| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Both authors: Department of Radiology, University of Manitoba, Health Sciences Centre, 820 Sherbrook St., Winnipeg, Manitoba R3A 1R9, Canada.
Received April 7, 2000;
accepted after revision August 24, 2000.
Address correspondence to I. D. C. Kirkpatrick.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. A retrospective review covering a 4-year period was performed of the charts and CT scans of 54 symptomatic patients with stool test results positive for C. difficile and of a control group of 56 patients with antibiotic-associated diarrhea with stool test results negative for C. difficile.
RESULTS. At our institution, C. difficile colitis was explicitly diagnosed at CT in these patients with a sensitivity of 52%, specificity of 93%, positive predictive value of 88%, and negative predictive value of 67%. The sensitivity can be raised to 70% with no change in specificity with more rigid adherence to diagnostic criteria of colon wall thickening of greater than 4 mm combined with any one or more findings of pericolonic stranding, colon wall nodularity, the "accordion" sign, or otherwise unexplained ascites.
CONCLUSION. Although routine CT screening of antibiotic-associated diarrhea is not advocated, the 88% positive predictive value of a diagnosis of C. difficile colitis in those who are scanned may merit consideration of treatment by clinicians on the basis of the CT results alone.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The gold standard for diagnosis is a microbiologic stool assay for C. difficile cytotoxin [1, 3]. Although this test has a high sensitivity and specificity, results are generally not available for 48 hr or longer. Other faster microbiologic assays are available but are not as sensitive [1, 3]. Endoscopy with biopsy has been used to make the diagnosis, but this method is also less sensitive, and mechanical manipulation of a diseased colon puts the patient at risk [2, 3].
CT has been increasingly studied as a non-invasive fast method of making the diagnosis of C. difficile colitis. Several recent articles have described CT findings in patients testing positive for C. difficile colitis on stool assay [4,5,6,7,8,9]. These CT findings include colonic wall thickening and nodularity, pericolonic stranding and edema, ascites, and the "accordion" sign, which describes the appearance of oral contrast material trapped between nodular thickened folds of bowel. In the largest study to date, Boland et al. [7] examined 64 patients and found that 85% showed some colonic abnormality on CT. Unfortunately, the most commonly seen abnormality, colonic wall thickening, is nonspecific and can be found in other colitides, although it is thought to be more pronounced in C. difficile colitis than in other causes of colitis [10]. The CT sensitivity and specificity for explicitly diagnosing C. difficile colitis have never been determined, to our knowledge.
The purpose of our study was to evaluate for the first time how successfully the diagnosis of C. difficile colitis could be made with CT. We retrospectively reviewed the CT examinations of 54 patients with C. difficile disease and 56 control patients with C. difficile—negative antibiotic-associated diarrhea.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We retrospectively reviewed all charts, CT reports from the original time of scanning, and CT scans. We were not blinded to the clinical history or stool assay findings. Scans were examined for evidence of colonic wall thickening (defined as a wall thickness measured with calipers of greater than 4 mm), wall nodularity, pericolonic stranding, ascites (defined as intraperitoneal free fluid collecting in the dependent abdomen), and the accordion sign. The quantitative measurement of wall thickness, distribution of disease throughout the colon, and any coexistent lesions were all noted. At our center, the diagnosis of C. difficile colitis on CT was made at the time of scanning on the basis of a colon wall thickness of greater than 4 mm, a compatible clinical history, and one or more of the following findings: colon wall nodularity, pericolonic stranding, or otherwise unexplained ascites. Equivocal cases, in which the reporting radiologist thought the findings may be present but too subtle to explicitly diagnose C. difficile colitis with confidence, received the diagnosis of a "nonspecific colitis."
For the purposes of determining sensitivity, specificity, positive predictive value, and negative predictive value of the CT diagnosis of C. difficile at our institution, results used were the diagnoses made at the original time the scans were obtained, as recorded in the patients' charts by radiologists blinded to the patient's C. difficile status. Results for the distribution and diagnostic value of CT findings, the findings themselves, and colon wall measurements in both groups were those determined at the time of the retrospective review.
CT scans were obtained with one of two helical CT scanners (Somotom-R; Siemens Medical Systems, Iselin, NJ). Our standard axial CT protocol used either an 8- or 10-mm slice thickness and intervals from the diaphragm to the symphysis pubis. All patients were given oral contrast material, but only 74% of the C. difficile—positive group and 79% of the control group received IV contrast material. Consequently, no comparisons of wall enhancement or halo effect were made.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Ten patients (19%) from the group with a positive stool assay and one patient (2%) from the negative control group were diagnosed with nonspecific colitis because CT findings were thought to be insufficient to make a more precise diagnosis at the time of scanning. The remaining patients in both groups were either diagnosed with a specific colonic disease other than C. difficile colitis, or the colon was thought to be unremarkable.
