Comprehensive diagnostic ability of endocytoscopy compared with biopsy for colorectal neoplasms: a prospective randomized noninferiority trial

• Introduction
• Patients and methods
• Patients
• Endoscopy equipment and endocytoscopy classification
• Randomization and masking
• Study procedure
• Assessment of histopathology
• Outcome measures
• Sample size and statistical methods
• Results
• Discussion
• References

Background and study aims: Endocytoscopy enables observation at 450-fold magnification during gastrointestinal endoscopy, allowing on-site “optical biopsy.” We compared the accuracies of endocytoscopy and standard biopsy for the diagnosis of colorectal neoplasms.

Patients and methods: We performed a randomized, controlled, open-label trial of patients with colorectal lesions ( ≥ 5 mm) detected during colonoscopy in a tertiary referral center. We randomly assigned the 203 detected lesions of 170 eligible patients to either the endocytoscopy or standard biopsy group. An on-site endoscopist assessed the histopathology of the endocytoscopy group lesions according to the endocytoscopic findings, whereas a pathologist later assessed standard biopsy group lesions by microscopic examination of the biopsy specimens. We calculated the diagnostic accuracies in both groups with reference to the final histopathology of the resected specimens. The primary endpoint was to determine whether the diagnostic accuracy of endocytoscopy for neoplastic lesions was noninferior to that of standard biopsy (with a predefined noninferiority margin of 10 %). Analyses were by intention-to-treat and per-protocol. The study is registered, number UMIN000003923.

Results: Overall, 102 lesions in the endocytoscopy group and 101 in the standard biopsy group were available for primary outcome analysis. There were no complications. The diagnostic accuracy of endocytoscopy for the discrimination of neoplastic lesions was 94.1 % (95 % confidence interval 87.6 % to 97.8 %), whereas that of standard biopsy was 96.0 % (90.2 % to 98.9 %), which is within the noninferiority margin (absolute difference − 1.9 %, − 8.6 % to + 5.0 %).

Conclusions: Endocytoscopy is noninferior to standard biopsy for the discrimination of neoplastic lesions. With its advantage of providing an on-site diagnosis, endocytoscopy could provide a novel alternative to standard biopsy in routine colonoscopy.

Introduction

The National Polyp Study found that colonoscopic polypectomy results in a lower than expected incidence of colorectal cancer [1]. However, because a fair percentage of polyps are hyperplastic polyps, and these are often followed up without resection, it is important to distinguish between adenomas and hyperplastic polyps. Thus, in many facilities, either standard biopsies are taken or polyps are resected prior to histopathological examination.

Current advances in endoscopic imaging modalities, such as magnifying endoscopy and narrow-band imaging (NBI), have changed these conventional diagnostic approaches. These imaging techniques, sometimes called “optical biopsies” [2] [3] have greatly improved the accuracy of diagnosis of colorectal neoplasms and allow on-site assessment of the histopathological features of lesions, obviating the need for a standard biopsy [2] [3] [4] [5]. However, they allow only indirect assessment of the morphology of crypts or vessels, and their accuracies vary. Standard biopsy has therefore remained the gold standard for reliable pretreatment diagnosis.

Endocytoscopy, which involves a contact light microscopy system integrated into the distal tip of a conventional colonoscope, is a novel emerging endoscopic system. In contrast to other modalities, the ultramagnification capability of endocytoscopy enables on-site observation not only of structural atypia, but also of cytological atypia. Endocytoscopy has demonstrated good consistency in assessing the histopathology of lesions in the gastrointestinal tract [6] [7] [8] [9] [10] [11] and shown a diagnostic accuracy of almost 100 % for discrimination of neoplastic lesions in the colorectum [8] [10] [11]. Endocytoscopy may thus be expected to provide the next-generation “optical biopsy.” However, no randomized controlled trials (RCTs) comparing endocytoscopy with standard biopsy have so far been performed.

We therefore conducted a prospective, randomized noninferiority trial to assess the potential of endocytoscopy, as compared with standard biopsy, for the diagnosis of colorectal neoplasms.

Patients and methods

Patients

The study was undertaken at Showa University Northern Yokohama Hospital, a tertiary referral center in Japan. Enrolment began in July 2010 and ended in March 2011. Consecutive patients with indications for screening, surveillance, or diagnostic colonoscopy were recruited, as permitted by the availability of the study colonoscopists and study instruments. Integrated-type endocytoscopes (CF Y-0001 or CF Y-0020-I, prototype; Olympus, Tokyo, Japan) were used for all colonoscopic examinations. Patients with undiagnosed colorectal lesions (≥  5 mm) detected during colonoscopy were eligible for inclusion. Patients who were taking antiplatelet or anticoagulant agents and those with inflammatory bowel disease were excluded. The patients underwent bowel preparation with 2 – 3 L polyethylene glycol solution. Diazepam and butylscopolamine were used intravenously for sedation and prevention of peristalsis. The study protocol was approved by the ethics committee of Showa University Northern Yokohama Hospital (No. 1006 – 02). All participants gave written informed consent and the study was conducted according to the Declaration of Helsinki.

