Multivariate Analysis of Patient and Graft Specific Factors among 330 Recipients of Unrelated Cord Blood Transplant (UCBT) To Predict Risk of Death from Opportunistic Infections in the First 6 Months after UCBT.


Conference Paper

Abstract Unrelated umbilical cord blood transplantation (UCBT) is a suitable option for those who lack HLA-matched sibling donors. However, opportunistic infections (OI), with the majority detected in the first 100 days, remain the major cause of of non-relapse mortality with almost all death ocurring in the first 6 months. Over the past 5 years we have studied the reconstitution of immunity in the immediate post-UCBT period in >150 pediatric UCBT recipients to identify those at risk for OI. Several graft and patient-specific variables were identified as significant confounders when the laboratory measurements of dendritic and T cell reconstitution was analyzed. Here, we set out to analyze the impact of these factors on the 6 months post-UCBT OI-related mortality by analyzing all consecutive patients transplanted during the same time period. Between June 1999 and Oct. 2005, 330 pediatric recipients of single UCB grafts were identified. Those receiving a second transplant for primary graft failure were not analyzed. 220 of the 330 patients were alive at 6 months (67%). Of the 110 children who died by 6 months, 46 patients were identified where OI (viral, fungal, protozoal infections) could not be implicated as a cause of death. After removing these from the analysis, 284 patients remained; 200 (77%) were alive at 6 onths and 64 (23%) died at or before 6 months related to OI. Using these 284 patients, nine predictors were tested in logistic regression models to test for risk to die due to OI by 6 months. First, six variables describing the demographic and clinical characteristics of the patients were entered into the model. These variables were gender, race (white/non-white), age at UCBT, CMV serology (pos/neg), HLA mismatches (0,1,2,3 antigen mismatch), and a 3-level categorical variable indicating whether indication for UCBT was non-malignant condition treated with chemotherapy alone (n=121, 42%), malignancy treated with regimens including TBI (n=116, 41%) or malignancy treated with no TBI (n=47, 17%). Gender and race did not predict death (p-values=0.45 and 0.99, respectively). A greater probability of death was found for patients who were CMV positive (p=0.0002), had a greater number of HLA mismatches (p=0.02), and were older (0.002). Patients with malignancy treated without TBI had the greatest probability of death (0.39), followed by those with no malignancy (0.25); patients with malignancy treated with TBI had the smallest probability of death related to OI (0.17) (p= 0.006). Since the three dosing variables, CD34+, CD3+, and total cell dose/kg were highly correlated with each other, they were entered separately into this model (resulting in p-values = 0.05, 0.23, 0.02, respectively). When all three dosing variables were entered simultaneously into the model the resulting p-values were 0.36, 0.27, and 0.07, respectively, demonstrating the superiority of total cell dose/kg to the CD34+ dose of the graft. This analysis demonstrates that patient and graft-specific confounders will have significant impact on mortality due to OI. Opportunistic infections (most being viral) is the most prevalent cause of death in the first 6 months after UCBT. This data can serve to develop risk models to stratify patients into risk groups and determine who may benefit from novel therapeutic interventions.

Full Text

Duke Authors

Cited Authors

  • Szabolcs, P; Peterson, BL; Niedzwiecki, D; Chao, N; Kurtzberg, J

Published Date

  • November 16, 2006

Published In

Volume / Issue

  • 108 / 11

Start / End Page

  • 2860 - 2860

Published By

Electronic International Standard Serial Number (EISSN)

  • 1528-0020

International Standard Serial Number (ISSN)

  • 0006-4971

Digital Object Identifier (DOI)

  • 10.1182/blood.v108.11.2860.2860