At review of the CT scans, the distribution of the findings of colon wall thickening (Fig. 1), colon wall nodularity, the accordion sign (Fig. 2A,2B), pericolonic stranding, and ascites for the two groups was determined and is shown in Table 1. Of the patients in the assay-positive group with colon wall thickening, the mean thickness was 11 mm (range, 5-25 mm).
|
|
|
|
For patients whose scans showed colon wall thickening, the sensitivity and specificity of each of the aforementioned signs and their combinations for the explicit diagnosis of C. difficile colitis are given in Table 2, as determined at the time of the retrospective review. Given the consensus that wall thickening is often more pronounced in patients with C. difficile colitis than in patients with other colitides, values were calculated first for all patients with colon wall measurements of greater than 4 mm and then for those with a thickness of 10 mm or greater.
|
In the patients with positive stool assay tests who had colon wall thickening, the distribution of the thickening was pancolonic in 17 (41%). Involvement of only the right colon (i.e., no changes beyond the mid transverse colon) was seen in four (10%) of the patients, and left-sided involvement only was seen in 10 (24%). Findings were limited to the cecum in two patients (5%) and to the rectum in two patients (5%). Rectal sparing was seen in 13 patients (32%) and rectosigmoid sparing in nine (22%).
The final diagnoses of the four patients with negative stool assays who were diagnosed at the time of scanning with C. difficile colitis were reviewed. Two patients underwent lower endoscopy with biopsy, which revealed cytomegalovirus colitis in one and ulcerative colitis in the other (Fig. 3). One patient was on multiple chemotherapeutic agents and IV antibiotics and was diagnosed clinically with a nonspecific drug—induced diarrhea. The fourth patient, who had abdominal pain, fever, and diarrhea, never received a formal diagnosis and his symptoms resolved on antibiotic therapy.
|
Of the other control patients with antibiotic-associated diarrhea, CT abnormalities of the colon included ischemic colitis, neutropenic enterocolitis, colonic lymphoma, appendicitis, Crohn's disease, and colonic carcinoma. All cases were successfully diagnosed on the basis of clinical history and CT findings, and none was mistaken for C. difficile colitis.
The use of IV contrast material caused no significant difference in CT sensitivity, with the diagnosis of C. difficile colitis correctly made at the time of scanning in 20 (50%) of the 40 patients who received IV contrast material and eight (57%) of the 14 who did not (p > 0.50). No significant differences in CT sensitivity could likewise be seen on the basis of the time of CT in relation to the time of stool assay testing. Of C. difficile—positive patients scanned within 3 days of stool testing, 23 (51%) of 45 were correctly diagnosed, compared with three (60%) of five scanned more than 3 days before stool testing (p > 0.50), and two (50%) of four scanned more than 3 days after stool testing (p > 0.50).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Other studies have suggested that findings on abdominal CT scans may be absent in a significant number of patients [6, 7]. Fishman et al. [6] found that three (12%) of 26 patients with the disease showed no colon wall abnormality. Furthermore, Boland et al. [7] found that 25 (39%) of 64 patients with the disease showed no colonic abnormality.
Our study found that at our center the sensitivity of CT for making the diagnosis of C. difficile colitis is 52%. Of the patients with positive findings for C. difficile, 13 (24%) showed no colonic abnormality on CT. Even in those who did show abnormalities, the findings were not always indicative of C. difficile colitis in particular. Clearly, this data suggests that CT would not be a test of choice for screening patients for the disease.
That having been said, the specificity of CT was shown to be 93% by our study. Perhaps of more interest to clinicians is the positive predictive value of 88%. In other words, our study found that a patient who received a diagnosis of C. difficile colitis on the basis of a CT scan had an 88% probability of testing positive on a stool assay. We believe that given this finding, it would be reasonable to consider initiating treatment based on the CT results alone, while a stool assay for confirmation is pending. In our chart review, only one case could be found in which antibiotic therapy was initiated on the basis of the CT diagnosis.
In examining the individual CT findings of the disease, our findings are similar to those described in previous studies with some exceptions [5,6,7]. The accordion sign was found in only 4% of our patients, compared with 19% in the study by Fishman et al. [6]. However, Boland et al. [7] found the sign in 5% of their patients, which resembles our findings. Ascites was also seen in a significantly higher number of our patients (52%) than those in the study by Boland et al., in which it was seen in only 17% of the patients. One contributing factor to this discrepancy is that six of the 28 C. difficile—positive patients in our study who showed ascites had other findings seen on CT that could explain the free fluid: cirrhosis (two patients), pancreatitis (two patients), and metastatic ovarian carcinoma (two patients). Five of these patients had no other findings to suggest C. difficile colitis and represented false-negative diagnoses. Likewise, six patients in the control group who had ascites had clear coexistent diseases shown on CT that could explain this finding. Even taking these factors into account, the percentage of patients with C. difficile colitis who had ascites remains significantly higher (41%). Other possible explanations could include the imaging of patients in our study at a more advanced point in the disease, on average, or perhaps patients who are more prone to ascites to begin with—be it caused by malnutrition or chronic disease (many patients imaged in our study were sent from either the intensive care unit or the oncology ward).
