CT Findings of Sigmoid Volvulus

Transcription

1 Gastrointestinal Imaging Original Research Levsky et al. CT of Sigmoid Volvulus Gastrointestinal Imaging Original Research Jeffrey M. Levsky 1 Elana I. Den Ronelle A. DuBrow Ellen L. Wolf Alla M. Rozenblit Levsky JM, Den EI, DuBrow RA, Wolf EL, Rozenblit AM Keywords: CT, sigmoid volvulus DOI: /AJR Received February 12, 2009; accepted after revision July 10, All authors: Department of Radiology, Montefiore Medical Center, Albert Einstein College of Medicine, 111 E 210th St., Bronx, NY Address correspondence to J. M. Levsky (jlevsky@montefiore.org). AJR 2010; 194: X/10/ American Roentgen Ray Society CT Findings of Sigmoid Volvulus OBJECTIVE. The purpose of this study was to evaluate the features of sigmoid volvulus on CT scanograms and cross-sectional images. MATERIALS AND METHODS. We retrospectively reviewed 21 cases of sigmoid volvulus in 15 men and six women. Three radiologists evaluated scanograms and cross-sectional images for several classic and two novel imaging signs of volvulus: crossing sigmoid transitions (called the X-marks-the-spot sign) and folding of the sigmoid wall by partial twisting (called the splitwall sign). A general impression was assigned to scanograms and cross-sectional images. CT findings suggesting bowel compromise were compared with pathologic and endoscopic findings. RESULTS. The most sensitive scanogram findings were absence of rectal gas (19 of 21 cases, 90%) and an inverted-u-shaped distended sigmoid (18 of 21 cases, 86%) followed by the coffee bean sign and disproportionate sigmoid enlargement (both 16 of 21 cases, 76%). The most sensitive cross-sectional findings were one sigmoid colon transition point (20 of 21 cases, 95%) and disproportionate enlargement of the sigmoid (18 of 21 cases, 86%). The X- marks-the-spot and split-wall signs were present in nine of 21 (43%) and 11 of 21 (52%) patients, but one of the two signs was present in 18 of 21 patients (86%). Classic radiographic and definitive cross-sectional findings were seen in 11 of 21 (52%) and 16 of 21 (76%) patients. CT findings were definitive in five of seven patients (71%) with indeterminate scanogram findings. Imaging signs suggesting bowel compromise correlated poorly with clinical ischemia, but CT features were present in all three patients with frank necrosis. CONCLUSION. Sigmoid volvulus has a spectrum of imaging findings. A classic appearance is absent on approximately one half of scanograms and one fourth of CT scans. Use of new signs that model the pathophysiologic characteristics of volvulus (X-marks-the-spot sign for more complete twisting and split-wall sign for less severe twisting) may improve diagnostic confidence. S igmoid volvulus is abnormal twisting of the sigmoid colon along its mesenteric axis that causes closed-loop obstruction and can lead to ischemia, perforation, and death. Sigmoid volvulus is thought to account for 2 5% of cases of large-bowel obstruction in the United States, which places it third in prevalence after cancer and diverticulitis [1]. The classic clinical presentation is a triad of abdominal pain, distention, and constipation [1, 2]. The nonspecific nature of these symptoms coupled with the relatively high prevalence of concomitant neuropsychiatric disease [2] make clinical diagnosis difficult and therefore present the opportunity for imaging to play a central role. Sigmoid volvulus has been described in the radiology literature in case series dating back several decades [3]. Radiographic findings include marked distention of an ahaustral sigmoid colon that assumes a coffee bean like configuration. The loop, which is classically described as pointing to the right upper quadrant, also can rise to the level of the left hemidiaphragm and overlap the liver shadow [3 6]. A more recently described [7] specific feature of volvulus is cephalic extension of the sigmoid in relation to the transverse colon (the northern exposure sign). Proximal colonic dilatation is present in approximately one half of cases [7], which prompts alternative diagnostic considerations, including simple distal colonic obstruction, colonic pseudoobstruction, and colonic ileus. Studies have shown that a substantial percentage of cases of sigmoid volvulus have an atypical radiographic appearance [4, 5]. In cases in 136 AJR:194, January 2010

