Clinical Radiology 70 (2015) 262e269
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Acute colitis: differential diagnosis using multidetector CT L. Plastaras a, L. Vuitton a, N. Badet b, S. Koch a, V. Di Martino c, E. Delabrousse b, d, * a
Department of Gastroenterology, University Hospital, 3 Boulevard Fleming, 25030 Besanc¸on, France Department of Radiology, University Hospital, 3 Boulevard Fleming, 25030 Besanc¸on, France c Department of Hepatology, University Hospital, 3 Boulevard Fleming, 25030 Besanc¸on, France d EA 4662 Nanomedicine Lab, Imagery and Therapeutics, University of Franche-Comt e, Besanc¸on, France b
article in formation Article history: Received 9 May 2014 Received in revised form 10 October 2014 Accepted 10 November 2014
AIM: To investigate the utility of multidetector CT (MDCT) in helping to establish the underlying cause of acute colitis. METHODS AND MATERIALS: All patients who had acute colitis with a well-identified cause and underwent abdomen 64-MDCT were included in the study. MDCT images were retrospectively analysed in a blinded fashion and the CT findings were correlated with the eventual aetiological diagnosis. RESULTS: The study population included 105 patients. Acute colitis was related to inflammatory bowel disease in 43 cases. MDCT was used to identify six relevant signs of inflammatory colitis: the “comb” sign (p < 0.001), enlarged lymph nodes (p < 0.001), abscess (p ¼ 0.026), fibro-fatty infiltration (p ¼ 0.007), small bowel involvement (p < 0.001), and the absence of an “empty colon” sign (p ¼ 0.045). Multivariate logistic regression analysis identified three independent signs of inflammatory colitis: the “comb” sign, small bowel involvement, and enlarged lymph nodes. Acute colitis was related to bacterial infection in 35 cases. Five signs were significantly associated with infectious colitis: continuous distribution (p ¼ 0.020), an “empty colon” sign (p ¼ 0.002), absence of fat stranding (p ¼ 0.013), and absence of a “comb” sign (p ¼ 0.010) and absence of enlarged lymph nodes (p ¼ 0.035). Multivariate analysis identified three independent signs: the “empty colon” sign and absence of fat stranding and of a “comb” sign. The remaining causes were ischaemic colitis (n ¼ 21) and drug-related colitis (n ¼ 6). MDCT examination provided five relevant signs of ischaemic colitis: fat stranding (p ¼ 0.002), discontinuous distribution (p < 0.001), and absence of enlarged lymph node (p < 0.001), a “comb” sign (p ¼ 0.006) and small bowel involvement (p ¼ 0.037). CONCLUSIONS: MDCT provides certain suggestive signs that may be helpful in distinguishing the underlying aetiological cause of acute colitis. Ó 2014 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
* Guarantor and correspondent: E. Delabrousse, Service de Radiologie rale, CHRU Besanc¸on, 25000 Besanc¸on, France. Tel.: þ33 3 81 66 93 80; visce fax: þ33 3 81 66 84 95. E-mail address: [email protected] (E. Delabrousse).
Introduction Acute colitis represents a group of diseases that are commonly encountered in routine gastroenterological
http://dx.doi.org/10.1016/j.crad.2014.11.008 0009-9260/Ó 2014 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
practice. The prognosis of this disease depends partly on the initiation of a specific treatment tailored to the aetiology of colitis. The aetiology of colitis can be difficult to determine as these diseases may manifest themselves in non-specific clinical presentations. CT is often the first imaging test performed in this context. In recent years, numerous publications have described the importance of CT in the initial confirmation of a diagnosis of acute colitis, definition of the extent of the disease process, and providing established criteria for the common types of acute colitis.1e5 Welldefined CT criteria exist for each type of acute colitis. However, to the authors’ knowledge, only one retrospective study, published in 1994, sought to establish a link between the diagnosis of acute colitis and suggestive CT signs that may help in establishing the underlying cause.6 Although the results of that study are still valid and used to guide the aetiological diagnosis, imaging techniques have improved since, with the objective of obtaining images of higher spatial resolution and better contrast with the latest generation of multidetector CT (MDCT) scanners. The present study retrospectively analysed various CT signs of acute colitis in a large cohort of patients, in an attempt to identify suggestive signs that may help in elucidating the underlying aetiological cause. To do this, the various MDCT signs from the literature were retrospectively studied. A new sign, termed the “empty colon” sign is described, which represents the complete emptiness of the colonic lumen. All the results were statistically analysed in light of the final diagnosis obtained for acute colitis.
Methods and materials Study design and data collection A retrospective, observational, single-centre study was conducted on all consecutive patients admitted to the gastroenterology department with a diagnosis of acute colitis and who underwent abdomen MDCT during a 4 year period. The list of all patients admitted for acute colitis (primary diagnosis) was established by the Department of Medical Information, University Hospital, Besanc¸on, France, using the International Classification of Diseases, 10th edition (ICD10). MDCT examinations are available on the hospital network, through the picture archiving and communicating system (PACS). For each patient, the following data were recorded from the charts: age, sex, history of hypertension, diabetes, excess weight, hypercholesterolemia, cardiovascular disease, family history of inflammatory bowel disease (IBD), recent use of antibiotics, recent changes in drugs received, clinical signs on admission including diarrhoea, abdominal pain, rectal bleeding, fever, high C-reactive protein (CRP) levels, high leukocyte counts, and anaemia. The cause of colitis, namely infectious, ischaemic, inflammatory [Crohn’s disease (CD) or ulcerative colitis (UC)] or drug-related colitis was retrospectively assessed by two gastroenterologists, by reviewing results of the pathological examinations, stool cultures, blood cultures, endoscopic
263
findings, and outcomes under appropriate therapy. Because of the small number of pseudomembranous colitis, these were included in the infectious colitis group. In the absence of a universal reference standard for the causal diagnosis of acute colitis,4,6 the aetiological diagnosis was considered clearly identified when concordant features (including endoscopic, pathological, bacteriological features, and outcome) were present. Only such cases were selected for the present study.
MDCT examinations MDCT examinations were performed on two different multidetector devices: a MDCT 64 sections (Somatom 64, Siemens, Erlangen, Germany), and a MDCT 64 section (Brilliance 64, Philips, Eindhoven, The Netherlands). For each patient concerned, the review included the acquisition of the abdomen 90 s after intravenous injection of 90e110 ml iodinated contrast agent (iomeprol, Iomeron 300, Bracco, Italy) using an automatic injector at a flow rate of 2 ml/s. For five patients, the acquisitions were performed without contrast injection due to contraindications. Thin contiguous axial sections and multiplanar reconstructions were performed routinely. The examinations were replayed in abdomen (W ¼ 350, L ¼ 50), pneumoperitoneum (W ¼ 500, L ¼ e50) and lung parenchymal windowing (W ¼ 1600, L ¼ e600). The images were retrospectively analysed and reviewed independently in agreement with two radiologists experienced in gastrointestinal imaging who were unaware of the patients’ clinical histories or the final diagnoses. The following CT signs from the literature were selected for the study and were recorded for each patient: wall thickness >3 mm,7 a “fat halo” sign or a “water halo” sign, a “comb” sign, an “accordion” sign, fibro-fatty proliferation, fat stranding, small bowel involvement, and the presence of abscess, fistula, ascites, or enlarged lymph nodes (short axis >10 mm). The distribution of continuous or noncontinuous colonic changes and the location of the diseased colon (right, left or diffuse) were also recorded. The “halo” sign is an aspect of the bowel wall usually represented by a stratification of the wall in two or three concentric rings of different density and enhancement. The appearance of the intermediate layer (corresponding to the submucosa) of fluid density or fat density was distinguished. The “comb” sign is a stretching and stiffening of the vessels on the mesenteric side of the bowel, leading to a combed appearance and hyperaemia of the mesocolon at imaging. The vessels are abnormally thick and tortuous. The “accordion” sign shows significant submucosal oedema, irregular thickening, pseudo-polypoid colonic wall, and a strong mucosal enhancement. Fibro-fatty proliferation corresponds to proliferation of fat and fibrous tissue in the mesentery. Fat stranding refers to intraperitoneal exudate leading to high-density of the peritoneal fat. Fistulas,2 abscesses on all bowel loops, and enlarged lymph nodes (>1 cm in short axis dimension) throughout the mesentery were searched for. Wall abnormality was considered continuous if the affected areas were not interspersed with healthy areas of colon. Distribution of colonic abnormality
264
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
was defined as right, left, or diffuse abnormality. Left distribution included lesions of the rectum, sigmoid, left colon, and the distal transverse; right distribution included lesions of the proximal transverse colon, ascending colon, and caecum. It should be noted that some types of colitis show abnormalities to the right and left without diffuse disease. Small bowel involvement was reported when an increase in small bowel wall thickness was seen, regardless of the segment. A new sign, the “empty colon” sign, which has never been described in the literature, was also specifically investigated. It was defined as a complete emptiness of the colonic lumen and resembled an empty colon without gas, fluid, or faeces.
