Introduction In patients with a biliary obstruction, tissue is acquired immediately before drainage during endoscopic retrograde cholangio-pancreatography (ERCP). This is performed by passing a brush inside the bile duct stricture. However, brush cytology has a modest sensitivity (30-57%) for the diagnosis of cancer. A device, called the "G.I.U.M." catheter, allows for the sampling of higher amounts of tissue during ERCP compared to brush cytology. The aim of this study is to compare the diagnostic yield of tissue sampling performed in patients with a suspected malignant biliary stricture using 2 techniques, namely a standard brush catheter and the G.I.U.M..

Protocol design Eligible patients will have tissue sampling performed using both techniques during ERCP, the first technique used being randomly assigned and immediately followed by the other one as well as biliary decompression. All specimens obtained will be subjected to cytopathological examination. After inclusion of the total number of patients, smears will be anonymized and analyzed for diagnosis, cell cellularity and quality. The final clinical diagnosis in each case will be based on cytologic results plus histological examination of biopsy specimens.

Introduction

Many patients with a suspected malignant biliary obstruction are considered unsuitable for surgery because of locally advanced or metastatic disease or poor clinical performance status. Management of these patients is facilitated by a tissue diagnosis at initial endoscopic retrograde cholangio-pancreatography (ERCP). This may obviate further invasive tests, and the most suitable nonsurgical treatment can be initiated without delay. Brush cytology is the most frequently used technique. The procedure is relatively easy to perform, requires little time, and generally is safe. Although its specificity is close to 100%, brush cytology has a modest sensitivity that ranges from 30% to 57% in most published studies .

A device, called the "G.I.U.M." catheter has been developed to increase the amount of tissue available for analysis 1. It consists of a basket with multiple wires that can be passed through the stricture, and grasp tissue between the wires. It has been shown in an uncontrolled study to allow for the diagnosis of malignancy with a high sensitivity (Endoscopy, submitted for publication).

The aim of this study is to compare the diagnostic yield of tissue sampling performed using a standard brush catheter and the G.I.U.M. in patients with a suspected malignant biliary stricture.

Selection of patients

Inclusion criteria:

• Suspicion of malignant biliary stricture without prior histopathologic confirmation;
• Informed consent for ERCP, including sampling and dilation modalities, obtained.

Exclusion criteria:

• Coagulopathy not corrected by administration of vitamin K or fresh frozen plasma;
• Inability to pass a guidewire through the stricture;
• Hilar stricture.

Protocol design and management policy Eligible patients will have tissue sampling performed using both techniques, the first technique used being randomly assigned and immediately followed by the other one. Randomization will be performed by opening an opaque sealed envelope numbered according to a table of random numbers with blocks of 6 patients; each center will receive a pack of 24 numbered envelopes (made by an investigation nurse). A listing of all patients with a biliary stricture diagnosed at ERCP will be maintained (name, surname, date of birth, and date of examination), and reason for non inclusion will be stated.

Methods of tissue sampling Antibiotics will be administered intravenously 30 minutes before ERCP. ERCP with biliary decompression will be performed with standard techniques. 2 After bile duct cannulation, iopromide (Ultravist, 300mgI/mL, Berlex, Montville, N.J.) will be injected, and the level and length of the biliary stricture will be determined. A guidewire will be passed through the stricture and intrahepatic opacification will be completed. A biliary sphincterotomy will be performed using a standard sphincterotome, to facilitate placement of a stent or of a naso-biliary drain. Pancreatography will possibly be attempted, especially if pancreatic disease is suspected.

Tissue sampling will be performed in the order assigned by randomization, according to the following technique:

• Brush: the RX-cytology brush (M00545000, Boston Scientific Corp., Natick, Mass.) will be passed as a unit within its 8 French sheath, over the guidewire through the stricture. Previous stricture dilation will be left at the discretion of the endoscopist. The brush/sheath will be withdrawn immediately below the stricture, the brush will be moved forth and back within the strictured segment at least 10 times. The brush will then be pulled into the catheter, still located immediately below the stricture, and the unit brush/catheter will be removed. A X-ray will be taken with the top of the brush inside the stricture, and the X-ray will be kept in a dedicated file, as well as a X-ray with the stricture opacified with contrast medium. Specimens obtained with the brush will be smeared on glass slides and fixed in 95% ethanol immediately after collection, in the ERCP suite. Immediately after, the brush will be agitated in 10 ml of saline placed in a vial. Finally, the brush segment will be cut from the supportive wire and placed into a container with CytoLytt solution (Cytyc, Crawley, United Kingdom).
• G.I.U.M.: immediately before using the G.I.U.M. catheter, the biliary stricture will be dilated using a 6-mm in diameter MaxForce balloon catheter (Microvasive Endoscopy, Boston Sc.). The G.I.U.M. catheter will be passed as a unit within its 8.5F sheath, alongside the guidewire. Once the basket/sheath is located above the stricture, the basket will be opened and passed through the strictured segment 1 time. The basket will then be pulled into the catheter, and the unit basket/catheter will be removed.. A X-ray will be taken with the meshes of the G.I.U.M. catheter opened inside the stricture, and the X-ray will be kept in a dedicated file. Specimens obtained with the G.I.U.M. will be smeared on glass slides and fixed in 95% ethanol immediately after collection, in the ERCP suite. Immediately after, the G.I.U.M. will be agitated in 10 ml of saline placed in a vial. Finally, all the material remaining between the wires of the G.I.U.M. will be rinsed through the sheath into a vial by perfusing 30 ml of CytoLytt solution (Cytyc) taken from the vial furnished by the manufacturer using a 20-ml sterile syringe into the same vial.

