Published online before print November 22, 2005, 10.1148/radiol.2381041756
(Radiology 2005;238:143-149.)
© RSNA, 2005
Colorectal Polyps: Detection with Dark-Lumen MR Colonography versus Conventional Colonoscopy1
Dirk Hartmann, MD,
Boris Bassler, MD,
Dieter Schilling, MD,
Henning E. Adamek, MD,
Ralf Jakobs, MD,
Barbara Pfeifer, MD,
Axel Eickhoff, MD,
Christoph Zindel, MD,
Jürgen F. Riemann, MD and
Günter Layer, MD
1 From the Departments of Medicine C (Gastroenterology) (D.H., D.S., H.E.A., R.J., A.E., J.F.R.) and Diagnostic and Interventional Radiology (B.B., B.P., G.L.), Hospital Ludwigshafen, Academic Teaching Hospital of Johannes-Gutenberg-University of Mainz, Bremserstrasse 79, 67063 Ludwigshafen am Rhein, Germany; and Siemens Medical Solutions, Erlangen, Germany (C.Z.). Received October 13, 2004; revision requested December 22; revision received January 11, 2005; accepted February 2; final version accepted March 9.
Address correspondence to G.L. (e-mail: guenter.layer{at}klilu.de).
 |
ABSTRACT
|
|---|
Purpose: To prospectively compare dark-lumen magnetic resonance (MR) colonography with conventional colonoscopy in the detection of colorectal polyps.
Materials and Methods: Local ethical committee approval and informed consent were obtained. One hundred consecutive patients (56 men, 44 women; mean age ± standard deviation, 67.7 years ± 14.7; range, 25–82 years) who were referred for conventional colonoscopy from January 2003 to January 2004 underwent MR colonography and conventional colonoscopy after standard precolonoscopic bowel cleansing. Colonoscopy was performed immediately after MR colonography. For MR colonography, the colon was filled with approximately 2000 mL of tap water. Imaging was performed with a 1.5-T MR unit with patients in the prone position. A T1-weighted three-dimensional volumetric interpolated breath-hold sequence was performed before and 75 seconds after intravenous administration of 0.2 mmol gadobenate dimeglumine per kilogram of body weight. Results of MR colonography were analyzed on a per-polyp and per-patient basis. Findings at colonoscopy were used as the reference for determining accuracy, sensitivity, specificity, and positive and negative predictive values of MR colonography.
Results: Of 100 patients recruited for study, 92 (52 men, 40 women; mean age, 61.5 years ± 14.5; range, 25–82 years) underwent complete MR and conventional colonoscopy examinations. Forty-three of the 92 patients (47%) had normal findings at conventional colonoscopy. In the other 49 patients (53%), conventional colonoscopy depicted 107 polyps (82 adenomas, 25 hyperplastic polyps) and seven carcinomas. At per-polyp analysis, sensitivity of MR colonography in the detection of adenomatous polyps was 100% for polyps at least 10 mm in diameter and 84.2% for polyps 6–9 mm in diameter. At per-patient analysis, the accuracy of MR colonography was 93.1% (sensitivity, 89%; specificity, 96%) if detection of adenomatous polyps of all sizes was considered.
Conclusion: Dark-lumen MR colonography is a promising modality with high accuracy for detecting colorectal polyps larger than 5 mm in diameter.
© RSNA, 2005
|
INTRODUCTION
|
|---|
To date, most evaluations of magnetic resonance (MR) colonography have considered only the bright-lumen technique (1,2). A recently developed and simplified variation of MR colonography is dark-lumen MR colonography (3). Dark-lumen MR colonography is based on the use of a heavily T1-weighted three-dimensional gradient-echo sequence after the rectal administration of a water enema and intravenous injection of paramagnetic contrast material.
Results of an early study (3) suggested that dark-lumen MR colonography may be an attractive alternative to existing diagnostic tests for colorectal cancer. To our knowledge, however, a large-scale prospective study to evaluate the performance of dark-lumen MR colonography has not been yet performed. To become an accepted method for polyp detection, dark-lumen MR colonography must be compared to conventional colonoscopy, which is considered the reference standard.
