| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (M.E.B., D.E., B.H., E.R.) and Colorectal Surgery (F.R., M.O., V.F.), Cleveland Clinic Foundation, 9500 Euclid Ave, Cleveland, OH 44195. From the 2001 RSNA scientific assembly. Received February 4, 2002; revision requested April 9; revision received June 7; accepted July 25. Address correspondence to M.E.B. (e-mail: bakerm@ccf.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: During a 4-year period, 92 of 702 patients underwent contrast medium–enhanced CT after a total proctocolectomy with an IPAA. These CT scans were retrospectively reviewed for portal vein thrombus presence, location, and occlusive nature, as well as any accompanying enhancement abnormalities of the hepatic parenchyma. Only 13 patients who had initial CT scans that were positive for thrombi underwent follow-up examinations, and these were reviewed for resolution or progression of the original findings.
RESULTS: Portal vein thrombi were present in 41 (45%) of the 92 patients; 24 (59%) of the 41 were isolated, often multiple, segmental right lobe thrombi. Five patients had both right and left segmental vein involvement. Eleven patients had various combinations of main portal vein, right and left portal vein, or segmental vein thrombi. One patient had an isolated superior mesenteric vein thrombus. Twenty-two of 25 superior mesenteric vein, main portal vein, and right and left portal vein thrombi were nonocclusive, while most (63 of 86) of the segmental vein thrombi were occlusive. Wedge-shaped, peripheral areas of hepatic parenchymal hyperenhancement that were distal to the thrombi were present in 30 (73%) of the 41 patients. Follow-up scans obtained in the 13 patients with portal vein thrombi showed thrombi resolved in five patients, progression to cavernous transformation occurred in one patient, and parenchymal enhancement changes persisted in seven patients. In the seven patients with persistent enhancement changes, four had complete resolution of thrombi.
CONCLUSION: Portal vein thrombi appear to be relatively common after IPAA surgery and are most likely segmental, multiple, and occlusive. Peripheral wedge-shaped areas of hepatic parenchymal hyperenhancement commonly accompany these thrombi.
© RSNA, 2003
Index terms: Portal vein, CT, 957.12915 • Portal vein, thrombosis, 957.442 • Surgery, complications, 957.442
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
At our institution, IPAA is the procedure of choice in all patients with ulcerative colitis, and our colorectal surgery department performed over 170 of these procedures per year for the past 5 years (8). Recently, we have identified several of these patients in whom nonocclusive and/or intrahepatic segmental portal vein thrombi were identified on postoperative computed tomographic (CT) images. As a result of this, we undertook a retrospective review to determine the presence and location of portal vein thrombi in post-IPAA patients scanned with CT.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Of the 702 patients who underwent the procedure, 94 (13%) underwent a postoperative CT examination. Two of these patients did not receive intravenously administered contrast medium and were subsequently excluded from the study. High-density thrombus in the right and left portal vein was identified in one of these patients and was confirmed as thrombi at ultrasonography; however, because segmental vein involvement could not be assessed, the patient was excluded from the analysis. Thus, the study population comprised 92 patients.
Information concerning the patient’s surgery, surgical findings, septic complications, pathologic diagnosis, and clinical course form the basis of another, detailed report (9). The 92 patients in our study included 48 male patients and 44 female patients. The average age was 37.9 years with a range of 7–64 years. All but one of the patients had inflammatory bowel disease. Reasons for performing CT included the following: leukocytosis in 55 (60%) of the 92 patients, fever greater than 38°C in 53 (58%) of the 92 patients; abdominal pain in 49 (53%) of the 92 patients; an ileus present on a kidney, ureter, and bladder radiograph in 17 (18%) of the 92 patients; abdominal distention in 17 (18%) of the 92 patients; abdominal tenderness in 12 (13%) of the 92 patients; and decreased stoma output in four (4%) of the 92 patients. Often more than one of these signs and symptoms were present in a patient. The CT examinations were performed after surgery with a mean interval of 20 days (range, 3–70 days).
