Blue-light imaging compared with high-definition white light for real-time histology prediction of colorectal polyps less than 1 centimeter: a prospective randomized study

ORIGINAL ARTICLE

Blue-light imaging compared with high-definition white light for real-time histology prediction of colorectal polyps less than 1 centimeter: a prospective randomized study Emanuele Rondonotti, MD, PhD,1 Silvia Paggi, MD,1 Arnaldo Amato, MD,1 Giuseppe Mogavero, MD,1,2 Alida Andrealli, MD,1 Francesco Simone Conforti, MD,1,3 Dario Conte, MD,1 Giancarlo Spinzi, MD,1 Franco Radaelli, MD1 Como, Italy

Background and Aims: Blue-light imaging (BLI) is a new chromoendoscopy technique, potentially useful for differentiating neoplastic from nonneoplastic lesions. The present study was aimed at comparing BLI with high-definition white light (HDWL) in the real-time histology prediction of colon polyps <10 mm. Methods: Consecutive outpatients undergoing colonoscopy with the ELUXEO 7000 endoscopy platform and 760 series video colonoscopes (Fujifilm Co, Tokyo, Japan) who had at least 1 polyp <10 mm were randomized to BLI or HDWL for polyp characterization. The accuracy of high-confidence real-time histology prediction (adenoma vs not adenoma) by either BLI or HDWL for polyps <10 mm (primary end-point) and diminutive (
5 mm) polyps was calculated, along with sensitivity, specificity, and positive and negative predictive values, with histopathology as the reference standard. Results: A total of 483 polyps were detected in 245 randomized patients (125 and 120 in the BLI and HDWL arms, respectively). A total of 358 were diminutive, and 283 were adenomas. Overall, 222 (85.7%) and 193 (86.1%) polyps were characterized with high confidence by BLI and HDWL, respectively (P Z .887), with an overall accuracy of 92% and 84%, respectively (P Z .011). The accuracy was significantly higher by BLI than HDWL, also for diminutive polyps (92% vs 83%; P Z .008). When BLI was used, the negative predictive value for diminutive rectosigmoid polyps was 88%, and the post-polypectomy surveillance interval was correctly attributed in 85.7% and 93.7% of patients, respectively, according to U.S. and European guidelines. Conclusion: BLI was superior to HDWL for the real-time prediction of histology in polyps <10 mm. A BLIdedicated classification might further improve the endoscopist performance. (Clinical trial registration number: NCT03274115.) (Gastrointest Endosc 2018;-:1-11.)

Abbreviations: BLI, blue-light imaging; HDWL, high-definition white light; NBI, narrow-band imaging; PIVI, Preservation and Incorporation of Valuable Endoscopic Innovations; SSL, sessile serrated lesions; WASP, workgroup on serrated polyps and polyposis. DISCLOSURE: All authors disclosed no financial relationships relevant to this article. Copyright ª 2018 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 https://doi.org/10.1016/j.gie.2018.09.027 Received June 4, 2018. Accepted September 20, 2018. Current affiliations: Gastroenterology Unit, Valduce Hospital, Como (1), Gastroenterology Unit, Biomedical Department of Internal and Specialized Medicine (Di.Bi.M.I.S.), University of Palermo, Palermo (2), Department of Pathophysiology and Transplantation, Università degli Studi di Milano, Milan, Italy (3). Reprint requests: Emanuele Rondonotti, MD, PhD, Gastroenterology Unit, Valduce Hospital, Via Dante 10, 22100 Como, Italy.

www.giejournal.org

BACKGROUND AND AIM During the past decades, several imaging techniques have been developed and tested in order to enable endoscopists to accurately differentiate neoplastic from nonneoplastic lesions and to predict deep submucosal invasion of cancer. In particular, digital chromoendoscopy techniques, such as Narrow-Band Imaging (NBI),1-4 i-Scan,1,4-6 and Fujifilm Intelligent ChromoEndoscopy1,4,7 have been demonstrated to be readily applicable in routine practice by properly trained endoscopists in order to accurately predict the histology of small and diminutive lesions, even without optical magnification.1-6 The possibility of real-time prediction of polyp histology is clinically relevant because diminutive polyps represent the majority of polyps detected during colonoscopy, Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 1

Blue-light imaging

Rondonotti et al

and they have a very low risk of harboring advanced histology.8-10 Thus, an optical diagnosis would allow diminutive polyps to be resected and discarded without pathology assessment or to be left in place without resection in the case of diminutive distal hyperplastic polyps. Discarding or leaving-in most diminutive lesions without pathology assessment has the potential for large cost saving with minimal risk. Indeed, several professional societies (eg, American Society for Gastrointestinal Endoscopy, European Society of Gastrointestinal Endoscopy, National Institutes of Health) endorsed this policy, providing that the endoscopist has been trained to use virtual chromoendoscopy and has been accredited to use the technique under a national accreditation pathway.1,4,11 Recently, blue-light imaging (BLI), a new chromoendoscopy technique based on the direct (ie, not filtered) emission of blue light with a short wavelength (410 nm) selectively absorbed by hemoglobin, has been integrated in the latest generation ELUXEO 7000 endoscopy platform (Fujifilm Co, Tokyo, Japan). This technique, combined with the new high-performance 4-LED Multi Light system (Fujifilm), generates bright images with enhanced visibility of both microvascular and superficial mucosal patterns (Figs. 1 and 2). Based on these features, this technique might be useful for polyp histology prediction, as observed for other chromoendoscopy techniques.12-15 Retrospective data from a video-based study suggest that BLI without magnification can improve diagnostic accuracy in discriminating neoplastic from non-neoplastic polyps, as compared with high-definition white light (HDWL).12-15 The present randomized controlled trial was planned to determine whether BLI is superior to HDWL for realtime histology prediction of polyps <10 mm during colonoscopy.

