Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online before print February 21, 2008, 10.1148/radiol.2471070439
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2471070439v1
247/1/122    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jensch, S.
Right arrow Articles by Stoker, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Jensch, S.
Right arrow Articles by Stoker, J.
(Radiology 2008;247:122-132.)
© RSNA, 2008

Gastrointestinal Imaging

CT Colonography with Limited Bowel Preparation: Performance Characteristics in an Increased-Risk Population1

Sebastiaan Jensch, MD, Ayso H. de Vries, MD, Jan Peringa, MD, Shandra Bipat, PhD, Evelien Dekker, MD, PhD, Lubbertus C. Baak, MD, PhD, Joep F. Bartelsman, MD, PhD, Anneke Heutinck, RN, Alexander D. Montauban van Swijndregt, MD, PhD, and Jaap Stoker, MD, PhD

1 From the Departments of Radiology (S.J., A.H.d.V., S.B., A.H., J.S.) and Gastroenterology (E.D., J.F.B.), Academic Medical Center, University of Amsterdam, Meibergdreef 9, 1100 DD, Amsterdam, the Netherlands; and Departments of Radiology (S.J., J.P., A.D.M.v.S.) and Gastroenterology (L.C.B.), Onze Lieve Vrouwe Gasthuis, Amsterdam, the Netherlands. Received March 7, 2007; revision requested May 16; revision received August 4; accepted August 28; final version accepted October 4. Supported by grant CKTO 2003-02 from the Dutch Cancer Society. Address correspondence to S.J. (e-mail: s.jensch{at}amc.uva.nl).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 IMPLICATION FOR PATIENT CARE
 References
 
Purpose: To prospectively evaluate the sensitivity and specificity of computed tomographic (CT) colonography with limited bowel preparation for the depiction of colonic polyps, by using colonoscopy as the reference standard.

Materials and Methods: Institutional review board approval and written informed consent were obtained. Patients at increased risk for colorectal cancer underwent CT colonography after fecal tagging, which consisted of 80 mL of barium sulfate and 180 mL of diatrizoate meglumine. Bisacodyl was added for stool softening. A radiologist and a research fellow evaluated all data independently by using a primary two-dimensional approach. Discrepant findings for lesions 6 mm or larger in diameter were solved with consensus. Segmental unblinding was performed. Per-patient sensitivity and specificity, per-polyp sensitivity, and number of false-positive findings were found (for lesions ≥ 6 mm and ≥ 10 mm in diameter). Per-patient sensitivities (blinded colonoscopy vs CT colonography) were tested for significance with McNemar statistics. Interobserver variability was analyzed per segment (prevalence-adjusted bias-adjusted {kappa} values [{kappa}p]).

Results: One hundred fourteen of 168 patients (105 men, 63 women; mean age, 56 years) had polyps, with 56 polyps 6 mm or larger and 17 polyps 10 mm or larger. Per-patient sensitivities were not significantly different for CT colonography (consensus reading) and colonoscopy (P ≥ .070). Sensitivity of CT colonography for patients with lesions 6 mm or larger and 10 mm or larger was 76% and 82%, respectively, and specificity of CT colonography was 79% and 97%, respectively. Blinded colonoscopy depicted 91% (lesions ≥ 6 mm) and 88% (lesions ≥ 10 mm) of disease in patients. Per-polyp sensitivity for CT colonography was 70% (lesions ≥ 6 mm) and 82% (lesions ≥ 10 mm). Number of false-positive findings was 42 (lesions ≥ 6 mm) and six (lesions ≥ 10 mm). {kappa}p Was 0.88 (lesions ≥ 6 mm) and 0.96 (lesions ≥ 10 mm).

Conclusion: CT colonography with limited bowel preparation has a sensitivity of 82% and specificity of 97% for patients with polyps 10 mm or larger.

© RSNA, 2008


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 IMPLICATION FOR PATIENT CARE
 References
 
As with other colonic examinations, computed tomographic (CT) colonography requires a clean colon for optimal assessment of the bowel wall (1–5). However, the necessary cathartic bowel preparation is often described by patients as the most burdensome part of colonic examinations (6–9). This might negatively affect patients' willingness to participate in a screening program. In 2001, limited bowel preparation for CT colonography with iodine- or barium-based contrast material for fecal tagging was introduced (10). With fecal tagging, any fecal material in the colon is labeled so that colorectal polyps or cancer can be distinguished from fecal material. Because no extensive bowel cleansing is necessary, this approach might increase patient compliance in a screening setting (11–13).

Feasibility studies with limited bowel preparation have revealed promising results (14–18). In addition, results of one large diagnostic accuracy study by Iannaccone et al (19) showed a sensitivity of 90% (71 of 79) and specificity of 92% (114 of 124) for patients with polyps, regardless of polyp size. These results are similar or superior to those of accuracy studies in which a cathartic bowel preparation was used (1–5).

To our knowledge, no confirmatory study on limited bowel preparation has been published since that of Iannaccone et al (19). Therefore, the aim of our study was to prospectively evaluate the sensitivity and specificity of CT colonography with limited bowel preparation for the depiction of colonic polyps, by using colonoscopy as the reference standard.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 IMPLICATION FOR PATIENT CARE
 References
 
This study received grant support from the Dutch Cancer Society (Koningin Wilhelmina Fonds) (No. CKTO 2003-02). The authors retained control of all aspects of the study.

