- Email: [email protected]
Halo Sign: Useful CT Sign for Differentiating Benign from Malignant Colonic Disease
Clinical Radiology (2003) 58: 306–310 doi:10.1016/S0009-9260(02)00520-2, available online at www.sciencedirect.com
Halo Sign: Useful CT Sign for Differentiating Benign from Malignant Colonic Disease M . G . H A R I SI N G H A N I , J. W I T T E N B E R G , M. A . B L A K E , S . C H E N , K . J H A V E R I , P . R. MU E L L E R Division of Abdominal Imaging and Intervention, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, USA Received: 30 July 2002 Revised: 21 October 2002
Accepted: 25 October 2002
AIM: To evaluate the halo sign for accurately distinguishing benign from malignant colonic wall thickening. MATERIALS AND METHODS: Computed tomography (CT) examinations of 92 patients (70 men; 22 women; mean age 57 years) with pathologically proven colonic wall thickening (51 benign and 41 malignant) were retrospectively reviewed in a blinded fashion. The affected segment was assessed for presence of the halo sign, degree and uniformity of thickness and density of the intramural stratum. RESULTS: The halo sign was present in 74.5% (38/51) patients with benign and 7.3% (3/41) patients with malignant bowel disease. The presence of the halo sign was 75.4% sensitive and 92.5% specific for benign bowel wall thickening. All 38 benign halos showed uniform, continuous stratification; only one of three malignant halos met the strict criteria for benign halo. CONCLUSION: The halo sign is a moderately sensitive and highly specific sign for distinguishing benign from malignant bowel wall thickening. However, it is not pathognomonic for benign disease. Detailed analysis of halo characteristics is necessary to improve the usefulness of this finding. Harisinghani M. G. et al. (2003). Clinical Radiology 58, 306– 310. q 2003 The Royal College of Radiologists. Published by Elsevier Science Ltd. All rights reserved. Key words: computed tomography, halo sign, colonic disease.
INTRODUCTION
MATERIAL AND METHODS
A stratified appearance of multiple continuous, uniformly annular layers in a thickened intestinal loop was originally labelled as the double halo sign [1] and has since also been referred to as the halo and “target” sign [1 –3]. We refer to all variations as the halo sign for sake of simplicity. Originally described in Crohn’s disease [1], this sign has subsequently been documented in a host of non-malignant bowel diseases [4] and has been widely accepted as a reliable indicator of benign bowel wall thickening [4 –9]. However, it is now known that certain types of malignancy, particularly metastatic scirrhous adenocarcinoma, [10–12] can also demonstrate the halo sign. The purpose of this study was to establish the usefulness of the halo sign for distinguishing benign from malignant colonic thickening.
We retrospectively reviewed the medical and pathological records of 92 consecutive patients (70 men; 22 women; mean age 57 years) with pathologically proven large bowel diseases who had undergone abdominal and pelvic computed tomography (CT) between January 1999 and June 2000. This Institutional Review Board (IRB) approved study group comprised 51 patients with benign bowel disease (35 pseudomembranous colitis, 13 inflammatory bowel disease, three diverticulitis) and 41 patients with malignant colonic disease (39 with primary adenocarcinoma, two with metastatic adenocarcinoma). The cause of bowel wall thickening was documented by reviewing endoscopic (n ¼ 38) or surgical (n ¼ 21) biopsy results in 59 patients and laboratory results in 33 patients (all having pseudomembranous colitis). Confirmation of Clostridium difficile colitis was based on positive stool assay for C. difficile cytotoxin. All patients underwent abdominal and pelvic CT using a helical CT system (HiSpeed CT/i; GE Medical Systems, Milwaukee, WI, U.S.A.). Helical data with contiguous 5 mm slices were acquired from the diaphragm to the symphysis pubis. All patients received 600 – 900 ml of 2.5% diluted sodium amidotrizoate and
Author for correspondence and guarantor of study: Dr Mukesh G. Harisinghani, Division of Abdominal Imaging and Intervention, Massachusetts General Hospital, Harvard Medical School, 55 Fruit Street, Boston, MA 02114, USA. Tel: þ1-617-726-8396; Fax: þ1-617726-4891; E-mail: [email protected] 0009-9260/03/$30.00/0
q 2003 The Royal College of Radiologists. Published by Elsevier Science Ltd. All rights reserved.
