In 2007 the WMDA responded to the publication of two manuscripts suggesting a causal link between G-CSF and myeloid malignancies in healthy donors by convening an international symposium to examine this issue. At the time, registries reviewed the long-term follow-up of their healthy donors, which suggested no excess of leukaemia in PBSC donors compared with BM donors. Although the evidence for an increased risk of malignancy in healthy donors was felt to be weak, it could not be excluded. The WMDA, therefore, issued a statement, to be included in all donor consent forms, stating that it was unknown whether G-CSF increased or decreased the risk of later developing cancer. In 2012, with 5 years of additional donor follow-up and the results of several genetic studies now available, the clinical working group of the WMDA again reviewed the data. On the basis of an assessment of a continuing lack of evidence for an increased risk of malignancy in donors receiving G-CSF, the WMDA has re-issued a more reassuring statement. The revised statement was circulated to all WMDA member registries in late 2012 to replace the existing statement in consent forms, which now conclusively states that, ‘Studies following large numbers of unrelated donors have shown that the risk of developing cancer within several years after the use of G-CSF is not increased compared with donors not receiving G-CSF’. Herein we review the evidence on which this statement is based.
PBSC collected from donors following stimulation by G-CSF injections are currently the most common form of hematopoietic stem cells used for both related and unrelated donor transplantation. In 2012, the EBMT reported that 72% of all allogeneic transplants were performed using PBSC.1 As the use of G-CSF was introduced in this setting, thousands of healthy individuals have received G-CSF for this indication. In several countries G-CSF is licensed for this indication, while in others its use in healthy donors is given as part of an ongoing research study. Two forms of G-CSF have been commonly used in healthy donors: Lenograstim (Granoctye-Chugai Pharma) and Filgrastim (Neupogen—Amgen). More recently, several biosimilar forms of Filgrastim have become available in the market, although to date their use in healthy donors has been limited. This review does not intend to draw any conclusions about the use of biosimilar G-CSF agents or other mobilizing agents such as plerixafor, as the published data in this review only include donors receiving the original filgrastim and lenograstim products. The use of these agents should currently be in the context of clinical studies.
In unrelated donors the use of G-CSF is generally standardized and donors are treated with 10 μg/kg/day for a maximum of 5 days. In some registries the dose is reduced, but it is seldom exceeded (except for rounding up dosages). Conversely, related donors often receive a higher dose, up to 20 μg/kg/day, and they may have a longer course of injections or an increased number of days of apheresis (classically a maximum of two apheresis sessions for unrelated donors). In vivo studies have shown that there is a significant increase in the levels of G-CSF in the peripheral blood following a single dose of G-CSF. This peaks at 4 h post injection (>25 000 pg/mL), and at 2 days post injection the levels have returned to their pre-injection baseline (<78 pg/mL).2 This increase greatly exceeds that which is found in individuals with a respiratory or bacterial infection (700–3000 pg/mL depending on the infectious agent), but tends to be more short lived.3
From the early years of G-CSF use many donor registries and transplant centres have had long-term follow-up protocols in place that included repeated blood tests for G-CSF-mobilised donors. For some, this was as part of their routine donor follow-up requirements. For example, the National Marrow Donor Program (NMDP) has given all donors G-CSF under an Investigational New Drug (IND) application (Food and Drug Administration approved) since 1997. The donor informed consent requires an agreement to long-term follow up.