Thus, the purpose of our study was to prospectively assess the diagnostic accuracy of dark-lumen MR colonography in the detection of polyps by using findings from histologic examination and conventional colonoscopy as the reference standards.
|
MATERIALS AND METHODS
|
|---|
Siemens Medical Solutions, Erlangen, Germany, developed the software for data interpretation. However, the authors who were not affiliated with Siemens Medical Solutions had control of the data and information included herein.
Study Group
One hundred consecutive patients (56 men, 44 women; mean age ± standard deviation, 67.7 years ± 14.7; age range, 25–82 years) who were older than 18 years and were referred for conventional colonoscopy from January 2003 to January 2004 were recruited for this study. Each patient was asked to undergo MR colonography before conventional colonoscopy. Patients were referred for colonoscopy because of gastrointestinal bleeding, to screen for colorectal cancer (at normal risk), to follow up an abnormal hemoccult test result, or to evaluate iron deficiency anemia or gastrointestinal symptoms (abdominal pain, diarrhea). Patients were excluded if they were younger than 18 years, had a personal or family history of a genetic polyp syndrome, had known intolerance to MR contrast material, or had contraindications to MR imaging (eg, pacemakers, intracorporeal metal parts, claustrophobia, pregnancy). The study was performed in accordance with all guidelines set forth by the local ethical committee, and all patients gave written informed consent.
Technique
Preparation for bowel cleansing consisted of oral ingestion of 4 L of a polyethylene glycol–electrolyte solution (Oralav; B. Braun, Melsungen, Germany). Dark-lumen MR colonography was performed immediately before conventional colonoscopy by using a 1.5-T MR system (Magnetom Sonata; Siemens Medical Solutions). A combination of two surface coils was used in conjunction with the built-in spine array coil for signal reception to permit coverage of the entire colon. To minimize bowel peristalsis, 40 mg of butyl scopolamine (Buscopan; Boehringer, Ingelheim, Germany) was injected intravenously. After the placement of a standard balloon-tipped rectal catheter, the colon was filled with 2000–2500 mL of warm tap water. To ensure safe and complete filling, the administration of the enema was monitored by using a T2-weighted fast imaging with steady-state precession (TrueFisp; Siemens Medical Solutions) sequence (2.4/1.2 [repetition time msec/echo time msec], 60° flip angle, 10-mm-thick sections), which enabled the acquisition of one image every 3 seconds. The filling process was stopped when the images demonstrated that the water reached the cecum.
Once complete filling and distention of the colon were assured, a first coronal unenhanced T1-weighted three-dimensional gradient-echo data set (volumetric interpolated breath-hold examination [VIBE; Siemens Medical Solutions]) was collected. The following parameters were used for the three-dimensional sequence: 3.1/1.17, 10° flip angle, 400 x 400-mm field of view, and effective section thickness of 1.5–2.0 mm (96 sections). Data acquisition was performed with the patient in the prone position only.
Contrast material–enhanced T1-weighted MR images were obtained by using identical parameters 75 seconds after the intravenous administration (3 mL/sec) of gadobenate dimeglumine (Multihance; Bracco, Milan, Italy) followed by injection of 20 mL of normal saline at the same injection rate. The entire acquisition was performed during 22 seconds in a single breath hold. After acquisition of the three-dimensional data set, the enema bag was placed on the floor to facilitate emptying of the colon.
Image Analysis
The three-dimensional data sets were processed by using proprietary software and hardware (Leonardo workstation; Siemens Medical Solutions). This software program automatically calculates multiplanar reconstructions from the three-dimensional data set. The diagnostic interface enables a virtual "fly through" tour of the three-dimensional image (surface rendering technique). Images from MR colonography were interpreted in the multiplanar reformation mode, by scrolling through the three-dimensional data set, and with virtual endoscopy. Detected polyps were measured with the electronic calipers of the workstation. For the purpose of determining location, the colon was divided into eight segments, as follows: rectum, sigmoid colon, descending colon, splenic flexure, transverse colon, hepatic flexure, ascending colon, and cecum. Five readers participated in the study: two radiologists (B.B. and G.L., with more than 5 and 15 years of experience in MR imaging of the abdomen, respectively) and three gastroenterologists (D.H., D.S., and H.E.A., all with more than 5 years of experience in performing colonoscopy). Each image was read by one MR radiologist (B.B., G.L.) and one gastroenterologist (D.H., D.S., H.E.A.) who were blinded to the findings from conventional colonoscopy. The location and size of all detected endoluminal lesions were recorded in a prospective consensus reading.
