Chromoscopy-Guided Endomicroscopy Increases the Diagnostic Yield of Intraepithelial Neoplasia in Ulcerative Colitis

GASTROENTEROLOGY 2007;132:874 882 Chromoscopy-Guided Endomicroscopy Increases the Diagnostic Yield of Intraepithelial Neoplasia in Ulcerative Colitis RALF KIESSLICH,* MARTIN GOETZ,* KATHARINA LAMMERSDORF,* CONSTANTIN SCHNEIDER,* JUERGEN BURG, MANFRED STOLTE, MICHAEL VIETH, BERNHARD NAFE, PETER R. GALLE,* and MARKUS F. NEURATH* *I. Medical Clinic, the Institute of Pathology, and the Institute for Statistics, Johannes Gutenberg University of Mainz, Germany; and the Institute of Pathology, Clinic of Bayreuth, Germany Background & Aims: Because of the large number of biopsy specimens, surveillance colonoscopy in ulcerative colitis (UC) is currently time consuming and significant flat lesions still may be missed. In this study we assessed the value of combined chromoscopy and endomicroscopy for the diagnosis of intraepithelial neoplasias in a randomized controlled trial. Methods: A total of 161 patients with long-term UC in clinical remission were randomized at a 1:1 ratio to undergo conventional colonoscopy or chromoscopy with endomicroscopy. Eight patients were excluded because of insufficient bowel preparation. In the conventional colonoscopic group (n 73), random biopsy examinations and targeted biopsy examinations were performed. In the endomicroscopy group (n 80), circumscribed mucosal lesions were identified by chromoscopy and evaluated for targeted biopsy examination by endomicroscopy. The primary outcome analysis was based on the detection of neoplasias. Results: By using chromoscopy with endomicroscopy, 4.75-fold more neoplasias could be detected (P.005) than with conventional colonoscopy, although 50% fewer biopsy specimens (P.008) were required. If only circumscribed lesions would have been biopsied in the first group, the total number of biopsy specimens could have been reduced by more than 90%. A total of 5580 confocal endomicroscopic images from 134 circumscribed lesions were compared with histologic results. The presence of neoplastic changes could be predicted by endomicroscopy with high accuracy (sensitivity, 94.7%; specificity, 98.3%; accuracy, 97.8%). Conclusions: Endomicroscopy based on in vivo histology can determine if UC lesions identified by chromoscopy should undergo biopsy examination, thereby increasing the diagnostic yield and reducing the need for biopsy examinations. Thus, chromoscopy-guided endomicroscopy may lead to significant improvements in the clinical management of UC. Ulcerative colitis (UC) is an immune cell mediated inflammatory bowel disease characterized by mucosal ulcerations, rectal bleeding, diarrhea, and abdominal pain. 1 3 Patients with long-standing UC face an increased risk for development of colitis-associated colorectal cancer. 4 7 Factors associated with increased risk for cancer development include the duration of the disease, extensive colonic involvement (pancolitis, backwash ileitis), primary sclerosing cholangitis, and severe chronic active inflammation. 4,8 Based on these observations, colonoscopic surveillance in patients with long-standing UC is highly recommended. The main objective of surveillance colonoscopy in UC is to detect neoplasia at a surgically curative and preferably preinvasive stage. However, in contrast to sporadic colorectal cancer, the growing pattern of neoplastic tissue in UC is often flat and multifocal. 9 11 Therefore, significant lesions during conventional colonoscopy in UC frequently are overlooked. 12 Chromoscopy with topically applied dyes such as methylene blue or indigo carmine facilitates the endoscopic detection of flat, circumscribed colitis-associated neoplastic changes in UC. In fact, 5 controlled studies showed that the diagnostic yield for the detection of intraepithelial neoplasia (IN) using chromoscopy is higher as compared with conventional colonoscopy with random biopsy specimens. 13 17 Based on the earlier-described studies, chromoscopy recently has been considered for incorporation into US guidelines for surveillance of patients with long-standing UC. 4 However, although this technique does allow for identification of mucosal lesions, it is not suitable for accurate endoscopic diagnosis of INs in UC because of the lack of cellular resolution and subsurface imaging. 12 For endoscopy, novel techniques allowing accurate diagnoses during ongoing examination are highly desirable and may allow appropriate and immediate therapeutic maneuvers (eg, resection vs biopsy). Recently, a miniaturized confocal microscope integrated into the distal tip of a conventional colonoscope was developed. 18 This new diagnostic technology for gastrointestinal endoscopy, denoted confocal endomicroscopy, enables histologic evaluation of the mucosal layer during ongoing colonoscopy. Furthermore, in patients screened for sporadic colorectal Abbreviation used in this paper: IN, intraepithelial neoplasia. 2007 by the AGA Institute 0016-5085/07/$32.00 doi:10.1053/j.gastro.2007.01.048