HALO SIGN: USEFUL CT SIGN FOR DIFFERENTIATING BENIGN FROM MALIGNANT COLONIC DISEASE
Fig. 1 – Pseudomembranous colitis affecting caecum. Contrast-enhanced axial CT image shows circumferential thickening of the caecum with a lowdensity halo (arrow) and enhancing muscularis propria (arrowhead). Accompanying pericolonic soft tissue inflammatory stranding is seen.
meglumine amidotrizoate mixture (Gastrografin, Schering, Berlin, Germany) approximately 4 h, 2 h, and immediately before CT. In addition 150 ml non-ionic water-soluble iodinated contrast medium (Oxilan, Cook Imaging Corporation; Bloomington, IN, U.S.A.) was mechanically injected (MCT Plus; Medrad, Pittsburgh, PA, U.S.A.) at a rate of 2 ml/s using a 20 gauge IV catheter and the data acquired after a 70 s delay. Two radiologists experienced in abdominal imaging performed qualitative assessment of the images unaware of the clinical or pathological findings in a blinded, independent fashion. Images were viewed on clinical workstations at standard soft tissue parameters (window width ca. 300 and level ca. þ40 HU units). Differences among readers were settled by consensus. The site and length of bowel thickening, bowel wall thickness, presence of halo sign, measurement of the intramural layers, and uniformity of the thickening of the halo components were evaluated. Abnormal colonic wall thickness with a well-distended lumen was defined as .3 mm. An outer layer of continuous high density surrounding a lower density parallel layer comprised the criteria for a halo sign (Fig. 1). Sensitivity and specificity of the halo sign for distinguishing benign from malignant bowel wall thickening was determined.
307
Fig. 2 – Chronic Crohn’s disease. Contrast-enhanced axial CT image shows halo sign with a central area of low attenuation halo in the submucosa (arrow) of the colon. The low attenuation measured 260 HU representing fat density.
Fig. 3 – Crohn’s disease in a young patient. Contrast-enhanced CT shows sigmoid colon thickening. A water density halo is seen in longitudinal section. The low-density inner layer measured 40 HU.
RESULTS
The halo sign was observed more commonly, but not exclusively, in benign colonic conditions. It was present in 38 of 51 (74.5%) patients with benign and three of 41 (7.3%) patients with malignant colonic disease. In the benign disease categories, the halo sign was present in 28 of 35 patients with pseudomembranous colitis and 10 of 13 patients with inflammatory bowel disease (Figs 2 and 3). The three patients with malignant disease who had a halo sign on CT included two patients with primary mucinous adenocarcinoma, and a patient with metastatic scirrhous adenocarcinoma of the sigmoid and rectum (Figs 4 and 5). The mean bowel wall thickness in patients those with the
Fig. 4 – Primary mucinous adenocarcinoma of transverse colon with a halo sign. Focal eccentric thickening of the colonic wall (arrow). While there is low-density within the thickened wall (arrowhead), the lack of uniform thickness in the inner layer is uncharacteristic of a benign halo. Note the adjacent lymph node (curved arrow).
308
CLINICAL RADIOLOGY
discontinuity were present. Also seen in these two patients was non-uniform intramural thickness manifested by unilateral thickening, as well as asymmetric thickening. In the single patient with metastatic gastric scirrhous adenocarcinoma and a halo sign the halo layers appeared uniform and uninterrupted. The sensitivity of halo sign for diagnosing benign colonic disease in our study group was 74.5%, and the specificity was 92.7%.
DISCUSSION
Fig. 5 – Metastasis involving the rectum from scirrhous adenocarcinoma of the stomach. (a ) On axial helical CT, wall thickening with target variation of the halo sign is seen. There is an abnormal inner ring of enhancement (arrows) seen in the mucosa. The submucosa is hypoattenuating (arrowhead). The outer rim of enhancement presumably represents the muscularis propria. (b ) Specimen micrograph from rectal biopsy demonstrates thickened mucosa infiltrated by inflammatory cells (arrow), as well as signet cells of gastric adenocarcinoma (arrowhead). Note that the mucosa crypts (curved arrow) are displaced by the infiltrative tumour but not destroyed. The muscularis mucosa (asterix) is also intact.
benign disease and a halo sign ranged from 1.1 to 2.4 cm, whereas in the malignant category the range was 1.6 to 3.2 cm. The mean length of involved colonic segment was 8.4 cm in benign compared with 3.4 cm in malignant disease. The density of the intramural stratum of the halo sign varied from 275 to 40 HU (mean 8 HU). Average attenuation values were calculated from three separate measurements. Of the 38 patients in the benign group with a halo sign, all had continuous circumferential, uniform thickness of both halo components. In two of the three patients with malignant disease and a halo sign the outer layer was uniform in thickness, although areas of
Bowel wall stratification or layering has proven to be a distinctive and reliable sign of intestinal injury. Its apparent unequivocality for the presence of intestinal disease overcomes the dilemma of pseudo wall thickening, which is a consequence of inadequate luminal distention or contrast streaming. Not all thickened bowel walls elicit stratification but when present the challenge becomes one of differential diagnosis. Originally observed in benign conditions of both the small and large bowel [1 –3], the halo sign has recently been described as occurring in malignancy [10,11,13,14]. This study was undertaken to determine its incidence in both aetiologies, and whether other halo characteristics distinguish between benign and malignant halo. Because that differential diagnostic challenge is much more common in the large than small intestine, we restricted our review to only the colonic halo group. The halo sign in our study group was highly specific (92.7%) for benign colonic disease. Its sensitivity was somewhat lower at 74.5% indicating that although this sign is not present in all benign bowel diseases, its presence is indicative of benign aetiology. These results compare favourably with other CT signs used to distinguish benign from malignant diseases. In a prospective study evaluating the ability of CT to distinguish between sigmoid diverticulitis and carcinoma, the most reliable signs for diverticulitis were pericolonic inflammation (sensitivity 95%, specificity 34%), and length of involved segment . 10 cm (sensitivity 95%, specificity 31%) [15]. As noted by Chintapalli et al. [15], the major difficulty in distinguishing benign from malignant bowel disease was in the relative infrequency of any given single sign. With multivariate analysis, the two best combined signs of benignity were absence of lymph nodes and presence of inflammation/oedema (sensitivity 92%, specificity 78%) [15]. Thus, the halo sign, even when used alone, compares favourably with other markers of benignity. In Frager’s [1] original description of the halo sign in Crohn’s disease, the inner ring was low in attenuation (2 8– 10 HU) and was felt to represent mucosal and submucosal oedema. The higher attenuation outer ring was attributed to the remainder of the bowel wall. While the pathophysiological aetiology of halo sign in benign diseases is the preferential infiltration of the submucosa by oedema or fat, [16] the aetiology of halo sign in malignancy is more complex. The most extensively documented malignant disease with this finding is scirrhous carcinoma of the gastrointestinal tract [10–12]. Primary scirrhous carcinoma originates more frequently in gastric than colonic carcinoma and is characterized less often by invasion of the mucosa or forming bulky tumour masses, but rather by spread along lymphatic channels of the
