Artificial intelligence in endocrinology: a comprehensive review.

BACKGROUND AND AIM: Artificial intelligence (AI) has emerged as a promising technology in the field of endocrinology, offering significant potential to revolutionize the diagnosis, treatment, and management of endocrine disorders. This comprehensive review aims to provide a concise overview of the current landscape of AI applications in endocrinology and metabolism, focusing on the fundamental concepts of AI, including machine learning algorithms and deep learning models. METHODS: The review explores various areas of endocrinology where AI has demonstrated its value, encompassing screening and diagnosis, risk prediction, translational research, and "pre-emptive medicine". Within each domain, relevant studies are discussed, offering insights into the methodology and main findings of AI in the treatment of different pathologies, such as diabetes mellitus and related disorders, thyroid disorders, adrenal tumors, and bone and mineral disorders. RESULTS: Collectively, these studies show the valuable contributions of AI in optimizing healthcare outcomes and unveiling new understandings of the intricate mechanisms underlying endocrine disorders. Furthermore, AI-driven approaches facilitate the development of precision medicine strategies, enabling tailored interventions for patients based on their individual characteristics and needs. CONCLUSIONS: By embracing AI in endocrinology, a future can be envisioned where medical professionals and AI systems synergistically collaborate, ultimately enhancing the lives of individuals affected by endocrine disorders.

New insights into the comorbid conditions of Turner syndrome: results from a long-term monocentric cohort study.

PURPOSE: Many questions concerning Turner syndrome (TS) remain unresolved, such as the long-term complications and, therefore, the optimal care setting for adults. The primary aim of this long-term cohort study was to estimate the incidence of comorbid conditions along the life course. METHODS: A total of 160 Italian patients with TS diagnosed from 1967 to 2010 were regularly and structurally monitored from the diagnosis to December 2019 at the University Hospital of Bologna using a structured multidisciplinary monitoring protocol. RESULTS: The study cohort was followed up for a median of 27 years (IQR 12-42). Autoimmune diseases were the comorbid condition with the highest incidence (61.2%), followed by osteoporosis and hypertension (23.8%), type 2 diabetes (16.2%) and tumours (15.1%). Median age of onset ranged from 22 years for autoimmune diseases to 39 years for type 2 diabetes. Malignant tumours were the most prominent type of neoplasm, with a cumulative incidence of 11.9%. Papillary thyroid carcinoma was the most common form of cancer, followed by skin cancer and cancer of the central nervous system. Only one major cardiovascular event (acute aortic dissection) was observed during follow-up. No cases of ischaemic heart disease, heart failure, stroke or death were recorded. CONCLUSIONS: This cohort study confirms the need for continuous, structured and multidisciplinary lifelong monitoring of TS, thus ensuring the early diagnosis of important comorbid conditions, including cancer, and their appropriate and timely treatment. In addition, these data highlight the need for the increased surveillance of specific types of cancer in TS, including thyroid carcinoma.

Computerized tomography texture analysis of pheochromocytoma: relationship with hormonal and histopathological data.

OBJECTIVES: Pheochromocytomas are rare tumors which can present with heterogeneous secretion profiles, clinical manifestations, and radiologic appearance. Under a histopathological point of view, they can be characterized as more or less aggressive with the Pheochromocytoma of the Adrenal gland Scaled Score (PASS) and the Grading system for Adrenal Pheochromocytoma and Paraganglioma (GAPP) score. The aim of this study is to analyze the texture analysis characteristics of pheochromocytoma and identify whether the texture analysis can yield information aiding in the diagnosis and the characterization of those tumors. METHODS: Radiological, biochemical, and histopathological data regarding 30 consecutive patients with histologically confirmed pheochromocytoma were analyzed. Images obtained in the unenhanced, late arterial, venous, and delayed phases were used for the texture analysis. RESULTS: Urinary epinephrine and metanephrine levels showed a significant correlation (R2 = 0.946; R2 = 699) in the multivariate linear model with texture features, as well as Ki-67 (R2 = 0.397), PASS score (R2 = 0.182), GAPP score (R2 = 0.705), and cellularity showed a significant correlation (R2 = 0.389). The cluster analysis based on radiomic features resulted in 2 clusters, with significative differences in terms of systolic and diastolic blood pressure values at the time of diagnosis (p = 0.025), GAPP score (4 vs 6, p = 0.05), histological pattern (1-2, p = 0.039), and comedonecrosis (0% vs 50%, p = 0.013). CONCLUSION: In conclusion, our study provides the proof of concept for the use of texture analysis on contrast-enhanced CT images as a noninvasive, quantitative tool for helping in the characterization of the clinical, biochemical, and histopathological features of pheochromocytoma.

Steroid biomarkers for identifying non-classic adrenal hyperplasia due to 21-hydroxylase deficiency in a population of PCOS with suspicious levels of 17OH-progesterone.