Patients
Patients with a personal or family history of colorectal polyps or cancer were invited to participate from January 10, 2004, until November 10, 2005. All patients were scheduled to undergo routine colonoscopy at the endoscopy department of the Academic Medical Center or the Onze Lieve Vrouwe Gasthuis. Exclusion criteria were age younger than 18 years, a personal history of inflammatory bowel disease or familial adenomatous polyposis, prior allergic reaction to an iodine-containing contrast agent, known colorectal polyps that were not removed at prior endoscopy, or participation in a research project that involved ionizing radiation within 12 months preceding CT colonographic examination. The institutional review boards of both hospitals approved the study. All patients gave written informed consent. Information about radiation associated with the examination was given to all patients before written consent was obtained.

Diagnostic Procedures
Bowel preparation.—Patients were instructed not to eat high-fiber foods for 2 days before the examination. A combination of 80 mL of barium sulfate suspension (Tagitol V; E-Z-Em, Westbury, NY) and 180 mL of diatrizoate meglumine (200 milligrams of iodine per milliliter, prepared by hospital pharmacy, Academic Medical Center) was prescribed for fecal tagging (Table 1). Bisacodyl (30 mg, prepared by hospital pharmacy, Academic Medical Center) was given to soften fecal material for optimal contrast material and feces mixing and to avoid solid, sticky stool.


View this table:
[in this window]
[in a new window]

  		Table 1. Bowel Preparation for CT Colonography

 
CT colonographic technique.—All examinations were performed at the Academic Medical Center with a CT scanner (Mx8000; Philips Medical Systems, Best, the Netherlands) 1–4 weeks prior to colonoscopy (mean, 25 days; median, 25 days). Scanning parameters were as follows: tube voltage, 120 kV; four detector rows; 2.5-mm collimation; rotation time, 0.75 second; pitch, 1.25; section thickness, 3.2 mm; and reconstruction interval, 1.6 mm. All patients were scanned in the supine and prone positions; average-size patients (abdominal circumference, <103 cm) were scanned with 50 mAs, and larger patients were scanned with 70 mAs. Butylscopolamine bromide (20 mg, Buscopan; Boehringer-Ingelheim, Ingelheim, Germany) or, if this was contraindicated, glucagon hydrochloride (1 mg, Glucagon; Novo-Nordisk, Bagsvaerd, Denmark) was given intravenously immediately prior to scanning. An automatic insufflator (ProtocO2l; E-Z-Em) was used to distend the colon with carbon dioxide. Scanner-room examination time for patients and the amount of insufflated carbon dioxide were recorded by a research nurse (A.H.).

CT colonographic image analysis.—CT colonographic data were evaluated for colorectal polyps and carcinoma independently by two observers who were blinded to clinical data. Observer 1 (J.P.) had 5 years of experience as an abdominal radiologist and had received training consisting of 50 cathartic untagged and 20 limited preparation tagged CT colonographic cases with colonoscopic verification. Observer 2 (A.H.d.V.) was a research fellow who had evaluated 100 cathartic untagged and 20 limited preparation tagged CT colonographic cases with colonoscopic verification as part of his fellowship in CT colonography. Evaluation of data was performed with a dedicated CT colonographic workstation (Viewforum; Philips Medical Systems) by using a primary two-dimensional evaluation technique, with instant onscreen correlation with multiplanar reformation images and three-dimensional endoluminal views for problem solving. Standard window settings were applied (window width, 1250 HU; window level, –50 HU), but observers were free to adjust these settings. Stool subtraction software was not used, and routine endoluminal fly-through was not performed.

The observers classified CT colonographic polyps with regard to size, morphology (sessile, pedunculated, or flat [ie, height < 3 mm]), and location (cecum, ascending colon, transverse colon, descending colon, or sigmoid colon and rectum) by using the same classification as our gastroenterologists. Consensus reading, or a double-read strategy, was performed in case one observer found a lesion 6 mm or larger in diameter but the other did not or if both observers found the same lesion but disagreed on morphology, size, or location. Agreement forms were completed after each examination for polyps 6 mm or larger in diameter and included two-dimensional and three-dimensional images of the polyp at CT colonography, with the location marked in the three-dimensionally rendered colon. The research nurse (A.H.) used these agreement forms to provide feedback for segmental unblinding during colonoscopy. Interpretation times, with the exclusion of reporting time and evaluation of extracolonic findings, were recorded for each examination by both observers.

CT colonographic stool-tagging analysis.—A questionnaire was filled out by observer 1 to subjectively assess the adequacy of stool tagging for every examination. A four-point scale was used in which a score of 1 indicated good tagging (homogeneous, diagnostic for all lesions); a score of 2, adequate tagging (fairly homogeneous and diagnostic for all lesions); a score of 3, poor tagging with untagged solid stool less than 6 mm (diagnostic for lesions ≥ 6 mm); and a score of 4, poor tagging with untagged solid stool 6 mm or larger (not diagnostic for lesions ≥ 6 mm).

Colonoscopy.—An experienced staff member (a gastroenterologist [n = 15] or a gastrointestinal surgeon [n = 4] with an average experience of 11 years; range, 1–26 years) or a gastroenterology fellow with direct supervision (including E.D., J.F.B., L.C.B.) performed the colonoscopic examinations with a standard colonoscope (CF-140L; Olympus, Tokyo, Japan). Patients received 2.0–10.0 mg midazolam (Dormicum; Roche, Basel, Switzerland) and 0.05–0.1 mg fentanyl (Fentanyl-Janssen; Janssen Pharmaceuticals, Beerse, Belgium) on request. The examination was recorded on videotape. Segmental location, morphology, and size of polyps were documented on a case-record form by the attending research nurse (A.H.). Polyp size was measured with open biopsy forceps (8 mm in length). Segmental unblinding was performed only for lesions 6 mm or larger that both observers agreed on (independently or through consensus). Histologic findings were obtained at colonoscopy, except in those cases in which it was technically not possible or when material was lost in the colonic lumen during the procedure.

