Granulocyte-stimulating factor and severe aplastic anemia: a survey by the European Group for Blood and Marrow Transplantation (EBMT)

Use of antithymocyte globulin (ATG) and cyclosporine (CsA) has led to a dramatic improvement in the survival of patients with severe aplastic anemia (SAA).¹  Combination of both drugs led to survival rates reaching 80%, in the long term.^1-3  More recently, granulocyte colony-stimulating factor (G-CSF) has been introduced as supportive therapy for patients with SAA.⁴  G-CSF should not be viewed as sole therapy for SAA.⁵  A small randomized trial by the European Group for Blood and Marrow Transplantation (EBMT), however, failed to demonstrate a survival advantage from adding G-CSF to ATG and CsA.⁶ 

Myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML) following immunosuppressive therapy (IST) for SAA have long been described. An EBMT survey in 1993 reported a 10-year incidence of 10% for MDS and 5% for AML.⁷  Of note, most of these patients had been treated with ATG alone without CsA. Since then, a few studies reported AML/MDS incidences ranging from 10% to 15% after combined treatment with ATG and CsA (for a review, see Socie et al⁸ ). However, in the late 1990s alarming data were published by Japanese groups reporting incidences as high as 45% in children who were treated with IST in combination with G-CSF.⁹  Although a retrospective Italian study failed to confirm such worrying figures,¹⁰  the question remains unsolved as to the potential role of G-CSF in the genesis of such complications.

The EBMT-SAA working party thus decided to reassess the incidence of MDS/AML following IST for SAA in the modern era of combined ATG and CsA therapy and to examine the role, if any, of G-CSF as part of the primary treatment in both malignant complications and its effect on overall survival.

The EBMT database was used to identify 881 patients who had received a first-line combined treatment with ATG and CsA with or without added G-CSF as first line. A questionnaire was sent to each participating center to confirm primary treatment; severity of the disease¹¹ ; the occurrence of MDS, AML, or solid tumor (clonal chromosomal abnormalities and date if any); relapse of SAA (and date if any); transplantation (and date if any); and use of G-CSF (dose and duration). The G-CSF dose used was 5 μg/kg/d given for less than 6 months. Forty-one patients were excluded from the analysis (32 because of lack of date of treatment, 1 patient with a transplant without date of transplantation available, 6 patients lost to follow-up, and 2 patients with a diagnosis of malignancy before treatment for SAA). Thus, 840 patients are included in this analysis.

The diagnosis of MDS, AML, and solid tumor was abstracted from each patient's chart. No central review of pathologic material was performed. The cumulative incidence of any clonal event was calculated with death (n = 169) or transplantation (n = 176) being considered as competing events. Risk factors for secondary malignancies and those affecting survival were analyzed by multivariate analysis using the Cox model, with time-dependent covariates for events (malignancies and relapse) occurring during follow-up.

This EBMT survey included 840 patients with SAA treated (as first-line therapy) with ATG plus CsA with (n = 363, 43%) or without (n = 477, 57%) G-CSF. Only a few patients (6%) received G-CSF before 1990, whereas 47% and 53%, respectively, received G-CSF in the time periods 1990-1994 and 1995-2002, respectively (P < .001). Median age at diagnosis was 26.8 years (interquartile range, 15.4-46.6 years); 470 were male, and the interval between diagnosis and treatment was 0.8 months. Most patients had idiopathic AA (n = 607, 72%). Younger patients were more likely to be treated by G-CSF (62% versus 37%; P < .001). IST plus G-CSF was delivered with a shorter time from diagnosis to treatment as compared to IST alone (86% versus 69% [P < .001] before or after 2 months, respectively). Otherwise no significant difference emerged between the 2 patients groups. The 10-year survival rate was 72.2%, as expected (2.1), with a median follow-up of almost 4.7 (0.3) years. Twenty-four, 26, and 10 cases of MDS, AML, and solid tumors were reported, at a median elapsed time (interquartile range) of 2.5 years (2.0-5.6 years), 2.2 years (0.9-5.4 years), and 7.1 years (1.2-11.0 years), respectively. Thus compared to our previous survey and results from other groups it appears that MDS and AML occurred earlier than previously described.^7,8  This may be linked to a more systematic search for secondary malignancies or to an earlier onset in the era of combined IST. The 10-year cumulative incidence of MDS, AML, and solid tumor was 4.3%, 4.6%, and 0.7%, respectively. These data do not reproduce the alarming figures from the Japanese group^9,12,13  but fit with those from the Italian study.¹⁰  The incidence of MDS and AML in patients who did or did not receive G-CSF is illustrated in Figure 1 (combining both MDS and AML 10-year cumulative incidences were 5.8% and 10.9% [P = .07] in no G-CSF and G-CSF groups, respectively). In multivariate risk factor analysis only older age was associated with the occurrence of MDS (age ≥ 45 years; hazard ratio [HR] = 2.9; 95% CI, 1.3-6.6; P = .01), and factors associated with increased risk of leukemia included older age (age ≥ 45 years; HR = 4.1; 95% CI, 1.8-9.6; P = .002) and use of G-CSF (HR = 2.5; 95% CI, 1.1-6.0; P = .03). Combining MDS and AML we found both age older than 45 (HR = 2.9 [1.6-5.5], P = .001) and G-CSF (HR = 1.9 [1.0-3.5] P = .04) to be significant. Thus, although this is a retrospective analysis, it appears that G-CSF is associated with a small but significantly increased risk of MDS/AML after combined IST, as recently reported in patients treated with G-CSF for severe congenital neutropenia.¹⁴ 

