- Email: [email protected]
Sensitivity of Narrow Band Imaging Compared With White Light Imaging for the Detection of Endometriosis
Accepted Manuscript The Sensitivity of Narrow Band Imaging Compared to White Light Imaging for the Detection of Endometriosis Fermin F. Barrueto, M.D., Kevin M. Audlin, M.D., Lisa Gallicchio, Ph.D., Charles Miller, M.D., Ryan MacDonald, Ph.D., Edgar Alonsozana, M.D., Mary Johnston, Kathy J. Helzlsouer, M.D., M.H.S. PII:
S1553-4650(15)00289-7
DOI:
10.1016/j.jmig.2015.04.005
Reference:
JMIG 2541
To appear in:
The Journal of Minimally Invasive Gynecology
Received Date: 30 January 2015 Revised Date:
2 April 2015
Accepted Date: 4 April 2015
Please cite this article as: Barrueto FF, Audlin KM, Gallicchio L, Miller C, MacDonald R, Alonsozana E, Johnston M, Helzlsouer KJ, The Sensitivity of Narrow Band Imaging Compared to White Light Imaging for the Detection of Endometriosis, The Journal of Minimally Invasive Gynecology (2015), doi: 10.1016/ j.jmig.2015.04.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
The Sensitivity of Narrow Band Imaging Compared to White Light Imaging for the
2
Detection of Endometriosis
3
Fermin F. Barrueto1, M.D., Kevin M. Audlin1, M.D., Lisa Gallicchio1,2, Ph.D., Charles Miller3,4,
5
M.D., Ryan MacDonald1, Ph.D., Edgar Alonsozana1, M.D., Mary Johnston3, and Kathy J.
6
Helzlsouer1,5, M.D., M.H.S.
RI PT
4
SC
7
Affiliations
9
1
Mercy Medical Center, Baltimore, MD
10
2
Department of Epidemiology and Public Health, University of Maryland School of Medicine,
11
Baltimore, MD
12
3
Advocate Lutheran General Hospital Park Ridge IL
13
4
Department of Obstetrics & Gynecology, University of Illinois at Chicago, Chicago, IL
14
5
Department of Epidemiology, Johns Hopkins University Bloomberg School of Public Health,
15
Baltimore, MD
TE D
M AN U
8
EP
16
Corresponding Author
18
Kathy J. Helzlsouer, M.D., M.H.S. The Prevention and Research Center Weinberg Women’s
19
Center for Health and Medicine, Mercy Medical Center, 301 St. Paul Place Baltimore, MD
20
21202
21
Phone: 410-951-7950; Fax: 410-951-7931; Email [email protected]
22
AC C
17
ACCEPTED MANUSCRIPT
Financial support and conflicts of interest statement: The study was funded by a research grant
24
provided by Olympus America, Inc. to Mercy Medical Center and Charles Miller, MDSC.
25
Additionally, Dr. Audlin reports personal fees from Gynecare and Olympus America, Inc outside
26
the submitted work. Dr. Barrueto reports personal fees in the form of honoraria for educational
27
programs from Olympus America, Inc. outside the submitted work.
AC C
EP
TE D
M AN U
SC
RI PT
23
ACCEPTED MANUSCRIPT
Abstract
29
STUDY OBJECTIVE: The primary objective of the study was to evaluate the ability of narrow
30
band imaging (NBI) in conjunction with standard white light imaging to improve the detection
31
and diagnosis of endometriosis during laparoscopic evaluation compared to white light imaging
32
alone. Sensitivity of NBI in detecting endometriosis was assessed and compared to white light
33
imaging.
34
DESIGN: A randomized controlled trial.
35
CLASSIFICATION OF STUDY DESIGN: LEVEL I: Evidence obtained from a properly
36
designed, randomized, controlled trial
37
SETTING: The trial was conducted in two medical centers.
38
PATIENTS:
39
endometriosis and/or infertility were recruited. Of these, 150 were evaluable for the primary aim
40
to determine sensitivity of NBI compared to white light imaging for the detection of
41
endometriotic lesions.
42
INTERVENTIONS: Patients were randomized in a 3:1 ratio to receive white light imaging
43
followed by NBI or white light imaging only. The pelvis was systematically visualized with
44
each assigned imaging modality; lesions were recorded under each visualization and then
45
resected. All patients had white light imaging on the first visualization followed by either a
46
second white light examination (control arm) or NBI examination (intervention arm.)
47
MEASUREMENT: Pathology of resected lesions was the gold standard for evaluating
48
sensitivity and was conducted at each institution. The method of detection of the lesion (white
49
light or NBI) was masked. Central pathology review was conducted for a randomly selected
50
10% sample of specimens and for those lesions visualized under only one imaging modality
M AN U
SC
RI PT
28
AC C
EP
TE D
A total of 167 women undergoing laparoscopic evaluation for suspected
ACCEPTED MANUSCRIPT
among patients assigned to the intervention arm. The sensitivity was assessed for each modality
52
(white light and NBI) and compared using a McNemar’s test.
53
MAIN RESULTS: Among the group randomized to receive both white light and NBI, four
54
patients had lesions detected with NBI but no lesions detected with white light. Among the 255
55
lesions confirmed as endometriosis by pathologic review, all were detected by NBI for a
56
sensitivity of 100%; 79% were detected by white light imaging (p<0.001).
57
CONCLUSION: The addition of NBI to white light imaging increased the number of
58
endometriotic lesions identified during laparoscopy and the diagnosis of endometriosis compared
59
to the use of white light imaging alone.
61 62
66 67 68 69 70 71 72 73
EP
65
AC C
64
TE D
63
M AN U
60
SC
RI PT
51
ACCEPTED MANUSCRIPT
74
Introduction
75
Endometriosis is a relatively common chronic gynecological condition that affects approximately
77
10% of all women of reproductive age (1). It is a pelvic inflammatory disease that is
78
characterized by the presence of endometrial glands and stroma outside of the uterine cavity (2).
79
Typical symptoms of endometriosis include dysmenorrhea, pelvic pain, and infertility; the
80
severity of pain associated with this disease often leads to a considerable decrease in quality of
81
life (3).
SC
RI PT
76
M AN U
82
A standard treatment for the severe pelvic pain and infertility associated with endometriosis is to
84
surgically remove endometriotic areas using laparoscopy, which is typically carried out under
85
white light. Identifying all endometriotic lesions is paramount to optimal endometriosis
86
debulking, as complete resection is thought to extend the pain-free interval, resulting in
87
improved clinical outcomes (4). Difficulty of identifying lesions has been well documented;
88
studies by multiple investigators have shown the positive predictive value for histologically
89
defined endometriosis is about 65% of resected lesions (5) and the predictive value varies based
90
on the stage of the patient (6). The lower the stage of the endometriosis, the less obvious the
91
lesions and the higher the probability that excised lesions may not histologically be consistent
92
with endometriosis. In the study by Walter et al. (7), the overall positive predictive value was as
93
low as 45% for excised lesions. In the Kazanegra et al. study (6), the positive predictive value
94
for resected tissue confirmed for endometriosis was as low as 66% for stage 1 disease and as
95
high as 81% to 92% for the more advanced stage 3 and 4 disease, demonstrating that the larger,
96
more frequent and deeper lesions are more easily identified. Thus, despite the marked
AC C
EP
TE D
83
ACCEPTED MANUSCRIPT
improvement in laparoscopic technology over the past decade, persistent endometriosis, and its
98
associated pelvic pain and infertility, remains an issue for many patients and is likely the result
99
of the inability to detect and excise all endometriotic lesions, especially early and superficial
100
lesions, under standard laparoscopic conditions.