HALO SIGN: USEFUL CT SIGN FOR DIFFERENTIATING BENIGN FROM MALIGNANT COLONIC DISEASE
bowel wall resulting in reactive proliferation of connective tissue and possible lymph oedema [10,11]. In previously reported series, the halo sign was seen in two of 22 patients (9.1%) with scirrhous carcinoma of the stomach [10], and in six of 22 patients (27%) with metastatic gastric carcinoma to the rectum [11]. The low attenuation inner ring was attributed to widening of submucosa by tumour-induced scirrhous reaction [10]. Both authors described the halo sign as consisting of an inner low attenuation ring and outer high attenuation ring. Recently, Jang et al. [12] described a reverse variation of the halo consisting of a higher density inner layer and lower density outer layer, in patients with intestinal metastases from gastric adenocarcinoma. Non-scirrhous colonic adenocarcinoma may also occasionally exhibit the halo sign [4,13,17]. In adenocarcinomas that grow concentrically, an inner ring of low attenuation may be caused by tumour necrosis [4,13,17]. Mucinous adenocarcinomas may also be of lower attenuation due to intracellular mucin deposition [4]. In our series, in two of the three patients who had a halo sign due to malignancy, the pathology of the resected specimen showed primary mucinous adenocarcinoma replacing the bowel layers. The low attenuation represented tumour full of mucin juxtaposed against a thin outer layer of enhancing granulation tissue, giving the appearance of a halo. Both of these cases had morphological features, as previously described, which distinguished them from benign entities. In the malignant halo indistinguishable from benign halo, pathology showed metastatic scirrhous adenocarcinoma of the stomach. The involved segment of colon showed an inner hyperattenuating zone, outer hyperattenuating zone and a hypoattenuating middle zone, referred to as the target sign [4,11,12]; the target sign is considered a subtype of the halo sign in this study. Pathological analysis showed a thickened mucosa infiltrated predominately by inflammatory cells, as well as occasional signet cells of gastric adenocarcinoma. The mucosal crypts were displaced by the infiltrative tumour but not destroyed. The muscularis mucosa was also intact. In two of the three malignant cases the internal morphology was grossly different from the benign halo. The thickness of the inner layer was more variable and the outer layer was discontinuous. The segmental distribution was also different in that most, but not all of the benign lesions, were multisegmental, (extending over a number of portions of colon divided into rectum, sigmoid, descending, transverse, ascending colon and caecum) while the carcinomas were segmental in distribution. However, the overlap of this differential distribution was inconsistent as a minority of benign diseases was also more focally distributed. Furthermore, a coexistence of ischaemic colitis proximal to colon carcinoma [18,19], a complication in 1– 7% of colon carcinoma can also be expected to be multisegmental in distribution. In a study of patients with ischaemic colitis proximal to colon cancer, the halo sign was seen in four of 20 patients (20%), and appears to be the most specific sign in distinguishing between segments of ischaemia and cancer [20]. Thus when the halo sign is present the involved segment should be evaluated for additional discriminators, such as length of segment involved, degree and uniformity of bowel wall thickening, homogeneous density of the bowel wall, and
309
consistency of the halo sign [15,17,21]. Evaluating these characteristics should increase confidence when distinguishing benign from malignant disease with a halo. For instance, in one of our patients with primary colon malignancy, the halo was only seen on one transaxial image, while the remainder of the large caecal tumour exhibited homogeneous attenuation. Conversely Fig. 5 demonstrates a malignant halo sign completely indistinguishable from the appearance of benign disease. Finally, age and symptoms are additional factors that influence differential diagnosis. A clearly defined internal fat layer may prove to be a sign, albeit uncommon, of benignity. In our series, the fat density halo was observed in two of 51 (3.9%) of benign processes. This finding, which was first described in Crohn’s disease, was shown to represent submucosal fat accumulation [16], and is now known to be a feature of chronic or subacute inflammatory bowel disease [7]. There are no documented reports in the literature of fatty halo in malignancy. However, it is expected that this may prove fallible in patients developing malignancy superimposed on an inflammatory bowel disease background. This study has some limitations. Due to consecutive patient selection of pathologically proven cases of bowel wall disease in our hospital, the benign category was dominated by pseudomembranous colitis; therefore, the true incidence of the halo sign in a larger variety of benign disease cannot be established. Our study used a standard CT technique and we did not investigate how different contrast medium dosage or timing parameters might affect the visualization of enhancement in the bowel wall that contributes to the halo sign. The “mucosal” attenuation of the target variation of the halo sign is also best appreciated when the bowel is distended with water-attenuation contrast material [22]. In summary, we report the sensitivity and specificity of the halo sign for benign bowel wall thickening in our study population to be 74.5 and 92.7%, respectively. When present, the halo sign is highly specific for benign process. Caution, however, needs to be exercised because malignant disease is infrequently present with a halo sign, as we and others have observed [10,11,13,14]. The halo sign should be used in conjunction with other CT and clinical signs for ultimate decision making.
