Background: Early recognition of cachexia is essential for ensuring the prompt intervention and treatment of cancer patients. However, the diagnosis of cancer cachexia (CC) usually is delayed. This study aimed to establish an accurate and high-efficiency diagnostic system for CC.
Methods: A total of 4834 cancer inpatients were enrolled in the INSCOC project from July 2013 to June 2020. All cancer patients in the study were randomly assigned to a development cohort (n=3384, 70%) and a validation cohort (n=1450, 30%). The least absolute shrinkage and selection operator (LASSO) method and multivariable logistic regression were used to identify the independent predictors for developing the dynamic nomogram. Discrimination and calibration were adopted to evaluate the ability of nomogram. A decision curve analysis (DCA) was used to evaluate clinical use.
Results: We combined 5 independent predictive factors (age, NRS2002, PG-SGA, QOL by the QLQ-C30, and cancer categories) to establish the online dynamic nomogram system. The C-index, sensitivity, and specificity of the nomo-system to predict CC was 0.925 (95%CI, 0.916-0.934, P < 0.001), 0.826, and 0.862 in the development set, while the values were 0.923 (95%CI, 0.909-0.937, P < 0.001), 0.854, and 0.829 in the validation set. In addition, the calibration curves of the diagnostic nomogram also presented good agreement with the actual situation. DCA showed that the model is clinically useful and can increase the clinical benefit in cancer patients.
Conclusions: This study developed an online dynamic nomogram system with outstanding accuracy to help clinicians and dieticians estimate the probability of cachexia. This simple-to-use online nomogram can increase the clinical benefit in cancer patients and is expected to be widely adopted.
Keywords: Cancer cachexia; Dynamic nomogram; LASSO regression; Prediction; Real-world cohort study.
© 2022. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.