101
RI PT
97
Narrow band imaging (NBI) is a technique that uses a specific narrow wavelength of light to
103
change the normal color contrasts of the endoscopic image and improve detection of
104
neovascularization, which is the pathological feature of endometriosis for both superficial and
105
deeper vascularization. Under NBI, two discrete bands of light, one blue at 415 nm and another
106
green at 540 mm, a high contrast image of the tissue surface is created thus enabling improved
107
visualization of blood vessels (8, 9). The clinical utility of NBI in detecting endometriotic
108
lesions was first described in a case report in 2007 by Farrugia et al. (10) who showed that
109
superficial endometriotic lesions were easier to identify with NBI and that their invasiveness was
110
easier to judge than under white light alone. Following this report, Barrueto & Audlin (11)
111
conducted a small pilot study of 20 patients of reproductive age with pelvic pain that
112
demonstrated that the use of NBI resulted in the diagnosis of pathologically-proven
113
endometriosis among four patients whose lesions were missed with visible white light only
114
examination. In addition, the number of confirmed endometriosis lesions increased with the use
115
of NBI in addition to white light. Despite the promise of NBI in identifying endometriotic
116
lesions and in increasing the positive predictive value of suspected lesions, NBI is not currently
117
used in standard laparoscopic surgeries for endometriosis. This is likely because there have been
118
no large multicenter studies of rigorous design that have been conducted to examine detection,
AC C
EP
TE D
M AN U
SC
102
ACCEPTED MANUSCRIPT
119
diagnosis, and discrimination of endometriotic lesions using NBI during the laparoscopic
120
procedure.
121
Thus, to fill this gap in knowledge, we conducted a multisite randomized controlled device trial
123
in two clinical centers to determine the degree to which NBI improves the detection and
124
diagnosis of endometriosis lesions over white light alone. The specific aims of the study were to
125
determine the degree to which NBI improves the diagnosis of endometriosis (diagnostic yield) of
126
laparoscopic examinations compared to use of visible white light-only laparoscopy and to
127
determine if NBI improves sensitivity in the detection of potential endometriosis lesions and
128
reduces false positives compared to visible white light.
M AN U
SC
RI PT
122
129 130
Methods
TE D
131
This study was a randomized controlled device trial conducted in two clinical centers (Mercy
133
Medical Center in Baltimore, Maryland and Lutheran General Hospital in Chicago, Illinois).
134
Participants were randomized at a 3:1 ratio to have laparoscopic examination with white light
135
followed by NBI (white light/NBI) or white light followed by repeat white light examination
136
(white light/white light). This study was approved by the Institutional Review Boards at Mercy
137
Medical Center and Lutheran General Hospital.
139
AC C
138
EP
132
140
Adult women of reproductive age (18 to 49 years of age) undergoing diagnostic laparoscopy for
141
pelvic pain, suspected endometriosis, or infertility were eligible, and were recruited in their
ACCEPTED MANUSCRIPT
physician’s office or on the day of surgery. Women were excluded from the study if they were
143
pregnant or had health issues that their surgeon determined would make laparoscopic surgery
144
unsafe. The pre-specified sample size of 157 was calculated using nQuery 7.0 (12) and
145
determined to be sufficient to detect a minimal difference in sensitivity of 12% between the two
146
imaging modalities with 90% power and a two-sided p-value of 0.05. The sample size accounted
147
for a potential dropout rate of 10%. As such, 141 evaluable patients were needed to detect an
148
estimated difference in sensitivity as little as 12%.
M AN U
150
SC
149
RI PT
142
A total of 167 patients were enrolled, consented, and randomized to either white light/NBI or
152
white light/white light; 124 were from Mercy Medical Center and 43 from Lutheran General
153
Hospital. Of these patients, three patients did not have the laparoscopic surgery for which they
154
were recruited into the study and three patients were withdrawn from the study during surgery at
155
the discretion of the surgeon due to extensive disease. In addition, 11 patients had protocol
156
deviations resulting in unusable primary outcome data. Thus, the final analytic study sample
157
consisted of 150 patients, 112 of whom were assigned to the white light/NBI intervention group
158
and 38 to the white light/white light control group for intention-to-treat (ITT) analysis. A total
159
of three patients (two assigned to white light/NBI and one assigned to white line/white light) did
160
not receive the assigned treatment either due to equipment failure (one patient) or error in
161
performing the randomization assignment (two patients.) Thus 111 patients were imaged with
162
white light and NBI and 39 patients were imaged with white light on both sweeps. Analyses
163
were conducted according to both ITT and to modality received.
164
AC C
EP
TE D
151
ACCEPTED MANUSCRIPT
The materials used were an Olympus HD Endoeye video telescope (Models WA50011A,
166
WA50013A, WA50013L, WA50013T, WA50015L) or Olympus HD Endoeye Laparo-Thoraco
167
Videoscope (Model LTF-VH OLYMPUS) connected to a CLV-180 EVIS EXERA II Xenon
168
Light Source and an Olympus CV-180 EVIS EXERA II Video System Center. An Olympus HD
169
LCD Surgical Monitor, Sony HD recorder PDW-70MD, and Printers OEP-4 were used.
RI PT
165
170
Prior to surgery, each participant provided informed consent and completed a short baseline
172
questionnaire, which collected information on demographics, pain, medical history, health
173
habits, and quality of life. Health-related quality of life was assessed using the question: “In
174
general, would you say your health is excellent, very good, good, fair, or poor?”
M AN U
SC
171
175
For all participants, the laparoscopic procedure was conducted in a standardized fashion with
177
visible white light first, followed by an additional sweep with either white light (control) or NBI
178
(intervention). The second modality was assigned by opening a randomization envelope
179
provided by the study biostatistician after completion of the first examination with white light.
EP
180
TE D
176
Each examination (either with white light or NBI) included systematically visualizing and
182
recording lesions found in the pelvis. For ease of recording the location of the lesions, the pelvis
183
was divided into four quadrants plus the cul-de-sac: right anterior, right posterior, left posterior,
184
left anterior and cul-de-sac. The broad ligaments and ovarian fossa were included in the
185
posterior quadrants. Any detected lesions were photographed, numbered and recorded by
186
method of detection to enable careful tracking for pathologic verification. Additionally, the
187
surgeons’ clinical impression regarding degree of suspicion that the lesions were endometriosis
AC C
181
ACCEPTED MANUSCRIPT
was recorded using a three-point Likert scale to assess their degree of clinical suspicion. All
189
suspected endometrial lesions were included in the documentation, even lesions that may have
190
been latent or healed (otherwise known as white lesions). Once both exams (white light/white
191
light or white light/NBI) were completed and documented, additional areas of the abdomen were
192
evaluated at the surgeon’s discretion. Those participants randomized to white light laparoscopic
193
examination only (white light/white light) had these additional areas evaluated only under white
194
light.
SC
RI PT
188
195
All lesions visualized on either modality, regardless of clinical suspicion were excised if
197
amenable to resection. Lesions were excised using ultrasonic energy and bipolar energy for
198
homeostasis; in very few cases, laparoscopic scissors were used.
M AN U
196
199
Data collected included location, method of detection and clinical diagnosis confidence score.
201
Pathologists were masked to method of detection (NBI, white light, or both NBI and white light).
202
Lesions were recorded as falling within three categories: visible white light-detected only; NBI-
203
detected only; or both NBI- and visible white light-detected. Pathology was performed at each
204
center following a standardized protocol for the diagnosis of endometriosis. All discordant
205
lesions (visualized under only one imaging modality) and a random 10% sample of specimens
206
were re-reviewed by a central independent reviewer at the coordinating center. If disagreement
207
occurred for those chosen as part of the random sampling, the slides for those specimens were
208
adjudicated by a third pathologist at the coordinating center. Thirteen lesions that were not
209
amendable to resection and were ablated during surgery were photographed only and were not
210
included in the analysis.
AC C
EP
TE D
200
ACCEPTED MANUSCRIPT
211
The baseline characteristics of individuals randomized to the two study arms were compared
213
using t-tests for continuous measures and Chi-square tests for categorical variables.
214
RI PT
212
The primary outcome of diagnostic yield and sensitivity was assessed using data from the
216
individuals randomized to receive both modalities (white light/NBI intervention arm). Pathology
217
assessment of resected lesions was the gold standard.
218
SC
215
Diagnostic yield is reported as the number of individuals diagnosed with endometriosis based on
220
findings of NBI versus findings from white light only examination. The number of lesions
221
detected by each modality and percent positive on pathological examination was also
222
determined.