Determination of lesion status.—Colonoscopy after unblinding of the CT colonographic findings served as a reference standard. For CT colonography, a polyp was considered true-positive if (a) the segment or adjacent segment corresponded with the reference standard segment and (b) the polyp size as estimated by the endoscopist corresponded with size as measured at CT colonography, considering a margin of error of 50%. Polyps 10 mm or larger at colonoscopy that were not identified by either observer were reevaluated with knowledge of the colonoscopic findings by an independent radiologist with experience in the evaluation of more than 1000 CT colonographic studies. This reevaluation was performed to differentiate between technical errors (polyps obscured by technical failure [eg, inhomogeneous tagging]), perception errors (polyps visible in retrospect), and occult lesions (polyps not visible in retrospect). This classification was also used by MacCarty et al (20). Furthermore, a research fellow in CT colonography (S.J.) who was not involved in evaluation of the CT colonographic assessed the nature of false-positive findings 10 mm or larger (eg, untagged stool or haustral folds).

Patient experience and preference.—To evaluate patient experience, we asked patients to fill out a questionnaire on the day of the CT colonographic examination. Patients were asked if they had experienced diarrhea as a side effect and graded it on a five-point scale (none, mild, moderate, severe, extreme, with 0 indicating none and 5 indicating extreme). Furthermore, the overall burden of the complete preparation was evaluated on the same scale. Patient preference was assessed 5 weeks after colonoscopy. This, like adverse reactions to tests, tends to temper with time (7,8), and the attitude at that time point will better reflect the attitude toward future screening. Patients were asked if they had experienced the bowel preparation for CT colonography or that for colonoscopy as most burdensome and if they would prefer CT colonography or colonoscopy in the future as the colonic examination of choice.

Statistical Analysis
Outcome parameters.—Observers were instructed before the start of the study that only polyps at CT colonography that were matched to true polyps (ie, adenomatous, hyperplastic, or hamartomatous polyps) on the basis of the histologic report or, if histologic findings were not acquired, on the basis of the endoscopic report would be considered true-positive findings. All other lesions at CT colonography that were matched to polypoid lesions at colonoscopy that have no malignant potential (eg, lipomas or pseudopolyps) would be classified as false-positive findings. For CT colonography, a patient was considered to have a true-positive finding if CT colonography depicted at least one polyp seen at colonoscopy, on the basis of the location and size criteria described previously. A patient was considered to have a false-negative finding if CT colonography depicted no polyps or only those of a lower size category in comparison to those depicted with the reference standard.

Sensitivity, specificity, and positive and negative predictive values were calculated for CT colonography on a per-patient basis and were stratified according to polyp size (all sizes, ≥6 mm, ≥7 mm, ≥8 mm, ≥9 mm, and ≥10 mm, as well for the size range 6–9 mm). Per-patient sensitivity was calculated for blinded colonoscopy. For calculation of the per-polyp sensitivity for CT colonography and blinded colonoscopy, generalized estimating equations (GEE) were used to revise the data clustering and dependency (21) because some patients had more than one polyp. For CT colonography, the number of false-positive findings was calculated on a per-lesion basis. For the per-polyp sensitivity and false-positive findings, data were also stratified according to polyp size (all sizes, ≥6 mm, ≥7 mm, ≥8 mm, ≥9 mm, and ≥10 mm, as well for the size range 6–9 mm).

Diagnostic value.—We used the McNemar test to compare per-patient sensitivity values between blinded conventional colonoscopy, consensus reading, and observer 1 and observer 2 (colonoscopy vs consensus reading, colonoscopy vs observer 1, colonoscopy vs observer 2, consensus reading vs observer 1, consensus reading vs observer 2, and observer 1 vs observer 2). We also used the McNemar test to compare per-patient specificity values between consensus reading and both observers (consensus reading vs observer 1, consensus reading vs observer 2, and observer 1 vs observer 2). With the GEE model, we took into account the comparison of per-polyp sensitivities between blinded colonoscopy, consensus reading, and both observers (colonoscopy vs consensus reading, colonoscopy vs observer 1, colonoscopy vs observer 2, consensus reading vs observer 1, consensus reading vs observer 2, and observer 1 vs observer 2).

Interobserver variability.—Per-segment agreement and prevalence-adjusted bias-adjusted {kappa} values ({kappa}p values) were calculated (22,23) for lesions 6 mm or larger and lesions 10 mm or larger. Observers were considered to agree if both recorded corresponding lesion(s) in the same segment or if both recorded no findings. Values were interpreted as those for the regular Cohen {kappa} statistic: {kappa}p ≤ 0.20 indicated poor agreement; {kappa}p = 0.21–0.40, fair agreement; {kappa}p = 0.41–0.60, moderate agreement; {kappa}p = 0.61–0.80, good agreement; and {kappa}p = 0.81–1.00, very good agreement.

Interpretation time.—Mean interpretation times and 95% confidence intervals are provided. Differences in evaluation time between the observers were tested for significance by using the paired Student t test.