Imaging
All the CT examinations were performed on spiral CT scanners and included both the abdomen and pelvis. Sixty-one of the 92 examinations were performed with a single-section spiral scanner (Somatom Plus-4; Siemens Medical Systems, Erlangen, Germany). Of these 61 examinations, 43 were performed with 8-mm collimation and pitch of 1 and were reconstructed with 8-mm section thickness. The remaining 18 examinations were performed with 5-mm collimation and pitch of 1.5 and were reconstructed with 5-mm section thickness. Eighteen of the 92 examinations were performed with a multisection spiral scanner (Somatom Plus-4 Volume Zoom; Siemens Medical Systems) with 1.5-mm collimation and variable pitch depending on z-axis coverage and were reconstructed with 5-mm section thickness. Eight examinations were performed with a single-section spiral scanner (Emotion; Siemens Medical Systems) with 5-mm collimation and pitch of 1.5 and were reconstructed with 5-mm section thickness. Five examinations were performed with a single-section spiral scanner (AR-S; Siemens Medical Systems) with 10-mm collimation and pitch of 1 and were reconstructed with 10-mm section thickness.
All the patients in the study population received intravenously administered contrast medium (ioversol, Optiray-300; Mallinckrodt, St Louis, Mo). Scanning started 70 seconds after the injection of the contrast medium. Thus, scanning occurred during the portal venous phase of hepatic parenchymal enhancement. A power injector (CT 9000; Liebel-Flarsheim, Cincinnati, Ohio) was used to deliver 150 mL of the contrast medium at 2.5 mL/sec. It is our practice to attempt to instill contrast medium into the ileal pouch in order to detect a pouch leak in these patients; however, this is not always documented in the report, and sometimes installation of contrast medium into the pouch is not attempted. Furthermore, in some patients, an attempt is made to instill the contrast medium and the patient refuses or cannot retain the contrast medium; therefore, it is impossible to determine the total number of patients in whom we attempted to instill contrast medium into the pouch. When performed, this was accomplished by instilling dilute water-soluble contrast medium via a rectal tube by using gravity. Generally only 50–100 mL of contrast medium is instilled, per patient tolerance.
Image Review
The CT scans were retrospectively reviewed by one abdominal radiologist (M.E.B.) with 17 years of experience interpreting abdominal CT scans. The CT scans were reviewed without knowledge of the clinical presentation of the patient or of the CT report. The scans were reviewed for the presence or absence of thrombi in the portal venous system. Thrombi were identified as low-attenuation, intraluminal filling defects in the veins (1). In those patients in whom thrombi were identified, the specific location was documented. This included the SMV, main portal vein, the right and left portal vein, and any segmental veins. Segmental thrombi were identified as tubular structures in the portal triad that connected to an enhanced portal venous structure more proximally placed in the liver. Segmental vein involvement was categorized by using the Couinaud nomenclature (10). An estimate of the degree of occlusion or nonocclusion was also noted. If the vein was completely filled with thrombus and no peripheral contrast medium surrounded the thrombus, the thrombus was deemed occlusive. If any peripheral contrast medium was present either adjacent to or surrounding the thrombus, the thrombus was deemed nonocclusive.
The presence of peripheral wedge-shaped areas of hyperenhancement distal to segmental thrombi was documented. While postoperative fluid collections were documented during the image review, the presence of an abscess was confirmed later by accessing the records of the ileal pouch registry. A fluid collection that was aspirated and subsequently proven to be infected was considered an abscess. When contrast medium was identified on the CT scan as being in the ileal pouch, the presence of extraluminal contrast medium (ie, pouch leak) was documented. Because it was impossible to determine whether administration of contrast medium into the pouch had been attempted, the presence of a pouch leak was confirmed later by accessing the records of the ileal pouch registry. Some of the pouch leaks were identified with a water-soluble contrast medium enema that was diluted by 50% with water (iothalamate meglumine, Conray; Mallinckrodt).
Follow-up CT scans were available in 13 patients with findings of portal vein thrombi. These were evaluated by the same experienced radiologist (M.E.B.) for persistence or change in the thrombi and persistence or resolution of hepatic parenchymal hyperenhancement. The follow-up CT examinations were performed a mean of 40 days (range, 2–87 days) after the initial CT examinations.
All the scans were reviewed on a dedicated radiology workstation (MagicView 1000; Siemens Medical Systems). This allowed for rapid manual cine evaluation through the liver and abdomen, thus enabling the physician to easily identify intraluminal filling defects and connect them to the portal venous system. It also allowed for complete freedom in changing the window and level settings on images of the hepatic parenchyma.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
|
Twenty-two of 25 SMV, main portal vein, and right and left portal vein thrombi were nonocclusive. Occlusive thrombi were identified in only one of the nine patients with right portal vein involvement and in two of the nine with left portal vein involvement. Of the segmental vein thrombi, 63 of 86 were occlusive (six of 11 left lobe segmental veins and 57 of 75 right lobe segmental veins).