Figure 1. Six-millimeter sessile sigmoid polyp evaluated with highdefinition white-light imaging.

PATIENTS AND METHODS This prospective, randomized controlled trial was conducted in a single open-access endoscopy center in Northern Italy. The institutional review board approved the protocol, which was registered in clinicaltrials.gov (NCT03274115). All patients provided their written informed consent. This was a no-profit study, and no funding was received or solicited. The study is reported according to the CONSORT criteria.

Figure 2. Six-millimeter sessile sigmoid polyp (same polyp as in Fig. 1) evaluated with blue-light imaging.

anti-thrombotic therapy precluding polyp resection, patient refusal, or incomplete colonoscopy. Among included patients, those with 1 polyp <10 mm were eventually randomized to receive histology prediction of all polyps <10 mm by either HDWL or BLI.

Patients All adult consecutive outpatients referred for elective colonoscopy for routine indications (eg, screening, fecal immunochemical test positivity, surveillance, symptoms) were assessed for eligibility. Exclusion criteria were as follows: previous colon resection, inadequate bowel preparation (defined as Boston Bowel Preparation Scale score <2 in at least 1 colon segment),16,17 history of inflammatory bowel disease or polyposis syndrome, ongoing 2 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

Colonoscopy procedures Four endoscopists (A.A., E.R., F.R., S.P.) participated in the present study. All had extensive experience with the use of NBI and were familiar with the Narrow-band Imaging International Colorectal Endoscopic Classification (NICE)18 and the Workgroup on Serrated Polyps and Polyposis (WASP) Classification.19 They had also participated in previous studies focused on colon polyp www.giejournal.org

Rondonotti et al

characterization with NBI.20,21 Before entering the study, all endoscopists had performed at least 50 colonoscopies with endoscopes equipped with the BLI system to become acquainted with this new technique. All procedures were performed with the ELUXEO 7000 endoscopy platform (including video processor ELUXEO VP-7000 and light source ELUXEO BL-7000; Fujifilm), high-definition 26-inch monitors (Barco NV, Kortijk, Belgium), and ELUXEO 760 series video colonoscopes with or without magnification up to 135 (EC-760ZPV and EC-760RV, respectively; Fujifilm). Bowel preparation consisted of a split regimen of low-volume polyethylene glycol solution plus bisacodyl (LOVOL-Dyl, LOVOL-Esse, AlfaSigma S.p.A, Milan, Italy). All procedures were performed with the patient under conscious sedation combining intravenous pethidine (dose range 25.0-50.0 mg) and intravenous midazolam (dose range 2.5-5.0 mg) or without sedation in case of contraindication or patient preference. Both colonoscope insertion and withdrawal were performed with carbon dioxide insufflation.

Blue-light imaging

color relative to surrounding mucosa (eg, same, lighter, brown/dark), (2) vascular pattern (eg, none, isolated lacy vessels, brown vessels surrounding white structures, areas with disrupted or missing vessels), and (3) surface pattern (eg, homogenous absence of pattern; dark or white spots of uniform size; oval, tubular, or branched white structures surrounded by brown vessels; or amorphous surface pattern). For each polyp categorized as non-adenoma, the endoscopist was eventually asked to predict whether it was a sessile serrated lesion (SSL) according to the presence of typical morphologic features (eg, mucus cap, clouded surface, indistinct border, irregular shape, dark spot inside crypts). All polyps detected during the insertion phase were evaluated for histology prediction during the withdrawal phase. Polyps were classified according to the Paris classification22 and were resected en bloc (by using forceps for <3 mm polyps or snare, cold or hot as appropriate, for larger ones), retrieved in separate specimen jars, and sent for histopathology assessment. Two expert pathologists, who had already participated in previous studies on polyp characterization, evaluated all the resected polyps.

Study design, randomization, and masking Cecal intubation was achieved by using white light as routine practice; it was confirmed by the documentation of cecal landmarks (ileocecal valve, appendicular orifice, or terminal ileum). When, during endoscope withdrawal, the first polyp <10 mm was identified (the polyp size was estimated by comparing the polyp to open standardcapacity forceps or snare), the patient was randomized (1:1) to receive polyp histology prediction by either HDWL or BLI by instructions given in an opaque, sealed envelope containing the allocation. The randomization sequence was generated through a computerized randomization scheme by using block sizes of 5. In patients randomized to HDWL, the endoscopist performed both polyp detection and polyp characterization by HDWL. Conversely, in patients randomized to BLI, the colon inspection was performed with HDWL, and the endoscopist switched to BLI only to perform polyp characterization. All further polyps <10 mm identified in the same patient during withdrawal were evaluated for histology prediction by using the same technique, according to patient randomization. Zoom magnification was always used and recorded for either BLI or HDWL evaluation whenever available. Regardless of the technique applied to predict polyp histology, each polyp was classified in real time by the endoscopist as adenoma or non-adenoma, and the endoscopist level of confidence (low or high) in predicting polyp histology was recorded. The analysis was restricted to polyps evaluated with high confidence only. Because at the time of study initiation there were no standardized and validated BLI-dedicated classifications, when predicting polyp histology with BLI, the endoscopist systematically assessed the following features, according to standard criteria for narrow-band imaging observation: (1) www.giejournal.org

