Transfusions and risk of failure in childhood acute lymphoblastic leukemia

Andrew S. Freiberg, Michael L. Hancock, Keith D. Kunkel, Gaston K. Rivera, William M. Crist

Research output: Contribution to journalArticlepeer-review

8 Scopus citations


To examine the relationship of blood product support to treatment outcome in childhood ALL, we reviewed records of all 358 patients with newly diagnosed ALL treated on St Jude Total Therapy Study XI (February 1984 to September 1988). All but six patients received blood products (median 7 units, range 0-246), with approximately 90% given during the 6-week induction period. Because all 16 patients who received ≥ 50 units failed, the number of units transfused was predictive of treatment failure in multivariate analysis (relative risk = 1.8, p = 0.02), although the number of units transfused was also associated with initial leukocyte count and age. Among the remaining 342 patients who received < 50 units, the number of units transfused was associated with reduced event-free survival in univariate analysis only, with maximal significance at > 7 units (p = 0.006). Because exclusion of the 16 patients who received the most blood eliminates the independent effect of transfusions on patient outcome, we believe that the number of transfusions is largely an epiphenomenon which reflects the effects of two risk factors not included in traditional outcome analysis in childhood ALL. These are acuity of illness during induction, and reduction of chemotherapy doses during induction therapy, due to the severity of illness. Immunomodulation caused by exposure to blood products appears unlikely to contribute strongly to outcome in childhood ALL.

Original languageEnglish (US)
Pages (from-to)1220-1223
Number of pages4
Issue number7
StatePublished - Jul 1994

All Science Journal Classification (ASJC) codes

  • Hematology
  • Oncology
  • Cancer Research


Dive into the research topics of 'Transfusions and risk of failure in childhood acute lymphoblastic leukemia'. Together they form a unique fingerprint.

Cite this