Walter Wiesner 1,2 Koenraad J. Mortelé 1 Jonathan N. Glickman 3 Hoon Ji 1 Pablo R. Ros 1

Transcription

1 Walter Wiesner 1,2 Koenraad J. Mortelé 1 Jonathan N. Glickman 3 Hoon Ji 1 Pablo R. Ros 1 Received March 19, 2001; accepted after revision May 31, Department of Radiology, Brigham and Women s Hospital, Harvard Medical School, 75 Francis St., Boston, MA Present address: Institute of Diagnostic Radiology, University Hospital Basel, Petersgraben 4, 4031 Basel, Switzerland. Address correspondence to W. Wiesner. 3 Department of Pathology, Brigham and Women s Hospital, Harvard Medical School, Boston, MA AJR 2001;177: X/01/ American Roentgen Ray Society Pneumatosis Intestinalis and Portomesenteric Venous Gas in Intestinal Ischemia: Correlation of CT Findings with Severity of Ischemia and Clinical Outcome OBJECTIVE. The purpose of this study was to analyze the correlation between pneumatosis or portomesenteric venous gas, or both, the severity of mural involvement, and the clinical outcome in patients with small- or large-bowel ischemia. MATERIALS AND METHODS. CT scans of 23 consecutive patients presenting with pneumatosis or portomesenteric venous gas caused by bowel ischemia were reviewed. The presence and extent of both CT findings were compared with the clinical outcome in all patients and with the severity and extent of ischemic bowel wall damage as determined by surgery (15 patients), autopsy (three patients), or follow-up (five patients). RESULTS. Seven patients showed isolated pneumatosis, and 16 patients showed portomesenteric venous gas with or without pneumatosis (11 and five patients, respectively). Pneumatosis and portomesenteric venous gas were associated with transmural bowel infarction in 14 (78%) of 18 patients and 13 (81%) of 16 patients, respectively. Nine patients (56%) with portomesenteric venous gas died. Of seven patients with infarction limited to one bowel segment (jejunum, ileum, or colon), only one patient (14%) died, whereas of the 10 patients with infarction of two or three bowel segments, eight patients (80%) died. CONCLUSION. CT findings of pneumatosis intestinalis and portomesenteric venous gas due to bowel ischemia do not generally allow prediction of transmural bowel infarction, because they may be observed in patients with only partial ischemic bowel wall damage. The clinical outcome of patients with bowel ischemia with these CT findings seems to depend mainly on the severity and extent of their underlying disease. P neumatosis intestinalis and portomesenteric venous gas are impressive, but uncommon, radiologic findings that most commonly develop because of bowel ischemia [1]. The presence of portomesenteric venous gas on radiographs has been associated with a mortality rate of more than 75% in earlier studies [2, 3]. Over the last decade, detection of pneumatosis intestinalis and portomesenteric venous gas has improved because of the use of CT, and therefore, these two processes may be detected in earlier stages [4 7]. Furthermore, pneumatosis and portomesenteric venous gas may also be the result of various nonischemic conditions, including iatrogenic, traumatic, inflammatory, infectious, neoplastic, obstructive, and idiopathic causes, all of which are apparently not associated with an unfavorable outcome [8 17]. Recent articles on pneumatosis or portomesenteric venous gas suggest that the clinical significance of these two CT findings depends on the individual clinical setting in which they are observed [18 22]. However, these articles have included patients in whom portomesenteric gas was caused by certain nonischemic conditions, and to the best of our knowledge, only one study including seven patients has focused exclusively on the clinical outcome of patients in whom portomesenteric venous gas was caused by bowel ischemia [22]. Pneumatosis intestinalis and portomesenteric venous gas are usually discussed separately in the literature, and, to our knowledge, the presence of pneumatosis intestinalis or portomesenteric venous gas has not been compared with the severity of bowel wall damage and with the clinical outcome in patients in whom these findings were exclusively caused by bowel ischemia. Therefore, the purpose of our study was to describe the CT presentation of pneumatosis and portomesenteric venous gas in mesenteric ischemia and to correlate AJR:177, December