Outcome measures The primary outcome measure was the endoscopist accuracy when using BLI and HDWL in predicting the histology of polyps <10 mm with high confidence, with pathology assessment as the reference standard. Sensitivity, specificity, and positive and negative predictive values also were calculated. The secondary outcomes were operative characteristics (sensitivity, specificity, and positive and negative predictive value as well as overall accuracy) for all diminutive (
5 mm) and for diminutive rectosigmoid polyps. The accuracy in predicting the serrated histology of polyps categorized as non-adenomas also was calculated. The post-polypectomy surveillance interval based on highconfidence histology prediction for polyps
5 mm, combined with the histopathology assessment of polyps >5 mm and for the diminutive ones characterized with low confidence was calculated according to both U.S. Multi-Society Task Force on Colorectal Cancer23 and European Society of Gastrointestinal Endoscopy guidelines,24 which are summarized in Appendix 1 (available online at www. giejournal.org). This was compared with the interval that would be recommended by using pathology findings for all identified polyps. Subgroup analyses according to the use of magnification and the endoscopist involved in the study were planned.

Sample size calculation and statistical analysis Based on current data, the accuracy of optical diagnosis by using HDWL in discriminating adenomas from non-adenomas was assumed to be 75%.25-27 An absolute difference of 15% between HDWL and BLI performances was considered as clinically relevant. Accordingly, at least Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 3

Blue-light imaging

Rondonotti et al

Outpatients referred for colonoscopy during the study period n=766

166 patients excluded as: • • • • •

Patients eligible n=600

on antithrombotic therapy (n=39) history of inflammatory bowel disease (n=17) history of polyposis (n=4) incomplete colonoscopies (n=53) poor bowel prep (n=53)

355 patients excluded as: • patients without polyps (n=323) or with only ≥10 polyps only (n=32)

Patients randomized n=245

HDWL Group n=120

BLI Group n=125

Polyps <10 mm detected n=224

Polyps <10 mm detected n=259

Polyps <10 mm evaluated with high confidence n=193

Polyps <10 mm evaluated with high confidence n=222

Figure 3. Patient and polyp flow. HDWL, high-definition white light; BLI, blue-light imaging.

100 polyps per group were estimated as necessary to have 80% power, assuming a type I error rate of 0.05 by 2-sided tests. Prudentially assuming a detection of only 1 polyp <10 mm per randomized patient and a low confidence rate of 15%,3,25,27 we planned to randomize 230 patients. Because in our center in the last 12 months, the percentage of patients with at least 1 polyp <10 mm was 38.3%, we planned to enroll 600 patients. Frequencies and percentages were used to summarize categorical variables, whereas quantitative variables were summarized by using means and standard deviation. The Fisher exact test or chi-square test (with Yates’s correction, when appropriate) were used to compare categorical variables, as appropriate. The t test was used to compare continuous variables. A P value < .05 was considered as statistically significant. The descriptive statistic analyses were carried out with Excel software (Office package V.2016) (Microsoft Co, Redmonton, Wash, USA), whereas the analysis of contingency tables was performed through the Vassarstat Statistical Computation Web site, available online at www.vassarstat.net. 4 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

RESULTS Patients As detailed in Figure 3, from July 2017 to December 2017, the 4 involved endoscopists evaluated a total of 766 outpatients referred for colonoscopy. After the exclusion of 166 of them (Fig. 3), 600 patients were enrolled. Of them, 323 had no polyps, and 32 had polyps only 10 mm. Overall, 245 patients with at least 1 polyp <10 mm were randomized (120 patients in HDWL and 125 patients in the BLI arm). The study arms were well balanced as concerns main clinical and demographic features (Table 1).

Primary end-point Accuracy in histology prediction of polyps <10 mm. In the overall study population, 483 polyps <10 mm were identified; in detail 224 and 259 in patients randomized to receive real-time histology prediction by HDWL and BLI, respectively (Table 2). We found 3 and 4 unspecific findings (eg, prominent folds, inflammatory areas) in the HDWL and BLI groups, respectively. www.giejournal.org

Rondonotti et al

Blue-light imaging

TABLE 1. Descriptive per-patient analysis Overall No. of patients

HDWL

BLI

P value

245

120

125

NA

136/109

72/48

64/61

.165

63.6  10.6

63.3  11.2

63.8  9.9

.960

FIT þve

68

30

38

Screening

42

26

16

Symptoms

57

23

34

Post-polypectomy surveillance

78

41

37

Sex (female/male) Age (mean  SD), y Colonoscopy indication

.135

Sedation regimen Conscious/none

.862 230/15

113/7

117/8

6/7

69/22

36/10

33/12

8/9

41/113

14/60

27/23

Bowel cleansing (BBPS)

Endoscopist

.84

.714

1

53

26

27

2

53

29

24

3

53

23

30

4

86

42

44

175/70

82/36

93/34

.517

Magnification No/yes

Advanced adenoma: adenoma with 1 of the following features: size 10 mm, high-grade dysplasia, or villous component. HDWL, High-definition white light; BLI, blue-light imaging; NA, not applicable; SD, standard deviation; FITþve, patients with positive fecal immunochemical test; BBPS, Boston Bowel Preparation Scale.

