A Comparison of Narrow Band Imaging (NBI) and Standard White Light Laparoscopy to Detect Endometriosis

Introduction:

Endometriosis is a relatively common chronic gynecological condition that affects approximately 10% of all women of reproductive age. It is a pelvic inflammatory disease that is characterized by the presence of endometrial glands and stroma outside of the uterine cavity. Typical symptoms of endometriosis include dysmenorrhea, pelvic pain, and infertility; the severity of pain associated with this disease often leads to a considerable decrease in quality of life.

The standard treatment for severe pelvic pain and infertility is to surgically remove endometriotic areas. Identifying all endometriotic lesions is paramount to "optimal endometriosis debulking." The inability to see all endometriosis lesions has been thought to be a factor for patients with little or no relief following surgery. Using the Narrow Band Imaging (NBI) method has the potential to improve visualization of endometriosis lesions, assist in debulking and thus, result in improved clinical outcomes.

NBI is a technique that uses a specific narrow wavelength of light to change the normal color contrasts of the endoscopic image and improve detection of neovascularization, which is the pathological feature of endometriosis for both superficial and deeper vascularization. This type of imaging has the potential to offer improved discrimination of lesions, increasing diagnostic yield as well as resulting in more complete debulking.

The results of a pilot study of 21 patients conducted at Mercy Medical Center demonstrated that use of NBI resulted in the diagnosis of pathologically-proven endometriosis among 4 patients whose lesions were missed with visible white light only examination. In addition, the number of confirmed endometriosis lesions increased with the use of NBI in addition to white light. The study also suggested that NBI may assist in discriminating endometriosis from other lesions, thus reducing false positives: one patient had no lesions detected with NBI and the lesions biopsied based on visible white light examination were not endometriosis at pathologic examination.

Lesion discrimination under white light examination based on lesion color has been previously described in the literature. Red (also known as clear and/or pink) lesions are considered to be active and early lesions on endometriosis. White lesions may indicate areas of fibrosis and may be latent or dormant lesion of endometriosis. The red and white lesions are also known as the "subtle" lesions of endometriosis. Black (also known as brown, chocolate green, purple or blue) represent a more severe stage of the disease. Black lesions are also known as "gunshot" or "powder-burn" lesions and are considered the "typical" lesions of endometriosis. Mixed color lesions can include a mixture of any color. It has been speculated that the subtle lesions of endometriosis are more likely to be positive lesions and to be missed under white light examination. The red lesions are also considered to have more vessels of smaller diameter and may also include peritoneal microlesions and therefore may be more easily detected under NBI.

The investigators will conduct a multi-center study with the following primary aims to determine the degree to which NBI improves the detection and diagnosis of endometriosis lesions.

1. Determine the degree to which NBI improves the diagnosis of endometriosis (diagnostic yield) of laparoscopic examinations compared to use of visible white light-only laparoscopy.

2. Determine if NBI improves sensitivity in the detection of potential endometriosis lesions and reduces false positives compared to visible white light.

Secondary aim:

Determine the impact of the use of NBI, and presumed more complete removal of endometriotic lesions, compared to use of white light examination on reported severity of pain at 6-weeks, 3-months, and 6 months following surgery.

The results of this study would significantly advance the state of art for the diagnosis and treatment of endometriosis. Improved detection and treatment of endometrial lesions should translate to a significant health benefit to women suffering from endometriosis. Potential health benefits include improved diagnosis of the etiology of pelvic pain, therefore guiding appropriate treatment management for women with pelvic pain due to endometriosis and improved debulking of lesions.

The proposed study design is a multisite randomized controlled device trial conducted in two clinical centers. Women will be randomized to have laparoscopic examination with white light followed by NBI (white light/NBI) examination or white light followed by repeat white light examination (white light/white light). Women of reproductive age (<50 years of age) undergoing diagnostic laparoscopy for pelvic pain, suspected endometriosis, or infertility will be eligible. A systematic examination covering the four quadrants of the pelvis will be conducted using white light followed by either NBI or white light to identify and document endometriotic lesions.

Study Aims

Hypotheses:

The use of NBI in addition to white light examination will improve the diagnostic yield of endometriotic lesions at the time of laparoscopy compared to only using white light examination.

The use of NBI in addition to white light examination will improve the sensitivity of detecting endometriotic lesions and reduce false positives at laparoscopy compared to only using white light examination.