March 2007 CHROMOSCOPY GUIDED ENDOMICROSCOPY 875 Table 1. Study Criteria Inclusion criteria Clinically and histologically verified UC Disease duration 8 y Colitis Activity Index 8 23a Activity index of Truelove and Witts: mild Exclusion criteria Known IN or colorectal cancer Coagulopathy (prothrombin time 50% of control, partial thromboplastin time 50 s) Impaired renal function (creatinine level 1.2 mg/dl) Pregnancy or breast feeding Inability to obtain informed consent Known allergy to methylene blue or fluorescein a Index parameters comprise bowel frequency, urgency of defecation, blood in stool, general well being, and extracolonic features. cancer, surface and subsurface analysis at cellular and subcellular resolutions can be used to predict INs with high accuracy. 18 However, because of the time required for examination of large surface areas, this technique is not suitable for screening of the entire colonic surface in UC to detect neoplasias in flat mucosa. In the present study, we used chromoscopy to identify potential neoplastic lesions and combined this with endomicroscopy for the endoscopic diagnosis of colitisassociated INs in UC. By using such chromoscopy-guided endomicroscopy, we could diagnose 4.75-fold more neoplastic lesions in UC in comparison with conventional video colonoscopy with random biopsy specimens, although significantly (50%) fewer biopsy specimens were required. Patients and Methods Consecutive patients with clinically inactive, longstanding UC (duration, 8 y) were recruited from the outpatient clinic of the First Department of Medicine at the University of Mainz (Mainz, Germany). Suitable patients were identified using the inclusion and exclusion criteria specified in Table 1. The identified patients and their primary care physicians were invited to participate in the study, and informed consent was obtained. The study was approved by the local ethical committee in Rheinland-Pfalz, Germany (no. 837.321.03). Primary and Secondary Analyses The a priori hypothesis was to show a 3.5-fold increase in the total number of lesions diagnosed with IN using chromoscopy in conjunction with endomicroscopy. Chromoscopy was used to unmask circumscribed lesions, and the accuracy of endomicroscopy to predict histologic findings was evaluated. Secondary analyses were the comparison of the predicted extent and severity of inflamed mucosa between the 2 groups as well as the reduction in the number of biopy specimens achieved by endomicroscopy without loss of diagnostic yield. Randomization Patients were randomized by random numbers at a 1:1 ratio into groups A and B (see later) using a computer-aided system. The respective randomization results were kept in sealed envelopes that were opened directly before the colonoscopy by an independent person. Chromoscopy and endomicroscopy during colonoscopy then were performed on patients in group A and conventional colonoscopy was performed on patients in group B. Confocal Laser Endoscope The components of the confocal laser endoscope are based on the integration of a confocal laser microscope into the distal tip of a conventional video endoscope (joint venture between Pentax in Tokyo, Japan and Optiscan in Melbourne, Australia), which enables confocal microscopy on a second monitor in addition and simultaneous to standard video endoscopy (See Supplementary Figure 1 online at www.gastrojournal.org). The diameter of both the distal tip and the insertion tube was 12.8 mm. The distal tip contained an air and water jet nozzle, 2 light guides, an auxiliary water jet channel, and a 2.8-mm working channel. Actuation of imaging the plane depth relative to the surface of the tissue was controlled using 2 additional buttons on the handpiece. During laser endoscopy, a single line laser delivered an excitation wavelength of 488 nm, and the maximum laser power output was 1 mw or less at the surface of the tissue. Confocal image data were collected at a scan rate of 0.8 frames/s (1024 1024 pixels) or 1.6 frames/s (1024 512 pixels). The optical slice thickness was 7 m, with a lateral resolution of 0.7 m. The field of view was 500 500 m, and the range of the z-axis was 0 250 m below the surface layer. Group A: Chromoscopy With Endomicroscopy The confocal laser endoscope was advanced into the terminal ileum or cecum. Subsequently, 5 ml of fluorescein (Fluorescein Alcon 10% Injektionslösung; Alcon Pharma GmbH, Freiburg, Germany) at a final concentration of 10% was injected intravenously to enable confocal imaging. On withdrawal of the endoscope, all parts of the colon were evaluated and chromoscopy according to the SURFACE guidelines 12 of the entire colon was performed. Methylene blue was used for staining at a final concentration of 0.1%. The colon was stained in a segmental fashion, 30 cm of colon at a time, using a spraying catheter (Olympus PW-1L; Hamburg, Germany) that allowed a homogeneous application of dye. Excess dye was removed immediately by suction. Staining was considered adequate when the tiny glandular duct openings of the mucosa (pits) were clearly visible (See Supplementary Figure 2 online at www.gastrojournal.org). The stained segment was inspected carefully and classified for the presence of circumscribed lesions (flat, depressed, polypoid) outside severely inflamed areas. Chromoscopy