2 Wiesner et al. these CT findings with the severity of bowel wall damage and the clinical outcome in 23 consecutive patients. Materials and Methods Patients Between February 1992 and March 2000, 23 consecutive patients presented on CT scans with pneumatosis intestinalis or portomesenteric venous gas, or both, caused by small- or large-bowel ischemia. Patients were identified by searching for key words such as hepatic gas or air, portal gas or air, portal venous gas or air, mesenteric venous gas or air, pneumatosis or intestinal ischemia, bowel ischemia, and bowel infarction in our computer database (IDXRAD; IDX, Burlington, VT). Patients in whom pneumatosis or portomesenteric venous gas was absent or not attributable to bowel ischemia were excluded from our study. Our study population included 16 men and seven women with an age range of years (mean age, 63.4 years). Imaging Technique and Analysis Patients were examined using a Somatom, a Somatom Plus 4, or a Volume Zoom CT scanner (Siemens Medical Systems, Erlangen, Germany). Seven patients underwent contrast-enhanced CT, and 16 patients underwent unenhanced CT. The CT scans were reviewed by two observers in consensus, and the presence or absence of pneumatosis and portomesenteric venous gas was assessed. If pneumatosis was present, it was classified as bubblelike if it consisted only or mainly of isolated bubbles of air in the bowel wall, whereas it was classified as bandlike, if it consisted of continuous bands of air in the affected bowel wall. If both types of pneumatosis were found, the more pronounced (bandlike pattern) was described. If pneumatosis intestinalis was present, it was also classified as involving the small or the large bowel, or both. The amount of portomesenteric gas at CT was assessed semiquantitatively. Portomesenteric venous gas was considered limited if visible mainly as focal gaseous inclusions on certain images, whereas it was considered pronounced if visible as multiple continuous bands of gas on several images. No distinction was made between gas in the mesenteric veins or gas in the intrahepatic branches of the portal vein, because all these findings were indicative of gas in the portomesenteric circulation (Figs. 1 and 2). Correlation Analysis Surgical and pathologic reports on the resected bowel segments were available in 15 patients. These reports were reviewed, and special emphasis was given to the presence or absence of transmural infarction. In 14 of these patients, the operation was performed the same day as CT, and one patient underwent surgery only the following day. Two patients who died the same day and one patient who died the next day underwent autopsy, and in these three patients, the autopsy reports were reviewed. In three patients who died from acute mesenteric infarction the same day as CT was performed, no surgery and no autopsy were performed. However, it could be assumed that they had undergone transmural small-bowel infarction due to their acute and dramatic clinical course. They all died within several hours from abdominal shock, and there was no suspicion of any other cause of death. In these three patients, CT showed pronounced pneumatosis of the jejunum and ileum and also showed extended portomesenteric venous gas. In another two patients, no histopathology was available because these two patients did not undergo surgery. Nevertheless, because they both recovered completely after conservative treatment, transmural bowel infarction could be excluded. On the basis of a combination of the previously mentioned radiologic, surgical, and pathologic findings, the extent of bowel ischemia was assessed semiquantitatively by dividing the entire bowel into three segments (jejunum, ileum, and colon) and by analyzing for each patient how many of these segments were involved by transmural and partial mural ischemia. Finally, all clinical records were reviewed, and the clinical outcome of all patients was compared with the presence of pneumatosis or portomesenteric gas, to the presence of transmural and only partial mural bowel ischemia, and also to the extent of transmural bowel infarction. Results Study Group In our 23 patients, the etiologies of intestinal ischemia included occlusions of visceral arteries by atherosclerosis or athero- or thromboembolism (11 