REFERENCES 1 Frager DH, Goldman M, Beneventano TC. Computed tomography in Crohn’s disease. J Comput Assist Tomogr 1983;7:819–824. 2 Berliner L, Redmond P, Purow E, et al. Computed tomography in Crohn’s disease. Am J Gastroenterol 1982;77:548 –553. 3 Goldberg HI, Gore RM, Margulis AR, et al. Computed tomography in the evaluation of Crohn’s disease. AJR Am J Roentgenol 1983;140: 277– 282. 4 Macari M, Balthazar EJ. CT of bowel wall thickening: significance and pitfalls of interpretation. AJR Am J Roentgenol 2001;176:1105– 1116. 5 Horton KM, Corl FM, Fishman EK. CT evaluation of the colon: inflammatory disease. RadioGraphics 2000;20:399–418. 6 Gore RM, Balthazar EJ, Ghahremani GG, et al. CT features of ulcerative colitis and Crohn’s disease. AJR Am J Roentgenol 1996;167: 3–15. 7 Philpotts LE, Heiken JP, Westcott MA, et al. Colitis: use of CT findings in differential diagnosis. Radiology 1994;190:445 –449.
310
CLINICAL RADIOLOGY
8 Balthazar EJ, Chako AC. Computerized tomography in acute gastrointestinal disorders. Am J Gastroenterol 1990;85:1445 –1452. 9 Balthazar EJ, Hulnick D, Megibow AJ, et al. Computed tomography of intramural intestinal hemorrhage and bowel ischemia. J Comput Assist Tomogr 1987;11:67–72. 10 Balthazar EJ, Siegel SE, Megibow AJ, et al. CT in patients with scirrhous carcinoma of the GI tract: imaging findings and value for tumor detection and staging. AJR Am J Roentgenol 1995;165: 839–845. 11 Ha HK, Jee KR, Yu E, et al. CT features of metastatic linitis plastica to the rectum in patients with peritoneal carcinomatosis. AJR Am J Roentgenol 2000;174:463 –466. 12 Jang HJ, Lim HK, Kim HS, et al. Intestinal metastases from gastric adenocarcinoma: helical CT findings. J Comput Assist Tomogr 2001;25: 61–67. 13 Kim AY, Ha HK, Seo BK, et al. CT of patients with right-sided colon cancer and distal ileal thickening. AJR Am J Roentgenol 2000;175: 1439–1444. 14 Moffat RE, Gourley WK. Ileal lymphatic metastases from cecal carcinoma. Radiology 1980;135:55–58. 15 Chintapalli KN, Chopra S, Ghiatas AA, et al. Diverticulitis versus colon
16
17 18 19
20
21
22
cancer: differentiation with helical CT findings. Radiology 1999;210: 429– 435. Jones B, Fishman EK, Hamilton SR, et al. Submucosal accumulation of fat in inflammatory bowel disease: CT/pathologic correlation. J Comput Assist Tomogr 1986;10:759 –763. Balthazar EJ. CT of the gastrointestinal tract: principles and interpretation. AJR Am J Roentgenol 1991;156:23–32. Rutledge RH. Pseudo-ulcerative colitis proximal to obstructing colon carcinoma. Am Surg 1969;35:384 –388. Toner M, Condell D, O’Briain DS. Obstructive colitis. Ulceroinflammatory lesions occurring proximal to colonic obstruction. Am J Surg Pathol 1990;14:719–728. Ko GY, Ha HK, Lee HJ, et al. Usefulness of CT in patients with ischemic colitis proximal to colonic cancer. AJR Am J Roentgenol 1997;168:951–956. Balthazar EJ, Megibow AJ, Hulnick D, et al. Carcinoma of the colon: detection and preoperative staging by CT. AJR Am J Roentgenol 1988; 150:301–306. Wittenberg J, Harisinghani MG, Jhaveri K, Varghese J, Mueller PR. Algorithmic approach to CT diagnosis of the abnormal bowel wall. RadioGraphics 2002;22:1093–1107.
