Penetrating Torso Trauma: Triple-Contrast Helical CT in Peritoneal Violation and Organ Injury A Prospective Study in 200 Patients 1

Transcription

1 Emergency Radiology Radiology K. Shanmuganathan, MD Stuart E. Mirvis, MD William C. Chiu, MD Karen L. Killeen, MD 2 Gerald J. F. Hogan, MD Thomas M. Scalea, MD Index terms: Computed tomography (CT), helical, Computed tomography (CT), multi detector row, Emergency radiology Trauma Published online before print /radiol Radiology 2004; 231: From the Department of Diagnostic Radiology (K.S., S.E.M., K.L.K., G.J.F.H.) and Maryland Shock-Trauma Center (K.S., S.E.M., W.C.C., T.M.S.), University of Maryland Medical Center, 22 S Greene St, Baltimore, MD From the 2002 RSNA scientific assembly. Received February 6, 2003; revision requested April 2; final revision received August 4; accepted September 29. Address correspondence to K.S. ( kshanmuganathan@umm.edu). Current address: 2 Commonwealth Radiologic Associates, Richmond, Va. Penetrating Torso Trauma: Triple-Contrast Helical CT in Peritoneal Violation and Organ Injury A Prospective Study in 200 Patients 1 PURPOSE: To assess the accuracy of computed tomography (CT) in demonstrating the presence or absence of peritoneal violation and type of intraperitoneal organ injury, if any, in hemodynamically stable patients with penetrating torso trauma but without definite peritoneal signs or radiographic evidence of free intraperitoneal air. MATERIALS AND METHODS: During a 29-month period, helical CT with oral, rectal, and intravenous contrast material (triple-contrast) was performed in 200 hemodynamically stable patients, including 169 men (age range, years; mean age, 31 years) and 31 women (age range, years; mean age, 28 years) with penetrating torso trauma. The study group included 86 patients with gunshot wounds, 111 with stab wounds, and three impaled by sharp objects. CT scans were evaluated prospectively by three trauma radiologists for evidence of peritoneal violation to determine injury to intra- or retroperitoneal solid organs, bowel, mesentery, vascular structures, diaphragm, and urinary tract. Sensitivity, specificity, and accuracy of CT in the diagnosis of peritoneal violation were determined. RESULTS: CT findings aided diagnosis of peritoneal violation in 34% of patients (68 of 200) and were negative for peritoneal violation in 66% of patients (132 of 200). Two patients with negative CT findings failed to improve with observation and underwent therapeutic laparotomy. CT had 97% sensitivity (66 of 68 findings), 98% specificity (130 of 132 findings), and 98% accuracy (196 of 200 findings) for peritoneal violation. CT aided diagnosis of 28 hepatic, 34 bowel or mesenteric, seven splenic, and six renal injuries. Laparotomy based on CT findings in 38 patients was considered therapeutic in 87% (33 of 38) and nontherapeutic in 8% (three of 38) and had negative results in 5% (two of 38). CONCLUSION: Triple-contrast helical CT accurately demonstrates peritoneal violation and visceral injury in patients with penetrating torso wounds. RSNA, 2004 Author contributions: Guarantors of integrity of entire study, K.S., S.E.M., W.C.C.; study concepts, K.S., S.E.M., W.C.C., T.M.S.; study design, K.S., S.E.M., W.C.C.; literature research, K.S.; clinical studies, K.S., W.C.C., K.L.K.; data acquisition, K.S., W.C.C., G.J.F.H.; data analysis/interpretation, K.S., S.E.M., G.J.F.H.; statistical analysis, K.S.; manuscript preparation, definition of intellectual content, editing, revision/review, and final version approval, K.S., S.E.M. RSNA, 2004 Although computed tomography (CT) has been the imaging modality of choice in the evaluation of hemodynamically stable patients with blunt abdominal trauma, routine use of CT in penetrating trauma has generally been implemented only in patients with entry sites in the back and flank (1 11). Asymptomatic patients with penetrating back and flank trauma infrequently have associated injury to critical retroperitoneal viscera, since these organs are well protected by the ribs, spine, and large back and paraspinal muscles. Patients who sustain these wounds may not present with overt clinical signs or symptoms and are not well evaluated by means of diagnostic peritoneal lavage (8,12). Use of CT with administration of oral, rectal, and intravenous contrast material (triple-contrast) has been advocated during the past decade because of its ability to optimally demonstrate injuries to the retroperitoneal organs in patients with penetrating back and flank injuries (8 11). In two studies (8,11), however, diagnostic peritoneal lavage or local wound exploration was used initially in addition to CT to evaluate concurrent intraperitoneal injury. 775

2 Surgeons have been reluctant to request CT routinely to evaluate hemodynamically stable patients without peritoneal signs who sustain penetrating torso wounds (abdomen, thoracoabdominal region, and pelvis), excluding the flank and back, because of concern about the uncertain accuracy of CT in the diagnosis of gastrointestinal injuries (13 15). Bowel injuries occur much more often after penetrating trauma than after blunt trauma and may remain clinically occult for several hours after trauma occurs (16 19). Both retrospective and prospective studies (20 23) have shown helical CT to be safe and highly accurate in the assessment of wound trajectory and in the determination of whether peritoneal violation has occurred in selected patients with torso gunshot wounds. The purpose of our prospective study was to assess the accuracy of CT in demonstrating the presence or absence of peritoneal violation and any associated visceral injury. MATERIALS AND METHODS Subjects The current report is an extension of our prior report (22) in which 104 patients were included. During a 29-month period, 540 patients with penetrating torso injury were admitted to our level 1 trauma center. Laparotomy was performed in 29% of patients (158 of 540) without the use of CT because the patient was either hemodynamically unstable or had evidence of peritonitis (rigidity, rebound tenderness, considerable tenderness away from the wound site) at clinical examination, evisceration, free intraperitoneal fluid at focused abdominal sonography for trauma, rectal bleeding, hematemesis, or free intraperitoneal air on radiographs of the abdomen or chest. CT was not performed