Endoscopic ultrasound fine-needle biopsy to assess DAXX/ATRX expression and alternative lengthening of telomeres status in non-functional pancreatic neuroendocrine tumors

Abstract

BACKGROUND/OBJECTIVES: Death domain-associated protein (DAXX) and/or α-thalassemia/mental retardation X-linked (ATRX) chromatin remodeling genes mutations and alternative lengthening of telomeres (ALT) activation are associated with more aggressive behavior of non-functional pancreatic neuroendocrine tumors (NF-PanNETs). We aimed to evaluate the reliability of such markers on endoscopic-ultrasound fine-needle biopsy (EUS-FNB) specimens. METHODS: Patients who underwent EUS-FNB and subsequent surgical resection for PanNETs between January 2017 and December 2019 were retrospectively identified. Immunohistochemistry (IHC) to evaluate DAXX/ATRX expression and fluorescence in situ hybridization (FISH) for ALT status were performed. Primary outcome was the concordance rate of markers expression between EUS-FNB and surgical specimens. Secondary aims were association between markers and lesion aggressiveness, their diagnostic performance in predicting aggressiveness, and agreement of preoperative and post-surgical Ki67-based grading. RESULTS: Forty-one NF-PanNETs (mean diameter 36.1 ± 26.5 mm) were included. Twenty-four showed features of lesion aggressiveness. Concordance of expressions of DAXX, ATRX, and ALT status between EUS-FNB and surgical specimens were 95.1% (κ = 0.828; p < 0.001), 92.7% (κ = 0.626; p < 0.001), and 100% (κ = 1; p < 0.001), respectively. DAXX/ATRX loss and ALT-positivity were significantly (p < 0.05) associated with metastatic lymphnodes and lymphovascular invasion. The combination of all tumor markers (DAXX/ATRX loss + ALT-positivity + grade 2) reached an accuracy of 73.2% (95%CI 57.1-85.8) in identifying aggressive lesions. Pre- and post-operative ki-67-based grading was concordant in 80.5% of cases (k = 0.573; p < 0.001). CONCLUSION: DAXX/ATRX expression and ALT status can be accurately evaluated in a preoperative setting on EUS-FNB samples, potentially improving the identification of patients with increased risk and poorer prognosis.