Study population
The population has been described in the first phase of the CONTACT2 trial published by Napoleon et al. [18]. The inclusion criteria were as follows: diagnosis of large (≥ 20 mm) single PCL identified with CT and/or MRI with at least one cavity larger than 13 mm (to allow for CEA and cytohistopathological analyses) without evidence of communication with the main pancreatic duct, without chronic calcifying pancreatitis; no evidence of criteria for malignancy (cyst containing solid masses or mural nodules, presence of metastatic nodes, presence of distant metastases, ascites and vascular infiltration); and scheduled for a diagnostic EUS-FNA procedure. The CONTACT study protocol was approved by the Institutional Review Board of the Institut Paoli Calmettes (Marseille, France), by the French Health Authority (Agence Française de Sécurité Sanitaire des Produits de Santé) and was registered on ClinicalTrials.gov with the following identifier: NCT01563133. The study was performed in accordance with Good Clinical Practice guidelines. All the co-authors had access to the study data and reviewed and approved the final manuscript.
Consensus on therapeutic management
Five pancreatic disease experts (three endosonographers with EUS-FNA expertise and two pancreato-biliary surgeons) from four tertiary hospitals were involved. They were independent from the CONTACT2 prospective study. A consensus to standardize the therapeutic management of patients with PCLs according to the diagnosis and its confidence level was first defined by the panel of experts as follows:
- (1)
Eight diagnostic options were retained: SCA, MCN, BD-IPMN, PC, NEN, indeterminate mucinous lesion, indeterminate lesion, and other cyst types (including cystic solid pseudopapillary neoplasm, congenital pancreatic cyst, and cystic lymphoma).
- (2)
Three therapeutic management options were retained: “neither surgery nor surveillance”, “surveillance”, or “surgery”. If the diagnosis was certain, “neither surgery nor surveillance” was recommended for SCAs and PCs; surgery was recommended for MCNs and NENs; either surgery or surveillance was recommended for BD-IPMNs depending on worrisome features observed in EUS; and either surgery or surveillance was recommended for indeterminate mucinous lesions and for indeterminate lesions. The therapeutic management of other rare types of cysts relied on cyst aetiology. Confidence level modified therapeutic management. For example, surveillance instead of “neither surveillance nor surgery” was recommended for SCAs associated with fair confidence level.
Expert training
The five independent pancreatic experts were informed about nCLE diagnostic performance by two nCLE experts (B.N and M.P) through teleconference. A slide deck was delivered summarizing the comprehensive diagnostic performance of nCLE criteria previously published for the most frequent types of PCLs (NEN, BD-IPMN, MCN, SCA, and PC) [18, 19]. Figure 1 provides illustrations of the common patterns of nCLE in PCLs. Prospectively validated nCLE diagnostic performances were presented, highlighting the very high specificity of nCLE for the diagnosis of SCA, IPMN, MCN and for the differentiation of mucinous from non-mucinous lesions [18, 19]. The five experts were also informed about nCLE limitations: first, the nCLE criterion “field of bright, gray or black particle” lacked specificity for PC characterization. It can be observed in other types of cysts such as the inflammatory cells that can be found in the cystic fluid of cystic tumors following infection, bleeding or previous procedures involving punctures; second, the nCLE criterion “dark spots of cell aggregates surrounded by gray areas of fibrosis and vessels” lacks specificity for NENs. It can be observed in other types of premalignant PCLs; third, the MCN nCLE criterion “epithelial borders” and the BD-IPMN nCLE criterion “papillae”, can be both observed in the same cyst, leading to the nCLE diagnosis of indeterminate mucinous lesions.
Expert evaluations (Fig. 2)
The panel of experts independently reviewed every case blinded to one another’s decisions, to the final diagnoses and to patient therapeutic managements. Two evaluations were performed. The information for each patient was sequentially disclosed in a stepwise manner: progressively from clinical data, to EUS report and to FNA reports comprising histocytological analysis, CEA and amylase levels. nCLE data were not available in evaluation 1, referred to as “EUS-FNA” (the current standard of care for diagnosis). In evaluation 2, referred as “EUS-nCLE-FNA”, the nCLE report, describing prospectively observed nCLE criteria, their associated diagnoses and the quality of nCLE images, was given before the FNA reports. Patients were randomly reordered between evaluations 1 and 2. Evaluation 2 was performed at least 15 days after evaluation 1. For each patient and each evaluation, each expert had to retain one of the eight diagnosis and one of the three therapeutic management options.
Statistical analyses
Baseline characteristics, including demographic and clinical data, were described as percentages and ranges or means and standard deviations, as appropriate. Diagnostic yields were defined as the ratio of conclusive tests (nCLE and FNA cytohistology) over the total number of patients.
To assess interobserver agreement (IOA) between the five experts, Fleiss’ kappas were calculated for the proposed diagnoses and therapeutic management for evaluations 1 and 2. The R package “raters” was used to determine Fleiss’ kappa with 95% CI. Fleiss’ Kappa was interpreted using Landis and Koch-Kappa’s benchmark scale.
For each patient, final consensus on diagnosis and therapeutic management were determined with a majority of at least 3/5 of the experts. The absence of a final consensus led to indeterminate answers. A consensus on the diagnosis of indeterminate mucinous lesions was determined when a majority of experts proposed a diagnosis belonging to the overall group of mucinous lesions (IPMN, MCN, indeterminate mucinous lesion) in the absence of a majority for a more specific diagnosis (IPMN, MCN). The impact on diagnosis was defined as the change in the final consensus diagnosis between evaluations 1 and 2. The impact on therapeutic management was defined as the change in the final therapeutic management consensus between evaluations 1 and 2.
Overall diagnostic and therapeutic management proportion changes between evaluations 1 and 2 were compared using the χ2 square test. Changes between evaluations 1 and 2 regarding the specific diagnostic and therapeutic management categories were compared using McNemar’s test. Fleiss’ kappas from evaluations 1 and 2 were compared using the Z test. p < 0.05 was considered statistically significant.
Endpoints
The main objective of the study was to evaluate the impact of nCLE on PCL therapeutic management: a significant proportion of final consensus changes between evaluations 1 and 2 for a given therapeutic management option was defined as the primary endpoint. The secondary objectives of the study were to evaluate the impact of nCLE on diagnosis and on IOAs for both diagnosis and therapeutic management: a significant proportion of final consensus changes for a given type of diagnosis, a significant increase of IOAs and an increased rate of full agreement between the five experts, between evaluations 1 and 2, for both diagnosis and therapeutic management, were also defined as secondary endpoints.