M AN U
219
TE D
223
The sensitivity was assessed for each modality (white light and NBI) and compared using a
225
McNemar’s test. Additional analyses of sensitivity and specificity by imaging modality were
226
calculated based on the clinical impression assessment among those assigned to have both white
227
light and NBI. For the lesions detected with both modalities, clinical impression was classified
228
with each sweep as definitely endometriosis, indefinite, or definitely not endometriosis.
229
Sensitivity and specificity based on clinical impression was calculated for lesions classified
230
clinically as endometriosis. Few lesions were classified as definitely not endometriosis based on
231
clinical impression; thus the indefinite and definitely not categories were combined.
232
AC C
EP
224
ACCEPTED MANUSCRIPT
233
Results were considered statistically significant with p-value ≤0.05 on two-sided tests. All
234
analyses were done using Stata 12.0.
235
Results
RI PT
236 237
Baseline characteristics of participants, according to their randomization group, are displayed in
239
Table 1. The two groups were comparable on age, education, body mass index, pain at baseline,
240
and average number of lesions excised per patient. Patients randomized to the white light/NBI
241
arm were more likely to have had prior laparoscopic surgery than those randomized to white
242
light only arm. The majority of patients reported pelvic pain as the reason for undergoing
243
surgery and taking part in the study, and most had experienced pelvic pain in the four weeks
244
prior to surgery.
M AN U
SC
238
TE D
245
Figure 1 displays 3 sets of photographs from 3 patients showing lesions visualized under NBI but
247
not with white light imaging. The circled lesions were confirmed to be endometriosis by
248
histologic examination.
249
EP
246
Among the group randomized to receive both white light and NBI, four patients had lesions
251
detected with NBI but no lesions detected with white light. Thus, these four patients were
252
diagnosed with endometriosis that would have gone undetected with white light imaging alone.
253
Seventy-two percent were diagnosed with endometriosis among those randomized to have NBI
254
compared to 65.8% among those in the white light imaging only arm. This difference, however,
255
was not statistically significant (Table 2).
AC C
250
ACCEPTED MANUSCRIPT
256
The number of lesions detected and pathology results are shown in Table 3. Among those
258
having both white light and NBI: four lesions were detected only with white light imaging, none
259
of which was confirmed to be endometriosis; 321 lesions were seen on both modalities; and 128
260
were detected only with NBI. NBI led to the detection of an additional 54 confirmed
261
endometriosis lesions, representing 42.2% of the endometriosis lesions seen only with NBI.
RI PT
257
SC
262
Table 4 displays the sensitivity of NBI and white light imaging to detect endometriosis lesions.
264
Among the 256 lesions confirmed as endometriosis by pathologic review, all were detected by
265
NBI for a sensitivity of 100%; 78.9% were detected by white light imaging. The difference in
266
sensitivity for detection of endometriosis between the two modalities was statistically significant
267
(p<.001).
M AN U
263
TE D
268
At the time of surgery, surgeons were asked to provide a clinical impression of the likelihood
270
that detected lesions were endometriosis under each imaging sweep. All lesions were to be
271
resected regardless of the clinical impression. Clinical impression was reported for 70.2% of
272
lesions visualized in patients randomized to receive both imaging modalities. Among those
273
randomized to receive both imaging modalities, and considering only those classified as
274
definitely endometriosis, the clinical impression was correct for 71.5% at the time of white light
275
imaging with a specificity of 35.9%. The sensitivity of the clinical impression improved to
276
84.1%, with a decrease in specificity to 23.9%, with NBI.
AC C
EP
269
277 278
Discussion
ACCEPTED MANUSCRIPT
279
Use of NBI, compared to white light imaging, improved the detection of endometriosis. . Of
281
the pathologically confirmed lesions resected after laparoscopic visualization with both
282
modalities, only 78.9% of the lesions were detected with white light. Thus, compared white light
283
imaging alone, NBI increased the detection of pathologically confirmed endometriosis. The
284
clinical identification of lesions at the time of laparoscopic surgery was also improved with NBI,
285
with 84.1% of pathologically confirmed lesions correctly identified as endometriosis at surgery
286
versus 71.5% with white light imaging. The use of NBI led to the diagnosis of endometriosis in
287
four patients whose endometriosis was not identified by white light examination.
288
. Improving the clinical assessment at the time of surgery can help in determining the extent of
289
resection or ablation that should be undertaken. All together, more lesions were visualized with
290
NBI in combination with white light imaging than with white light imaging alone. NBI
291
improves the detection of early or more superficial endometriosis and thus may help to prevent
292
or extend the interval until recurrence of symptoms (13). Deep lesions of endometriosis or
293
extensive disease are readily visualized with standard illumination so the benefit of NBI-assisted
294
detection of such disease would be would be lessened.
295
The results of this multi-center randomized clinical trial confirm and expand upon the findings of
296
a single-center pilot study of 21 patients (11). In that study, 14 of the 21 patients had additional
297
lesions identified with NBI that were not identified with white light imaging alone, and four
298
patients were diagnosed with endometriosis due to lesions detected only with NBI (11). The
299
results of this multi-center trial demonstrate that NBI is a useful adjunct to standard white light
300
imaging during laparoscopy.
301
AC C
EP
TE D
M AN U
SC
RI PT
280
ACCEPTED MANUSCRIPT
To our knowledge, this is the first randomized clinical trial to evaluate the use of NBI for the
303
detection of endometriosis. NBI is well suited to the detection of active endometriosis since the
304
filtered wavelength selected (415 and 540-nm wavelengths) include the highest hemoglobin
305
absorption (8, 9). This improved ability to visualize vascularization was demonstrated in the
306
current study by the improved sensitivity of NBI compared to white light imaging alone.
RI PT
302
307
The primary aim of the study was to determine the sensitivity of combined white light and NBI
309
compared to the use of standard white light imaging alone. NBI increased sensitivity for
310
detection of endometriosis. However, an increased sensitivity comes at the expense of a lowered
311
specificity. This study protocol led to a higher number of non-endometriotic lesions being
312
removed than might occur in usual clinical practice where removal of lesions may be more
313
targeted. The potential impact of improved detection of subtle endometriosis on symptoms
314
remains unanswered.
TE D
315
M AN U
SC
308
.
While maximum debulking is a goal, the disadvantage of increased identification and
317
excision is the increased risk for surgical complications. Despite the design of the trial requiring
318
excision of all lesions regardless of clinical suspicion, no complications occurred. In clinical
319
practice, NBI may help with discrimination of lesions seen with white light imaging, as well as
320
identifying new lesions, and thus avoid unnecessary excisions. NBI may serve as a training tool,
321
assisting surgeons to better identify lesions with white light imaging. NBI is easy to incorporate
322
into clinical practice when using laparoscopes that are NBI compatible because of the simplicity
323
of switching to and from standard white light and NBI during laparoscopic examination.
AC C
EP
316
ACCEPTED MANUSCRIPT
324
Whether increased debulking of endometriosis deters recurrence and improves overall clinical
325
outcomes requires additional research using a different study design.
326
Conclusion
RI PT
327 328
The addition of NBI to white light imaging increased the number of endometriosis lesions
330
identified during laparoscopy and the diagnosis of endometriosis compared to the use of white
331
light imaging alone. Improved detection of lesions can optimize debulking of endometriosis.
SC
329
M AN U
332 333 334 335
AC C
EP
TE D
336
ACCEPTED MANUSCRIPT
337
References
338
(1) Agic A, Xu H, Finas D, Banz C, Diedrich K, Hornung D. Is endometriosis associated
340
with systemic subclinical inflammation? Gynecol Obstet Invest 2006; 62(3):139-147.
341
(2) Redwine DB. Pelvic endometriosis--the same or different entities in disguise? Fertil Steril
344 345
(3) Garry R, Clayton R, Hawe J. The effect of endometriosis and its radical laparoscopic
SC
343
1998; 70(3):588-589.
excision on quality of life indicators. BJOG 2000; 107(1):44-54.