876 KIESSLICH ET AL GASTROENTEROLOGY Vol. 132, No. 3 Table 2. Confocal Pattern Classification to Predict Colorectal Pathology in Circumscribed Lesions on Chromoscopy Grading Crypt architecture Vessel architecture Normal Regeneration Neoplasia Regular luminal openings and distribution of the crypts covered by a homogeneous layer of epithelial cells, including goblet cells Star-shaped luminal crypt openings or focal aggregation of regular-shaped crypts with a regular or reduced amount of goblet cells Ridged-lined irregular epithelial layer with loss of crypts and goblet cells, irregular cell architecture with little or no mucin Hexagonal, honeycomb appearance that presents a network of capillaries outlining the stroma surrounding the luminal openings of the crypts Hexagonal, honeycomb appearance with no or mild increase in the number of capillaries Dilated and distorted vessels with increased leakage, irregular architecture with little or no orientation to adjunct tissue with methylene blue did not interfere with confocal imaging. After video endoscopic inspection, endomicroscopy was performed on circumscribed lesions in the colon. Furthermore, endomicroscopy was performed every 10 15 cm within the colon to assess the potential presence of mononuclear cells in the mucosa. The distal end of the confocal laser endoscope was placed in a gentle fashion at the mucosal surface of areas of interest. The position of the endoscope was stabilized and suction of tissue was performed to further reduce movement artifacts if necessary. The position of the imaging plane within the mucosa was adjusted by using the buttons on the endoscope. In every area of interest images from the surface to deeper parts of the mucosal layer were obtained and stored digitally in a specific folder labeled according to the site of data collection (0 110 cm from the anal verge). Circumscribed lesions were graded for the presence of INs according to cellular and vascular changes using the endoscopic confocal pattern classification (Table 2). Randomly selected areas or diffuse changes were graded for the severity of inflammatory changes according to crypt and vessel architecture and the presence of inflammatory cells within the lamina propria (Table 3). All digitally stored images within a specific area were re-assessed by a blinded investigator grading (or scoring) the quality of images. Images were scored as good (no moving artifacts, single cells can be differentiated), average (artifacts present but tissue structure can be recognized), or poor (artifacts not permitting evaluation of the image). Group B: Standard Video Endoscopy With Random Biopsy Specimens In the second group, colonoscopy was performed using conventional video endoscopes (Pentax EC 3830FK). Larger areas of inflamed mucosa were classified according to the degree of mucosal destruction. Intact mucosa without visible mucosal changes was classified as normal. A reticular surface pattern with scattered erosions or multiple erosive changes with partially preserved mucosa were classified as mild to moderate inflammatory changes. An ulcerated mucosal surface was classified as a severe inflammatory change. Biopsy Protocol Mucosal abnormalities were recorded in both groups with regard to location (distance from the anus in centimeters), morphology (polypoid, flat, depressed), and size. On withdrawal of the colonoscope from the cecum to the anus, sequential biopsy specimens were taken in a systematic fashion in both groups. In group A, endomicroscopy was performed every 10 15 cm and biopsy specimens were taken only in the presence of in vivo mucosal irregularities. However, if normal colonic mucosa appeared, the cecum, splenic flexure, and rectum had to be biopsied to determine the histological extent and severity of ulcerative colitis. In group B, every 10 cm there were 4 biopsy specimens taken, which were placed in a specimen container. Table 3. Classification to Predict Activity of Inflammation in UC Grading of inflammation Crypt architecture Cellular infiltration Vessel architecture No Mild to moderate Regular luminal openings and distribution of crypts covered by a homogeneous layer of epithelial cells, including goblet cells Differences in shape, size and distribution of crypts; increased distance between crypts, focal crypt destruction Absent Present; 50% of crypts involved Severe Unequivocal crypt destruction Present; 50% of crypts involved Normal hexagonal, honeycomb appearance that presents a network of capillaries outlining the stroma surrounding the luminal openings of the crypts Mild to moderate increase of capillaries, dilated and distorted capillaries Marked increase of dilated and distorted capillaries; leakage of fluorescein

March 2007 CHROMOSCOPY GUIDED ENDOMICROSCOPY 877 In addition, biopsy specimens were taken in a targeted fashion whenever possible in both groups depending on the presence of circumscribed lesions. Biopsy specimens from targeted areas in both groups were sent separately. All specimen containers were labeled by the distance from the anal verge (in centimeters). The endoscopist classified the degree of inflammation for each area that was sampled (ie, for each specimen container) during ongoing endoscopy. In addition, the presence of neoplastic changes was predicted prospectively for circumscribed lesions based on chromoscopic and endomicroscopic (group A) or only macroscopic (group B) changes during ongoing endoscopy. Specifically, lesions were judged as neoplastic or nonneoplastic by the endoscopist. Pathologic Evaluation The fixed biopsy specimens were oriented with a microscope for reverse light, embedded in paraffin, and sectioned vertically and transversely to facilitate the comparison between histology and confocal images. Afterward the serial sections (4 m) were stained with H&E for histopathologic examination. The pathologist was blinded to the recorded assessment of the endoscopist and classified the inflammatory activity in the samples of each specimen container into the following categories: mild inflammation, moderate inflammation, or severe inflammation. The longitudinal extent of the inflammation in the colon was assessed by the presence of inflammation in the various sequential segments sampled. Neoplastic changes were classified according to the new Vienna classification. 