Statistical analysis Three groups of colitis were studied and inter-group comparison performed: ischaemic colitis, inflammatory colitis, and infectious colitis. No specific analysis was conducted regarding drug-related colitis because of its low prevalence in the present study. Statistical comparisons between two groups used the chi-square test or Fisher’s exact test for categorical variables, and Student’s t-test or a ManneWhitney test for quantitative variables. The formulation of hypotheses was two-sided and a p-value under 0.05 was considered as significant. When the radiological signs appeared to be significant in the univariate analysis, their sensitivity, specificity, positive (PPV), and negative (NPV) predictive values were calculated. Multivariate analyses were performed for each group of colitis using logistic regression models. Parameters analysed though multivariate analyses were selected based on univariate analysis of the results and the absence of co-linearity. The diagnostic performance of each model was established using the area under the receiver operating characteristic (ROC) curve. Statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC, USA) and NCSS (Jerry Hintze, Visual Components, Sybase, Kaysville, UT, USA) software for Windows.
Results Study population During the study period, 1816 hospitalizations involving 1452 patients with acute colitis were recorded. Among these patients, 1186 were excluded because these patients did not undergo CT. Of the 266 remaining patients, 161 were excluded because the causal diagnosis was not clear after reviewing the medical records. The study population included 105 patients, 61 women and 44 men; average age 52 years (23e92 years). The causes of acute colitis were as follows: ischaemic colitis (n ¼ 21), inflammatory colitis (n ¼ 43, including 31 CD and 12 UC), infectious colitis (n ¼ 35), and drug-related colitis (n ¼ 6). The aetiological diagnosis was mainly based on endoscopic findings (76 patients), bacteriological and histopathological examination of biopsy samples (61 patients), stool cultures (nine patients), or positive blood culture (one patient). For
patients without bacteriological, endoscopic, or histopathological evidence, the causal diagnosis was assessed based on compatible features, suggestive clinical symptoms at baseline, and suggestive outcome after empirical therapy (n ¼ 17). The main characteristics of the study population are summarized in Table 1. Patients with inflammatory colitis were younger (39 years), especially when compared to patients with ischaemic colitis (76 years, p < 0.001); the sex ratio did not differ among the three groups of colitis; patients with ischaemic colitis usually suffered from a spectrum of suggestive comorbidities such as hypertension (p < 0.001), diabetes (p ¼ 0.028), excess weight (p ¼ 0.053), hypercholesterolemia (p ¼ 0.002), and a history of cardiovascular disease (p < 0.001). Patients with inflammatory colitis had more often a family history of IBD (p ¼ 0.018). Patients with infectious colitis more often received antibiotics prior to the onset of colitis (p < 0.001). The clinical presentation more often involved rectal bleeding (p ¼ 0.012) or anaemia (p ¼ 0.008) in the event of ischaemic colitis, abdominal pain in the event of inflammatory colitis (p ¼ 0.033), and in the event of infectious colitis, fever (p ¼ 0.004), a high leukocyte count (p ¼ 0.002), and high CRP levels (p < 0.001).
MDCT analysis results Overall, wall thickness >3 mm was present in 97% of patients, a “water halo” sign was present in 71% of patients, and conversely, a “fat halo” sign (Fig 1) was encountered in 6%. An accordion sign (Fig 2) was present in only 6%, and ascites in only 20% of patients in the present study. There was a variable distribution of colonic abnormality: 40% was in the right colon, 57% in the left colon, and 16% diffuse. These signs failed to discriminate between the three groups of acute colitis. The odds ratio (OR), specificity, sensitivity, PPV, and NPV of relevant MDCT signs for different groups of acute colitis are shown in Table 2. A “comb” sign (Fig 3) was present in 47% of IBD cases versus 3% of other acute colitis (p < 0.001), mainly associated with CD (61%) rather than UC (8%). A “comb” sign was never encountered in the event of ischaemic colitis (p ¼ 0.006) and was detected in only 6% of cases of infectious colitis (p ¼ 0.009). Fat stranding (Fig 4) was found in 71% of ischaemic colitis (p ¼ 0.001) versus 40% of other types of acute colitis. Fat stranding was detected in only 23% of cases of infectious colitis (p ¼ 0.012). Enlarged lymph nodes (Fig 5) were found in 67% of IBD (p < 0.001), versus 16% of other types of acute colitis. Enlarged lymph nodes were detected in only 23% of cases of infectious colitis (p ¼ 0.035), and they were detected in only 5% of cases of ischaemic colitis (p < 0.001). Small bowel involvement was found in 30% of IBD (p < 0.001) versus 5% of other types of acute colitis. Small bowel involvement was never encountered in the event of ischaemic colitis (p ¼ 0.037). Fibrofatty proliferation was found in 21% of IBD (p ¼ 0.007), versus 3% of other types of acute colitis. The newly described “empty colon” sign (Fig 6) was found in 23% of infectious colitis (p ¼ 0.002), versus 3% of other types of
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
265
Table 1 Demographics of the study population. Infectious colitisa
Inflammatory colitisa
Ischaemic colitisa
33
41
20
51.8 22.8 42 24 7 5 11 7 16 10
53 22.5 40 17 6 3 9 3 9 29b
39 17.5b 47 14 2 2 2 14b 2 0
76.2 11.8b 29 57b 19b 14b 33b 0 57b 0
25 65 83 29
43b 80 34 14
21 58 49b 44
5 48 17 57b
44 61 11
66b 83b 3
28 44 10
38 57 29b
Total Prevalence (%) Demographics Age Male (%) Hypertension (%) Diabetes (%) Overweight (%) Hypercholesterolaemia (%) ATCD familial IBD (%) ATCD vascular (%) Antibiotics (%) Clinical data fever (%) Diarrhoea (%) Abdominal pain (%) Rectal bleeding (%) Biological data Leukocytosis (%) High CRP (%) Anaemia (%)
ATCD, antecedent; IBD, inflammatory bowel disease; CRP, C-reactive protein. a Each group of colitis was compared with all other colitis. b Significant data, p < 0.05.
acute colitis. A “empty colon” sign was detected in only 2% of cases of IBD (p ¼ 0.045). Abscesses (Fig 4) were found in 9% of IBD (p ¼ 0.026), and were encountered in other types of acute colitis. Continuous colonic abnormality was found in 66% of infectious colitis (p ¼ 0.020) versus 41% of other types of acute colitis. Continuous colonic abnormality was
detected in only 14% of cases of ischaemic colitis (p < 0.001). The results of multivariate analysis are presented in Table 3 and Table 4. Three MDCT signs independently associated with infectious colitis were identified: the presence of the “empty colon” sign (p ¼ 0.013), the absence of fat stranding
Figure 1 A “fat halo” sign in a patient with proven ulcerative colitis. Contrast-enhanced axial MDCT image shows stratification of the caecal wall with fat-density of the intermediate layer corresponding to the submucosa (arrowheads).