2 CT of Sigmoid Volvulus which the radiographic findings are unclear, a contrast enema is considered a highly sensitive alternative [5, 6]. CT is being increasingly used for evaluation of undiagnosed abdominal symptoms. Although detection of sigmoid volvulus by CT has been described in case reports [8], the largest series of patients found was two [9]. The purpose of our study was to assess the sensitivity of CT features in the diagnosis of sigmoid volvulus and to investigate the utility of novel CT signs of volvulus. Materials and Methods Patient Sample This study was approved by our institutional review board with a waiver of informed consent. We retrospectively identified consecutive patients given the diagnosis code volvulus of the intestine (International Classification of Diseases, 9th edition, 560.2) at our institution from January 2004 through June Records of 159 of 161 clinical encounters were obtained. From these, we selected 22 patients who had a discharge diagnosis of volvulus and had one of the following: endoscopy report describing or mentioning sigmoid volvulus, operative report describing or mentioning sigmoid volvulus, or pathology specimen described as sigmoid volvulus. The 21 of these patients (15 men, six women) who underwent CT before an interventional procedure comprised our study population. The median interval between CT and the interventional procedure was less than 24 hours (range, 2 hours 6 days). Clinical Data Medical charts were reviewed for clinical characteristics that included demographic features, reason for initial evaluation, hospital course, imaging studies performed, comorbid conditions, and clinical outcome. Pathologic, surgical, and endoscopic reports were reviewed for the following descriptions of bowel compromise: ischemia, hemorrhage, ulcer, necrosis, and perforation. The sole finding of bowel congestion at pathologic examination was considered negative for bowel compromise. Imaging Techniques CT was performed with a 64-MDCT scanner (LightSpeed VCT, GE Healthcare) in nine cases, a 4-MDCT scanner (LightSpeed Plus, GE Healthcare) in eight cases, and a single-detector scanner (HiSpeed CTi, GE Healthcare) in four cases. Detector collimation varied between and 2.5 mm for MDCT equipment and was 5 mm on the singledetector scanner. Twenty of 21 studies were reconstructed at 5-mm slice thickness, and one study was reconstructed at 2.5-mm slice thickness. In nine of 21 cases, coronal reconstructions with 3-mm slice thickness with 1.5-mm overlap were made at the time of clinical imaging, and these images were reviewed. Seventeen of 21 studies were performed with iohexol (Omnipaque 300, GE Healthcare), iopromide (Ultravist 300, Bayer HealthCare), or iodixanol (Visipaque 320, GE Healthcare) administered IV with a power injector at 2 3 ml/s. In 16 cases, images were obtained in the portal venous phase with an injection-to-scan delay of seconds. In one case, acquisition was in the arterial phase with a 30-second delay. Nineteen of 21 studies were performed with 3% diatrizoate meglumine (Gastrografin, Bracco Diagnostics) as a positive oral contrast agent. Rectal contrast administration was not used for imaging of any patient in this study. Image Analysis Three attending radiologists with considerable experience (30, 29, and 24 years) and subspecialty training in abdominal or gastrointestinal radiology independently reviewed the original images. The reviewers were blinded to details of clinical management, outcome, and findings of subsequent imaging studies but knew that each patient had confirmed sigmoid volvulus. Because only eight of 21 patients had undergone abdominal radiography before intervention, we used the scanogram in lieu of the radiograph to evaluate the radiographic findings. Because only two of the 21 patients underwent a barium enema before intervention, correlations with CT findings were not analyzed. Scanograms were analyzed first and followed by axial cross-sectional images and coronal crosssectional images (when available). Disagreements between reviewers were resolved through consensus assessment by all three readers. Each scanogram was examined for 1, disproportionate enlargement of the sigmoid compared with the rest of the bowel; 2, distended sigmoid with an inverted U shape; 3, coalescence of the medial walls of the ascending and descending sigmoid loops forming a coffee bean appearance; 4, location of the apex of the distended loop (right, left, midline); 5, liver overlap sign; 6, extension of the sigmoid to the left hemidiaphragm; 7, cephalad position of the sigmoid colon relative to the transverse colon (northern exposure sign [7]); 8, proximal colonic dilatation; and 9, absence of rectal gas. Three scanograms showed no obvious distended sigmoid loop and were excluded from analysis of the loop apex (sign 4), and signs 5 7 were automatically scored as absent. The overall impression of sigmoid volvulus was graded on the basis