Endoscopy equipment and endocytoscopy classification

All colonic examinations were performed by highly trained endoscopists (Y. M., N. I., K. W., Y. W., M. K.), who had carried out more than 2000 colonoscopies and 50 endocytoscopies. Integrated-type endocytoscopes have two separate observation modes: standard video endoscopy and endocytoscopy. Using a one-touch switch or hand lever, endoscopists can consecutively perform endocytoscopic observations, in addition to standard video endoscopy, without changing the scope. Endocytoscopic observation is based on the principles of contact light microscopy and the CF Y-0001 has a 450-fold magnification with a focusing depth of 50 μm and a field of view of 400 × 400 μm (CF Y-0020-I; 380-fold magnification with a field of view of 700 × 600 μm). Endocytoscopic images were obtained after staining with 1 % methylene blue and 0.05 % crystal violet, thus providing morphological images of gland duct lumens and the shape of the nuclei in the epithelial superficial layer.

Endocytoscopy provides images that are similar to micrographic images of fixed specimens stained with hematoxylin and eosin (H&E); cell nuclei are clearly stained with methylene blue and cytoplasm with crystal violet, making it easy to identify the lumens of the glands ([Fig. 1]). The endocytoscopic images were analyzed according to the classification proposed by Kudo et al. [11] and Sasajima et al. [8] ([Table 1]). In this classification, predicted histopathology was expressed using the revised Vienna classification category (VCC) [12] as follows: endocytoscopic pattern of “normal mucosa or hyperplastic polyp” corresponds to “VCC 1 or 2 (negative for neoplasia or indefinite for neoplasia)”; that of “low grade adenoma” corresponds to “VCC 3 (mucosal low grade neoplasia)”; and that of “high grade adenoma or invasive cancer” corresponds to “VCC 4 or 5 (mucosal high grade neoplasia or submucosal invasion by carcinoma).”

As a substudy, the interobserver and intraobserver agreements of the endocytoscopic diagnosis were calculated by using 50 randomly selected endocytoscopy images among the five enrolled endoscopists at intervals of 1 month.

Randomization and masking

When a lesion was detected, it was randomly assigned to either the endocytoscopy or standard biopsy group, using a sequential numbered-container method, without stratification. An independent data manager who was not involved in the study analysis prepared the numbered container, generated the randomization sequence, and issued the randomly assigned group after checking patient eligibility. The numbered container was locked, only the data manager having access. If a single patient had multiple lesions that were eligible for inclusion, each lesion was randomized separately, and a per-lesion analysis subsequently performed. Patients and clinicians were not masked to the intervention assigned.

Study procedure

In the endocytoscopy group, the lesions were observed using endocytoscopy mode and their histopathology assessed by the on-site endoscopist according to the endocytoscopic diagnosis described above. In the standard biopsy group, a standard biopsy specimen was taken from each lesion using a Radial Jaw 4 (Boston Scientific, Natick, Massachusetts, USA) and the histopathology of the standard biopsy specimens was assessed by a pathologist at a later date. Prior to these procedures, chromoendoscopy with methylene blue was performed to provide contrast in protruded or depressed portions of the subject lesion and thus allow identification of the area with the most severe histopathology for endocytoscopic observation or standard biopsy. Following colonoscopic examination, endoscopic or surgical resection was performed to obtain the reference diagnosis for each lesion.

Assessment of histopathology

In the present study, the final histopathology of the resected specimens (not that of the biopsies) was used as the gold standard for calculating the diagnostic accuracies in the endocytoscopy and standard biopsy groups. Both standard biopsy and resected histopathological specimens were stained with H & E and allocated to a single gastrointestinal pathologist (S. H.) who was blinded to the endoscopic findings. In order to achieve complete blinding of the pathologist to both standard biopsy and resected specimens, the pathologist first diagnosed only the standard biopsy specimens, then, a week later, diagnosed the resected specimens, which had been randomly mixed to ensure blinding. The pathologist then made histopathological diagnoses and classified the lesions into the following three groups according to the VCC [12]: VCC 1 or 2, VCC 3, and VCC 4 or 5. In this study, hyperplastic polyps and sessile serrated adenomas/polyps (SSA/Ps) were provisionally classified as non-neoplastic and traditional serrated adenomas were classified as neoplastic.

Outcome measures

The primary endpoint was to determine whether the diagnostic accuracy for neoplastic lesions of endocytoscopy was noninferior to that of standard biopsy. The secondary endpoint was to investigate the diagnostic ability of endocytoscopy for identifying invasive cancers which have the potential to metastasize and require surgical rather than endoscopic treatment. Lesion size, location, shape (Paris classification [13]), complications (e. g., bleeding or perforation), and procedure time were also measured and recorded. Lesion size was estimated using a nontraumatic catheter (3 mm, Olympus 6233064; Olympus) during endoscopy. Procedure time was defined as the time from detection of a lesion to completing endocytoscopic observation or obtaining a standard biopsy. As a subanalysis, the ability of endocytoscopy to discriminate low grade adenomas (VCC 3) from high grade adenomas or invasive cancers (VCC 4 or 5) was compared with that of standard biopsy.

Sample size and statistical methods

Because endocytoscopy was deemed comparable to standard biopsy with regard to accuracy in diagnosing neoplastic changes, a noninferiority design was used to assess the primary outcome. According to our own previous unpublished experience, the diagnostic accuracy of standard biopsy is around 95 %. A noninferiority margin of 10 % was predefined. Thus, it was calculated that a sample size of at least 200 lesions in total (100 per arm) was the minimum required to demonstrate noninferiority with a power of 80 % and a one-sided type I error of 2.5 %. With regard to analysis of the primary endpoint, the diagnostic accuracy of endocytoscopy was defined as noninferior to standard biopsy if the lower limit of the 95 % confidence interval (95 %CI) for the estimated difference in the primary outcome was higher than − 10 %.

