To date, no reports have compared the diagnostic efficacy of narrow-band imaging (NBI) and i-scan for the histologic prediction of intermediate-to-large colorectal polyps. We aimed to compare the diagnostic accuracy of NBI and i-scan in predicting histology, and their inter-/intra-observer agreement.
We performed a prospective, randomized study that included 66 patients (NBI,
The overall diagnostic accuracies in the NBI and i-scan groups were 73.7% (73/99) and 75.8% (75/99), respectively, and there was no statistical significance between the two groups (
NBI and i-scan have similar diagnostic accuracies for the histologic prediction of intermediate-to-large colorectal polyps. Furthermore, the inter-/intra-observer agreement was acceptable for both modalities when the JNET classification was applied.
Colonoscopy is considered an optimal modality for the detection and treatment of colorectal neoplasms [
Narrow-band imaging (NBI; Olympus, Tokyo, Japan) is an IEE technique that has been widely used in clinical practice. NBI is a method of image enhancement that modifies white light via narrowed bandwidth filters to allow the visualization of the surface pattern and capillary vessels. This enables endoscopists to identify neoplastic lesions [
One previous report compared the diagnostic efficacy of NBI and i-scan in terms of their ability to predict the histology of diminutive colonic polyps [
We performed a prospective, randomized pilot study to compare NBI and i-scan for the prediction of the histology of colorectal polyps during colonoscopic resection. A total of 66 patients (one lesion per patient) who underwent colonoscopic resection for intermediate-to-large colorectal polyps between December 2019 and June 2020 were considered eligible. The inclusion criteria were as follows: (1) age >18 years, (2) colonoscopic resection for the treatment of colorectal polyps, (3) polyp size >10 mm but <50 mm, and (4) no underlying disease that is associated with a higher likelihood of bleeding. The exclusion criteria were as follows: (1) previous diagnosis of a colorectal malignancy, familial adenomatous polyposis, or inflammatory bowel disease; (2) history of diverticulosis and radiation treatment; (3) pregnant status in women; and (4) refusal to participate in this study. All procedures were performed by three experienced (>5 years of therapeutic endoscopy experience) endoscopists (JSL, SWJ, and YHK) who were familiar with the IEE modalities of NBI and i-scan. The study was reviewed and approved by the institutional review board of Kyungpook National University Hospital (KNUCH 2019-06-017).
We adapted the JNET classification for histologic prediction. JNET type 1 is characterized by changes in normal vessels and mucosal patterns, as well as the presence of dark and white spots. Type 1 indicates a hyperplastic polyp (HP) or a sessile serrated adenoma/polyp (SSA/P). Type 2A is characterized by vessels of a regular caliber and a normal distribution pattern. Type 2A indicates low-grade dysplasia (LGD) and includes tubular adenoma and tubulovillous adenoma. Type 2B is characterized by vessels of a variable caliber with an irregular distribution pattern and an obscure mucosal pattern. Type 2B indicates high-grade dysplasia (HGD), intramucosal cancer (IMC), and superficial submucosal invasive cancer (sSMIC). Type 3 is defined as a vessel pattern that is typified by areas of interrupted thick vessels and a mucosal pattern involving amorphous areas. Type 3 indicates deep submucosal invasive cancer (dSMIC) [
The i-scan method combines high-resolution colonoscopy with three adjustable modes of image enhancement: SE, CE, and TE. The TE mode is similar to NBI and may be suitable for the characterization and prediction of the histology of colorectal polyps [
Before study initiation, the three endoscopists participated in an interactive training session that included 20 representative images. The images comprised five images of each of the four JNET classifications acquired with NBI and i-scan. The patients were randomly assigned to the NBI group (
The primary outcome of this study was the comparison of the diagnostic accuracy of the histologic prediction of NBI and i-scan in intermediate-to-large colorectal polyps. The secondary outcome was the estimation of the inter-/intra-observer agreement of the JNET classification between the NBI and i-scan methods.