Halo Sign: Useful CT Sign for Differentiating Benign from Malignant Colonic Disease M . G . H A R I SI N G H A N I , J. W I T T E N B E R G , M. A . B L A K E , S . C H E N , K . J H A V E R I , P . R. MU E L L E R Division of Abdominal Imaging and Intervention, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, USA Received: 30 July 2002 Revised: 21 October 2002
Accepted: 25 October 2002
AIM: To evaluate the halo sign for accurately distinguishing benign from malignant colonic wall thickening. MATERIALS AND METHODS: Computed tomography (CT) examinations of 92 patients (70 men; 22 women; mean age 57 years) with pathologically proven colonic wall thickening (51 benign and 41 malignant) were retrospectively reviewed in a blinded fashion. The affected segment was assessed for presence of the halo sign, degree and uniformity of thickness and density of the intramural stratum. RESULTS: The halo sign was present in 74.5% (38/51) patients with benign and 7.3% (3/41) patients with malignant bowel disease. The presence of the halo sign was 75.4% sensitive and 92.5% specific for benign bowel wall thickening. All 38 benign halos showed uniform, continuous stratification; only one of three malignant halos met the strict criteria for benign halo. CONCLUSION: The halo sign is a moderately sensitive and highly specific sign for distinguishing benign from malignant bowel wall thickening. However, it is not pathognomonic for benign disease. Detailed analysis of halo characteristics is necessary to improve the usefulness of this finding. Harisinghani M. G. et al. (2003). Clinical Radiology 58, 306– 310. q 2003 The Royal College of Radiologists. Published by Elsevier Science Ltd. All rights reserved. Key words: computed tomography, halo sign, colonic disease.
INTRODUCTION
MATERIAL AND METHODS
A stratified appearance of multiple continuous, uniformly annular layers in a thickened intestinal loop was originally labelled as the double halo sign [1] and has since also been referred to as the halo and “target” sign [1 –3]. We refer to all variations as the halo sign for sake of simplicity. Originally described in Crohn’s disease [1], this sign has subsequently been documented in a host of non-malignant bowel diseases [4] and has been widely accepted as a reliable indicator of benign bowel wall thickening [4 –9]. However, it is now known that certain types of malignancy, particularly metastatic scirrhous adenocarcinoma, [10–12] can also demonstrate the halo sign. The purpose of this study was to establish the usefulness of the halo sign for distinguishing benign from malignant colonic thickening.
We retrospectively reviewed the medical and pathological records of 92 consecutive patients (70 men; 22 women; mean age 57 years) with pathologically proven large bowel diseases who had undergone abdominal and pelvic computed tomography (CT) between January 1999 and June 2000. This Institutional Review Board (IRB) approved study group comprised 51 patients with benign bowel disease (35 pseudomembranous colitis, 13 inflammatory bowel disease, three diverticulitis) and 41 patients with malignant colonic disease (39 with primary adenocarcinoma, two with metastatic adenocarcinoma). The cause of bowel wall thickening was documented by reviewing endoscopic (n ¼ 38) or surgical (n ¼ 21) biopsy results in 59 patients and laboratory results in 33 patients (all having pseudomembranous colitis). Confirmation of Clostridium difficile colitis was based on positive stool assay for C. difficile cytotoxin. All patients underwent abdominal and pelvic CT using a helical CT system (HiSpeed CT/i; GE Medical Systems, Milwaukee, WI, U.S.A.). Helical data with contiguous 5 mm slices were acquired from the diaphragm to the symphysis pubis. All patients received 600 – 900 ml of 2.5% diluted sodium amidotrizoate and
Author for correspondence and guarantor of study: Dr Mukesh G. Harisinghani, Division of Abdominal Imaging and Intervention, Massachusetts General Hospital, Harvard Medical School, 55 Fruit Street, Boston, MA 02114, USA. Tel: þ1-617-726-8396; Fax: þ1-617726-4891; E-mail: [email protected] 0009-9260/03/$30.00/0
q 2003 The Royal College of Radiologists. Published by Elsevier Science Ltd. All rights reserved.
HALO SIGN: USEFUL CT SIGN FOR DIFFERENTIATING BENIGN FROM MALIGNANT COLONIC DISEASE
Fig. 1 – Pseudomembranous colitis affecting caecum. Contrast-enhanced axial CT image shows circumferential thickening of the caecum with a lowdensity halo (arrow) and enhancing muscularis propria (arrowhead). Accompanying pericolonic soft tissue inflammatory stranding is seen.