Smears as well as the 2 vials of saline and the 2 vials of CytoLytt will be labeled with the patient's name and the mark "G.I.U.M." or "brush".

Complications possibly detected during ERCP or during the 30 following days will be noted and assessed by using established consensus criteria. 3

Preparation of tissue sample Cytolytt vials: specimens in Cytolytt will be prepared according to the ThinPrep processor operator manual (http://www.thinprep.com/85506Prd/gencyt.htm). Specimens obtained with the ThinPrep technique will be processed for 1 slide as described by the manufacturer. Cell block inclusion will be performed whenever possible.

Smears: specimens will be stained by the Papanicolaou technique for standard cytologic examination.

Cytopathological examination After inclusion of the total number of patients, labels and marks on the smears will be removed and replaced by random numbers. Two non-consecutive random numbers from 1 to 1000 will be selected for each patient, one for the smear obtained with the G.I.U.M., the other for the smear obtained with the brush by JMD. Smears will be re-read by two cytopathologists blinded to the name of the patient, the technique of tissue sampling, previous diagnosis, as well as the relationship between the 2 random numbers for each pair of samples collected from the same patient (so avoiding interpretation of a sample obtained using one of the 2 techniques with the knowledge of the sample obtained from the same patient using the other technique). Indeed, their knowledge will be limited to the fact that a biliary stricture was identified at ERCP. Final diagnosis will be reached by agreement between the 2 cytopathologists. Specimens will be interpreted as normal, atypical (considered benign), highly atypical (suspicious for cancer), and malignant. Cell cellularity and single epithelial cell cellularity will be graded as absent, rare, moderate, or numerous. Finally, nuclear detail will be graded as poor, satisfactory, or excellent. Other data will be recorded as indicated in Table 1.

A list of patients' names for whom cell block inclusion has been performed will be kept, with indication if it was obtained from material collected with the cytobrush or with the G.I.U.M.

Histopathological examination Surgical specimens obtained from patients who undergo duodenopancreatectomy will be subjected to histopathological examination, in particular to detect carcinomatous lymphangitis.

Statistical analysis The final clinical diagnosis in each case will be based on cytologic results plus specimens obtained at surgery, autopsy, via percutaneous puncture or endoscopic ultrasonography with fine needle aspiration, and disease course, including signs of clinical deterioration, death, or a stable course and/or improvement during follow-up. Information will be collected by reviewing hospital records and telephone contact with patients/families and referring physicians 1, 6 and 12 months after ERCP.

For the purpose of calculating sensitivity and specificity, all highly atypical (suspicious for cancer) and malignant diagnoses at cytopathologic examination will be regarded as "positive", and diagnoses of normal and atypical (considered benign) will be regarded as "negative".4 Sensitivity and specificity will be calculated using the Fischer exact test. A p value less than 0.05 will be considered statistically significant.

Based on the hypothesis that the sensitivity for the detection of cancer would be 45 4 and 70% on specimens obtained with the brush and the G.I.U.M. catheter respectively, we calculate that at least 68 patients with a final clinical diagnosis of cancer should be included to reach statistical significance with 5% and 20% alpha and beta error, respectively.

An interim analysis will be performed after collection of resection specimens in 5 patients to evaluate possible lesions to the biliary tract and surrounding tissues.

Inclusion Criteria:

• Suspicion of malignant biliary stricture without prior histopathologic confirmation;
• Informed consent for ERCP, including sampling and dilation modalities, obtained.

Exclusion Criteria:

• Coagulopathy not corrected by administration of vitamin K or fresh frozen plasma;
• Inability to pass a guidewire through the stricture;
• Hilar stricture