(4) Jacobson TZ, Duffy JM, Barlow D, Koninckx PR, Garry R. Laparoscopic surgery for
M AN U
342
RI PT
339
346
pelvic pain associated with endometriosis. Cochrane Database Syst Rev
347
2009;(4):CD001300.
(5) Marchino GL, Gennarelli G, Enria R, Bongioanni F, Lipari G, Massobrio M. Diagnosis
349
of pelvic endometriosis with use of macroscopic versus histologic findings. Fertil Steril
350
2005; 84(1):12-15.
351
TE D
348
(6) Kazanegra R, Zaritsky E, Lathi RB, Clopton P, Nezhat C. Diagnosis of stage I endometriosis: comparing visual inspection to histologic biopsy specimen. J Minim
353
Invasive Gynecol 2008; 15(2):176-180.
355 356
(7) Walter AJ, Hentz JG, Magtibay PM, Cornella JL, Magrina JF. Endometriosis: correlation
AC C
354
EP
352
between histologic and visual findings at laparoscopy. Am J Obstet Gynecol 2001; 184(7):1407-1411.
357
(8) Surico D, Vigone A, Bonvini D, Tinelli R, Leo L, Surico N. Narrow-band imaging in
358
diagnosis of endometrial cancer and hyperplasia: a new option? J Minim Invasive
359
Gynecol 2010; 17(5):620-625.
ACCEPTED MANUSCRIPT
361 362 363 364 365
(9) Surico D, Vigone A, Leo L. Narrow band imaging in endometrial lesions. J Minim Invasive Gynecol 2009; 16(1):9-10. (10) Farrugia M, Nair MS, Kotronis KV. Narrow band imaging in endometriosis. J Minim Invasive Gynecol 2007; 14(4):393-394.
RI PT
360
(11) Barrueto FF, Audlin KM. The use of narrowband imaging for identification of endometriosis. J Minim Invasive Gynecol 2008; 15(5):636-639.
(12) Elashoff, J.D. nQuery Advisor Version 7.0. Los Angeles, CA: 2007.
367
(13) Kuroda K, Kitade M, Kikuchi I, Kumakiri J, Matsuoka S, Jinushi M et al. Vascular
SC
366
density of peritoneal endometriosis using narrow-band imaging system and vascular
369
analysis software. J Minim Invasive Gynecol 2009; 16(5):618-621.
M AN U
368
AC C
EP
TE D
370
ACCEPTED MANUSCRIPT
Legend
372
Figure 1.
373
Example of three pathologically-confirmed endometriosis lesions from three different patients
374
that were visualized only with NBI. The Figure shows photos of three areas as visualized with
375
standard white light (a, c, e) and NBI (b, d, f ).NBI = narrow band imaging
376
Patient 1:
Patient 2:
381
Patient 3:
M AN U
d.) Cul de sac, NBI; lesion circled
379 380
c.) Cul de sac, white light; no lesion
SC
b.) Left posterior quadrant, NBI; lesion circled
377 378
a.) Left posterior quadrant, white light; no lesion
RI PT
371
e.) Cul de sac, white light; no lesion f.) Cul de sac, NBI; lesion circled
382
AC C
EP
TE D
383
ACCEPTED MANUSCRIPT Barrueto 1 Table 1. Baseline characteristics according to study arm White light/NBI
White Light/
(n = 112)
White light (n = 38)
SD
Mean
SD
p-value
33.2
7.4
30.6
7.1
.07
Race/Ethnicity
N
%
N
%
White
79
70.5
29
76.3
African American
22
19.6
7
18.4
Hispanic
SC
M AN U
Age at time of surgery, years
RI PT
Mean
6
5.4
1
2.6
5
4.5
1
2.6
Mean
SD
Mean
SD
14.6
2.4
15
2.3
N
%
N
%
52
46.4
19
50.0
47
42.0
16
42.1
13
11.6
3
7.9
Body mass index, kg/m2
N
%
N
%
Less than 25
49
43.7
16
42.1
25 to 29
30
26.8
12
31.6
30 or greater
32
28.6
10
26.3
Missing
1
0.9
0
0.0
Other
General Health
Good
AC C
Fair/Poor
EP
Excellent/Very good
TE D
Education, number of years
.84
.43
.78
.86
ACCEPTED MANUSCRIPT Barrueto 2
SD
Mean
SD
Average pain (Scale 0-10)
4.6
2.7
4.4
3.5
Prior laparoscopic surgery
N
%
Yes
73
65.2
No
38
33.9
Missing
1
0.9
N
%
15
39.5
23
60.5
0
0.0
SC
M AN U
Number of prior laparoscopic surgeries
RI PT
Mean
N
%
N
%
38
33.9
23
60.5
40
35.7
9
23.7
12.5
4
10.5
17
15.2
2
5.3
3
2.7
0
0.0
Reason for study participation
N
%
N
%
Pelvic Pain
84
75.0
28
73.7
Infertility
3
2.7
3
7.9
Pelvic Pain and Infertility
16
14.3
3
7.9
Other
9
8.0
4
10.5
Pelvic pain over the past 4 weeks
N
%
N
%
None 1
14
TE D
2 3 or more
AC C
EP
Missing
.76
.004
.039
.38
ACCEPTED MANUSCRIPT Barrueto 3 101
90.2
35
92.1
No
11
9.8
3
7.9
RI PT
Yes
AC C
EP
TE D
M AN U
SC
NBI = narrow band imaging
.72
ACCEPTED MANUSCRIPT
Barrueto 4 Table 2. Diagnostic yield among individuals randomized to the NBI/WL group versus the WL/WL group (N = 150)
N
%
N
112
81a
72.3
25
N
%
111
65.8
TE D
NBI = narrow band imaging; WL = white light
.45
81*
WL/WL Group N
%
p-valueb
64.1
.29
39
73.0
25
In the NBI/WL group, 4 patients were diagnosed with endometriosis using NBI who would not have been diagnosed
with endometriosis using WL alone b
AC C
Chi-square test for group differences
EP
a
p-valueb
38
Number diagnosed with endometriosis
%
NBI/WL Group
M AN U
Total number in group
WL/WL Group
SC
NBI/WL Group
RI PT
Analysis by Modality Received
Intention to Treat Analysis
ACCEPTED MANUSCRIPT Barrueto 5 Table 3. The number of lesions detected by each modality among individuals who received the combined NBI/WL modality (N=111) Positive by pathology N
4
0
Detected with WL and NBI
321
202
Detected with NBI alone
128
54
Total
AC C
EP
TE D
NBI = narrow band imaging; WL = white light
453
256
M AN U
Detected with WL alone
%
0.0
62.9
42.2
SC
N
RI PT
Total
ACCEPTED MANUSCRIPT Barrueto 6 Table 4. Sensitivity for the detection of endometriosis lesions according to the modality among those receiving the combined NBI/WL modality arm (N=111)
Pathology
Positive Negative
Positive Negative
Yes
256a
193
No
0
4
100.0%
202
123
No
54
74
78.9%
NBI = narrow band imaging; WL = white light a
Detected Yes
RI PT
Pathology
SC
Sensitivityb
WL Sweep
M AN U
Detected
NBI Sweep
54 additional endometrial lesions confirmed pathology positive were detected with NBI
b
AC C
EP
TE D
p<.001 for the McNemar’s test for differences in sensitivity
ACCEPTED MANUSCRIPT Barrueto 7 Table 5. Sensitivity and specificity for the clinical judgment of endometriosis against pathology gold standard by modality among those receiving the combined NBI/WL modality arm (N=111) WL a
NBI
Pathology
RI PT
Pathology Clinical Impression at the time Negative
Total
Positive
Negative
Total
Endometriosis
169
89
258
143
75
218
Not Endometriosis/Indefinite
32
28
60
57
42
99
Sensitivity
84.1%
Specificity
EP
TE D
1 lesion missing white light sweep clinical impression
AC C
a
71.5%
23.9%
NBI = narrow band imaging; WL = white light
SC
Positive
M AN U
of surgery
35.9%
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Precis:
AC C
EP
TE D
M AN U
SC
RI PT
A randomized trial was conducted to determine the sensitivity of laparoscopic examination using narrow band imaging in conjunction white light imaging compared to white light imaging alone to detect endometriosis.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
http://www.AAGL.org/jmig-22-5-JMIG-D-15-00065
S1553-4650(15)00289-7
DOI:
10.1016/j.jmig.2015.04.005
Reference:
JMIG 2541
To appear in:
The Journal of Minimally Invasive Gynecology
Received Date: 30 January 2015 Revised Date:
2 April 2015
Accepted Date: 4 April 2015
Please cite this article as: Barrueto FF, Audlin KM, Gallicchio L, Miller C, MacDonald R, Alonsozana E, Johnston M, Helzlsouer KJ, The Sensitivity of Narrow Band Imaging Compared to White Light Imaging for the Detection of Endometriosis, The Journal of Minimally Invasive Gynecology (2015), doi: 10.1016/ j.jmig.2015.04.005. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
The Sensitivity of Narrow Band Imaging Compared to White Light Imaging for the
2
Detection of Endometriosis
3
Fermin F. Barrueto1, M.D., Kevin M. Audlin1, M.D., Lisa Gallicchio1,2, Ph.D., Charles Miller3,4,
5
M.D., Ryan MacDonald1, Ph.D., Edgar Alonsozana1, M.D., Mary Johnston3, and Kathy J.