Statistical analysis was based on both intention-to-treat (ITT) and per protocol. For the ITT analysis, all randomized lesions were analyzed according to their allocated intervention, whereas in the per-protocol analysis, lesions with protocol violations, such as failure to perform resection or histological analysis, were excluded. In the ITT analysis, worst-case scenario analysis was used for missing data; lesions lacking either endocytoscopic or standard biopsy findings or final histopathology of resected specimens were treated as misdiagnosed lesions for analysis. Continuous variables were compared using Student’s t test, while categorical variables were compared using the chi-squared test. As for validation of the endocytoscopic diagnosis, unweighted kappa values were used.

In this study, the validity of the statistical tests could theoretically have been affected by some patients having more than one lesion. However, because first-detected lesions were deemed not to affect the diagnostic procedures for subsequently detected polyps, it was decided to ignore possible within-patient correlations in the analysis. Mean values were expressed as mean ±  standard deviation (SD) or with a 95 %CI. A two-sided P value of ≤ 0.05 was considered statistically significant. All statistical analyses were performed using R version 2.10.0. (http://www.r-project.org).

Interim data on safety, adherence to randomization strategies, and efficacy of endocytoscopy and standard biopsy were reviewed by an independent clinician on 13 January 2011, at which point the sample size had reached 100 and there had been no major changes to the study protocol since commencement of the trial. Data were frozen for analysis on 11 June 2011. This study is reported according to an extension of the CONSORT statement [14] and is registered with UMIN Clinical Trials, number UMIN000003923 (http://www.umin.ac.jp/ctr/index.htm).

Results

[Fig. 2] shows the trial profile. In all, 669 consecutive patients were screened for inclusion, 499 being classified as ineligible. Thus, 170 patients with 203 lesions were enrolled in the study (116 men and 54 women; mean age, 63.8 ± 10.8 years). The 170 patients actually had 307 lesions altogether, 253 of which were ≥  5 mm and 54 being < 5 mm. Of the lesions ≥ 5 mm, 80.2 % (203 /253) were enrolled in the study and subject to randomization. The main reason for omitting the remainder was that these polyps were left in situ because some examinations had to be stopped prematurely due to colonic peristalsis or patient discomfort. The indications for colonoscopy were screening in 23 patients, surveillance in 33, positive fecal occult blood tests in 76, abdominal pain in 14, bloody stools in 11, constipation in 5, and other reasons in 8. The 203 lesions were randomly allocated to the endocytoscopy (n = 102) or standard biopsy groups (n = 101); all lesions were entered in the ITT analysis. Of the ITT subjects, nine (4.4 %) were excluded from the per-protocol analysis because of protocol violations (five in the endocytoscopy, four in the standard biopsy group), leaving 97 (95.1 %) in the endocytoscopy group and 97 (96.0 %) in the standard biopsy group to be included in the per-protocol analysis. Of the 196 histopathologically analyzable resected lesions, 26 (13.3 %) were non-neoplastic and 170 (86.7 %) neoplastic. Lesion baseline characteristics were well balanced between the two groups except for the rate of tubular adenoma, which was significantly lower in the endocytoscopy than the standard biopsy group (45.1 % vs. 64.4 %, P = 0.006) ([Table 2]). As for the prevalence (ratio of neoplastic lesions), there was also no significant difference between the two groups: 81.4 % (83 /102) in the endocytoscopy group vs. 86.1 % (87/101) in the standard biopsy group (P  =  0.358).

[Table 3] shows the agreement between endocytoscopic and final histopathological diagnoses in the endocytoscopy group of the per-protocol population and [Table 4] shows the agreement between diagnoses on standard biopsy specimens and the final histopathological diagnoses in the standard biopsy group of the per-protocol population. Diagnoses in both the endocytoscopy and standard biopsy groups showed good agreement with the final histopathological diagnoses: overall accuracies were 77.3 % (75/97) and 77.3 % (75/97), respectively.

[Table 5] shows the main clinical outcomes in the endocytoscopy and standard biopsy groups. Concerning the primary outcome of ability to distinguish neoplastic changes in the ITT analysis, endocytoscopy correctly identified 15 of 15 non-neoplastic lesions and 81of 83 neoplastic lesions; the final histopathological diagnoses were unknown in four patients. This represents a diagnostic accuracy of 94.1 % (96/102; 95 % CI 87.6 % to 97.8 %). Standard biopsy correctly identified 11 of 11 non-neoplastic lesions and 86 of 87 neoplastic lesions; the final histopathological diagnoses were unknown in three. This represents a diagnostic accuracy of 96.0 % (97/101; 90.2 % to 98.9 %). The absolute difference between the two groups was − 1.9 % (− 8.6 to + 5.0 %), which is well within the predefined non-inferiority margin of 10 %. Per-protocol analysis did not change these results: 99.0 % (96/97; 94.4 % to 100.0 %) in the endocytoscopy group vs. 100 % (97/97; 96.3 % to 100.0 %) in the standard biopsy group, with an absolute difference of − 1.0 % (− 4.1 % to + 2.0 %).