All MR data sets were assessed for the presence of nonpolypoid findings. Criteria used for inflammatory bowel diseases included bowel wall thickening, increased contrast material uptake in segmental parts of the colon, and loss of haustral folds. Extracolonic findings (eg, cholecystolithiasis, hepatic cysts, hepatic metastasis) at MR imaging were also recorded and categorized as representing a condition of potentially high or low clinical importance.
Conventional Colonoscopy and Histologic Examination
Conventional colonoscopy was performed with a standard endoscope (GIF CF 160 HI; Olympus Optical Europe, Hamburg, Germany) by a gastroenterologist (R.J., A.E., J.F.R.) with at least 5 years of experience in performing endoscopy. Colonoscopy was performed immediately after MR colonography. The gastroenterologist was blinded to the results of MR colonography. The location of each endoscopically detected lesion was documented with fluoroscopy. The size was measured by comparison with an open biopsy forceps. All lesions identified were removed or sampled for biopsy and analyzed at histopathologic examination. Histologic evaluation was performed by a pathologist with more than 10 years of experience in colon polyp analysis.
Data and Statistical Analysis
We regarded the results of conventional colonoscopy as the reference standard with which to compare the results of MR colonography. Patients who did not undergo complete MR colonography or conventional colonoscopy examinations were excluded from analysis. The results of MR colonography were analyzed on a per-polyp and per-patient basis. Analysis included an evaluation of the agreement between the two methods of colonoscopy with respect to both the size and location of the polyp. Colorectal lesions were classified according to size, as follows: (a) 5 mm in diameter or smaller, (b) 6–9 mm in diameter, and (c) 10 mm in diameter or larger. For a true-positive result, the lesion identified at MR colonography had to have been matched according to location, size, and morphologic features to a lesion found at conventional colonoscopy. In the per-patient evaluation, a result was considered to be a true-positive finding only when at least one polyp identified at MR colonography was matched to a lesion seen at conventional colonoscopy. All other results were considered to be false-positive findings. The findings at colonoscopy were used as the standard of reference for determining the accuracy, sensitivity, specificity, and positive and negative predictive values of MR colonography. Results of the statistical analysis of patient data are expressed as the mean ± standard deviation.
|
RESULTS
|
|---|
Baseline Patient Characteristics
Of the 100 recruited patients, 92 (52 men, 40 women; mean age, 61.5 years ± 14.5; age range, 25–82 years) underwent complete MR and colonoscopic examinations (Table 1). Six patients were excluded because conventional colonoscopy could not be completed owing to the tortuosity of the colon in one patient and an occlusive stricture that precluded passage of the colonoscope in five patients. Two additional patients were excluded because MR colonography could not be completed owing to claustrophobia.
Conventional Colonoscopy
Of the 92 patients who underwent colonoscopy, 43 (47%) had normal findings. The remaining 49 patients had 107 polyps and seven carcinomas. Forty-four of the 107 polyps (41%) were 1–5 mm in diameter, 41 (38%) were 6–9 mm in diameter, and 22 (21%) were 10 mm in diameter or larger (Table 2). All lesions were successfully removed and examined histologically. In seven lesions with an endoscopic appearance that was suspicious for malignancy, only biopsy was performed. Eighty-two of the 107 polyps were adenomatous (76.6%) and 25 were hyperplastic (23.4%). All seven adenocarcinomas (in which only biopsy was performed) were sessile; three were located in the rectum, two were located in the sigmoid colon, and one each was located in the splenic flexure and the ascending colon. The prevalence of adenomatous polyps with diameters of at least 10 mm was 27% (n = 22); the prevalence of adenomatous polyps with diameters of 6–9 mm was 46% (n = 38), and the prevalence of adenomatous polyps with diameters of 5 mm or less was 27% (n = 22).