Figure 2 An “accordion” sign in a patient with confirmed pseudomembranous colitis. Contrast-enhanced MDCT image with coronal reformation shows a significant thickening of the colonic wall with significant submucosal oedema and a strong mucosal enhancement.
266
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
Table 2 Odds ratio (OR), specificity (Sp), sensitivity (Se), positive (PPV) and negative (NPV) predictive values of relevant MDCT signs for different groups of colitis. Sign
Infectious colitis OR
Comb sign Fat stranding
0.15 0.31 0.37 Lymph nodes 2.71 Continuous distribution e Small bowel involvement e 12.32 Fibro-fatty proliferation e Empty colon sign Abscesses
Inflammatory colitis
Ischaemic colitis
Se (%) Sp (%) PPV (%) NPV (%) OR
Se (%) Sp (%) PPV (%) NPV (%) OR
5.7 22.9 22.9 65.7 e e 23.1 e
46.5 e 67.4 e 30.2 20.9 2.4 9.3
71.4 51.4 55.7 58.6 e e 97.1 e
9.1 19 20.5 44.2 e e 80 e
60.2 57.1 59.1 77.4 e e 71.6 e
26.1 e 10.77 e 8.52 7.94 0.14 14.24
96.8 e 83.9 e 95.2 96.8 85.5 100
90.9 e 74.4 e 81.2 81.8 10 100
72.3 e 78.8 e 66.3 e 55.8 61.4
Se (%) Sp (%) PPV (%) NPV (%)
0.06 0 5.28 71.4 0.06 4.8 0.12 14.3 0.01 0 e e e e e
73.8 67.9 54.8 41.7 80.9 e e
0 35.7 2.6 5.8 0 e e e
74.7 90.5 69.7 66 76.4 e e e
e, Not significant.
(p ¼ 0.014), and the absence of a “comb” sign (p ¼ 0.031), whereas continuous colonic abnormality and the absence of enlarged lymph nodes were not significant (Table 3). The area under the ROC curve of the logistic regression model was 0.802. Three separate/different signs independently associated with inflammatory colitis were identified: the presence of a “comb” sign (p ¼ 0.006), small bowel involvement (p ¼ 0.037), and enlarged lymph nodes (p < 0.001), whereas an “empty colon” sign, fibro-fatty proliferation, and the presence of an abscess were not significant (Table 4). The area under the ROC curve of the regression model was 0.904. Given the small number of patients in the ischaemic colitis group, no multivariate analysis was performed for this group.
Figure 3 A “comb” sign, fistula and abscess in a patient with proven Crohn’s disease. Contrast-enhanced axial MDCT image shows colonic wall thickening with a characteristic stretching and stiffening of the mesocolic vessels (arrowheads). Note the presence of an associated fistula and abscess (arrow).
Discussion Currently, MDCT is an indispensable examination in the initial evaluation of patients presenting with an acute abdomen. This includes individuals with acute colitis, where the initial diagnosis is clinically suspected in less than half of cases, as the presentation may be rather vague with atypical and/or febrile acute abdomen.6 Although there have been many previous descriptions of the classical CT findings in acute colitis, only one study has attempted to define suggestive CT signs that can provide strong clues as to the underlying aetiology in a large patient cohort.6 The strengths of the present study include its novelty, large patient cohort, and rigorous design. In the present study, multivariate analysis could identify three discriminating factors for the diagnosis of inflammatory
Figure 4 Fat stranding in a patient with confirmed ischaemia of the descending colon. Contrast-enhanced axial MDCT image shows a loss of transparency of the peritoneal fat (arrow) adjacent to the left colon.
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
267
Table 3 Comparison of multidetector CT signs of infectious colitis. Multivariate analysis
p-Value
Odds ratio (95% CI) Empty colon sign (%) Continuous distribution (%) Lymph nodes (%) Fat stranding (%) Comb sign (%)
12.32 2.1 0.33 0.26 0.17
(1.68e90.15) (0.78e5.7) (0.11e1.01) (0.09e0.77) (0.03e0.85)
0.013 0.143 0.053 0.014 0.031
Not significant at p > 0.2.
Figure 5 Enlarged lymph nodes in a patient with proven Crohn’s disease. Contrast-enhanced axial MDCT image shows an enlarged and strongly enhanced lymph node (arrowhead) in the vicinity of the right colon.
colitis, i.e., a “comb” sign, small bowel involvement, and the presence of enlarged lymph nodes; and three for the diagnosis of infectious colitis, i.e., the presence of an “empty colon” sign, continuous distribution of colonic abnormality, and absence of enlarged lymph nodes. The univariate analysis presented the following signs for ischaemic colitis: presence of fat stranding and absence of enlarged lymph nodes, “comb” sign, small bowel involvement, and continuous distribution of colonic abnormality. An increase in wall thickness, as reported in the study of Philpotts et al.,6 was found in the majority of cases and does not appear to discriminate between specific aetiologies; therefore, thickness of the colonic wall was not measured in
the present study. Thus, there may be a link between the aetiology of acute colitis and wall thickness, although this has never been clearly identified.7,8 The “fat halo” sign and the “water halo” sign were not associated with a specific aetiology of colitis. However, a “water halo” sign seems to appear when the submucosal oedema of the colonic wall is recent, whereas a “fat halo” sign may be more present in subacute or chronic colitis.9,10 The present study included only patients with acute colitis, the low percentage of “fat halo” sign found may explain this recruitment bias. A “comb” sign shows engorgement of mesenteric vessels and has been described in CD, especially when it is active.11 In the present series, the “comb” sign was found in 8% of cases of UC. This result seems surprising as UC is not transmural and does not reach the serosa. Theoretically, a “comb” sign should not be seen. However, some important aspects of fibro-fatty proliferation that are visible in UC could be interpreted as a “comb” sign, thus explaining these results. Moreover, a “comb” sign may also be viewed in other diseases such as vasculitis.12,13 Small bowel involvement is also correlated with the presence of an inflammatory disease, especially CD, but also by a UC reflux ileitis. Although small bowel involvement is reported to be specifically associated with inflammatory colitis, some infectious diseases, such as Yersinia infections, are likely to affect the small bowel and colon. Furthermore, in the present study, 9% of cases of infectious colitis showed this sign. It is interesting to find the presence of enlarged lymph nodes as a discriminating factor for inflammatory colitis and not for infectious
Figure 6 A “empty colon” sign in a patient with confirmed infectious colitis. Contrast-enhanced MDCT. (a) Coronal reformation image shows significant thickening of the transverse colon wall and a complete emptiness of the transverse lumen. (b) Idem for the left colon. (c) Axial image shows a “water halo” sign and a total emptiness of the left colon lumen.
