A novel predictive algorithm to personalize autologous T-cell harvest for chimeric antigen receptor T-cell manufacture

Abstract

Background aimsThe most widely accepted starting materials for chimeric antigen receptor T-cell manufacture are autologous CD3+ T cells obtained via the process of leukapheresis, also known as T-cell harvest. As this treatment modality gains momentum and apheresis units struggle to meet demand for harvest slots, strategies to streamline this critical step are warranted. MethodsThis retrospective review of 262 T-cell harvests, with a control cohort of healthy donors, analyzed the parameters impacting CD3+ T-cell yield in adults with B-cell malignancies. The overall aim was to design a novel predictive algorithm to guide the required processed blood volume (PBV) (L) on the apheresis machine to achieve a specific CD3+ target yield. ResultsFactors associated with CD3+ T-cell yield on multivariate analysis included peripheral blood CD3+ count (natural log, ×109/L), hematocrit (HCT) and PBV with coefficients of 0.86 (95% confidence interval [CI], 0.80–0.92, P < 0.001), 1.30 (95% CI, 0.51–2.08, P = 0.001) and 0.09 (95% CI, 0.07–0.11, P < 0.001), respectively. The authors’ model, incorporating CD3+ cell count, HCT and PBV (L), with an adjusted R2 of 0.87 and root-mean-square error of 0.26 in the training dataset, was highly predictive of CD3+ cell yield in the testing dataset. An online application to estimate PBV using this algorithm can be accessed at https://cd3yield.shinyapps.io/cd3yield/. ConclusionsThe authors propose a transferrable model that incorporates clinical and laboratory variables accessible pre-harvest for use across the field of T-cell therapy. Pending further validation, such a model may be used to generate an individual leukapheresis plan and streamline the process of cell harvest, a well-recognized bottleneck in the industry.