No previous studies on which the sample size should be based have been published; thus, the number of enrolled patients was determined according to general feasibility. The chi-square test or Fisher’s exact test was used for categorical variables, the results of which are presented as absolute values and percentages. Quantitative data are summarized as means and standard deviations. For comparison, two-sample
The baseline characteristics of the enrolled patients are summarized in
The diagnostic accuracies of NBI versus i-scan are shown in
The JNET classification as applied to NBI and i-scan showed a substantial level of inter-observer agreement (NBI κ-value 0.612,
This study demonstrated that NBI and i-scan have comparable efficacy in predicting the histology of intermediate-to-large colorectal polyps. The overall diagnostic accuracies of NBI and i-scan were 73.7% and 75.8%, respectively. In addition, the inter-observer and intra-observer agreements were acceptable among experienced endoscopists. To our knowledge, this is the first randomized study to compare the diagnostic accuracy and agreement between NBI and i-scan for intermediate-to-large colorectal polyps.
Typically, according to clinical practice consensus, adenoma, IMC, and sSMIC are candidates for endoscopic resection. Additionally, dSMIC should be surgically treated because of potential regional lymph node metastasis [
In this regard, the use of IEE can more accurately predict histology. IEE offers the advantage of convenience because it does not require additional steps such as the application of dye to the mucosa. To date, the usefulness of NBI for predicting the histology of colorectal polyps has been documented in several reports [
The i-scan technology, another IEE modality, was recently introduced to clinical practice. To date, several reports have been published on the efficacy of the histologic prediction ability of i-scan [
We estimated the inter-observer and intra-observer agreement for the histologic prediction of colorectal polyps using the JNET classification. The κ-value of the inter-observer agreement among three experienced endoscopists was 0.612 for NBI and 0.662 for i-scan, which indicates acceptable agreement. This result is in concordance with the findings of previous reports. A retrospective analysis of the JNET 2B classification suggested an inter-observer agreement value of 0.743 [
Several important strengths of our study should be acknowledged. To our knowledge, our study is the first to estimate and compare the diagnostic accuracy of NBI and i-scan in intermediate-to-large colorectal polyps using the JNET classification. As mentioned above, no validated classification of colorectal polyps has been performed with i-scan. The current study showed that the two modalities yielded comparable diagnostic accuracies, particularly when the JNET classification was used for the i-scan mode settings. Additionally, unlike previous retrospective reports in which still images were reviewed, we analyzed the diagnostic accuracy and inter-/intra-observer agreement using a video recording system, which can enhance the reliability and reproducibility of the results.
Despite these strengths, our study also had some limitations. This study was performed by three experienced endoscopists who used NBI and i-scan at a tertiary referral center. In addition, the small sample size might have resulted in misleading diagnostic yields, including positive predictive value, negative predictive value, and accuracy. Therefore, further large-scale trials conducted by endoscopists with different levels of experience are needed to determine the superiority and generalizability of IEE. Moreover, the current study did not alternatively compare each lesion with the different modalities of NBI and i-scan. In addition, the current study was performed using first-generation NBI and an advanced version of i-scan. Thus, a technical gap between the two modalities possibly exists, which may have influenced the results. However, the JNET classification was introduced using first-generation NBI, and consequently, application of the JNET classification to i-scan with this generation of NBI is needed to reduce bias [
In conclusion, this study demonstrated that NBI and i-scan have comparable diagnostic accuracies for the histologic prediction of intermediate-to-large colorectal polyps. Furthermore, the inter-observer agreement and intra-observer agreement were acceptable for both NBI and i-scan when the JNET classification was applied.
Conceptualization: Joon Seop Lee, Seong Woo Jeon
Data curation: JSL
Formal analysis: JSL
Funding acquisition: SWJ
Investigation: SWJ
Methodology: JSL
Project administration: JSL, SWJ, Yong Hwan Kwon
Writing-original draft: JSL
Writing-review&editing: SWJ, YHK
This work was supported by Pentax Medical, Tokyo, Japan.
Representative images of white-light endoscopy and narrow-band imaging (NBI) obtained from the enrolled patients. (A) Japan NBI Expert Team (JNET) classification 1. (B) JNET classification 2A. (C) JNET classification 2B. (D) JNET classification 3.
Representative images of white-light endoscopy and i-scan obtained from the enrolled patients. (A) Japan Narrow-band Imaging Expert Team (JNET) classification 1. (B) JNET classification 2A. (C) JNET classification 2B. (D) JNET classification 3.