of the reviewers gestalt perceptions. This analysis was based solely on personal experience rather than scoring one or more of the nine imaging features because these findings have not been validated previously. The following three grades were used: indeterminate, no radiographic suggestion of sigmoid volvulus; probable, suggestive of sigmoid volvulus; and classic, appearance typical of sigmoid volvulus and no serious consideration of alternative diagnosis. Cross-sectional images were examined for 1, disproportionate enlargement of the sigmoid compared with the rest of the bowel; 2, degree of sigmoid dilatation (mild, moderate, or severe); 3, proximal colonic dilatation; 4, rectal decompression; 5, the whirl sign [8]; 6, number of transition points in the sigmoid colon; 7, two crossing sigmoid transition points projecting from a single location (X-marksthe-spot sign) (Fig. 1); and 8, apparent separation of the sigmoid walls by adjacent mesenteric fat secondary to incomplete twisting or folding (split-wall sign) (Fig. 2). Overall impression of sigmoid volvu- TABLE 1: Clinical Characteristics of Study Sample (n = 21) Sex Characteristic Value Men 15 (71) Women 6 (29) Age (y) Average ± SD 74 ± 15 Range Reason for initial evaluation Abdominal pain 13 (62) Abdominal distention 11 (52) Constipation 10 (43) Comorbid conditions Hypertension 15 (71) Neuropsychiatric disease 13 (62) Diabetes mellitus 8 (38) Previous abdominal surgery 5 (24) Cancer 4 (19) Diagnosis Endoscopy 16 (76) Surgery 11 (52) Pathologic corroboration 9 (43) Hospital course Radiography before intervention 8 (38) Barium enema before intervention 2 (10) Endoscopy 18 (86) Surgery 13 (62) Death 3 (14) Note Except for age, values are numbers of patients with percentages in parentheses. AJR:194, January

3 Levsky et al. A Fig year-old woman with sigmoid volvulus (patient 18). A, Drawing shows X-marks-the-spot sign caused by crossing transition points. Horizontal lines and schematic axial insets indicate levels of B and C. Oblique lines in insets indicate two transition zones oriented in opposite directions, producing an X shape. Courtesy of Jordan Winick. B, Contrast-enhanced axial CT image through proximal superior transition (arrows). C, Axial CT image 15 mm caudad to B through distal inferior sigmoid transition (arrowheads), which is oriented in the direction opposite to the proximal transition. Fig year-old man with sigmoid volvulus (patient 20). A, Drawing shows split-wall sign caused by apparent separation of sigmoid walls by adjacent mesenteric fat (arrow) secondary to incomplete twisting. Horizontal lines and schematic axial insets indicate levels of B D. Courtesy of Jordan Winick. B, Contrast-enhanced axial CT image through the single beak-shaped transition point (arrow). C, Axial CT image 25 mm caudal to B shows mesenteric fat (arrow) separating the two sigmoid walls (arrowheads). D, Axial CT image 25 mm caudad to C shows the decompressed rectosigmoid (arrow). lus was graded on the basis of the reviewers gestalt perceptions. The following four grades were used: not evident, no evidence of sigmoid volvulus; possible, evidence that might represent sigmoid volvulus; probable, evidence most suggestive of sigmoid volvulus; and definite, appearance typical of sigmoid volvulus and no serious consideration of alternative diagnosis. The following findings were noted as evidence of bowel wall ischemia or compromise: sigmoid wall thickening, surrounding stranding, pneumatosis intestinalis, free intraperitoneal fluid, and free intraperitoneal gas. Statistical Analysis The sensitivity of each sign was calculated as the proportion of patients with the sign divided by the total number of patients, and the 95% CI was determined with a modified Wald method [10]. Kappa statistics were computed to assess interobserver variation [11]. An odds ratio was computed to assess B A C the association between the X-marks-the-spot and split-wall signs [12]. Subgroup analyses were performed for patients with classic and nonclassic scanogram findings. Correlations were calculated for disproportionate enlargement of sigmoid, proximal colonic dilatation, and rectal decompression determined by scanogram and CT. An odds ratio was computed to evaluate the association of CT findings with clinically proven bowel ischemia [12]. Imaging findings and mortality were correlated. C B D 138 AJR:194, January 2010