Because there were few unspecific findings, they were equally distributed across the study groups, and they were all correctly classified as non-adenomas, we excluded them from further analysis. The histology prediction was performed with high confidence in 193 polyps (86%) and in 222 polyps (86%) in the HDWL and BLI study arms, respectively (P Z .887). The polyps evaluated with high confidence in the study groups were comparable as concerns size, location, morphology, and rate of adenomas (Table 3). The number of polyps evaluated by each endoscopist as well as the type of endoscope used (with or without magnification) were similar across the study groups (Table 3). The overall accuracy in predicting histology was 84% and 92% in the HDWL and BLI groups, respectively (P Z .011). The sensitivity, specificity, and positive and negative predictive values of high-confidence prediction for adenomatous histology were 87%, 79%, 89%, and 75% in the HDWL arm and 94%, 88%, 95%, and 87% in the BLI arm, respectively (Table 4).

Secondary end-points Accuracy in histology prediction of diminutive polyps. Overall, 358 diminutive polyps, representing most of the evaluated polyps (86.2%), were included in the present study. The overall accuracy in predicting www.giejournal.org

histology was 83% and 92% in the HDWL and BLI groups, respectively (P Z .008). Corresponding sensitivity, specificity, and positive and negative predictive values are reported in Table 4. Accuracy in histology prediction of diminutive rectosigmoid polyps. When we restricted the analysis to rectosigmoid polyps
5 mm (124 overall, 47 and 77 in the HDWL and BLI groups, respectively), the overall accuracy was 80% in the HDWL arm and 92% in the BLI arm (P Z .110). The sensitivity, specificity, positive and negative predictive values were 82%, 78%, 82%, and 78%, and 92%, 92%, 88%, and 88% in the HDWL and the BLI groups, respectively (Table 4). The negative predictive value for rectosigmoid polyps was comparable across the study arms (78% vs 88%; P Z .223). Agreement of post-polypectomy surveillance intervals. The agreement between the prediction-based and the histology-based post-polypectomy surveillance interval was calculated in 219 patients (107 patients evaluated with HDWL and 112 with BLI) because 8 patients had invasive carcinoma (2 in the HDWL and 6 in the BLI group), and 18 patients (11 in the HDWL and 7 in the BLI group) had all polyps evaluated with low confidence. According to the U.S. Multi-Society Task Force on Colorectal Cancer recommendations,23 the post-polypectomy Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 5

Blue-light imaging

Rondonotti et al

TABLE 2. Descriptive per-polyp analysis Overall

HDWL, no. (%)

BLI, no. (%)

P value

483

224

259

NA

Polyps <10 mm Size

NA


5 mm

422

199 (88.8)

223 (86.1)

6-9 mm

61

25 (11.2)

36 (13.9)

Location

.198

Right side of colon

242

114 (50.9)

128 (49.4)

Transverse colon

49

27 (12.0)

22 (8.5)

Descending colon

43

23 (10.3)

20 (7.7)

Sigmoid colon/rectum

149

60 (26.8)

89 (34.4)

Ip

20

8 (3.6)

12 (4.7)

Is

346

163 (72.7)

183 (70.6)

Morphology*

.842

Isp

20

8 (3.6)

12 (4.7)

IIa

61

26 (11.6)

35 (13.5)

IIb

36

19 (8.5)

17 (6.5)

Histology Adenoma Low risk/advanced

322

147 (65.6)

175 (67.6)

292/30

136/11

156/19

161

77 (34.4)

84 (32.4)

138/23

64/13

74/10

Non-adenoma Hyperplastic/SSL

.654y

Confidence in histology prediction

.887

High/low

415/68

193 (86.1)/31 (13.9)

222 (85.7)/37 (14.3)

Endoscopist

.536

1

120

58 (25.8)

62 (23.9)

2

107

55 (24.5)

52 (20.1)

3

89

40 (17.7)

49 (18.9)

4

168

72 (32.0)

96 (37.1)

338/145

154 (68.7)/72 (31.2)

184 (71.0)/73 (29.0)

Magnification No/yes

.409

HDWL, High-definition white light; BLI, blue-light imaging; NA, not applicable; SSL, sessile serrated lesion. *Polyp morphology is reported according to Paris classification 22. yAdenoma vs non-adenoma. Advanced adenoma: adenoma with 1 of the following features: size 10 mm, high-grade dysplasia, or villous component.

surveillance interval was correctly predicted in 78% (84/ 107) and in 86% (96/112) of patients receiving polyp characterization by HDWL and BLI, respectively (P Z .223) (Table 5). The same figures were 92% (98/107) and in 94% (105/112) with HDWL and BLI, respectively (P Z .609), when we used European Society of Gastrointestinal Endoscopy guidelines24 (Table 6). Accuracy in histology prediction of SSLs. Out of 138 polyps endoscopically categorized as non-adenomas (67 by HDWL and 71 by BLI), 21 (16%) were SSLs, most of which (85%) were located in the right side of the colon. Twelve SSLs were evaluated with HDWL and 9 with BLI. The endoscopist accuracy in predicting serrated histology was 82% by HDWL and 79% by BLI, respectively (P Z 1.00). 6 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