The sensitivities and specificities for predicting neoplastic change in both the endocytoscopy and standard biopsy groups were also excellent, there being no significant differences between them. Sensitivity was 97.6 % (81/83; 91.6 % to 99.7 %) in the endocytoscopy group and 98.9 % (86/87; 93.8 % to 100.0 %) in the standard biopsy group, whereas specificity was 100 % (15/15; 78.2 % to 100.0 %) in the endocytoscopy group and 100 % (11/11; 71.5 % to 100 %) in the standard biopsy group (P = 1.000 and P = 1.000, respectively).

As for the secondary outcome, endocytoscopy had an accuracy of 91.2 % (93/102), sensitivity of 75.0 % (15/20), and specificity of 100 % (78/78) with regard to the discrimination of invasive cancers.

Regarding the ability to distinguish low grade adenomas (VCC 3) from high grade adenomas or invasive cancers (VCC 4 or 5), the accuracies were comparable between the two groups: 73.2 % (60/82; 62.2 % to 82.4 %) in the endocytoscopy group vs. 74.4 % (64/86; 63.9 % to 83.2 %) in the standard biopsy group (P = 0.854). However, endocytoscopy showed significantly lower sensitivity and higher specificity than standard biopsy: 50.0 % (17/34; 32.4 % to 67.6 %) vs. 100 % (43/43; 91.8 to 100.0 %), P < 0.0001, respectively, for sensitivity, and 89.6 % (43/48, 77.3 % to 96.5 %) vs. 48.8 % (21/43, 33.3 % to 64.5 %), P < 0.0001, respectively, for specificity.

As for the reliability of the endocytoscopic diagnosis, there was excellent interobserver and intraobserver agreement. The kappa scores were 0.841 (0.796 to 0.886) and 0.834 (0.704 to 0.965), respectively.

No complications were observed in either group. The procedure time was significantly longer in the endocytoscopy group than in the standard biopsy group (13.4 ± 6.0 min versus 7.9 ± 4.8 min, P < 0.0001). However, it took 5.9 ± 2.9 days for confirmation of the histopathological diagnoses in the standard biopsy group, compared with the on-site diagnoses made in the endocytoscopy group. 

Discussion

The results of this study demonstrate that endocytoscopy, which was combined with sequential white light endoscopy (WLE) and chromoendoscopy, is noninferior to standard biopsy for discriminating neoplastic lesions within the predefined noninferiority margin of 10 %. This suggests that endocytoscopy has the potential to provide a valid alternative to standard biopsy in routine colonoscopy.

In our study, the accuracy of endocytoscopy for identifying neoplastic lesions was 94.1 % with a sensitivity of 97.6 % and a specificity of 100 %, according to ITT analysis. Although this degree of accuracy is high enough for routine examination, it is slightly lower than that reported previously [8] [10] [11]. This discrepancy could be because previous researchers did not perform ITT analyses. Generally, diagnostic ability according to ITT analysis tends to appear lower than that according to per-protocol analysis. Discrimination of neoplastic lesions is considered important because, with the exception of SSA/Ps [15] [16], most non-neoplastic lesions do not have malignant potential and therefore do not require resection. Recently, emerging endoscopic imaging modalities such as magnifying chromoendoscopy and NBI have demonstrated their high accuracies (of between 69 % and 96.8 %) in diagnosing neoplastic colorectal lesions [2] [3] [4] [5]. However, most of these studies are not RCTs with ITT analysis and their results vary. In the present study, endocytoscopy showed greater accuracy (94.1 % according to ITT analysis and 99.0 % according to per-protocol analysis) and could be genuine “optical biopsy” with diagnostic accuracies as good as conventional biopsies. Endocytoscopy also revealed excellent interobserver and intraobserver agreement, ensuring its reproducibility as a routine diagnostic modality.

For the endocytoscopy group lesions, we assessed the diagnostic accuracy of the sequential use of WLE, chromoendoscopy, and endocytoscopy. Kudo et.al., using downloaded endoscopic images, have already reported that endocytoscopy alone has significantly greater diagnostic accuracy than WLE alone even in experienced hands (100 % vs. 96.7 %, P = 0.015) [11]. However, that was in an experimental setting; thus, we designed the current RCT to evaluate the overall accuracy of the endocytoscopy procedure as routinely used (not endocytoscopy alone, but the sequential use of WLE, chromoendoscopy, and endocytoscopy).

The primary outcome was also supported by the secondary outcome: endocytoscopy had a high accuracy of 91.2 % for discriminating invasive lesions, with a sensitivity of 75.0 % and specificity of 100 %. This is in line with the results of a pilot study that demonstrated an accuracy of 95.9 % [11]. It is clinically crucial to identify invasive cancers because, compared with other neoplastic lesions, invasive cancers are strongly associated with nodal metastasis and should therefore be treated surgically [17] [18]. The excellent specificity of endocytoscopy could contribute to the prevention of the unnecessary surgery that is sometimes performed for even adenomas in the absence of precise diagnoses.