in another 26% of patients (140 of 540) for the following reasons: The penetrating injury was considered to be superficial and did not require further evaluation with CT, the patient had a known major allergy to intravenous contrast material, or CT would introduce an unacceptable delay in treatment of injuries, as judged by the attending surgeon. Our institutional review board approved our study. Two hundred forty-two hemodynamically stable patients underwent triplecontrast CT during the study period. These patients maintained a systolic blood pressure higher than 90 mm Hg, base deficit less than 5, and pulse rate higher than 50 but lower than 110 beats per minute. Clinical examination findings were normal or equivocal for signs of peritonitis, and there was no evidence of free intraperitoneal air on conventional radiographs of the abdomen or chest. Informed consent to participate in the study was obtained from 200 patients, who form our study population. There were 169 male patients (age range, years; mean age, 31 years) and 31 female patients (age range, years; mean age, 28 years) with an age range of years (mean age, 30 years). No statistical difference (P.22) was observed for the reporting of age distribution according to patient sex. Mechanism of injury included gunshot wounds in 43% of patients (86 of 200), stab wounds in 56% (111 of 200), and impalement on sharp objects in 2% (three of 200). Multiple entry sites were seen in 20% of patients (17 of 86) admitted with gunshot wounds and 9% of patients (10 of 111) with stab wounds. Eleven patients with multiple entry sites had two torso penetrations. The torso is defined surgically by five anatomic regions to localize penetration site(s) and includes the thoracoabdominal region, abdomen, flank, back, and pelvis. The thoracoabdominal region is defined by the internipple line superiorly and costal margin inferiorly extending posteriorly up to the inferior tip of the scapula. The abdomen is defined by the costal margin superiorly, anterior axillary lines laterally, and inguinal ligaments and symphysis pubis inferiorly. The flank is defined by the costal margin superiorly, anterior and posterior axillary lines laterally, and iliac crest inferiorly. The interscapular line, between the inferior scapular tips superiorly and the iliac crests inferiorly, defines the back. The pelvis is defined as extending inferiorly from the inguinal ligaments and iliac crests through the upper thirds of the thigh inferiorly. Techniques Triple-contrast single detector row helical CT or multi detector row helical CT scans (Plus 4, Siemens Medical Systems, Iselin, NJ; or MX 8000, Philips Medical Systems, Best, the Netherlands) were obtained from the internipple line to the symphysis pubis with a collimation of 8 mm and table speed of 8 mm/sec or a collimation of 2.5 mm (2.5 4 images per rotation) and table speed of 10 mm/ sec, respectively. Change in scanning protocol occurred because the single detector row helical CT scanner in the trauma center was replaced by a multi detector row helical CT scanner during the study period. CT scans were acquired by using the single detector row CT scanner in 136 patients and the multi detector row CT scanner in 64 patients. Intravenous contrast material (240 mg of iodine per milliliter, 150 ml) was administered at 3 ml/sec by using a power injector (Medrad 4; Medrad, Pittsburgh, Pa). A total volume of 600 ml of 2% sodium diatrizoate (Hypaque sodium; Nycomed, Princeton, NJ) oral contrast material was administered 30 minutes before and immediately prior to initiation of scanning. An enema with L of 2% sodium diatrizoate (Hypaque sodium; Nycomed) was also administered in the CT suite prior to initiation of scanning. Delayed images (2 3 minutes following initiation of intravenous contrast material injection) were routinely obtained in the renal excretory phase to evaluate the renal collecting systems. Definitions and Image Analysis Three trauma radiologists with 2 17 years of experience (K.S., 12 years; S.E.M., 17 years; K.L.K., 2 years) interpreted the CT scans prospectively. The scans were assessed for evidence of peritoneal violation to determine injury to intra- or retroperitoneal solid organs, bowel, mesentery, vascular structures, diaphragm, and renal collecting system and bladder. CT findings were considered positive if there was peritoneal violation, injury to the retroperitoneal bowel or major vessels, specific signs of diaphragm injury, or injury to the renal collecting system or bladder. CT findings of peritoneal violation included a wound track outlined by hemorrhage, air, bullet, or bone fragments that clearly extended into the peritoneal cavity; the presence of intraperitoneal free fluid, free air, or bullet fragment(s); and intraperitoneal organ injury. CT findings were considered negative if there was no evidence of peritoneal violation with or without minor extraperitoneal injuries that did not require intervention, such as minor renal injuries or small retroperitoneal or pelvic hematomas. Specific CT findings of bowel or mesenteric injury included extravasation of oral or rectal contrast material; a penetrating injury track outlined by hemorrhage, air, or bullet fragments extending to the bowel wall; focal thickening or discontinuity of the bowel wall; active mesenteric bleeding; or infiltration or focal thickening of the mesentery (24,25). 776 Radiology June 2004 Shanmuganathan et al