patients), a combination of vasculoocclusive factors and a low flow state (six patients), cardiogenic shock (two patients), occlusions of the mesenteric arteries caused by vasculitis (one patient), thrombotic occlusions of the mesenteric veins caused by a thrombotic thrombocytopenic purpura (one patient), strangulation in small bowel volvulus (one patient), and pronounced prestenotic dilatation of the colon (one patient). Radiologic Findings Seven patients presented with isolated pneumatosis, and 16 patients showed portomesenteric venous gas with (11 patients) or without (five patients) pneumatosis (Fig. 1). In those patients who presented with pneumatosis but with- A B Fig year-old woman with transmural infarction of distal colon. A, Unenhanced CT scan shows intrahepatic portal venous gas in left liver lobe. B, Unenhanced CT scan shows marked wall thickening of infarcted colon (arrows). Note absence of pneumatosis. Patient survived AJR:177, December 2001

3 CT Findings in Intestinal Ischemia Fig year-old woman with only partial mural small-bowel ischemia. Contrast-enhanced CT scan shows bubblelike pneumatosis (arrow) and bandlike pneumatosis (arrowhead) in proximal small bowel and pronounced mesenteric venous gas. Patient survived. out portomesenteric venous gas, pneumatosis was present in the small bowel in three and in the large bowel in four patients. In six of these patients, pneumatosis was bubblelike (Figs. 2 4), whereas in one of these patients pneumatosis was bandlike (Figs. 5 and 6). In those 11 patients who had shown pneumatosis and portomesenteric venous gas, seven patients presented with pneumatosis of their small bowel, one patient showed pneumatosis of the large bowel, and in three patients, pneumatosis was found in both the small and large bowel. In seven patients, pneumatosis was bandlike, whereas in another four patients, pneumatosis was bubblelike. Correlation with Severity of Bowel Ischemia Of all 18 patients who presented with pneumatosis, four showed partial mural bowel ischemia (two of the small bowel and two of the large bowel), and the other 14 patients (78%) showed transmural infarction of their large bowel (three patients), small bowel (eight patients), or both (three patients). Overall bubblelike pneumatosis was associated with transmural bowel infarction in seven (70%) of 10 patients, whereas bandlike pneumatosis was associated with transmural bowel infarction in seven (88%) of eight patients. Of the seven patients who presented with isolated pneumatosis, four patients (57%) showed transmural infarction of their large bowel (two patients) or their small bowel (two patients). The remaining three patients with isolated pneumatosis showed only partial mural ischemia of their small bowel (one patient) or of their large bowel (two patients). In one patient, pneumatosis was bandlike, whereas in the remaining six patients, it was bubblelike. Overall, 13 (81%) of the 16 patients with portomesenteric venous gas showed transmural infarction of their small bowel (six patients), of their large bowel (two patients), or of both small and large bowel (five patients) (Figs. 2, 4, and 5). Three (19%) of 16 patients with portomesenteric venous gas showed only a partial mural ischemia of their large bowel (one patient) and small bowel (two patients). Of the five patients who presented with portomesenteric venous gas, but without pneumatosis, one patient showed transmural infarction of the large bowel and two patients showed transmural infarction of both small and large bowel, whereas the other two patients showed only partial mural ischemia of the large bowel (one patient) or of the small bowel (one patient). Of the 11 patients with portomesenteric venous gas and pneumatosis, all but one showed transmural bowel infarction involving the large bowel (one patient), the small bowel (six patients), or both (three patients). Correlation with Clinical Outcome All patients (n = 6) who showed only partial mural bowel ischemia survived and only patients with transmural bowel infarction died. Fig year-old man with transmural small-bowel infarction. Unenhanced CT scan shows bubblelike pneumatosis in small bowel (arrows), but no mesenteric venous gas. Patient survived. Fig year-old man with only partial mural ischemia of cecum and ascending colon. Unenhanced CT scan shows bubblelike pneumatosis of ascending colon (arrows), but no mesenteric venous gas. Patient survived. AJR:177, December