Secondarily, the use of NBI will be associated with a greater reduction in pain at the 6-week, 3-month, and 6-month follow-up compared to the use of white light examination alone because of improved lesion identification and debulking.

Pass/Fail Criteria for Primary Aims. In this randomized clinical trial, failure is defined as no statistically significant difference in the primary outcomes of diagnostic yield or sensitivity between the white only examination arm and the white light followed by NBI examination arm.

Study Design and Methods The investigators will test the hypotheses by conducting a multicenter device trial. Participants will be randomized to either white light/NBI or white light/white light at a 3:1 ratio. In both study groups, each of the quadrants of the pelvis will first be examined by white light.

Once a full exam is performed using the white light with lesion location documented with photographs, a randomization envelope will be opened to reveal the patient's second modality (NBI or white light). Each area will then be inspected by this second modality with photographs taken for each quadrant, documenting lesion location. All identified lesions in areas amenable to resection will then be resected, labeled by method of detection, and sent for pathologic examination.

Study Outcomes:

The primary outcomes are the comparison of the diagnostic yield, sensitivity, and false positive rate of visible white light (white light/white light) versus NBI (white light/NBI) based on pathology gold standard.

Methods:

Investigator Training:

The investigators will be trained on the visualization of endometriosis under NBI

- Viewing WL and NBI images of potential endometrial lesions of various types

- The procedural use of NBI

- Completion of study forms to ensure data integrity

The participating surgeons and assistants from each site will be trained on the procedural use of NBI by Olympus and the coordinating center. A minimum of three cases of patients suspected to be positive for endometriosis will be watched by surgeons and coordinators from remaining sites. The training surgeon will perform the sweep through the quadrants of the pelvis. Each area will be viewed under white light. The training surgeon will ask the surgeons to identify suspicious areas that they would biopsy. Then the training surgeon will switch to NBI and once again ask the training surgeons to identify suspicious areas that they would biopsy. This will be done to ensure that all participating surgeons are able to identify suspicious areas under both imaging modalities to learn what potential endometrial lesions look like under NBI.

Olympus will visit each site prior to the first procedure and train the assistant(s) on the completion of the study case report forms. Olympus will also monitor adherence to the protocol and the completion of the first study case report forms during the time of study start (first patient enrollment) to ensure the surgeon is performing the procedure according to the protocol and that the assistant is accurately capturing all of the required information.

Study Population:

The population will include women of reproductive age undergoing diagnostic laparoscopy for suspected endometriosis. The study population may include:

- Patients undergoing initial diagnostic laparoscopy for pelvic pain, suspected endometriosis or infertility

- Patients previously evaluated with laparoscopy undergoing repeat laparoscopy for suspected recurrence or onset of endometriosis

Recruitment:

Women will be recruited through the gynecologic practices of the participating physicians. At the pre-surgery visit, patients will be informed of the purpose of the study as well as the risks and benefits of the procedure and will be asked to sign an informed consent.

After signing the informed consent, women will be asked to complete a baseline questionnaire to obtain information on demographic characteristics, medication history, and symptoms. A 10-centimeter (cm) visual analog scale and a validated endometriosis symptom questionnaire [the Endometriosis Health Profile-30 (EHP-30)] will also be administered to assess pain. The EHP-30 is an instrument specifically designed to measure pain and quality of life among women diagnosed with endometriosis. The EHP-30 consists of a core questionnaire with 30 items and five scales and six modular parts consisting of 23 questions. The EHP-30 has demonstrated both high reliability and validity.

Equipment:

- OLYMPUS HD EndoEYE Video Telescope (Models WA50011A, WA50013A, WA50013L, WA50013T, WA50015L) or OLYMPUS HD EndoEYE Laparo-Thoraco Videoscope LTF-VHOLYMPUS CLV-180 EVIS EXERA II Xenon Light Source OLYMPUS CV-180 EVIS EXERA II Video System Center OLYMPUS HD LCD Surgical Monitor

- SONY HD recorder PDW-70MD

- Printers OEP-4

Examination Procedures:

The laparoscopic examination will be conducted in a standardized fashion with visible white light (standard), followed by an additional sweep with either white light (control) or NBI. The second look examination will be determined by random assignment after completion of the first examination with white light.

The examination will include visualizing each quadrant of the pelvis (Right Anterior, Right Posterior, Left Posterior, Left Anterior).

