Correspondence

Recombinant Granulocyte Colony-Stimulating Factor in Acute Myelogenous Leukemia

N Engl J Med 1995; 333:1155-1156October 26, 1995

Article

To the Editor:

Dombret et al. (June 22 issue)1 report a striking difference in the rates of complete response between patients treated with granulocyte colony-stimulating factor (G-CSF) after chemotherapy for acute myeloid leukemia (AML) and patients not given G-CSF. Although treatment with G-CSF shortened the duration of neutropenia, the mortality rate was not reduced during induction chemotherapy. Since the chemotherapy doses were equivalent, a direct antileukemic effect of G-CSF was postulated. This higher rate of complete response was translated into neither a longer period of complete response nor better event-free or overall survival. An unexpected and rather extraordinary antileukemic effect cannot be ruled out, and if such is the case, it offers a new avenue for the treatment of AML.

Given their use of an arbitrary cutoff of 5 percent blasts in the definition of a complete response and that its significance in terms of hematopoietic growth factors remains to be established, the authors may have misinterpreted partial responses, which can be associated with peripheral neutrophil and platelet recovery, as complete responses.2 Complete responses were evaluated sooner in the G-CSF group (24 days) than in the placebo group (33 days), and this difference in the timing of the response assessment may have introduced a bias.

We have reviewed the charts of the few patients treated in our department with salvage therapy followed by G-CSF. In the four patients from whom bone marrow aspirates were obtained five to seven days after the interruption of G-CSF, hypergranulopoiesis was evidenced, with less than 5 percent blasts. One to two weeks later, however, in the absence of stimulation of the normal myeloid compartment, partial responses were diagnosed. Thus, we wonder whether the expansion of the normal myeloid lineage induced by G-CSF in the study by Dombret et al. may have altered blast counts at the time of the evaluation of response, artificially reducing the percentage of blasts, which otherwise would have been in excess of 5 percent. How long after the cessation of G-CSF was the bone marrow assessment performed? Before consolidation chemotherapy, was the bone marrow reevaluated after a sufficient interval from the first evaluation?

Jean-Nicolas Munck, M.D.
Didier Deacaudin, M.D.
Serge Koscielny, Ph.D.
Institut Gustave-Roussy, 94805 Villejuif, France

2 References

1. 1

   Dombret H, Chastang C, Fenaux P, et al. A controlled study of recombinant human granulocyte colony-stimulating factor in elderly patients after treatment for acute myelogenous leukemia. N Engl J Med 1995;332:1678-1683
   Full Text | Web of Science | Medline

2. 2

   Cheson BD, Cassileth PA, Head DR, et al. Report of the National Cancer Institute-sponsored workshop on definitions of diagnosis and response in acute myeloid leukemia. J Clin Oncol 1990;8:813-819
   Web of Science | Medline

Author/Editor Response

The authors reply:

To the Editor: Since the primary objective of our trial was to evaluate the effect of lenograstim (glycosylated recombinant human G-CSF) on mortality at eight weeks, we used only the usual criteria to assess the response to induction therapy. Bone marrow was aspirated just after hematopoietic recovery, and we only reassessed peripheral-blood samples before the first course of consolidation therapy. As noted by Munck et al., the standard criteria (especially the time requirement) used to define complete and partial responses might have to be redefined in patients treated with a granulocytic growth factor. Nevertheless, we believe that the difference in the rate of complete response observed between the two treatment groups in this double-blind trial (with the criteria that we used) cannot be explained only by the G-CSF–induced stimulation of normal granulopoiesis.

First, all but two patients classified as having a complete response continued to meet the criteria for remission, according to peripheral-blood analysis, just before the initiation of the first consolidation course. This peripheral-blood analysis was performed a median of 18 days after the discontinuation of the study medication (range, 9 to 43 days in the G-CSF group and 8 to 36 days in the placebo group; P = 0.62 by the Kruskal–Wallis test). In patients with a complete response who had received G-CSF, the median platelet count was 273,000 per cubic millimeter (range, 118,000 to 957,000 per cubic millimeter), whereas it was 245,000 per cubic millimeter (range, 126,000 to 544,000 per cubic millimeter) in patients with a complete response who had received placebo (P = 0.47 by the Kruskal–Wallis test). A normal platelet count at that time is not usual in patients with primary AML who do not have a complete response. At the same time, the median polymorphonuclear-cell count was 4700 per cubic millimeter (range, 1100 to 20,300 per cubic millimeter) in the G-CSF group and 5200 per cubic millimeter (range, 2100 to 15,400 per cubic millimeter) in the placebo group (P = 0.65 by the Kruskal–Wallis test).

Second, even if the increase in the rate of complete response did not translate into a longer event-free survival in the G-CSF group when we used the log-rank test, two observations argue against our misinterpreting complete responses in the G-CSF group: the duration of complete remission was similar in the two treatment groups, and short-term event-free survival was better in the G-CSF group than in the placebo group, with median event-free survivals of 168 and 78 days, respectively.

The quality of complete response in terms of minimal residual disease could be lower in patients treated with G-CSF, but the poor outcome may also result from inadequate post-remission therapy. We think that a more effective consolidation or maintenance treatment might prolong event-free survival.

Hervé Dombret, M.D.
Claude Chastang, M.D., Ph.D.
Hôpital Saint-Louis, 75010 Paris, France

for the AML Cooperative Study Group