3 Figure 1. Transverse CT scans show peritoneal violation in a 37-year-old woman stabbed multiple times in the chest and abdomen. (a) Large liver laceration (straight arrow) with active bleeding (curved arrow). Free intraperitoneal fluid (black arrowheads) is seen anterior to the liver. Air bubbles (white arrowheads) are seen along the knife track. (b, c) Linear laceration (arrow in b) in pancreatic body with free fluid in lesser sac (arrowheads). Gastric contrast material extravasation (arrow in c) is also seen in the lesser sac. Injuries to the liver, stomach, and pancreas were confirmed surgically. (Adapted and reprinted, with permission, from reference 27.) Free intraperitoneal air can be introduced into the peritoneal cavity by a bullet or knife wound. Isolated free intraperitoneal fluid (blood) can enter the peritoneal cavity from an extraperitoneal injury through the wound track or can be seen after injury occurs in the peritoneal lining itself. Thus, CT findings of either free peritoneal air or fluid were considered to be signs of peritoneal violation but not bowel injury. Vascular injury was diagnosed in the presence of active bleeding, pseudoaneurysm, or posttraumatic arteriovenous fistula. Active bleeding was diagnosed when a linear or irregular area of contrast material, similar in attenuation to intravenous contrast material seen within the aorta or an adjacent major artery (26), was seen surrounded by hematoma. At multi detector row CT, the area of active bleeding identified during the arterial phase may appear to expand when reexamined in the delayed renal excretory phase because of continuous bleeding. Posttraumatic pseudoaneurysms or arteriovenous fistulas appear on intravenous contrast material enhanced CT scans as a well-circumscribed area(s) of attenuation similar to that of an adjacent contrast-enhanced artery. These vascular injuries may be surrounded by organ parenchyma and/or hematoma. At delayed imaging during the renal excretory phase, vascular lesions typically lose attenuation from washout of intravenous contrast material and become isoattenuating to slightly hyperattenuating to the adjacent normal organ parenchyma at multi detector row CT (22). Patients with a wound track that extended to the diaphragm were considered to potentially have a diaphragm injury. CT findings of herniated abdominal content at the site of a diaphragmatic rent, CT collar sign (constriction of a herniated viscus at the site of diaphragmatic rent), and a single gunshot or stab wound and contiguous organ injury on either side of the diaphragm were considered positive for complete diaphragm tear. CT findings of potential diaphragm injury included a penetrating injury track that extended to the diaphragm, thickening of the diaphragm, and an isolated focal defect in the normal continuity of the diaphragm without adjacent hemorrhage. Laparotomy was defined as therapeutic if there was a visceral injury that required surgical repair, nontherapeutic if a visceral injury did not require surgical repair (eg, small mesenteric hematoma, nonbleeding hepatic or splenic injuries), and negative if no injury was found. Medical records were reviewed by three authors (K.S., W.C.C., G.J.F.H.) to determine surgical findings, length of hospital stay, and associated injuries in patients without peritoneal violation. CT diagnoses were compared with laparotomy findings and clinical course until discharge. Statistical Analysis Statistical analysis was performed to determine the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy with 95% CIs to quantify uncertainty in the CT depiction of peritoneal violation. RESULTS Site of Entry Wound and Peritoneal Violation Entry wounds were thoracoabdominal in 42% of patients (83 of 200), abdominal in 18% (36 of 200), in the flank in 15% (30 of 200), pelvic in 15% (30 of 200), and in the back in 16% (32 of 200). A second entry site was seen in the torso of 11 patients. CT findings were considered positive in 35% of patients (70 of 200). Among 70 patients with positive CT findings, peritoneal violation was seen in 97% (68 of 70) (Fig 1). CT findings were negative for peritoneal violation in 66% of patients (132 of 200) (Fig 2). Among the 68 patients with peritoneal violation, the entry site was thoracoabdominal in 37% (31 of 83), abdominal in 50% (18 of 36), in the flank in 27% (eight of 30), pelvic in 23% (seven of 30), and in the back in 16% (five of 32). The most common CT finding of peritoneal violation was intraperitoneal free fluid, which was seen in 85% of patients (58 of 68). Other CT findings of peritoneal violation included pneumoperitoneum, as seen in 35% of patients (24 of 68); intraperitoneal organ injury, as seen in 75% (51 of 68); and a wound track that extended into the peritoneal cavity or an intraperitoneal bullet fragment, which was seen in 65% (44 of 68). Among the 66% of patients (132 of 200) with CT findings negative for peri- Volume 231 Number 3 Penetrating Torso Trauma: Helical CT 777

4 toneal violation, CT findings were positive for other injuries in three, including injury to the retroperitoneal bowel in two and a large retroperitoneal hematoma and injury to the inferior vena cava (Fig 3) in the other. All three of these patients underwent laparotomy. Hematomas were seen in 14% of patients (27 of 200) (retroperitoneal, n 15; pelvic, n 9; paraspinal, n 3; abdominal wall, n 3), lower thoracic injuries were seen in 18% (36 of 200) (pneumothorax, n 24; hemothorax, n 22; lung contusion or laceration, n 11), and fractures were seen in 7% (13 of 200) (pelvis, n 6; spine, n 5; femur, n 2). Triple-contrast helical CT had a sensitivity of 97% (66 of 68 patients; 95% CI: 89.8%, 99.6%), specificity of 98% (130 of 132 patients; 95% CI: 94.6%, 99.8%), positive predictive value of 97% (66 of 68 patients), negative predictive value of 98% (130 of 132 patients), and accuracy of 98% (196 of 200 patients; 95% CI: 95.0%, 99.5%) for depiction of peritoneal violation. Review of medical records and surgical findings indicated that CT findings were false-positive for peritoneal violation in two patients and false-negative in two. Both patients with false-negative CT findings for peritoneal violation were stabbed in the left thoracoabdominal region and underwent therapeutic laparotomy. One patient had thickening of the left side of the diaphragm and a small left-sided hemothorax (Fig 4). This patient developed abdominal pain during observation, and repeat CT showed a large actively bleeding mesenteric hematoma in the upper left abdomen and injury to the left side of the diaphragm. At laparotomy, the mesenteric bleeding was controlled, and the injury to the left side of the diaphragm was repaired. The other patient s CT findings showed a stab wound that extended to the left side of the diaphragm with thickening of the diaphragm. At laparotomy of the left side of the diaphragm, the injury was repaired. A small amount of free intraperitoneal blood was present at surgery. Both patients with false-positive CT findings underwent laparotomy, which yielded negative findings. One patient underwent laparotomy for a diaphragm injury on the basis of CT findings. The other had CT evidence of mesenteric contusion and was suspected of having sigmoid colonic injury. At