Our study is similar to that performed by Boland et al. [7] in that although colon wall thickening was the most sensitive abnormality in C. difficile colitis, other findings such as wall nodularity, the accordion sign, and ascites were more specific. Wall thickening is the key CT finding in this disease, and, except for the five patients who showed ascites explained by other coexistent diseases on their scans, all the patients in the group with positive assay results showing any abnormality suggestive of C. difficile had thickened colon walls. As Table 2 shows, once colon wall thickening is identified, the ancillary findings of pericolonic stranding, ascites, and colon wall nodularity (including the accordion sign) increase the specificity of the diagnosis for C. difficile, with additive effects. A wall thickness of 10 mm or greater is by itself a specific finding for C. difficile colitis, but not a sensitive one. The combination of wall thickening of greater than 4 mm and any one or more of the ancillary findings described in Table 2 provides criteria that are still specific but that have a significantly increased sensitivity. At our center, the criteria used for diagnosis by our radiologists (based on prior studies) over the last 4 years have been a colon wall thickness of greater than 4 mm in conjunction with any findings of wall nodularity, the accordion sign, pericolonic stranding, or unexplained ascites [6, 7, 10]. In fact, on review these criteria were not always used. If these criteria had been rigidly applied, 10 patients initially diagnosed with nonspecific colitis would have been correctly labeled by CT as having C. difficile colitis. The sensitivity and specificity in this case would have, in fact, been 70% and 93%, respectively, with a positive predictive value of 90% and negative predictive value of 76%. Using the criteria of either a wall thickness of 10 mm or greater or a wall thickness of greater than 4 mm and any of the more specific findings does not add significantly to the diagnosis based on our data (sensitivity, 72%; specificity, 91%) but gives equally satisfactory results.
Of the 16 patients with positive stool assays who were not initially diagnosed with either C. difficile or nonspecific colitis, three patients showed colon wall thickening only, which was missed at the original time of scanning. The remaining 13 patients showed none of the previously described findings on their scans. The most likely explanation is that there is a spectrum of findings in this disease, and these patients were imaged before the disease was significant enough to show radiographic changes. Unlike CT, a stool assay for toxin is not as dependent on the severity and length of illness [1].
There are a number of limitations to our study. Patients imaged were predominantly those who showed more marked clinical symptoms, including abdominal pain, fever, and signs of sepsis. Many patients with diarrhea alone did not undergo diagnostic imaging. The sensitivity shown in our study is, therefore, likely more representative of testing in patients with more advanced disease. Results of our study are also representative of the radiologists in our CT department and could, thus, be expected to vary from center to center. This is compounded by the lack of clear CT diagnostic criteria for C. difficile colitis in the literature to date. Finally, we are using the stool cytotoxin assay as the microbiologic gold standard, and no allowance was made for microbiologic misdiagnosis.
In conclusion, we found that at our center the CT diagnosis of C. difficile colitis was made with a sensitivity of 52%, specificity of 93%, positive predictive value of 88%, and negative predictive value of 67%. Although the sensitivity is not satisfactory for screening purposes, the positive predictive value is impressive. Sensitivity could have been improved—to 70%—with no change in specificity by adhering strictly to diagnostic guidelines of colon wall thickness of greater than 4 mm in conjunction with any one or more findings of colon wall nodularity, the accordion sign, pericolonic stranding, or otherwise unexplained ascites. We suggest that patients with antibiotic-associated diarrhea who have CT findings diagnostic of C. difficile colitis merit consideration for treatment on that basis.
Acknowledgments
We thank John Embil and Debbie Nichol of the Infection Control Department
of the Winnipeg Health Sciences Centre for their assistance in providing stool
assay data.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. F. Thoeni and J. P. Cello CT Imaging of Colitis Radiology, September 1, 2006; 240(3): 623 - 638. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Ash, M. E. Baker, C. M. O'Malley Jr., S. M. Gordon, C. P. Delaney, and N. A. Obuchowski Colonic Abnormalities on CT in Adult Hospitalized Patients with Clostridium difficile Colitis: Prevalence and Significance of Findings. Am. J. Roentgenol., May 1, 2006; 186(5): 1393 - 1400. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Image of the Month--Diagnosis Arch Surg, May 1, 2004; 139(5): 566 - 566. [Full Text] [PDF]
|
|
|
|
|
|
I. D. C. Kirkpatrick and H. M. Greenberg Gastrointestinal Complications in the Neutropenic Patient: Characterization and Differentiation with Abdominal CT Radiology, March 1, 2003; 226(3): 668 - 674. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