4 CT of Sigmoid Volvulus Results The characteristics of the study sample are shown in Table 1. A clinical triad of constipation, distention, and abdominal pain was present in the small minority of patients (10%), and most of the patients (62%) had neuropsychiatric disease. Only eight of 21 patients (38%) underwent radiography before CT. The scanogram signs of sigmoid volvulus and the overall radiologic impressions are shown in Table 2. Only 11 of 21 cases (52%) were assigned a classic radiographic impression (Fig. 3); three (14%) were probable (Fig. 4); and seven (33%) were indeterminate (Fig. 5). The most sensitive radiographic signs of volvulus were absence of rectal gas (19/21, 90%) and distended sigmoid loop with an inverted-u appearance (18/21, 86%) followed by the coffee bean sign and disproportionate sigmoid enlargement (both 16/21,76%). The northern exposure sign was present in 12 cases (57%). No patients were found to have sigmoid colon overlapping the left hemidiaphragm. The cross-sectional imaging signs of sigmoid volvulus and bowel wall compromise are shown in Table 3. A definite impression of sigmoid volvulus was assigned in 16 of 21 cases (76%). The most sensitive findings of volvulus were presence of at least one sigmoid colon transition point (20/21, 95%) and disproportionate enlargement of the sigmoid (18/21, 86%). The whirl sign was present in 12 of 21 cases (57%). We identified the X- TABLE 2: Scanogram Findings in Patients With Confirmed Sigmoid Volvulus (n = 21) Patient No. Disproportionate Sigmoid Dilatation Distended, Inverted-U Sigmoid Coffee Bean Sign Loop Apex Liver Overlap Northern Exposure Sign marks-the-spot sign (Fig. 1) in only nine patients (43%) and the split-wall sign (Fig. 2) in only 11 (52%). One of these two signs was present in 18 patients (86%), and both signs were present in two patients (10%). The two signs were statistically less likely to be present together (odds ratio, 0.095; Wald 95% CI, ). Findings evaluated on both scanograms and cross-sectional images were correlated. Agreement in disproportionate enlargement of the sigmoid, proximal bowel dilatation, and rectal decompression was found in 19, 17, and 10 of the 21 cases. In all 16 cases in which the scanogram depicted disproportionate sigmoid enlargement and in the seven cases in which it depicted proximal colonic dilatation, cross- Proximal Colonic Dilatation Rectal Gas 1 Present Present Present Right Present Present Absent Absent Probable 2 Present Present Present Left Absent Absent Present Absent Classic Overall Impression of Volvulus 3 Present Present Present Right Present Present Present Absent Indeterminate 4 Present Present Present Right Absent Absent Absent Absent Classic 5 Present Present Present Right Absent Absent Absent Absent Probable 6 Absent Present Absent Midline Absent Absent Present Absent Indeterminate 7 Present Present Present Right Present Present Absent Absent Probable 8 Present Present Present Right Absent Present Absent Absent Classic 9 Present Present Present Right Present Present Present Absent Classic 10 Present Absent Absent NA Absent Absent Absent Absent Indeterminate 11 Present Present Present Left Absent Present Absent Absent Classic 12 Present Present Present Left Absent Present Absent Absent Classic 13 Absent Absent Absent NA Absent Absent Absent Absent Indeterminate 14 Present Present Present Midline Present Absent Absent Present Classic 15 Absent Present Present Midline Present Present Present Absent Indeterminate 16 Present Present Present Left Absent Absent Absent Absent Classic 17 Present Present Present Right Present Present Absent Absent Classic 18 Absent Absent Absent NA Absent Absent Present Present Indeterminate 19 Present Present Present Left Absent Present Absent Absent Classic 20 Absent Present Absent Right Present Present Present Absent Indeterminate 21 Present Present Present Midline Present Present Absent Absent Classic Sensitivity (%) 76 (16 present, 5 absent) 86 (18 present, 3 absent) 76 (16 present, 5 absent) a (9 right, 4 midline, 5 left) 43 (9 present, 12 absent) 57 (12 present, 9 absent) 33 (7 present, 14 absent) 90 (19 absent, 2 present) 52 (11 classic), 14 (3 probable), 33 (7 indeterminate) 95% CI (%) a Classic, 32 72; probable, 4 35; indeterminate, Average κ Note Overall impression of volvulus was graded by gestalt perception rather than use of a specified combination of imaging features. NA indicates not applicable. a Sensitivity for right loop apex was 43% (95% CI, 24 63%); for midline loop apex was 19% (95% CI, 7 41%); and for left loop apex was 24% (95% CI, 10 45%). AJR:194, January