Subgroup analyses Accuracy in histology prediction according to endoscopist. The accuracy in discriminating adenomas from non-adenomas varied across endoscopists, ranging from 73% to 89% by HDWL and from 83% to 98% by BLI. The improvement in the diagnostic accuracy was consistent across all endoscopists. In 3 of 4 endoscopists, the accuracy in histology prediction increased over 90% (Table 7). Accuracy in histology prediction according to magnification. Fifty-eight (30.1%) and 68 (30.6%) polyps were evaluated with magnification in the HDWL and BLI study groups, respectively (P Z .887). The overall accuracy as well as sensitivity, specificity, and negative and positive predictive values according to the use of magnification are reported in Table 8. In predicting polyp histology, www.giejournal.org

Rondonotti et al

Blue-light imaging

TABLE 3. Descriptive per-polyp analysis including only polyps evaluated with high confidence

Polyps <10 mm

Overall

HDWL, no. (%)

BLI, no. (%)

415

193

222

Size

P value NA .314


5 mm

358

170 (88.1)

188 (84.7)

6-9 mm

57

23 (11.9)

34 (15.3)

Location

.075

Right side of colon

209

100 (51.8)

109 (49.0)

Transverse colon

47

26 (13.5)

21 (9.4)

Descending colon

35

20 (10.4)

15 (6.7)

Sigmoid colon/rectum

124

47 (24.3)

77 (34.9)

Ip

19

8 (4.1%)

11 (4.9%)

Is

297

137 (71.0%)

160 (72.1%)

Isp

17

8 (4.1%)

9 (4.1%)

IIa

53

25 (13%)

28 (12.6%)

IIb

29

15 (7.8%)

14 (6.3%)

Morphology*

.974

Histology Adenoma Low risk/advanced Non-adenoma Hyperplastic/SSL

.752y 283

130 (67.4)

255/28

119/11

153 (68.9) 136/17

132

63 (32.6)

69 (31.1)

111/21

51/12

60/9

Endoscopist

.239

1

95

48 (24.9)

47 (21.2)

2

94

49 (25.4)

45 (20.3)

3

79

37 (19.1)

42 (18.9)

4

147

59 (30.6)

88 (39.6)

Magnification No/yes

.887 126/289

135 (69.9)/58 (30.1)

154 (69.4)/68 (30.6)

HDWL, High-definition white light; BLI, blue-light imaging; SSL, sessile serrated lesion. *Polyp morphology is reported according to Paris classification 22. yAdenoma vs non-adenoma. Advanced adenoma: adenoma with 1 of the following features: size 10 mm, high-grade dysplasia, or villous component.

BLI without zoom magnification was significantly more accurate than HDWL with zoom (92% vs 81%; P Z .019).

DISCUSSION The present study demonstrated that the use of BLI significantly improved endoscopist accuracy in real-time characterization of colon polyps <10 mm and diminutive polyps, when compared with HDWL. The analysis of cumulative data indicated that HDWL was associated with suboptimal performance in the accuracy of polyp characterization, whereas the adjunct use of BLI to HDWL for polyp characterization increased this accuracy over 90%. Thus, present data confirm that narrow-band light observation, by further emphasizing the mucosal and vascular details of the polyp surface, is essential for an accurate histology prediction, even when latest generation HDWL imaging is used. www.giejournal.org

This finding is relevant because, at present, sufficient evidence on the usefulness of BLI for polyp characterization was lacking. The few published reports on BLI for polyp characterization were based on retrospective evaluations of high-quality still images or short videos, representing a setting far from being comparable to current clinical practice.14,15 To the best of our knowledge, this is the first randomized study comparing the recently introduced BLI technique with HDWL as concerns the performances of endoscopists in real-time histology prediction. Because the comparison between any new chromoendoscopy system and HDWL represents the necessary step before implementing the technique in clinical practice, we designed this comparative study mostly as a feasibility study. For this purpose we did not focused on diminutive rectosigmoid polyps only, but we included polyps <10 mm in size regardless of their location, in line with similar previous studies evaluating other advanced imaging systems for optical Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 7

Blue-light imaging

Rondonotti et al

TABLE 4. Sensitivity, specificity, positive and negative predictive values, and overall accuracy of HDWL imaging and BLI in predicting histology of polyps <10 mm, with high confidence Overall polyps (n [ 415) HDWL (n [ 93)

BLI (n [ 222)

Diminutive polyps (≤5 mm) (n [ 358)

Diminutive rectosigmoid polyps (≤5 mm) (n [ 124)

Small polyps (6-9 mm) (n [ 57)

HDWL (n [ 170)

HDWL (n [ 47)

BLI (n [ 77)

HDWL (n [ 23)

BLI (n [ 34)

BLI (n [ 188)

Sensitivity (%)

87

94

85

94

82

92

100

96

Specificity (%)

79

88

78

89

78

92

86

86

PPV (%)

89

95

89

94

82

88

94

96

NPV (%)

75

87

72

87

78

88

100

85

Accuracy (%)

84

92

83

92

80

92

95

94

P value*

.011

.008

.110

1.000

HDWL, High-definition white light; BLI, blue-light imaging; PPV, positive predictive value; NPV, negative predictive value. *P value refers to the accuracy comparison.