There were also interesting data regarding the subanalysis: endocytoscopy had accuracy comparable to that of standard biopsy (73.2 % and 74.4 %, respectively) for distinguishing low grade adenomas (VCC3) from high grade adenomas or invasive cancers (VCC 4 or 5). Regarding sensitivity and specificity, endocytoscopy showed significantly lower sensitivity and higher specificity than standard biopsy (50.0 % vs. 100 % for sensitivity, and 89.6 % vs. 48.8 % for specificity). These disparities between the two groups might have been because endoscopists tended to diagnose lesions as “high grade adenomas or worse” as a result of putting weight on a small number of swollen nuclei in the small imaging area under the ultramagnification provided by endocytoscopy. Another finding concerning the subanalysis is that standard biopsy specimens do not necessarily provide a perfect reference arm for endoscopic diagnosis: in the present study, the diagnostic accuracy of standard biopsy for discrimination of low grade adenomas was 74.4 %, not 100 %. This lack of accuracy may be because a relatively small standard biopsy specimen may not be representative of the whole resected specimen, which is sometimes heterogeneous.

Confocal laser endomicroscopy (CLE) is another attractive technique for in vivo microscopic imaging; this also allows the endoscopist to obtain real-time in vivo “optical biopsy” during ongoing endoscopy. Various studies have shown the potential of CLE [19] [20] [21] [22] [23] [24], its diagnostic accuracies for neoplastic changes ranging from 82 % to 99.2 % (most being over 90 %). Because there are some differences in study design between those studies and the present study, it is not possible to determine accurately the relative superiority and inferiority of these modalities. The fact that both devices provide images that are similar to those of fixed specimens stained with H&E explains the excellent diagnostic accuracies of both endocytoscopy and CLE. However, both modalities have some limitations in addition to their prolongation of procedure time: CLE usually requires intravenous injection of exogenous fluorescence agents and provides no direct nuclear visualization, whereas endocytoscopy requires a staining procedure which sometimes makes the lumen darker and can cause peristalsis of the colon.

The present study provides further data on complications and procedure times. We encountered no complications in either the endocytoscopy or standard biopsy groups, although a 2.2 % incidence of bleeding due to standard biopsy has previously been reported [25]. This difference may be due to the fact that patients who were taking antiplatelet or anticoagulant agents were excluded from the present study. However physicians must accept an increased risk of bleeding during routine practice if they perform standard biopsy on patients who are taking these medications. The procedure time was significantly longer in the endocytoscopy group than in the standard biopsy group. This could be because additional time is required for washing, staining, and assessing lesions when employing endocytoscopy. However, the advantage of the shorter on-site procedure time in the standard biopsy group is minor compared with the disadvantage of the total time required to confirm the histopathological diagnosis (5.9 days in the present study).

Regarding the practicability, the greatest advantage of using endocytoscopy during routine colonoscopy is the ability to make an accurate diagnosis of any lesion prior to its resection. Endocytoscopy can be useful for assessing the necessity for subsequent endoscopic treatment. Additionally, endocytoscopy can be useful for patients whose lesions cannot be biopsied or resected safely due to strong antiplatelet or anticoagulant medications. Finally, endocytoscopy could reduce the cost of taking biopsies and the workloads of pathologists. Prototype endoscopes, which might be expensive with commercial use, were used in the present study. However, savings in histopathological assessment of standard biopsy specimens would cover the extra cost of acquiring and using this equipment; for one patient with one colorectal lesion undergoing colonoscopy under Japanese National Health Insurance the endocytoscopy group procedure costs 16 100 yen whereas the standard biopsy group procedure costs 30 400 yen.

Consensus guidelines of the American Society for Gastrointestinal Endoscopy broadly advise removal of all mucosal lesions of > 5 mm irrespective of their histology [26] [27], but we have chosen to study lesions of ≥ 5 mm in the present trial. The rationale is as follows: in Japan non-neoplastic lesions such as hyperplastic, inflammatory, or juvenile polyps are sometimes left in situ, even when they are more than 5 mm in size, for various reasons (e. g., number of lesions, patient’s age, discomfort, co-morbid conditions, or antiplatelet medications). In such situations, the endoscopist usually takes tentative biopsies, which has the disadvantages of additional costs for histopathology, as well as delay in informing patients of their diagnoses. Therefore, real-time assessment of the histopathology of even lesions of  ≥ 5 mm is desirable.

The present study has several limitations. First, because colonic lesions are sometimes heterogeneous (e. g. carcinoma in adenoma or dysplasia in hyperplastic components), the target areas identified for both endocytoscopy observation and standard biopsy might not have been the sites of the worst histopathology. To reduce the bias caused by heterogeneity, we performed chromoendoscopy to define a target area in the lesion prior to endocytoscopy observation or standard biopsy. Second, both hyperplastic polyps and SSA/Ps were provisionally treated as non-neoplastic lesions in the present study. Concerning serrated polyps, SSA/Ps should be distinguished from hyperplastic polyps because they are considered to be precursors of colorectal cancers, though cytological dysplasia is not present in uncomplicated SSA/Ps [28]. However, it is currently very difficult to distinguish them endoscopically from hyperplastic polyps. The discrimination of SSA/Ps by means of endocytoscopy is currently being investigated in an ongoing trial. A third limitation was that we chose diagnostic accuracy as a primary outcome measure for evaluating the diagnostic potential of endocytoscopy, though sensitivity and specificity are clinically preferable measures. We chose accuracy because this index integrates the factors of specificity, sensitivity, and prevalence. Provided that the prevalence in the two groups is equal, accuracy can be reliably compared between two groups. In the present study, there was no statistically significant difference (81.4 % vs. 86.1 %, P = 0.358) between the prevalence (ratio of neoplastic lesions) in the endocytoscopy and standard biopsy groups. Fourth, we defined the noninferiority margin for clinical rather than statistical reasons. We defined it as 10 %, because this was clinically acceptable to both physicians and patients. Finally, only highly trained endoscopists participated; there was a very high exclusion rate due to absence of eligible lesions; and a high proportion of otherwise eligible patients refused to participate. We acknowledge that these limitations might compromise generalization of the study results.