Each area will first be viewed and photographed under white light. The absence or presence of suspected endometrial lesions will be noted. If suspected lesions are seen, the number of lesions, the extent of the lesions and the surgeon's confidence in the lesion being endometriosis will be recorded. Photographs will be labeled by quadrant and all lesions will be circled and numbered. Clinicians will record their degree of confidence on a 5 point likert scale that the lesion to be excised is a lesion of endometriosis.

Following visualization, photographic identification and documentation of the quadrants under white light has been completed, the study coordinator will open the randomization envelope and reveal the second modality to be used. The physician will then repeat the exam and photographs with either white light (control) or under NBI (switching on NBI).

Again, under the second modality, the absence or presence of suspected endometrial lesions will be noted. If suspected lesions are seen, the number of lesions, the extent of the lesions, the lesion color and the surgeon's confidence in the lesion being endometriosis will be recorded. A photograph will be taken, labeled and all lesions will be circled and numbered. Clinicians will record their degree of confidence on a 5 point likert scale that the lesion to be excised is a lesion of endometriosis. The surgeon will record if the lesion was also visualized under the 1st imaging modality or under only one of the modalities.

The order at which the areas are examined through the sweep will be at the surgeon's discretion. Examination by both sweeps will be photographed. Photos of all areas and detected lesions will be printed, labeled and lesions will be circled and numbered.

All suspected endometrial lesions will be included in the analysis, even lesions that may be latent or healed (otherwise known as white lesions).

Once both exams are complete, additional areas of the abdomen can be evaluated at the surgeon's discretion; those randomized to white light only will continue with only white light examination.

All lesions detected by either method that are amenable to resection will be numbered and biopsied. Data collected will include location, method of detection and clinical diagnosis confidence score. Pathologists will be masked to method of detection (NBI versus white light).

Lesions will be recorded according as falling within three categories: visible white light-detected only; NBI-detected only; or both NBI- and visible white light-detected. Pathology will be performed at each center following a standardized protocol for the diagnosis of endometriosis. Pathologists at each institution will be masked to the method of examination used to identify the specimen. All discordant lesions (visualized under only one imaging modality), a random 10% sample of specimens and a random 10% of images from each center will be re-reviewed by a central independent reviewer at the coordinating center.

Lesions that are not amendable to resection will be photographed only and will not be included in the data analysis.

Data Collection:

1. Patient baseline questionnaire: symptoms, medication history, demographic information, pain measurement (including 10-cm visual analog scale and EHP-30)

2. Photo documentation of each light examination

3. Case report forms:

- Recording of location of lesion by method of examination

- Notation of lesion color

- Clinical impression of detected lesions - degree of confidence on a scale of 1 to 5 that the lesions observed are endometriosis

- Notation of lesions excised

- Photographic documentation of lesions

- Change in surgical treatment (larger excision/smaller excision/organ removal) based on results of visualization under the different imaging modalities.

4. Pathology reports

5. Medical record review

6. Provider questionnaire: planned follow-up treatment after surgery

7. Follow-up questionnaires for pain at 6-weeks, 3-months and 6-months; follow-up surveys will include both the 10-cm visual analog scale and the EHP-30

Quality Control:

Quality control will be maintained by standardizing the protocols across the centers. As stated above, in each patient, the locations of all lesions suspected to be endometriosis detected with each modality will be documented with photographs. All discordant lesions (seen under only one imaging modality) and a random 10% sample of the images and specimen samples from each center will be re-reviewed by a central independent reviewer at the coordinating center. The results between the site physician and the central reviewer will be compared. If the site and central reviewer have disagreement on the photographed lesion or pathology, adjudication will be performed between the site and the central reviewer. If a consensus still cannot be reached, a third party (surgeon and/or pathologist (from the remaining site) will be used as the tie breaker.

Randomization Procedure Subjects will be assigned to one of two groups in a stratified randomized design. All subjects will undergo a white light examination followed by either a second examination with white light or an examination using NBI. Investigators will be blinded to second look imaging method until after the initial white light examination is complete. Randomization to either white light/NBI or white light/white light will occur at a 3:1 ratio. Each site is treated as independent and subjects will be combined across sites for analyses.

A randomization table will be generated using Microsoft Excel at the beginning of study recruitment. The randomization number will be a three digit number and will form the last 3 digits of the study ID. Sealed envelopes will be provided to each site by the coordinating center to be opened after the initial white light examination.