5 Levsky et al. TABLE 3: Cross-Sectional Imaging Findings in Patients With Confirmed Sigmoid Volvulus (n = 21) Patient No. Disproportionate Sigmoid Dilatation Degree Sigmoid Dilatation Proximal Colonic Dilatation Rectal Decompression Whirl Sign No. of Transition Points X-Marksthe-Spot Sign Split-Wall Sign Overall Impression of Volvulus Signs of Bowel Compromise 1 Present Severe Absent Absent Absent 1 Absent Present Possible 2 Present Severe Present Present Present 2 Present Present Definite Surrounding stranding, free fluid, pneumatosis intestinalis 3 Present Moderate Present Absent Present 1 Absent Present Definite 4 Present Severe Absent Absent Present 2 Absent Present Definite Surrounding stranding, free fluid 5 Present Severe Absent Absent Absent 0 Absent Absent Not evident 6 Absent Moderate Present Present Absent 1 Absent Present Probable Surrounding stranding 7 Present Severe Absent Absent Present 1 Absent Present Definite 8 Present Severe Absent Absent Absent 1 Absent Present Probable 9 Present Severe Present Present Absent 2 Present Absent Definite 10 Present Moderate Present Absent Present 1 Absent Present Definite Free fluid 11 Present Moderate Absent Present Present 1 Absent Absent Definite 12 Present Severe Present Present Present 2 Present Absent Definite Wall thickening, free fluid 13 Present Severe Absent Absent Absent 2 Absent Absent Not evident 14 Present Severe Absent Absent Present 2 Absent Present Definite Surrounding stranding, free fluid 15 Absent Moderate Present Present Present 2 Present Absent Definite 16 Present Moderate Absent Present Present 2 Present Absent Definite Surrounding stranding, free fluid 17 Present Severe Present Present Present 2 Present Absent Definite Surrounding stranding 18 Absent Severe Present Present Absent 2 Present Absent Definite 19 Present Mild Present Absent Absent 2 Present Absent Definite Surrounding stranding, free fluid 20 Present Moderate Present Present Present 1 Absent Present Definite Free fluid 21 Present Moderate Absent Absent Absent 1 Present Present Definite Surrounding stranding 52 (11 present, 10 absent) Sensitivity (%) 86 (18 present, 3 absent) a (12 severe, 8 moderate, 1 mild) 52 (11 present, 10 absent) 48 (10 present, 11 absent) 57 (12 present, 9 absent) b (2 points, 12; 1 point, 8; 0 points, 1) 43 (9 present, 12 absent) 52 (11 present, 10 absent) 76 (16 definite), 10 (2 probable), 5 (1 possible), 10 (2 not evident) 95% CI (%) a b Definite, 55 90; probable, 1 30; possible, 0 24; not evident, 1 30 Average κ c Note Overall impression of volvulus was graded by gestalt perception rather than using a specified combination of imaging features. a Sensitivity for severe sigmoid dilatation was 57% (95% CI, 37 76%); for moderate sigmoid dilatation was 38% (95% CI, 21 59%); and for mild sigmoid dilation was 5% (95% CI, 1 30%). b Sensitivity for two transition points was 57% (95% CI, 37 76%); for one transition point was 38% (95% CI, 21 59%); and for either one or two transition points was 95% (95% CI, %). c Average kappa values for signs of bowel compromise were bowel wall thickening, 0.33; surrounding stranding, 0.65; free fluid, 0.46; pneumatosis intestinalis, AJR:194, January 2010

6 CT of Sigmoid Volvulus Fig year-old man with sigmoid volvulus (patient 21). Scanogram graded classic for sigmoid volvulus shows distended sigmoid loop with inverted U configuration and coffee bean sign. Sigmoid overlaps liver and extends cephalad to transverse colon (northern exposure sign). There is no proximal colonic dilatation. sectional imaging corroborated the finding. When the scanogram did not depict disproportionate sigmoid enlargement or proximal colonic dilatation, the cross-sectional imaging finding was present in two of five and four of 14 cases, respectively. There was poor correlation between absence of rectal gas on scanograms and rectal decompression on cross-sectional images (9/21, 43%). To evaluate the contribution of cross-sectional imaging signs in the setting of atypical radiographic findings, we compared the subgroups of patients with findings scored classic (n = 11) and those with findings not scored classic (n = 10). Patients with classic scanogram impressions were more likely to have disproportionate sigmoid dilatation (11/11), two sigmoid colon transition points (9/11), and the X-marks-the-spot sign (7/11) compared with patients with nonclassic scanogram impressions (7/10, 3/10, and 2/10). Patients with classic and nonclassic scanogram findings had approximately equal proportions of positive whirl (7/11 and 5/10) and split-wall (5/11 and 6/10) signs. Five of seven patients with scanograms scored as indeterminate had cross-sectional imaging that was scored as definite sigmoid volvulus. Endoscopic or pathologic evidence of bowel wall compromise was found in 11 of 21 patients (52%), as follows: hemorrhage (patients 12, 13, 14, and 15), necrosis (patients 2, 17, and 20), ulcer (patients 7 and 10), ischemia Fig