Role of Recipient CD34 Specific Chimerism (CSC) in the Bone Marrow (BM) to Predict BM Relapse Post Allogeneic (Allo) Stem Cell Transplant (SCT)

Abstract

BM relapse is a major cause of treatment failure post Allo-SCT in hematological malignancies (HM) with prior BM involvement. Unfractionated whole bone marrow chimerism (WMC) and CSC are frequently used to monitor engraftment after Allo-SCT. Since several of these HM express CD 34 antigen, one can hypothesize that persistence of recipient CSC in the BM post Allo-SCT may be a tool to predict impending relapse. We performed a single institution retrospective analysis of all patients who underwent Allo-SCT for AML, ALL, MDS/MPN, and CLL between October 2009 and December 2013 on an IRB approved protocol. We included patients with at least two separately recorded WMC and CSC studies who survived at least 100 days after Allo-SCT. Engraftment analysis was performed on the whole BM and the CD34 subset using Promega PowerPlex 24 loci STR system. CD34 subset was obtained by positive selection from the BM using antibody coated magnetic particles from STEMCELL Technologies. >5% or ≥10% chimerism of recipient stem cells detected on either whole BM or CD34 subset were used for prediction of relapse. Sensitivity, specificity, false positive, false negative and receiver operator curve (ROC) by the 4 test criteria were computed using Bayes’ Theorem and compared for best utility. 76 patients with AML (n=45), ALL (n=14), MDS/MPN (n=14), and CLL (n=3) were identified in the data base meeting the above chimerism and survival criteria. Bone marrow aspirates were performed around scheduled day 30, day 100, 6 months, 1 year, or as clinically indicated post Allo-SCT and sent for WMC and CSC. 17 out of the 76 patients (22.4%) developed morphologic, cytogenetic, and/or molecular relapse at time of follow up by September 2015. 16 of the 17 relapses had CSC measurements whereas all 17 had WMC measurements. Of the 16 patients who relapsed with CSC measurements, 10 (62.5%) had relapse predicted by recipient CSC > 5%, and 6 (37.5%) had relapse predicted by CSC ≥ 10%. Median time ahead of relapse was 168 days (range 28-379 days) from first CSC > 5% and 99 days (range 28-257 days) from first CSC ≥10%. Only 2 (11.8%) of the 17 patients with WMC measurements had relapse predicted by WMC > 5%, and 0 were predicted using WMC ≥10%. CSC ≥10% showed the best sensitivity (82.4%), specificity (84.7%), false positive rate (39.2%), false negative rate (5.6%), and ROC curve of 0.836 (95% CI, 0.731 - 0.940). CSC > 5% had the best sensitivity at 88.2% but at the expense of a lower specificity (62.7%), higher false positive rate (59.6%), and lower ROC curve of 0.755 (95% CI 0.654-0.855). WMC > 5% and WMC ≥10% were poor with sensitivity both at 41.2% and ROC curves of 0.706 and 0.697 respectively. Recipient CSC ≥ 10% was the best test in correlating with relapse. However, recipient CSC > 5% was the most sensitive test for predicting relapse but had inferior specificity.