Dark-Lumen MR Colonography
Table 3 shows the diagnostic performance of MR colonography in the detection of polyps according to size and histologic type. With a per-polyp analysis, 58 of the 107 polyps (54.2%) seen at conventional colonoscopy were correctly identified on the basis of location and size. Four of the 44 polyps (9.1%) with diameters of 1–5 mm, 32 of 41 polyps (78%) with diameters of 6–9 mm, and all polyps with diameters of at least 10 mm (100%) were correctly identified with MR colonography. The sensitivity of MR colonography in the detection of polyps with diameters of 6–9 mm and 10 mm or larger was higher than that in the detection of polyps with diameters of 1–5 mm (78% and 100%, respectively). The performance of MR colonography was also related to histologic type. No hyperplastic polyp could be detected independently of size. Four of the 22 adenomas (18%) with diameters of 1–5 mm, 32 of 38 adenomas (84%) with diameters of 6–9 mm, and all adenomas with diameter of at least 10 mm (100%) were correctly identified with MR colonography. The sensitivity of MR colonography in the detection of all adenomatous polyps was 71%. All seven carcinomas were identified with MR colonography. There were eight false-positive findings at MR colonography: two lesions were 1–5 mm in diameter, five were 6–9 mm in diameter, and one was at least 10 mm in diameter.
When the results of MR colonography were analyzed on a per-patient basis, the performance was improved (Table 4). For polyps of all sizes, there were 41 true-positive, 44 true-negative, two false-positive, and five false-negative findings. Therefore, when polyps of all sizes were included, the per-patient sensitivity of MR colonography was 89% (positive predictive value, 95%) and the specificity was 96% (negative predictive value, 90%). The sensitivity and specificity in the detection of adenomas were 24% and 96%, respectively, for lesions with diameters of 1–5 mm; 84% and 99%, respectively, for lesions with diameters of 6–9 mm; and 100% and 100% for lesions with diameters of at least 10 mm. The accuracy in the detection of adenomas independently of size was 92%.
Nonpolypoid and Extracolonic Findings
Conventional colonoscopy revealed acute diverticulitis of the sigmoid colon in one patient, Crohn disease in five patients, ulcerative colitis in four patients, and inflammatory stenoses on the basis of nonsteroidal antiinflammatory drug, or NSAID, intake in one patient.
MR colonography enabled the correct diagnosis of inflammatory changes in all patients. Ulcerative colitis demonstrated a loss of haustral markings at MR colonography and revealed increased contrast material uptake in the affected bowel segments and pseudopolyps. The patient with Crohn disease had skip lesions characterized by a thickened wall and increased contrast material uptake. MR colonography of acute diverticulitis showed a thickened colonic wall and perifocal edema in the small segment of the sigmoid colon. In one patient, MR colonography showed a tumor in the ileum that was not reached with conventional colonoscopy. After resection, histopathologic analysis revealed an adenocarcinoma of the small bowel. In the two patients with NSAID colonopathy, MR colonography showed a thickened colonic wall with multiple stenoses.
Extraintestinal nonrelevant clinical findings included renal cysts (n = 12), hepatic cysts (n = 5), cholecystolithiasis (n = 6), and hepatic hemangioma (n = 4). Relevant findings were hepatic metastasis in two patients and pelvic lymphoma in one woman with a history of cervical cancer.
|
DISCUSSION
|
|---|
Initial studies with computed tomographic (CT) colonography have revealed a sensitivity of 91% relative to conventional colonoscopy for polyps larger than 10 mm (4–8). Exposure to ionizing radiation, however, may potentially limit the applicability of CT as a screening method. The typical radiation load per examination is up to 10 mSv with CT colonography.
Thus, our efforts have been focused on MR colonography. To date, most approaches to MR colonography have been based on the administration of a rectal enema containing paramagnetic contrast material (1,2). This method, called bright-lumen MR colonography, has been shown to be accurate for detecting polyps larger than 8 mm. On three-dimensional gradient-echo images, only the colonic lumen containing contrast material is bright; the surrounding tissues, including the colonic wall and polyps, have low signal intensity. Polypoid masses appear as dark filling defects within the bright colonic lumen—an appearance that is difficult to differentiate from that of residual fecal material and small pockets of air.