19 Areas that were suspicious for neoplasia were sent to a second experienced pathologist for review. In the absence of definite criteria for the differentiation between adenoma and colitis-associated dysplasia in biopsy material from patients with UC, 20 no attempt was made to differentiate between adenomas and dysplasia-associated lesional masses in this study. Instead, the term intraepithelial neoplasia according to the new Vienna classification was used. 19 The final histopathologic findings then were compared with the endoscopic assessment, with regard to the extent and activity of the inflammation and the presence of IN and colorectal cancer. Statistical Analysis Absolute and relative frequencies for qualitative variables were calculated for the patient groups. The relationship between qualitative variables and the comparison of relative figures were examined using contingency tables. Results were confirmed by using the Fisher exact test. Quantitative variables that were not distributed normally were tested for differences between groups with the Wilcoxon rank sum test. (probability for error) was set at.05. For the results of the previously defined main criteria (detection of neoplasia), a probability of error of.05 was set. For the secondary end points, test results with a P value of.05 or less (2-sided) were considered significant. Multiple adjustments were not made. Results therefore were judged in a descriptive fashion. Statistical analysis was performed by computer using the statistical software package SAS (release 6.08; SAS Institute Inc., Cary, NC). Calculation of Sample Size For the purpose of this study, we set the probability for error ( ) to.05 and the error to.1 (reflecting a power of.90). For conventional colonoscopy, a neoplasia risk of 10% in the specific patient population was assumed based on previous studies. 13 17 We defined that chromoscopy increases the diagnostic yield of neoplasia 3.5-fold as compared with conventional endoscopy. This resulted in a calculated sample size of 114 patients (57 per group). Results Patient Characteristics A total of 192 consecutive patients with previously diagnosed UC from the outpatient clinic of the First Department of Medicine at the University of Mainz were screened for possible inclusion in the present study, 31 of whom were excluded (see CONSORT statement in Supplementary Figure 3 online at www.gastrojournal.org). A total of 161 patients fulfilled all inclusion criteria and were enrolled in the study. Patients were randomly assigned at a 1:1 ratio to undergo colonoscopy with chromoscopy and endomicroscopy (group A) or conventional colonoscopy (group B). Eight patients had to be excluded from further analysis because of insufficient bowel preparation that precluded satisfactory examination, and 153 patients completed the study protocol (group A, n 80 [ 1]; group B, n 73 [ 7]). As shown in Table 4, baseline characteristics such as age, disease duration, or body mass index were not significantly different between both groups. According to histopathologic criteria, 38 patients had pancolitis (group A, 12; group B, 26), 53 patients had left-sided colitis (group A, 23; group B, 30), and 62 patients had proctitis (group A, 45; group B, 17). Thus, in spite of clinically inactive UC in all patients, there was on average more extended colonic inflammation in group B compared with group A, as determined by histologic criteria. Confocal Imaging in UC Confocal imaging allowed visualization of cells, connective tissue, and vasculature during ongoing endoscopy in healthy and diseased mucosa in UC. Moreover, inflammatory and neoplastic changes in UC could be diagnosed in vivo based on predefined characteristics (Figures 1 and 2). When comparing the endoscopic pre-

878 KIESSLICH ET AL GASTROENTEROLOGY Vol. 132, No. 3 Table 4. Patient Characteristics Group A Group B P value N 80 73 Age, y 46.2 (18 70; 23.1) 41.9 (21 65; 19.8).519 Duration of UC, y 12.3 (8 40; 9.2) 14.5 (8 44; 7.9).357 Body mass index 24.1 (18.1 28.9; 2.2) 25.9 (19.4 27.4; 3.4).405 Stool frequency per day 4 4.617 Prevalence of primary sclerosing cholangitis 6 (7.5%) 9 (12.3%).262 Maintenance mesalamine therapy 51 (63.8%) 59 (80.8%).205 NOTE. P value was nonsignificant for all parameters. Data are given as mean/median; ranges and SDs are given in parentheses. diction of the extent of inflammatory activity with the histologic results from the corresponding specimens, we found an agreement of 95.0% (76 of 80) in group A, but only of 34.2% (25 of 73) in group B (P.0001) (See Supplementary Table 1 online at www.gastrojournal.org). Biopsy specimens from 24 patients with UC showed histologic evidence of severe inflammation. Seventy-two patients had moderate inflammation and 57 patients had mild inflammation (severe inflammation, 8 