|Home | About | Journals | Submit | Contact Us | Français|
In patients with severe congenital neutropenia (SCN), long-term therapy with granulocyte colony-stimulating factor (G-CSF) has reduced mortality from sepsis, revealing an underlying predisposition to myelodysplastic syndrome and acute myeloid leukaemia (MDS/AML). We have reported the early pattern of evolution to MDS/AML, but the long-term risk remains uncertain. We updated a prospective study of 374 SCN patients on long-term G-CSF enrolled in the Severe Chronic Neutropenia International Registry (SCNIR) with longer follow-up. After 10 years on G-CSF, the annual risk of MDS/AML was 2.3%/year. After 15 years on G-CSF, the cumulative incidence was 10% for death from sepsis and 22% for MDS/AML. The data continue to support the hypothesis that SCN patients with high G-CSF requirements are also at high risk of MDS/AML. The risk per year of MDS/AML in SCN now appears similar to, rather than higher than, the reported risk of AML in Fanconi anemia and dyskeratosis congenita.
Severe congenital neutropenia (SCN) is a genetically heterogeneous disorder of myelopoeisis that is diagnosed clinically on the basis of absolute neutrophil counts (ANC) persistently below the threshold of 0.5 ×109/l, with maturation arrest of neutrophil precursors in the bone marrow (Ancliff 2003). Maintenance therapy with granulocyte colony-stimulating factor (G-CSF) is standard of care to prevent life-threatening bacterial sepsis(Dale, et al 1993). The success of G-CSF in averting mortality from sepsis (sepsis deaths) has unmasked a notable predisposition to myelodysplastic syndrome and acute myeloid leukaemia (MDS/AML)(Freedman, et al 2000). This susceptibility appears greatest in patients who require higher doses of G-CSF to achieve an adequate neutrophil response (Rosenberg, et al 2006).
We previously characterized the natural history of SCN patients maintained on G-CSF, using prospective observational study of 374 patients enrolled in the Severe Chronic Neutropenia International Registry (SCNIR) (Rosenberg, et al 2006). At the time of that report, long-term follow-up was limited, but available data suggested that the risk of MDS/AML in SCN rose to extraordinary levels, perhaps 8%/year after 12 years on G-CSF, albeit with a substantial margin for error. The apparent increase in the hazard curve for MDS/AML has been a cause for concern among patients and their physicians, and raised therapeutic and etiological questions (Dale and Link 2009). Hence, obtaining a more precise characterization of the risk profile is of considerable clinical and scientific importance. To this end, we updated prospective follow-up of the same cohort to obtain more accurate estimates.
We updated prospective follow-up of 374 well-characterized SCN patients on long-term G-CSF enrolled in the SCNIR. We ascertained event-free time, sepsis deaths, and MDS/AML events that accrued since our previous report, which ended 26 February 2001, through 10 July 2009. For purposes of this analysis, follow-up was censored at the time of bone marrow transplant. Data on baseline G-CSF dose at month 6 of treatment was abstracted from registry and medical records, and converted to units of micrograms per kilogram per day (μg/kg/d), as described previously. Data on mean ANC counts (cells ×109/l) during months 6 – 18 on treatment were also abstracted. The study was conducted in accordance with the Declaration of Helsinki under the auspices of the Human Subjects Committee of the University of Washington and other participating institutions. Patients provided informed consent.
Patients are at risk of both MDS/AML and sepsis death, which we analyzed as competing adverse events. The time scale was years on G-CSF therapy. We obtained flexible and smooth estimates of the cause-specific hazards of MDS/AML and sepsis death using spline functions(Rosenberg 1995). We estimated the cumulative incidence of each adverse event using the nonparametric maximum likelihood estimator (Gaynor, et al 1993). We estimated the effects of baseline G-CSF dose on the hazard of each adverse event using Cox proportional hazards models applied to follow-up from month 6 onwards. All statistical tests were two-sided. CI denotes confidence interval.
The update yielded a total of 3,590 person-years of follow-up, versus 2,043 in the prior report. Follow-up was censored in 19 patients who received a bone marrow transplant. There were 849 person-years among 176 patients treated for 10 or more years; versus just 67 person-years among 60 patients previously. In all, there were 61 MDS/AML events and 29 sepsis deaths, versus prior totals of 44 and 19, respectively. After including up-to-date follow-up, the estimated annual hazard of sepsis death remained qualitatively stable, at 0.81%/year (95% CI: 0.56 – 1.16%/year). Similarly, during the first five years after the start of G-CSF therapy, the updated estimate of the hazard curve for MDS/AML showed the same increasing trend as the previous estimate (Figure 1A). However, in contrast to the prior estimate that showed an increasing trend after year 5 (with a large margin for error), the updated hazard curve attained a plateau, with confidence intervals for the new hazard curve considerably narrower than corresponding intervals in our earlier report. After 10 years on G-CSF, the estimated hazard of MDS/AML was 2.3%/year (95% CI: 1.7 – 2.9%/year).