OBJECTIVE: We aimed at defining the most effective routine immunoassay- or liquid chromatography-tandem mass spectrometry (LC-MS/MS)-determined steroid biomarkers for identifying non-classic adrenal hyperplasia due to 21-hydroxylase deficiency (21-NCAH) in a PCOS-like population before genotyping. METHODS: Seventy PCOS-like patients in reproductive age with immunoassay-determined follicular 17OH-progesterone (17OHP) ≥ 2.00 ng/mL underwent CYP21A2 gene analysis and 1-24ACTH test. Serum steroids were measured by immunoassays at baseline and 60 min after ACTH stimulation; basal steroid profile was measured by LC-MS/MS. RESULTS: Genotyping revealed 23 21-NCAH, 15 single allele heterozygous CYP21A2 mutations (21-HTZ) and 32 PCOS patients displaying similar clinical and metabolic features. Immunoassays revealed higher baseline 17OHP and testosterone, and after ACTH stimulation, higher 17OHP (17OHP60) and lower cortisol, whereas LC-MS/MS revealed higher 17OHP (17OHPLC-MS/MS), progesterone and 21-deoxycortisol and lower corticosterone in 21-NCAH compared with both 21-HTZ and PCOS patients. Steroid thresholds best discriminating 21-NCAH from 21-HTZ and PCOS were estimated, and their diagnostic accuracy in identifying 21-NCAH from PCOS was established by ROC analysis. The highest accuracy was observed for 21-deoxycortisol ≥ 0.087 ng/mL, showing 100% sensitivity, while the combination of 17OHPLC-MS/MS ≥ 1.79 ng/mL and corticosterone ≤ 8.76 ng/mL, as well as the combination of ACTH-stimulated 17OHP ≥ 6.77 ng/mL and cortisol ≤ 240 ng/mL by immunoassay, showed 100% specificity. CONCLUSIONS: LC-MS/MS measurement of basal follicular 21-deoxycortisol, 17OHP and corticosterone seems the most convenient method for diagnosing 21-NCAH in a population of PCOS with a positive first level screening, providing high accuracy and reducing the need for ACTH stimulation test.

Radiologically defined lipid-poor adrenal adenomas: histopathological characteristics.

BACKGROUND: Adrenal lipid-poor adenomas (LPA) are defined by high unenhanced density (≥ 10 HU), and absolute and relative contrast medium washout > 60% and > 40%, respectively, at computerized tomography (CT). To date, no thorough histopathological characterization has been performed in those frequent lesions (one-third of adrenal adenomas). Our aim was to analyze the histopathological characteristics of adrenal LPA. METHODS: Patients with LPA (n = 57) were selected among consecutive subjects referred for an adrenal incidentaloma or ACTH-independent Cushing syndrome. FluoroDeoxyGlucose-Positron Emission Tomography (FDG-PET) was performed in 37 patients. In patients treated by adrenalectomy (n = 17), Weiss score and Lin-Weiss-Bisceglia score (in tumors composed entirely or predominantly of oncocytes) were calculated. RESULTS: Radiological parameters did not differ among patients with ACTH-independent Cushing syndrome (n = 6) and those with adrenal incidentalomas associated with primary aldosteronism (n = 2), autonomous cortisol secretion (n = 14), or non-functioning (n = 35). Patients treated by adrenalectomy had larger tumors (28.9 ± 11.2 vs 17.3 ± 8.4 mm, P < 0.001), higher CT unenhanced density (29.1 ± 11.0 vs 23.1 ± 9.0 HU, P = 0.043), and FDG-PET adrenal uptake (9.0 ± 6.4 vs 4.4 ± 2.3 SUV, P = 0.003) than non-operated ones. Oncocytic features > 75% of the tumor were detected in 12/17 cases (70.6%). Five of those showed borderline-malignant histopathological characteristics by Lin-Weiss-Bisceglia score. Among remaining non-oncocytic tumors, 1/5 had a Weiss score ≥ 3. Overall, 6/17 tumors (35.3%) had borderline-malignant potential. Radiological parameters were similar between patients with benign and borderline-malignant tumors. CONCLUSIONS: Adrenal LPA are a heterogeneous group of tumors, mostly composed of oncocytomas. Up to 1/3 of those tumors may have a borderline-malignant potential at histopathology.

Thyroid dysfunctions secondary to cancer immunotherapy.

BACKGROUND: Immunotherapy is a firmly established pillar in the treatment of cancer, alongside the traditional approaches of surgery, radiotherapy, and chemotherapy. Like every treatment, also cancer immunotherapy causes a diverse spectrum of side effects, collectively referred to as immune-related adverse events. OBJECTIVE: This review will examine the main forms of immunotherapy, the proposed mechanism(s) of action, and the incidence of thyroid dysfunctions. METHODS: A comprehensive MEDLINE search was performed for articles published up to March 30, 2017. RESULTS: Following the pioneering efforts with administration of cytokines such as IL-2 and IFN-g, which caused a broad spectrum of thyroid dysfunctions (ranging in incidence from 1 to 50%), current cancer immunotherapy strategies comprise immune checkpoint inhibitors, oncolytic viruses, adoptive T-cell transfer, and cancer vaccines. Oncolytic viruses, adoptive T-cell transfer, and cancer vaccines cause thyroid dysfunctions only rarely. In contrast, immune checkpoint blockers (such as anti-CTLA-4, anti-PD-1, anti-PD-L1) are associated with a high risk of thyroid autoimmunity. This risk is highest for anti-PD-1 and increases further when a combination of checkpoint inhibitors is used. CONCLUSIONS: Cancer patients treated with monoclonal antibodies that block immune checkpoint inhibitors are at risk of developing thyroid dysfunctions. Their thyroid status should be assessed at baseline and periodically after initiation of the immunotherapy.