GEE analyses were performed with software (SPSS 15 for Windows; SPSS, Chicago, Ill). All other analyses (per-patient sensitivity, per-patient specificity, and interpretation time) were performed with software (SPSS for Windows, version 12.0.2; SPSS). P values less than .05 were considered to indicate statistically significant differences.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 ADVANCES IN KNOWLEDGE
 IMPLICATION FOR PATIENT CARE
 References
 
Patients
Of 468 eligible patients who were scheduled to undergo colonoscopy during the inclusion period, 180 participated in the study. Twelve (7%) of 180 patients were excluded for different reasons, which left 168 patients for analysis (Fig 1, Table 2).


Figure 1
View larger version (23K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1: Flowchart shows patient participation. AMC = Academic Medical Center, CTC = CT colonography, OLVG = Onze Lieve Vrouwe Gasthuis.

 

View this table:
[in this window]
[in a new window]

  		Table 2. Baseline Patient Characteristics

 
Colonoscopy
Blinded and unblinded colonoscopic results revealed that 45 (27%) of 168 patients had a polyp 6 mm or larger and 17 (10%) patients had a polyp 10 mm or larger. Table 3 shows results of histologic and morphologic characterization, subdivided for lesions 6–9 mm and 10 mm or larger. No colorectal carcinomas were found. Blinded colonoscopy revealed a lesion 6 mm or larger in 41 (91%) of 45 patients and a polyp 10 mm or larger in 15 (88%) of 17 patients (Table 4).


View this table:
[in this window]
[in a new window]

  		Table 3. Histologic Findings and Morphology of Polyps at Colonoscopy

 

View this table:
[in this window]
[in a new window]

  		Table 4. Per-Patient Sensitivity and Specificity according to Polyp Size Group

 
CT Colonography
The mean scanner-room examination time for patients was 21 minutes (range, 13–48 minutes). Buscopan was administered in 144 patients, glucagon was administered in 21, and three patients received no spasmolytic agent. Eighty-eight patients were scanned with 50 mAs, and 80 patients were scanned with 70 mAs. The average amount of carbon dioxide insufflated was 3.9 L (range, 2.0–8.0 L). No complications occurred.

Interpretation Time
Observer 1 needed significantly more time to evaluate a complete study than observer 2: a mean interpretation time of 18 minutes (95% confidence interval: 17, 18 minutes) versus a mean interpretation time of 13 minutes (95% confidence interval: 12, 14 minutes) (P < .001).

Per-Patient Analysis
No patient had more than one polyp 10 mm or larger. The consensus reading resulted in 34 (76%) of 45 patients with a lesion 6 mm or larger, 20 (71%) of 28 patients with a 6–9-mm lesion, and 14 (82%) of 17 patients with a lesion 10 mm or larger (Table 4, Fig 2). Specificity was 79% (97 of 123) for the identification of patients with a lesion 6 mm or larger, 81% (113 of 140) for the identification of patients with lesions 6–9 mm, and 97% (146 of 151) for the identification of patients with a lesion 10 mm or larger (Table 4). No significant differences in detection rates were observed between the consensus reading (CT colonography) and colonoscopy (all P values, ≥.070). Colonoscopy did help identify significantly more patients with a lesion of any size than did observers 1 (P < .001) and 2 (P < .001) and significantly more patients with a lesion 6–9 mm (P = .012) or 6 mm or larger (P = .008) than did observer 2. No other significant differences in detection rates were found between colonoscopy and separate observers (all P values, ≥.057). Furthermore, no significant differences in sensitivities or specificities were observed between consensus reading and observers (sensitivity: all P values, ≥.063; specificity: all P values, ≥.092) or between separate observers (sensitivity: all P values, ≥.267; specificity: all P values, ≥.115). The consensus reading resulted in 34 (76%) of 45 patients with a polyp 6 mm or larger, in comparison to 29 (64%) of 45 patients for observer 2 (P = .267); no increase was observed compared with observer 1. With regard to larger lesions, the consensus reading resulted in the identification of 14 (82%) of 17 patients with a polyp 10 mm or larger, in comparison to 13 (76%) of 17 patients for observer 1 (P > .99) and 12 (71%) of 17 patients for observer 2 (P = .50). Table 5 shows negative and positive predictive values.


Figure 2A
View larger version (141K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a: CT colonographic images in patients with true-positive findings for a lesion 10 mm or larger or 6 mm or larger. (a) Image of 20-mm villous polyp (arrow) in sigmoid colon with adherent contrast agent on surface of lesion. A smaller polypoid lesion (arrowhead) with contrast agent in center was visible and was correctly classified as tagged stool. (b) Image of 6-mm polyp (arrow) in descending colon that is situated within a contrast material–tagged fluid level. Both lesions were correctly identified by both observers.

 

Figure 2B
View larger version (151K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b: CT colonographic images in patients with true-positive findings for a lesion 10 mm or larger or 6 mm or larger. (a) Image of 20-mm villous polyp (arrow) in sigmoid colon with adherent contrast agent on surface of lesion. A smaller polypoid lesion (arrowhead) with contrast agent in center was visible and was correctly classified as tagged stool. (b) Image of 6-mm polyp (arrow) in descending colon that is situated within a contrast material–tagged fluid level. Both lesions were correctly identified by both observers.