268
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
Table 4 Comparison of MDCT signs of inflammatory colitis. Multivariate analysis
p-Value
Odds ratio (95% CI) Comb sign (%) Lymph nodes (%) Small bowel involvement (%) Fibro-fatty proliferation (%) Empty colon sign (%)
15.16 16.66 6.35 2.37 0.03
(2.19e105.1) (4.94e56.21) (1.11e36.29) (0.2e28.41) (0.01e1.19)
0.006 0.2.
colitis. However, the enlarged lymph nodes, reflecting the inflammation of the intestinal structures, should not mask the enlarged lymph nodes indicative of carcinoma. Patients with IBD are at a higher risk of developing colon or small bowel adenocarcinomas,14,15 and any lymph node >10 mm in short axis dimension should attract attention. The present study found continuous distribution of colonic abnormality in infectious colitis without predictive location. The population of infectious colitis was made up of all the different bacteria mixed together and subgroups were not used. The probability of the specific location according to the organism (such as the right colon for Yersinia or the left colon for Shigella) was therefore not observed. This study included five patients with confirmed Clostridium difficile colitis. The classic “accordion” sign was found in only one patient, and other patients with this sign were mainly those with non-infectious colitis pseudomembranous. These data seem to confirm that, as pointed out by Macari et al.,16 the “accordion” sign may be found with an organism other than Clostridium difficile or even in cases of other causes of acute colitis.17,18 In the present study, the “accordion” sign was seen in only 4e19% of cases of colitis with Clostridium difficile. Regarding infectious colitis, the multivariate model allowed the “empty colon” sign to be highlighted. The “empty colon” sign corresponds to a complete emptiness of the lumen of the colon. Whereas evidence of infectious colitis is often difficult to obtain (the causative agent is found in only 50% of cases), and it is often the clinical outcome after the initiation of antibiotic treatment that will confirm the diagnosis, the presence of an “empty colon” sign at CT, with a specificity of 0.97, seems a good indicator of this aetiology. The localization of ischaemic colitis in the present study was predominantly on the left (67%) as in the study of Philpotts et al. (77%).6 However, there was no statistically significant difference for that location when ischaemic colitis was compared to colitis of all other causes. According to the literature, diffuse colitis was present in 10% of cases, illustrating the fact that ischaemic colitis is usually segmental.19 It is also interesting to note that arterial or venous vascular abnormalities that could explain the ischaemic phenomena were not found in the present study. In most cases, these anomalies are not present; reperfusion was restored in most cases when MDCT was performed. Only one set of vascular abnormalities was found in up to 25% of patients.20 The present study has several limitations. Firstly, there was no control group of patients without colitis. Such a group would have been useful for assessing the specificity
of the MDCT signs described with more accuracy. However, the present study was pragmatic and only focused on patients with acute colitis. Secondly, the study was monocentric and retrospective, thus a substantial number of cases with an uncertain aetiology were excluded. Such procedures may compromise the applicability of the present results because of the small number of patients in each group. However, although “intent-to diagnose” results were not provided, well-documented cases were carefully selected regarding causal diagnosis for the accurate assessment of MDCT performance in characterizing acute colitis even if a pathological confirmation was not always obtained. Further studies, which would ideally be prospective and include protocol endoscopy, bacteriology, and colonic biopsy samples, should be conducted to better investigate uncertain cases and enable external validation. The present study only studied patients hospitalized in a gastroenterology department, i.e., those who did not require immediate surgical treatment. This may be considered as a selection bias, and may thus have decreased the prevalence of serious forms of acute colitis in the present series, especially ischaemic gangrenous colitis. However, these forms often have typical clinical presentations and are easy to diagnose at MDCT, which shows a complete lack of mucosal enhancement, corresponding to a transmural infarction or the presence of intestinal pneumatosis. In conclusion, MDCT provides certain suggestive signs that may be helpful in distinguishing the underlying aetiological cause of acute colitis, i.e., the presence of lymph nodes, comb sign, small bowel involvement, and fibro-fatty proliferation for inflammatory colitis; the presence of the “empty colon” sign and continuous distribution for infectious colitis; and the presence of fat stranding for ischaemic colitis. A larger prospective study should be conducted to confirm these results, including patients who are candidates for surgery. This further study would also validate the value of the “empty colon” sign for identifying infectious colitis.
References 1. Balthazar EJ. CT of the gastrointestinal tract: principles and interpretation. AJR Am J Roentgenol 1991;156:23e32. 2. Horton KM, Corl FM, Fishman EK. CT evaluation of the colon: inflammatory disease. RadioGraphics 2000;20:399e418. 3. Johnson GL, Johnson PT, Fishman EK. CT evaluation of the acute abdomen: bowel pathology spectrum of disease. Crit Rev Diagn Imaging 1996;37:163e90. 4. Thoeni RF, Cello JP. CT imaging of colitis. Radiology 2006;240:623e38. 5. Delabrousse E, Ferreira FR, Badet N, et al. Coping with the problems of diagnosis of acute colitis. Diagn Interv Imaging 2013;94:793e804. 6. Philpotts LE, Heiken JP, Westcott MA, et al. Colitis: use of CT findings in differential diagnosis. Radiology 1994;190:445e9. 7. Fisher JK. Normal colon wall thickness on CT. Radiology 1982;145:415e8. 8. Macari M, Balthazar EJ. CT of bowel wall thickening: significance and pitfalls of interpretation. AJR Am J Roentgenol 2001;176:1105e16. 9. Jones B, Fishman EK, Hamilton SR, et al. Submucosal accumulation of fat in inflammatory bowel disease: CTepathologic correlation. J Comput Assist Tomogr 1986;10:759e63. 10. Wittenberg J, Harisinghani MG, Jhaveri K, et al. Algorithmic approach to CT diagnosis of the abnormal bowel wall. RadioGraphics 2002;22:1093e107. discussion 1107e1109. 11. Lee SS, Ha HK, Yang S-K, et al. CT of prominent pericolic or perienteric vasculature in patients with Crohn’s disease: correlation with clinical
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
12. 13.
14. 15.
disease activity and findings on barium studies. AJR Am J Roentgenol 2002;179:1029e36. Madureira AJ. The “comb” sign. Radiology 2004;230:783e4. Byun JY, Ha HK, Yu SY, et al. CT features of systemic lupus erythematosus in patients with acute abdominal pain: emphasis on ischemic bowel disease. Radiology 1999;211:203e9. Eaden JA, Abrams KR, Mayberry JF. The risk of colorectal cancer in ulcerative colitis: a meta-analysis. Gut 2001;48:526e35. Jess T, Gamborg M, Matzen P, et al. Increased risk of intestinal cancer in Crohn’s disease: a meta-analysis of population-based cohort studies. Am J Gastroenterol 2005;100:2724e9.
269
16. Macari M, Balthazar EJ, Megibow AJ. The accordion sign at CT: a nonspecific finding in patients with colonic edema. Radiology 1999;211:743e6. 17. Fishman EK, Kavuru M, Jones B, et al. Pseudomembranous colitis: CT evaluation of 26 cases. Radiology 1991;180:57e60. 18. Ash L, Baker ME, O’Malley Jr CM, et al. Colonic abnormalities on CT in adult hospitalized patients with Clostridium difficile colitis: prevalence and significance of findings. AJR Am J Roentgenol 2006;186:1393e400. 19. Balthazar EJ, Yen BC, Gordon RB. Ischemic colitis: CT evaluation of 54 cases. Radiology 1999;211:381e8. 20. Guttormson NL, Bubrick MP. Mortality from ischemic colitis. Dis Colon Rectum 1989;32:469e72.