6
Helzlsouer1,5, M.D., M.H.S.
RI PT
4
SC
7
Affiliations
9
1
Mercy Medical Center, Baltimore, MD
10
2
Department of Epidemiology and Public Health, University of Maryland School of Medicine,
11
Baltimore, MD
12
3
Advocate Lutheran General Hospital Park Ridge IL
13
4
Department of Obstetrics & Gynecology, University of Illinois at Chicago, Chicago, IL
14
5
Department of Epidemiology, Johns Hopkins University Bloomberg School of Public Health,
15
Baltimore, MD
TE D
M AN U
8
EP
16
Corresponding Author
18
Kathy J. Helzlsouer, M.D., M.H.S. The Prevention and Research Center Weinberg Women’s
19
Center for Health and Medicine, Mercy Medical Center, 301 St. Paul Place Baltimore, MD
20
21202
21
Phone: 410-951-7950; Fax: 410-951-7931; Email [email protected]
22
AC C
17
ACCEPTED MANUSCRIPT
Financial support and conflicts of interest statement: The study was funded by a research grant
24
provided by Olympus America, Inc. to Mercy Medical Center and Charles Miller, MDSC.
25
Additionally, Dr. Audlin reports personal fees from Gynecare and Olympus America, Inc outside
26
the submitted work. Dr. Barrueto reports personal fees in the form of honoraria for educational
27
programs from Olympus America, Inc. outside the submitted work.
AC C
EP
TE D
M AN U
SC
RI PT
23
ACCEPTED MANUSCRIPT
Abstract
29
STUDY OBJECTIVE: The primary objective of the study was to evaluate the ability of narrow
30
band imaging (NBI) in conjunction with standard white light imaging to improve the detection
31
and diagnosis of endometriosis during laparoscopic evaluation compared to white light imaging
32
alone. Sensitivity of NBI in detecting endometriosis was assessed and compared to white light
33
imaging.
34
DESIGN: A randomized controlled trial.
35
CLASSIFICATION OF STUDY DESIGN: LEVEL I: Evidence obtained from a properly
36
designed, randomized, controlled trial
37
SETTING: The trial was conducted in two medical centers.
38
PATIENTS:
39
endometriosis and/or infertility were recruited. Of these, 150 were evaluable for the primary aim
40
to determine sensitivity of NBI compared to white light imaging for the detection of
41
endometriotic lesions.
42
INTERVENTIONS: Patients were randomized in a 3:1 ratio to receive white light imaging
43
followed by NBI or white light imaging only. The pelvis was systematically visualized with
44
each assigned imaging modality; lesions were recorded under each visualization and then
45
resected. All patients had white light imaging on the first visualization followed by either a
46
second white light examination (control arm) or NBI examination (intervention arm.)
47
MEASUREMENT: Pathology of resected lesions was the gold standard for evaluating
48
sensitivity and was conducted at each institution. The method of detection of the lesion (white
49
light or NBI) was masked. Central pathology review was conducted for a randomly selected
50
10% sample of specimens and for those lesions visualized under only one imaging modality
M AN U
SC
RI PT
28
AC C
EP
TE D
A total of 167 women undergoing laparoscopic evaluation for suspected
ACCEPTED MANUSCRIPT
among patients assigned to the intervention arm. The sensitivity was assessed for each modality
52
(white light and NBI) and compared using a McNemar’s test.
53
MAIN RESULTS: Among the group randomized to receive both white light and NBI, four
54
patients had lesions detected with NBI but no lesions detected with white light. Among the 255
55
lesions confirmed as endometriosis by pathologic review, all were detected by NBI for a
56
sensitivity of 100%; 79% were detected by white light imaging (p<0.001).
57
CONCLUSION: The addition of NBI to white light imaging increased the number of
58
endometriotic lesions identified during laparoscopy and the diagnosis of endometriosis compared
59
to the use of white light imaging alone.
61 62
66 67 68 69 70 71 72 73
EP
65
AC C
64
TE D
63
M AN U
60
SC
RI PT
51
ACCEPTED MANUSCRIPT
74
Introduction
75
Endometriosis is a relatively common chronic gynecological condition that affects approximately
77
10% of all women of reproductive age (1). It is a pelvic inflammatory disease that is
78
characterized by the presence of endometrial glands and stroma outside of the uterine cavity (2).
79
Typical symptoms of endometriosis include dysmenorrhea, pelvic pain, and infertility; the
80
severity of pain associated with this disease often leads to a considerable decrease in quality of
81
life (3).
SC
RI PT
76
M AN U
82
A standard treatment for the severe pelvic pain and infertility associated with endometriosis is to
84
surgically remove endometriotic areas using laparoscopy, which is typically carried out under
85
white light. Identifying all endometriotic lesions is paramount to optimal endometriosis
86
debulking, as complete resection is thought to extend the pain-free interval, resulting in
87
improved clinical outcomes (4). Difficulty of identifying lesions has been well documented;
88
studies by multiple investigators have shown the positive predictive value for histologically
89
defined endometriosis is about 65% of resected lesions (5) and the predictive value varies based
90
on the stage of the patient (6). The lower the stage of the endometriosis, the less obvious the
91
lesions and the higher the probability that excised lesions may not histologically be consistent
92
with endometriosis. In the study by Walter et al. (7), the overall positive predictive value was as
93
low as 45% for excised lesions. In the Kazanegra et al. study (6), the positive predictive value
94
for resected tissue confirmed for endometriosis was as low as 66% for stage 1 disease and as
95
high as 81% to 92% for the more advanced stage 3 and 4 disease, demonstrating that the larger,
96
more frequent and deeper lesions are more easily identified. Thus, despite the marked
AC C
EP
TE D
83
ACCEPTED MANUSCRIPT
improvement in laparoscopic technology over the past decade, persistent endometriosis, and its
98
associated pelvic pain and infertility, remains an issue for many patients and is likely the result
99
of the inability to detect and excise all endometriotic lesions, especially early and superficial
100
lesions, under standard laparoscopic conditions.
