Citations

Background: The use of artificial intelligence (AI) in health care has grown exponentially with the promise of facilitating biomedical research and enhancing diagnosis, treatment, monitoring, disease prevention, and health care delivery. We aim to examine the current state, limitations, and future directions of AI in thyroidology. Summary: AI has been explored in thyroidology since the 1990s, and currently, there is an increasing interest in applying AI to improve the care of patients with thyroid nodules (TNODs), thyroid cancer, and functional or autoimmune thyroid disease. These applications aim to automate processes, improve the accuracy and consistency of diagnosis, personalize treatment, decrease the burden for health care professionals, improve access to specialized care in areas lacking expertise, deepen the understanding of subtle pathophysiologic patterns, and accelerate the learning curve of less experienced clinicians. There are promising results for many of these applications. Yet, most are in the validation or early clinical evaluation stages. Only a few are currently adopted for risk stratification of TNODs by ultrasound and determination of the malignant nature of indeterminate TNODs by molecular testing. Challenges of the currently available AI applications include the lack of prospective and multicenter validations and utility studies, small and low diversity of training data sets, differences in data sources, lack of explainability, unclear clinical impact, inadequate stakeholder engagement, and inability to use outside of the research setting, which might limit the value of their future adoption. Conclusions: AI has the potential to improve many aspects of thyroidology; however, addressing the limitations affecting the suitability of AI interventions in thyroidology is a prerequisite to ensure that AI provides added value for patients with thyroid disease.

e18807 Background: CUP represents both a diagnostic and a management challenge. Despite improvements through sophisticated imaging modalities, specific and sensitive testing using immunohistochemistry, a concrete determination of the site of origin of the primary tumor can still be a diagnostic challenge Hereby, we described the genomic landscape of patients with CUP treated at our institution. Methods: Retrospective study of patients with CUP treated at Mayo Clinic Rochester MN from 2010 to 2021. Next generation sequencing (NGS) information was abstracted, and alterations were ranked into different levels of actionability following the Onco KB framework. All data were analyzed by the SPSS version 22.0 software (SPSS, Chicago, IL, United States). Results: From a total of 214 patients, we identified 67 (32%) patients in whom NGS was performed. Among patients with available NGS the median age was 64 years (range: 35-86) and 51% were female. the most prevalent histology included adenocarcinoma (39%), carcinoma NOS (31%), and neuroendocrine tumors (12%). NGS on tumor (t-NGS) was available in 56 (84 %) patients, circulating tumor NGS (b-NGS) was available in 12 (18%) patients, and only 1 patient had both testing available. Among the 56 patients with available t-NGS, MSI-high was detected in 9 patients, whereas among the 12 patients with available b-NGS, MSI-high was detected in 8 patients; yielding an MSI-high detection rate of 16% and 67% for t-NGS and b-NGS. In this cohort, there were a total of 474 genomic alterations detected from which 103 (22 %) were able to be ranked into one therapeutic level following the OncoKB framework. Genomic alterations were most observed among patients with carcinoma (n = 50, 48.53%) adenocarcinoma (n = 33, 32.03 %), squamous cell carcinoma (n = 3, 2.91 %) and neuroendocrine tumor (n = 2, 1.94%). Following the OncoKB framework, 39-level 1 genomic alterations were detected in a total of 14 genes. The most common level 1 genomic alterations comprised the BRCA2 (20.51%), NF1 ATM (17.95%) and HER2/NEU (ERBB2) (15.38%) genes. Therapeutic level 2 included 51 genomic alterations in 13 genes of which KRAS (17.65%), BRAF (15.69%), and BRCA2 (15.69%) were the most prevalent mutated genes. Additionally, level R comprised 13 genomic alterations in the following genes: KRAS (69.23) NRAS (23.08%) and PDGFRA (7.69%). Conclusions: CUP are malignancies rich in potentially actionable alterations, and NGS of tissue and blood samples can inform systemic therapy decisions. Alteration-specific targeted therapy may be an option for a substantial proportion of patients. NGS of tumor tissue or blood should be considered for all patients with CUP.

e18802 Background: CUP is a rare cancer with an incidence of 7–12 cases per 100,000/year. CUP is defined as metastatic cancer with no identifiable primary tumor despite comprehensive evaluation. Despite its heterogeneous clinicopathological presentation, the treatment of CUP has primarily been with platinum-based combination chemotherapies but results are poor and survival is short. Here, we describe survival outcomes in real-world dataset of patients with CUP seen at a tertiary referral center. Methods: Retrospective study of patients treated at Mayo Clinic, Rochester MN, from 2010 to 2021. Survival probabilities were estimated with the Kaplan-Meier method. We used Cox proportional hazard regression with forward modeling to explore the association between each variable and death. Overall survival (OS) analysis was conducted in the entire cohort, and it was stratified by histological subtype. For the hypothesis test, we considered p < 0.05 to be statistically significant; all tests were two-sided. All data were analyzed by the SPSS version 22.0 software (SPSS, Chicago, IL, United States). Results: From a total of 214 patients, 198 patients were included in the survival analysis. The patients were predominantly white (87%), female (50.2%), with a mean age of 63 (SD: 13) without history of smoking (51%). The predominant histology was adenocarcinoma (30%), followed by neuroendocrine neoplasms (NEN) (27%), carcinoma NOS (25%), squamous cell carcinoma (SCC) (10%) and non-classifiable tumors (7%). The median follow-up time was 23 months, and medial OS was 23 months in the entire cohort. OS by histology was as follows: 51 months for NEN, 34 months for SCC, 18 months for carcinomas NOS and 11 months for adenocarcinomas (p < 0.001) (fig.1). Similar OS trends were observed when we restricted the analysis to poorly differentiated tumors. The 3- and 5-year cumulative survival by histology were as follows: 7% and 0% for adenocarcinoma, 25% and 8% for carcinoma NOS, 39% and 17% for SCC, and 56% and 22% for NEN. In Cox regression models we did not identify any prognostic factors among patients with adenocarcinoma. Conclusions: Patients with adenocarcinoma of unknown primary possess a considerable challenge and the OS is short. CUP of other histologies have higher OS. This study highlights the need of accurate tumor identification and better treatment, perhaps as a multiomics therapeutic approach.