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Acute colitis: differential diagnosis using multidetector CT L. Plastaras a, L. Vuitton a, N. Badet b, S. Koch a, V. Di Martino c, E. Delabrousse b, d, * a
Department of Gastroenterology, University Hospital, 3 Boulevard Fleming, 25030 Besanc¸on, France Department of Radiology, University Hospital, 3 Boulevard Fleming, 25030 Besanc¸on, France c Department of Hepatology, University Hospital, 3 Boulevard Fleming, 25030 Besanc¸on, France d EA 4662 Nanomedicine Lab, Imagery and Therapeutics, University of Franche-Comt e, Besanc¸on, France b
article in formation Article history: Received 9 May 2014 Received in revised form 10 October 2014 Accepted 10 November 2014
AIM: To investigate the utility of multidetector CT (MDCT) in helping to establish the underlying cause of acute colitis. METHODS AND MATERIALS: All patients who had acute colitis with a well-identified cause and underwent abdomen 64-MDCT were included in the study. MDCT images were retrospectively analysed in a blinded fashion and the CT findings were correlated with the eventual aetiological diagnosis. RESULTS: The study population included 105 patients. Acute colitis was related to inflammatory bowel disease in 43 cases. MDCT was used to identify six relevant signs of inflammatory colitis: the “comb” sign (p < 0.001), enlarged lymph nodes (p < 0.001), abscess (p ¼ 0.026), fibro-fatty infiltration (p ¼ 0.007), small bowel involvement (p < 0.001), and the absence of an “empty colon” sign (p ¼ 0.045). Multivariate logistic regression analysis identified three independent signs of inflammatory colitis: the “comb” sign, small bowel involvement, and enlarged lymph nodes. Acute colitis was related to bacterial infection in 35 cases. Five signs were significantly associated with infectious colitis: continuous distribution (p ¼ 0.020), an “empty colon” sign (p ¼ 0.002), absence of fat stranding (p ¼ 0.013), and absence of a “comb” sign (p ¼ 0.010) and absence of enlarged lymph nodes (p ¼ 0.035). Multivariate analysis identified three independent signs: the “empty colon” sign and absence of fat stranding and of a “comb” sign. The remaining causes were ischaemic colitis (n ¼ 21) and drug-related colitis (n ¼ 6). MDCT examination provided five relevant signs of ischaemic colitis: fat stranding (p ¼ 0.002), discontinuous distribution (p < 0.001), and absence of enlarged lymph node (p < 0.001), a “comb” sign (p ¼ 0.006) and small bowel involvement (p ¼ 0.037). CONCLUSIONS: MDCT provides certain suggestive signs that may be helpful in distinguishing the underlying aetiological cause of acute colitis. Ó 2014 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
* Guarantor and correspondent: E. Delabrousse, Service de Radiologie rale, CHRU Besanc¸on, 25000 Besanc¸on, France. Tel.: þ33 3 81 66 93 80; visce fax: þ33 3 81 66 84 95. E-mail address: [email protected] (E. Delabrousse).
Introduction Acute colitis represents a group of diseases that are commonly encountered in routine gastroenterological
http://dx.doi.org/10.1016/j.crad.2014.11.008 0009-9260/Ó 2014 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
practice. The prognosis of this disease depends partly on the initiation of a specific treatment tailored to the aetiology of colitis. The aetiology of colitis can be difficult to determine as these diseases may manifest themselves in non-specific clinical presentations. CT is often the first imaging test performed in this context. In recent years, numerous publications have described the importance of CT in the initial confirmation of a diagnosis of acute colitis, definition of the extent of the disease process, and providing established criteria for the common types of acute colitis.1e5 Welldefined CT criteria exist for each type of acute colitis. However, to the authors’ knowledge, only one retrospective study, published in 1994, sought to establish a link between the diagnosis of acute colitis and suggestive CT signs that may help in establishing the underlying cause.6 Although the results of that study are still valid and used to guide the aetiological diagnosis, imaging techniques have improved since, with the objective of obtaining images of higher spatial resolution and better contrast with the latest generation of multidetector CT (MDCT) scanners. The present study retrospectively analysed various CT signs of acute colitis in a large cohort of patients, in an attempt to identify suggestive signs that may help in elucidating the underlying aetiological cause. To do this, the various MDCT signs from the literature were retrospectively studied. A new sign, termed the “empty colon” sign is described, which represents the complete emptiness of the colonic lumen. All the results were statistically analysed in light of the final diagnosis obtained for acute colitis.
Methods and materials Study design and data collection A retrospective, observational, single-centre study was conducted on all consecutive patients admitted to the gastroenterology department with a diagnosis of acute colitis and who underwent abdomen MDCT during a 4 year period. The list of all patients admitted for acute colitis (primary diagnosis) was established by the Department of Medical Information, University Hospital, Besanc¸on, France, using the International Classification of Diseases, 10th edition (ICD10). MDCT examinations are available on the hospital network, through the picture archiving and communicating system (PACS). For each patient, the following data were recorded from the charts: age, sex, history of hypertension, diabetes, excess weight, hypercholesterolemia, cardiovascular disease, family history of inflammatory bowel disease (IBD), recent use of antibiotics, recent changes in drugs received, clinical signs on admission including diarrhoea, abdominal pain, rectal bleeding, fever, high C-reactive protein (CRP) levels, high leukocyte counts, and anaemia. The cause of colitis, namely infectious, ischaemic, inflammatory [Crohn’s disease (CD) or ulcerative colitis (UC)] or drug-related colitis was retrospectively assessed by two gastroenterologists, by reviewing results of the pathological examinations, stool cultures, blood cultures, endoscopic
263
findings, and outcomes under appropriate therapy. Because of the small number of pseudomembranous colitis, these were included in the infectious colitis group. In the absence of a universal reference standard for the causal diagnosis of acute colitis,4,6 the aetiological diagnosis was considered clearly identified when concordant features (including endoscopic, pathological, bacteriological features, and outcome) were present. Only such cases were selected for the present study.
MDCT examinations MDCT examinations were performed on two different multidetector devices: a MDCT 64 sections (Somatom 64, Siemens, Erlangen, Germany), and a MDCT 64 section (Brilliance 64, Philips, Eindhoven, The Netherlands). For each patient concerned, the review included the acquisition of the abdomen 90 s after intravenous injection of 90e110 ml iodinated contrast agent (iomeprol, Iomeron 300, Bracco, Italy) using an automatic injector at a flow rate of 2 ml/s. For five patients, the acquisitions were performed without contrast injection due to contraindications. Thin contiguous axial sections and multiplanar reconstructions were performed routinely. The examinations were replayed in abdomen (W ¼ 350, L ¼ 50), pneumoperitoneum (W ¼ 500, L ¼ e50) and lung parenchymal windowing (W ¼ 1600, L ¼ e600). The images were retrospectively analysed and reviewed independently in agreement with two radiologists experienced in gastrointestinal imaging who were unaware of the patients’ clinical histories or the final diagnoses. The following CT signs from the literature were selected for the study and were recorded for each patient: wall thickness >3 mm,7 a “fat halo” sign or a “water halo” sign, a “comb” sign, an “accordion” sign, fibro-fatty proliferation, fat stranding, small bowel involvement, and the presence of abscess, fistula, ascites, or enlarged lymph nodes (short axis >10 mm). The distribution of continuous or noncontinuous colonic changes and the location of the diseased colon (right, left or diffuse) were also recorded. The “halo” sign is an aspect of the bowel wall usually represented by a stratification of the wall in two or three concentric rings of different density and enhancement. The appearance of the intermediate layer (corresponding to the submucosa) of fluid density or fat density was distinguished. The “comb” sign is a stretching and stiffening of the vessels on the mesenteric side of the bowel, leading to a combed appearance and hyperaemia of the mesocolon at imaging. The vessels are abnormally thick and tortuous. The “accordion” sign shows significant submucosal oedema, irregular thickening, pseudo-polypoid colonic wall, and a strong mucosal enhancement. Fibro-fatty proliferation corresponds to proliferation of fat and fibrous tissue in the mesentery. Fat stranding refers to intraperitoneal exudate leading to high-density of the peritoneal fat. Fistulas,2 abscesses on all bowel loops, and enlarged lymph nodes (>1 cm in short axis dimension) throughout the mesentery were searched for. Wall abnormality was considered continuous if the affected areas were not interspersed with healthy areas of colon. Distribution of colonic abnormality
264
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
was defined as right, left, or diffuse abnormality. Left distribution included lesions of the rectum, sigmoid, left colon, and the distal transverse; right distribution included lesions of the proximal transverse colon, ascending colon, and caecum. It should be noted that some types of colitis show abnormalities to the right and left without diffuse disease. Small bowel involvement was reported when an increase in small bowel wall thickness was seen, regardless of the segment. A new sign, the “empty colon” sign, which has never been described in the literature, was also specifically investigated. It was defined as a complete emptiness of the colonic lumen and resembled an empty colon without gas, fluid, or faeces.