Distal and corresponding to the segmental location of the thrombus, wedge-shaped areas of hyperenhancement were identified in 30 (73%) of 41 patients (Fig 4). This finding was often subtle and could be identified only with narrow window and level settings that were manually adjusted by the reader. This finding was not present on any of the scans without thrombi.
In the patient population examined, an intraabdominal abscess was present in seven patients, and a pouch leak was present in five patients.
Follow-up Images
The follow-up scans indicated that the thrombi resolved completely in five of the 13 patients. These five patients demonstrated the following findings: Two patients had occlusive right and left segmental vein thrombi, one patient had nonocclusive right portal vein thrombus with occlusive segmental involvement, one patient had isolated occlusive right segmental vein thrombi, and one patient had isolated nonocclusive SMV thrombus. One patient had nonocclusive SMV and main portal vein thrombi along with occlusive segmental vein thrombi. Despite anticoagulation, the thrombi progressed to complete main portal vein thrombosis and cavernous transformation. The follow-up CT scans in the remaining seven patients showed either no resolution of the identifiable thrombi or resolution in some, but not all, of the identifiable thrombi.
On the follow-up scans, the wedge-shaped area of hyperenhancement persisted in seven of the 13 patients evaluated. Interestingly, four of the five patients who showed complete resolution of identifiable thrombi had persistence in these areas of hepatic parenchymal hyperenhancement.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Infection and preexisting abdominal inflammatory disease is also an important, but less common, cause of portal vein thrombosis (11–19). Relative to our study, important preexisting conditions that may predispose to or cause portal vein thrombosis include portal pyemia, sepsis after abdominal surgery, and inflammatory bowel disease. The exact mechanism of this in patients with inflammatory bowel disease is speculative. There are substantial differences in the hemostatic mechanism between patients with inflammatory bowel disease and other control patients. Patients with inflammatory bowel disease have increased platelet counts, factor V and VIII levels, and fibrinogen levels. Furthermore, antithrombin III levels are decreased (20,21). These findings suggest that patients with inflammatory bowel disease are hypercoagulable. These patients may also be dehydrated depending on the degree of diarrhea present.
Portal vein thrombosis has been reported following major bowel surgery (1–5,22). The mechanism of this after abdominal surgery is often unknown. In a recent study of 43 cases of portal vein thrombosis detected with CT, 20 patients had a history of bowel surgery 3 months prior to the CT (23). These procedures included resection of the small bowel, large bowel, or both. The mechanisms of this after total proctocolectomy with IPAA are likely multifactorial. The following causes may play a role: mobilization of the small bowel to the root of the mesentery, tension of the pouch anastomosis, and manipulation of the mesenteric vessels with high ligation of ileocolic vessels where they arise off the superior mesenteric vessels. Furthermore, inflammatory bowel disease itself and septic complications may also play a role. It is very likely that any small portal venous thrombi can embolize to the liver, which acts as an end organ sieve, much like the pulmonary veins in cases of deep venous thrombi in the lower extremities.
In portal vein thrombosis, the affected vein is completely occluded. Most of the clinical and radiologic literature is concerned with the cause, detection, and treatment of occlusive portal venous disease (1–5,12,14–19,23–31). There is almost no literature on the nonocclusive nature of thrombi in the portal venous system. Because many of our patients had nonocclusive disease, we prefer the term thrombus or thrombi to describe our experience.
Our results suggest that portal vein thrombi are common in patients after IPAA or restorative proctocolectomy. In our series, this condition was present in at least 6% of the patients after they underwent surgery. If all the patients in the 4-year period had undergone CT, the number would likely be greater.
The majority of our patients did not have typical CT findings of portal vein thrombosis, in that segmental, often peripheral veins were most commonly affected. Furthermore, when thrombi were present in the proximal larger veins (SMV and main portal, right portal, and left portal veins), only three of 25 were occluded. It was only in the more distal segmental branches that occlusion commonly occurred. The classic finding of a tubular low-attenuation filling defect completely filling and often expanding the SMV or portal vein was not the most common manifestation of thrombi in our series. There is little to no emphasis in the literature on nonocclusive thrombi. Furthermore, in our reading of the literature, segmental portal vein thrombi are rarely presented in figure form, and, when they are, they are not identified as such (14,23,28,29). When shown, they are almost always associated with occlusive proximal vein thrombi.