4 Wiesner et al. Fig year-old man with transmural colonic infarction. Unenhanced CT scan shows bandlike pneumatosis of ascending and transverse colon (arrows), but no portomesenteric venous gas. Patient survived. Overall, of 17 patients with transmural bowel infarction, nine patients (53%) died. Of those seven patients with transmural infarction of one bowel segment, only one patient (14%) died. Nevertheless, of those patients with transmural infarction of two or three bowel segments, five (71%) of seven and three (100%) of three patients died respectively. All seven patients who presented with isolated pneumatosis had ischemia of only one bowel segment, and all these patients survived. Conversely, of 16 patients with portomesenteric venous gas, nine patients (56%) died. Eight of these nine patients had transmural infarction of two or three bowel segments, whereas the remaining patient had transmural bowel infarction involving only one segment. Four patients (25%) with portomesenteric venous gas survived despite transmural infarction of the colon (two patients), of the small bowel (one patient), or of both the small and large bowel (one patient). The remaining three patients (19%) with portomesenteric venous gas also survived, all of whom had only partial mural ischemia of their large bowel (one patient) or small bowel (two patients). Discussion The CT findings of pneumatosis intestinalis and portomesenteric venous gas usually indicate the presence of mesenteric infarction. However, both findings may also be observed occasionally in nonischemic conditions [1 7]. Pneumatosis and portomesenteric venous gas may occur after damage to the gastrointestinal wall caused by infection and inflammation, but also by neoplastic bowel wall damage, ulceration, or overdistention [8, 9]. Besides, portomesenteric venous gas may occur without pneumatosis in cases of mesenteric abscess formation, portal pylephlebitis, sepsis, abdominal trauma and, especially, after gastrointestinal surgery and liver transplantation [9 17]. For these reasons, the presence of pneumatosis intestinalis, and the presence of portomesenteric venous gas should not be always regarded as ominous signs, because they may not be associated with increased mortality rates in some conditions [18 20]. Previously, patients having findings of portomesenteric venous gas on radiographs had a reported mortality rate of 75%; if iatrogenic causes were excluded, this rate was as high as 84%. Thus, the presence of portomesenteric venous gas has been previously indicative of an unfavorable clinical outcome, with a poor prognosis in patients having bowel ischemia [2, 3]. Because portomesenteric venous gas is currently usually detected at CT and not by unenhanced radiographs, the clinical outcome of patients with portal venous gas might have improved because of its earlier detection. Therefore, some more recent articles focused on the clinical impact of portomesenteric venous gas at CT. However, most of these studies represented only case reports including up to seven patients. Similar to one larger study that reported the clinical outcome of 17 patients with portal venous gas, they were not focused on patients with the same cause, but also included patients in whom portomesenteric venous gas was caused by nonischemic conditions [18 21]. Although one recent article reported on Fig year-old man with extended transmural infarction of small and large bowel. Contrast-enhanced CT scan shows bandlike pneumatosis of multiple smallbowel loops and colon and mesenteric venous gas (arrow). Patient died. seven patients in whom portomesenteric venous gas was exclusively caused by intestinal ischemia, to our knowledge, no other study has addressed the question of whether the clinical outcome of these patients might have changed over the last decade [22]. Furthermore, to our knowledge, no one has analyzed whether the presence of pneumatosis intestinalis and portomesenteric venous gas allows prediction of the severity of ischemic bowel wall damage and whether the clinical outcome of patients with bowel ischemia and isolated pneumatosis differs from that of patients who additionally show portomesenteric venous gas at CT. Therefore, we focused our study on