To avoid false-positive findings and to compensate for the presence of residual air, the three-dimensional acquisition is performed in both the prone and supine positions. Turning the patient in the midst of the examination prolongs the examination and necessitates that a new landmark be chosen to localize the sequence to ensure full coverage of the colon in the subsequent three-dimensional acquisition. Although most authors suggest a gadolinium-water dilution of 1:100 (1), some have recommended the use of a 1:50 dilution (2). Assuming a colonic volume of 3000 mL, 30–60 mL of costly paramagnetic contrast material is needed for the rectal enema alone. In addition, most bright-lumen MR colonography protocols call for the additional intravenous administration of paramagnetic contrast material at a dose of 0.1 mmol per kilogram of body weight (1,2).
Dark-lumen MR colonography overcomes the limitations inherent to bright-lumen MR colonography. The intravenous application of paramagnetic contrast material enables the direct depiction of the colorectal wall. Occasionally, the presence of residual stool or air bubbles may require direct comparison with the three-dimensional data set collected before the administration of paramagnetic contrast material. To date, only limited data exist comparing dark-lumen MR colonography to conventional colonoscopy (3).
In our study, all seven carcinomas and all 22 adenomas with diameters of at least 10 mm were identified with MR colonography. With a per-polyp analysis, only five of 38 adenomas with diameters of 6–9 mm were missed with MR colonography. The sensitivity in the detection of adenomas smaller than 5 mm was very low. In a recently published retrospective study, Ajaj et al (3) performed dark-lumen MR colonography followed by conventional colonoscopy in 122 patients suspected of having colorectal disease (48 polyps). None of 30 polyps measuring 5 mm or smaller identified at conventional colonoscopy were detected at MR colonography. In polyps with diameters of 6–10 mm, MR colonography correctly depicted 16 of 18 lesions documented at conventional colonoscopy. In addition, two polyps at least 10 mm and all nine colorectal carcinomas were correctly seen at MR colonography. Unfortunately, the authors did not differentiate the histologic type into adenomatous and hyperplastic and so it is unclear whether hyperplastic polyps were seen in this trial. In the present study, no hyperplastic polyps were detected independently of size. This may be because their vascular architecture is similar to that of normal mucosa, which resulted in no pathologic wall enhancement. Fenlon et al (4), in a study using CT colonography, suggested that hyperplastic polyps may be effaced when the colon is distended. Regardless of the reason, it should be noted that malignant transformation is less common with hyperplastic polyps than with adenomas.
The performance of MR colonography in our study was highly dependent on the size and histologic type of the lesions. The threshold for the reliable detection of small lesions was approximately 5 mm. The detection rate for larger adenomas (>5 mm), however, was much better, approaching 90% compared with conventional colonoscopy. It should also be noted that conventional colonoscopy itself is not free of false results (9–11). Between 10% and 20% of colonic polyps and up to 5% of colorectal cancers may be missed at conventional colonoscopy. For this reason, care must be taken when reporting false-positive results with MR colonography. In a large study (n = 1233) (8) comparing CT colonography with conventional colonoscopy, 55 additional polyps with diameters of at least 5 mm were missed at initial colonoscopy but were seen retrospectively at conventional colonoscopy. Consequently, it is possible that the true specificity and positive predictive value of MR colonography are even higher than those reported herein.
From the standpoint of patient care, it is not useful to analyze the results on a per-polyp basis alone. A per-patient analysis is more important because each patient with one positive finding at MR colonography must undergo conventional colonoscopy for therapeutic planning. In our study, the per-patient performance led to a sensitivity of 89% in the detection of all lesions with a possible risk for colorectal cancer, independently of size. For this reason, the low sensitivity in the detection of adenomas with diameters of 1–5 mm can be considered acceptable. Nevertheless, there is controversy about what constitutes a clinically significant polyp with regard to size. A recent prospective study (12) of 1000 consecutive patients who underwent routine colonoscopy revealed that lesions smaller than 10 mm in diameter, whether flat or polypoid, were unlikely to contain early cancer. The risk of cancer for small polyps and flat lesions was 6% and 4%, respectively, whereas that for polyps and flat lesions larger than 10 mm was 16% and 29%, respectively.