We do not know why there has been scant experience and emphasis in the literature on segmental portal venous thrombi. We speculate that segmental portal venous thrombi may be common only after proctocolectomy with IPAA. Our institution performs over 170 of these procedures a year, and it is likely that few other institutions perform as many. Alternatively, it may be that other centers and radiologists are unfamiliar with these thrombi; therefore, the segmental vein thrombi were not detected.
Another interesting observation is the multiplicity of segmental veins affected. The multiplicity of segmental vein involvement suggests that the mechanism of injury is a shower embolism phenomenon that occurs sometime during the surgical procedure. Most of the centrally located thrombi were nonocclusive, as opposed to the more peripherally located segmental thrombi. This would also support the theory that the thrombi are embolic.
As has been previously reported, abnormal hepatic parenchymal enhancement was identified in over 70% of our patients. Previous reports (14,23,28–30) have described areas of both hypoenhancement and hyperenhancement distal to the venous thrombus. In our patients, the only alteration in hepatic enhancement was peripheral, wedge-shaped areas of hyperenhancement corresponding to and extending distal from the affected segmental veins. The presence of this finding has been, in our subsequent, more recent experience, helpful in alerting us to the presence of subtle, isolated, segmental thrombi in post-IPAA patients. By narrowing the window and level setting on the workstation and by manual cine evaluation of multiple images, the reader can easily detect these changes. Once detected, a careful and thorough search for discrete, small, tubular structures extending from normally enhanced portal vein branches must be performed. Manual cine evaluation on the workstation, much like with a small-bowel obstruction, is, in our experience, essential for proper evaluation and detection. In fact, anecdotally we find that detecting these very small thrombi is very difficult on filmed CT scans.
The cause of hepatic parenchymal enhancement changes is likely secondary to compensatory increases in arterial flow when the portal vein is obstructed or when portal vein flow is impaired. These are not the changes associated with biphasic spiral CT (32).
We do not know the long-term outcome in these patients, as only a small number were subsequently studied. It is disconcerting that in one of our patients, nonocclusive SMV and portal vein thrombus progressed to complete thrombosis and cavernous transformation of the portal vein. This patient was prospectively identified with use of CT and was treated with anticoagulant therapy. In our limited experience with follow-up scans, some of the thrombi resolve, but many persist. It is also interesting to note that in most patients, when present on the initial scan, the hyperenhancing hepatic parenchyma remains even when the thrombus cannot be detected. This would suggest that the segmental portal veins affected have completely sclerosed or that blood flow has been altered.
We do not know the clinical importance of portal vein thrombi in IPAA patients or how these patients should be treated, especially those with only segmental thrombi. It is likely that many thrombi, especially those that occur in the segmental veins, are clinically unimportant. Our limited follow-up experience suggests that even when detected, they do not progress; however, we do not know if there are any untoward effects on the liver and portal venous system. We do have some theoretic concerns about occlusive, distal segmental venous thrombi. Empirically, we currently treat with anticoagulants all patients in whom portal venous thrombi are identified after IPAA, regardless of the location. We are also designing a prospective, consecutive study to evaluate a large cohort of patients after IPAA for the presence of portal venous thrombi. Included in this study will be an assessment of liver function.
In summary, both nonocclusive and segmental portal vein thrombi appear to be common after IPAA, especially in patients referred for CT evaluation after surgery. These patients present with typical signs and symptoms suggesting postoperative infection. Segmental portal vein thrombi are subtle and are identified as linear, thin, tubular structures coursing through the liver in the portal triad and connecting to enhancing normal, intrahepatic portal veins. They are commonly associated with peripheral, wedge-shaped areas of hyperenhancement in the hepatic parenchyma.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, M.E.B., F.R.; study concepts, M.E.B., F.R., D.E.; study design, M.E.B., F.R.; literature research, M.E.B., F.R., M.O.; clinical studies, M.E.B., D.E., F.R.; data acquisition and analysis/interpretation, M.E.B., M.O.; manuscript preparation, M.E.B., F.R., V.F.; manuscript definition of intellectual content, M.E.B., D.E., F.R.; manuscript editing, M.E.B., B.H., E.R., D.E., V.F.; manuscript revision/review, M.E.B., F.R., D.E., E.R., B.H., V.F.; manuscript final version approval, M.E.B., F.R.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. H. Choi, J. M. Lee, K. H. Lee, S. H. Kim, Y. J. Kim, S. K. An, J. K. Han, and B. I. Choi Relationship Between Various Patterns of Transient Increased Hepatic Attenuation on CT and Portal Vein Thrombosis Related to Acute Cholecystitis Am. J. Roentgenol., August 1, 2004; 183(2): 437 - 442. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