surgery, no peritoneal violation or colonic injury was seen. Bowel and Mesenteric Injury Bowel and mesenteric injuries were seen at CT performed at admission in 17% of patients (34 of 200), including bowel injury in 13% (26 of 200) and isolated mesenteric injuries in 4% (eight of 200). Among the patients with CT findings positive for peritoneal violation, bowel or mesenteric injuries were seen in 47% (32 of 68). The other two patients had only retroperitoneal violation at CT. Of the 26 patients with bowel injury, isolated bowel injury was seen in 6% (13 of 200), and bowel plus mesenteric injury was seen in 6% (13 of 200). The most sensitive CT finding of bowel injury was a wound track that clearly extended to the bowel (Fig 5), which was seen in 77% of patients (20 of 26). Other CT findings of bowel injury included subjective bowel wall thickening, which was seen in 42% of patients (11 of 26), and oral or rectal contrast material extravasation, which was seen in 19% of patients (five of 26) (Figs 1, 6). Laparotomy was performed in 88% of patients with bowel injury (23 of 26) and was based on CT findings in 78% (18 of 23), peritoneal violation with a large amount of free intraperitoneal air or free intraperitoneal fluid in 9% (two of 23), lack of improvement during observation for peritoneal violation in 9% (two of 23), and local wound exploration that showed evidence of peritoneal violation in 4% (one of 23). Laparotomy was performed in the two patients who failed to improve with observation because one developed hematemesis and the other developed peritonitis. All patients with oral or rectal contrast material extravastion (n 5) and bowel wall thickening and adjacent mesenteric contusion (n 13) required surgical intervention. CT findings were false-positive for bowel injury in four patients. Injury to the stomach was diagnosed at CT in three patients, and duodenal injury was diagnosed in one. All three patients with CT findings false-positive for gastric injury had a wound track that extended to the stomach, and two of these patients also had blood in the lesser sac. At surgery, the hemorrhage in the lesser sac arose from a splenic injury in one patient and from a pancreatic injury in the other. Another patient who had CT findings false-positive for duodenal injury had a metal shot pellet lodged between the second part of the duodenum and the pancreatic head at CT (Fig 7). No duodenal injury was found at surgery. Follow-up CT was performed in 12% of patients who were observed (three of 26). Figure 2. Transverse CT scan in a man from a mental institution who was admitted after impaling himself with a metallic wire. The wire (arrow) is superficial within the anterior abdominal wall, without peritoneal violation. Figure 3. Transverse CT scan in a 23-year-old man shot multiple times in the back shows major vascular injury in the retroperitoneum. Note retroperitoneal hematoma (curved white arrows) within and along the right psoas muscle. A defect (black arrow) is seen in the posterior wall of the inferior vena cava. Air bubbles are present within the spinal canal (arrowhead) and left psoas muscle (straight arrows) along bullet tracks. CT findings included focal bowel wall thickening in one and mesenteric contusion in two. In all three patients, the initial CT findings had resolved at follow-up CT. Isolated mesenteric injuries were seen in 4% of patients (eight of 200). Active mesenteric bleeding was seen in two patients, who both underwent therapeutic laparotomy. Among the six patients with infiltration or focal thickening of the mesentery at CT, laparotomy was performed in four, and mesenteric injury was confirmed in three. The other patient had negative findings at laparotomy. Diaphragm Injury CT findings indicated a potential diaphragm injury in 25% of patients (50 of 778 Radiology June 2004 Shanmuganathan et al

5 Figure 4. CT findings false-negative for peritoneal violation in a 25-year-old man stabbed in the left thoracoabdominal region. (a) Transverse CT scan obtained at admission shows an air bubble (white arrowhead) in the left thoracoabdominal region at the entry site. Localized thickening (arrows) is seen in the left side of the diaphragm adjacent to the wound site. A small hemothorax (black arrowheads) is seen in the lower chest. (b) Follow-up CT scan obtained because of abdominal pain during observation shows active bleeding (curved arrows) within a large mesenteric hematoma (white arrowheads) and free intraperitoneal blood (black arrowheads) around the liver. (c) Follow-up CT scan shows a subtle defect (arrowheads) in the left side of the diaphragm. (Adapted and reprinted, with permission, from reference 27.) 200), including right-sided diaphragm injury in 54% (27 of 50) and left-sided diaphragm injury in 46% (23 of 50). The most common CT finding was a wound track that extended up to the diaphragm in 88% of patients (44 of 50). Other nonspecific CT findings of diaphragm injury included (a) thickening of the hemidiaphragm from blood or edema, either alongside or within the diaphragm in 44% of patients (22 of 50) (Fig 4) or (b) an isolated defect in the normal continuity of the diaphragm without herniation or adjacent hematoma in 10% of patients (five of 50). Specific CT findings of diaphragm injury were seen in 40% of patients (20 of 50) and included contiguous organ injury on either side of the diaphragm in a patient with a single gunshot (Figs 4, 8) or stab wound in 90% of patients (18 of 20) and herniation of abdominal fat into the thoracic cavity through a diaphragmatic defect in 4% (two of 50) (Fig 6). Laparotomy (n 12) or thoracoscopy (n 5) were performed in 34% of patients (17 of 50) with CT findings suggestive of diaphragm injury. Surgical reports confirmed a diaphragm injury in 71% of patients (12 of 17). Surgical reports did not confirm a diaphragm injury in two patients with specific CT findings of diaphragm injury of contiguous organ injury on either side of the diaphragm in a patient with a single gunshot or stab wound and in three patients with nonspecific signs of diaphragm injury. Solid Organ Injury Figure 5. Transverse CT scan shows a wound track that extends to the bowel in a 33-year-old man shot in the right hemipelvis. Note bullet and bone fragments (black arrowheads) outlining the wound track that extends to a loop of small bowel. Free