vein thrombosis: a systematic re-view and meta-analysis of its incidence and rate of gastroin-testinal bleeding. HPB (Oxford) 2011;13:839–845.
19. Gonzelez HJ, Sahay SJ, Samadi B, et al. Splanchnic veinthrombosis in severe acute pancreatitis: a 2-year, single-institution experience. HPB 2011;13:860–864.
20. Mounzer R, Langmead CJ, Wu BU, et al. Comparison of existingclinical scoring systems to predict persistent organ failure inpatients with acute pancreatitis. Gastroenterology 2012;142:1476–1482.
21. Gallego C, Velasco M, Marcuello P, et al. Congenital and ac-quired anomalies of the portal venous system. Radiographics2002;22:141–159.
22. Gislason H, Horn A, Hoem D, et al. Acute pancreatitis in Bergen,Norway: a study on incidence, etiology and severity. Scand JSurg 2004;93:29–33.
23. Lankisch PG, Karimi M, Bruns A, et al. Temporal trends inincidence and severity of acute pancreatitis in Luneburg County,Germany: a population-based study. Pancreatology 2009;9:420–426.
24. Condat B, Pessione F, Hillaire S, et al. Current outcome of portalvein thrombosis in adults: risk and benefit of anticoagulanttherapy. Gastroenterology 2001;120:490–497.
25. Kumar S, Kamath PS. Acute superior mesenteric venousthrombosis: one disease or two? Am J Gastroenterol 2003;98:1299–1304.
26. Plessier A, Darwish-Murad S, Hernandez-Guerra M, et al.Acute portal vein thrombosis unrelated to cirrhosis: a pro-spective multicenter follow-up study. Hepatology 2010;51:210–218.
27. Mortele KJ, Ip IK, Wu BU, et al. Acute pancreatitis: imagingutilization practices in an urban teaching hospital—analysis oftrends with assessment of independent predictors in correlationwith patient outcomes. Radiology 2011;258:174–181.
Reprint requestsAddress requests for reprints to: Dhiraj Yadav, MD, MPH, Division of Gastro-enterology and Hepatology, University of Pittsburgh Medical Center, 200Lothrop Street, M2, C-wing, Pittsburgh, Pennsylvania 15213. e-mail: [email protected]; fax: (412) 383-8992.
AcknowledgmentsPresented in part at the Digestive Diseases Week 2012 and published in ab-stract form in Gastroenterology 2012;142(Suppl 1):S-320.
Conflicts of interestThe authors disclose no conflicts.