Statistical analysis Three groups of colitis were studied and inter-group comparison performed: ischaemic colitis, inflammatory colitis, and infectious colitis. No specific analysis was conducted regarding drug-related colitis because of its low prevalence in the present study. Statistical comparisons between two groups used the chi-square test or Fisher’s exact test for categorical variables, and Student’s t-test or a ManneWhitney test for quantitative variables. The formulation of hypotheses was two-sided and a p-value under 0.05 was considered as significant. When the radiological signs appeared to be significant in the univariate analysis, their sensitivity, specificity, positive (PPV), and negative (NPV) predictive values were calculated. Multivariate analyses were performed for each group of colitis using logistic regression models. Parameters analysed though multivariate analyses were selected based on univariate analysis of the results and the absence of co-linearity. The diagnostic performance of each model was established using the area under the receiver operating characteristic (ROC) curve. Statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC, USA) and NCSS (Jerry Hintze, Visual Components, Sybase, Kaysville, UT, USA) software for Windows.
Results Study population During the study period, 1816 hospitalizations involving 1452 patients with acute colitis were recorded. Among these patients, 1186 were excluded because these patients did not undergo CT. Of the 266 remaining patients, 161 were excluded because the causal diagnosis was not clear after reviewing the medical records. The study population included 105 patients, 61 women and 44 men; average age 52 years (23e92 years). The causes of acute colitis were as follows: ischaemic colitis (n ¼ 21), inflammatory colitis (n ¼ 43, including 31 CD and 12 UC), infectious colitis (n ¼ 35), and drug-related colitis (n ¼ 6). The aetiological diagnosis was mainly based on endoscopic findings (76 patients), bacteriological and histopathological examination of biopsy samples (61 patients), stool cultures (nine patients), or positive blood culture (one patient). For
patients without bacteriological, endoscopic, or histopathological evidence, the causal diagnosis was assessed based on compatible features, suggestive clinical symptoms at baseline, and suggestive outcome after empirical therapy (n ¼ 17). The main characteristics of the study population are summarized in Table 1. Patients with inflammatory colitis were younger (39 years), especially when compared to patients with ischaemic colitis (76 years, p < 0.001); the sex ratio did not differ among the three groups of colitis; patients with ischaemic colitis usually suffered from a spectrum of suggestive comorbidities such as hypertension (p < 0.001), diabetes (p ¼ 0.028), excess weight (p ¼ 0.053), hypercholesterolemia (p ¼ 0.002), and a history of cardiovascular disease (p < 0.001). Patients with inflammatory colitis had more often a family history of IBD (p ¼ 0.018). Patients with infectious colitis more often received antibiotics prior to the onset of colitis (p < 0.001). The clinical presentation more often involved rectal bleeding (p ¼ 0.012) or anaemia (p ¼ 0.008) in the event of ischaemic colitis, abdominal pain in the event of inflammatory colitis (p ¼ 0.033), and in the event of infectious colitis, fever (p ¼ 0.004), a high leukocyte count (p ¼ 0.002), and high CRP levels (p < 0.001).
MDCT analysis results Overall, wall thickness >3 mm was present in 97% of patients, a “water halo” sign was present in 71% of patients, and conversely, a “fat halo” sign (Fig 1) was encountered in 6%. An accordion sign (Fig 2) was present in only 6%, and ascites in only 20% of patients in the present study. There was a variable distribution of colonic abnormality: 40% was in the right colon, 57% in the left colon, and 16% diffuse. These signs failed to discriminate between the three groups of acute colitis. The odds ratio (OR), specificity, sensitivity, PPV, and NPV of relevant MDCT signs for different groups of acute colitis are shown in Table 2. A “comb” sign (Fig 3) was present in 47% of IBD cases versus 3% of other acute colitis (p < 0.001), mainly associated with CD (61%) rather than UC (8%). A “comb” sign was never encountered in the event of ischaemic colitis (p ¼ 0.006) and was detected in only 6% of cases of infectious colitis (p ¼ 0.009). Fat stranding (Fig 4) was found in 71% of ischaemic colitis (p ¼ 0.001) versus 40% of other types of acute colitis. Fat stranding was detected in only 23% of cases of infectious colitis (p ¼ 0.012). Enlarged lymph nodes (Fig 5) were found in 67% of IBD (p < 0.001), versus 16% of other types of acute colitis. Enlarged lymph nodes were detected in only 23% of cases of infectious colitis (p ¼ 0.035), and they were detected in only 5% of cases of ischaemic colitis (p < 0.001). Small bowel involvement was found in 30% of IBD (p < 0.001) versus 5% of other types of acute colitis. Small bowel involvement was never encountered in the event of ischaemic colitis (p ¼ 0.037). Fibrofatty proliferation was found in 21% of IBD (p ¼ 0.007), versus 3% of other types of acute colitis. The newly described “empty colon” sign (Fig 6) was found in 23% of infectious colitis (p ¼ 0.002), versus 3% of other types of
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
265
Table 1 Demographics of the study population. Infectious colitisa
Inflammatory colitisa
Ischaemic colitisa
33
41
20
51.8 22.8 42 24 7 5 11 7 16 10
53 22.5 40 17 6 3 9 3 9 29b
39 17.5b 47 14 2 2 2 14b 2 0
76.2 11.8b 29 57b 19b 14b 33b 0 57b 0
25 65 83 29
43b 80 34 14
21 58 49b 44
5 48 17 57b
44 61 11
66b 83b 3
28 44 10
38 57 29b
Total Prevalence (%) Demographics Age Male (%) Hypertension (%) Diabetes (%) Overweight (%) Hypercholesterolaemia (%) ATCD familial IBD (%) ATCD vascular (%) Antibiotics (%) Clinical data fever (%) Diarrhoea (%) Abdominal pain (%) Rectal bleeding (%) Biological data Leukocytosis (%) High CRP (%) Anaemia (%)
ATCD, antecedent; IBD, inflammatory bowel disease; CRP, C-reactive protein. a Each group of colitis was compared with all other colitis. b Significant data, p < 0.05.
acute colitis. A “empty colon” sign was detected in only 2% of cases of IBD (p ¼ 0.045). Abscesses (Fig 4) were found in 9% of IBD (p ¼ 0.026), and were encountered in other types of acute colitis. Continuous colonic abnormality was found in 66% of infectious colitis (p ¼ 0.020) versus 41% of other types of acute colitis. Continuous colonic abnormality was
detected in only 14% of cases of ischaemic colitis (p < 0.001). The results of multivariate analysis are presented in Table 3 and Table 4. Three MDCT signs independently associated with infectious colitis were identified: the presence of the “empty colon” sign (p ¼ 0.013), the absence of fat stranding
Figure 1 A “fat halo” sign in a patient with proven ulcerative colitis. Contrast-enhanced axial MDCT image shows stratification of the caecal wall with fat-density of the intermediate layer corresponding to the submucosa (arrowheads).