In summary, endocytoscopy, which was combined with sequential WLE and chromoendoscopy, is as effective as standard biopsy for differentiating between neoplastic and non-neoplastic lesions. In addition, it provides the possibility of distiguishing invasive cancers. Although endocytoscopy has the disadvantage of prolonged procedure time, it allows on-site diagnosis of colorectal lesions with no complications. Endocytoscopy potentially provides an attractive alternative to standard biopsy in routine colonoscopy.

Competing interests: None

Acknowledgments

We would like to express our gratitude to Drs. Seiko Hayashi, Katsuro Ichimasa, Akihiro Yamauchi, and Yushi Ogawa, Digestive Disease Center, Showa University Northern Yokohama Hospital, Yokohama, Japan, and Naoya Toyoshima, Digestive Disease Center, Nagatsuda Kose Hospital, Yokohama, Japan, for their invaluable support for the study. The authors also thank Dr. Trish Reynolds of Edanz Group for language editing of the manuscript.

• References

• 1 Winawer SJ, Zauber AG, Ho MN. The National Polyp Study Workgroup et al. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med 1993; 329: 1977-1981
  CrossrefPubMedSearch in Google Scholar
• 2 East JE, Tan EK, Bergman JJ et al. Meta-analysis: narrow band imaging for lesion characterization in the colon, oesophagus, duodenal ampulla and lung. Aliment Pharmacol Ther 2008; 28: 854-867
  CrossrefPubMedSearch in Google Scholar
• 3 Hurlstone DP, Cross SS, Adam I et al. Efficacy of high magnification chromoscopic colonoscopy for the diagnosis of neoplasia in flat and depressed lesions of the colorectum: a prospective analysis. Gut 2004; 53: 284-290
  CrossrefPubMedSearch in Google Scholar
• 4 East JE, Suzuki N, Saunders BP. Comparison of magnified pit pattern interpretation with narrow band imaging versus chromoendoscopy for diminutive colonic polyps: a pilot study. Gastrointest Endosc 2007; 66: 310-316
  CrossrefPubMedSearch in Google Scholar
• 5 Wada Y, Kudo SE, Kashida H et al. Diagnosis of colorectal lesions with the magnifying narrow-band imaging system. Gastrointest Endosc 2009; 70: 522-531
  CrossrefPubMedSearch in Google Scholar
• 6 Neumann H, Fuchs FS, Vieth M et al. Review article: in vivo imaging by endocytoscopy. Aliment Pharmacol Ther 2011; 33: 1183-1193
  CrossrefPubMedSearch in Google Scholar
• 7 Pohl H, Koch M, Khalifa A et al. Evaluation of endocytoscopy in the surveillance of patients with Barrett's esophagus. Endoscopy 2007; 39: 492-496
  Thieme ConnectPubMedSearch in Google Scholar
• 8 Sasajima K, Kudo SE, Inoue H et al. Real-time in vivo virtual histology of colorectal lesions when using the endocytoscopy system. Gastrointest Endosc 2006; 63: 1010-1017
  CrossrefPubMedSearch in Google Scholar
• 9 Kumagai Y, Monma K, Kawada K. Magnifying chromoendoscopy of the esophagus: in-vivo pathological diagnosis using an endocytoscopy system. Endoscopy 2004; 36: 590-594
  Thieme ConnectPubMedSearch in Google Scholar
• 10 Rotondano G, Bianco MA, Salerno R et al. Endocytoscopic classification of preneoplastic lesions in the colorectum. Int J Colorectal Dis 2010; 25: 1111-1116
  CrossrefPubMedSearch in Google Scholar
• 11 Kudo S, Wakamura K, Ikehara N et al. Diagnosis of colorectal lesions with a novel endocytoscopic classification – a pilot study. Endoscopy 2011; 43: 869-875
  Thieme ConnectPubMedSearch in Google Scholar
• 12 Dixon MF. Gastrointestinal epithelial neoplasia: Vienna revisited. Gut 2002; 51: 130-131
  CrossrefPubMedSearch in Google Scholar