meglumine amidotrizoate mixture (Gastrografin, Schering, Berlin, Germany) approximately 4 h, 2 h, and immediately before CT. In addition 150 ml non-ionic water-soluble iodinated contrast medium (Oxilan, Cook Imaging Corporation; Bloomington, IN, U.S.A.) was mechanically injected (MCT Plus; Medrad, Pittsburgh, PA, U.S.A.) at a rate of 2 ml/s using a 20 gauge IV catheter and the data acquired after a 70 s delay. Two radiologists experienced in abdominal imaging performed qualitative assessment of the images unaware of the clinical or pathological findings in a blinded, independent fashion. Images were viewed on clinical workstations at standard soft tissue parameters (window width ca. 300 and level ca. þ40 HU units). Differences among readers were settled by consensus. The site and length of bowel thickening, bowel wall thickness, presence of halo sign, measurement of the intramural layers, and uniformity of the thickening of the halo components were evaluated. Abnormal colonic wall thickness with a well-distended lumen was defined as .3 mm. An outer layer of continuous high density surrounding a lower density parallel layer comprised the criteria for a halo sign (Fig. 1). Sensitivity and specificity of the halo sign for distinguishing benign from malignant bowel wall thickening was determined.
307
Fig. 2 – Chronic Crohn’s disease. Contrast-enhanced axial CT image shows halo sign with a central area of low attenuation halo in the submucosa (arrow) of the colon. The low attenuation measured 260 HU representing fat density.
Fig. 3 – Crohn’s disease in a young patient. Contrast-enhanced CT shows sigmoid colon thickening. A water density halo is seen in longitudinal section. The low-density inner layer measured 40 HU.
RESULTS
The halo sign was observed more commonly, but not exclusively, in benign colonic conditions. It was present in 38 of 51 (74.5%) patients with benign and three of 41 (7.3%) patients with malignant colonic disease. In the benign disease categories, the halo sign was present in 28 of 35 patients with pseudomembranous colitis and 10 of 13 patients with inflammatory bowel disease (Figs 2 and 3). The three patients with malignant disease who had a halo sign on CT included two patients with primary mucinous adenocarcinoma, and a patient with metastatic scirrhous adenocarcinoma of the sigmoid and rectum (Figs 4 and 5). The mean bowel wall thickness in patients those with the
Fig. 4 – Primary mucinous adenocarcinoma of transverse colon with a halo sign. Focal eccentric thickening of the colonic wall (arrow). While there is low-density within the thickened wall (arrowhead), the lack of uniform thickness in the inner layer is uncharacteristic of a benign halo. Note the adjacent lymph node (curved arrow).
308
CLINICAL RADIOLOGY
discontinuity were present. Also seen in these two patients was non-uniform intramural thickness manifested by unilateral thickening, as well as asymmetric thickening. In the single patient with metastatic gastric scirrhous adenocarcinoma and a halo sign the halo layers appeared uniform and uninterrupted. The sensitivity of halo sign for diagnosing benign colonic disease in our study group was 74.5%, and the specificity was 92.7%.
DISCUSSION
Fig. 5 – Metastasis involving the rectum from scirrhous adenocarcinoma of the stomach. (a ) On axial helical CT, wall thickening with target variation of the halo sign is seen. There is an abnormal inner ring of enhancement (arrows) seen in the mucosa. The submucosa is hypoattenuating (arrowhead). The outer rim of enhancement presumably represents the muscularis propria. (b ) Specimen micrograph from rectal biopsy demonstrates thickened mucosa infiltrated by inflammatory cells (arrow), as well as signet cells of gastric adenocarcinoma (arrowhead). Note that the mucosa crypts (curved arrow) are displaced by the infiltrative tumour but not destroyed. The muscularis mucosa (asterix) is also intact.
benign disease and a halo sign ranged from 1.1 to 2.4 cm, whereas in the malignant category the range was 1.6 to 3.2 cm. The mean length of involved colonic segment was 8.4 cm in benign compared with 3.4 cm in malignant disease. The density of the intramural stratum of the halo sign varied from 275 to 40 HU (mean 8 HU). Average attenuation values were calculated from three separate measurements. Of the 38 patients in the benign group with a halo sign, all had continuous circumferential, uniform thickness of both halo components. In two of the three patients with malignant disease and a halo sign the outer layer was uniform in thickness, although areas of
Bowel wall stratification or layering has proven to be a distinctive and reliable sign of intestinal injury. Its apparent unequivocality for the presence of intestinal disease overcomes the dilemma of pseudo wall thickening, which is a consequence of inadequate luminal distention or contrast streaming. Not all thickened bowel walls elicit stratification but when present the challenge becomes one of differential diagnosis. Originally observed in benign conditions of both the small and large bowel [1 –3], the halo sign has recently been described as occurring in malignancy [10,11,13,14]. This study was undertaken to determine its incidence in both aetiologies, and whether other halo characteristics distinguish between benign and malignant halo. Because that differential diagnostic challenge is much more common in the large than small intestine, we restricted our review to only the colonic halo group. The halo sign in our study group was highly specific (92.7%) for benign colonic disease. Its sensitivity was somewhat lower at 74.5% indicating that although this sign is not present in all benign bowel diseases, its presence is indicative of benign aetiology. These results compare favourably with other CT signs used to distinguish benign from malignant diseases. In a prospective study evaluating the ability of CT to distinguish between sigmoid diverticulitis and carcinoma, the most reliable signs for diverticulitis were pericolonic inflammation (sensitivity 95%, specificity 34%), and length of involved segment . 10 cm (sensitivity 95%, specificity 31%) [15]. As noted by Chintapalli et al. [15], the major difficulty in distinguishing benign from malignant bowel disease was in the relative infrequency of any given single sign. With multivariate analysis, the two best combined signs of benignity were absence of lymph nodes and presence of inflammation/oedema (sensitivity 92%, specificity 78%) [15]. Thus, the halo sign, even when used alone, compares favourably with other markers of benignity. In Frager’s [1] original description of the halo sign in Crohn’s disease, the inner ring was low in attenuation (2 8– 10 HU) and was felt to represent mucosal and submucosal oedema. The higher attenuation outer ring was attributed to the remainder of the bowel wall. While the pathophysiological aetiology of halo sign in benign diseases is the preferential infiltration of the submucosa by oedema or fat, [16] the aetiology of halo sign in malignancy is more complex. The most extensively documented malignant disease with this finding is scirrhous carcinoma of the gastrointestinal tract [10–12]. Primary scirrhous carcinoma originates more frequently in gastric than colonic carcinoma and is characterized less often by invasion of the mucosa or forming bulky tumour masses, but rather by spread along lymphatic channels of the