In the mid-2000s, two publications raised concerns about the use of G-CSF in healthy donors. Bennett et al.4 reported on several cases of related donors who had developed haematological malignancies after exposure to growth factors. Two of the donors who developed AML had received G-CSF to mobilise stem cells for donation to a sibling with AML. Nagler et al.5 published a study that reported lymphocytes of G-CSF mobilized donors showed genetic and epigenetic abnormalities similar to those found in the lymphocytes of cancer patients. The abnormalities in replication timing were transient and returned to the same level as that seen in controls by a few months after G-CSF exposure; however, the aneuploidy in lymphocytes persisted (the last time points tested were at ~6 months post G-CSF exposure). Although several methodological issues about this analysis were raised,3, 6 the study nevertheless provoked concerns. In addition, it had been reported by some, but not all,7, 8 studies in patients with haematological disorders (such as aplastic anaemia or congenital neutropaenia) who receive recurrent doses of G-CSF, that there was an increase in myeloid leukaemia and MDS, in particular those associated with monosomy 7.9, 10, 11
As a result of these concerns the WMDA convened a special symposium at the European Group for Blood and Marrow Transplantation (EBMT) meeting in Lyon in 2007. International experts reviewed the available literature, as well as published and unpublished experience from several donor registries, reporting on long-term outcomes in donors who had received G-CSF. None of these reports found an increase in myeloid malignancies in healthy donors (reviewed in detail below and in Table 4). Several important editorial and other comments from leading experts in the field were also reviewed. These urged caution in over interpreting the data published by Bennett et al., in particular as the sample size was small, various mobilising agents were used (including some that had been withdrawn) and that the cases were all sibling donors (who may have a higher likelihood of developing AML due to familial clustering or predisposition to this disease).12, 13 The experts at this meeting did not believe that the evidence for a potentially causative role in the development of haematological malignancies was sufficient to halt the use of G-CSF in normal donors. Several recommendations were put forward, however, including a statement that the WMDA requested to be included in all unrelated donor consent forms.
‘Normal individuals are at risk for developing cancer, including leukemia, lymphoma or other blood diseases throughout their life time. G-CSF stimulates normal blood cell growth. In some patients with cancer or abnormal blood cells, it has been shown to stimulate leukemic blood cells. It is unknown whether G-CSF increases or decreases an individual’s risk of developing cancer. On the basis of available data from healthy people who have received G-CSF, no long-term risks have been found so far. The data being collected during follow-up will help establish whether there are any positive or negative long-term effects from receiving G-CSF. There was also a requirement to ask all donors whether there was a history of leukaemia in their family and a request for long-term follow-up and reporting of new onset haematological malignancies at any time following G-CSF exposure.
The following section will review the current evidence that has been published or presented to support a revision of the statement issued in 2007. The short-term and self-limiting side effects of G-CSF have been well described elsewhere,14, 15, 16, 17, 18 and it is not the intention of this review to repeat that here. The biology of G-CSF in relation to potential malignant transformations has been recently reviewed by Avalos et al.19 The goal of this review is to focus on malignancies post G-CSF exposure; other potential late effects are not considered. The discussion details data in the following areas on donors who have received G-CSF: (1) changes in blood counts/immunological parameters, (2) changes at the chromosomal level, (3) changes in gene expression profiles and (4) the incidence of haematological malignancies after donation compared with either BM donor or normal population controls.
Changes in blood counts and cellular or immunological markers
Several studies have investigated whether there are long-term changes in peripheral blood counts or other immunological markers in healthy donors receiving G-CSF (Table 1). In most studies changes in the blood count are transient and rarely persist beyond a maximum of 1 year following the G-CSF exposure.20, 21, 22 Even in studies where longer lasting abnormalities in particular cell sets have been found in donors (minor changes in neutrophil, lymphocyte or monocyte counts), these have not been associated with any health problems in the donor and hence the clinical significance of this is unknown.18, 20, 23, 24
Changes at the chromosomal level
Several studies have directly addressed the question of whether G-CSF administered to healthy individuals causes persistent chromosomal abnormalities (Table 2). Kaplinsky et al.25 noted a low level of tetraploidy in normal individuals given G-CSF; however, abnormalities were not found in the CD34+ cell fraction, nor were they noted in patients following the infusion of cells, suggesting that these are limited to mature non-progenitor cells. Similarly, Shapira et al.26 found transient increases in DNA destabilization during the G-CSF administration period that returned to the basal level 25 days after exposure.