intraperitoneal air (white arrowheads) and fluid (arrows) are seen from peritoneal violation. An injury to the ileum was confirmed surgically. Solid organ injuries were seen in 19% of patients (38 of 200). The liver was the intraperitoneal organ injured most commonly in this study, as seen in 14% of patients (28 of 200). Triple-contrast CT showed injury to the spleen in 4% of patients (seven of 200) and to the kidney in 3% (six of 200). Isolated hepatic injuries were seen in 82% of patients with hepatic injury (23 of 28). Nonsurgical management of hepatic injury was attempted and was successful in 91% of patients with isolated hepatic injury (21 of 23). Hepatic arteriography and transcatheter embolization were performed in four patients on the basis of CT evidence of active bleeding in one patient and clinical evidence of ongoing hemorrhage in three patients. One patient in the nonsurgical repair group developed a biliary-pleural fistula (Fig 8) that required CT-guided aspiration of the pleural collection. Laparotomy was performed in two patients with isolated hepatic injury. In one patient, laparotomy was performed for hypotension soon after CT. Surgery followed by hepatic arteriography and embolization was needed to control hepatic hemorrhage. A right hemidiaphragm was also repaired at laparotomy in this patient. The second patient with isolated hepatic injury underwent laparotomy to repair a right hemidiaphragm injury. Among the seven patients with splenic injury, active splenic bleeding was seen at CT in three. Laparotomy was performed in six patients, and splenic arteriography and transcatheter embolization were performed in one (Fig 9). Splenic hemorrhage was seen at the time of surgery in five of the six patients who required therapeutic intervention. None of the six patients with renal injury had injury to the renal collecting system or required surgical intervention. Outcome Laparotomy was performed in 20% of patients (41 of 200). In this group, 93% of patients (38 of 41) had CT evidence of intraperitoneal injury, and 5% (two of 41) had only evidence of retroperitoneal injury. Laparotomy was based on CT findings in 93% of patients (38 of 41), local wound exploration in 2% (one of 41), development of peritonitis during observation in 2% (one of 41), and hematoma seen at sigmoidoscopy but yielding negative CT findings in 2% (one of 40). Laparotomy based on CT findings was considered therapeutic in 87% of patients (33 of 38) and nontherapeutic in Volume 231 Number 3 Penetrating Torso Trauma: Helical CT 779

6 8% (three of 38) and yielded negative findings in 5% (two of 38). Among the three patients who underwent nontherapeutic laparotomy on the basis of CT findings, CT showed infiltration or focal thickening of the mesentery in two. In the other patient, CT showed a shotgun pellet lodged between the head of the pancreas and second part of the duodenum (Fig 7). In the two patients who had negative laparotomy results, CT showed evidence of a mesenteric contusion, and findings were suggestive of a sigmoid colonic injury following a gunshot wound to the pelvis. The laparotomies based on local wound exploration and sigmoidoscopy were nontherapeutic and yielded negative results, respectively. Among the 66% of patients (132 of 200) without evidence of peritoneal violation but with minor injuries (soft-tissue lacerations; small abdominal-wall, pelvic, or retroperitoneal hematomas) at CT, 58% (76 of 132) were discharged from the hospital within 24 hours of admission. The average length of hospital stay was 5.6 days in 42% of patients (56 of 132) who needed treatment for associated thoracic injuries in 57% (32 of 56), orthopedic injuries in 23% (13 of 56), head or spinal injuries in 9% (five of 56), and other surgical interventions in 11% (six of 56). DISCUSSION The surgical practice of mandatory laparotomy for the diagnosis and management of all penetrating torso injuries, which was in effect prior to the 1960s, was an extension of policy developed during the two World Wars (18 20,28,29). This practice was based on the assumption that these wounds were associated with a high incidence of visceral injury that required surgical intervention and that delay in laparotomy resulted in unacceptably high morbidity and mortality. Further, it was assumed that laparotomy added no substantial increase to length of hospitalization. Since the 1960s, selective conservative management has become popular in many U.S. trauma centers and has been shown to be successful without increased morbidity or mortality (18 20,28,29). This management approach was used initially in hemodynamically stable patients with stab wounds to the torso without overt signs of peritonitis (18 20,29 32). Currently, in most trauma centers, the most popular approaches in the diagnosis of abdominal injury in stable patients include serial observation and local wound Figure 6. Transverse CT scans show gastrointestinal contrast material extravasation in a 29- year-old man shot in the left thoracoabdominal region. (a) Scan shows focal thickening of the splenic flexure of the colon (curved black arrow) and rectally administered contrast material (black arrowhead) within the chest wall. Free intraperitoneal air (white arrowhead) is seen from peritoneal violation. Peritoneal fat (curved white arrow) herniates through a defect (straight arrows) in the left side of the diaphragm. (b) Scan obtained in the middle of the abdomen shows rectally administered contrast material extravasation (curved arrow) from the descending colon (straight arrow) and free intraperitoneal air (arrowhead). Colon and diaphragm injuries were repaired surgically. exploration, followed by diagnostic peritoneal lavage. Careful serial abdominal examinations ideally performed by the same experienced surgeon can be used to treat 34% 50% of patients with penetrating abdominal injury and can lead to a delayed laparotomy rate of 3% 12% (31,33). Unnecessary laparotomy is avoidable in 34% 47% of these patients. A limitation of this approach is the demand placed on medical staff in a busy urban trauma center and extended duration of hospitalization. To expedite management and minimize missed injuries, local wound exploration followed by diagnostic peritoneal lavage was introduced (31,34). When local wound exploration reveals no evidence of peritoneal violation, the