A limitation of our study is the high prevalence of polyps compared with the numbers reported from most studies in which CT or MR colonography had been compared with conventional colonoscopy (1,8). The reason is that we examined a symptomatic patient population or patients with a higher risk for colorectal diseases rather than a screening population. In a large prospective study (n = 1233) Pickhardt et al (8) evaluated the performance characteristics of CT colonography in a typical asymptomatic screening population. The prevalence of adenomatous polyps with diameters of at least 10 mm was 3.9%, whereas that for adenomatous polyps at least 8 or 6 mm in diameter was 6.7% and 13.6%, respectively. The corresponding sensitivities for detection were 88.7%, 93.9% and 93.8%, respectively. The authors concluded that CT colonography is an accurate method for screening asymptomatic adults who have an average risk of colorectal cancer. On the basis of these study results, the relative prevalence of polyps and adenomas may not be a major concern for CT colonography. For MR colonography, the proof is not yet evident.
Dark-lumen MR colonography is a promising modality with a high accuracy for detecting colorectal polyps larger than 5 mm in diameter. MR colonography in symptomatic patients helped correctly diagnose more than 90% of polyps when the results were compared with those from conventional colonoscopy. Further, MR colonography shows comparable results to CT colonography data in the literature, but without x-ray exposure. The findings support the further evaluation of MR colonography in a large-scale study involving an asymptomatic screening population.
|
ACKNOWLEDGMENTS
|
|---|
We thank Manfred Bohrer, MD, from the Institute of Pathology, Klinikum Ludwigshafen, Germany, for histologic analysis.
|
FOOTNOTES
|
|---|
See Materials and Methods for pertinent disclosures.
Author contributions: Guarantors of integrity of entire study, J.F.R., G.L.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, D.H., B.P., G.L.; clinical studies, D.H., B.B., D.S., H.E.A., R.J., B.P., A.E., C.Z.; statistical analysis, D.H., B.B., D.S.; and manuscript editing, D.H., G.L.
|
References
|
|---|
- Pappalardo G, Polettini E, Frattaroli FM, et al. Magnetic resonance colonography versus conventional colonoscopy for the detection of colonic endoluminal lesions. Gastroenterology 2000;119:300–304.[CrossRef][Medline]
- Luboldt W, Steiner P, Bauerfeind P, Pelkonen P, Debatin JF. Detection of mass lesions with MR colonography: preliminary report. Radiology 1998;207:59–65.[Abstract/Free Full Text]
- Ajaj W, Pelster G, Treichel U, et al. Dark lumen magnetic resonance colonography: comparison with conventional colonoscopy for the detection of colorectal pathology. Gut 2003;52:1738–1743.[Abstract/Free Full Text]
- Fenlon HM, Nunes DP, Schroy PC III, Barish MA, Clarke PD, Ferrucci JT. A comparison of virtual and conventional colonoscopy for the detection of colorectal polyps. N Engl J Med 1999;341:1496–1503.[Abstract/Free Full Text]
- Yee J, Akerkar GA, Hung RK, Steinauer-Gebauer AM, Wall SD, McQuaid KR. Colorectal neoplasia: performance characteristics of CT colonography for detection in 300 patients. Radiology 2001;219:685–692.[Abstract/Free Full Text]
- Johnson CD, Toledano AY, Herman BA, et al. Computerized tomographic colonography: performance evaluation in a retrospective multicenter setting. Gastroenterology 2003;125:688–695.[CrossRef][Medline]
- 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;125:304–310.[CrossRef][Medline]
- 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;349:2191–2200.[Abstract/Free Full Text]
- Rex DK, Cutler CS, Lemmel GT, et al. Colonoscopic miss rates of adenomas determined by back-to-back colonoscopies. Gastroenterology 1997;112:24–28.[CrossRef][Medline]
- Hixson LJ, Fennerty MB, Sampliner RE, McGee D, Garewal H. Prospective study of the frequency and size distribution of polyps missed by colonoscopy. J Natl Cancer Inst 1990;82:1769–1772.[Abstract/Free Full Text]
- Hixson LJ, Fennerty MB, Sampliner RE, Garewal HS. Prospective blinded trial of the colonoscopic miss-rate of large colorectal polyps. Gastrointest Endosc 1991;37:125–127.[Medline]
- Rembacken BJ, Fujii T, Cairns A, et al. Flat and depressed colonic neoplasms: a prospective study of 1000 colonoscopies in the UK. Lancet 2000;355:1211–1214.[CrossRef][Medline]
TOP
ABSTRACT
INTRODUCTION