Hirsch et al.27 followed 22 donors (20 related, 2 unrelated), with repeated blood tests up to 1 year post donation. Eight additional donors provided samples for the 1 year assessment and results were compared with those from 22 age and gender-matched controls. Nine chromosomal loci, from six different chromosomes were tested, either due to their relevance to myeloid malignancies, or to allow comparison with previous studies. PBL were processed by two methods either directly or following culture (again for comparison with previous studies). FISH was used to evaluate aneuploidy and replication timing. Overall, their results showed no significant differences in the rates of absolute aneuploidy between the donors who received G-CSF and the controls for any of the chromosomes tested. Similarly, there were no consistent significant increases in asynchronous cells in the G-CSF donors by replication timing. The overall percentage of asynchronous cells was less than that seen in the study by Nagler et al.5 Although these two studies were not performed on CD34+ selected cells, the direct, uncultured preparations are expected to include the mobilized CD34+ cell population.
To directly address the issue of changes in progenitor cells, Marmier-Savet et al.,28 studied cytogenetic abnormalities in both CD34- and CD34+ cell fractions, isolated from a subset of donors mobilized with G-CSF. Although they found an increase in aneuploidy (of chromosomes 8 and 17) in samples taken at early time points post exposure to G-CSF, this was not found in the CD34+ cell fraction when compared with control samples. In addition, they found aneuploid nuclei only in the interphase fractions (not in metaphase preparations) leading them to speculate that cells with aneuploidy may be unable to complete cell division, potentially conferring a protective effect against malignant transformation. Similarly, in 35 stem cell donors, Olnes et al. 29 found no increase in monosomy 7 or trisomy 8 aneuploidy in CD34+ selected PBSC, either in stimulated or unstimulated preparations.
The UK unrelated donor registries have performed a study of chromosomal changes with both a retrospective (50 BM and 50 PBSC donors 3–5 years post donation) and prospective arm (50 PBSC donors tested on the day of donation, then 90 and 180 of G-CSF exposure). Chromosomal analysis on samples from PBSC (n=100), BM (n=50) and normal controls (n=50) was performed. Dual colour FISH probes were designed to test for abnormalities in Chromosomes 7, 8 and 17, with techniques resembling those employed by Nagler et al.5 as closely as possible. Both metaphase and interphase preparations were assessed. Using international agreed standards, no increase in abnormalities (either aneuploidy or unbalanced structural abnormalities) were found in the PBSC group compared with either of the other two groups (E Nacheva, personal communication).
In summary, none of these studies show any persistent chromosomal abnormalities in the mononuclear cells (or CD34+ cell fractions, where tested) when using modern and standardised analytical techniques. Both by number and quality, these studies convincingly refute the worrisome findings of protracted chromosomal changes noted by Nagler et al.5
Changes in gene expression profiles
Several studies have used microarray technology to examine either mononuclear or CD34+ cells isolated after G-CSF stimulation to assess potential changes in the gene profiles of these cells (Table 3). Graf et al. analysed CD34+ cells pooled from several donors either post G-CSF mobilization or BM donation. They found a small set of sequences to be either over- or underexpressed in the PBSC samples compared with the BM samples. Only a single time point was tested.30 A few recent studies have addressed the issues of whether these changes in donor cells are persistent in-vivo. Baez et al.31 examined isolated CD34+ cells from six healthy donors before G-CSF exposure and on day 5, 30 and 365 after G-CSF mobilisation. They found that changes in both gene expression profiles and miRNAs were present and persisted until at least 1 year after the administration of G-CSF. miRNA was mostly overexpressed and involved in processes such as cell cycle, proliferation, angiogenesis and the immune response. Controls had injections of saline (n=6) and none of these showed any alteration of miRNA expression between the pre and day 30 samples. The authors question whether sustained overexpression of these miRNA could lead to modifications in the implicated biological processes in these donors. In contrast, two studies have analysed the gene expression profiles in the mononuclear cell fraction, rather than CD34+ cells. In the first, Amariglio et al.32 reported that, although alterations in the expression levels of multiple genes were found in sibling donors (n=4) following G-CSF mobilisation, these were essentially all transient and returned to normal within 4 months of exposure to the drug. Similarly Hernandez et al.33 found that differences in gene expression had returned to normal within 2 months of G-CSF exposure.