• 13 Baba H. The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon: November 30 to December 1, 2002. Gastrointest Endosc 2003; 58: 3-43
  CrossrefPubMedSearch in Google Scholar
• 14 Piaggio G, Elbourne DR, Altman DG et al. Reporting of noninferiority and equivalence randomized trials: an extension of the CONSORT statement. JAMA 2006; 295: 1152-1160
  CrossrefPubMedSearch in Google Scholar
• 15 Hiraoka S, Kato J, Fujiki S et al. The presence of large serrated polyps increases risk for colorectal cancer. Gastroenterology 2010; 139: 1503-1510 , 1510 e1501-1503
  CrossrefPubMedSearch in Google Scholar
• 16 Schreiner MA, Weiss DG, Lieberman DA. Proximal and large hyperplastic and nondysplastic serrated polyps detected by colonoscopy are associated with neoplasia. Gastroenterology 2010; 139: 1497-1502
  CrossrefPubMedSearch in Google Scholar
• 17 Kudo S, Tamegai Y, Yamano H et al. Endoscopic mucosal resection of the colon: the Japanese technique. Gastrointest Endosc Clin N Am 2001; 11: 519-535
  PubMedSearch in Google Scholar
• 18 Kitajima K, Fujimori T, Fujii S et al. Correlations between lymph node metastasis and depth of submucosal invasion in submucosal invasive colorectal carcinoma: a Japanese collaborative study. J Gastroenterol 2004; 39: 534-543
  PubMedSearch in Google Scholar
• 19 Goetz M, Toermer T, Vieth M et al. Simultaneous confocal laser endomicroscopy and chromoendoscopy with topical cresyl violet. Gastrointest Endosc 2009; 70: 959-968
  CrossrefPubMedSearch in Google Scholar
• 20 Shahid MW, Buchner AM, Heckman MG et al. Diagnostic accuracy of probe-based confocal laser endomicroscopy and narrow band imaging for small colorectal polyps: a feasibility study. Am J Gastroenterol 2012; 107: 231-239 Epub 2011 Nov 8
  CrossrefPubMedSearch in Google Scholar
• 21 Kuiper T, van den Broek FJ, van Eeden S et al. New classification for probe-based confocal laser endomicroscopy in the colon. Endoscopy 2011; 43: 1076-1081
  Thieme ConnectPubMedSearch in Google Scholar
• 22 Neumann H, Kiesslich R, Wallace MB et al. Confocal laser endomicroscopy: technical advances and clinical applications. Gastroenterology 2010; 139: 388-392 , 392 e381-382
  CrossrefPubMedSearch in Google Scholar
• 23 Buchner AM, Shahid MW, Heckman MG et al. Comparison of probe-based confocal laser endomicroscopy with virtual chromoendoscopy for classification of colon polyps. Gastroenterology 2010; 138: 834-842
  CrossrefPubMedSearch in Google Scholar
• 24 Kiesslich R, Burg J, Vieth M et al. Confocal laser endoscopy for diagnosing intraepithelial neoplasias and colorectal cancer in vivo. Gastroenterology 2004; 127: 706-713
  CrossrefPubMedSearch in Google Scholar
• 25 Shiffman ML, Farrel MT, Yee YS. Risk of bleeding after endoscopic biopsy or polypectomy in patients taking aspirin or other NSAIDS. Gastrointest Endosc 1994; 40: 458-462
  CrossrefPubMedSearch in Google Scholar
• 26 Rex DK, Kahi C, O’Brien M et al. The American Society for Gastrointestinal Endoscopy PIVI (Preservation and Incorporation of Valuable Endoscopic Innovations) on real-time endoscopic assessment of the histology of diminutive colorectal polyps. Gastrointest Endosc 2011; 73: 419-422
  CrossrefPubMedSearch in Google Scholar
• 27 Davila RE, Rajan E, Baron TH et al. ASGE guideline: Colorectal cancer screening and surveillance. Gastrointest Endosc 2006; 63: 546-557
  CrossrefPubMedSearch in Google Scholar
• 28 Fred TB, Fatima C, Ralph HH, Neil DD. WHO classification of tumours of the digestive system. 4th. edn. Lyon, France: International Agency for Research on Cancer; 2010
  Search in Google Scholar