The impact of G-CSF on overall survival was then studied. By univariate Cox analysis G-CSF was associated with an improved survival; patients who did not receive G-CSF had a worse outcome (HR = 1.4, P = .02). Other significant risk factors for worse survival in univariate analysis were development of MDS, AML, solid tumors (HR = 11.7, 21.5, 12.5, all with P < .001, respectively), relapse (HR = 1.7, P = .03), year of treatment before 1990 (HR = 1.54, P = .02), and age older than 45 years (HR = 4.2, P < .001). By multivariate analysis, G-CSF was marginally significant (HR = 0.7, P = .03) if age was omitted from the model, but was no longer significant if it was included (P = .3). Factors affecting survival are summarized in Table 1. Finally, 2 Cox models were run separately in patients who received or did not receive G-CSF as part of their first-line treatment (Table 2). From these analyses, relapse was no longer associated with worse outcome in patients who did not receive G-CSF, whereas relapse was significantly associated with worse outcome in those who received G-CSF. This strongly suggests an interaction between G-CSF and relapse (with worse outcome of patients who relapse if they had been previously exposed to G-CSF).

As expected from this and previous studies,⁸  the occurrence of secondary malignancy strongly affected the overall survival. However, survival data were more conflicting with regard to relapse. In keeping with results from other studies¹⁵  relapse did not translate into a worse outcome in patients who did not receive G-CSF, but it did affect survival in patients who received G-CSF. In patients who have been previously exposed to G-CSF, relapse may theoretically be associated with a worse outcome due to hematopoietic stem cell exhaustion. Finally, as previously reported older age was associated with a worse outcome.^2,16,17 

Although previous phase 2 trials suggested that the addition of G-CSF to ATG and CsA contributed to an improved survival rate in the 90% range,^1,18  we have failed to show evidence for such an improvement. Furthermore, a previous small randomized trial of the EBMT also failed to demonstrate a benefit for G-CSF in this situation. However, our data also show that nearly half of European patients received G-CSF as part of their first-line therapy, as most probably do patients in the United States.

The results of the present survey suggest an increased incidence of MDS/AML and that G-CSF is a potent risk factor, but it has had no clear role in improvement in survival. Although based on the largest survey ever performed, the retrospective nature of this analysis precludes any definitive statements about the role, if any, of G-CSF in the treatment of SAA or in the genesis of secondary malignancy. This reinforces our view for the need for a larger randomized trial comparing ATG and CsA with or without G-CSF, which the EBMT is currently running, and to alert physicians that adding G-CSF to combined IST is currently not standard treatment for SAA.

Contribution: G.S. designed the study, collected the data, and wrote the paper; J.Y.M. set the database, performed the statistical analysis, and wrote the paper; H.S., J.C.W.M., A.B., A.L., M.F., A.B., A.T., and J.P. are active members of the Aplastic Anemia working party of the EBMT (Chairman, J. Passweg); all participated actively in the design of the study, collected data, and participated in the writing of the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

A complete list of the members of the European Group for Blood and Marrow Transplantation is provided in Document S1, available on the Blood website; see the Supplemental Document link at the top of the online article.

Correspondence: Gérard Socié, Service d'Hématologie Greffe and U728 INSERM, Université Paris VII, Hôpital Saint Louis, 1 Ave Claude Vellefaux, 75475, Paris Cedex 10, France; e-mail: gerard.socie@paris7.jussieu.fr.