vs 16 patients; moderate inflammation, 41 vs 31 patients; mild inflammation, 31 vs 26 patients in groups A and B, respectively). The overall agreement between the endoscopic prediction of disease activity and histologic findings was 92.5% (74 of 80) in group A as compared with 58.9% (43 of 73) in group B (P.046) (See Supplementary Table 2 online at www.gastrojournal.org). Confocal Imaging of Neoplasias A total of 134 circumscribed lesions could be identified after pan-chromoscopy with methylene blue (See Supplementary Figure 2 online at www.gastrojournal.org). Targeted endomicroscopic analysis of these lesions allowed the analysis of cellular and subcellular details at high resolution. Cellular vascular changes and connective tissue could be analyzed and graduated with the help of the confocal pattern classification from deeper parts of the mucosal layer up to the surface. A total of 5580 confocal images were analyzed during ongoing endoscopy and the results were compared with 311 targeted biopsy specimens. By using the confocal pattern classification, the presence of IN could be predicted with a sensitivity of 94.7%, and a specificity of 98.3% (accuracy, 97.8%), Figure 1. Comparison between endomicroscopy and histology for inflammatory changes in UC. (A) Normal crypt architecture. Typical round crypts with regular lumen (arrow) are visible. Black spots within the epithelial layer represent mucin in goblet cells. (B) Corresponding histologic specimen (H&E staining). Crypt lumen, mucin, and nuclei are present in regularly distributed crypts. (C) Dilated and distorted capillaries are brightly displayed by endomicroscopy. Black dots within the capillaries represent blood cells. Crypt architecture shows different shape and size and increased distance between the crypts. (D) Corresponding histology shows similar changes. However, delineation of single vessels is more difficult. (E) A mixed cellular infiltration is present in the lamina propria. The crypts are not affected. (F) Corresponding histology revealed similar changes. The cellular infiltration is highlighted by an arrow.

March 2007 CHROMOSCOPY GUIDED ENDOMICROSCOPY 879 Figure 2. Diagnosis of nonneoplastic and neoplastic tissue using confocal endomicroscopy in circumscribed lesions in UC. (A) Differences in shape and size of crypts are visible. In addition, enlarged luminal opening is present. However, distribution of goblet cells is normal and epithelial cells show a regular pattern indicating nonneoplastic tissue. (B) These results are confirmed ex vivo by H&E staining of biopsy specimens. (C) Ridged line epithelial layer is present with different shape and size of cells, but a clear basal border. In addition, leakage of vessels is present and depletion of goblet cells indicating neoplasia. (D) Corresponding histology revealed low-grade IN. (E) Different shape and size of epithelial cells and irregular configuration of crypts is present. In addition, the basement membrane cannot be delineated. Goblet cells are depleted. The overall pattern indicates neoplasia. (F) Corresponding histology revealed similar changes and high-grade IN was diagnosed. respectively (Table 5). The negative predictive value for mucosa with a normal endomicroscopic appearance not containing IN was 99.1%. A single circumscribed neoplastic lesion containing low-grade IN within diffuse inflammatory changes was misdiagnosed as inflammation. The blinded re-assessor judged 84% of all obtained images as good quality (4687 of 5580 images). Neoplasias A total of 23 neoplastic lesions were identified in 15 patients in groups A and B (Table 6). All lesions were INs (15 low grade, 8 high grade). With the use of chromoscopy in conjunction with endomicroscopy, significantly more INs were diagnosed in group A as compared with group B (P.005). In group A, 19 intraepithelial neoplastic lesions (12 low grade, 7 high grade) were found in 11 of the 80 patients. Fourteen of the 19 INs Table 5. Correlation Between Endomicroscopy (In Vivo Histology) and Conventional Histology for Circumscribed Lesions Histology Confocal analysis Nonneoplastic Low-grade INs High-grade INs Nonneoplastic 113 1 0 Neoplastic 2 11 7 Sensitivity, 94.7%; specificity, 98.3; positive predictive value, 90%; negative predictive value, 99.1%; accuracy, 97.8%. were not seen on conventional viewing and were identified only after chromoscopy with methylene blue as flat circumscribed lesions suspicious for neoplasia. Thus, the patient number increased significantly from 3 patients (with 5 IN) to 11 patients (with 19 IN) (P.036). In group A, the intraepithelial lesions were polypoid in 3 cases (mean size, 0.9 cm), flat in 15 cases (mean size, 1.1 cm), and 1 flat lesion had a central depression (mean size, 0.6 cm). In contrast, in group B, only 4 of the 73 patients were found to have INs and 3 low-grade and 1 high-grade INs were identified. In comparison, 2 of the 4 INs in group B were polypoid lesions (mean size, 1.1 cm) and 2 were found by random biopsy examinations. Statistical analysis proved a statistically significant increase in the number of detected flat INs in group A as compared with group B(P.002). Table 6. Intraepithelial Neoplasias Group A Group B P value N 80 73 Patients with IN 11 4.097 Number of INs a 19 4.005 Low-grade INs 12 3 High-grade INs 7 1 Polypoid INs 3 2 In flat mucosa a 16 2.002 a Fisher exact test.