101
RI PT
97
Narrow band imaging (NBI) is a technique that uses a specific narrow wavelength of light to
103
change the normal color contrasts of the endoscopic image and improve detection of
104
neovascularization, which is the pathological feature of endometriosis for both superficial and
105
deeper vascularization. Under NBI, two discrete bands of light, one blue at 415 nm and another
106
green at 540 mm, a high contrast image of the tissue surface is created thus enabling improved
107
visualization of blood vessels (8, 9). The clinical utility of NBI in detecting endometriotic
108
lesions was first described in a case report in 2007 by Farrugia et al. (10) who showed that
109
superficial endometriotic lesions were easier to identify with NBI and that their invasiveness was
110
easier to judge than under white light alone. Following this report, Barrueto & Audlin (11)
111
conducted a small pilot study of 20 patients of reproductive age with pelvic pain that
112
demonstrated that the use of NBI resulted in the diagnosis of pathologically-proven
113
endometriosis among four patients whose lesions were missed with visible white light only
114
examination. In addition, the number of confirmed endometriosis lesions increased with the use
115
of NBI in addition to white light. Despite the promise of NBI in identifying endometriotic
116
lesions and in increasing the positive predictive value of suspected lesions, NBI is not currently
117
used in standard laparoscopic surgeries for endometriosis. This is likely because there have been
118
no large multicenter studies of rigorous design that have been conducted to examine detection,
AC C
EP
TE D
M AN U
SC
102
ACCEPTED MANUSCRIPT
119
diagnosis, and discrimination of endometriotic lesions using NBI during the laparoscopic
120
procedure.
121
Thus, to fill this gap in knowledge, we conducted a multisite randomized controlled device trial
123
in two clinical centers to determine the degree to which NBI improves the detection and
124
diagnosis of endometriosis lesions over white light alone. The specific aims of the study were to
125
determine the degree to which NBI improves the diagnosis of endometriosis (diagnostic yield) of
126
laparoscopic examinations compared to use of visible white light-only laparoscopy and to
127
determine if NBI improves sensitivity in the detection of potential endometriosis lesions and
128
reduces false positives compared to visible white light.
M AN U
SC
RI PT
122
129 130
Methods
TE D
131
This study was a randomized controlled device trial conducted in two clinical centers (Mercy
133
Medical Center in Baltimore, Maryland and Lutheran General Hospital in Chicago, Illinois).
134
Participants were randomized at a 3:1 ratio to have laparoscopic examination with white light
135
followed by NBI (white light/NBI) or white light followed by repeat white light examination
136
(white light/white light). This study was approved by the Institutional Review Boards at Mercy
137
Medical Center and Lutheran General Hospital.
139
AC C
138
EP
132
140
Adult women of reproductive age (18 to 49 years of age) undergoing diagnostic laparoscopy for
141
pelvic pain, suspected endometriosis, or infertility were eligible, and were recruited in their
ACCEPTED MANUSCRIPT
physician’s office or on the day of surgery. Women were excluded from the study if they were
143
pregnant or had health issues that their surgeon determined would make laparoscopic surgery
144
unsafe. The pre-specified sample size of 157 was calculated using nQuery 7.0 (12) and
145
determined to be sufficient to detect a minimal difference in sensitivity of 12% between the two
146
imaging modalities with 90% power and a two-sided p-value of 0.05. The sample size accounted
147
for a potential dropout rate of 10%. As such, 141 evaluable patients were needed to detect an
148
estimated difference in sensitivity as little as 12%.
M AN U
150
SC
149
RI PT
142
A total of 167 patients were enrolled, consented, and randomized to either white light/NBI or
152
white light/white light; 124 were from Mercy Medical Center and 43 from Lutheran General
153
Hospital. Of these patients, three patients did not have the laparoscopic surgery for which they
154
were recruited into the study and three patients were withdrawn from the study during surgery at
155
the discretion of the surgeon due to extensive disease. In addition, 11 patients had protocol
156
deviations resulting in unusable primary outcome data. Thus, the final analytic study sample
157
consisted of 150 patients, 112 of whom were assigned to the white light/NBI intervention group
158
and 38 to the white light/white light control group for intention-to-treat (ITT) analysis. A total
159
of three patients (two assigned to white light/NBI and one assigned to white line/white light) did
160
not receive the assigned treatment either due to equipment failure (one patient) or error in
161
performing the randomization assignment (two patients.) Thus 111 patients were imaged with
162
white light and NBI and 39 patients were imaged with white light on both sweeps. Analyses
163
were conducted according to both ITT and to modality received.
164
AC C
EP
TE D
151
ACCEPTED MANUSCRIPT
The materials used were an Olympus HD Endoeye video telescope (Models WA50011A,
166
WA50013A, WA50013L, WA50013T, WA50015L) or Olympus HD Endoeye Laparo-Thoraco
167
Videoscope (Model LTF-VH OLYMPUS) connected to a CLV-180 EVIS EXERA II Xenon
168
Light Source and an Olympus CV-180 EVIS EXERA II Video System Center. An Olympus HD
169
LCD Surgical Monitor, Sony HD recorder PDW-70MD, and Printers OEP-4 were used.
RI PT
165
170
Prior to surgery, each participant provided informed consent and completed a short baseline
172
questionnaire, which collected information on demographics, pain, medical history, health
173
habits, and quality of life. Health-related quality of life was assessed using the question: “In
174
general, would you say your health is excellent, very good, good, fair, or poor?”
M AN U
SC
171
175
For all participants, the laparoscopic procedure was conducted in a standardized fashion with
177
visible white light first, followed by an additional sweep with either white light (control) or NBI
178
(intervention). The second modality was assigned by opening a randomization envelope
179
provided by the study biostatistician after completion of the first examination with white light.
EP
180
TE D
176
Each examination (either with white light or NBI) included systematically visualizing and
182
recording lesions found in the pelvis. For ease of recording the location of the lesions, the pelvis
183
was divided into four quadrants plus the cul-de-sac: right anterior, right posterior, left posterior,
184
left anterior and cul-de-sac. The broad ligaments and ovarian fossa were included in the
185
posterior quadrants. Any detected lesions were photographed, numbered and recorded by
186
method of detection to enable careful tracking for pathologic verification. Additionally, the
187
surgeons’ clinical impression regarding degree of suspicion that the lesions were endometriosis
AC C
181
ACCEPTED MANUSCRIPT
was recorded using a three-point Likert scale to assess their degree of clinical suspicion. All
189
suspected endometrial lesions were included in the documentation, even lesions that may have
190
been latent or healed (otherwise known as white lesions). Once both exams (white light/white
191
light or white light/NBI) were completed and documented, additional areas of the abdomen were
192
evaluated at the surgeon’s discretion. Those participants randomized to white light laparoscopic
193
examination only (white light/white light) had these additional areas evaluated only under white
194
light.
SC
RI PT
188
195
All lesions visualized on either modality, regardless of clinical suspicion were excised if
197
amenable to resection. Lesions were excised using ultrasonic energy and bipolar energy for
198
homeostasis; in very few cases, laparoscopic scissors were used.
M AN U
196
199
Data collected included location, method of detection and clinical diagnosis confidence score.
201
Pathologists were masked to method of detection (NBI, white light, or both NBI and white light).
202
Lesions were recorded as falling within three categories: visible white light-detected only; NBI-
203
detected only; or both NBI- and visible white light-detected. Pathology was performed at each
204
center following a standardized protocol for the diagnosis of endometriosis. All discordant
205
lesions (visualized under only one imaging modality) and a random 10% sample of specimens
206
were re-reviewed by a central independent reviewer at the coordinating center. If disagreement
207
occurred for those chosen as part of the random sampling, the slides for those specimens were
208
adjudicated by a third pathologist at the coordinating center. Thirteen lesions that were not
209
amendable to resection and were ablated during surgery were photographed only and were not
210
included in the analysis.
AC C
EP
TE D
200
ACCEPTED MANUSCRIPT
211
The baseline characteristics of individuals randomized to the two study arms were compared
213
using t-tests for continuous measures and Chi-square tests for categorical variables.
214
RI PT
212
The primary outcome of diagnostic yield and sensitivity was assessed using data from the
216
individuals randomized to receive both modalities (white light/NBI intervention arm). Pathology
217
assessment of resected lesions was the gold standard.
218
SC
215
Diagnostic yield is reported as the number of individuals diagnosed with endometriosis based on
220
findings of NBI versus findings from white light only examination. The number of lesions
221
detected by each modality and percent positive on pathological examination was also
222
determined.
M AN U
219
TE D
223
The sensitivity was assessed for each modality (white light and NBI) and compared using a
225
McNemar’s test. Additional analyses of sensitivity and specificity by imaging modality were
226
calculated based on the clinical impression assessment among those assigned to have both white
227
light and NBI. For the lesions detected with both modalities, clinical impression was classified
228
with each sweep as definitely endometriosis, indefinite, or definitely not endometriosis.
