Biomarkers for classification and risk assessment of pancreatic cystic neoplasms

• Main Author: Antunes, Sandra de Jesus Reis Faias
• Publication Date: 2020
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10400.6/11138
• topic: Quisto do pâncreas; Ecoendoscopia com punção com agulha fina (EUS-FNA); Líquido quístico pancreático; CEA; Citologia; Análise Molecular; Biomarcador

Pancreatic cystic lesions (PCLs) are increasing incidental findings due to increased ageing of the population and widespread use of imaging. The main problem in clinical practice has to do with distinguishing the high-risk premalignant and malignant cysts that require surgical treatment from the benign or low-grade dysplastic cysts, which should not be over-treated and might not even require surveillance. The goal of the present work is to perform a comprehensive analysis of biomarkers and diagnostic approaches by Endoscopic Ultrasound with Fine-needle Aspiration (EUS-FNA), in a cohort of patients harboring mostly low-risk cysts under surveillance, which are far more frequent in clinical practice. The PCF analysis performed in this cohort includes studies of genomics (DNA mutations), epigenomics (methylation analysis), metabolomics (glucose), and proteomics (CEA, chromogranin A, NSE), with putative biomarkers encompassing the diagnosis of mucinous and malignant cysts, that require surveillance and surgical resection, respectively. We performed a first meta-analysis comparing current diagnostic methods - CEA and cytology - with KRAS mutations for the diagnosis of mucinous cysts. CEA was the best test for clinically significant cysts (AUC=0.69), cytology performed better in malignant cysts (AUC=0.78), surpassing KRAS mutations (AUC=0.53 and AUC=0.56, respectively). In a second meta-analysis we compared the accuracy of molecular analysis versus micro forceps biopsy (MFB) in the diagnosis of PCLs. The two approaches were identical for diagnosing benign cysts, but molecular analysis was superior for diagnosing both low and high-risk mucinous cysts. In addition to these two meta-analyses, we performed a retrospective study evaluating the added value of KRAS and GNAS mutations in PCF of 52 frozen PCF samples. We conclude that, as compared with conventional tests, these had no added value in the differential diagnosis of PCLs. In another publication, we compared glucose level in PCF with CEA in 82 patients. For mucinous cyst diagnosis, a CEA >192 ng/ml showed an AUC of 0.84 while glucose <50 mg/dl revealed an AUC of 0.86. Besides its higher accuracy, PCF glucose evaluated “on site” with a glucometer is easy, immediate, and requires a minimal amount of PCF. In the next study we sought to determine whether a second EUS-FNA changed the diagnosis or management of pancreatic cysts. We compared the outcome of 105 patients with a single EUS-FNA with that of 23 patients who had a second EUS-FNA. EUS-FNA may be recommended, as it changed management toward surgery in approximately 20% of the patients, particularly with diagnosis of cystic NETs. Following these results, we explored the role of EUS-FNA in small PCLs (<3 cm) in 115 patients with PCLs <3 cm who underwent EUS-FNA. 19/115 were submitted to surgery with 15 malignant or premalignant lesions and the remaining 4 were benign lesions. We conclude that EUS-FNA in lesions <3 cm may improve outcome and cost-effectiveness of surveillance programs, as it confirmed malignancy in 2 out of 5 resected lesions, while it also diagnosed benign cysts who could be released from these programs. In a pilot study with 16 patients, including 4 cystic NETs we aimed at assessing the value of Chromogranin A (CroA) and neuron-specific enolase (NSE) levels in PCF. CroA and NSE levels were higher in cystic NETs with an AUC of 0.94 for CroA and 1 for NSE. These are promising biomarkers to identify pancreatic cystic NETs. Finally, we studied epigenetic changes in the diagnosis of malignant cysts. Methylation changes of GNAS locus were evaluated to understand whether they may contribute to malignant progression of PCLs. Fifty-two samples of PCF were studied. We observed that GNAS locus methylation changes were significantly associated with malignancy. This is the first study to identify methylation changes in the GNAS locus improving diagnosis of malignant PCLs. We end this work proposing a revised diagnostic organogram of PCLs established by current guidelines, that incorporates the results obtained in this dissertation’s research.