 

View this table:
[in this window]
[in a new window]

  		Table 5. Per-Patient Negative and Positive Predictive Values at CT Colonography according to Polyp Size Group

 
Per-Polyp Analysis
With regard to the three missed lesions 10 mm or larger (two flat polyps—one in the transverse and one in the ascending colon—and one sessile polyp in the ascending colon), no perception or technical errors were observed for the consensus reading (Table 6, Fig 3). Two (33%) of the six false-positive findings for lesions 10 mm or larger were matched to polypoid lesions at colonoscopy: an inverted appendix stump (Fig 4) and a mucosal bleb (cecum); one finding proved to be a lipomatous ileocecal valve, and three findings were attributed to poorly tagged stool. In general, detection rates for colonoscopy were better for lesions 6 mm or larger and 6–9 mm lesions (Table 6). For the other size thresholds, no significant differences in per-polyp sensitivity were found between CT colonography (consensus or observer 1 or observer 2) and colonoscopy (all P values, ≥.101), except for observer 2, whose detection rates were lower for lesions 7 mm or larger (P = .016). All data on per-polyp sensitivity values, with significant corresponding P values for all comparisons (colonoscopy vs consensus reading, colonoscopy vs observer 1, colonoscopy vs observer 2, consensus reading vs observer 1, consensus reading vs observer 2, and observer 1 vs observer 2), are provided in Table 6.


View this table:
[in this window]
[in a new window]

  		Table 6. Per-Polyp Sensitivity and Number of False-Positive Findings according to Size Group

 

Figure 3
View larger version (57K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3: Images from colonoscopy of lesions 10 mm or larger missed at CT colonography. Two flat adenomatous polyps (arrows) (left and middle) and sessile adenomatous polyp (arrow) (right) were missed by reviewers. These polyps were not visible in retrospect on CT colonographic images.

 

Figure 4
View larger version (46K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4: False-positive findings for lesions 10 mm or larger at CT colonography. Transverse (left) and three-dimensional (middle) CT colonographic images of false-positive finding (arrow) that had true polypoid morphology at endoscopy (right) but proved to be an inverted appendix stump.

 
Interobserver Agreement
Observers 1 and 2 recorded lesion(s) 10 mm or larger in 19 and 23 segments, respectively: Concordant findings were reported in 11 segments. No findings 10 mm or larger were recorded in 966 segments by both observers. This resulted in a per-segment agreement of 98% (977 of 1000 segments) and a corresponding very good {kappa}p value of 0.96. For lesions 6 mm or larger, the per-segment agreement was 94% (940 of 1000 segments), with an associated {kappa}p value of 0.88.

Image Quality
Four (2%) of 172 evaluated studies were considered nondiagnostic because of poor tagging and were therefore excluded from analysis (Fig 1). The radiologist evaluated stool tagging as good in 160 (95%) of the 168 included studies and as adequate in seven (4%) of 168 studies. In one (1%) of 168 studies, image quality was rated as poor, with untagged solid stool less than 6 mm. Figure 5 shows different tagging examples.


Figure 5A
View larger version (113K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5a: (a–d) CT colonographic images of different levels of homogeneity of stool tagging. (a) Image was evaluated as showing good stool tagging; a 15-mm adenomatous polyp (arrow) is visible in sigmoid colon; (b) image shows adequate tagging; (c) image shows poor tagging in sigmoid colon, but study was considered diagnostic for lesions 6 mm or larger; (d) image shows poor tagging, and this study was excluded because our reviewers rated it as nondiagnostic.

 

Figure 5B
View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5b: (a–d) CT colonographic images of different levels of homogeneity of stool tagging. (a) Image was evaluated as showing good stool tagging; a 15-mm adenomatous polyp (arrow) is visible in sigmoid colon; (b) image shows adequate tagging; (c) image shows poor tagging in sigmoid colon, but study was considered diagnostic for lesions 6 mm or larger; (d) image shows poor tagging, and this study was excluded because our reviewers rated it as nondiagnostic.

 

Figure 5C
View larger version (97K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5c: (a–d) CT colonographic images of different levels of homogeneity of stool tagging. (a) Image was evaluated as showing good stool tagging; a 15-mm adenomatous polyp (arrow) is visible in sigmoid colon; (b) image shows adequate tagging; (c) image shows poor tagging in sigmoid colon, but study was considered diagnostic for lesions 6 mm or larger; (d) image shows poor tagging, and this study was excluded because our reviewers rated it as nondiagnostic.

 

Figure 5D
View larger version (97K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5d: (a–d) CT colonographic images of different levels of homogeneity of stool tagging. (a) Image was evaluated as showing good stool tagging; a 15-mm adenomatous polyp (arrow) is visible in sigmoid colon; (b) image shows adequate tagging; (c) image shows poor tagging in sigmoid colon, but study was considered diagnostic for lesions 6 mm or larger; (d) image shows poor tagging, and this study was excluded because our reviewers rated it as nondiagnostic.

 
Patient Experience and Preference
Diarrhea was present in 152 (93%) of 163 patients who filled out the questionnaire with regard to diarrhea (Table 7). From among the 165 patients who returned the questionnaire after 5 weeks, 144 (87%) rated the bowel preparation for colonoscopy as more burdensome than the bowel preparation for CT colonography. With regard to patient preference, 114 (70%) of 164 patients preferred CT colonography to colonoscopy, 13 (8%) were indifferent, and 36 (22%) favored colonoscopy as a colonic examination.