In summary, all studies to date have demonstrated changes in the gene expression profiles of either CD34+ or mononuclear cells in patients receiving G-CSF. Two studies show complete resolution of these changes, while in one several of the changes (in CD34+ cells) were still persistent at 1 year. The clinical significance of these findings is currently unknown. No studies have followed donors for longer than 1 year. Future studies in this area are recommended.
Incidence of haematological malignancies in healthy donors
Given that G-CSF stimulates BM myeloid cells, much of the monitoring for increased risk of cancer has focused on AML/MDS. Estimates regarding numbers of donors that would need to be followed long-term have been put forward,34 but may be difficult to calculate due to variability in study methodology. Below we will review data from several studies that show little or no signal in over 50 000 donors (Table 4). Virtually all of these donors have been followed for longer than 2 years, with the median length of follow-up at 3–5 years for most studies. The follow-up in terms of donor years is>150 000 even when only considering the studies that have reported outcomes in this way. The studies vary in their comparator groups, some comparing directly with BM donors, who may be more likely to have an age and health status like PBSC donors, and some with the general population, who may be slightly less healthy than volunteer donors. Both types of comparisons, however, do have value in trying to detect whether cancer risk is increased with G-CSF use by donors, and both give reassuring results.
Most recently the NMDP published their results in a very large data set of 2726 BM and 6768 PBSC unrelated donors enrolled between January 2004 and July 2009.35 The overall incidence of cancer was not significantly different between BM and PBSC donors. In addition, it was lower than that of the general population. There were only three haematological malignancies reported (one case each of AML (PBSC), Myeloma (PBSC) and Lymphoma (BM)). The NMDP donor outcomes from the earlier years of G-CSF use have also been published, in a separate and non-overlapping analysis.14 Malignancy rates in 2408 PBSC donors, who donated between 1999 and 2004, were evaluated. The median follow-up was 49 months (range: 2 days to 99 months). Only one haematological malignancy, chronic lymphocytic leukaemia, was reported in these donors.
A second large study by Holig et al.18 reported the outcomes in 3928 German unrelated donors who donated PBSCs between January 1996 and January 2008. In a follow-up period of 8234 donor years, 12 donors were reported to develop malignancies. Of these, four were haematological malignancies (AML (1), CLL (1), Hodgkin Disease (2)). For AML and CLL, this incidence is not different from that reported in the general population; however, in Hodgkin disease the incidence was higher than the general German population, showing a deviation from the expected rate in an age-adjusted control population (standardised incidence ratio 8.86; 95% confidence interval, 1.06–31.98). Although the incidence of Hodgkin disease is statistically increased, not a single donor in the NMDP experience had this disorder. The fact that only two donors in the cohort developing this disorder led to a significant value puts this observation at high risk of being an anomaly.
A later update by Holig,36 which extended the data set to 2012 and included 8290 donors, added four additional cases with haematological malignancies (once each of HD, ALL, CML and AML). They found that the incidence of HD, as previously, was higher than that in the normal German population, but, in this study, AML, unlike previously, was also found to be higher than the general population. A total of two AML cases in the population, with none in the NMDP study, carries the same risk mentioned above of being a statistical anomaly.
