Su1371 Prospective Study of the Diagnostic Yield From Histological Versus Cytological Preparation of Specimens Acquired Using Standard Eus Fna Needles

Su1371 Prospective Study of the Diagnostic Yield From Histological Versus Cytological Preparation of Specimens Acquired Using Standard Eus Fna Needles

Abstracts Su1369 A New Model Construted by Logistic Regression of Eus-Elastography for the Differential Diagnosis of Pancreatic Masses: A Two-Step Pr...

289KB Sizes 0 Downloads 21 Views

Abstracts

Su1369 A New Model Construted by Logistic Regression of Eus-Elastography for the Differential Diagnosis of Pancreatic Masses: A Two-Step Prospective Study in 115 Patients Marc Giovannini*, Chiara De Cassan, Stefano Baraldo Paraboli Silva, Erwan Bories endoscopy, Paoli-Calmettes Institute, Aubagne, France Introduction: Endoscopic ultrasound elastography (EUS-EG) is used for the realtime visualization and evaluation of tissue elasticity. Interesting results have been obtained to differentiate malignant and benign pancreatic masses. Strain histogram (SH) corresponds to a graphical representation of the color distribution in a region of interest (ROI), and mean SH (mSH) value is a SH-derived quantitative measure of the global hardness in the evaluated ROI. Matherial and methods: The aim of the present study was: 1st step to quantify the EUS-GCEG pattern of solid malignant pancreatic masses and to differentiate within malignant pancreatic masses using the mSH value. A group of normal patients was enrolled to define the normal pancreatic elasticity. This is a prospective, observational, monocentric study. SH was calculated automatically by machine integrated software in a ROI manually selected by the operator. Four parameters were measured: mSH, standard deviation, kurtosis and skewness. A FNA was performed when a lesion was observed and used as reference standard for the diagnosis. The second step will be to evaluate prospectively these four parameters in patients with a pancreatic mass. Results: A total of 63 patients were included: 9 normals, 45 PA, 9 pNET (5 grade 1, 4 grade 2-3). At elastography, we obtained a statistically significant difference between malignant lesions (p-Net grade 2/3 and adenocarcinoma) and normal pancreas, considering all the parameters evaluated (mSH, standard deviation, kurtosis and skewness), p<0.05. No differences were found in elastographic measurement between p-NET and PA, neither between p-NET grade 1 and p-NET grade 2/3. At univariate analysis we found that all the elastographic parameters measured were capable to differentiate benign versus malignant tissue, with a cut-off calculated with ROC-curves of 43.250 for mSH, 44.63 for standard deviation, 1.5 for the skewness and 5 for kurtosis. At multivariate analysis only mSH resulted statistically significative to distinguish benign versus malign lesions. At a model constructed by logistic regression (9.8958 – 0.3170*moyenne + 0.2989*SD + 0.7935*kurtosis – 7.0642*skewness) a cut off of 0.57 was found to distinguish benign versus malign lesions. The second step of the study has included 52 patients: 10 benign and 42 malign, we obtained a sensibility of 100%, specificity of 86,67%, a positive predictive value of 95,45% and a negative predictive value of 100%. Conclusions: the MM seems able to differentiate pancreatic cancer from benign pancreas. EUS elastography could potentially be used in negative EUS–FNA cases, which represent up to 10-15% of patients with focal masses. A strong suspicion of pancreatic cancer is indicated by values of the strain histogram multivariate model analysis higher than 0.57.

Su1370 Impact of Needle Based Confocal Endomicroscopy (NCLE) in Improving Diagnosis of Pancreatic Cystic Neoplasms : Single Center Experience Virendra Joshi*1, Farnaz Shariati2, Elizabeth D. Verter2, Maya Lucas2, Cris Molina1, John Adams3 1 gastroenterology, ochsner, New Orleans, LA; 2Ochsner, New Orleans, LA; 3 Poindexter Labs, Dallas, TX Background: Pancreatic cystic neoplasms are found in only 0.7% of individuals who have had abdominal imaging. Endoscopic Ultrasound (EUS) has been widely accepted as a diagnostic modality due to better accuracy in identifying pancreatic tumors than Computed Tomography (CT) or Magnetic Resonance Imaging (MRI). EUS-Fine Needle Aspiration (EUS-FNA) has been found to have a specificity of 96.5%. Carcinoembryonic antigen (CEA) tumor marker has also been used to differentiate pancreatic cystic lesions is the most accurate marker of mucinous cystic neoplasms. Recently, another diagnostic technique, needle-based conofocal laser endomicroscopy (nCLE), has been increasingly used for the diagnosis of pancreatic cystic lesions. nCLE with 100% specificity allows evaluation of pancreatic cysts with results similar to that of a pathological diagnosis. In this study, we will compare our standard of care, EUS-FNA with combined CEA and nCLE to determine which combination of diagnostic modalities is a better predictor of definitive diagnoses in pancreatic cystic neoplasms and pancreatic cancer. Methods: In this retrospective chart review of 22 patients with pancreatic cyst in electronic medical record ( EMR) at Ochsner Health system under an IRB approved protocol, specificity, specificity, PPV and NPV of EUS only, or EUS with CEA, nCLE combined were evaluated and diagnostic accuracy were compared with FNA/pathology using McNemar’s test. In evaluation of data knitter, caret and cluster methods were used. worrisome features (cyst size, thick wall, MDP size, non enhanced mural nodule, abrupt changes, distal atrophy and lymphadenopathy) were tested by determining dissimilar calculations using Euclidean distance and later were used in hierarchical clustering to create two clusters based on Euclidean distance Results: Our study shows that regarding diagnosis of pancreatic