Figure 2 An “accordion” sign in a patient with confirmed pseudomembranous colitis. Contrast-enhanced MDCT image with coronal reformation shows a significant thickening of the colonic wall with significant submucosal oedema and a strong mucosal enhancement.
266
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
Table 2 Odds ratio (OR), specificity (Sp), sensitivity (Se), positive (PPV) and negative (NPV) predictive values of relevant MDCT signs for different groups of colitis. Sign
Infectious colitis OR
Comb sign Fat stranding
0.15 0.31 0.37 Lymph nodes 2.71 Continuous distribution e Small bowel involvement e 12.32 Fibro-fatty proliferation e Empty colon sign Abscesses
Inflammatory colitis
Ischaemic colitis
Se (%) Sp (%) PPV (%) NPV (%) OR
Se (%) Sp (%) PPV (%) NPV (%) OR
5.7 22.9 22.9 65.7 e e 23.1 e
46.5 e 67.4 e 30.2 20.9 2.4 9.3
71.4 51.4 55.7 58.6 e e 97.1 e
9.1 19 20.5 44.2 e e 80 e
60.2 57.1 59.1 77.4 e e 71.6 e
26.1 e 10.77 e 8.52 7.94 0.14 14.24
96.8 e 83.9 e 95.2 96.8 85.5 100
90.9 e 74.4 e 81.2 81.8 10 100
72.3 e 78.8 e 66.3 e 55.8 61.4
Se (%) Sp (%) PPV (%) NPV (%)
0.06 0 5.28 71.4 0.06 4.8 0.12 14.3 0.01 0 e e e e e
73.8 67.9 54.8 41.7 80.9 e e
0 35.7 2.6 5.8 0 e e e
74.7 90.5 69.7 66 76.4 e e e
e, Not significant.
(p ¼ 0.014), and the absence of a “comb” sign (p ¼ 0.031), whereas continuous colonic abnormality and the absence of enlarged lymph nodes were not significant (Table 3). The area under the ROC curve of the logistic regression model was 0.802. Three separate/different signs independently associated with inflammatory colitis were identified: the presence of a “comb” sign (p ¼ 0.006), small bowel involvement (p ¼ 0.037), and enlarged lymph nodes (p < 0.001), whereas an “empty colon” sign, fibro-fatty proliferation, and the presence of an abscess were not significant (Table 4). The area under the ROC curve of the regression model was 0.904. Given the small number of patients in the ischaemic colitis group, no multivariate analysis was performed for this group.
Figure 3 A “comb” sign, fistula and abscess in a patient with proven Crohn’s disease. Contrast-enhanced axial MDCT image shows colonic wall thickening with a characteristic stretching and stiffening of the mesocolic vessels (arrowheads). Note the presence of an associated fistula and abscess (arrow).
Discussion Currently, MDCT is an indispensable examination in the initial evaluation of patients presenting with an acute abdomen. This includes individuals with acute colitis, where the initial diagnosis is clinically suspected in less than half of cases, as the presentation may be rather vague with atypical and/or febrile acute abdomen.6 Although there have been many previous descriptions of the classical CT findings in acute colitis, only one study has attempted to define suggestive CT signs that can provide strong clues as to the underlying aetiology in a large patient cohort.6 The strengths of the present study include its novelty, large patient cohort, and rigorous design. In the present study, multivariate analysis could identify three discriminating factors for the diagnosis of inflammatory
Figure 4 Fat stranding in a patient with confirmed ischaemia of the descending colon. Contrast-enhanced axial MDCT image shows a loss of transparency of the peritoneal fat (arrow) adjacent to the left colon.
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
267
Table 3 Comparison of multidetector CT signs of infectious colitis. Multivariate analysis
p-Value
Odds ratio (95% CI) Empty colon sign (%) Continuous distribution (%) Lymph nodes (%) Fat stranding (%) Comb sign (%)
12.32 2.1 0.33 0.26 0.17
(1.68e90.15) (0.78e5.7) (0.11e1.01) (0.09e0.77) (0.03e0.85)
0.013 0.143 0.053 0.014 0.031
Not significant at p > 0.2.
Figure 5 Enlarged lymph nodes in a patient with proven Crohn’s disease. Contrast-enhanced axial MDCT image shows an enlarged and strongly enhanced lymph node (arrowhead) in the vicinity of the right colon.
colitis, i.e., a “comb” sign, small bowel involvement, and the presence of enlarged lymph nodes; and three for the diagnosis of infectious colitis, i.e., the presence of an “empty colon” sign, continuous distribution of colonic abnormality, and absence of enlarged lymph nodes. The univariate analysis presented the following signs for ischaemic colitis: presence of fat stranding and absence of enlarged lymph nodes, “comb” sign, small bowel involvement, and continuous distribution of colonic abnormality. An increase in wall thickness, as reported in the study of Philpotts et al.,6 was found in the majority of cases and does not appear to discriminate between specific aetiologies; therefore, thickness of the colonic wall was not measured in
the present study. Thus, there may be a link between the aetiology of acute colitis and wall thickness, although this has never been clearly identified.7,8 The “fat halo” sign and the “water halo” sign were not associated with a specific aetiology of colitis. However, a “water halo” sign seems to appear when the submucosal oedema of the colonic wall is recent, whereas a “fat halo” sign may be more present in subacute or chronic colitis.9,10 The present study included only patients with acute colitis, the low percentage of “fat halo” sign found may explain this recruitment bias. A “comb” sign shows engorgement of mesenteric vessels and has been described in CD, especially when it is active.11 In the present series, the “comb” sign was found in 8% of cases of UC. This result seems surprising as UC is not transmural and does not reach the serosa. Theoretically, a “comb” sign should not be seen. However, some important aspects of fibro-fatty proliferation that are visible in UC could be interpreted as a “comb” sign, thus explaining these results. Moreover, a “comb” sign may also be viewed in other diseases such as vasculitis.12,13 Small bowel involvement is also correlated with the presence of an inflammatory disease, especially CD, but also by a UC reflux ileitis. Although small bowel involvement is reported to be specifically associated with inflammatory colitis, some infectious diseases, such as Yersinia infections, are likely to affect the small bowel and colon. Furthermore, in the present study, 9% of cases of infectious colitis showed this sign. It is interesting to find the presence of enlarged lymph nodes as a discriminating factor for inflammatory colitis and not for infectious
Figure 6 A “empty colon” sign in a patient with confirmed infectious colitis. Contrast-enhanced MDCT. (a) Coronal reformation image shows significant thickening of the transverse colon wall and a complete emptiness of the transverse lumen. (b) Idem for the left colon. (c) Axial image shows a “water halo” sign and a total emptiness of the left colon lumen.
