Determinants of fatal bleeding during induction therapy for acute promyelocytic leukemia in the ATRA era

Simon Mantha, Debra A. Goldman, Sean M. Devlin, Ju-Whei Lee, Diana Zannino, Marnie Collins, Dan Douer, Harry J. Iland, Mark R. Litzow, Eytan M. Stein, Frederick R. Appelbaum, Richard A. Larson, Richard Stone, Bayard L. Powell, Susan Geyer, Kristina Laumann, Jacob M. Rowe, Harry Erba, Steven Coutre, Megan Othus, Jae H. Park, Peter H. Wiernik and Martin S. Tallman

Published e-Letters

Compose eLetter

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Author Information
First or given name, e.g. 'Peter'.
Your last, or family, name, e.g. 'MacMoody'.
Your email address, e.g.
Your role and/or occupation, e.g. 'Orthopedic Surgeon'.
Your organization or institution (if applicable), e.g. 'Royal Free Hospital'.
Statement of Competing Interests
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.
Enter the characters shown in the image.

Vertical Tabs

Jump to comment:

  • RE: APL Letter to Editor
    • Sean Devlin, statistician Memorial Sloan Kettering Cancer Center
    • Other Contributors:
      • Debra Goldman, statistician
      • Simon Mantha, hematologist
      • Martin Tallman, hematologist

    Thank you for your letter and for raising this important issue. The number of events per variable is an important consideration in model building in the time-to-event setting, and we agree with your assessment of our final model and choice for building. The parsimonious multivariate model included in the manuscript was in part driven by the limited number of HD events in the patient population. In addition, multivariate model building was designed based on including factors that achieved a univariate threshold of α=0.05, which we defined a priori.
    We note that the analyses of FAB classification and to a lesser degree PT are limited in our study by the degree of missing values for these factors across the studies. For PT, we are happy to provide the multivariate model which includes this factor in addition to WBC count and ECOG performance status. As evident by the results in the table below, inclusion of PT does not change the final interpretation of our manuscript. We agree that additional larger studies are needed to further elucidate the determinants of HD.

    Sean Devlin, Debra A. Goldman, Simon Mantha and Martin Tallman


    Conflict of Interest:
    None declared.
  • RE: Determinants of fatal bleeding during induction therapy for acute promyelocytic leukemia in the ATRA era

    Mantha et al. conducted a 30-day prospective study to clarify the potential predictors of hemorrhagic death (HD) during induction therapy in 1009 patients with acute promyelocytic leukemia (APL) (1). There were 37 cases of HD, and the adjusted hazard ratios (95% confidence intervals) of increased white blood cell (WBC) count and decreased performance status for HD were 5.20 (2.70-10.02) and 2.17 (0.84-5.62), respectively. I have a query on their study with special reference to the statistical validity of their results.
    The author entered the variables identified by univariate analysis as significant into the multivariate Cox proportional hazard model. The authors considered multicollinearity and also included shared frailty as a random factor. According to their description, only the WBC count and performance status were used for their analysis, although the prothrombin time and FAB classification also approached the significance level. The number of HD cases was limited, and I suppose that the authors could not use a sufficient number of independent variables for obtaining a stable estimate (2, 3). The authors compiled five trials for the analysis, and they considered adjustments by including shared frailty as a random factor. On the basis of the results of the sensitivity analysis, further study is needed to confirm the determinants of HD among patients receiving induction therapy for APL.


    1. Mantha S, Goldman DA, Devlin SM, et al. Determin...

    Show More
    Conflict of Interest:
    None declared.