TABLE 5. Agreement between histology-based and both HDWL-based and BLI-based imaging post-polypectomy surveillance intervals calculated according to U.S. Multi-Society Task Force on Colorectal Cancer recommendations23

TABLE 6. Agreement between histology-based and both HDWL-based and BLI-based imaging post-polypectomy surveillance intervals calculated according to European Society of Gastrointestinal Endoscopy guidelines24

Histology-directed surveillance, y Surveillance, y

3

5

5-10

Histology-directed surveillance, y

10

HDWL-directed

Surveillance, y

3

10

HDWL-directed

3

15*

0

7

1

3

15*

8

5

0

4*

0

0

10

1

83*

5-10

1

0

48*

4

10

0

4

5

18*

3

18*

6

10

1

87*

3

18*

0

5

1

5

0

4*

0

0

5-10

1

1

54*

2

10

0

3

3

20*

BLI-directed

HDWL, High-definition white light; BLI, blue-light imaging. *No. of patients in whom post-polypectomy surveillance is correctly attributed.

diagnosis.4,7,20,21,25,28 Moreover, we settled as the primary outcome the overall accuracy in histology prediction, rather than other more relevant clinical performance measures (eg, negative predictive value for rectosigmoid diminutive polyps, agreement between the prediction-based and the histology-based post-polypectomy surveillance interval). The study design with a direct comparison of BLI with HDWL, by which polyps usually are detected and thus primarily evaluated as concerns their features, allowed to objectively assess the actual additional value of BLI over HDWL in polyp characterization, avoiding the risk of overestimating the positive impact of this new technique. The randomization of patients, instead of polyps, allowed us not only to evaluate the performance of HDWL and BLI in predicting polyp histology but also to perform a perpatient analysis. Interestingly, the results obtained for polyps <10 mm were mostly driven by BLI performances on diminutive 8 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

BLI-directed

HDWL, High-definition white light; BLI, blue-light imaging. *No. of patients in whom post-polypectomy surveillance is correctly attributed.

polyps. Indeed, we found that BLI significantly improved the accuracy of real-time optical diagnosis of polyps
5 mm, whereas its actual additional value over HDWL was negligible for polyps of 6 to 9 mm in size. This finding is highly relevant clinically because diminutive polyps represent the large majority of polyps encountered during colonoscopy and are the target of the resect and discard strategy.1,4 According to other studies, the subgroup analysis disclosed that the overall accuracy of BLI-based optical diagnosis largely varied across endoscopists, ranging from 83% to 98%. All endoscopists had an accuracy of histology prediction by HDWL lower than 90%, but they all consistently improved their performances with BLI, and 3 of them achieved high accuracy levels (over 90%). The magnitude of the accuracy increase appears to be relevant (5%-10%), albeit not statistically significant, because of the relatively low number of polyps evaluated by each endoscopist. Indeed, considering all endoscopists together, the increase from 84% to 92% reached statistical significance. When we restricted the analysis to diminutive rectosigmoid polyps, which represent the target for the leave-in strategy, we observed with BLI a 10% absolute increase www.giejournal.org

Rondonotti et al

Blue-light imaging

TABLE 7. Sensitivity, specificity, positive and negative predictive values, and overall accuracy of HDWL imaging and BLI in predicting histology of polyps <10 mm with high confidence, according to endoscopists Endoscopist 1 (n [ 95 polyps)

Endoscopist 2 (n [ 94 polyps)

Endoscopist 3 (n [ 79 polyps)

Endoscopist 4 (n [ 147 polyps)

HDWL (n [ 48 polyps)

BLI (n [ 47 polyps)

HDWL (n [ 49 polyps)

BLI (n [ 45 polyps)

HDWL (n [ 37 polyps)

BLI (n [ 42 polyps)

HDWL (n [ 59 polyps)

BLI (n [ 88 polyps)

Sensitivity (%)

68

85

90

96

96

100

97

94

Specificity (%)

90

77

83

85

75

90

74

94

PPV (%)

96

90

90

93

88

97

85

96

NPV (%)

43

67

83

92

90

100

94

91

Accuracy (%)

73

83

88

93

89

98

88

P value*

.323

.489

.180

94 .224

HDWL, High-definition white light; BLI, blue-light imaging; PPV, positive predictive value; NPV, negative predictive value. *P value refers to the accuracy comparison.

TABLE 8. Sensitivity, specificity, positive and negative predictive value and overall accuracy of HDWL imaging and BLI in predicting histology of polyps <10 mm with high confidence, according to the use of magnification With zoom magnification (n [ 116)

Without zoom magnification (n [ 289)

HDWL (n [ 58)

BLI (n [ 68)

HDWL (n [ 135)

BLI (n [ 154)

Sensitivity (%)

77

91

90

95

Specificity (%)

86

90

75

87

PPV (%)

90

95

89

94

NPV (%)

70

83

77

89

Accuracy (%)

81

91

85

P value*

.119

92 .091

HDWL, High-definition white light; BLI, blue-light imaging; PPV, positive predictive value; NPV, negative predictive value. *P value refers to the accuracy comparison.

of negative predictive value for adenomatous histology (78%-88%), that however remained below the 90% performance threshold established by Preservation and Incorporation of Valuable Endoscopic Innovations (PIVI)1 initiative to consider the technique suitable for applying a policy of leaving in place suspected nonneoplastic polyps. The overall number of diminutive rectosigmoid polyps was 124; thus, the paucity of polyps analyzed does not allow us to draw definite conclusions, and further studies adequately powered to this clinical end-point are warranted. Similarly, the per-patient analysis is limited by the low numbers of patients evaluated. However, our data showed that the additional use of BLI allowed overcoming the 90% threshold settled by the PIVI initiative for postpolypectomy surveillance interval accuracy, at least in the frame of European Union guidelines. Interestingly, in our study the threshold was reached also by using HDWL alone, suggesting that the technological improvement in the white light observation significantly contributed to the characterization process. As concerns SSLs, the present study suggests substantial equivalence between BLI and HDWL, with an overall accuracy in their identification of 79% and 82%, respectively. Actually, many morphologic features typical of www.giejournal.org