Corresponding author

• References

• 1 Winawer SJ, Zauber AG, Ho MN. The National Polyp Study Workgroup et al. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med 1993; 329: 1977-1981
  CrossrefPubMedSearch in Google Scholar
• 2 East JE, Tan EK, Bergman JJ et al. Meta-analysis: narrow band imaging for lesion characterization in the colon, oesophagus, duodenal ampulla and lung. Aliment Pharmacol Ther 2008; 28: 854-867
  CrossrefPubMedSearch in Google Scholar
• 3 Hurlstone DP, Cross SS, Adam I et al. Efficacy of high magnification chromoscopic colonoscopy for the diagnosis of neoplasia in flat and depressed lesions of the colorectum: a prospective analysis. Gut 2004; 53: 284-290
  CrossrefPubMedSearch in Google Scholar
• 4 East JE, Suzuki N, Saunders BP. Comparison of magnified pit pattern interpretation with narrow band imaging versus chromoendoscopy for diminutive colonic polyps: a pilot study. Gastrointest Endosc 2007; 66: 310-316
  CrossrefPubMedSearch in Google Scholar
• 5 Wada Y, Kudo SE, Kashida H et al. Diagnosis of colorectal lesions with the magnifying narrow-band imaging system. Gastrointest Endosc 2009; 70: 522-531
  CrossrefPubMedSearch in Google Scholar
• 6 Neumann H, Fuchs FS, Vieth M et al. Review article: in vivo imaging by endocytoscopy. Aliment Pharmacol Ther 2011; 33: 1183-1193
  CrossrefPubMedSearch in Google Scholar
• 7 Pohl H, Koch M, Khalifa A et al. Evaluation of endocytoscopy in the surveillance of patients with Barrett's esophagus. Endoscopy 2007; 39: 492-496
  Thieme ConnectPubMedSearch in Google Scholar
• 8 Sasajima K, Kudo SE, Inoue H et al. Real-time in vivo virtual histology of colorectal lesions when using the endocytoscopy system. Gastrointest Endosc 2006; 63: 1010-1017
  CrossrefPubMedSearch in Google Scholar
• 9 Kumagai Y, Monma K, Kawada K. Magnifying chromoendoscopy of the esophagus: in-vivo pathological diagnosis using an endocytoscopy system. Endoscopy 2004; 36: 590-594
  Thieme ConnectPubMedSearch in Google Scholar
• 10 Rotondano G, Bianco MA, Salerno R et al. Endocytoscopic classification of preneoplastic lesions in the colorectum. Int J Colorectal Dis 2010; 25: 1111-1116
  CrossrefPubMedSearch in Google Scholar
• 11 Kudo S, Wakamura K, Ikehara N et al. Diagnosis of colorectal lesions with a novel endocytoscopic classification – a pilot study. Endoscopy 2011; 43: 869-875
  Thieme ConnectPubMedSearch in Google Scholar
• 12 Dixon MF. Gastrointestinal epithelial neoplasia: Vienna revisited. Gut 2002; 51: 130-131
  CrossrefPubMedSearch in Google Scholar
• 13 Baba H. The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon: November 30 to December 1, 2002. Gastrointest Endosc 2003; 58: 3-43
  CrossrefPubMedSearch in Google Scholar
• 14 Piaggio G, Elbourne DR, Altman DG et al. Reporting of noninferiority and equivalence randomized trials: an extension of the CONSORT statement. JAMA 2006; 295: 1152-1160
  CrossrefPubMedSearch in Google Scholar
• 15 Hiraoka S, Kato J, Fujiki S et al. The presence of large serrated polyps increases risk for colorectal cancer. Gastroenterology 2010; 139: 1503-1510 , 1510 e1501-1503
  CrossrefPubMedSearch in Google Scholar
• 16 Schreiner MA, Weiss DG, Lieberman DA. Proximal and large hyperplastic and nondysplastic serrated polyps detected by colonoscopy are associated with neoplasia. Gastroenterology 2010; 139: 1497-1502
  CrossrefPubMedSearch in Google Scholar
• 17 Kudo S, Tamegai Y, Yamano H et al. Endoscopic mucosal resection of the colon: the Japanese technique. Gastrointest Endosc Clin N Am 2001; 11: 519-535
  PubMedSearch in Google Scholar
• 18 Kitajima K, Fujimori T, Fujii S et al. Correlations between lymph node metastasis and depth of submucosal invasion in submucosal invasive colorectal carcinoma: a Japanese collaborative study. J Gastroenterol 2004; 39: 534-543
  PubMedSearch in Google Scholar
• 19 Goetz M, Toermer T, Vieth M et al. Simultaneous confocal laser endomicroscopy and chromoendoscopy with topical cresyl violet. Gastrointest Endosc 2009; 70: 959-968
  CrossrefPubMedSearch in Google Scholar
• 20 Shahid MW, Buchner AM, Heckman MG et al. Diagnostic accuracy of probe-based confocal laser endomicroscopy and narrow band imaging for small colorectal polyps: a feasibility study. Am J Gastroenterol 2012; 107: 231-239 Epub 2011 Nov 8
  CrossrefPubMedSearch in Google Scholar
• 21 Kuiper T, van den Broek FJ, van Eeden S et al. New classification for probe-based confocal laser endomicroscopy in the colon. Endoscopy 2011; 43: 1076-1081
  Thieme ConnectPubMedSearch in Google Scholar
• 22 Neumann H, Kiesslich R, Wallace MB et al. Confocal laser endomicroscopy: technical advances and clinical applications. Gastroenterology 2010; 139: 388-392 , 392 e381-382
  CrossrefPubMedSearch in Google Scholar
• 23 Buchner AM, Shahid MW, Heckman MG et al. Comparison of probe-based confocal laser endomicroscopy with virtual chromoendoscopy for classification of colon polyps. Gastroenterology 2010; 138: 834-842
  CrossrefPubMedSearch in Google Scholar
• 24 Kiesslich R, Burg J, Vieth M et al. Confocal laser endoscopy for diagnosing intraepithelial neoplasias and colorectal cancer in vivo. Gastroenterology 2004; 127: 706-713
  CrossrefPubMedSearch in Google Scholar
• 25 Shiffman ML, Farrel MT, Yee YS. Risk of bleeding after endoscopic biopsy or polypectomy in patients taking aspirin or other NSAIDS. Gastrointest Endosc 1994; 40: 458-462
  CrossrefPubMedSearch in Google Scholar
• 26 Rex DK, Kahi C, O’Brien M et al. The American Society for Gastrointestinal Endoscopy PIVI (Preservation and Incorporation of Valuable Endoscopic Innovations) on real-time endoscopic assessment of the histology of diminutive colorectal polyps. Gastrointest Endosc 2011; 73: 419-422
  CrossrefPubMedSearch in Google Scholar
• 27 Davila RE, Rajan E, Baron TH et al. ASGE guideline: Colorectal cancer screening and surveillance. Gastrointest Endosc 2006; 63: 546-557
  CrossrefPubMedSearch in Google Scholar
• 28 Fred TB, Fatima C, Ralph HH, Neil DD. WHO classification of tumours of the digestive system. 4th. edn. Lyon, France: International Agency for Research on Cancer; 2010
  Search in Google Scholar