A later updated and extended data set from the DKMS in 12 559 unrelated donors was presented at the ASH annual meeting in 2010.37 The DKMS sent questionnaires to their donors with an 81% response rate. They calculated that there were a total of 55 228 donor observation years (30 777 observation years for 8730 PBSC donors, 23 037 observation years for 3556 BM donors and 1414 observation years for 273 donors of both PBSCs and BM). In those for whom the outcome is known, there were six cases of haematological malignancies in PBSC donors: two Hodgkin's disease (as previously reported) and one plasmacytoma; in BM donors: one case of AML and one of non-Hodgkin's lymphoma and one case of chronic lymphocytic leukaemia in a donor who had donated via both routes). There was no increase in the incidence of haematological malignancies between BM and PBSC. In addition, the standard incidence ratio values for PBSC and BM donors are 1.12 (0.82–1.50) and 0.84 (0.56–1.19), respectively, showing no increase compared with the general population. One further case of Hodgkin’s Disease was reported in a study from the Netherlands,38 but this did not show an increased incidence above the age and gender-matched normal population.
The experience of the Spanish National Donor Registry,39 reporting on long-term outcomes in 736 PBSC donors (of whom 320 had been followed for 2 years or more), showed no cases of haematological malignancies, as did a study of 198 donors from Germany.40
The EBMT conducted two survey-based studies to investigate the rate of events in both related and unrelated donors.17 The first survey covered the years 1993–2002, while the second survey covered the years 2003–2005. Overall 20 hematologic malignancies were reported, 8 among BM donors and 12 among PB donors. Only one case was in an unrelated donor (AML), while the remainder were in siblings. The rate could not be compared with the general population as some of the donors had missing information (for example, age or sex); however, taking this limitation into account the observed incidence of haematological malignancies was not greater than that would be expected.
In the related donor setting, a study from Japan reports the outcomes in 3114 PBSC donors who were enrolled in a prospective mandatory national registration protocol beginning in 2000.41 There was just one donor who developed a haematological malignancy (AML) in this cohort. To attempt to estimate the incidence of late adverse events among BM family donors in Japan over a similar time period, questionnaires were sent to 286 transplant teams belonging to Japan Society for Hematopoietic Cell Transplantation (JSHCT). In all, 67% of those surveyed responded with information from 5921 BM harvests from family donors performed between 1991 and 2003. In that group, two donors had developed AML.
Smaller single centre studies in related donors (reporting on a total of 667 donors who could be contacted months to years following their donation) have reported no new cases of haematological malignancies in PBSC donors (with one study reporting 1 each of ALL and CLL in BM donors).3, 22, 42
The WMDA instituted a system for the reporting of both short- and long-term serious adverse reactions in unrelated donors in 2002. Beginning as a voluntary scheme, participation has since become mandatory for WMDA accredited registries and is strongly encouraged for all other registries; however, this may not cover all unrelated donations that currently occur. The cases reported to WMDA are also likely to be represented in the large publications from individual registries, which are mentioned above, except for the most recent years where the WMDA data are more contemporary. In 2012, three cases of haematological malignancies were reported to SEAR: one each of AML, Hodgkin’s disease and MDS, although without a reliable denominator, it is difficult to draw conclusions from these data. A registry for all donors, as is already mandatory in certain countries (for example, Japan, Switzerland), could enhance our ability to track complications like these in all donors.