SSLs, such as the presence of a mucus cap, clouded surface, and indistinct borders or irregular shape can be recognized easily with new-generation HDWL colonoscopes, and this might explain the acceptable accuracy achieved with HDWL. Of note, the levels of accuracy of optical diagnosis of SSLs reached in the present study were comparable to those achieved with NBI, by using a specific NBI-based endoscopic classification (WASP classification).19 The use of zoom magnification in addition to either HDWL or BLI, allowing a more detailed evaluation of the superficial mucosa and capillary pattern, was expected to facilitate the distinction between adenoma and non-adenoma. Conversely, we did not observe any improvement in the overall accuracy of histology prediction in both the study arms. We have instead found a significantly higher accuracy in optical diagnosis when polyps were evaluated by BLI without zoom rather than HDWL plus zoom (92% vs 81%), suggesting that the substantial improvement of optical diagnosis accuracy might be related to BLI rather than to zoom technology. These observations may reinforce the concept, already suggested by the studies with NBI1,4,11,18 that BLI enables us to reach acceptable accuracy values, even without any magnification. However, it is worth Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 9

Blue-light imaging

Rondonotti et al

noting that the study was not designed to explore this issue, and although magnification was used whenever available, the overall number of polyps evaluated with zoom is small. Thus, further studies to better clarify the contribution of zoom in predicting polyp histology are needed. We acknowledge some limitations in the present study: first, this was a single-center study, all the involved endoscopists were highly experienced in polyp characterization, and this may limit the generalizability of the results. Second, when the present study was designed, the BLI dedicated classification for optical diagnosis of polyps (eg, BLI Adenoma Serrated International ClassificationBASIC) was not available yet.29 Thus, the histology prediction was generally performed evaluating the same domains (color, vessels, and surface pattern), which the endoscopists currently evaluated with the NICE and WASP classifications. Third, we did not perform any specific training before starting the present study. Recent data demonstrate that the use of a bespoke BLI classification system29 with adequate training significantly improves both sensitivity and negative predictive value in polyp characterization.14 Last but not least, we did not set any accuracy monitoring system during the trial; therefore, we do not know whether the performance could have been improved over time during the study period. In conclusion, as from present results, BLI significantly improved the overall accuracy in the real-time characterization of polyps <10 mm and diminutive colon polyps, the improvement of accuracy being consistent across all endoscopists. However, further studies adequately powered and specifically designed to assess whether BLI can allow endoscopists to meet the PIVI thresholds are needed before the technique is implemented into clinical practice. The recently developed BLI-based polyp classification, along with formal BLI specific training sessions, could help to achieve this goal.

REFERENCES 1. ASGE Technology Committee; Abu Dayyeh BK, Thosani N, Konda V, et al. ASGE Technology Committee systematic review and metaanalysis assessing the ASGE PIVI thresholds for adopting real-time endoscopic assessment of the histology of diminutive colorectal polyps. Gastrointest Endosc 2015;81:502.e1-502. 2. Rastogi A, Rao DS, Gupta N, et al. Impact of a computer-based teaching module on characterization of diminutive colon polyps by using narrow-band imaging by non-experts in academic and community practice: a video-based study. Gastrointest Endosc 2014;79:390-8. 3. Repici A, Hassan C, Radaelli F, et al. Accuracy of narrow-band imaging in predicting colonoscopy surveillance intervals and histology of distal diminutive polyps: results from a multicenter, prospective trial. Gastrointest Endosc 2013;78:106-14. 4. East JE, Vleugels JL, Roelandt P, et al. Advanced endoscopic imaging: European Society of Gastrointestinal Endoscopy (ESGE) Technology Review. Endoscopy 2016;48:1029-45. 5. Basford PJ, Longcroft-Wheaton G, Higgins B, et al. High-definition endoscopy with i-Scan for evaluation of small colon polyps: the HiSCOPE study. Gastrointest Endosc 2014;79:111-8.