880 KIESSLICH ET AL GASTROENTEROLOGY Vol. 132, No. 3 Table 7. Number of Biopsy Specimens per Group Biopsy specimens Group A (n 80) Group B (n 73) P value Total no. of biopsy specimens 1688 3081.008 No. of targeted biopsy 312 227.0001 specimens a No. of targeted biopsy specimens 62 using endomicroscopy (suspicious in vivo architecture) No. of targeted biopsy specimens 57 13.0001 containing INs Total no. of biopsy specimens containing INs 57 7.0001 a Group A circumscribed lesions after chromoendoscopy; group B visible lesions. Duration of Colonoscopy and Biopsy Specimens The mean duration of the examination for patients in group A was 42 minutes (range, 29 64 min), whereas conventional colonoscopy (group B) required 31 minutes (range, 18 48 min). This difference was not statistically significant (P.276). Endomicroscopic-guided biopsy examinations led to a total of 21.2 biopsy specimens per patient (group A), whereas 42.2 biopsy specimens were necessary on average performing random biopsy specimens (group B) (Table 7). This difference was statistically significant (P.008). Furthermore, if only circumscribed lesions would have been biopsied in group A, the total number of biopsy specimens could have been limited to 3.9 biopsy specimens per patient only, without reducing the number of identified INs (Table 7). In addition, endomicroscopy theoretically could further decrease 5 times (62 vs 312 biopsy specimens) the number of biopsy specimens needed to diagnose INs if only suspicious in vivo architecture would have been biopsied (0.78 biopsy specimens per patient). Discussion The immediate endoscopic identification and diagnosis of INs are unmet goals in the management of patients with UC, particularly those who have risk factors for cancer development. 4,11 Our results show that chromoscopy with endomicroscopy is a new diagnostic tool that enables rapid diagnosis of INs in patients with long-standing UC with high accuracy during ongoing endoscopy. Such chromoscopy-guided endomicroscopy combines the strengths of both techniques in UC. Although pan-chromoscopy with methylene blue facilitates the detection of flat lesions in UC, subsequent targeted confocal endomicroscopy can be used to differentiate between neoplastic and nonneoplastic tissue. By using this diagnostic approach, 4.75-fold more INs could be diagnosed (group A) as compared with conventional colonoscopy with random biopsy specimens (group B), despite the fact that 50% fewer biopsy specimens were taken. These findings are of particular importance in light of the finding that colorectal cancer occurs more frequently in patients with long-standing UC 4,6,11 and suggest that chromoscopy-guided endomicroscopy may lead to significant improvements in the clinical management of UC. It should be noted, however, that the primary end point of the present study focused on the number of detected INs rather than on the number of patients with neoplasias. Although highly significant differences were detected between groups A and B in the number of detected neoplasias, there was only a clear trend toward the identification of more patients with neoplasias in group A. Thus, future multicenter studies with a larger number of patients are warranted to prove that chromoendoscopy in conjunction with endomicroscopy identifies more UC patients with neoplasias as compared with standard video endoscopy. In contrast to sporadic colorectal cancer, INs in UC grow in flat mucosa and are missed easily during conventional colonoscopy. 9 Thirty to 50 random biopsy specimens in macroscopically normal mucosa have been proposed to circumvent this problem but this approach is time consuming and not well suited to identify the majority of relevant lesions in UC. 4,12 Recently, chromoscopy has been suggested as a helpful tool for the detection of lesions in flat mucosa in UC, but this technique does not allow a detailed subsurface analysis of the identified lesions. 13 17 Because endomicroscopy may be helpful to further characterize unmasked, circumscribed lesions seen after chromoendoscopy, we have combined in the present study chromoscopy as a red flag technique with confocal laser endomicroscopy to detect and analyze INs in UC during ongoing endoscopy. The advantage of this approach as compared with chromoendoscopy alone is the possibility to further select and analyze areas of interest with suspicious in vivo architecture, and this method may reduce the number of biopsy specimens required to diagnose neoplasias in UC. In fact, the number of biopsy specimens theoretically could be decreased in our study 5 times when comparing standard chromoendoscopy with chromoscopic-guided endomicroscopy (312 vs 62; Table 7), underlining the advantage of the latter technique as compared with chromoendoscopy alone. Confocal laser endomicroscopy provides confocal microscopic imaging simultaneously to the macroscopic view. 18 Subsurface visualization of circumscribed lesions in UC allowed analysis of crypt disarrangement and changes of the vascular architecture. Optical, noninvasive sectioning of the mucosal layer was possible in conjunction with the fluorescence agent fluorescein, which offers a homogenous distribution and staining of the mucosal layer. Fluorescein does not usually stain nuclei because of its pharmacologic properties; thus a differentiation between low-grade IN, high-grade IN, and mucosal cancer