Multiple exposures to G-CSF (granulocyte donors)
Most PBSC donors are not exposed to multiple courses of G-CSF, even in the related donor setting. In contrast, this may well be the case in granulocyte donors, where only a single dose of G-CSF is given, but these may be repeated on multiple occasions. Studies in both related and unrelated donors have addressed this issue and have considered haematological, genetic as well as clinical outcomes. Quillen et al.24 followed up 92 donors who had donated a mean of 13.5 times. The median follow–up was 10.5 years (with a minimum follow up period of 5 years). The granulocyte donors were matched to platelet donors, who had experienced a similar number of apheresis procedures, but had not been exposed to G-CSF. No increase in long-term health problems was identified between the two groups (with specific emphasis on cardiac, vascular, haematological disorders or cancer). Interestingly, although the absolute neutrophil and lymphocyte counts in granulocyte donors were in the normal range at all time points (and the donors clinically well), they did show a significant decrease from the baseline test to the final blood test (which was not seen in the platelet donors). Olnes et al.29 included 38 granulocyte donors in their genetic study mentioned above. Donors had been exposed to a median of 12 doses of G-CSF (range: 3–42) at least 6 months before samples being taken for this analysis. As in PBSC donors, the authors found no chromosome 7 or 8 aberrations by FISH or SKY techniques. These studies provide reassuring evidence against long-term clinically relevant health problems or myeloid malignancies in donors with multiple G-CSF exposures, at a cumulative lifetime dose exceeding that of PBSC donors, although it is possible that homeostatic stressors differ in the context of multiple, widely spaced exposures.
Summary and updated statement
The evidence summarised here suggests that, to date, there is no compelling evidence for an increase in haemtological malignancies among healthy donors who receive G-CSF to mobilise cells for donation. Only one study has found the incidence to be higher than the age and gender-matched population for certain malignancies. While clinically insignificant abnormalities in blood counts (and other immunological markers) have been shown, most studies to date have noted that these are transient. A few studies have shown persistence in small changes in blood counts several years post donation, but in no studies is this correlated with any clinical sequelae in the donor.
This conclusion is largely supported by the available in vitro data from several studies, which have examined both mononuclear and CD34+ cell populations, using standardised and validated methodology. Although not all studies can be directly compared with each other due to differences in the cellular population, time points at testing, or testing methodology, the majority have shown only transient abnormalities in either chromosomal number (or other abnormalities) or gene expression profiles. The exception is a study by Baez et al.,31 which does show persistent changes at 1 year post exposure. Further analyses of this sort are warranted to investigate the significance of these findings and correlation with the clinical condition for these donors is necessary.
In view of this evidence the WMDA released a revised statement for inclusion in all WMDA donor consent forms in 2012. As related donors may receive G-CSF by varying doses and schedules and may have a different underlying cancer risk compared with unrelated donors, it is not certain that the impact of G-CSF is identical in the related donor setting. Thus, the WMDA statement if focused on unrelated donors: ‘normal individuals are at risk for developing cancer, including leukemia, lymphoma or other blood diseases throughout their life time. G-CSF stimulates normal blood cell growth. In some patients with cancer or abnormal blood cells, it has been shown to stimulate leukemic blood cells. Studies following large numbers of unrelated donors have shown that the risk of developing cancer within several years after the use of G-CSF is not increased compared with donors not receiving G-CSF.’
The WMDA continues to recommend that all donors are asked about a family history of leukaemia and that all donors have long-term follow-up. Although the WMDA does not make recommendations or inspect institutions that care for related donors, there are other groups, such as the Worldwide Network for Blood and Marrow Transplantation (WBMT), the Foundation for the Accreditation of Cellular Therapy (FACT) and Joint Accreditation Committee-ISCT & EBMT (JACIE), who have stressed the importance of long-term follow-up of related donors. The international donor and transplant communities are working towards the harmonisation of reporting procedures. Importantly, the WMDA revised statement clearly states that the long-term results reported to date are all in donors who have received ‘originator product G-CSF’ (Neupogen (Filgrastim, Amgen) or granocyte (Lenograstim, Chugai)) and that the conclusions drawn here cannot necessarily be drawn for donors who are exposed to any other mobilising agent.
In conclusion, to date, long-term outcome studies and genetic analyses have provided reassuring evidence that the use of G-CSF to mobilise stem cells for transplantation is a safe and acceptable method of hematopoietic cell donation.
About this article
Myelodysplastic syndrome and acute myeloid leukemia following adjuvant chemotherapy with and without granulocyte colony-stimulating factors for breast cancer
Breast Cancer Research and Treatment (2015)