year-old man with sigmoid volvulus (patient 5). Scanogram graded probable for volvulus shows disproportionate sigmoid dilatation and coffee bean sign. Loop does not extend cephalad to transverse colon or substantially overlap liver shadow. (patient 18), and perforation (patient 9). Six of 11 patients (55%) with endoscopic or pathologic findings of ischemia had one or more suggestive imaging signs, and five of 10 patients (50%) without ischemia had one or more imaging signs of bowel wall compromise. There was no statistically significant correlation between imaging signs and the endoscopic or pathologic finding of ischemia (odds ratio, 1.2; Wald 95% CI, ). CT signs of ischemia were present in all three patients with pathologically proven bowel necrosis. No patients were found to have free air. Patients 9, 17, and 18 died during hospitalization. Only one of the three patients who died had any CT sign of bowel compromise. Discussion Recognition of the spectrum of CT appearances of sigmoid volvulus is important even though this entity has been diagnosed radiographically in the past. In the largest published retrospective study we identified (from Turkey, an area with endemic sigmoid volvulus), plain radiographs were diagnostic in 460 of 708 cases (65%) [13]. Only 38% of the patients in our study had undergone radiography and 10% had undergone barium enema before intervention. We suspect that similar imaging practices are widespread. CT is the current diagnostic examination of choice for nonspecific gastrointestinal problems and is often performed first. Fig year-old woman with sigmoid volvulus (patient 15). Scanogram graded indeterminate for volvulus shows gaseous dilatation of entire colon. Although coffee bean sign is present, sigmoid dilatation is not disproportionate. We evaluated the sensitivity of cross-sectional imaging findings that have not, to our knowledge, been studied before. The whirl sign is well recognized as a sign of volvulus of various parts of the gastrointestinal tract [8], but its sensitivity for sigmoid volvulus has not been evaluated. In our series, a whirl sign was present in only 12 of 21 cases (57%), which indicates the need for additional imaging signs of sigmoid volvulus. Moderate or severe dilatation of the sigmoid and the presence of at least one sigmoid colon transition point were highly sensitive, but obstruction of any cause can cause dilatation and a transition point, which renders these signs nonspecific. Assessment for the presence of two transition points would be expected to be more specific. To our surprise, two transition points were detected in only 12 of 21 cases (57%). In addition, we observed that five of 21 cases (24%) could not be confidently confirmed as sigmoid volvulus with cross-sectional imaging. This finding emphasizes that a previously unappreciated spectrum of disease appearances is present on CT scans and that not every patient will have findings one would expect with complete mesenteric twisting. We identified two new signs of sigmoid volvulus intended to reflect the pathophysiologic features of volvulus. We call the presence of two crossing sigmoid transition points at a single location the X-marks-thespot sign (Fig. 1). This sign reflects a com- AJR:194, January

7 Levsky et al. plete twist of two limbs of a bowel loop on themselves, resembling a twist-tie. This finding was expected to be sensitive, but it was detected in only nine of 21 cases (43%). Barium enema studies generally do not assess the status of the proximal sigmoid because the contrast column does not pass the distal transition. It was therefore generally assumed that the proximal sigmoid also had a sharp transition due to a complete twist, but this assumption was not borne out at crosssectional imaging. Several patients had subtler twisting of a single limb of the distal sigmoid that caused visual separation of the loop walls by mesenteric fat (split-wall sign) (Fig. 2). The single loop assumes a bilobed or C shape due to mass effect from the adjacent fat at its midportion, making it appear to be two decompressed bowel loops. A split-wall sign was found in 11 of the 21 patients (52%). We suspect that the split-wall sign represents an earlier stage in twisting or a twist of lesser extent and that the X-marks-the-spot sign results from a greater degree of twist. Twisting of the distal portion of the sigmoid may be a precursor to a more complete twist involving both the ascending and descending limbs. Alternatively, the split-wall sign may be the result of untwisting of a previous complete twist. When the two limbs are completely twisted around each other, the tightness