Whereas the long-term hazard of MDS/AML now appears significantly lower than first suggested, substantial numbers of sepsis deaths and cases of MDS/AML accumulated over time. After 15 years on G-CSF, the cumulative incidence was 10% (95% CI: 6 – 14%) for sepsis death and 22% (95% CI: 17 – 28%) for MDS/AML Figure 1B). In the initial dataset, the corresponding estimates of cumulative incidence attained similar values earlier, 8% and 21% at 10 years, respectively.
With additional follow-up, the association of G-CSF dose at 6 months with the relative hazard of MDS/AML became more strongly statistically significant (P = 0.003 versus P = 0.024; the hazard of MDS/AML increased by 1.24-fold [95% CI: 1.08-1.43-fold] per doubling of the G-CSF dose). In contrast, the association of G-CSF dose at 6 months with the relative hazard of sepsis death was slightly attenuated (P = 0.053 versus P = 0.039; the hazard of sepsis death increased by 1.25-fold [95% CI: 1.00 – 1.56-fold] per doubling).
As in the previous analysis, the subset of patients who failed to achieve a mean ANC count at or above the median for the cohort (2.188×109/l) despite doses of G-CSF at or above the median (8 μg/kg/day), were at elevated risk of both sepsis death and MDS/AML, compared with patients who achieved a good response at a lower dose (i.e., median ANC count above 2.188 ×109/l on G-CSF below 8 μg/kg/day). In the low-risk group, the cumulative incidence after 15 years on G-CSF was 5% (95% CI: 0 – 12%) for sepsis death and 15% (95% CI: 4 – 25%) for MDS/AML (Figure 2A), versus 18% (95% CI: 7 – 28%) and 34% (95% CI: 21 – 47%), respectively, in the high-risk group (Figure 2B).
This analysis incorporates extended follow-up of the largest existing cohort of patients with SCN, which has allowed us to estimate the long-term risks with greater precision. Overall, comparisons of the new results (2009) with the old (2001) are very consistent. However, there is one important exception: in all patients combined, the hazard of MDS/AML now appears to be around 2.3%/year after 10 years on G-CSF, substantially below the range of 4 – 12%/year suggested by prior unstable data.
This is good news for patients and their physicians. Also, the reduced hazard estimate for MDS/AML helps resolve an etiological conundrum. From a molecular perspective, it was not entirely clear why susceptibility to leukaemia appeared higher in SCN than in other high-risk inherited bone marrow failure syndromes, including the DNA repair disorder of Fanconi anemia (FA) (Rosenberg, et al 2003) or the telomere maintenance syndrome of dyskeratosis congenita (DC) (Alter, et al 2009). Indeed, the high incidence of leukaemic transformation in SCN has raised concerns that G-CSF may promote malignant clones (Donadieu, et al 2005). In light of these new data, it now appears that the rate of MDS/AML in SCN is qualitatively quite similar to the rate of AML in both FA and DC. Furthermore, a plateau is now seen in SCN, similar to that for FA.
The positive news in this report must be put in perspective. Although the hazard curve for MDS/AML in SCN now appears to plateau, the cumulative incidence still attains high levels, albeit more slowly. Hence, it is imperative that all patients continue to be closely monitored for leukemic transformation. Furthermore, the risks and benefits of early haematopoietic stem cell transplantation should be evaluated for the subset of patients who respond poorly to a high dose of G-CSF.
We thank the patients and doctors for participating in the registry, and the data collection centers in Seattle, WA, and Hannover, Germany, for maintaining the registry. This research was supported in part by the Intramural Research Program of the NIH, National Cancer Institute, Division of Cancer Epidemiology and Genetics; NIH Grant 2R24AI049393; a gift from the Amgen Foundation; and a grant for the Rare Diseases Program “German Network on Congenital Bone Marrow Failure Syndromes” by the German Ministry for Education and Research, BMBF Grant 01 GM0618.
Disclosures David C. Dale has research support from Amgen, and is a consultant and speaker for Amgen (Honorarium). Laurence A. Boxer owns Amgen stock.
Off-label Use: There is no information to disclose.