HALO SIGN: USEFUL CT SIGN FOR DIFFERENTIATING BENIGN FROM MALIGNANT COLONIC DISEASE
bowel wall resulting in reactive proliferation of connective tissue and possible lymph oedema [10,11]. In previously reported series, the halo sign was seen in two of 22 patients (9.1%) with scirrhous carcinoma of the stomach [10], and in six of 22 patients (27%) with metastatic gastric carcinoma to the rectum [11]. The low attenuation inner ring was attributed to widening of submucosa by tumour-induced scirrhous reaction [10]. Both authors described the halo sign as consisting of an inner low attenuation ring and outer high attenuation ring. Recently, Jang et al. [12] described a reverse variation of the halo consisting of a higher density inner layer and lower density outer layer, in patients with intestinal metastases from gastric adenocarcinoma. Non-scirrhous colonic adenocarcinoma may also occasionally exhibit the halo sign [4,13,17]. In adenocarcinomas that grow concentrically, an inner ring of low attenuation may be caused by tumour necrosis [4,13,17]. Mucinous adenocarcinomas may also be of lower attenuation due to intracellular mucin deposition [4]. In our series, in two of the three patients who had a halo sign due to malignancy, the pathology of the resected specimen showed primary mucinous adenocarcinoma replacing the bowel layers. The low attenuation represented tumour full of mucin juxtaposed against a thin outer layer of enhancing granulation tissue, giving the appearance of a halo. Both of these cases had morphological features, as previously described, which distinguished them from benign entities. In the malignant halo indistinguishable from benign halo, pathology showed metastatic scirrhous adenocarcinoma of the stomach. The involved segment of colon showed an inner hyperattenuating zone, outer hyperattenuating zone and a hypoattenuating middle zone, referred to as the target sign [4,11,12]; the target sign is considered a subtype of the halo sign in this study. Pathological analysis showed a thickened mucosa infiltrated predominately by inflammatory cells, as well as occasional signet cells of gastric adenocarcinoma. The mucosal crypts were displaced by the infiltrative tumour but not destroyed. The muscularis mucosa was also intact. In two of the three malignant cases the internal morphology was grossly different from the benign halo. The thickness of the inner layer was more variable and the outer layer was discontinuous. The segmental distribution was also different in that most, but not all of the benign lesions, were multisegmental, (extending over a number of portions of colon divided into rectum, sigmoid, descending, transverse, ascending colon and caecum) while the carcinomas were segmental in distribution. However, the overlap of this differential distribution was inconsistent as a minority of benign diseases was also more focally distributed. Furthermore, a coexistence of ischaemic colitis proximal to colon carcinoma [18,19], a complication in 1– 7% of colon carcinoma can also be expected to be multisegmental in distribution. In a study of patients with ischaemic colitis proximal to colon cancer, the halo sign was seen in four of 20 patients (20%), and appears to be the most specific sign in distinguishing between segments of ischaemia and cancer [20]. Thus when the halo sign is present the involved segment should be evaluated for additional discriminators, such as length of segment involved, degree and uniformity of bowel wall thickening, homogeneous density of the bowel wall, and
309
consistency of the halo sign [15,17,21]. Evaluating these characteristics should increase confidence when distinguishing benign from malignant disease with a halo. For instance, in one of our patients with primary colon malignancy, the halo was only seen on one transaxial image, while the remainder of the large caecal tumour exhibited homogeneous attenuation. Conversely Fig. 5 demonstrates a malignant halo sign completely indistinguishable from the appearance of benign disease. Finally, age and symptoms are additional factors that influence differential diagnosis. A clearly defined internal fat layer may prove to be a sign, albeit uncommon, of benignity. In our series, the fat density halo was observed in two of 51 (3.9%) of benign processes. This finding, which was first described in Crohn’s disease, was shown to represent submucosal fat accumulation [16], and is now known to be a feature of chronic or subacute inflammatory bowel disease [7]. There are no documented reports in the literature of fatty halo in malignancy. However, it is expected that this may prove fallible in patients developing malignancy superimposed on an inflammatory bowel disease background. This study has some limitations. Due to consecutive patient selection of pathologically proven cases of bowel wall disease in our hospital, the benign category was dominated by pseudomembranous colitis; therefore, the true incidence of the halo sign in a larger variety of benign disease cannot be established. Our study used a standard CT technique and we did not investigate how different contrast medium dosage or timing parameters might affect the visualization of enhancement in the bowel wall that contributes to the halo sign. The “mucosal” attenuation of the target variation of the halo sign is also best appreciated when the bowel is distended with water-attenuation contrast material [22]. In summary, we report the sensitivity and specificity of the halo sign for benign bowel wall thickening in our study population to be 74.5 and 92.7%, respectively. When present, the halo sign is highly specific for benign process. Caution, however, needs to be exercised because malignant disease is infrequently present with a halo sign, as we and others have observed [10,11,13,14]. The halo sign should be used in conjunction with other CT and clinical signs for ultimate decision making.