of the turn may preclude visualization of the boundup mesenteric fat. This suggestion was confirmed statistically when the odds ratio analysis showed an inverse correlation between the two signs. The utility of the X-marks-the-spot and split-wall signs is supported by the small number of cases (3/21,14%) in which neither finding was made. We assessed several established radiographic findings. In our series, an indeterminate radiographic impression was assigned in seven of 21 patients (33%). This finding is similar to that in a previous report [4] in which one third of cases were deemed diagnostically challenging. We identified a distended ahaustral sigmoid with an inverted U shape in 86% of cases. This finding is similar to a previous study [6] that showed the finding in 94% of cases. The same report, however, suggested that this sign is nonspecific. Instead, extension of the distended sigmoid loop to the left hemidiaphragm was found to be both sensitive and specific. We did not identify this sign in our study sample. The original study describing the northern exposure sign reported a sensitivity of 86% [7], however, we found this sign in only 57% of cases. These disparities may be attributable in part to differences in technique or scoring of these signs. We used scanograms, which are obtained with the patient supine, whereas in several previous studies supine and upright radiographs were assessed. Scanogram technique also may accentuate aspects of bowel findings differently than does radiography. The coffee bean sign is thought to be specific for the diagnosis of sigmoid volvulus [6]. This sign was identified in 76% of patients in our series and may be the best initial feature suggesting sigmoid volvulus. The location of the apex of the loop, however, does not significantly contribute to the diagnosis despite the popular belief that sigmoid volvulus classically points toward the right upper quadrant. In our study, equal numbers of patients had sigmoid loops either at midline or to the left. The most sensitive radiographic finding in our series was the absence of rectal gas on the scanogram, but this finding would be expected in any case of obstruction. The absence of rectal gas, however, correlated poorly with rectal decompression on cross-sectional images. Care should be exercised in drawing conclusions about the presence or absence of rectal content on supine radiographs. We found proximal colonic dilatation in approximately one half of cases, similar to findings in a previous study [7]. Five of the seven scanograms deemed indeterminate showed proximal colonic dilatation, and this finding contributed to diagnostic uncertainty. Because no radiographic sign is highly sensitive, there is a role for cross-sectional imaging signs. The results of our subgroup analysis suggest that if radiographs are indeterminate, the split-wall and whirl signs may be more helpful than signs of more complete twisting (two transition points and the X- marks-the-spot sign). The additive value of CT is further illustrated by the observation that in five of seven cases (71%) in which scanograms were graded indeterminate, cross-sectional images were scored as definitely showing sigmoid volvulus. CT is often used to assess bowel ischemia, the fundamental complication of sigmoid volvulus. Bowel ischemia can progress to infarction, perforation, and death. In our analysis, CT signs correlated poorly with endoscopically or surgically proven ischemia in general. However, CT findings were present in all three patients with bowel necrosis, suggesting higher sensitivity if frank infarction is present. Results of some previous studies suggest that the CT findings of ischemia are insensitive [14] and nonspecific [15, 16], but others have shown high sensitivity in the detection of bowel wall compromise [17] and good correlation with extent of ischemia or infarction [18, 19]. Our study had several limitations. First, the sample size analyzed was relatively small. Low disease prevalence limits the feasibility of a large-scale study. Second, all images were analyzed by radiologists who knew the final diagnosis, and this knowledge might have biased the scoring of specific imaging signs based on a scorer s predetermined assumptions. We controlled for this factor to an extent by using three reviewers. Third, some imaging features of sigmoid volvulus were assessed subjectively and without numeric criteria. To correct for this factor, we used consensus in cases of interobserver discrepancy. Fourth, cases were identified by a specific International Classification of Diseases code and endoscopic or surgical pathologic