268
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
Table 4 Comparison of MDCT signs of inflammatory colitis. Multivariate analysis
p-Value
Odds ratio (95% CI) Comb sign (%) Lymph nodes (%) Small bowel involvement (%) Fibro-fatty proliferation (%) Empty colon sign (%)
15.16 16.66 6.35 2.37 0.03
(2.19e105.1) (4.94e56.21) (1.11e36.29) (0.2e28.41) (0.01e1.19)
0.006 0.2.
colitis. However, the enlarged lymph nodes, reflecting the inflammation of the intestinal structures, should not mask the enlarged lymph nodes indicative of carcinoma. Patients with IBD are at a higher risk of developing colon or small bowel adenocarcinomas,14,15 and any lymph node >10 mm in short axis dimension should attract attention. The present study found continuous distribution of colonic abnormality in infectious colitis without predictive location. The population of infectious colitis was made up of all the different bacteria mixed together and subgroups were not used. The probability of the specific location according to the organism (such as the right colon for Yersinia or the left colon for Shigella) was therefore not observed. This study included five patients with confirmed Clostridium difficile colitis. The classic “accordion” sign was found in only one patient, and other patients with this sign were mainly those with non-infectious colitis pseudomembranous. These data seem to confirm that, as pointed out by Macari et al.,16 the “accordion” sign may be found with an organism other than Clostridium difficile or even in cases of other causes of acute colitis.17,18 In the present study, the “accordion” sign was seen in only 4e19% of cases of colitis with Clostridium difficile. Regarding infectious colitis, the multivariate model allowed the “empty colon” sign to be highlighted. The “empty colon” sign corresponds to a complete emptiness of the lumen of the colon. Whereas evidence of infectious colitis is often difficult to obtain (the causative agent is found in only 50% of cases), and it is often the clinical outcome after the initiation of antibiotic treatment that will confirm the diagnosis, the presence of an “empty colon” sign at CT, with a specificity of 0.97, seems a good indicator of this aetiology. The localization of ischaemic colitis in the present study was predominantly on the left (67%) as in the study of Philpotts et al. (77%).6 However, there was no statistically significant difference for that location when ischaemic colitis was compared to colitis of all other causes. According to the literature, diffuse colitis was present in 10% of cases, illustrating the fact that ischaemic colitis is usually segmental.19 It is also interesting to note that arterial or venous vascular abnormalities that could explain the ischaemic phenomena were not found in the present study. In most cases, these anomalies are not present; reperfusion was restored in most cases when MDCT was performed. Only one set of vascular abnormalities was found in up to 25% of patients.20 The present study has several limitations. Firstly, there was no control group of patients without colitis. Such a group would have been useful for assessing the specificity
of the MDCT signs described with more accuracy. However, the present study was pragmatic and only focused on patients with acute colitis. Secondly, the study was monocentric and retrospective, thus a substantial number of cases with an uncertain aetiology were excluded. Such procedures may compromise the applicability of the present results because of the small number of patients in each group. However, although “intent-to diagnose” results were not provided, well-documented cases were carefully selected regarding causal diagnosis for the accurate assessment of MDCT performance in characterizing acute colitis even if a pathological confirmation was not always obtained. Further studies, which would ideally be prospective and include protocol endoscopy, bacteriology, and colonic biopsy samples, should be conducted to better investigate uncertain cases and enable external validation. The present study only studied patients hospitalized in a gastroenterology department, i.e., those who did not require immediate surgical treatment. This may be considered as a selection bias, and may thus have decreased the prevalence of serious forms of acute colitis in the present series, especially ischaemic gangrenous colitis. However, these forms often have typical clinical presentations and are easy to diagnose at MDCT, which shows a complete lack of mucosal enhancement, corresponding to a transmural infarction or the presence of intestinal pneumatosis. In conclusion, MDCT provides certain suggestive signs that may be helpful in distinguishing the underlying aetiological cause of acute colitis, i.e., the presence of lymph nodes, comb sign, small bowel involvement, and fibro-fatty proliferation for inflammatory colitis; the presence of the “empty colon” sign and continuous distribution for infectious colitis; and the presence of fat stranding for ischaemic colitis. A larger prospective study should be conducted to confirm these results, including patients who are candidates for surgery. This further study would also validate the value of the “empty colon” sign for identifying infectious colitis.
References 1. Balthazar EJ. CT of the gastrointestinal tract: principles and interpretation. AJR Am J Roentgenol 1991;156:23e32. 2. Horton KM, Corl FM, Fishman EK. CT evaluation of the colon: inflammatory disease. RadioGraphics 2000;20:399e418. 3. Johnson GL, Johnson PT, Fishman EK. CT evaluation of the acute abdomen: bowel pathology spectrum of disease. Crit Rev Diagn Imaging 1996;37:163e90. 4. Thoeni RF, Cello JP. CT imaging of colitis. Radiology 2006;240:623e38. 5. Delabrousse E, Ferreira FR, Badet N, et al. Coping with the problems of diagnosis of acute colitis. Diagn Interv Imaging 2013;94:793e804. 6. Philpotts LE, Heiken JP, Westcott MA, et al. Colitis: use of CT findings in differential diagnosis. Radiology 1994;190:445e9. 7. Fisher JK. Normal colon wall thickness on CT. Radiology 1982;145:415e8. 8. Macari M, Balthazar EJ. CT of bowel wall thickening: significance and pitfalls of interpretation. AJR Am J Roentgenol 2001;176:1105e16. 9. Jones B, Fishman EK, Hamilton SR, et al. Submucosal accumulation of fat in inflammatory bowel disease: CTepathologic correlation. J Comput Assist Tomogr 1986;10:759e63. 10. Wittenberg J, Harisinghani MG, Jhaveri K, et al. Algorithmic approach to CT diagnosis of the abnormal bowel wall. RadioGraphics 2002;22:1093e107. discussion 1107e1109. 11. Lee SS, Ha HK, Yang S-K, et al. CT of prominent pericolic or perienteric vasculature in patients with Crohn’s disease: correlation with clinical
L. Plastaras et al. / Clinical Radiology 70 (2015) 262e269
12. 13.
14. 15.
disease activity and findings on barium studies. AJR Am J Roentgenol 2002;179:1029e36. Madureira AJ. The “comb” sign. Radiology 2004;230:783e4. Byun JY, Ha HK, Yu SY, et al. CT features of systemic lupus erythematosus in patients with acute abdominal pain: emphasis on ischemic bowel disease. Radiology 1999;211:203e9. Eaden JA, Abrams KR, Mayberry JF. The risk of colorectal cancer in ulcerative colitis: a meta-analysis. Gut 2001;48:526e35. Jess T, Gamborg M, Matzen P, et al. Increased risk of intestinal cancer in Crohn’s disease: a meta-analysis of population-based cohort studies. Am J Gastroenterol 2005;100:2724e9.
269
16. Macari M, Balthazar EJ, Megibow AJ. The accordion sign at CT: a nonspecific finding in patients with colonic edema. Radiology 1999;211:743e6. 17. Fishman EK, Kavuru M, Jones B, et al. Pseudomembranous colitis: CT evaluation of 26 cases. Radiology 1991;180:57e60. 18. Ash L, Baker ME, O’Malley Jr CM, et al. Colonic abnormalities on CT in adult hospitalized patients with Clostridium difficile colitis: prevalence and significance of findings. AJR Am J Roentgenol 2006;186:1393e400. 19. Balthazar EJ, Yen BC, Gordon RB. Ischemic colitis: CT evaluation of 54 cases. Radiology 1999;211:381e8. 20. Guttormson NL, Bubrick MP. Mortality from ischemic colitis. Dis Colon Rectum 1989;32:469e72.