AB354 GASTROINTESTINAL ENDOSCOPY Volume 85, No. 5S : 2017

cystic neoplasms, EUS alone had a specificity of 0.75 and a negative predictive value of 0.88. EUS and CEA combined had a specificity of 0.95 and a negative predictive value of 0.90. Finally, EUS with CEA and nCLE combined had a specificity of 0.80 and a negative predictive value of 0.94. Worrisome features clustering was able to discern positive result for pathology with p values of 2.89(-4) Conclusion: C Specificity and NPV are positively impacted by the addition of CEA and nCLE. C Clustering of worrisome factors does predict pathology C A larger cohort is required to further validate our findings. Results:

Sensitivity Specificity EUS EUS and CEA EUS with CEA and nCLE

Negative Predict Value

Postitive Predictive Value

McNemars’ test comparison with FNA/pathology

0 0

0.75 95

0.88 0.90

0 0

PZ 0.44 PZ1

0.5

0.80

0.94

0.2

PZ0.37

Su1371 Prospective Study of the Diagnostic Yield From Histological Versus Cytological Preparation of Specimens Acquired Using Standard Eus Fna Needles Venkatswamy Mahesh*, Uche Nosegbe, Wei Fan Tay, Kathryn Davies, Ashwini Banerjee, Sameer Shaktawat Gastroenterology and Hepatology, Blackpool Teaching hospitals NHS trust, Blackpool, United Kingdom Introduction: Improvements in the EUS FNA needle design and optimisation of tissue acquisition techniques have resulted in better diagnostic yield. Recently there has been considerable interest in histological processing of tissue acquired using newer ‘core’ FNA needles, which are often more expensive than standard needles. We hereby provide data on diagnostic yield of tissue acquired using standard FNA needles. Aim: To compare the diagnostic yield of histological versus cytological preparation of tissue acquired using standard EUS FNA needle and to compare overall diagnostic yield. Methodology Prospective non-blinded randomised study. All patients undergoing EUS guided FNA from November 2015 to November 2016, for solid lesions were included. FNA was performed using Cook Echotip UltraTM or Boston Scientific ExpectTM 19, 22 or 25G needles based on stock availability and location of the lesion, Olympus EU-ME2TM processor and linear echoendoscopes was used. All patients had 4 passes done with the same needle, specimens from each pass were randomly collected in BD CytorichTM or Formalin. Each preservative had 2 passes of material. Results: 108 patients had EUS FNA. Of these 76 had samples both for cytology and histology. 36M:40F. Average age was 68.9 years (9.7). 19G needle was utilized in 16 cases, 20G in 28 cases, 22G in 30 cases and 25G in 2 cases respectively. Adequacy of specimen was reported in 74/76 ( 97.3%). Cytological diagnosis was made in 62/76, DY 81.5%; with cytological and histological diagnosis being entered in 70/76, DY 92.1%. There were 8/76 instances (an increase in diagnostic yield of 10.6%) where histological processing produced a diagnosis in the absence of cytological confirmation. Conclusion: Our study results show that adequate samples can be obtained for histological processing even with standard FNA needles and incremental gain in DY by additional histological processing. Limitations of the study include varying size of needles used, no blinding being used for randomisation of the FNA specimen and the same pathologist reporting both cytology and histology slides.

Su1372 Endoscopic Ultrasound-Guided Fine Needle Aspiration Cytology Vs Biopsy in Diagnosis of Solid Pancreatic Lesions Shahab Ahmadzadeh1, Hosein Shokouh-Amiri1, Gazi Zibari1, Greg Wellman3,1, Jill Ferguson3,1, Sathya Jaganmohan*2,1 1 Willis Knighton Health system, Shreveport, LA; 2Therapeutic Endoscopy, Gastrointestinal Specialists, Shreveport, LA; 3Delta Pathology, Shreveport, LA Introduction: Endoscopic ultrasound guided Fine needle aspiration (EUS-FNA) with rapid onsite evaluation of cytology (ROSE) is the most accepted first line diagnostic method to obtain tissue biopsies for solid pancreatic neoplasm (SPN).

www.giejournal.org