March 2007 CHROMOSCOPY GUIDED ENDOMICROSCOPY 881 was not possible by evaluating the nucleus cytoplasm ratio. However, confocal imaging could be used for immediate diagnosis during colonoscopy of neoplastic changes in UC because of distinct changes of crypt architecture in circumscribed lesions. By using the confocal pattern classification, 18 the shape and size of single cells and the structure of goblet cells could be used for in vivo diagnosis of INs in UC with remarkably high sensitivity (94.7%) and specificity (98.3%). Endomicroscopy is the first diagnostic tool enabling in vivo histology of subsurface structures at the cellular and subcellular levels during ongoing endoscopy. Alternative techniques such as chromoscopy, magnifying endoscopy, or narrow band imaging can be used to predict pathologic changes to some extent, but they are limited to the mucosal surface and have failed to show convincing results for the endoscopic diagnosis of neoplasia owing to limited resolution and accuracy. 21 In the present study, the combination of chromoscopy with endomicroscopy could be used to target biopsy specimens to relevant areas because the confocal window at the distal tip of the endoscope is located adjacent to the working channel. Endomicroscopic-guided biopsy examinations significantly reduced the number of biopsy specimens by 50% and provided the pathologist with a high percentage of diagnostically positive specimens from neoplastic tissue. Furthermore, biopsy specimens from endomicroscopically normal tissue revealed with 99% predictive accuracy nonneoplastic tissue by histopathologic analysis, suggesting that random biopsy specimens in UC are not necessary if normal crypt architecture is present on endomicroscopic analysis. Endomicroscopy and chromoscopy prolonged colonoscopy for about 10 minutes on average, although this was not significantly different from the conventional colonoscopy group owing to large variability among patients. However, if biopsy specimens in group A had been limited to previously identified suspicious areas using chromoendoscopy, fewer than 4 biopsy specimens per patient (as compared with 42.2 in group B) would have been necessary. In addition, targeted endomicroscopy of chromoendoscopically defined areas theoretically could further reduce the number of biopsy specimens to less than 1 per patient, with an associated reduction of costs and efforts for histopathologic analysis. At the same time, there was a 4.75-fold increase in the diagnostic yield of neoplasias, suggesting that chromoscopy and endomicroscopy offer substantial advantages for cancer surveillance in UC over conventional colonoscopy. Furthermore, patients with multifocal neoplastic lesions could be identified and immediately sent to surgery. Because very recent data suggest that local resection of adenomas might be appropriate in patients with UC, 22 endomicroscopy also could become an ideal tool to plan endoscopic resections of lesions and avoid complications caused by resection of nonneoplastic tissue. Gastroenterologists will require additional education and close interactions with pathologists to interpret endomicroscopic images of the gastrointestinal tract. A potential disadvantage of the currently used endomicroscopy system is the limitation of the imaging plane depth to the mucosal layer. Thus, submucosal malignant infiltration of tumor cells cannot be seen. Finally, future multicenter studies will have to show the cost effectiveness of endomicroscopy for surveillance endoscopy in UC. In conclusion, an endoscopic differentiation between neoplastic and nonneoplastic changes in UC is highly desirable because in the presence of neoplasia, proctocolectomy currently is recommended. 8,22 Although several clinical, immunologic, and histologic criteria for such differentiation have been proposed, a key problem is the large number of untargeted biopsy specimens during colonoscopy that may still underestimate the frequency of INs and colitis-associated cancers in patients with UC. 4,5,10,20 The present study clearly indicates that chromoscopy in combination with endomicroscopy allows the diagnosis of significantly more INs in patients with UC as compared with standard colonoscopy with random biopsy specimens. Although chromoscopy helps to unmask flat lesions, subsequent targeted endomicroscopy identifies the presence of neoplastic changes during ongoing colonoscopy. This newly developed surveillance strategy will have important implications for medical and surgical therapeutic interventions in patients with longstanding UC. Supplementary Data Supplementary data associated with this article can be found, in the online version, at doi:10.1053/ j.gastro.2007.01.048. References 1. Podolsky DK. Medical progress: inflammatory bowel disease. N Engl J Med 2002;347:417 429. 