10 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

6. Klenske E, Zopf S, Neufert C, et al. I-Scan optical enhancement for the in vivo prediction of diminutive colorectal polyp histology: results from a prospective three-phased multicentre trial. PLoS One 2018;13: e0197520. 7. Repici A, Ciscato C, Correale L, et al. Narrow-band Imaging International Colorectal Endoscopic Classification to predict polyp histology: REDEFINE study (with videos). Gastrointest Endosc 2016;84:479-86. 8. Ponugoti PL, Cummings OW, Rex DK. Risk of cancer in small and diminutive colorectal polyps. Dig Liver Dis 2017;49:34-7. 9. Brenner H, Hoffmeister M, Stegmaier C, et al. Risk of progression of advanced adenomas to colorectal cancer by age and sex: estimates based on 840,149 screening colonoscopies. Gut 2007;56:1585-9. 10. Nolthenius CJT, Boellaard TN, de Haan MC, et al. Evolution of screen detected small (6-9 mm) polyps after a 3-year surveillance interval: assessment of growth with CT colonography compared with histopathology. Am J Gastroenterol 2015;110:1682-90. 11. Kami nski MF, Hassan C, Bisschops R, et al. Advanced imaging for detection and differentiation of colorectal neoplasia: European Society of Gastrointestinal Endoscopy (ESGE) Guideline. Endoscopy 2014;46:435-49. 12. Yoshida N, Yagi N, Inada Y, et al. Ability of a novel blue laser imaging system for the diagnosis of colorectal polyps. Dig Endosc 2014;26: 250-8. 13. Utano K, Nemoto D, Kashida H, et al. Endoscopic characterization of small colorectal polyps using blue laser imaging [abstract]. Gastrointest Endosc 2015;81:AB278. 14. Subramaniam S, Alkandari A, Kandlah K, et al. Blue light imaging for the optical diagnosis of small colorectal polyps: the impact of a training intervention. Endoscopy 2018;50:S1-216. 15. Szalai M, Oczella L, Dubravcsik Z, et al. Differentiation between neoplastic and non-neoplastic diminutive colorectal polyps with Fujinon Blue Light Imaging electronic chromoendoscopy with and without optical magnificationda randomized, prospective, multicentre trial. Endoscopy 2018;50:S1-216. 16. Calderwood AH, Schroy PC 3rd, Lieberman DA, et al. Boston Bowel Preparation Scale scores provide a standardized definition of adequate for describing bowel cleanliness. Gastrointest Endosc 2014;80:269-76. 17. Lai EJ, Calderwood AH, Doros G, et al. The Boston bowel preparation scale: a valid and reliable instrument for colonoscopy-oriented research. Gastrointest Endosc 2009;69:620-5. 18. Hewett DG, Kaltenbach T, Sano Y, et al. Validation of a simple classification system for endoscopic diagnosis of small colorectal polyps using narrow-band imaging. Gastroenterology 2012;143:599-607. 19. IJspeert JE, Bastiaansen BA, van Leerdam ME, et al. Development and validation of the WASP classification system for optical diagnosis of adenomas, hyperplastic polyps and sessile serrated adenomas/polyps. Gut 2016;65:963-70. 20. Paggi S, Rondonotti E, Amato A, et al. Narrow-band imaging in the prediction of surveillance intervals after polypectomy in community practice. Endoscopy 2015;47:808-14. 21. Paggi S, Rondonotti E, Amato A, et al. Resect and discard strategy in clinical practice: a prospective cohort study. Endoscopy 2012;44: 899-904. 22. Endoscopic Classification Review Group. Update on the Paris classification of superficial neoplastic lesions in the digestive tract. Endoscopy 2005;37:570-8. 23. Lieberman DA, Rex DK, Winawer SJ, et al. Guidelines for colonoscopy surveillance after screening and polypectomy: a consensus update by the US Multi-Society Task Force on Colorectal Cancer. Gastroenterology 2012;143:844-57. 24. Hassan C, Quintero E, Dumonceau JM, et al. Post-polypectomy colonoscopy surveillance: European Society of Gastrointestinal Endoscopy (ESGE) Guideline. Endoscopy 2013;45:842-51. 25. Longcroft-Wheaton G, Brown J, Cowlishaw D, et al. High-definition vs. standard-definition colonoscopy in the characterization of small colonic polyps: results from a randomized trial. Endoscopy 2012;44: 905-10.

www.giejournal.org

Rondonotti et al 26. Klare P, Haller B, Wormbt S, et al. Narrow-band imaging vs. high definition white light for optical diagnosis of small colorectal polyps: a randomized multicenter trial. Endoscopy 2016;48:909-15. 27. Rastogi A, Early DS, Gupta N, et al. Randomized, controlled trial of standard-definition white-light, high-definition white-light, and narrowband imaging colonoscopy for the detection of colon polyps and prediction of polyp histology. Gastrointest Endosc 2011;74:593-602.

www.giejournal.org

Blue-light imaging 28. Iacucci M, Trovato C, Daperno M, et al. Development and validation of the SIMPLE endoscopic classification of diminutive and small colorectal polyps. Endoscopy. Epub 2018 Apr 17. 29. Bisschops R, Hassan C, Bhandari P, et al. BASIC (BLI Adenoma Serrated International Classification) classification for colorectal polyp characterization with blue light imaging. Endoscopy 2018;50: 211-20.

Volume

-,

No.

-

: 2018 GASTROINTESTINAL ENDOSCOPY 11

Blue-light imaging

Rondonotti et al

APPENDIX 1. Practice guidelines for post-polypectomy surveillance according to the U.S. Multi-Society Task Force on Colorectal Cancer23 and the European Society of Gastrointestinal Endoscopy24 recommendations Surveillance intervals, y Patient features <10 mm hyperplastic polyp in the rectosigmoid colon 1 or 2 tubular adenomas <1 cm 3-10 tubular adenomas >10 adenomas

U.S. Multi-Society Task Force on Colorectal Cancer

European Society of Gastrointestinal Endoscopy

10

10

5-10

10

3

3

<3

3

Any adenoma 1 cm, villous features, or high-grade dysplasia

3

3

Sessile serrated adenoma/polyp <10 mm

5

10

11.e1 GASTROINTESTINAL ENDOSCOPY Volume

-,

No.

-

: 2018

www.giejournal.org