Minimally Invasive Detection of High-Risk Pancreatic Cystic Neoplasms Using a Novel Multiparametric Single-Molecule Biosensor

Early detection of pancreatic cancer represents the only chance for cure of this extraordinarily aggressive disease. This implies the detection and surgical resection of high-risk cystic lesions, that is, intraductal papillary mucinous neoplasms (IPMN) and mucinous cystic neoplasms (MCN) bearing high-grade (HG) dysplasia. Application of multidisciplinary guidelines has improved the detection of high-risk precursors, but the proportion of false negative and false positive cases remains high.1 A further improvement in diagnostics has been reached by the introduction of liquid biopsy approaches focusing on the detection of lesion-specific genetic variants in the cyst fluid obtained during endoscopic ultrasound (EUS) investigation.2 For example, the PancreaSeq approach, based on a 74-gene DNA/RNA next generation sequencing panel, reports 82% sensitivity (95% confidence interval [CI]: 0.61-0.93) and 100% specificity (95% CI: 0.91-1.00) for the detection of HG-dysplasia and/or invasive cancer.3 However, their high cost and complexity limit such approaches to highly specialized centers, far away from largely accessible point-of-care (POC) testing. We have previously reported Single Molecule Bio-Electronic Smart System Array for Clinical Testing (SiMBiT), a bioelectronic multiplex array based on single-molecule-with-a-large-transistor (SiMoT) technology.4,5 Briefly, SiMBiT is an ultrasensitive portable diagnostic device that detects biomarkers at the single-molecule level from small fluid samples and delivers results rapidly without complex processing. Here, we report on the results obtained on a series of 92 liquid samples (73 pancreatic cyst fluids and 19 plasma samples) consecutively collected during routine diagnostics from 82 patients (10 with both cyst fluid and plasma) with newly diagnosed cystic lesions of the pancreas (Figure 1A). Clinical guidelines6,7 were applied to determine whether EUS-guided fine-needle aspiration (FNA) or fine-needle biopsy (FNB) was necessary. Cytopathological and/or histopathological evaluation of samples and next generation sequencing-based analysis of cyst fluids was performed at the Institute of Pathology of the University Hospital of Dusseldorf, Germany, a referral center for pancreatobiliary pathology, as described previously.8 Diagnosis was reached by integrating clinical, morphological, and sequencing data. Thirty-three patients (40%) underwent surgical resection. The results of cytopathologic/histopathologic (biopsy or resection specimen) analysis were considered the diagnostic gold standard; if no cells/tissue were available, the most probable diagnosis was rendered based on all other parameters. Samples were then analyzed using SiMBiT, focusing on three parameters: KRASG12D mutation, the protein MUC1, which is associated with pancreatobiliary IPMN (the most aggressive IPMN subtype) and invasive cancer,9 and the protein CD55, a marker of HG-dysplasia in IPMN.10 Applying stringent quality criteria, 13 samples (14%; 12 cyst fluids and 1 plasma sample) were excluded from the analysis, which was finally performed on 79 fluids (61 cyst fluids and 18 plasma samples) from 71 patients (34 with non-neoplastic cysts, mostly pseudocysts, and 37 with neoplastic cysts), using the previously described methods (Figure 1B and C, Supplementary Data).5 Sensitivities, specificities, positive and negative predictive values (PPV/NPV) were computed 1) for the detection of mucin-producing cysts and 2) the detection of HG lesions (including cancer) in cyst fluids and blood plasma. Statistical analyses were performed using SAS (SAS Institute Inc, Cary, NC, USA), Version 9.4, FREQ procedure.