both, pneumatosis and portomesenteric venous gas, but exclusively in patients in whom these two CT findings were related to bowel ischemia. In seven of these patients, pneumatosis intestinalis occurred as an isolated finding; in 11 patients, it was combined with portomesenteric venous gas; and in five patients, portomesenteric venous gas occurred as an isolated finding. Pneumatosis intestinalis may logically occur without portomesenteric venous gas in bowel ischemia because accumulation of intramural gas in an ischemic bowel segment does not automatically mean that the intramural gas has also found its way into some mesenteric veins. However, if this scenario develops, portomesenteric venous gas usually occurs as an additional finding with pneumatosis. Nonetheless, even if rare, occasionally after an ischemic damage to the intestinal wall, intraluminal gas may enter the portomesenteric veins directly without producing a radiologically detectable pneumatosis (Fig. 1) AJR:177, December 2001

5 CT Findings in Intestinal Ischemia Overall, pneumatosis was associated with transmural bowel infarction in 78% of patients, and this result shows that pneumatosis intestinalis may occasionally occur in ischemic bowel segments that have not yet undergone transmural infarction. Bubblelike pneumatosis was associated with transmural bowel infarction in only 70% of patients, whereas bandlike pneumatosis was associated with transmural bowel infarction in almost 90% of patients and, therefore, was more specific for a full-thickness necrosis of the affected bowel wall. Overall, portomesenteric venous gas was associated with transmural bowel infarction in 81% of our patients, and pronounced portomesenteric gas was also associated slightly more often with transmural bowel infarction (86%) than mild portomesenteric venous gas (78%) was. However if both CT findings of pneumatosis and portomesenteric venous gas were seen, their presence was associated with transmural bowel infarction in 91% of patients, regardless of their aspect and extent. Our results show that pneumatosis and portomesenteric venous gas may occasionally be found even in patients with only partial mural bowel ischemia. Therefore, neither pneumatosis nor portomesenteric venous gas differentiate transmural bowel infarction and only partial mural bowel ischemia, if they are encountered as mild and isolated findings. However, transmural infarction of the affected bowel becomes likely if pneumatosis is pronounced and bandlike, and, especially, if it is combined with portomesenteric venous gas. All our patients with isolated pneumatosis survived, but these results are surely influenced by the fact that in all these seven patients, bowel ischemia or bowel infarction involved only one bowel segment. Overall, pneumatosis was associated with a mortality rate of 44%, and portomesenteric venous gas was associated with a mortality rate of 56%. In those patients who showed portomesenteric venous gas and pneumatosis at CT, the mortality rate increased to 72%. However, only patients with transmural bowel infarction died, and there was a clear difference in the mortality rate in patients in whom transmural infarction involved only one bowel segment (14%) and in the mortality rate in patients in whom bowel infarction involved two (71%) or three bowel segments (100%). These results support the theory that it is not only the severity of bowel ischemia and the presence of a full-thickness bowel wall necrosis but also the extent of transmural bowel infarction that mainly determine the clinical outcome of affected patients. The clinical outcome of patients with intestinal ischemia also depends on the time delay between diagnosis and surgical exploration. By using CT, pneumatosis and portomesenteric venous gas may be detected earlier and in less pronounced stages than by using radiography, and besides some improvements in intensive care medicine, this may be the most important reason why the presence of portomesenteric venous gas in our study was not associated with the same high mortality rate as that reported in the late 1970s and 1980s. One may, however, argue that the CT finding of portomesenteric venous gas is associated with a mortality rate of 56% in bowel ischemia and that it, therefore, has not yet completely lost its ominous character in these high-risk