wound is closed. Patients with indeterminate or documented peritoneal penetration by means of local wound exploration undergo diagnostic peritoneal tap or lavage. A major controversy regarding the use of diagnostic peritoneal lavage in penetrating torso trauma is the red blood cell count that constitutes positive findings. The proposed values for a clinically important red blood cell count in diagnostic peritoneal lavage fluid that indicates a need for exploration range from 1,000 to 100,000 red blood cells per cubic milliliter (17,34 36). In a prospective study Figure 7. Transverse CT scan shows findings false-positive for duodenal injury in a 21-yearold man shot in the right flank. Note the pellet (arrowhead) between the pancreatic head (curved arrow) and the second part of the duodenum (straight arrow). At surgery, no bowel injury was found. performed to determine the role of red blood cells in diagnostic peritoneal lavage performed for abdominal stab wounds, Muckart and McDonald (34) reported that the standard red blood cell count in blunt trauma (100,000 red blood cells per cubic milliliter) is only 780 Radiology June 2004 Shanmuganathan et al

7 69% sensitive and results in an unacceptably high 31% missed injury rate. A low red blood cell count threshold (5,000 red blood cells per cubic milliliter) would be more sensitive for injury but would invariably increase the nontherapeutic or negative laparotomy rate to 26%. For the past decade, triple-contrast CT has been used for triage of patients with penetrating trauma to the flank and back (8 11). Use of CT in this limited context has shown high sensitivity (89% 100%) and accuracy (97% 98%) while maintaining a low nontherapeutic laparotomy rate of 3.2% (8 11). To our knowledge, the present prospective study of 200 patients with penetrating torso trauma examined with helical or multi detector row CT to determine both peritoneal violation and extent of organ injury is the largest reported to date. In our study, Figure 8. Transverse CT scans show right-sided diaphragm injury and biliary-pleural fistula in a 16-year-old adolescent shot in the right side of the back. (a) CT scan obtained at admission shows a posterior right-lobe liver laceration (arrow) and air bubbles (white arrowheads) within the liver. A bullet track (black arrowheads) is seen in the back. (b) CT scan obtained in the lower chest shows extensive lung contusion (arrows) along the bullet track, indicating contiguous organ injury on either side of the diaphragm. Anteroposterior radiograph (not shown) obtained 5 days after admission showed a right lung cavity with an air-fluid level. (c) CT scan obtained after chest radiography shows a cavity (straight arrow) with an air-fluid level (curved arrow) in the posterior right lower lobe. Bile was aspirated from this cavity with CT guidance. 34% of patients (68 of 200) who had no clinical or radiographic findings to suggest peritoneal perforation in fact had peritoneal violation diagnosed at CT. These patients are at high risk for intraperitoneal injury, and laparotomy was performed in 57% (39 of 68). On the basis of CT results, however, 43% of patients (29 of 68) with peritoneal violation who had only isolated hepatic injury, a small quantity of intraperitoneal free fluid or air, or mesenteric contusions were treated successfully without surgery. Thus, triple-contrast helical CT was helpful in facilitating nonsurgical management, even in patients with peritoneal violation. A high sensitivity of 97% (66 of 68 patients) and negative predictive value of 98% (130 of 132 patients) for the detection of peritoneal violation indicate that triple-contrast helical CT is very useful in the work-up of stable patients with penetrating torso trauma. A high accuracy of 98% (196 of 200 patients) and specificity of 98% (130 of 132 patients) also indicates that contrast-enhanced helical CT is reliable in the demonstration of intra- and extraperitoneal visceral injury. Among 66% of patients (132 of 200) without CT findings of peritoneal violation, 58% (76 of 132) were discharged within 24 hours. To our knowledge, no patient in this group had an injury that was missed and later required treatment. The 42% of patients (56 of 132) without peritoneal violation had other injuries that required admission for further treatment. Information derived from triplecontrast helical CT clearly facilitates appropriate use of medical resources. Triple-contrast multi detector row CT is currently in the routine imaging algorithm for stable patients with penetrating trauma in our trauma center, as shown in Figure 10. Multi detector row CT, followed by diagnostic peritoneal lavage, have substantially replaced both observation and local wound exploration in the assessment of stable patients with penetrating torso trauma without clinical signs of peritonitis or pneumoperitoneum seen radiographically. A high incidence of hollow visceral injury is commonly seen in patients with entry sites to the thoracoabdominal, abdominal, and pelvic regions (16 19). Some of these bowel injuries may be clinically occult at admission because of minimal bleeding or peritoneal contamination. The varying sensitivity, specificity, and accuracy of CT in the diagnosis of bowel and mesenteric injuries, as reported in the literature (13 15), has discouraged many trauma surgeons from routinely using CT to evaluate patients with penetrating torso trauma. Many surgeons were concerned that CT findings would not show intraperitoneal bowel injury and continued to rely on local wound exploration or diagnostic peritoneal lavage for diagnosis of these injuries (8,9,11). In our study, bowel and mesenteric injuries were diagnosed in 50% of patients (34 of 68) with positive CT findings. Bowel injury was intraperitoneal in 24 patients and retroperitoneal in two. The most sensitive CT finding of bowel injury was a wound track that extended up to the injured bowel, which was seen in 77% of patients (20 of 26). It is important to know the precise wound entry site(s) and number of wounds and to em- Volume 231 Number 3 Penetrating Torso Trauma: Helical CT 781

