Improved Prediction of CD34+ Cell Yield before Peripheral Blood Hematopoietic Progenitor Cell Collection Using a...

The most commonly used stem cell source for both autologous and allogeneic transplantation is mobilized peripheral blood hematopoietic progenitor cells collected by apheresis. In the 1990s, an Italian group used the correlation between the preapheresis peripheral blood CD34+ cell count and the final number of CD34+ cells collected to devise a formula for "target value-tailored" (TVT) apheresis. Using local patient data, the Canadian Blood Services Stem Cell Laboratory created a similar model to determine the blood volume to process during apheresis collection. The objectives of this study were to determine the correlation between the number of CD34+ cells predicted by the TVT formula and the actual number of CD34+ cells collected and to determine whether the TVT formula remains predictive when applied to an external data set. All apheresis collections performed at the Ottawa Hospital between January 1, 2003 and October 2, 2013 were reviewed. The primary outcome was the correlation between the number of CD34+ cells predicted by the TVT formula and the actual number of CD34+ cells collected on day 1 of apheresis. For the external data set, all autologous collections performed at the London Health Sciences Centre between December 1, 2008 and December 1, 2013 were reviewed. The external data set was divided into test and validation sets to determine whether a model could be created to predict the final number of CD34+ cells collected on day 1 based on the preapheresis CD34+ count. A total of 1252 collections were included in the analysis. The Ottawa data set included 1012 collections, 836 of which were autologous and 176 of which were from donors. Of the autologous collections in Ottawa, 764 (92.5%) were first collections. In 759 (91%) collections, chemotherapy plus granulocyte colony-stimulating factor (G-CSF) was used as the mobilization regimen. In 747 collections (89%), only 1 collection day was required to achieve the desired number of CD34+ cells. The TVT estimate was highly predictive of the number of CD34+ cells × 10(6)/kg actually collected on apheresis day 1 (r = .90, P < .0001). The London data set included 240 autologous collections. All mobilizations were with G-CSF alone. For the test set, the precollection CD34+ count was highly predictive of the number of CD34+ cells × 10(6)/kg collected on day 1 of apheresis. Applying this model to the validation set, the correlation between the predicted and final and day 1 CD34+ cells × 10(6)/kg count was .9186 (P < .0001). Using a modified TVT approach, the preapheresis CD34+ count can be used to accurately predict the number of CD34+ cells × 10(6)/kg collected on day 1. This approach can be applied at other centers and for different diseases and mobilization regimens. This method can be used to individualize the blood volume processed and, thus, optimize resource utilization.