Prediction of metastatic pheochromocytoma and paraganglioma: a machine learning modelling study using data from a cross-sectional cohort

Laboratory Testing for Endocrine Hypertension: Current and Future Perspectives

BACKGROUND

Secondary hypertension (SH) is a form of high blood pressure caused by an identifiable underlying condition. Although, it accounts for a small fraction of the overall hypertensive population, detection and management of SH is of utmost importance, because SH phenotypes carry a high cardiovascular risk and can possibly be cured by timely treatment.

CONTENT

This review focuses on the endocrine causes of SH, such as primary aldosteronism, Cushing syndrome, thyroid disease, pheochromocytoma and paraganglioma, acromegaly, and rare monogenic forms. It discusses current biomarkers, analytical methods, and diagnostic strategies, highlighting advantages and limitations of each approach. It also explores the emerging -omics technologies that can provide a comprehensive and multidimensional assessment of SH and its underlying mechanisms.

SUMMARY

Endocrine SH is a heterogeneous and complex condition that requires proper screening and confirmatory tests to avoid diagnostic delays and improve patient outcomes. Careful biomarker interpretation is essential due to potential interferences, variability, and method-dependent differences. Liquid chromatography-tandem mass spectrometry is a superior method for measuring low-concentration hormones and metabolites involved in SH, but it requires expertise. Omics approaches have great potential to identify novel biomarkers, pathways, and targets for SH diagnosis and treatment, especially considering its multifactorial nature.

Preoperative prediction of metastatic pheochromocytoma and paraganglioma using clinical, genetic, and biochemical markers: A cohort study

BACKGROUND

The prevalence of metastatic pheochromocytoma and paraganglioma (PPGL) is approximately 15%-20%. Although there are indicators to assess metastatic risks, none of them predict metastasis reliably. Therefore, we aimed to develop and validate a scoring system using clinical, genetic, and biochemical risk factors to preoperatively predict the metastatic risk of PPGL.

METHODS

In the cross-sectional cohort (n = 180), clinical, genetic, and biochemical risk factors for metastasis were identified using multivariate logistic regression analysis, and a novel scoring system was developed. The scoring system was validated and compared with the age, size of tumor, extra-adrenal location, and secretory type (ASES) score in the longitudinal cohort (n = 114).

RESULTS

In the cross-sectional cohort, pseudohypoxia group-related gene variants (SDHB, SDHD, or VHL), methoxytyramine >0.16 nmol/L, and tumor size >6.0 cm were independently associated with metastasis after multivariate logistic regression. Using them, the gene variant, methoxytyramine, and size of tumor (GMS) score were developed. In the longitudinal cohort, Harrell's concordance index of the GMS score (0.873, 95% confidence interval [CI]: 0.738-0.941) was higher than that of the ASES score (0.713, 95% CI: 0.567-0.814, p = 0.007). In the longitudinal cohort, a GMS score ≥2 was significantly associated with a higher risk of metastasis (hazard ratio = 25.07, 95% CI: 5.65-111.20). A GMS score ≥2 (p < 0.001), but not ASES score ≥2 (p = 0.090), was associated with shorter progression-free survival.

CONCLUSION

The GMS scoring system, which integrates gene variant, methoxytyramine level, and tumor size, provides a valuable preoperative approach to assess metastatic risk in PPGL.

Intraoperative radiofrequency ablation for unresectable abdominal paraganglioma: a case report

For pheochromocytoma and paraganglioma (PPGL), the efficacy of percutaneous ablative therapies in achieving control of metastatic tumors measuring <3 cm had been demonstrated in only few reports, and intraoperative radiofrequency ablation (RFA) of locally invasive primary PPGLs has not been reported. We presented the case of a 31-year-old man who had a 9-cm functioning unresectable PPGL. He was treated with 13 cycles of cytotoxic chemotherapy without objective tumor response, according to the Response Evaluation Criteria in Solid Tumors (RECIST). Subsequently, magnetic resonance imaging revealed a 9.0 × 8.6 × 6.0-cm retroperitoneal mass that extended to the inferior portion of the inferior vena cava, the inferior mesenteric artery, and the infrarenal aorta. Biochemical evaluation demonstrated high level of plasma normetanephrine (20.2 nmol/L, normal range <0.9 nmol/L). Genetic investigation showed the germline pathogenic variant c.1591delC (p. Ser198Alafs*22) in the SDHB gene. I131-metaiodobenzylguanidine scintigraphy was negative and Ga68-dotatate PET-CT scan showed high tumor uptake without distant metastases. On open laparotomy, tumor debulking was not possible. Therefore, intraoperative RFA was performed by a highly experienced team of interventional radiologists. At 12 months after the RFA, the tumor volume decreased from 208 to 45 mL (78%), plasma normetanephrine decreased from 20.2 to 2.6 nmol/L (87%), and the doxazosin dose was reduced from 16 to 8 mg/day. To our best knowledge, this was the first report on intraoperative RFA that markedly reduced the size of a large primary unresectable PPGL, along with clinical and biochemical responses.

Phaeochromocytoma and paraganglioma

Phaeochromocytomas and paragangliomas are rare catecholamine-producing neuroendocrine tumours which can potentially cause catastrophic crises with high morbidity and mortality. This best practice article considers the causes and presentation of such tumours, screening and diagnostic tests, management of these patients and consideration of family members at risk.

