Although plasma methoxytyramine provides some utility to predict metastases among patients with pheochromocytomas or paragangliomas, sensitivity is limited. However, incorporation of plasma methoxytyramine in machine learning models, along with other clinical features such as primary tumour location and size, provides a highly accurate, non-invasive approach to predict metastases in patients with pheochromocytomas and paragangliomas, and can thereby guide individualised patient management and follow-up strategies.
Findings from the study that introduces machine learning models to predict metastatic disease than previously possible more accurately are published by Dr. Christina Pamporaki of the Department of Medicine III, University Hospital Carl Gustav Carus, TU Dresden in Dresden, Germany, and colleagues on 18 July 2023 in The Lancet Digital Health.
Pheochromocytomas and paragangliomas are neuroendocrine tumours with up to a 35% hereditary predisposition and an approximately 20% prevalence of metastatic disease. Unlike other tumours, there are no histopathological methods to identify metastatic disease and all pheochromocytomas and paragangliomas must be considered to have variable potential to metastasise. Currently only presence of metastases at sites where no chromaffin tissue should be expected (e.g. bones and lymph nodes) establishes a definitive diagnosis of metastatic disease. Therefore, long-term follow-up is recommended for all patients with pheochromocytomas or paragangliomas.
As metastatic disease cannot be reliably predicted, the present study aimed to prospectively validate the use of the dopamine metabolite, methoxytyramine as a preoperative predictor of metastases in patients with pheochromocytomas or paragangliomas, establish machine learning models that incorporate methoxytyramine with other features to predict metastatic pheochromocytomas or paragangliomas preoperatively, and compare the performance of the selected machine learning models with the predictions of 12 clinical care specialists with expertise in the management of patients with pheochromocytomas or paragangliomas.
In this machine learning modelling study, the study team used cross-sectional cohort data from the PMT trial, based in Germany, Poland, and the Netherlands, to prospectively examine the utility of methoxytyramine to predict metastatic disease in 267 patients with pheochromocytomas or paragangliomas and positive biochemical test results at initial screening. Another retrospective dataset of 493 patients with these tumours enrolled under clinical protocols at National Institutes of Health (00-CH-0093) and the Netherlands (PRESCRIPT trial) was used to train and validate machine learning models according to selections of additional features. The best performing machine learning models were then externally validated using data for all patients in the PMT trial. For comparison, 12 specialists provided predictions of metastatic disease using data from the training and external validation datasets.
Prospective predictions indicated that plasma methoxytyramine could identify metastatic disease at sensitivities of 52% and specificities of 85%. The best performing machine learning model was based on an ensemble tree classifier algorithm that used nine features: plasma methoxytyramine, metanephrine, normetanephrine, age, sex, previous history of pheochromocytoma or paraganglioma, location and size of primary tumours, and presence of multifocal disease.
This model had an area under the receiver operating characteristic curve of 0.942 (95% confidence interval [CI] 0.894–0.969) that was larger (p < 0.0001) than that of the best performing specialist before (0.815, 95% CI 0.778–0.853) and after (0.812, 95% CI 0.781–0.854) provision of SDHB variant data. Sensitivity for prediction of metastatic disease in the external validation cohort reached 83% at a specificity of 92%.
The authors commented that their findings support emerging concepts that machine learning mathematical processes will gain traction in medicine and oncology for their potential to facilitate robust non-invasive diagnostic stratification and guide personalised patient management. Clinicians will benefit from the assistance of the selected machine learning models, as they provide suitable prediction of metastatic pheochromocytomas and paragangliomas and can be easily implemented in digital health care systems.
The study was funded by Deutsche Forschungsgemeinschaft, the Free State of Saxony and TU Dresden, the National Institutes of Health and by the Clinical Research Priority Program of the University of Zurich for the CRPP HYRENE.
Reference
Pamporaki C, Berends AMA, Filippatos A, et al. Prediction of metastatic pheochromocytoma and paraganglioma: a machine learning modelling study using data from a cross-sectional cohort. The Lancet Digital Health; Published online 18 July 2023. DOI: https://doi.org/10.1016/S2589-7500(23)00094-8