View this table:
[in this window]
[in a new window]

  		Table 7. Burden of Diarrhea and Overall Burden of Bowel Preparation in 168 Patients

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
ADVANCES IN KNOWLEDGE
 IMPLICATION FOR PATIENT CARE
 References
 
Our study results had a sensitivity for the consensus reading of 76% (34 of 45) and 82% (14 of 17) for patients with a colorectal polyp 6 mm or larger and those with a polyp 10 mm or larger, respectively. Detection rates were higher for colonoscopy, for both patients with a polyp 6 mm or larger (41 [91%] of 45) and patients with a lesion 10 mm or larger (15 [88%] of 17), than those at CT colonography, but this difference was not statistically significant. CT colonography correctly depicted 97 (79%) of 123 patients with polyps 6 mm or larger and 146 (97%) of 151 patients with polyps 10 mm or larger. For these size thresholds, negative predictive values were 90% (97 of 108) and 98% (146 of 149), and positive predictive values were 57% (34 of 60) and 74% (14 of 19) for polyps 6 mm or larger and polyps 10 mm or larger, respectively. Interobserver agreement per segment was 94% (940 of 1000) and 98% (977 of 1000) for lesions 6 mm or larger and lesions 10 mm or larger, respectively.

Our ability to identify patients with lesions 6 mm or larger and those with lesions 10 mm or larger was not as good as in the study by Iannaccone et al (19), in which sensitivity ranged from 90% (44 of 49) for patients with a polyp 6 mm or larger to 100% (24 of 24) for patients with a polyp 10 mm or larger. These differences in accuracy could be explained by interpretation problems, but could also possibly be at least partly explained by visualization problems. An argument in favor of an interpretation problem is that the observers in our study were less experienced (70 and 120 cases) than those in the other study (100, 200, and 300 cases, respectively, for the three observers). Although we believe that a higher level of experience might have improved the detection of medium-size (6–9-mm) polyps, this is not an adequate explanation for the difference with regard to the detection rates of lesions 10 mm or larger. Namely, no lesions 10 mm or larger that were missed at the consensus reading were visible in retrospect. Therefore, the problem with the identification of polyps 10 mm or larger seems to be a visualization rather than an interpretation problem.

There are two major differences between both studies that can possibly explain these visualization errors. First, we used a different bowel preparation with a lower dose of diatrizoate meglumine but added barium and bisacodyl. In theory, inhomogeneous tagging might result in the masking of true lesions. However, stool tagging was considered good in 160 (95%) of 168 of the studies, including the cases with missed lesions, and only four studies were excluded because of poor stool tagging. Therefore, we do not believe that our tagging regimen was the cause of the visualization errors. Second, we investigated only patients at increased risk for colorectal cancer, while the other study had a mixed population, of which 106 (52%) of 203 were high-risk patients. Recently, MacCarty et al (20) suggested that patients at increased risk who regularly undergo colonoscopy tend to have lesions that are hard to detect with CT colonography because easy-to-see lesions are detected and removed at previous colonoscopic examinations, while less conspicuous lesions remain in situ. Van Gelder et al (5), who studied a patient group comparable to ours, found one-third (14 of 48) of polyps 10 mm or larger to be flat, and 71% (10 of 14) of these flat polyps were missed at CT colonography. In our study, four (24%) of the 17 polyps that were 10 mm or larger were flat, and two (50%) of these were missed. Our results underscore the findings of MacCarty et al; this indicates that CT colonography with fecal tagging might therefore perform better in screening populations (eg, patients at average risk who have not previously undergone screening with colonoscopy).

The relatively high number of 42 false-positive findings by using 6 mm as a size threshold was somewhat disappointing because fecal tagging is supposed to be associated with a lower rate of false-positive findings because stool is not mistaken for polyps (15,17,24). The actual polyp size threshold for colonoscopy referral in a screening setting is still debated, but many institutions consider a size threshold of 6 mm clinically relevant. An explanation for our high number of false-positive findings in this category is that although stool tagging was homogeneous, the average volume-rendering effect at a section thickness of 3.2 mm might have caused small (6- or 7-mm) pieces of stool to look like small polyps. This explains the sharp decline in the number of false-positive findings at a size threshold of 8 mm or larger, which can be deduced from the fact that only 13 false-positive findings 8 mm or larger were observed, and six of these were 10 mm or larger. Furthermore, the observers were relatively inexperienced, which, in our opinion, undoubtedly contributed to the number of false-positive findings. As mentioned before, we did not use stool subtraction software, which might have helped lower the number of false-positive findings. Despite the high number of medium-sized false-positive findings, specificity was 79% (97 of 123) for patients with a lesion 6 mm or larger and increased to 94% (136 of 145) at a size threshold of 8 mm. Furthermore, negative predictive values were high (lesions ≥ 6 mm, 90% [97 of 108]; lesions ≥ 10 mm, 98% [146 of 149]).

We applied a double-reading strategy (consensus reading for discrepant lesions) for lesions 6 mm or larger. Although this is a time-consuming and costly strategy, it has been suggested to improve sensitivity (4). Although a higher detection rate for the consensus reading compared with that for the separate observers was observed, no significant improvement was found by using double reading (P ≥ .063). Alternatives to a double reading by radiologists are computed-aided detection (25,26) or the deployment of radiographers (27,28).

The radiologist needed significantly (P < .001) more time to review the cases in comparison to the radiology fellow (18 vs 13 minutes). An explanation is that the research fellow had more experience in CT colonography prior to the study than the radiologist (120 vs 70 cases). However, the longer review time might also reflect a more careful evaluation of the data by the radiologist. This can be substantiated by the fact that higher sensitivities were calculated for the radiologist than for the research fellow.

