Amino acid-based risk stratification model improves prognostic precision in diffuse large B-cell lymphoma

Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma. We hope to provide a new conceptual basis for evaluating patient prognosis and guiding diagnosis and treatment. Twenty patients newly diagnosed with DLBCL were selected as the experimental group, and 10 healthy volunteers composed the control group. The expression of amino acids in the plasma of patients and healthy controls was detected via liquid chromatography-tandem mass spectrometry (LC‒MS/MS). The sparse partial least squares discriminant analysis (sPLS-DA) model was established. Pathway enrichment analysis was performed on the selected differential amino acids. Tryptophan and glutamine, are significantly correlated with prognosis, which can be used as potential DLBCL biomarkers. MATLAB was used to create a partial least squares regression (PLSR) prognostic model and a support vector regression (SVR) machine learning prognostic model. The R²of the PLSR model is 0.33, and the RMSE is 14.22. A paired - sample T - test was conducted on the predicted values and the actual values, with P = 0.999. The R² of the SVR model is 0.89, the MAE is 1.95, and the MBE is 0.77. After training, the PLSR prognostic model and SVR machine can predict the prognosis of DLBCL and provide convenient guidance for the treatment of DLBCL.