Baseline Characteristics of the Enrolled Patients
NBI ( |
i-scan ( |
||
---|---|---|---|
Sex, male | 18 (54.5) | 21 (63.6) | 0.453 |
Age, yr | 62.50±8.63 | 61.50±11.35 | 0.913 |
Comorbidity | 0.300 | ||
None | 15 (45.5) | 22 (66.7) | 0.129 |
Hypertension | 6 (18.2) | 7 (21.2) | |
Diabetes mellitus | 4 (12.1) | 2 (6.1) | |
Others |
8 (24.2) | 2 (6.1) | |
History of abdominal surgery | 1 (3.0) | 5 (15.2) | 0.087 |
Lesion size, mm | 18.20±6.98 | 19.80±7.13 | 0.264 |
Lesion morphology | 0.473 | ||
Pedunculated | 4 (12.1) | 3 (9.1) | |
Semi-pedunculated | 9 (27.3) | 12 (36.4) | |
Sessile | 10 (30.3) | 7 (21.2) | |
LST-G | 5 (15.2) | 9 (27.3) | |
LST-NG | 5 (15.2) | 2 (6.1) | |
Endoscopist who performed the procedures | 0.422 | ||
Endoscopist 1 | 16 (48.5) | 12 (36.4) | |
Endoscopist 2 | 8 (24.2) | 7 (21.2) | |
Endoscopist 3 | 9 (27.3) | 14 (42.4) |
Data are summarized as mean±standard deviation or number (%).
LST-G, laterally spreading tumor granular type; LST-NG, laterally spreading tumor non-granular type; NBI, narrow-band imaging.
Including cerebrovascular disease, chronic kidney disease, gout, hypothyroidism, and hyperthyroidism.
Diagnostic Accuracy of Narrow-Band Imaging and i-scan according to Endoscopist
Narrow-band imaging | i-scan | ||
---|---|---|---|
All endoscopists | 73/99 (73.7) | 75/99 (75.8) | 0.744 |
Endoscopist 1 | 25/33 (75.8) | 24/33 (72.7) | 0.778 |
Endoscopist 2 | 24/33 (72.7) | 26/33 (78.8) | 0.566 |
Endoscopist 3 | 24/33 (72.7) | 25/33 (75.8) | 0.778 |
Data are
Diagnostic Yield of Narrow-Band Imaging and i-scan according to Japan Narrow-Band Imaging Expert Team Classification
Sensitivity | Specificity | PPV | NPV | Accuracy | |
---|---|---|---|---|---|
NBI | |||||
JNET 1 ( |
100.0 | 100.0 | 100.0 | 100.0 | 100.0 |
JNET 2A ( |
66.7 | 77.8 | 63.2 | 80.3 | 66.7 |
JNET 2B ( |
72.6 | 81.3 | 81.3 | 73.3 | 72.5 |
JNET 3 ( |
100.0 | 100.0 | 100.0 | 100.0 | 100.0 |
i-scan | |||||
JNET 1 ( |
88.9 | 97.8 | 86.7 | 98.9 | 88.9 |
JNET 2A ( |
73.3 | 82.1 | 42.1 | 94.6 | 73.3 |
JNET 2B ( |
73.2 | 82.6 | 91.6 | 54.9 | 73.6 |
JNET 3 ( |
100.0 | 100.0 | 100.0 | 100.0 | 100.0 |
Data are %.
JNET, Japan NBI Expert Team; NBI, narrow-band imaging; NPV, negative predictive value; PPV, positive predictive value.
Inter-/Intra-Observer Agreements of Narrow-Band Imaging and i-scan according to Japan Narrow-Band Imaging Expert Team Classification
NBI | i-scan | |||
---|---|---|---|---|
Inter-observer agreement | 0.612 | 0.001 | 0.662 | 0.002 |
Intra-observer agreement | ||||
Endoscopist 1 | 0.467 | 0.016 | 0.571 | 0.028 |
Endoscopist 2 | 0.385 | 0.168 | 0.741 | 0.013 |
Endoscopist 3 | 0.660 | 0.004 | 0.364 | 0.085 |
NBI, narrow-band imaging.