Management of phaeochromocytoma and paraganglioma in patients with germline SDHB pathogenic variants: an international expert Consensus statement

Adult and paediatric patients with pathogenic variants in the gene encoding succinate dehydrogenase (SDH) subunit B (SDHB) often have locally aggressive, recurrent or metastatic phaeochromocytomas and paragangliomas (PPGLs). Furthermore, SDHB PPGLs have the highest rates of disease-specific morbidity and mortality compared with other hereditary PPGLs. PPGLs with SDHB pathogenic variants are often less differentiated and do not produce substantial amounts of catecholamines (in some patients, they produce only dopamine) compared with other hereditary subtypes, which enables these tumours to grow subclinically for a long time. In addition, SDHB pathogenic variants support tumour growth through high levels of the oncometabolite succinate and other mechanisms related to cancer initiation and progression. As a result, pseudohypoxia and upregulation of genes related to the hypoxia signalling pathway occur, promoting the growth, migration, invasiveness and metastasis of cancer cells. These factors, along with a high rate of metastasis, support early surgical intervention and total resection of PPGLs, regardless of the tumour size. The treatment of metastases is challenging and relies on either local or systemic therapies, or sometimes both. This Consensus statement should help guide clinicians in the diagnosis and management of patients with SDHB PPGLs.

Large language models streamline automated machine learning for clinical studies

A knowledge gap persists between machine learning (ML) developers (e.g., data scientists) and practitioners (e.g., clinicians), hampering the full utilization of ML for clinical data analysis. We investigated the potential of the ChatGPT Advanced Data Analysis (ADA), an extension of GPT-4, to bridge this gap and perform ML analyses efficiently. Real-world clinical datasets and study details from large trials across various medical specialties were presented to ChatGPT ADA without specific guidance. ChatGPT ADA autonomously developed state-of-the-art ML models based on the original study's training data to predict clinical outcomes such as cancer development, cancer progression, disease complications, or biomarkers such as pathogenic gene sequences. Following the re-implementation and optimization of the published models, the head-to-head comparison of the ChatGPT ADA-crafted ML models and their respective manually crafted counterparts revealed no significant differences in traditional performance metrics (p ≥ 0.072). Strikingly, the ChatGPT ADA-crafted ML models often outperformed their counterparts. In conclusion, ChatGPT ADA offers a promising avenue to democratize ML in medicine by simplifying complex data analyses, yet should enhance, not replace, specialized training and resources, to promote broader applications in medical research and practice.

A comprehensive characterisation of phaeochromocytoma and paraganglioma tumours through histone protein profiling, DNA methylation and transcriptomic analysis genome wide

BACKGROUND

Phaeochromocytomas and paragangliomas (PPGLs) are rare neuroendocrine tumours. Pathogenic variants have been identified in more than 15 susceptibility genes; associated tumours are grouped into three Clusters, reinforced by their transcriptional profiles. Cluster 1A PPGLs have pathogenic variants affecting enzymes of the tricarboxylic acid cycle, including succinate dehydrogenase. Within inherited PPGLs, these are the most common. PPGL tumours are known to undergo epigenetic reprograming, and here, we report on global histone post-translational modifications and DNA methylation levels, alongside clinical phenotypes.

RESULTS

Out of the 25 histone post-translational modifications examined, Cluster 1A PPGLs were distinguished from other tumours by a decrease in hyper-acetylated peptides and an increase in H3K4me2. DNA methylation was compared between tumours from individuals who developed metastatic disease versus those that did not. The majority of differentially methylated sites identified tended to be completely methylated or unmethylated in non-metastatic tumours, with low inter-sample variance. Metastatic tumours by contrast consistently had an intermediate DNA methylation state, including the ephrin receptor EPHA4 and its ligand EFNA3. Gene expression analyses performed to identify genes involved in metastatic tumour behaviour pin-pointed a number of genes previously described as mis-regulated in Cluster 1A tumours, as well as highlighting the tumour suppressor RGS22 and the pituitary tumour-transforming gene PTTG1.

CONCLUSIONS

Combined transcriptomic and DNA methylation analyses revealed aberrant pathways, including ones that could be implicated in metastatic phenotypes and, for the first time, we report a decrease in hyper-acetylated histone marks in Cluster 1 PPGLs.

Development and validation of machine-learning models for the difficulty of retroperitoneal laparoscopic adrenalectomy based on radiomics

Objective

The aim is to construct machine learning (ML) prediction models for the difficulty of retroperitoneal laparoscopic adrenalectomy (RPLA) based on clinical and radiomic characteristics and to validate the models.

Methods

Patients who had undergone RPLA at Shanxi Bethune Hospital between August 2014 and December 2020 were retrospectively gathered. They were then randomly split into a training set and a validation set, maintaining a ratio of 7:3. The model was constructed using the training set and validated using the validation set. Furthermore, a total of 117 patients were gathered between January and December 2021 to form a prospective set for validation. Radiomic features were extracted by drawing the region of interest using the 3D slicer image computing platform and Python. Key features were selected through LASSO, and the radiomics score (Rad-score) was calculated. Various ML models were constructed by combining Rad-score with clinical characteristics. The optimal models were selected based on precision, recall, the area under the curve, F1 score, calibration curve, receiver operating characteristic curve, and decision curve analysis in the training, validation, and prospective sets. Shapley Additive exPlanations (SHAP) was used to demonstrate the impact of each variable in the respective models.

Results

After comparing the performance of 7 ML models in the training, validation, and prospective sets, it was found that the RF model had a more stable predictive performance, while xGBoost can significantly benefit patients. According to SHAP, the variable importance of the two models is similar, and both can reflect that the Rad-score has the most significant impact. At the same time, clinical characteristics such as hemoglobin, age, body mass index, gender, and diabetes mellitus also influenced the difficulty.

Conclusion

This study constructed ML models for predicting the difficulty of RPLA by combining clinical and radiomic characteristics. The models can help surgeons evaluate surgical difficulty, reduce risks, and improve patient benefits.