229
Sensitivity and specificity based on clinical impression was calculated for lesions classified
230
clinically as endometriosis. Few lesions were classified as definitely not endometriosis based on
231
clinical impression; thus the indefinite and definitely not categories were combined.
232
AC C
EP
224
ACCEPTED MANUSCRIPT
233
Results were considered statistically significant with p-value ≤0.05 on two-sided tests. All
234
analyses were done using Stata 12.0.
235
Results
RI PT
236 237
Baseline characteristics of participants, according to their randomization group, are displayed in
239
Table 1. The two groups were comparable on age, education, body mass index, pain at baseline,
240
and average number of lesions excised per patient. Patients randomized to the white light/NBI
241
arm were more likely to have had prior laparoscopic surgery than those randomized to white
242
light only arm. The majority of patients reported pelvic pain as the reason for undergoing
243
surgery and taking part in the study, and most had experienced pelvic pain in the four weeks
244
prior to surgery.
M AN U
SC
238
TE D
245
Figure 1 displays 3 sets of photographs from 3 patients showing lesions visualized under NBI but
247
not with white light imaging. The circled lesions were confirmed to be endometriosis by
248
histologic examination.
249
EP
246
Among the group randomized to receive both white light and NBI, four patients had lesions
251
detected with NBI but no lesions detected with white light. Thus, these four patients were
252
diagnosed with endometriosis that would have gone undetected with white light imaging alone.
253
Seventy-two percent were diagnosed with endometriosis among those randomized to have NBI
254
compared to 65.8% among those in the white light imaging only arm. This difference, however,
255
was not statistically significant (Table 2).
AC C
250
ACCEPTED MANUSCRIPT
256
The number of lesions detected and pathology results are shown in Table 3. Among those
258
having both white light and NBI: four lesions were detected only with white light imaging, none
259
of which was confirmed to be endometriosis; 321 lesions were seen on both modalities; and 128
260
were detected only with NBI. NBI led to the detection of an additional 54 confirmed
261
endometriosis lesions, representing 42.2% of the endometriosis lesions seen only with NBI.
RI PT
257
SC
262
Table 4 displays the sensitivity of NBI and white light imaging to detect endometriosis lesions.
264
Among the 256 lesions confirmed as endometriosis by pathologic review, all were detected by
265
NBI for a sensitivity of 100%; 78.9% were detected by white light imaging. The difference in
266
sensitivity for detection of endometriosis between the two modalities was statistically significant
267
(p<.001).
M AN U
263
TE D
268
At the time of surgery, surgeons were asked to provide a clinical impression of the likelihood
270
that detected lesions were endometriosis under each imaging sweep. All lesions were to be
271
resected regardless of the clinical impression. Clinical impression was reported for 70.2% of
272
lesions visualized in patients randomized to receive both imaging modalities. Among those
273
randomized to receive both imaging modalities, and considering only those classified as
274
definitely endometriosis, the clinical impression was correct for 71.5% at the time of white light
275
imaging with a specificity of 35.9%. The sensitivity of the clinical impression improved to
276
84.1%, with a decrease in specificity to 23.9%, with NBI.
AC C
EP
269
277 278
Discussion
ACCEPTED MANUSCRIPT
279
Use of NBI, compared to white light imaging, improved the detection of endometriosis. . Of
281
the pathologically confirmed lesions resected after laparoscopic visualization with both
282
modalities, only 78.9% of the lesions were detected with white light. Thus, compared white light
283
imaging alone, NBI increased the detection of pathologically confirmed endometriosis. The
284
clinical identification of lesions at the time of laparoscopic surgery was also improved with NBI,
285
with 84.1% of pathologically confirmed lesions correctly identified as endometriosis at surgery
286
versus 71.5% with white light imaging. The use of NBI led to the diagnosis of endometriosis in
287
four patients whose endometriosis was not identified by white light examination.
288
. Improving the clinical assessment at the time of surgery can help in determining the extent of
289
resection or ablation that should be undertaken. All together, more lesions were visualized with
290
NBI in combination with white light imaging than with white light imaging alone. NBI
291
improves the detection of early or more superficial endometriosis and thus may help to prevent
292
or extend the interval until recurrence of symptoms (13). Deep lesions of endometriosis or
293
extensive disease are readily visualized with standard illumination so the benefit of NBI-assisted
294
detection of such disease would be would be lessened.
295
The results of this multi-center randomized clinical trial confirm and expand upon the findings of
296
a single-center pilot study of 21 patients (11). In that study, 14 of the 21 patients had additional
297
lesions identified with NBI that were not identified with white light imaging alone, and four
298
patients were diagnosed with endometriosis due to lesions detected only with NBI (11). The
299
results of this multi-center trial demonstrate that NBI is a useful adjunct to standard white light
300
imaging during laparoscopy.
301
AC C
EP
TE D
M AN U
SC
RI PT
280
ACCEPTED MANUSCRIPT
To our knowledge, this is the first randomized clinical trial to evaluate the use of NBI for the
303
detection of endometriosis. NBI is well suited to the detection of active endometriosis since the
304
filtered wavelength selected (415 and 540-nm wavelengths) include the highest hemoglobin
305
absorption (8, 9). This improved ability to visualize vascularization was demonstrated in the
306
current study by the improved sensitivity of NBI compared to white light imaging alone.
RI PT
302
307
The primary aim of the study was to determine the sensitivity of combined white light and NBI
309
compared to the use of standard white light imaging alone. NBI increased sensitivity for
310
detection of endometriosis. However, an increased sensitivity comes at the expense of a lowered
311
specificity. This study protocol led to a higher number of non-endometriotic lesions being
312
removed than might occur in usual clinical practice where removal of lesions may be more
313
targeted. The potential impact of improved detection of subtle endometriosis on symptoms
314
remains unanswered.
TE D
315
M AN U
SC
308
.
While maximum debulking is a goal, the disadvantage of increased identification and
317
excision is the increased risk for surgical complications. Despite the design of the trial requiring
318
excision of all lesions regardless of clinical suspicion, no complications occurred. In clinical
319
practice, NBI may help with discrimination of lesions seen with white light imaging, as well as
320
identifying new lesions, and thus avoid unnecessary excisions. NBI may serve as a training tool,
321
assisting surgeons to better identify lesions with white light imaging. NBI is easy to incorporate
322
into clinical practice when using laparoscopes that are NBI compatible because of the simplicity
323
of switching to and from standard white light and NBI during laparoscopic examination.
AC C
EP
316
ACCEPTED MANUSCRIPT
324
Whether increased debulking of endometriosis deters recurrence and improves overall clinical
325
outcomes requires additional research using a different study design.
326
Conclusion
RI PT
327 328
The addition of NBI to white light imaging increased the number of endometriosis lesions
330
identified during laparoscopy and the diagnosis of endometriosis compared to the use of white
331
light imaging alone. Improved detection of lesions can optimize debulking of endometriosis.
SC
329
M AN U
332 333 334 335
AC C
EP
TE D
336
ACCEPTED MANUSCRIPT
337
References
338
(1) Agic A, Xu H, Finas D, Banz C, Diedrich K, Hornung D. Is endometriosis associated
340
with systemic subclinical inflammation? Gynecol Obstet Invest 2006; 62(3):139-147.
341
(2) Redwine DB. Pelvic endometriosis--the same or different entities in disguise? Fertil Steril
344 345
(3) Garry R, Clayton R, Hawe J. The effect of endometriosis and its radical laparoscopic
SC
343
1998; 70(3):588-589.
excision on quality of life indicators. BJOG 2000; 107(1):44-54.
(4) Jacobson TZ, Duffy JM, Barlow D, Koninckx PR, Garry R. Laparoscopic surgery for
M AN U
342
RI PT
339
346
pelvic pain associated with endometriosis. Cochrane Database Syst Rev
347
2009;(4):CD001300.
