Incidence and Prognostic Factors for Infectious and Hemorrhagic Death in Patients with Acute Myeloid Leukemia: A Single-Centre Study.

INTRODUCTION: Death is as a common cause of remission induction failure in patients with acute myeloid leukemia (AML), mainly due to hemorrhage and infection. The relative incidence and chronology of each of these categories of induction failure, as well as their prognostic factors, have been investigated critically and in detail in rare studies only.

OBJECTIVES: We report the incidence, chronology, and prognostic factors for induction death, analyzing separately hemorrhagic and infectious death, in a large series of 946 patients with AML who received induction therapy in a single institution over the last 30 years.

PATIENTS AND METHODS: Adult patients were consecutively diagnosed of AML and started first induction chemotherapy in our institution. AML was classified according to the FAB criteria. Induction therapy consisted of the classic combination of cytarabine and anthracyclines (with or without a third agent) in 50% of patients, cytarabine plus adriamicine and thioguanine or vincristine in 17%, ATRA with chemotherapy in 9%, monochemotherapy with anthracycline in 7%, high dose cytarabine in 7%, and other regimens in 10%. Causes of induction death include the following categories:

1. Infection, when death was due to a clinical, radiological or microbiologically documented infection,

2. Hemorrhage, when a major bleeding occured in a vital organ (central nervous system, lungs). Gastrointestinal hemorrhage required massive melena or hematemesis accompanied by fall in blood pressure, and

3. Other, i.e., any other cause not classified as infection or hemorrhage.

RESULTS: From 1977 to 2007, 946 consecutive patients with diagnosis of AML received induction chemotherapy, 24% in the period 1 (1977–1986), 28% in the period 2 (1987–1993), 28% in the period 3 (1994–2000), and 20% in the period 4 (2001–2007). Median age was 55 years (range 13–83 years). One hundred and sixty-seven patients (18%) had antecedents of myelodysplastic/myeloprolipherative disease (10%) or other neoplasia (8%). Two hundred and thirty-seven patients (25%) died during induction therapy, 13% due to infection, 7% due to hemorrhage, 2% due to hemorrhage and infection, and 3% due to other causes. The induction mortality rates decreased gradually over the 4 periods (31% vs 24% vs 18% vs 18%), due to reduction of both hemorrhagic and non-hemorrhagic deaths. Overall, 42% of hemorrhagic deaths occurred within the first 10 days of induction therapy, whereas 86% of infectious deaths occurred after 10 days. In multivariate analysis, the following characteristics had an unfavorable impact on overall induction mortality: age >60 years, WBC >50x10⁹/L, Quick index <65%, ECOG >1, and albumin serum levels <3.5mg/dL. Multivariate analysis identified the following factors predicting for infectious mortality: albumin <3.5mg/dL, age >50 years, AML secondary to neoplasia, ECOG >1, and fever at presentation. The following factors were associated with hemorrhagic mortality: WBC >50x10⁹/L, FAB-M3, age >60 years, de novo AML, and ECOG >1.

CONCLUSIONS: The main causes of induction death in AML patients, infection and hemorrhage, shows a different chronologic pattern and can be separately predicted by their own specific prognostic factors.