confirmation, so cases may have been missed if they were not coded appropriately. Our inclusion criteria might have imposed selection bias toward patients who had more severe disease that would result in surgery and the resulting surgical pathologic evaluation. Fifth, sensitivity for the whirl sign might have been underestimated because multiplanar reformations were not made. Coronal reconstructions were used when available, and they were generally considered helpful. Finally, we used no control group of potential entities that can mimic volvulus to evaluate the predictive values and specificity of imaging signs. Although it would be challenging to choose an appropriate comparison group in a systematic way, this design would be an appropriate avenue of further research. We conclude that CT of patients with sigmoid volvulus shows a spectrum of findings that can be approached with the use of established and novel imaging signs but that indeterminate features can be present in one fourth of patients. The need to assess for multiple imaging signs is supported by the relatively low sensitivity of several findings that have been generally assumed to be sensitive (e.g., the whirl sign and the northern exposure sign). We suggest the X-marks-the-spot sign as a marker of more complete volvulus and the split-wall sign as a marker of less severe twisting. CT signs were ineffective for prediction of the presence of pathologically proven ischemia until there was frank bowel necrosis. 142 AJR:194, January 2010

8 CT of Sigmoid Volvulus Additional investigation to determine whether imaging signs of sigmoid volvulus have prognostic value would be of interest. 6. Burrell HC, Baker DM, Wardrop P, Evans AJ. Significant plain film findings in sigmoid volvulus. Clin Radiol 1994; 49: J Gastroenterol Hepatol 2009; 24: Alpern MB, Glazer GM, Francis IR. Ischemic or infarcted bowel: CT findings. Radiology 1988; Acknowledgments We thank Jordan Winick for preparing Figures 1A and 2A. We thank Joseph DiVito, Steven Kramer, and Raisa Goldman for assistance in obtaining archived CT studies. References 1. Lal SK, Morgenstern R, Vinjirayer EP, Matin A. Sigmoid volvulus an update. Gastrointest Endosc Clin N Am 2006; 16: Safioleas M, Chatziconstantinou C, Felekouras E, et al. Clinical considerations and therapeutic strategy for sigmoid volvulus in the elderly: a study of 33 cases. World J Gastroenterol 2007; 13: Rigler LG, Lipschultz O. Roentgenologic findings in acute obstruction of the colon: with particular reference to acute volvulus of the sigmoid. Radiology 1940; 35: Young WS, Engelbrecht HE, Stoker A. Plain film analysis in sigmoid volvulus. Clin Radiol 1978; 29: Agrez M, Cameron D. Radiology of sigmoid volvulus. Dis Colon Rectum 1981; 24: Javors BR, Baker SR, Miller JA. The northern exposure sign: a newly described finding in sigmoid volvulus. AJR 1999; 173: Shaff MI, Himmelfarb E, Sacks GA, Burks DD, Kulkarni MV. The whirl sign: a CT finding in volvulus of the large bowel. J Comput Assist Tomogr 1985; 9: Catalano O. Computed tomographic appearance of sigmoid volvulus. Abdom Imaging 1996; 21: QuickCalcs. GraphPad Software Website. Confidence interval of a proportion or a count. www. graphpad.com/quickcalcs/confinterval1.cfm. Accessed June 19, Geertzen J. Cohen s kappa for more than two annotators with multiple classes. cosmion.net/jeroen/ software/kappa. Accessed June 22, Two by two table analysis online. SISA Simple Interactive Statistical Analysis Website. www. quantitativeskills.com/sisa/statistics/twoby2. htm. Accessed June 19, Atamanalp SS, Yildirgan MI, Başoğlu M, et al. Clinical presentation and diagnosis of sigmoid volvulus: outcomes of 40-year and 859-patient experience. 166: Kernagis LY, Levine MS, Jacobs JE. Pneumatosis intestinalis in patients with ischemia: correlation of CT findings with viability of the bowel. AJR 2003; 180: Taourel P, Garibaldi F, Arrigoni J, Le Guen V, Lesnik A, Bruel JM. Cecal pneumatosis in patients with obstructive colon cancer: correlation of CT findings with bowel viability. AJR 2004; 183: Frager D, Baer JW, Medwid SW, Rothpearl A, Bossart P. Detection of intestinal ischemia in patients with acute small-bowel obstruction due to adhesions or hernia: efficacy of CT. AJR 1996; 166: Romano S, Romano L, Grassi R. Multidetector row computed tomography findings from ischemia to infarction of the large bowel. Eur J Radiol 2007; 61: Lee SS, Ha HK, Park SH, et al. Usefulness of computed tomography in differentiating transmural infarction from nontransmural ischemia of the small intestine in patients with acute mesenteric venous thrombosis. J Comput Assist Tomogr 2008; 32: AJR:194, January