(5) Marchino GL, Gennarelli G, Enria R, Bongioanni F, Lipari G, Massobrio M. Diagnosis
349
of pelvic endometriosis with use of macroscopic versus histologic findings. Fertil Steril
350
2005; 84(1):12-15.
351
TE D
348
(6) Kazanegra R, Zaritsky E, Lathi RB, Clopton P, Nezhat C. Diagnosis of stage I endometriosis: comparing visual inspection to histologic biopsy specimen. J Minim
353
Invasive Gynecol 2008; 15(2):176-180.
355 356
(7) Walter AJ, Hentz JG, Magtibay PM, Cornella JL, Magrina JF. Endometriosis: correlation
AC C
354
EP
352
between histologic and visual findings at laparoscopy. Am J Obstet Gynecol 2001; 184(7):1407-1411.
357
(8) Surico D, Vigone A, Bonvini D, Tinelli R, Leo L, Surico N. Narrow-band imaging in
358
diagnosis of endometrial cancer and hyperplasia: a new option? J Minim Invasive
359
Gynecol 2010; 17(5):620-625.
ACCEPTED MANUSCRIPT
361 362 363 364 365
(9) Surico D, Vigone A, Leo L. Narrow band imaging in endometrial lesions. J Minim Invasive Gynecol 2009; 16(1):9-10. (10) Farrugia M, Nair MS, Kotronis KV. Narrow band imaging in endometriosis. J Minim Invasive Gynecol 2007; 14(4):393-394.
RI PT
360
(11) Barrueto FF, Audlin KM. The use of narrowband imaging for identification of endometriosis. J Minim Invasive Gynecol 2008; 15(5):636-639.
(12) Elashoff, J.D. nQuery Advisor Version 7.0. Los Angeles, CA: 2007.
367
(13) Kuroda K, Kitade M, Kikuchi I, Kumakiri J, Matsuoka S, Jinushi M et al. Vascular
SC
366
density of peritoneal endometriosis using narrow-band imaging system and vascular
369
analysis software. J Minim Invasive Gynecol 2009; 16(5):618-621.
M AN U
368
AC C
EP
TE D
370
ACCEPTED MANUSCRIPT
Legend
372
Figure 1.
373
Example of three pathologically-confirmed endometriosis lesions from three different patients
374
that were visualized only with NBI. The Figure shows photos of three areas as visualized with
375
standard white light (a, c, e) and NBI (b, d, f ).NBI = narrow band imaging
376
Patient 1:
Patient 2:
381
Patient 3:
M AN U
d.) Cul de sac, NBI; lesion circled
379 380
c.) Cul de sac, white light; no lesion
SC
b.) Left posterior quadrant, NBI; lesion circled
377 378
a.) Left posterior quadrant, white light; no lesion
RI PT
371
e.) Cul de sac, white light; no lesion f.) Cul de sac, NBI; lesion circled
382
AC C
EP
TE D
383
ACCEPTED MANUSCRIPT Barrueto 1 Table 1. Baseline characteristics according to study arm White light/NBI
White Light/
(n = 112)
White light (n = 38)
SD
Mean
SD
p-value
33.2
7.4
30.6
7.1
.07
Race/Ethnicity
N
%
N
%
White
79
70.5
29
76.3
African American
22
19.6
7
18.4
Hispanic
SC
M AN U
Age at time of surgery, years
RI PT
Mean
6
5.4
1
2.6
5
4.5
1
2.6
Mean
SD
Mean
SD
14.6
2.4
15
2.3
N
%
N
%
52
46.4
19
50.0
47
42.0
16
42.1
13
11.6
3
7.9
Body mass index, kg/m2
N
%
N
%
Less than 25
49
43.7
16
42.1
25 to 29
30
26.8
12
31.6
30 or greater
32
28.6
10
26.3
Missing
1
0.9
0
0.0
Other
General Health
Good
AC C
Fair/Poor
EP
Excellent/Very good
TE D
Education, number of years
.84
.43
.78
.86
ACCEPTED MANUSCRIPT Barrueto 2
SD
Mean
SD
Average pain (Scale 0-10)
4.6
2.7
4.4
3.5
Prior laparoscopic surgery
N
%
Yes
73
65.2
No
38
33.9
Missing
1
0.9
N
%
15
39.5
23
60.5
0
0.0
SC
M AN U
Number of prior laparoscopic surgeries
RI PT
Mean
N
%
N
%
38
33.9
23
60.5
40
35.7
9
23.7
12.5
4
10.5
17
15.2
2
5.3
3
2.7
0
0.0
Reason for study participation
N
%
N
%
Pelvic Pain
84
75.0
28
73.7
Infertility
3
2.7
3
7.9
Pelvic Pain and Infertility
16
14.3
3
7.9
Other
9
8.0
4
10.5
Pelvic pain over the past 4 weeks
N
%
N
%
None 1
14
TE D
2 3 or more
AC C
EP
Missing
.76
.004
.039
.38
ACCEPTED MANUSCRIPT Barrueto 3 101
90.2
35
92.1
No
11
9.8
3
7.9
RI PT
Yes
AC C
EP
TE D
M AN U
SC
NBI = narrow band imaging
.72
ACCEPTED MANUSCRIPT
Barrueto 4 Table 2. Diagnostic yield among individuals randomized to the NBI/WL group versus the WL/WL group (N = 150)
N
%
N
112
81a
72.3
25
N
%
111
65.8
TE D
NBI = narrow band imaging; WL = white light
.45
81*
WL/WL Group N
%
p-valueb
64.1
.29
39
73.0
25
In the NBI/WL group, 4 patients were diagnosed with endometriosis using NBI who would not have been diagnosed
with endometriosis using WL alone b
AC C
Chi-square test for group differences
EP
a
p-valueb
38
Number diagnosed with endometriosis
%
NBI/WL Group
M AN U
Total number in group
WL/WL Group
SC
NBI/WL Group
RI PT
Analysis by Modality Received
Intention to Treat Analysis
ACCEPTED MANUSCRIPT Barrueto 5 Table 3. The number of lesions detected by each modality among individuals who received the combined NBI/WL modality (N=111) Positive by pathology N
4
0
Detected with WL and NBI
321
202
Detected with NBI alone
128
54
Total
AC C
EP
TE D
NBI = narrow band imaging; WL = white light
453
256
M AN U
Detected with WL alone
%
0.0
62.9
42.2
SC
N
RI PT
Total
ACCEPTED MANUSCRIPT Barrueto 6 Table 4. Sensitivity for the detection of endometriosis lesions according to the modality among those receiving the combined NBI/WL modality arm (N=111)
Pathology
Positive Negative
Positive Negative
Yes
256a
193
No
0
4
100.0%
202
123
No
54
74
78.9%
NBI = narrow band imaging; WL = white light a
Detected Yes
RI PT
Pathology
SC
Sensitivityb
WL Sweep
M AN U
Detected
NBI Sweep
54 additional endometrial lesions confirmed pathology positive were detected with NBI
b
AC C
EP
TE D
p<.001 for the McNemar’s test for differences in sensitivity
ACCEPTED MANUSCRIPT Barrueto 7 Table 5. Sensitivity and specificity for the clinical judgment of endometriosis against pathology gold standard by modality among those receiving the combined NBI/WL modality arm (N=111) WL a
NBI
Pathology
RI PT
Pathology Clinical Impression at the time Negative
Total
Positive
Negative
Total
Endometriosis
169
89
258
143
75
218
Not Endometriosis/Indefinite
32
28
60
57
42
99
Sensitivity
84.1%
Specificity
EP
TE D
1 lesion missing white light sweep clinical impression
AC C
a
71.5%
23.9%
NBI = narrow band imaging; WL = white light
SC
Positive
M AN U
of surgery
35.9%
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Precis:
AC C
EP
TE D
M AN U
SC
RI PT
A randomized trial was conducted to determine the sensitivity of laparoscopic examination using narrow band imaging in conjunction white light imaging compared to white light imaging alone to detect endometriosis.
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
http://www.AAGL.org/jmig-22-5-JMIG-D-15-00065