REFERENCES 1 Frager DH, Goldman M, Beneventano TC. Computed tomography in Crohn’s disease. J Comput Assist Tomogr 1983;7:819–824. 2 Berliner L, Redmond P, Purow E, et al. Computed tomography in Crohn’s disease. Am J Gastroenterol 1982;77:548 –553. 3 Goldberg HI, Gore RM, Margulis AR, et al. Computed tomography in the evaluation of Crohn’s disease. AJR Am J Roentgenol 1983;140: 277– 282. 4 Macari M, Balthazar EJ. CT of bowel wall thickening: significance and pitfalls of interpretation. AJR Am J Roentgenol 2001;176:1105– 1116. 5 Horton KM, Corl FM, Fishman EK. CT evaluation of the colon: inflammatory disease. RadioGraphics 2000;20:399–418. 6 Gore RM, Balthazar EJ, Ghahremani GG, et al. CT features of ulcerative colitis and Crohn’s disease. AJR Am J Roentgenol 1996;167: 3–15. 7 Philpotts LE, Heiken JP, Westcott MA, et al. Colitis: use of CT findings in differential diagnosis. Radiology 1994;190:445 –449.
310
CLINICAL RADIOLOGY
8 Balthazar EJ, Chako AC. Computerized tomography in acute gastrointestinal disorders. Am J Gastroenterol 1990;85:1445 –1452. 9 Balthazar EJ, Hulnick D, Megibow AJ, et al. Computed tomography of intramural intestinal hemorrhage and bowel ischemia. J Comput Assist Tomogr 1987;11:67–72. 10 Balthazar EJ, Siegel SE, Megibow AJ, et al. CT in patients with scirrhous carcinoma of the GI tract: imaging findings and value for tumor detection and staging. AJR Am J Roentgenol 1995;165: 839–845. 11 Ha HK, Jee KR, Yu E, et al. CT features of metastatic linitis plastica to the rectum in patients with peritoneal carcinomatosis. AJR Am J Roentgenol 2000;174:463 –466. 12 Jang HJ, Lim HK, Kim HS, et al. Intestinal metastases from gastric adenocarcinoma: helical CT findings. J Comput Assist Tomogr 2001;25: 61–67. 13 Kim AY, Ha HK, Seo BK, et al. CT of patients with right-sided colon cancer and distal ileal thickening. AJR Am J Roentgenol 2000;175: 1439–1444. 14 Moffat RE, Gourley WK. Ileal lymphatic metastases from cecal carcinoma. Radiology 1980;135:55–58. 15 Chintapalli KN, Chopra S, Ghiatas AA, et al. Diverticulitis versus colon
16
17 18 19
20
21
22
cancer: differentiation with helical CT findings. Radiology 1999;210: 429– 435. Jones B, Fishman EK, Hamilton SR, et al. Submucosal accumulation of fat in inflammatory bowel disease: CT/pathologic correlation. J Comput Assist Tomogr 1986;10:759 –763. Balthazar EJ. CT of the gastrointestinal tract: principles and interpretation. AJR Am J Roentgenol 1991;156:23–32. Rutledge RH. Pseudo-ulcerative colitis proximal to obstructing colon carcinoma. Am Surg 1969;35:384 –388. Toner M, Condell D, O’Briain DS. Obstructive colitis. Ulceroinflammatory lesions occurring proximal to colonic obstruction. Am J Surg Pathol 1990;14:719–728. Ko GY, Ha HK, Lee HJ, et al. Usefulness of CT in patients with ischemic colitis proximal to colonic cancer. AJR Am J Roentgenol 1997;168:951–956. Balthazar EJ, Megibow AJ, Hulnick D, et al. Carcinoma of the colon: detection and preoperative staging by CT. AJR Am J Roentgenol 1988; 150:301–306. Wittenberg J, Harisinghani MG, Jhaveri K, Varghese J, Mueller PR. Algorithmic approach to CT diagnosis of the abnormal bowel wall. RadioGraphics 2002;22:1093–1107.