2. Rutgeerts P, Sandborn WJ, Feagan BG, Reinisch W, Olson A, Johanns J, Travers S, Rachmilewitz D, Hanauer SB, Lichtenstein GR, de Villiers WJS, Present D, Sands BE, Colombel JF. Infliximab for induction and maintenance therapy for ulcerative colitis. N Engl J Med 2005;353:2462 2476. 3. Neurath MF, Finotto S, Glimcher LH. The role of Th1/Th2 polarization in mucosal immunity. Nat Med 2002;8:567 573. 4. Itzkowitz SH, Present DH. Consensus conference: colorectal cancer screening and surveillance in inflammatory bowel disease. Inflamm Bowel Dis 2005;11:314 321. 5. Chambers WM, Warren BF, Jewell DP, Mortensen NJ. Cancer surveillance in ulcerative colitis. Br J Surg 2005;92:928 936. 6. Ekbom A, Helmick C, Zack M, Adami HO. Ulcerative colitis and colorectal cancer. A population-based study. N Engl J Med 1990; 323:1228 1233. 7. Collins RH, Feldman M, Fordtran JS. Colon cancer, dysplasia, and surveillance in patients with ulcerative colitis. N Engl J Med 1987;316:1654 1658. 8. Heuschen UA, Hinz U, Allemeyer EH, Stern J, Lucas M, Autschbach F, Herfarth C, Heuschen G. Backwash ileitis is strongly

882 KIESSLICH ET AL GASTROENTEROLOGY Vol. 132, No. 3 associated with colorectal carcinoma in ulcerative colitis. Gastroenterology 2001;120:841 847. 9. Tytgat GN, Dhir V, Gopinath N. Endoscopic appearance of dysplasia and cancer in inflammatory bowel disease. Eur J Cancer 1995;31A:1174 1177. 10. Bernstein CN, Shanahan F, Weinstein WM. Are we telling patients the truth about surveillance colonoscopy in ulcerative colitis? Lancet 1994;343:71 74. 11. Itzkowitz SH, Harpaz N. Diagnosis and management of dysplasia in patients with inflammatory bowel diseases. Gastroenterology 2004;126:1634 1648. 12. Rutter M, Bernstein C, Matsumoto T, Kiesslich R, Neurath MF. Endoscopic appearance of dysplasia in ulcerative colitis and the role of staining. Endoscopy 2004;36:1109 1114. 13. Kiesslich R, Fritsch J, Holtmann M, et al. Methylene blue aided chromoendoscopy for the detection of intraepithelial neoplasia and colon cancer in ulcerative colitis. Gastroenterology 2003; 124:880 888. 14. Hurlstone DP. Further validation of high-magnification-chromoendoscopic colonoscopy for the detection of intraepithelial neoplasia and colon cancer in ulcerative colitis. Gastroenterology 2004; 126:376 378. 15. Rutter MD, Saunders BP, Schofield G, et al. Pancolonic indigo carmine dye spraying for the detection of dysplasia in ulcerative colitis. Gut 2004;53:256 260. 16. Sada M, Igarashi M, Yoshizawa S, et al. Dye spraying and magnifying endoscopy for dysplasia and cancer surveillance in ulcerative colitis. Dis Colon Rectum 2004;47:1816 1823. 17. Hurlstone DP, Sanders DS, Lobo AJ, McAlindon ME, Cross SS. Indigo carmine-assisted high-magnification chromoendoscopic colonoscopy for the detection and characterisation of intraepithelial neoplasia in ulcerative colitis: a prospective evaluation. Endoscopy 2005;12:1186 1192. 18. Kiesslich R, Burg J, Vieth M, Gnaendiger J, Enders M, Delaney P, Polglase A, McLaren W, Janell D, Thomas S, Nafe B, Galle PR, Neurath MF. Confocal laser endoscopy for diagnosing intraepithelial neoplasias and colorectal cancer in vivo. Gastroenterology 2004;127:706 713. 19. Schlemper RJ, Riddell RH, Kato Y, Borchard F, Cooper HS, Dawsey, SM, Dixon MF, Fenoglio-Preiser CM, Flejou JF, Geboes K, Hattori T, Hirota T, Itabashi M, Iwafuchi M, Iwashita A, Kim YI, Kirchner T, Klimpfinger M, Koike M, Lauwers GY, Lewin KJ, Oberhuber G, Offner F, Price AB, Rubio CA, Shimizu M, Shimoda T, Sipponen P, Solcia E, Stolte M, Watanabe H, Yamabe H. The Vienna classification of gastrointestinal epithelial neoplasia. Gut 2000;47:251 255. 20. Vieth M, Stolte M, Mueller E, Mueller J. Bioptical differential diagnosis of adenomas, dysplasias and carcinomas in patients with ulcerative colitis. Leber Magen Darm 2000;3:125 132. 21. Kiesslich R, Hoffman A, Neurath MF. Colonoscopy, tumors, and inflammatory bowel disease new diagnostic methods. Endoscopy 2006;38:5 10. 22. Vieth M, Behrens H, Stolte M. Sporadic adenoma in ulcerative colitis: endoscopic resection is an adequate treatment. Gut 2006;55:1151 1155. 23. Walmsley RS, Ayres RCS, Pounder RE, Allan RN. A simple clinical colitis activity index. Gut 1998;43:29 32. Received August 15, 2006. Accepted December 7, 2006. Address requests for reprints to: Ralf Kiesslich, MD, PhD, I. Medical Clinic, University of Mainz, Langenbeckstr. 1, 55131 Mainz, Germany. e-mail: info@ralf-kiesslich.de; fax: (49) 6131-175552. Presented in part as a plenary lecture during Digestive Disease Week, May 16, 2005, Chicago, IL.