A viral infection prediction model for patients with r/r B-cell malignancies after CAR-T therapy: a retrospective analysis.

BACKGROUND

Chimeric antigen receptor T cell (CAR-T) therapy for relapsed/refractory (r/r) B cell acute lymphoblastic leukemia (B-ALL) and B cell non-Hodgkin lymphoma (B-NHL) patients has shown promising effects, but side effects such as viral infections have been observed.

METHODS

A total of 45 patients with r/r B-ALL and r/r B-NHL were included in this retrospective study. Patient demographics were recorded, with the primary endpoint being viral infection within 3 months post CAR-T treatment. Univariate and multivariate logistic regression analyses and least absolute shrinkage and selection operator (LASSO) regression analysis were used to analyze independent factors. The patients were divided into a training cohort of 28 and a validation cohort of 17 to construct a prediction model based on determined independent factors. The model's discrimination and calibration were assessed using the receiver operating characteristic curve (ROC), calibration plot, and decision curve analysis (DCA curve).

RESULTS

The univariate and multivariate logistic regression analyses of the 43 patients showed that low baseline lymphocyte ratio was an independent risk factor and using granulocyte colony-stimulating factor (G-CSF) early was a protective factor for viral infection after CAR-T therapy in patients with B-ALL and B-NHL. Based on that, the area under the ROC curve (AUC) of the training cohort and validation cohort was 0.935 (95% CI 0.837-1.000) and 0.869 (95%CI 0.696-1.000), respectively, showing excellent predictive value.

CONCLUSIONS

We established a nomogram to predict the factors' influence on viral infection after CAR-T therapy and found that the ratio of baseline lymphocytes and using G-CSF early or lately were able to predict viral infection after CAR-T therapy in r/r B-ALL and B-NHL.