Sample size

Sample size in the original protocol for this study was calculated as described in section 12.1 below. Because data from a third site of this multicenter trial (Atlanta Northside) will not be used, the total sample size goal (N = 170) in the original protocol will not be reached. Mercy was approved by the IRB in September 2012 to enroll 100 patients total; the Chicago site is approved to enroll 57 patients. Thus, the total sample size goal has been revised to 157; minimal detectable differences under the conditions of 90% power are described in section 12.2 below.

Sample size calculations from original protocol, dated 5/14/2010

Sample size calculations were based on the primary outcomes of diagnostic yield and sensitivity as well as the secondary outcome of impact on pain symptoms. The sample size is largely driven by the primary outcome of diagnostic yield. A sample size of 170 will provide 90% power to detect an improvement in diagnostic yield of 20% (95% CIs 72%-85% white light and 95%-100% NBI), based on a randomization scheme of 3:1 for white light/NBI versus white light/white light examination. This sample size takes into consideration a 25% drop out/not evaluable rate for the following reasons: incomplete visualization of the pelvis, loss to follow-up for pain assessment or patients not evaluable for pain assessment (a subset of infertility patient who may not have pain as a presenting symptom).

This sample size provides 90% power to detect an average increase of one additional endometriosis lesion with the NBI examination (white light/NBI group) compared to white light only examination (white light/white light group). To note, this power calculation was based on data from the pilot study that showed an average of 2 lesions (per patient) found using white light and 3 lesions (per patient) detected using NBI; the calculation assumes a common standard deviation of 1.

Data used for sample size calculations:

Diagnostic yield: The calculations for sample size were based on information obtained from a pilot study that examined patients using white light followed by NBI during diagnostic laparoscopy. In that study, 4 additional patients were diagnosed with endometriosis with the addition of NBI to white light examination for an absolute improvement of 27% (4/15 patients with endometriosis were found with NBI examination only) (Barrueto and Audlin, 2008). Using these data, sample size calculations were conducted using nQuery 6.01 (Elashoff, 2005) for a two group chi-square test with unequal Ns (3:1 ratio). A two group chi-square test with a 0.05 two-sided significance level will have 90% power to detect the difference between a Group 1 (white light/white light) diagnostic yield of 0.79 and a Group 2 (white light/NBI) diagnostic yield of 0.99 when the sample sizes in each group are 34 and 101, respectively (a total sample size of 135).

Sensitivity and false positive rate:

In the pilot study, 15 of 20 patients examined were diagnosed with endometriosis on histology. All 15 patients were detected with NBI and white light. Four patients were diagnosed by lesions that were detected only with NBI. Based on these data, the estimated sensitivities of white light and NBI are 73% and 100% respectively. Given these sensitivities and the small total sample size, 95% confidence intervals are wide (48%-90% white light and 80%-100% NBI). At the lesion level, pilot data have shown an average of 2 lesions seen under white light compared to an average of 3 lesions using NBI. Using a common standard deviation of 1, the current sample size has at least 90% power to detect an average difference of at least one additional lesion.

Pain difference:

The study will also consider pain reduction as a clinically relevant outcome. A recently published study by Abbott et al (2003) reported that participants undergoing laparoscopy for potential endometriosis on average rated non-menstrual pelvic pain prior to surgery as 6 on a 10-point visual analogue scale, with 10 being the worst pain. In the Abbott et al (2003) and other similar studies, laparoscopic surgery using white light was shown to be associated with a 30% reduction in pain at 6 month follow up and 50%-75% at 12 month follow up with increasing variance (Abbott et al., 2004, Jarrell et al., 2005). An RCT from Jarrell et al. (2005) showed a reduction in pain score of 1.9 (SD 0.5) at 6 month follow up and 2.6 (SD 0.5) at 12 month follow up, although the sample size of this study was small (N=16 overall, N=7 for laparoscopy arm). This study will test the hypothesis that NBI will result in an additional 20% decrease in pain compared to what has been observed in these previously published studies examining the reduction in pain using white light laparoscopy.

