There is a substantial body of experience using the CliniMacs Plus immunomagnetic cell separator to produce modified grafts for hematopoietic progenitor transplantation. In general, graft preparation has been performed on starting products of varying cellularity, and the total recovered product from CliniMacs manipulation has been administered in its entirety. This, in turn, leaves cell dosing as a confounding variable when evaluating clinical outcomes resulting from treatment with CliniMacs processed grafts. We hypothesize that the relationship between starting numbers of targeted cell types and final numbers may be sufficiently linear that final cell numbers might be pre-determined by controlling the number of targeted cells in the pre-processing products. We have reviewed the data of pre-processing cell numbers and the resultant targeted cell recoveries from our CD34 selection and CD45RA depletion single step processes and from our NK cell selection double step process. Using a Pearson's linear regression model, we analyzed each identified cell type, comparing the number of those cells input into the CliniMacs process with the number of the same cell type recovered at the end of processing. From these analyses, we obtained for each cell type a regression equation, y=ax+b, wherein a is the slope of the regression line and also the average recovery of the targeted cell, and b is the y axis intercept of the best fit line. Additionally, we obtained an R2 value, which identifies the fractional certainty that the input cell value predicts the output value. This analysis has revealed that for the identified cell types, there is a linear relationship, suggesting a predictability of final product counts from starting product counts with 65–94 percent certainty (R2). The resultant predictive model incorporates all processing steps, including washes, antibody incubations and the actual CliniMacs separation processing. It should be noted in particular that a linear relationship is present even in our NK cell enrichment processes which comprise a CD3 depletion followed serially by a CD56 selection. We conclude that maintaining minimal variation in the conduct of processing steps, combined with a non-varying CliniMacs process, produces a sufficiently predictable final product to allow for increased precision in cell dosing. Table ICliniMacs plus processing linearity. Manipulation Cell Type Targeted N Regression Line Slope R-square CD34 Selection CD34 positive 309 0.60 0.94 CD45RA depletion CD34 positive 64 0.56 0.88 CD45RA depletion CD3 positive/CD45RA negative 64 0.39 0.65 CD3 depletion followed by CD56 selection CD56 positive 151 0.53 0.80 CD3 depletion followed by CD56 selection CD56 positive/CD3 negative 151 0.53 0.80 Open table in a new tab
Read full abstract