Endocrine Abstracts

Background: The diagnosis of neuroendocrine tumor disease has been tardy due to a lack of sensitive, noninvasive technology to facilitate identification. In the 1960’s, diagnosis required tumor tissue attained by surgery or biopsy. Identification was by histopathology and immunohistochemistry (1975). Acquisition of information was limited by the morbidity of invasive strategies, inability to reassess and interpretation was user-dependent with low kappa values. The introduction of molecular strategies (1990–2000) using gene expression assays was comparable to histology with NET detection analogous to accuracy in breast or colon cancer. The limitation remained the requirement of tissue.

Aim: To define molecular genomic signatures for NET in blood by developing automation and micro-assay techniques.

Results: In 2005, we defined transcriptome profiles of gastrointestinal and pancreatic NETs and. In 2010, we developed molecular-based tissue classifiers utilizing machine learning and multigene analyses. Although this strategy was >95% accurate it still relied on tissue. Between 2010–2015, we investigated if blood was an alternative compartment for NET gene identification and demonstrated that blood transcriptomes and mRNA expression in whole blood (1ml) identified 51 genes that were co-expressed in tissue (R>0.8). Scores derived from genes co-expressed in blood and tissue were equivalent and provided the basis for a ‘liquid’ biopsy (NETest). Metrics were >90% sensitivity and specificity. Clinical evaluation (n>5,000 patients) demonstrated the multigene assay to have clinical utility as a prognostic biomarker. In 2017, to facilitate measurement, we then automated the assay using targeted PCR and spotted plate technology with significant concordance (P<0.0001) between standard qPCR (R>0.95; n=280 NETs, n=125 controls). To obviate the cold-chain problem, we created an RNA stabilization buffer that maintains NETest signature integrity at room temperature for up to 10 days. Comparison of 120 matched samples show significant concordance in NETest levels (R>0.93). Venipuncture, although safer than biopsy, is still a patient-encumbrance. To move beyond venipuncture, we developed a 50ul of blood fingerprick micro-assay system which is concordant (n=50) (R>0.95) with venipuncture.

Conclusion: We have demonstrated that a NET gene expression assay in blood is accurate and can be automated and miniaturized. The implications for facilitation of diagnosis and point-of-care management warrant investigation.