Using the unbalanced randomization design with 1 patient randomized to white light/white light for every 3 patients randomized to white light/NBI , a two group t-test with a 0.050 two-sided significance level will have 90% power to detect the difference in pain change with a Group 1 (white light/white light) mean change of 1.9 and a Group 2 (white light/NBI) mean of 2.28, a difference in means of 0.38 with a common SD of 0.5, when the sample sizes in the two groups are 25 and 75, respectively (a total sample size of 100). We project that the majority of patients enrolled (over 70%) will be symptomatic with pelvic pain despite the indication for surgery (i.e. pelvic pain or infertility). Only 1 of the 2 study center sites has a significant population of women undergoing laparoscopy for evaluation and treatment of infertility.

Amended sample size The sample size of the study has been adjusted downward to 157 due to the loss of one of the study centers for accrual. The revised sample size considers the outcomes of diagnostic yield and sensitivity of clinical determination of identified lesions at the time of surgery, by modality. A review of the first 69 patients enrolled and assigned to the combined white light/NBI arm shows an average detection rate for any lesion to be 4.2 per patient using NBI versus 2.9 per patient with white light. The minimal detectable difference in diagnostic yield and sensitivity was determined using nQuery 6.01 (Elashoff, 2005). With the current projected sample size of 157 patients, the minimal detectable difference between the two modalities, with 90% power, in diagnostic yield proportions and sensitivity estimates is 12.2%. Thus, the power of the study remains sufficient to detect a clinically meaningful difference in detection of endometriosis by white light and NBI.

Data Analysis The investigators hypothesize that NBI will improve detection of endometriosis lesions thus improving the diagnosis of endometriosis. For the primary outcome of diagnostic yield, the reference standard will be the pathologic confirmation of excised lesions where any one lesion within an individual detected by either method will confirm the diagnosis of endometriosis. The investigators will address the detection of endometriosis by comparing the diagnostic yield of the white light/white light group versus the white light/NBI group. The investigators will use a McNemar's test for the proportion comparison, which accounts for the pairing of assessment within a participant and makes no assumptions about the distribution of samples. At the lesion level, the investigators will assess the average change in the number of lesions detected between the white light/white light group to the white light/NBI group using independent samples t-test. The investigators will also calculate and compare the sensitivities and false positive rates for the white light/white light group and the white light/NBI group. Pain across the study period will be assessed by comparing average change in pain score (using both the VAS and the EHP-30 scores) between the two groups. The difference in average pain change scores will be assessed using an independent samples t-test. Differences in average pain between baseline and follow-up will be assessed using paired t-tests.

Data Management A copy of the consent form will be sent to the Data Coordinating Center (DCC). Report forms will be sent by fax or electronically (secured/encrypted) to the DCC data entry. The data will be checked for consistency by using selected double entry where 1/10th of the entered data will be re-entered and checked for errors. Data cleaning procedures will be done using SAS with extensive program files to ensure the correctness of skip patterns and responses. After baseline information is collected and entered, an analysis will be done to confirm comparable baseline characteristics across the two groups.

Confidentiality:

Hard copies of questionnaires and interview records will be locked in file cabinet with limited keyed access to research personnel. Computer data files are also kept on the Mercy Medical Center Intranet in a password protected file that is accessible only by research staff.

Any results from this study will be published using data from the group as a whole. No individual will be identified in any reports.

Risks/Benefits Assessment

Risks/Discomforts:

White light examination is the current standard of care. NBI itself does not incur any additional risk. However, the use of NBI may result in the identification of additional lesions and thus incur the risk of additional resection at the time of laparoscopy. Additional sites identified for biopsy using NBI have no additional risks for removal as compared to white light. Lesions identified which are associated with higher risk structures such as the bowel and the diaphragm will not be removed.

Benefits:

It is uncertain to what extent participants may benefit clinically and this will be clearly stated in the consent form. If diagnosis of endometriosis is improved, then patients may benefit from improved medical management. If additional lesions are identified and resected, patients may benefit from an improved clinical outcome.

Withdrawal from protocol All participants are informed that taking part in this study is completely voluntary and they have the right to not participate. If they do not want to join the study, or if they wish to withdraw from the study, they will still receive the same quality of medical care.

Modifications to the Protocol Any modifications to the protocol will be sent to the IRB for approval.

Protocol Deviations The IRB (and other review boards as required for this study) will be notified of any protocol deviations.

Reporting of Serious Adverse Events and Unanticipated Problems The protocol has minimal additional risks beyond the usual risks associated with diagnostic laparoscopy. Should any unanticipated problems develop, the investigators will attend to/address the problem and notify institutional IRBs and the sponsor.