patients over the last decade. However, this correlation is presumably only indirect and not causative because the clinical outcome of affected patients seems to depend mainly on the presence and extent of transmural bowel infarction. Nevertheless, because it is still possible that portomesenteric venous gas additionally worsens the already bad prognosis of patients with transmural bowel infarction, future studies will be needed to prove if patients with bowel infarction and portomesenteric venous gas have a higher mortality rate compared with patients with similar severity and extent of bowel ischemia but without portomesenteric venous gas. In conclusion, bandlike pneumatosis and the combination of pneumatosis and portomesenteric venous gas at CT are highly associated with transmural bowel infarction. On the other hand, bubblelike pneumatosis and isolated portomesenteric gas may be related to only partial mural bowel ischemia in approximately one third of cases. Furthermore, although in bowel ischemia, the presence of portomesenteric venous gas at CT is associated with a mortality rate of 56%, this association is presumably only indirect. Therefore, the ominous character of this CT finding seems to be justified only in patients with extended transmural bowel infarction, and their clinical outcome seems to depend mainly on the severity and extent of their underlying disease. References 1. Schulze CG, Blum U, Haag K. Hepatic portal venous gas: imaging modalities and clinical significance. Acta Radiol 1995;36: Liebman PR, Patten MT, Manny J, Benfield JR, Hechtman HB. Hepatic-portal venous gas in adults: etiology, pathophysiology and clinical significance. Ann Surg 1978;187: Griffiths DM, Gough MH. Gas in the hepatic portal veins. Br J Surg 1986;73: Smerud MJ, Johnson CD, Stephens DH. Diagnosis of bowel infarction: a comparison of plain films and CT scans in 23 cases. AJR 1990;154: Alpern MB, Glazer GM, Francis IR. Ischemic or infarcted bowel: CT findings. Radiology 1988;166: Kelvin FM, Korobkin M, Rauch RF, Rice RP, Silverman PM. Computed tomography of pneumatosis intestinalis. J Comput Assist Tomogr 1984;8: Connor R, Jones B, Fishman EK, Siegelman SS. Pneumatosis intestinalis: role of computed tomography in diagnosis and treatment. J Comput Assist Tomogr 1984;8: Feczko PJ, Mezwa DG, Farah MC, White BD. Clinical significance of pneumatosis of the bowel wall. RadioGraphics 1992;12: Sebastia C, Quiroga S, Espin E, Boye R, Alvarez- Castells A, Armengol M. Portomesenteric vein gas: pathologic mechanism, CT findings and prognosis. RadioGraphics 2000;20: Zhang D, Weltman D, Baykal A. Portal vein gas and colonic pneumatosis after enema, with spontaneous resolution. AJR 1999;173: Quirke TE. Hepatic-portal venous gas associated with ileus. Am Surg 1995;61: Zielke A, Hasse C, Nies C, Rothmund M. Hepatic-portal venous gas in acute colonic diverticulitis. Surg Endosc 1999;32: Gurland B, Dolgin SE, Shlasko E, Kim U. Pneumatosis intestinalis and portal vein gas after blunt abdominal trauma. J Pediatr Surg 1998;3: Brown MA, Hauschildt JP, Casola G, Gosink BB, Hoyt DB. Intravascular gas as an incidental finding at US after blunt abdominal trauma. Radiology 1999;210: Wakisaka M, Mori H, Kiyosue H, Kamegawa T, Uragami S. Septic thrombosis of the portal vein due to peripancreatic ligamental abscess. Eur Radiol 1999;9: Mallens WM, Schepers-Bok R, Nicolai JJ, Jakobs FA, Heyerman HGM. Portal and systemic venous gas in a patient with cystic fibrosis: CT findings. AJR 1995;165: Chezmar JL, Nelson RC, Bernardino ME. Portal venous gas after hepatic transplantation: sonographic detection and clinical significance. AJR 1989;153: Muscari F, Suc B, Lagarrigue J. Hepatic portal venous gas: is it always a sign of severity and surgical emergency [in French]? Chirurgie 1999; 124: Scheidler J, Stabler A, Kleber G, Neidhardt D. Computed tomography in pneumatosis intestinalis: differential diagnosis and therapeutic consequences. Abdom Imaging 1995;20: Hong JJ, Gadaleta D, Rossi P, Esquivel J, Davis JM. Portal vein gas, a changing clinical entity: report of 7 patients and review of the literature. Arch Surg 1997;132: Faberman RS, Mayo-Smith WW. Outcome of 17 patients with portal venous gas detected by CT. AJR 1997;169: Monneuse O, Gruner L, Henry L, et al. Hepatic portal venous gas. Ann Chir 2000;125: AJR:177, December