8 ploy optimal CT window and level settings (window width, 550 HU; level, 75 HU) to accurately identify wound tracks (22). The most specific CT findings of bowel injury include gastrointestinal contrast material extravasation and bowel wall thickening with adjacent mesenteric injury, which were seen in 19% of patients (five of 26) and 42% (11 of 26), respectively. All patients with both of these CT findings had bowel injuries that required surgical intervention, and all underwent therapeutic laparotomy. In the current study, the authors considered a CT finding of free intraperitoneal air or free intraperitoneal fluid to be evidence of peritoneal violation and not a direct sign of bowel injury. Pneumoperitoneum may be seen after penetrating injury because air is introduced into the peritoneal space by the bullet or knife at the time of peritoneal violation. Air may also subsequently enter the peritoneal cavity through a peritoneal defect that communicates with the exterior (37). Similarly, isolated free intraperitoneal fluid may be seen within the peritoneum as a result of bleeding from the injured peritoneal lining or from extraperitoneal bleeding that enters the peritoneal cavity along a knife or bullet track. This CT finding alone should not be considered diagnostic of bowel injury. It is important to routinely use both oral and rectal contrast material to opacify bowel tissue to improve the ability to identify bowel wall thickening, small mesenteric hematomas, and gastrointestinal contrast material extravasation (8,22). Gastrointestinal contrast material extravasation was seen in 19% of bowel injuries (five of 26) and was one of the most specific signs of bowel injury that indicated the need for surgery. Opacification of bowel was also subjectively helpful in the present study in the identification of focal bowel wall thickening and small mesenteric hematomas and likely reduced the number of false-negative CT findings of bowel or mesenteric injury. The role of follow-up CT in penetrating trauma has not been determined. In our study, only three patients with initial CT findings positive for peritoneal violation but no clinical evidence of peritonitis underwent follow-up CT. Follow-up CT performed 4 6 hours after initial CT showed resolution of the initial CT findings, including bowel wall thickening in one patient and mesenteric infiltration in the other two. These follow-up CT results increased confidence that the nonspecific CTfindings at initial CT were lesions that would not be repaired surgically. Figure 9. Active bleeding spleen in a 33-yearold man stabbed in the left thoracoabdominal region. (a) CT scans obtained in the portal venous phase show active bleeding (arrowheads) from a splenic laceration (straight arrow). Perisplenic blood (curved arrows) is also seen. (b) Delayed CT scan obtained in the same region shows an increase in the amount of active bleeding (curved arrows) seen within the perisplenic hematoma (straight arrows). (c) Splenic arteriogram confirms bleeding (arrow) from a splenic artery branch. Hemorrhage was stopped by using transcatheter embolization (not shown). Given the high sensitivity and specificity of CT for the detection of peritoneal violation and organ injury reported in this study, our trauma surgeons are currently more enthusiastic to obtain follow-up CT findings if results of initial CT are nonspecific or equivocal for intraperitoneal injury that might require exploration. Penetrating injury to the thoracoabdominal region often results in injury to organs in the thoracic and abdominal cavity (38). Investigators in prior studies (22,39,40) reported that CT may help stable patients to be selected for nonsurgical treatment. The liver was the intraperitoneal organ injured most commonly in our study, as seen in 14% of patients (28 of 200). Ninety-one percent of patients (21 of 23) with isolated hepatic injury were treated without surgery, with the adjunct of hepatic arteriography and embolization in three patients. CT fostered the use of nonsurgical treatment of isolated penetrating liver trauma by allowing exclusion of other intraperitoneal injuries that would require surgery. To our knowledge, this is the largest group of patients with penetrating trauma and isolated hepatic injury treated successfully by using CT prospectively for diagnosis and treatment planning. The incidence of diaphragm injury varies from 24% to 42% of patients with penetrating injury to the thoracoabdominal region or lower chest (40,41). Isolated injuries to the diaphragm are usually asymptomatic at admission and often lack specific findings on chest radiographs (22,40,41). It is well recognized, however, that small diaphragmatic hernias can lead to fatal complications remote from the initial injury (42,43). In our study, specificctfindings of diaphragm injury were seen in 40% of patients (20 of 50), and nonspecific CT findings were seen in 88% (44 of 50). The most common specific CT finding of diaphragm injury was contiguous organ in- 782 Radiology June 2004 Shanmuganathan et al

9 Figure 10. Schematic shows imaging algorithm for hemodynamically stable patients with penetrating torso trauma. jury seen on either side of the diaphragm in patients with single penetrating wounds, which occurred in 36% of patients (18 of 50). The most common nonspecific CTfinding of a diaphragm injury was a wound track that extended to the diaphragm. A major limitation of our study was lack of a standard test to confirm that there was no peritoneal violation in patients with negative CT findings. These patients were only evaluated clinically for evidence of a missed injury. Another limitation was that only 34% of patients (17 of 50) with CT findings of potential diaphragm injury underwent optimal work-up to confirm or exclude this injury. To overcome this limitation, the authors are currently performing a prospective study to better correlate CT findings of diaphragm injury with laparotomy or thoracoscopic results. In conclusion, triple-contrast multi detector row helical CT is highly accurate in the diagnosis and exclusion of peritoneal violation and visceral injury in hemodynamically stable patients with penetrating torso trauma. Further studies are necessary to determine the accuracy of CT in the diagnosis of specific sites of injury, such as the diaphragm. References 1. Federle MP, Goldberg HI, Kaiser JA, et al. Evaluation of trauma by computed tomography. Radiology 1981; 138: Federle MP. Computed tomography in blunt abdominal trauma. Radiol Clin North Am 1983; 21: Mirvis SE, Rodriguez A. Abdominal and pelvic