Validation of diagnostic characteristics of needle based confocal laser endomicroscopy in differentiation of pancreatic cystic lesions

Abstract

Background and aims: Endoscopic ultrasound (EUS)-guided needle-based Confocal Laser Endomicroscopy (nCLE) characteristics of pancreatic cystic lesions (PCLs) have been identified in studies where the gold standard surgical histopathology was available in a minority of patients. There are diverging reports of interobserver agreement (IOA) and paucity of intraobserver reliability (IOR). Thus, we sought to validate current EUS-nCLE criteria of PCLs in a larger consecutive series of surgical patients. Methods: A retrospective analysis of patients who underwent EUS-nCLE at a single center was performed. For calculation of IOA (Fleiss’ kappa) and IOR (Cohen’s kappa), blinded nCLE-naïve observers (n = 6) reviewed nCLE videos of PCLs in two phases separated by a 2-week washout period. Results: EUS-nCLE was performed in 49 subjects, and a definitive diagnosis was available in 26 patients. The overall sensitivity, specificity, and accuracy for diagnosing a mucinous PCL were 94 %, 82 %, and 89 %, respectively. The IOA for differentiating mucinous vs. non-mucinous PCL was “substantial” (κ = 0.67, 95 %CI 0.57, 0.77). The mean (± standard deviation) IOR was “substantial” (κ = 0.78 ± 0.13) for diagnosing mucinous PCLs. Both the IOAs and mean IORs were “substantial” for detection of known nCLE image patterns of papillae/epithelial bands of mucinous PCLs (IOA κ = 0.63; IOR κ = 0.76 ± 0.11), bright particles on a dark background of pseudocysts (IOA κ = 0.71; IOR κ = 0.78 ± 0.12), and fern-pattern or superficial vascular network of serous cystadenomas (IOA κ = 0.62; IOR κ = 0.68 ± 0.20). Three (6.1 % of 49) patients developed post-fine needle aspiration (FNA) pancreatitis. Conclusion: Characteristic EUS-nCLE patterns can be consistently identified and improve the diagnostic accuracy of PCLs. These results support further investigations to optimize EUS-nCLE while minimizing adverse events.Study registration: NCT02516488

Impact of needle-based confocal laser endomicroscopy on the therapeutic management of single pancreatic cystic lesions

“…nCLE feasibility and safety for the assessment of PCLs were demonstrated in a pilot study [12] and were validated in an international multicenter study [13]. Subsequent clinical investigations described the correlation between nCLE images and histological features, and comprehensive nCLE criteria were established for the characterization of the most frequent types of PCL (NEN, MCN, BD-IPMN, SCA, and PC) [14][15][16]. A recent international study confirmed excellent interobserver and intraobserver agreement for the interpretation of nCLE images for the diagnosis of PCLs [17].…”

Endoscopic ultrasonography evaluation for pancreatic cysts: Necessity or overkill?

“…The results demonstrated a high level of confidence of 72.6% and interobserver agreement showed kappa of 0.71, 0.65, and 0.9 for SCN, MCN, and pseudocyst, respectively. 67,68 A validation study by Krishna et al 62 also showed similar results with moderate to substantial interobserver agreement. However, a study from a different group found only poor to fair agreement in vessels, villi, dark clumps, reticular pattern and debris among experts with a mean accuracy of 46%.…”

“…59 The diagnostic performance of EUS-guided nCLE ranges from 70% to 96% (Table 9). [60][61][62][63][64][65][66] Recently, several studies have been carried out aiming to validate the nCLE findings. The results demonstrated a high level of confidence of 72.6% and interobserver agreement showed kappa of 0.71, 0.65, and 0.9 for SCN, MCN, and pseudocyst, respectively.…”

Endoscopic diagnosis of cystic lesions of the pancreas

“…40 A validation study of EUS-guided nCLE criteria in 49 patients with pancreatic cystic lesions found that the interobserver agreement (j = 0.67) and interobserver reliability (j = 0.78) for diagnosing mucinous cysts were high and that the overall sensitivity, specificity, and accuracy for diagnosing mucinous cysts were 94%, 82%, and 89%, respectively. 41 Thus, EUS-guided nCLE has been carried out to identify IPMN, SN, MCN, and/or pseudocysts, and especially to differentiate mucinous and non-mucinous pancreatic cysts. A limitation of this novel technique is that the diagnostic yield for the differential diagnosis of benign and malignant cysts is unknown.…”