Stochastic model based on preharvest peripheral CD34‐positive cell count and collection efficiency predicting processed blood volume in peripheral hematopoietic stem cell apheresis. Issue 2 (28th December 2018)

• Record Type: Journal Article
• Authors: Ono, Yuichiro; Yoshioka, Satoshi; Inoue, Kazuhisa; Yoshida, Masahiro; Maruoka, Hayato; Ishikawa, Takayuki
• Is Part Of: Transfusion. Volume 59:Issue 2(2019)
• Page Start: 671
• Page End: 680
• Publication Date: 2018-12-28
• Subjects: Hematology -- Periodicals; Blood -- Transfusion -- Periodicals; Blood Group Antigens -- Periodicals; Blood Preservation -- Periodicals; Blood Transfusion -- Periodicals; 615
• DOI: 10.1111/trf.15096
• Languages: English
• ISSNs: 0041-1132

Abstract

Abstract : BACKGROUND: Processed blood volume (PBV) required to obtain a predefined number of stem cells can be estimated from peripheral blood CD34+ cell concentration, body weight, and collection efficiency (CE). Because CE is indefinite, this study was designed to formulate and validate a new model of PBV based on stochastic CE distribution. STUDY DESIGN AND METHODS: Data were retrospectively collected on 146 peripheral blood stem cell harvests from 114 patients and donors in a single institution from April 2014 to February 2018. The training set consisted of all procedures performed from April 2014 to June 2016 and the validation set of all procedures performed from July 2016 to February 2018. A new algorithm, based on CE2 distribution of the training set, was affirmed using the validation set. The positive predictive value of the model was estimated from the expected percentage of procedures that reached the target CD34‐positive dose, with predicted PBV processed as the gold standard. RESULTS: The 10th and 50th percentiles of CE2 were 33.4% and 52.5%, respectively. When PBV was assorted into three categories, defined as greater than 90%, 50% to 90%, and less than 50% of procedures reaching the targeted CD34‐positive dose, the positive predictive values of the new model were 100%, 70%, and 100%, respectively. CONCLUSION: The new model was validated with a high positive predictive value and can reliably estimate the required PBV and the success rate corresponding to the PBV. The model can be utilized easily with a Web‐based tool.