Differential dose‐related haematological effects of GM‐CSF in pancytopenia: evidence supporting the advantage of low‐ over high‐dose administration in selected patients

Abstract

Granulocyte-macrophage colony-stimulating factor (GM-CSF) is a multifunctional haematopoietin which can promote production of several blood cell lineages, though the predominant target cells are neutrophils, monocytes, and their precursors. Occasional undesirable clinical effects include eosinophilia, an increase in blasts, or thrombocytopenia. Here, we describe four patients who were treated with GM-CSF, at subcutaneous doses significantly lower than are conventional, and experienced an unusual response pattern. Three patients had severe pancytopenia associated with chronic lymphocytic leukaemia (CLL) or myelodysplastic syndrome (MDS) and exhibited an unexpected switch in the responsive lineage on high- versus very low-dose therapy. The two CLL patients developed marked eosinophilia (up to 10.0 x 10(9) cells/l) without an increase in neutrophils on 125-300 micrograms/m2/d of GM-CSF. In contrast, when the dose was lowered to 10 micrograms/m2/d, the neutrophils rose to physiological levels, without significant eosinophilia. The MDS patient showed a rapid rise in peripheral blasts (baseline level = 0; post-therapy level = 5.0 x 10(9)/l), without a change in other cell types, when receiving 60 micrograms/m2/d of GM-CSF. After GM-CSF was held, blasts returned to baseline levels; reinstituting therapy at the very low dose of 6 micrograms/m2/d was followed by an increase in platelet counts from 50 to 185 x 10(9)/l with only a minor increase in blasts. The fourth patient, who suffered from severe aplastic anaemia complicated by recurrent gastrointestinal haemorrhage, was only treated with the low-dose regimen. He showed a predominant platelet effect with counts rising from 9 to 169 x 10(9)/l. Very low-dose GM-CSF therapy was devoid of constitutional side effects. The biological implications of these GM-CSF responses are discussed. Our results indicate that, in some patients, GM-CSF may stimulate different target cells depending on the dose. Therefore, in contrast to the results of administration of many classical drugs, there may not always be a direct relationship between the amount of GM-CSF given and the optimal effect.