trauma. In: Mirvis SE, Young JWR, eds. Imaging in trauma and critical care. Baltimore, Md: Williams & Wilkins, 1992; Fabian TC, Mangiante EC, White TJ, et al. A prospective study of 91 patients undergoing computed tomography and peritoneal lavage following blunt abdominal trauma. J Trauma 1986; 26: Mirvis SE, Shanmuganathan K. Abdominal computed tomography in blunt abdominal trauma. Semin Roentgenol 1992; 27: Jhirad R, Boone D. Computed tomography for evaluation of blunt abdominal trauma in the low-volume nondesignated trauma center: the procedure of choice? J Trauma 1998; 45: Livingston DH, Lavery RF, Passannante MR, et al. Admission or observation is not necessary after a negative abdominal computed tomographic scan in patients with suspected blunt abdominal trauma: results of a prospective, multi-institutional trial. J Trauma 1998; 44: Phillips T, Sclafani SJ, Goldstein A, Scalea T, Panetta T, Shaftan G. Use of contrastenhanced CT enema in the management of penetrating injuries trauma to the flank and back. J Trauma 1986; 26: Kirton OC, Wint D, Thrasher B, et al. Stab wounds to the back and flank in hemodynamically stable patients: a decision algorithm based on contrast-enhanced computed tomography with colonic opacification. Am J Surg 1997; 173: Maldjian PD, Zurlo JV, Sebastiano L. Role of abdominal computed tomography in the evaluation and management of stab wounds to the back and flank. Emerg Radiol 1997; 4: Meyer DM, Thal E, Weigelt JA, et al. Role of abdominal CT in the evaluation of stab wounds to the back. J Trauma 1989; 29: Boyle EM, Maier RV, Salazar JD, et al. Diagnosis of injuries after stab wounds to back and flank. J Trauma 1997; 42: Marx JA, Moore EE, Jorden RC, et al. Limitations of computed tomography in the evaluation of acute abdominal trauma: a prospective comparison with diagnostic peritoneal lavage. J Trauma 1985; 25: Sherck JP, Oakes DD. Intestinal injuries missed by computed tomography. J Trauma 1990; 30: Butela ST, Federle MP, Chang PJ, et al. Performance of CT in detection of bowel injury. AJR Am J Roentgenol 2001; 176: Fabian TC, Croce MA, Stewart RM, et al. A prospective analysis of diagnostic laparoscopy in trauma. Ann Surg 1993; 217: Feliciano DV, Rozycki GS. The management of penetrating abdominal trauma. Adv Surg 1995; 28: Nance FC, Cohn I Jr. Surgical judgment in the management of stab wounds of the abdomen: a retrospective and prospective analysis based on a study of 600-stabbed patients. Ann Surg 1969; 170: Henneman PL. Penetrating abdominal trauma. Emerg Med Clin North Am 1989; 7: Grossman MD, May AK, Schwab CW, et al. Determine anatomic injury with computed tomography in selected torso gunshot wounds. J Trauma 1998; 45: Ginzburg E, Carrillo EH, Kopelman T, et al. The role of computed tomography in selective management of gunshot wounds to the abdomen and flank. J Trauma 1998; 45: Shanmuganathan K, Mirvis SE, Chiu W, et al. Value of triple-contrast spiral CT in penetrating torso trauma: a prospective study to determine peritoneal violation and the need for laparotomy. AJR Am J Roentgenol 2001; 177: Chiu WC, Shanmuganathan K, Mirvis SE, et al. Determining the need for laparotomy in penetrating torso trauma: a prospective study using triple-contrast enhanced abdominopelvic computed tomography. J Trauma 2001; 51: Rizzo MJ, Federle MP, Griffiths BG. Bowel Volume 231 Number 3 Penetrating Torso Trauma: Helical CT 783

10 and mesenteric injury following blunt abdominal trauma: evaluation with CT. Radiology 1989; 173: Dowe MF, Shanmuganathan K, Mirvis SE, et al. CT findings of mesenteric injury after blunt trauma: implications for surgical intervention. AJR Am J Roentgenol 1997; 168: Shanmuganathan K, Mirvis SE, Sover ER. Value of contrast-enhanced CT in detecting active hemorrhage in patients with blunt abdominal or pelvic trauma. AJR Am J Roentgenol 1993; 161: Shanmuganthan K, Scalea TM. Imaging of penetrating trauma of the torso and chest. In: Mirvis SE, Shanmuganathan K, eds. Imaging in trauma and critical care. 2nd ed. Philadelphia, Pa: Saunders, 2003; Nance ML, Nance FC. It is time we told the emperor about his clothes (editorial). J Trauma 1996; 40: Moore EE, Moore JB, Van Duzer-Moore S, et al. Mandatory laparotomy for gunshot wounds penetrating the abdomen. Am J Surg 1980; 140: Shaftan GW. Indications for operation in abdominal trauma. Am J Surg 1960; 99: Feliciano DV, Bitondo CG, Steed G, Mattox KL, Burch JM, Jordan GL Jr. Five hundred open taps or lavages in patients with abdominal stab wounds. Am J Surg 1984; 148: Demetriades D, Rabinowitz B. Indications for operation in abdominal stab wounds: a prospective study of 651 patients. Ann Surg 1987; 205: Demetriades D, Velmahos G, Cornwell E III, et al. Selective nonoperative management of gunshot wounds of the anterior abdomen. Arch Surg 1997; 132: Muckart DJ, McDonald MA. Unreliability of standard quantitative criteria in diagnostic peritoneal lavage performed for suspected penetrating abdominal stab wounds. Am J Surg 1991; 162: Oreskovich MR, Carrico JC. Stab wound to anterior abdomen: analysis of a management plan using local wound exploration and quantitative peritoneal lavage. Ann Surg 1983; 198: Thompson JS, Moore EE, Van Duzer- Moore S, et al. The evolution of abdominal stab wound management. J Trauma 1980; 20: Killeen KL, Shanmuganathan K, Poletti PA, et al. Helical computed tomography of bowel and mesenteric injuries. J Trauma 2001; 51: Asensio JA, Arroyo H Jr, Veloz W, et al. Penetrating thoracoabdominal injuries: ongoing dilemma which cavity and when? World J Surg 2002; 26: Demetriades D, Gomez H, Chahwan S, et al. Gunshot injuries to the liver: the role of selective nonoperative management. J Am Coll Surg 1999; 188: Renz BM, Feliciano DV. Gunshot wounds to the right thoracoabdomen: a prospective study of nonoperative management. J Trauma 1994; 37: Freeman RK, Al-Dossari G, Hutchenso KA, et al. Indications for using video-assisted thoracoscopic surgery to diagnose diaphragmatic injuries after penetrating chest trauma. Ann Thorac Surg 2001; 72: Murray JA, Demeteriades D, Asensio JA, et al. Occult injuries to the diaphragm: prospective evaluation of laparoscopy in penetrating injuries to the left lower chest. J Am Coll Surg 1998; 187: Murray JA, Berne J, Asensio JA. Penetrating thoracoabdominal trauma. Emerg Med Clin North Am 1998; 16: Radiology June 2004 Shanmuganathan et al