Despite the limited bowel preparation in our study, almost all patients had some form of diarrhea. We believe the high occurrence of diarrhea was caused by a combination of the laxative side effect of diatrizoate meglumine and bisacodyl. Diarrhea undoubtedly increases patient burden but, in theory, also leads to homogeneous tagging of fecal material by providing a liquidized colonic lumen. The fact that good homogeneity of stool was found in 160 (95%) of 168 patients supports this hypothesis. Regardless of the presence of diarrhea, the overall burden of the CT colonographic preparation was perceived as none or mild by 123 (74%) of 166 patients. Furthermore, 144 (87%) of 165 patients rated the bowel preparation for colonoscopy as more burdensome than the bowel preparation used for CT colonography. Therefore, our results underscore the idea that a limited bowel preparation is less burdensome for patients than an extensive bowel preparation and at the same time can provide good CT colonographic image quality.

Some limitations of our study should be considered. A cohort of 168 patients at increased risk for colorectal carcinoma was evaluated in our study. Our results can therefore not be extrapolated to a screening setting of patients at average risk for developing colorectal cancer. Two observers were used in our study, and although interobserver variability was high, the use of two observers might limit generalizibility of our data. Relatively short training of 70 cases for the radiologist and 120 cases for the radiology fellow was applied, and this is probably not enough to achieve a high level of competence with a primary two-dimensional evaluation method (29). We used a section thickness of 3.2 mm with our four-section CT scanner to enable scanning in one breath hold, and thin collimation is associated with better accuracy in the clean colon (30–32). Presumably, thin collimation is also important in the limitedly prepared colon because inhomogeneously tagged stool can be better differentiated from polyps, which subsequently lowers false-positive rates and improves detection of particularly flat polyps. Finally, a disadvantage of using a limited bowel preparation for CT colonography in comparison to using a cathartic bowel preparation is that same-day colonoscopy for polyp removal cannot be performed.

In conclusion, we report that CT colonography with limited bowel preparation in a population at increased risk for colorectal cancer had a sensitivity and specificity of 82% (14 of 17) and 97% (146 of 151), respectively, for patients with polyps 10 mm or larger. These results are in line with those of performance studies with extensive bowel preparation (4,5). Because a cathartic bowel preparation could potentially be a deterrent to screening, we believe that limited bowel preparation can be of value for the implementation of CT colonography as a screening technique for colorectal polyps and cancer.


   ADVANCES IN KNOWLEDGE
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
IMPLICATION FOR PATIENT CARE
 References
 


   IMPLICATION FOR PATIENT CARE
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 IMPLICATION FOR PATIENT CARE
References
 


   ACKNOWLEDGMENTS
 
The authors thank Jasper Florie, MD, Karin Horsthuis, MD, and Henk W. Venema, PhD, for their critical review of the manuscript. We thank C. Yung Nio, MD, for his critical review of the CT colonographic cases with missed lesions and Koos A. H. Zwinderman, PhD, for his valuable contribution with regard to the statistical analyses (GEE) of our data.


   FOOTNOTES
 

Abbreviations: GEE = generalized estimating equations

Author contributions: Guarantors of integrity of entire study, S.J., A.H.d.V., S.B., J.S.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, S.J., A.H.d.V., S.B., J.S.; clinical studies, S.J., A.H.d.V., J.P., E.D., L.C.B., J.F.B., A.H., J.S.; statistical analysis, S.J., S.B.; and manuscript editing, all authors

Authors stated no financial relationship to disclose.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 ADVANCES IN KNOWLEDGE
 IMPLICATION FOR PATIENT CARE
 References
 

  1. Fenlon HM, Nunes DP, Schroy PC, et al. A comparison of virtual and conventional colonoscopy for the detection of colorectal polyps. N Engl J Med 1999;341(20):1496–1503. [Published correction appears in N Engl J Med 2000;342(7):524.][Abstract/Free Full Text]
  2. Pickhardt PJ, Choi JR, Hwang I, et al. Computed tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med 2003;349(23):2191–2200.[Abstract/Free Full Text]
  3. Pineau BC, Paskett ED, Chen GJ, et al. Virtual colonoscopy using oral contrast compared with colonoscopy for the detection of patients with colorectal polyps. Gastroenterology 2003;125(2):304–310.[CrossRef][Medline]
  4. Johnson CD, Harmsen WS, Wilson LA, et al. Prospective blinded evaluation of computed tomographic colonography for screen detection of colorectal polyps. Gastroenterology 2003;125(2):311–319.[CrossRef][Medline]
  5. Van Gelder RE, Nio CY, Florie J, et al. Computed tomographic colonography compared with colonoscopy in patients at increased risk for colorectal cancer. Gastroenterology 2004;127(1):41–48.[CrossRef]
  6. Harewood GC, Wiersema MJ, Melton LJ 3rd. A prospective, controlled assessment of factors influencing acceptance of screening colonoscopy. Am J Gastroenterol 2002;97(12):3186–3194.[CrossRef][Medline]
  7. van Gelder RE, Birnie E, Florie J, et al. CT colonography and colonoscopy: assessment of patient preference in a 5-week follow-up study. Radiology 2004;233(2):328–337.[Abstract/Free Full Text]
  8. Ristvedt SL, McFarland EG, Weinstock LB, Thyssen EP. Patient preferences for CT colonography, conventional colonoscopy, and bowel preparation. Am J Gastroenterol 2003;98(3):578–585.[CrossRef][Medline]
  9. Zalis ME, Perumpillichira JJ, Magee C, Kohlberg G, Hahn PF. Tagging-based, electronically cleansed CT colonography: evaluation of patient comfort and image readability. Radiology 2006;239(1):149–159. [Published correction appears in Radiology 2006;240(1):304.][Abstract/Free Full Text]
  10. Callstrom MR, Johnson CD, Fletcher JG, et al. CT colonography without cathartic preparation: feasibility study. Radiology 2001;219(3):693–698.[Abstract/Free Full Text]
  11. Weitzman ER, Zapka J, Estabrook B, Goins KV. Risk and reluctance: understanding impediments to colorectal cancer screening. Prev Med 2001;32(6):502–513.[CrossRef][Medline]
  12. Gluecker TM, Johnson CD, Harmsen WS, et al. Colorectal cancer screening with CT colonography, colonoscopy, and double-contrast barium enema examination prospective assessment of patient perceptions and preference. Radiology 2003;227(2):378–384.[Abstract/Free Full Text]
  13. Rex DK. Current colorectal cancer screening strategies overview and obstacles to implementation. Rev Gastroenterol Disord 2002;2(suppl 1):S2–S11.
  14. Lefere P, Gryspeerdt S, Marrannes J, et al. CT colonography after fecal tagging with a reduced cathartic cleansing and a reduced volume of barium. AJR Am J Roentgenol 2005;184(6):1836–1842.[Abstract/Free Full Text]
  15. Lefere PA, Gryspeerdt SS, Dewyspelaere J, et al. Dietary fecal tagging as a cleansing method before CT colonography: initial results—polyp detection and patient acceptance. Radiology 2002;224(2):393–403.[Abstract/Free Full Text]
  16. McFarland EG, Zalis ME. CT colonography: progress toward colorectal evaluation without catharsis. Gastroenterology 2004;127(5):1623–1626.[CrossRef][Medline]
  17. Thomeer M, Carbone I, Bosmans H, et al. Stool tagging applied in thin-slice multidetector computed tomography colonography. J Comput Assist Tomogr 2003;27(2):132–139.[CrossRef][Medline]
  18. Bielen D, Thomeer M, Vanbeckevoort D, et al. Dry preparation for virtual CT colonography with fecal tagging using water-soluble contrast medium: initial results. Eur Radiol 2003;13(3):453–458.[Medline]
  19. Iannaccone R, Laghi A, Catalano C, et al. Computed tomographic colonography without cathartic preparation for the detection of colorectal polyps. Gastroenterology 2004;127(5):1300–1311.[CrossRef][Medline]
  20. MacCarty RL, Johnson CD, Fletcher JG, Wilson LA. Occult colorectal polyps on CT colonography: implications for surveillance. AJR Am J Roentgenol 2006;186(5):1380–1383.[Abstract/Free Full Text]
  21. Pepe MS. The statistical evaluation of medical tests for classification and prediction: Oxford statistical science series 28. New York, NY: Oxford University Press, 2003; 35–65.
  22. Byrt T, Bishop J, Carlin J. Bias, prevalence and kappa. J Clin Epidemiol 1993;46(5):423–429.[CrossRef][Medline]
  23. Petersen HH, Enøe C, Nielsen EO. Observer agreement on pen level prevalence of clinical signs in finishing pigs. Prev Vet Med 2004;64(2-4):147–156.[CrossRef][Medline]
  24. Zalis ME, Hahn PF. Digital subtraction bowel cleansing in CT colonography. AJR Am J Roentgenol 2001;176(3):646–648.[Free Full Text]
  25. Summers RM, Jerebko AK, Franaszek M, et al. Colonic polyps: complementary role of computer-aided detection in CT colonography. Radiology 2002;225(2):391–399.[Abstract/Free Full Text]
  26. Halligan S, Altman DG, Mallett S, et al. Computed tomographic colonography: assessment of radiologist performance with and without computer-aided detection. Gastroenterology 2006;131(6):1690–1699.[CrossRef][Medline]
  27. Bodily KD, Fletcher JG, Engelby T, et al. Nonradiologists as second readers for intraluminal findings at CT colonography. Acad Radiol 2005;12(1):67–73.[CrossRef][Medline]
  28. Jensch S, Van Gelder RE, Florie J, et al. Performance of radiographers in the evaluation of CT colonographic images. AJR Am J Roentgenol 2007;188(3):W249–W255.[Abstract/Free Full Text]
  29. Taylor SA, Halligan S, Burling D, et al. CT colonography: effect of experience and training on reader performance. Eur Radiol 2004;14(6):1025–1033.[CrossRef][Medline]
  30. Lui YW, Macari M, Israel G, Bini EJ, Wang H, Babb J. CT colonography data interpretation: effect of different section thicknesses—preliminary observations. Radiology 2003;229(3):791–797.[Abstract/Free Full Text]
  31. Taylor SA, Halligan S, Bartram CI, et al. Multi–detector row CT colonography: effect of collimation, pitch, and orientation on polyp detection in a human colectomy specimen. Radiology 2003;229(1):109–118.[Abstract/Free Full Text]
  32. Jensch S, van Gelder RE, Venema HW, et al. Effective radiation doses in CT colonography: results of an inventory among research institutions. Eur Radiol 2006;16(5):981–987.[CrossRef][Medline]




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
2471070439v1
247/1/122    most recent
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Jensch, S.
Right arrow Articles by Stoker, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Jensch, S.
Right arrow Articles by Stoker, J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE