Risk of Acute Myeloid Leukemia and Myelodysplastic Syndrome in Trials of Adjuvant Epirubicin for Early Breast Cancer: Correlation With Doses
http://www.100md.com
《临床肿瘤学》
the Centre Oscar Lambret, Lille
Centre Georges-Franois Leclerc, Dijon
Centre Eugène Marquis, Rennes, France
Karolinska Hospital, Stockholm
University Hospital, Uppsala, Sweden
Institute of Cancer Research, Sutton
Cancer Research United Kingdom Laboratories, Imperial College, London, United Kingdom
Institut Jules Bordet and European Organisation for Research and Treatment of Cancer Data Center, Brussels, Belgium
National Cancer Institute of Canada Clinical Trials Group, Kingston, Ontario, Canada
Laurentius Hospital, Roermond, the Netherlands
Centre Hospitalier Princesse Grace, Monaco
Università degli Studi, Brescia
Ospedale Maggiore di Milano, Istituto di Ricovero e Cura a Carattere Scientifico
Safety Surveillance and Reporting, Pharmacia, Gruppo Pfizer, Milan, Italy
ABSTRACT
PURPOSE: We reviewed follow-up of patients treated in 19 randomized trials of adjuvant epirubicin in early breast cancer to determine incidence, risk, and risk factors for subsequent acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS).
PATIENTS AND METHODS: The patients (N = 9,796) were observed from the start of adjuvant treatment (53,080 patient-years). Cases of AML or MDS (AML/MDS) were reported, with disease characteristics. Incidence and cumulative risk were compared for possible risk factors, for assigned regimens, and for administered cumulative doses of epirubicin and cyclophosphamide.
RESULTS: In 7,110 patients treated with epirubicin-containing regimens (92% of whom also received cyclophosphamide), 8-year cumulative probability of AML/MDS was 0.55% (95% CI, 0.33% to 0.78%). The risk of developing AML/MDS increased in relation to planned epirubicin dose per cycle, planned epirubicin dose-intensity, and administered cumulative doses of epirubicin and cyclophosphamide. Patients with administered cumulative doses of both epirubicin and cyclophosphamide not exceeding those used in standard regimens ( 720 mg/m2 and 6,300 mg/m2, respectively) had an 8-year cumulative probability of developing AML/MDS of 0.37% (95% CI, 0.13% to 0.61%) compared with 4.97% (95% CI, 2.06% to 7.87%) for patients administered higher cumulative doses of both epirubicin and cyclophosphamide.
CONCLUSION: Patients treated with standard cumulative doses of adjuvant epirubicin ( 720 mg/m2) and cyclophosphamide ( 6,300 mg/m2) for early breast cancer have a lower probability of secondary leukemia than patients treated with higher cumulative doses. Increased risk of secondary leukemia must be considered when assessing the potential benefit to risk ratio of higher than standard doses.
INTRODUCTION
Over the last 20 years, adjuvant therapy for early breast cancer with DNA-targeted antiproliferative drugs has substantially increased the number of long-term survivors. However, a small proportion of these survivors subsequently develop acute myelogenous leukemia (AML), with or without a preleukemic myelodysplastic syndrome (MDS). When long-term survival is expected, estimation of the risk of secondary AML/MDS after chemotherapy is important to determine the risk to benefit ratio of treatment
with this aim, patients treated with epipodophyllotoxins for pediatric leukemias1 and solid tumors2 and with the combination of doxorubicin and high-dose cyclophosphamide for breast cancer3 have been observed in monitoring plans. These studies have raised concerns over the roles of topoisomerase II inhibitors, alkylating agents, and growth factors in the development of new primary malignancies.
Adjuvant chemotherapy regimens containing epirubicin, a DNA topoisomerase II–targeted anthracycline, are widely used in the treatment of breast cancer. In 1992, a Danish survey reported five cases of AML in 360 patients with advanced breast cancer treated with a regimen of epirubicin and either cyclophosphamide or cisplatin or after previous exposure to other alkylating agents (melphalan or lomustine).4 Cases of secondary leukemia were then reported in patients with early breast cancer treated with epirubicin.5 In response to these reports, the manufacturer (then Pharmacia & Upjohn, now Pfizer
New York, NY) extended its routine pharmacovigilance to include a monitoring plan for secondary leukemia in patients treated with epirubicin-containing regimens in controlled clinical trials of adjuvant therapy for early breast cancer.
PATIENTS AND METHODS
Study Population
The Monitoring Plan included all patients entered onto 19 randomized clinical trials, which were all the research ethics committee–approved trials for adjuvant treatment of early breast cancer known to the manufacturer that included epirubicin in at least one arm (Table 1). 6-19 All the trials defined early breast cancer as node positive or negative, not previously treated, and without distant metastases. The plan was activated in June 1995. The cutoff date for this analysis was December 31, 2001, when all but one trial (GFEA08) had completed accrual. Results of eight trials have been published,6,8,9,12-14,16,19 and preliminary results of other trials have been presented as abstracts.10,11,17,18,20 The trials included patients enrolled as early as 1984, so the analysis covers more than 15 years of clinical experience.
The trials contained 44 different treatment arms. These arms were pooled into the following five groups: chemotherapy including epirubicin at an assigned dose per cycle of 100 mg/m2
chemotherapy including epirubicin at an assigned dose per cycle of more than 100 mg/m2
chemotherapy not including epirubicin
hormone therapy without chemotherapy
no chemotherapy or hormone therapy. The cutoff point of 100 mg/m2/cycle for assigned epirubicin dose was derived from the fluorouracil, epirubicin, and cyclophosphamide (FEC) FEC100 regimen commonly used in Europe.9
Data Management and Statistical Analysis
Patient characteristics, treatment, and follow-up were extracted from the original study databases and merged into a SAS database (SAS Institute, Cary, NC). All cases of leukemia (any type) or MDS were reported together with disease characteristics, treatment (within and outside the original protocol), and follow-up.
Crude rates for AML/MDS (number of AML/MDS cases divided by number of patient-years [PY] at risk) with 95% CIs were compared for the five groups using the Score test.21 The hazard function for AML/MDS with 95% CIs was calculated for all epirubicin-treated patients, using the life-table method, for each 12-month period for up to 96 months of follow-up, when at least 20% of the patients remained at risk. Median cumulative doses of epirubicin and cyclophosphamide were compared between granulocyte colony-stimulating factor (G-CSF) –treated or untreated patients and tamoxifen-treated or untreated patients using the Mann-Whitney U test.
The rates of incidence of AML/MDS were modeled by multivariate backward Poisson regression models based on classes of patients according to age ( 49 or > 49 years), assigned cumulative doses of epirubicin and cyclophosphamide, and whether they received radiotherapy, were mandated to receive G-CSF, or actually received tamoxifen, together with the corresponding interactions.22
Kaplan-Meier cumulative probability estimates for AML/MDS with time were calculated for all epirubicin-treated patients. No allowance was made for competing risks on the grounds that there were no true competing risks for the AML/MDS event and that, for patient and physician, the most relevant risk was that of developing AML/MDS if the patient was a survivor. Because both the Kaplan-Meier and the direct estimates of cumulative incidence are small, any bias in the Kaplan-Meier estimate resulting from censoring is small and has negligible clinical implications. Cumulative probabilities of AML/MDS over the follow-up time were calculated for subgroups of patients according to planned epirubicin dose per cycle ( 100 v > 100 mg/m2/cycle), planned dose per week ( 33.3 v > 33.3 mg/m2/wk), and administered cumulative dose ( 720 v > 720 mg/m2) across all study arms containing epirubicin. Survival curves were censored for death and loss to follow-up to give an estimate of risk in the surviving population and compared for the subgroups using the log-rank test.23 The Cox proportional hazards model was used for assessing the independent role of each subgroup risk factor. Kaplan-Meier cumulative probability estimates for AML/MDS risk were calculated for the administered cumulative doses (mg/m2) of epirubicin and cyclophosphamide.
RESULTS
Patients
The population at risk, which was defined as all randomized patients for whom follow-up information was available, was 9,796 (97% of the 10,111 patients randomized in the trials). The distribution of follow-up across the clinical trials is shown in Table 1. The 9,796 assessable patients contributed 53,080.7 PY of follow-up before death or last contact. As of December 31, 2001, 2,542 assessable patients were reported to have died
cumulative survival probability was 78.2% (95% CI, 77.2% to 79.1%) at 5 years, 67.5% (95% CI, 66.3% to 68.7%) at 8 years, and 61.5% (95% CI, 60.0% to 63.0%) at 10 years.
Events
Up to December 31, 2001, 30 (0.306%) of the total population of 9,796 patients had presented with AML/MDS. In addition, three patients presented with acute lymphoblastic leukemia, one presented with hairy cell leukemia, and one presented with chronic lymphocytic leukemia (Table 1). Secondary leukemias occurred in 12 trials, with the number of cases ranging from none to four per trial except in NCIC/MA5 (n = 6) and SBG9401 (n = 10).
Of the 7,110 patients randomly assigned to epirubicin-containing regimens, 28 (0.394%) developed AML/MDS. Only one (patient 27, EORTC10921) of these patients experienced a breast cancer relapse before presenting with AML/MDS
the relapse had been treated with taxanes and fluorouracil.
One (0.07%) of the 1,427 patients randomly assigned to chemotherapy regimens not including epirubicin developed AML/MDS. This patient (patient 25, NCIC/MA5 trial), who was treated initially with cyclophosphamide, methotrexate, and fluorouracil (CMF), had a breast cancer relapse (treated only with radiotherapy and tamoxifen) 7 months before presenting with leukemia.
One (0.11%) of the 903 patients randomly assigned to hormone therapy without chemotherapy developed AML/MDS. This patient (patient 16, GFEA02 trial), who was treated initially with tamoxifen, had a breast cancer relapse, which was treated with mitoxantrone, cyclophosphamide, fluorouracil, epirubicin, vinorelbine, and mitomycin, 3 years before presenting with leukemia.
Individual characteristics of the patients are listed in Table 2, where the AML cases are listed according the French-American-British (FAB) morphologic criteria.24 The most frequent subtype of AML was M4 (n = 10), followed by M2 (n = 6), M5 (n = 4), and M3 (n = 1)
in four patients, FAB classification was unknown. Five patients presented with MDS, and a preleukemic myelodysplastic phase was suspected in five additional patients who eventually progressed to AML (patients 2, 5, 10, 24, and 25).
The time elapsed from the start of adjuvant treatment to the diagnosis of AML/MDS ranged from 8 to 126 months, with a median of 33 months. Latency was shorter (median, 17.5 months
range 8 to 81 months) in the 14 M4/M5 patients than in the six M2 patients (median, 44.5 months
range, 19 to 126 months). Of the 29 patients with AML/MDS who had received epirubicin and/or cyclophosphamide per protocol, 13 developed AML/MDS (nine of these patients developed M4/M5) within the first 2 years from first chemotherapy
these patients were younger (median age, 46 years) than the 16 patients (median age, 51 years) who developed AML/MDS more than 2 years after first chemotherapy.
Cytogenetic analyses were available for 24 AML/MDS patients and one acute lymphoblastic leukemia (ALL) patient. Of the AML patients, three had normal karyotype, 12 had balanced chromosomal translocations with or without a single chromosome abnormality, seven had abnormalities of only one or two chromosomes (deletion 5q
monosomy/abnormality 7, trisomy 8
inversion 16), and two had complex karyotypes with three or more chromosome abnormalities.
As of December 31, 2001, 12 AML/MDS patients had survived 1 year or more from leukemia diagnosis
six of these patients were still alive at the last follow-up, at 13, 14, 17, 34, 75, and 78 months from diagnosis. Five patients survived less than 1 month from leukemia diagnosis. The two M4/M5 patients with the longest survival from leukemia diagnosis (14 months and > 34 months) had received only epirubicin and tamoxifen as adjuvant treatment for their early breast cancer.
Risk Factors Evaluation: Incidence Rates and Multivariate Analysis
The median reported age for the 9,796 assessable patients at start of adjuvant treatment was 49 years (range, 23 to 93 years)
age was not reported for 73 patients. Incidence rates for AML/MDS were similar in the age groups of 49 or older than 49 years (Table 3).
After surgery, 7,216 patients (81%) received local radiotherapy (details of dose, field, and modality were not requested), whereas 1,664 (19%) patients did not receive radiotherapy
for 916 patients, no information on radiotherapy was reported. Twenty-seven cases of AML/MDS were reported in patients who received radiotherapy, and three cases were reported in patients who did not receive radiotherapy
incidence rates were 0.666 and 0.329/1,000 PY, respectively, which do not differ significantly (P = .2354, Table 3). No cases were reported in the patients without information on radiotherapy.
Treatment with G-CSF was required by protocol in four study arms with intensive chemotherapy (GETIS02, arm I
EORTC10921, arm I
SBG9401, arms I and II). Although in some of the other studies G-CSF was to be administered as needed, G-CSF administration other than that required by protocol was rarely reported. Thirteen cases of AML/MDS occurred in the 796 patients mandated to receive G-CSF by protocol, giving a significantly (P = .0001, Table 3) higher incidence rate (5.453/1,000 PY) versus patients without G-CSF administration (0.336/1,000 PY). The patients mandated to receive G-CSF according to protocol received median cumulative doses of 634 mg/m2 of epirubicin and 8,267 mg/m2 of cyclophosphamide, both of which were significantly (P = .0001) higher than the corresponding median cumulative doses (304 mg/m2 and 4,547 mg/m2, respectively) in patients not mandated to receive G-CSF.
In the 3,230 patients who actually received tamoxifen, regardless of their study protocol (median age, 54 years
range, 22 to 93 years), there were 19 cases of AML/MDS. Eleven cases occurred in the 6,567 patients who did not receive tamoxifen (median age, 46 years
range, 23 to 85 years)
the corresponding AML/MDS rates were significantly different (P = .0001
1.260 and 0.289/1,000 PY, respectively
Table 3). Median epirubicin and cyclophosphamide cumulative doses were 516 and 4,775 mg/m2, respectively, in the group of tamoxifen-treated patients and were significantly higher (P = .0001) than the corresponding doses (300 and 4,684 mg/m2, respectively) in the group of patients who did not receive tamoxifen.
At Poisson regression analysis, assigned cumulative doses of epirubicin and cyclophosphamide were found to be the only independent risk factors. Age, reported radiotherapy, mandated use of G-CSF, and reported treatment with tamoxifen were not found to be independent risk factors
patients who received these treatments also tended to receive high cumulative doses of epirubicin and cyclophosphamide.
AML/MDS Incidence Rates and Hazard Function
The distribution of assessable patients and follow-up across the five assigned treatment groups is shown in Table 4, which also lists the average patient age at start of adjuvant treatment, the percentage of patients who were mandated to receive G-CSF according to the original study protocol, the number of reported AML/MDS cases, and the rate/1,000 PY of the AML/MDS event together with its 8-year cumulative probability. There were no significant differences in incidence rates between patients treated with epirubicin at a planned dose 100 mg/m2, patients receiving chemotherapy not including epirubicin, and patients treated with hormone therapy. The incidence rate was 3.620/1,000 PY (95% CI, 2.260 to 5.798/1,000 PY) in patients treated with epirubicin administered in various combinations at a planned dose per cycle of more than 100 mg/m2, which was significantly (P = .0001) higher than in all other treatment groups (rates ranging from 0.117 to 0.345/1,000 PY).
The hazard function for AML/MDS events in all epirubicin-treated patients is shown in Figure 1. A first peak (0.15 x 10–3) occurs in the second year
thereafter, the hazard function seems to plateau, increasing slowly to a less pronounced peak (0.09 x 10–3) in the fifth year. The third peak (0.07 x 10–3) in the seventh year is a result of two cases occurring when only a limited number of patients remained at risk.
AML/MDS Cumulative Probability Over Time in Relation to Treatment
The cumulative probability of AML/MDS in the 7,110 patients treated with epirubicin was 0.27% (95% CI, 0.14% to 0.40%) at 3 years, 0.46% (95% CI, 0.28% to 0.65%) at 5 years, and 0.55% (95% CI, 0.33% to 0.78%) at 8 years. Figure 2 shows that the cumulative probability of AML/MDS was significantly higher in the patients assigned to the epirubicin regimens with a planned dose of more than 100 mg/m2/cycle (P = .0001) or more than 33.3 mg/m2/wk (P = .0001) or who were administered a cumulative dose of more than 720 mg/m2 (P = .0001) compared with patients assigned to epirubicin doses that were less than these amounts
the cumulative probabilities of AML/MDS were 2.3% v 0.27% at 8 years, 3.2% v 0.31% at 6 years (graph truncated at 6 years because, after this, too few patients remained at risk in the > 33.3 mg/m2/wk group), and 4.4% v 0.62% at 8 years, respectively. Only the epirubicin planned dose ( 100 or > 100 mg/m2/cycle) and the epirubicin administered cumulative dose ( 720 or > 720 mg/m2) remained as independent factors in the Cox model (both P = .0001)
the adjusted relative risks of having the event of AML/MDS were 5.76 (95% CI, 2.36 to 14.04) for the more than 100 mg/m2/cycle dose group versus the 100 mg/m2/cycle dose group and 6.80 (95% CI, 2.86 to 16.13) for the more than 720 mg/m2 dose group versus the 720 mg/m2 dose group. Epirubicin planned dose-intensity ( 33.3 v > 33.3 mg/m2/wk) was not found to be an independent factor.
AML/MDS Cumulative Probability in Relation to Administered Cumulative Doses of Epirubicin and Cyclophosphamide
Epirubicin and cyclophosphamide were administered in combination in most patients. Given that the Poisson distribution analysis demonstrated that the planned cumulative dose of each was an independent risk factor, the cumulative probability of AML/MDS was plotted in relation to the administered cumulative dose of each agent. The plot against administered epirubicin cumulative dose (Fig 3) shows a progressive increase in risk, with cumulative probabilities of 0.07% at 300 mg/m2, 0.34% at 600 mg/m2, and 1.53% at 720 mg/m2. Thereafter, the cumulative probability increases steeply, reaching 9.94% (95% CI, 4.46% to 15.41%) at 900 mg/m2. The AML/MDS cumulative probability also increases with administered cyclophosphamide oral or intravenous cumulative dose, with cumulative probabilities of 0.08% at 3,000 mg/m2, 0.29% at 6,000 mg/m2, and 0.98% at 9,000 mg/m2
thereafter, the cumulative probability increases steeply, reaching 6.4% (95% CI, 1.92% to 10.90%) at 12,000 mg/m2.
Table 5 shows the cumulative probability of AML/MDS over time for subgroups according to whether administered cumulative doses of either epirubicin or cyclophosphamide or both were within or above the standard doses. The CIs overlap for the two epirubicin subgroups (< 720 mg/m2 or > 720 mg/m2) when cyclophosphamide was administered within the standard cumulative dose (6,300 mg/m2), but when cyclophosphamide was administered above the standard cumulative dose, there was a significant increase in incidence at all three time points when epirubicin was administered above the standard cumulative dose. In the 4,760 patients whose administered cumulative doses of both epirubicin and cyclophosphamide did not exceed those used in standard regimens, the 8-year cumulative probability of developing AML/MDS was 0.37% (95% CI, 0.13% to 0.61%) compared with 4.97% (95% CI, 2.06% to 7. 87%) in the 261 patients who received higher cumulative doses of both epirubicin and cyclophosphamide.
DISCUSSION
The success of adjuvant chemotherapy in prolonging disease-free survival has resulted in the treatment of thousands of women each year. The risk of AML/MDS is an important consideration in a patient’s treatment decision. Secondary leukemia risk has been evaluated in many studies of commonly used alkylating agents and anthracyclines.
The Epirubicin Monitoring Plan is the most comprehensive analysis of secondary leukemia after treatment with adjuvant epirubicin completed to date, with 9,796 patients observed for a total of 53,080 PYs. The analysis confirms a small but significant risk of AML/MDS in patients treated with adjuvant epirubicin and cyclophosphamide for early breast cancer. In all of the 7,110 patients treated with epirubicin-containing regimens, the 8-year cumulative probability of AML/MDS was 0.55%. The risk of developing AML/MDS increased in relation to total dose of epirubicin and total dose of cyclophosphamide, with a steep increase at cumulative doses above those used in recommended regimens for adjuvant therapy (epirubicin up to 720 mg/m2 and cyclophosphamide up to 6,300 mg/m2). The highest risk (4.97% cumulative probability at 8 years) was seen in patients who had been administered greater than standard doses of both epirubicin and cyclophosphamide.
The monitoring plan covered a period of clinical experience extending over more than 15 years. During this period, the therapeutic approach to early breast cancer adjuvant treatment became more aggressive, with the epirubicin dose increasing from approximately 50 mg/m2 every 3 weeks in the earlier studies up to 120 mg/m2 every 3 or 4 weeks in the later studies. In the six-cycle regimens that are approved for adjuvant therapy by the US Food and Drug Administration, epirubicin (in combination with cyclophosphamide and fluorouracil) is administered at 100 mg/m2 on day 1 every 3 weeks or at 60 mg/m2 on day 1 and on day 8 every 4 weeks, for total cumulative doses of 600 and 720 mg/m2, respectively. To evaluate the effect of the epirubicin dose on the risk of AML/MDS, cutoff points for dose per cycle, dose per week, and total cumulative dose were chosen to correspond to upper limits of dosing in these regimens
the cutoff point for dose per cycle was chosen for the first of these regimens, which is the more widely used regimen in Europe.
The incidence rate (3.62/1,000 PY) in the group of patients treated with epirubicin administered at a planned dose per cycle of more than 100 mg/m2 was significantly (P = .001) higher than the incidence rate (0.345/1,000 PY) in the group of patients treated with epirubicin at a planned dose of 100 mg/m2, suggesting an appreciable dose effect. At a planned dose of 100 mg/m2, the low risk of secondary leukemia does not differ significantly from the risk in patients not treated with epirubicin or who received hormonal therapy only.
Given that the risk of AML/MDS is not constant over time, as demonstrated by the hazard function (Fig 1), the AML/MDS cumulative probabilities were calculated and compared. Significant epirubicin dose effects were seen when the groups treated with a high planned epirubicin dose per cycle (> 100 mg/m2) or per week (> 33.3 mg/m2) or with a high administered cumulative dose (> 720 mg/m2) were compared with the corresponding groups treated at a lower dose (Fig 2). However, in the Cox model analysis, only planned dose per cycle and administered cumulative dose remained as independent factors, indicating that, in this analysis, dose-intensity does not have a significant independent association with AML/MDS risk.
Calculation of the AML/MDS cumulative probability as a function of the administered cumulative dose of epirubicin provides a method for estimating the risk in the individual patient according to the total dose received, similar to the risk estimation for anthracycline cardiotoxicity. If the cumulative dose remains within the maximum recommended for this indication (720 mg/m2), the AML/MDS cumulative probability increases slowly, up to 1.53% at 720 mg/m2, whereas above this dose, the incidence increases rapidly to reach 9.94% at a cumulative dose of 900 mg/m2 (Fig 3).
The monitoring plan was not designed specifically to evaluate the role of cyclophosphamide because trial selection was determined by inclusion of epirubicin in a study arm and subgroup analysis was determined by epirubicin dose schedule. However, 92% of patients treated with epirubicin also received cyclophosphamide, and the multivariate analysis identified epirubicin and cyclophosphamide as the only independent determinants of the leukemia risk. Secondary leukemias seen in patients treated with alkylating agents, such as cyclophosphamide, typically differ from those seen with DNA topoisomerase II inhibitors, such as the anthracyclines. For example, a myelodysplastic preleukemic phase has been observed in patients treated with alkylating agents but is not common in patients treated with DNA topoisomerase II inhibitors.25 In addition, AML in patients treated with topoisomerase II inhibitors typically has an early onset and an M4/M5 subtype, whereas AML after alkylating agents has a later onset and a M1/M2 subtype. Chromosomal changes also differ
chromosomes 5 and 7 are typically altered, usually with unbalanced changes, in AML/MDS after treatment with alkylating agents, whereas abnormalities of chromosomes 11 and 21, especially the balanced translocations 11q23 and 21q22, are more typical of AML after treatment with topoisomerase II inhibitors.
The features typical of AML/MDS caused by alkylating agents were seen in a number of patients in the present study, suggesting a major contribution by cyclophosphamide to the development of AML/MDS. For instance, four patients with MDS for whom cytogenetic information was available had unbalanced abnormalities of chromosomes 5 or 7. The only AML patient who did not received epirubicin but who did receive a large cumulative dose of cyclophosphamide had a long preleukemic myelodysplastic phase and also had an unbalanced chromosome 5 deletion. Six patients had M2 FAB subtype
of these six patients, five had onset of disease more than 2 years after start of adjuvant therapy, although only one patient had a typical unbalanced chromosome abnormality (deletions of both chromosomes 5 and 7).
Consistent with a role of cyclophosphamide, the plot of AML/MDS probability against administered cumulative dose of cyclophosphamide (Fig 3) shows that, if the cumulative dose remains within the maximum recommended dose in the combination regimens with epirubicin (6,300 mg/m2), the AML/MDS cumulative probability increases slowly up to 0.33%, whereas, above this dose, the cumulative probability increases more rapidly to reach 6.4% at a cumulative dose of 12,000 mg/m2.
The majority of patients with AML responded to induction chemotherapy, and 11 patients with AML/MDS (MDS, n = 3
M2, n = 4
M3, n = 1
M4, n = 2
unclassified AML, n = 1) had survived more than 1 year from diagnosis at the date chosen for this analysis. In treatment-related AML after treatment with epipodophyllotoxins and anthracyclines, balanced translocations [t(8;21), t(15;17),] and inv(16) have been reported as favorable cytogenetic abnormalities, with response rates and survival similar to de novo adult AML.26,27 In contrast, for alkylating agent–related secondary leukemia, the results of chemotherapy have generally been disappointing.14 However, in the present survey, better survival was observed in the M2 subgroup (median latency, 44.5 months), two of whom showed t(8;21), than in the M4/M5 subgroup (median latency, 17.5 months). This is similar to the poor prognosis reported by Linassier et al28 in a group of 10 patients with AML (M1/M2, n = 3
M4/M5, n = 6
M3, n = 1) and early breast cancer treated with mitoxantrone, cyclophosphamide, and fluorouracil (latency, 16 months).
The effect of escalating doses of cyclophosphamide (600, 1,200, and 2,400 mg/m2/cycle for two or four cycles) associated with a fixed dose (60 mg/m2/cycle for four cycles) of another anthracycline, doxorubicin, on the risk of secondary leukemia in early breast cancer patients is being monitored in six National Surgical Adjuvant Breast and Bowel Project (NSABP) adjuvant trials. In a recent analysis of 8,563 patients,3 43 AML/MDS cases were reported
intensified cyclophosphamide dosing requiring filgrastim support was associated with an increased incidence of AML and MDS. Although a significant correlation between planned cyclophosphamide dose-intensity and the leukemia rate was reported, the authors did not find a significant relationship with planned cumulative dose of cyclophosphamide, even if there was a clear trend towards increased risk with the highest doses. However, the maximum cumulative dose of cyclophosphamide planned to be administered in the NSABP studies was 9,600 mg/m2, which, in the curve of AML/MDS probability against cyclophosphamide cumulative dose in the present analysis, is equivalent to a cumulative risk of 1.57% (95% CI, 0.19% to 2.95%), so that the range of incidence in terms of cyclophosphamide dose in the NSABP studies is likely to be narrow. Previous studies carried out in patients with early breast cancer suggested that the risk of leukemia after CMF regimens for adjuvant chemotherapy using cyclophosphamide at standard dose was not much higher than that in the general population.1,29,30
Age, radiotherapy, mandated use of G-CSF, and assigned treatment with tamoxifen were not found to be independent risk factors in the Poisson regression analysis. Lack of any age-related leukemia risk is at variance from the NSABP experience with doxorubicin, in which a significantly higher leukemia rate was observed in women older than 49 years. Other studies have reported an increased risk of AML/MDS after radiotherapy in patients treated with adjuvant chemotherapy for breast cancer.1,31,32 That no significant effect was seen in the present study is probably a result of the large proportion of patients who received radiotherapy, making the comparison unbalanced, and a result of the lack of homogeneity of radiotherapy administration between the individual studies.
Although 13 of the 30 AML/MDS patients in this study were mandated to receive G-CSF, the present study was not designed to determine whether G-CSF treatment was causally linked to AML/MDS in this population. The apparent effects of G-CSF may be explained by the high doses of epirubicin and cyclophosphamide received by the patients included in the treatment arms to which G-CSF was assigned. However, it has been suggested that growth factors may be leukemogenic,33 and a possible link between the use of supportive G-CSF in adjuvant doxorubicin treatment of breast cancer and secondary leukemia has been suggested. Future analyses might investigate the possible role of G-CSF cumulative dose and duration of exposure in development of leukemia in patients treated with epirubicin.
The AML/MDS rate was also significantly higher in the patient population who received tamoxifen, but most patients who received tamoxifen were also found to have received high doses of epirubicin and cyclophosphamide. An increase in estrogen-dependent secondary malignancies has been reported after treatment with tamoxifen,34,35 but the occurrence of secondary AML/MDS is an extremely rare event. Moreover, the extensive randomized direct comparison of tamoxifen with observation or placebo summarized in the Early Breast Cancer Trialists’ Collaborative Group overview36 has shown no excess of AML/MDS in the patients receiving tamoxifen. In the present study, one case (patient 16) of AML M4 occurred 6 years after enrollment in the group of 903 patients treated with hormone therapy (mostly tamoxifen) alone
this patient, however, had suffered a breast cancer relapse, which was treated with cytotoxic therapy, including epirubicin, cyclophosphamide, and mitoxantrone, 3 years before leukemia onset. Thus, it is likely that the increased rate of AML/MDS seen with tamoxifen therapy was a result of coadministration of epirubicin and cyclophosphamide.
Secondary leukemia after adjuvant therapy for early breast cancer has also been investigated for regimens containing the anthracycline doxorubicin in combination with cyclophosphamide. Without direct access to the databases of these studies, interstudy comparisons are possible only from the published data, which are sometimes available only as abstracts, and using, as a common evaluation parameter, the leukemia frequency (number of AML/MDS cases divided by the number of patients treated in each study). Comparison of these frequencies with the results of the present analysis and between themselves is fraught with difficulty because of differences among and within other studies in the doses of doxorubicin and cyclophosphamide, the number of cycles, the duration of treatment, the additional treatment administered after completion of doxorubicin therapy, and the length of follow-up.
In the NSABP experience,3 doxorubicin has always been administered for four cycles only, to give a planned cumulative dose of 240 mg/m2. The frequency of AML/MDS increased in the three arms of the study in parallel with the cyclophosphamide dose-intensity. Including all three arms of the study, 43 AML/MDS cases in 8,563 treated patients were reported, corresponding to an overall frequency of 0.50%. Six-cycle regimens used by Albain et al37 and Haskell et al38 resulted in planned doxorubicin cumulative doses of 360 and 254 mg/m2, respectively, with a planned cyclophosphamide cumulative dose of 7,200 mg/m2 in both studies
the AML/MDS frequencies in the two six-cycle regimens were similar (0.77% and 0.81%, respectively) but higher than the frequency in the NSABP study (0.50%), in which the mean planned cumulative dose of cyclophosphamide was lower (4,500 mg/m2).
A risk evaluation of secondary leukemia in a number of adjuvant studies was carried out at the M.D. Anderson Cancer Center, and Diamandidou et al32 published results for the combination of doxorubicin plus cyclophosphamide in 1996. In 1,474 patients treated with a variable number of cycles and doses of doxorubicin and cyclophosphamide, an overall frequency of 0.95% was reported. The highest AML/MDS frequency (1.32%) was reported in a subgroup of patients treated with 24 cycles of chemotherapy, with a planned cumulative cyclophosphamide dose greater than 14,000 mg/m2 and a planned doxorubicin cumulative dose of 300 mg/m2. Finally, in a French adjuvant study, in which a low dose of doxorubicin per cycle (30 mg/m2) was combined with cyclophosphamide (1,200 mg/m2/cycle), fluorouracil, and vincristine, no cases of secondary leukemia were reported after a follow-up of 15 years.39
All these results are consistent with the suggested relationship between the risk of AML/MDS and increasing doses of doxorubicin and cyclophosphamide in combination. Overall, in these published studies, 91 AML/MDS cases were reported among 14,869 patients treated with various doxorubicin combinations, with an overall frequency of 0.61% and ranging from 0% to 1.32% according to doxorubicin study arm. In comparison, the overall frequency in published studies6,12,13,15,16,19,20 included in the Epirubicin Monitoring Plan was 0.49% (25 cases of AML/MDS in 5,085 patients treated with epirubicin in various doses and combinations), ranging from 0% to 3.58% (SBG9401 arm I: tailored FEC with dosing escalated to tolerance). These comparisons suggest that doxorubicin and epirubicin regimens normally used for adjuvant treatment of early breast cancer carry similar risks of secondary leukemia.
Anthracyclines and anthraquinones are biochemically related compounds, and therefore, it is not surprising that the risk of secondary leukemia was found by Chaplain et al40 to increase with cumulative doses of mitoxantrone. The 4-year cumulative rate of AML/MDS was reported to be 0.63% for mitoxantrone cumulative doses lower or equal to 12 mg/m2 and 3.54% for cumulative doses equal to or greater than 13 mg/m2.
For both epirubicin and cyclophosphamide, the analyses of cumulative probability in relation to cumulative dose indicate a low risk of AML/MDS at the standard cumulative doses used in recommended regimens. There were only 11 cases (0.23%) of AML/MDS in the 4,758 patients who were known to have received cumulative doses of epirubicin 720 mg/m2 and cyclophosphamide 6,300 mg/m2, in contrast to 15 cases of AML/MDS (1.2%) in the 1,263 patients known to have received cumulative doses (epirubicin > 720 mg/m2 and/or cyclophosphamide > 6,300 mg/m2) higher than those in currently recommended regimens. Although analysis of survival benefit was intentionally not a part of the Monitoring Plan, review of the available published reports from studies included in the Plan confirms the therapeutic advantage for patients treated with these agents, in both recommended and intensive regimens. The overall survival was significantly prolonged in the arms containing the intensive regimens of studies MA513 and GFEA05,9 whereas a significant advantage in relapse-free survival was reported for the tailored FEC regimen in the SBG9401 study19 and for the higher-dose regimen in the BE87001 study.6 When comparing the risk of AML/MDS between studies, it should be noted that the 95% CIs for studies MA5 and SBG9401 overlap.
Thus, systematic follow-up of patients with early breast cancer treated in controlled clinical trials shows that the cumulative probability of AML/MDS may be correlated with the cumulative doses of epirubicin and cyclophosphamide. Risk is low at low cumulative doses of epirubicin and cyclophosphamide but increases sharply with increasing cumulative doses. The therapeutic benefit from adjuvant therapy with both agents in early breast cancer has been convincingly demonstrated and is of a different order of magnitude from the risk of secondary leukemia. In clinical practice, the described regimens have a strong benefit to risk ratio, and patients and clinicians can be reassured that the risk of leukemia is likely to be low if the cumulative doses of these regimens are not exceeded. Clinical trials attempting to improve therapeutic benefit by dose escalation need to take the resulting increased risk of leukemia into account in assessing potential benefit and risk.
Appendix
The following Epirubicin Plan Investigators participated in this study: GFEA01, GFEA02, GFEA03, GFEA04, GFEA05, GFEA06, GFEA07, GFEA08, GFEA09, GETIS02: Jacques Bonneterre (Centre Oscar Lambret, Lille), Pierre Fargeot (Centre Georges Franois Leclerc, Dijon), Pierre Fumoleau (Centre Rene Gauducheau, Nantes), Michel Hery (Monaco), Pierre Kerbrat (Centre Eugène Marquis, Rennes), Henri Roche (Institut Claudius Regaud, Toulouse), Moise Namer (Nice), Martine Bardonnet, Isabelle Chapelle-Marcillac (Pharmacia/Pfizer). EORTC10921, EORTC10902, BE 87001: Martine Piccart, Rosa Giuliani (Institut Jules Bordet, Brussels), Patrick Therasse (European Organisation for Research and Treatment of Cancer), Jean Pierre Lobelle (Pharmacia/Pfizer). NCIC/MA5: Mark Levine, Lois Shepherd (National Cancer Institute of Canada Clinical Trials Group, Kingston). ICCGC/4/87, ICCGC/2/84, ICCGC/6/89, ICCGC/9/91: Judith Bliss (Institute of Cancer Research, Sutton), R. Charles Coombes (Hammersmith Hospital, London), Michel Marty (Institut Gustave Roussy, Villejuif), Jacques Wils (Laurentius Hospital, Roermond), Jan Smeets (Pharmacia/Pfizer). SBG9401: Jonas Bergh, Annika Folin (Karolinska Hospital, Stockholm), Jonas Nilsson (University Hospital, Uppsala). Analysis and Writing: Claudio Praga (Former Senior Director, Pharmacia Global Drug Safety Europe, retired), Bruno Cesana (Università degli Studi, Brescia), Francesco Onida (Ospedale Maggiore di Milano, Istituto di Ricovero e Cura a Carattere Scientifico, Milan), David Rogers (Therapeutic Area Director, Oncology, Safety Surveillance and Reporting, Pharmacia, Gruppo Pfizer).
Authors' Disclosures of Potential Conflicts of Interest
The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Employment: David Rogers, Pfizer. Consultant/Advisory Role: Bruno Cesana, Pfizer
R. Charles Coombes, Argenta
Pierre Fumoleau, Pfizer
Francesco Onida, Pfizer. Stock Ownership: Claudio Praga, Pfizer
David Rogers, Pfizer. Honoraria: Jonas Bergh, Pfizer
R. Charles Coombes, Pfizer. Research Funding: Claudio Praga, Pfizer
Jonas Bergh, Pfizer
Judith Bliss, Pfizer
Bruno Cesana, Pfizer
R. Charles Coombes, AstraZeneca, Pfizer. For a detailed description of these categories, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration and Disclosures of Potential Conflicts of Interest found in Information for Contributors in the front of each issue.
Acknowledgment
We thank Maria Grazia Zurlo, Benjamin Winograd, Jan Smeets, Jean-Pierre Lobelle, Andrea Ashford, Rebecca Rosenstein, and Susan Pitman Lowenthal (all Pharmacia/Pfizer employees) for helpful advice and suggestions.
NOTES
Supported by Pharmacia, Gruppo Pfizer (previously Pharmacia & Upjohn).
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Smith MA, Ungerleider RS: Therapy-related acute myeloid leukemia following treatment with epipodophyllotoxins: Estimating the risks. Med Pediatr Oncol 23:86-98, 1994
Smith MA, Rubinstein L, Anderson JR, et al: Secondary leukemia or myelodysplastic syndrome after treatment with epipodophyllotoxins. J Clin Oncol 17:569-577, 1999
Smith RE, Bryant J, DeCillis A, et al: Acute myeloid leukemia and myelodysplastic syndrome following doxorubicin-cyclophosphamide adjuvant therapy for operable breast cancer: The National Surgical Adjuvant Breast and Bowel Project Experience. J Clin Oncol 21:1195-1204, 2003
Pedersen-Bjergaard J, Sigsgaard TC, Nielsen D, et al: Acute monocytic or myelomonocytic leukemia with balanced chromosome translocations to band 11q23 after therapy with 4-epi-doxorubicin and cisplatin or cyclophosphamide for breast cancer. J Clin Oncol 10:1444-1515, 1992
Riggi M, Riva A: Therapy-related leukemia: What is the role of 4-epi-doxorubicin J Clin Oncol 11:1430-1431, 1993
Piccart MJ, Di Leo A, Beauduin M, et al: Phase III trial comparing two dose levels of epirubicin combined with cyclophosphamide with cyclophosphamide, methotrexate and fluorouracil in node-positive breast cancer. J Clin Oncol 19:3103-3110, 2001
Bernard-Marty C, Mano M, Paesmans M, et al: Second malignancies following adjuvant chemotherapy: 6-Year results from a Belgian randomized study comparing cyclophosphamide, methotrexate and 5-fluorouracil (CMF) with an anthracycline-based regimen in adjuvant treatment of node-positive breast cancer patients. Ann Oncol 14:693-698, 2003
Fumoleau P, Kerbrat P, Romestaing P, et al: Randomized trial comparing six versus three cycles of epirubicin-based adjuvant chemotherapy in premenopausal, node-positive breast cancer patients: 10-Year follow-up results of the French Adjuvant Study Group 01 trial. J Clin Oncol 21:298-305, 2003
French Adjuvant Study Group: Benefit of a high-dose epirubicin regimen in adjuvant chemotherapy for node-positive breast cancer patients with poor prognostic factors: 5-Year follow-up results of French Adjuvant Study Group 05 Randomized Trial. J Clin Oncol 19:602-611, 2001
Fargeot P, Roche H, Bonneterre J, et al: Disease-free survival advantage of weekly epirubicin plus tamoxifen vs tamoxifen alone as adjuvant treatment of operable, node positive elderly breast cancer: 5-year follow-up results of French Adjuvant Study Group, FASG-08 trial. Proc Am Soc Clin Oncol 21:37a, 2002 (abstr 145)
Kerbrat P, Roche H, Bonneterre J, et al: Can we substitute an epirubicin-vinorelbine regimen to 6 cycles of FEC 100 as adjuvant chemotherapy of operable, node positive breast cancer patients 5-Year follow-up results of French Adjuvant Study Group, FASG-09. Proc Am Soc Clin Oncol 21:48a, 2002 (abstr 189)
van der Hage JA, van de Velde CJH, Julien JP, et al: Preoperative chemotherapy in primary operable breast cancer: Results from the European Organization for Research and Treatment of Cancer Trial 10902. J Clin Oncol 19:4224-4237, 2001
Levine MN, Bramwell VH, Pritchard KI, et al: Randomized trial of intensive cyclophosphamide, epirubicin and fluorouracil chemotherapy compared with cyclophosphamide, methotrexate and fluorouracil in premenopausal women with node-positive breast cancer. J Clin Oncol 16:2651-2658, 1998
Crump M, Tu D, Shepherd L, et al: Risk of acute leukemia following epirubicin-base adjuvant chemotherapy: A report from the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol 21:3066-3071, 2003
Wils JA, Bliss JM, Marty M, et al: Epirubicin plus tamoxifen versus tamoxifen alone in node-positive postmenopausal patients with breast cancer: A randomized trial of the International Collaborative Cancer Group. J Clin Oncol 17:1988-1998, 1999
Coombes RC, Bliss JM, Wils J, et al: Adjuvant cyclophosphamide, methotrexate, 5-fluorouracil (CMF) versus 5-fluorouracil, epirubicin, cyclophosphamide (FEC) chemotherapy in premenopausal women with axillary node positive operable breast cancer: Results of a randomised trial—The International Collaborative Cancer Group. J Clin Oncol 14:35-45, 1996
Marty M, Hall E, Wils J, et al: Evaluation of tolerability of CMF versus FEC in a randomised trial in node negative poor risk primary breast cancer patients. Proc Am Soc Clin Oncol 21:69a, 2002 (abstr 273)
Bliss JM, Wils J, Marty M, et al: Evaluation of the tolerability of FE50C versus FE75C in a prospective randomized trial in adjuvant breast cancer patients. Proc Am Soc Clin Oncol 21:51b, 2002 (abstr 2017)
Bergh J, Wiklund T, Erikstein B, et al: Tailored fluorouracil, epirubicin, and cyclophosphamide compared with marrow-supported high-dose chemotherapy as adjuvant treatment for high-risk breast cancer: A randomized trial. Lancet 356:1384-1391, 2000
Campone M, Roche H, Kerbrat P, et al: Secondary leukemias after epirubicin-based adjuvant chemotherapy for operable breast cancer patients: 15 years experience of French Adjuvant Study Group. Breast Cancer Res Treat 76:S157, 2002
Clayton D, Hills M: Statistical Models in Epidemiology. Oxford, United Kingdom, Oxford Science Publications, 1993, pp 96-109, 122-132
Stokes ME, Davis CS, Koch GG: Categorical Data Analysis Using the SAS System. Cary, NC, SAS Institute Inc, 1995, pp 463-481
Peto R, Pike MC, Armitage P, et al: Design and analysis of randomised clinical trials requiring prolonged observation of each patient: I. Introduction and design. Br J Cancer 34:585-612, 1976
Bennet JM, Catovsky D, Daniel MT, et al: Proposed revised criteria for the classification of the acute myeloid leukemia: A report of the French-American-British (FAB) Cooperative Group. Ann Intern Med 103:620-625, 1985
Pedersen-Bjergaard J, Philip P: Two different classes of therapy-related and de-novo acute myeloid leukemia Cancer Genet Cytogenet 55:119-124, 1991
Quesnel B, Kantarjian H, Pedersen-Bjergaard J, et al: Therapy-related acute myeloid leukemia with t(8;21), inv(16) and t(8;16): A report on 25 cases and review of the literature. J Clin Oncol 11:2370-2379, 1993
Hoffmann L, Mo P, Pedersen-Bjergaard J, et al: Therapy-related acute promyelocytic leukemia with t(15;17)(q22;q12) following chemotherapy with drugs targeting DNA topoisomerase-II: A report of two cases and a review of the literature. Ann Oncol 6:781-788, 1995
Linassier C, Barin C, Calais G, et al: Early secondary acute myelogenous leukemia in breast cancer patients after treatment with mitoxantrone, cyclophosphamide, fluorouracil and radiation therapy. Ann Oncol 11:1289-1294, 2000
Valagussa P, Moliterni A, Terenziani M, et al: Second malignancies following CMF-based adjuvant chemotherapy in resectable breast cancer. Ann Oncol 5:803-808, 1994
Tallman MS, Gray R, Bennet JM, et al: Leukemogenic potential of adjuvant chemotherapy for early-stage breast cancer: The Eastern Cooperative Oncology Group experience. J Clin Oncol 13:1557-1563, 1995
Curtis RE, Boice JD Jr, Stovall M, et al: Risk of leukemia after chemotherapy and radiation treatment for breast cancer. N Engl J Med 326:1745-1751, 1992
Diamandidou E, Buzdar AU, Smith TL, et al: Treatment-related leukemia in breast cancer patients treated with fluorouracil-doxorubicin-cyclophosphamide combination adjuvant chemotherapy: The University of Texas M.D. Anderson Cancer Center experience. J Clin Oncol 14:2722-2730, 1996
Brodsky RA, Bedi A, Jones RJ: Are growth factors leukemogenic Leukemia 10:175-177, 1996
Morgan RW: Risk of endometrial cancer after tamoxifen treatment. Oncology 11:25-33, 1997
Matsuyama Y, Tominaga T, Nomura Y, et al: Second cancers after adjuvant tamoxifen therapy for breast cancer in Japan. Ann Oncol 11:1537-1543, 2000
Early Breast Cancer Trialists' Collaborative Group: Tamoxifen for early breast cancer: An overview of the randomized trials. Lancet 351:1451-1467, 1998
Albain K, Green S, Ravdin P, et al: Overall survival after cyclophosphamide, Adriamycin, 5-FU, and tamoxifen (CAFT) is superior to T alone in postmenopausal, receptor(+), node(+) breast cancer: New Findings from Phase III South West Oncology Group Intergroup Trial S8814 (INT-0100). Proc Am Soc Clin Oncol 20:24a, 2001 (abstr 94)
Haskell CM, Green SJ, Sledge GW, et al: Phase III comparison of adjuvant high-dose doxorubicin plus cyclophosphamide (AC) versus sequential doxorubicin followed by cyclophosphamide (AC) in breast cancer patients with 0-3 positive nodes (INTERGROUP 0137). Proc Am Soc Clin Oncol 21:36a, 2002 (abstr 142)
Delaloge S, PalmaM D, Guerin S, et al: Adjuvant doxorubicin, vincristine, cyclophosphamide, fluorouracil (AVCF) for breast cancer (BC): No treatment related leukemia at 13 year follow-up. Proc Am Soc Clin Oncol 17:117a, 1998 (abstr 449)
Chaplain G, Milan C, Sgro C, et al: Increased risk of acute leukemia after adjuvant chemotherapy for breast cancer: A population-based study. J Clin Oncol 18:2836-2842, 2000(Claudio Praga, Jonas Berg)
Centre Georges-Franois Leclerc, Dijon
Centre Eugène Marquis, Rennes, France
Karolinska Hospital, Stockholm
University Hospital, Uppsala, Sweden
Institute of Cancer Research, Sutton
Cancer Research United Kingdom Laboratories, Imperial College, London, United Kingdom
Institut Jules Bordet and European Organisation for Research and Treatment of Cancer Data Center, Brussels, Belgium
National Cancer Institute of Canada Clinical Trials Group, Kingston, Ontario, Canada
Laurentius Hospital, Roermond, the Netherlands
Centre Hospitalier Princesse Grace, Monaco
Università degli Studi, Brescia
Ospedale Maggiore di Milano, Istituto di Ricovero e Cura a Carattere Scientifico
Safety Surveillance and Reporting, Pharmacia, Gruppo Pfizer, Milan, Italy
ABSTRACT
PURPOSE: We reviewed follow-up of patients treated in 19 randomized trials of adjuvant epirubicin in early breast cancer to determine incidence, risk, and risk factors for subsequent acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS).
PATIENTS AND METHODS: The patients (N = 9,796) were observed from the start of adjuvant treatment (53,080 patient-years). Cases of AML or MDS (AML/MDS) were reported, with disease characteristics. Incidence and cumulative risk were compared for possible risk factors, for assigned regimens, and for administered cumulative doses of epirubicin and cyclophosphamide.
RESULTS: In 7,110 patients treated with epirubicin-containing regimens (92% of whom also received cyclophosphamide), 8-year cumulative probability of AML/MDS was 0.55% (95% CI, 0.33% to 0.78%). The risk of developing AML/MDS increased in relation to planned epirubicin dose per cycle, planned epirubicin dose-intensity, and administered cumulative doses of epirubicin and cyclophosphamide. Patients with administered cumulative doses of both epirubicin and cyclophosphamide not exceeding those used in standard regimens ( 720 mg/m2 and 6,300 mg/m2, respectively) had an 8-year cumulative probability of developing AML/MDS of 0.37% (95% CI, 0.13% to 0.61%) compared with 4.97% (95% CI, 2.06% to 7.87%) for patients administered higher cumulative doses of both epirubicin and cyclophosphamide.
CONCLUSION: Patients treated with standard cumulative doses of adjuvant epirubicin ( 720 mg/m2) and cyclophosphamide ( 6,300 mg/m2) for early breast cancer have a lower probability of secondary leukemia than patients treated with higher cumulative doses. Increased risk of secondary leukemia must be considered when assessing the potential benefit to risk ratio of higher than standard doses.
INTRODUCTION
Over the last 20 years, adjuvant therapy for early breast cancer with DNA-targeted antiproliferative drugs has substantially increased the number of long-term survivors. However, a small proportion of these survivors subsequently develop acute myelogenous leukemia (AML), with or without a preleukemic myelodysplastic syndrome (MDS). When long-term survival is expected, estimation of the risk of secondary AML/MDS after chemotherapy is important to determine the risk to benefit ratio of treatment
with this aim, patients treated with epipodophyllotoxins for pediatric leukemias1 and solid tumors2 and with the combination of doxorubicin and high-dose cyclophosphamide for breast cancer3 have been observed in monitoring plans. These studies have raised concerns over the roles of topoisomerase II inhibitors, alkylating agents, and growth factors in the development of new primary malignancies.
Adjuvant chemotherapy regimens containing epirubicin, a DNA topoisomerase II–targeted anthracycline, are widely used in the treatment of breast cancer. In 1992, a Danish survey reported five cases of AML in 360 patients with advanced breast cancer treated with a regimen of epirubicin and either cyclophosphamide or cisplatin or after previous exposure to other alkylating agents (melphalan or lomustine).4 Cases of secondary leukemia were then reported in patients with early breast cancer treated with epirubicin.5 In response to these reports, the manufacturer (then Pharmacia & Upjohn, now Pfizer
New York, NY) extended its routine pharmacovigilance to include a monitoring plan for secondary leukemia in patients treated with epirubicin-containing regimens in controlled clinical trials of adjuvant therapy for early breast cancer.
PATIENTS AND METHODS
Study Population
The Monitoring Plan included all patients entered onto 19 randomized clinical trials, which were all the research ethics committee–approved trials for adjuvant treatment of early breast cancer known to the manufacturer that included epirubicin in at least one arm (Table 1). 6-19 All the trials defined early breast cancer as node positive or negative, not previously treated, and without distant metastases. The plan was activated in June 1995. The cutoff date for this analysis was December 31, 2001, when all but one trial (GFEA08) had completed accrual. Results of eight trials have been published,6,8,9,12-14,16,19 and preliminary results of other trials have been presented as abstracts.10,11,17,18,20 The trials included patients enrolled as early as 1984, so the analysis covers more than 15 years of clinical experience.
The trials contained 44 different treatment arms. These arms were pooled into the following five groups: chemotherapy including epirubicin at an assigned dose per cycle of 100 mg/m2
chemotherapy including epirubicin at an assigned dose per cycle of more than 100 mg/m2
chemotherapy not including epirubicin
hormone therapy without chemotherapy
no chemotherapy or hormone therapy. The cutoff point of 100 mg/m2/cycle for assigned epirubicin dose was derived from the fluorouracil, epirubicin, and cyclophosphamide (FEC) FEC100 regimen commonly used in Europe.9
Data Management and Statistical Analysis
Patient characteristics, treatment, and follow-up were extracted from the original study databases and merged into a SAS database (SAS Institute, Cary, NC). All cases of leukemia (any type) or MDS were reported together with disease characteristics, treatment (within and outside the original protocol), and follow-up.
Crude rates for AML/MDS (number of AML/MDS cases divided by number of patient-years [PY] at risk) with 95% CIs were compared for the five groups using the Score test.21 The hazard function for AML/MDS with 95% CIs was calculated for all epirubicin-treated patients, using the life-table method, for each 12-month period for up to 96 months of follow-up, when at least 20% of the patients remained at risk. Median cumulative doses of epirubicin and cyclophosphamide were compared between granulocyte colony-stimulating factor (G-CSF) –treated or untreated patients and tamoxifen-treated or untreated patients using the Mann-Whitney U test.
The rates of incidence of AML/MDS were modeled by multivariate backward Poisson regression models based on classes of patients according to age ( 49 or > 49 years), assigned cumulative doses of epirubicin and cyclophosphamide, and whether they received radiotherapy, were mandated to receive G-CSF, or actually received tamoxifen, together with the corresponding interactions.22
Kaplan-Meier cumulative probability estimates for AML/MDS with time were calculated for all epirubicin-treated patients. No allowance was made for competing risks on the grounds that there were no true competing risks for the AML/MDS event and that, for patient and physician, the most relevant risk was that of developing AML/MDS if the patient was a survivor. Because both the Kaplan-Meier and the direct estimates of cumulative incidence are small, any bias in the Kaplan-Meier estimate resulting from censoring is small and has negligible clinical implications. Cumulative probabilities of AML/MDS over the follow-up time were calculated for subgroups of patients according to planned epirubicin dose per cycle ( 100 v > 100 mg/m2/cycle), planned dose per week ( 33.3 v > 33.3 mg/m2/wk), and administered cumulative dose ( 720 v > 720 mg/m2) across all study arms containing epirubicin. Survival curves were censored for death and loss to follow-up to give an estimate of risk in the surviving population and compared for the subgroups using the log-rank test.23 The Cox proportional hazards model was used for assessing the independent role of each subgroup risk factor. Kaplan-Meier cumulative probability estimates for AML/MDS risk were calculated for the administered cumulative doses (mg/m2) of epirubicin and cyclophosphamide.
RESULTS
Patients
The population at risk, which was defined as all randomized patients for whom follow-up information was available, was 9,796 (97% of the 10,111 patients randomized in the trials). The distribution of follow-up across the clinical trials is shown in Table 1. The 9,796 assessable patients contributed 53,080.7 PY of follow-up before death or last contact. As of December 31, 2001, 2,542 assessable patients were reported to have died
cumulative survival probability was 78.2% (95% CI, 77.2% to 79.1%) at 5 years, 67.5% (95% CI, 66.3% to 68.7%) at 8 years, and 61.5% (95% CI, 60.0% to 63.0%) at 10 years.
Events
Up to December 31, 2001, 30 (0.306%) of the total population of 9,796 patients had presented with AML/MDS. In addition, three patients presented with acute lymphoblastic leukemia, one presented with hairy cell leukemia, and one presented with chronic lymphocytic leukemia (Table 1). Secondary leukemias occurred in 12 trials, with the number of cases ranging from none to four per trial except in NCIC/MA5 (n = 6) and SBG9401 (n = 10).
Of the 7,110 patients randomly assigned to epirubicin-containing regimens, 28 (0.394%) developed AML/MDS. Only one (patient 27, EORTC10921) of these patients experienced a breast cancer relapse before presenting with AML/MDS
the relapse had been treated with taxanes and fluorouracil.
One (0.07%) of the 1,427 patients randomly assigned to chemotherapy regimens not including epirubicin developed AML/MDS. This patient (patient 25, NCIC/MA5 trial), who was treated initially with cyclophosphamide, methotrexate, and fluorouracil (CMF), had a breast cancer relapse (treated only with radiotherapy and tamoxifen) 7 months before presenting with leukemia.
One (0.11%) of the 903 patients randomly assigned to hormone therapy without chemotherapy developed AML/MDS. This patient (patient 16, GFEA02 trial), who was treated initially with tamoxifen, had a breast cancer relapse, which was treated with mitoxantrone, cyclophosphamide, fluorouracil, epirubicin, vinorelbine, and mitomycin, 3 years before presenting with leukemia.
Individual characteristics of the patients are listed in Table 2, where the AML cases are listed according the French-American-British (FAB) morphologic criteria.24 The most frequent subtype of AML was M4 (n = 10), followed by M2 (n = 6), M5 (n = 4), and M3 (n = 1)
in four patients, FAB classification was unknown. Five patients presented with MDS, and a preleukemic myelodysplastic phase was suspected in five additional patients who eventually progressed to AML (patients 2, 5, 10, 24, and 25).
The time elapsed from the start of adjuvant treatment to the diagnosis of AML/MDS ranged from 8 to 126 months, with a median of 33 months. Latency was shorter (median, 17.5 months
range 8 to 81 months) in the 14 M4/M5 patients than in the six M2 patients (median, 44.5 months
range, 19 to 126 months). Of the 29 patients with AML/MDS who had received epirubicin and/or cyclophosphamide per protocol, 13 developed AML/MDS (nine of these patients developed M4/M5) within the first 2 years from first chemotherapy
these patients were younger (median age, 46 years) than the 16 patients (median age, 51 years) who developed AML/MDS more than 2 years after first chemotherapy.
Cytogenetic analyses were available for 24 AML/MDS patients and one acute lymphoblastic leukemia (ALL) patient. Of the AML patients, three had normal karyotype, 12 had balanced chromosomal translocations with or without a single chromosome abnormality, seven had abnormalities of only one or two chromosomes (deletion 5q
monosomy/abnormality 7, trisomy 8
inversion 16), and two had complex karyotypes with three or more chromosome abnormalities.
As of December 31, 2001, 12 AML/MDS patients had survived 1 year or more from leukemia diagnosis
six of these patients were still alive at the last follow-up, at 13, 14, 17, 34, 75, and 78 months from diagnosis. Five patients survived less than 1 month from leukemia diagnosis. The two M4/M5 patients with the longest survival from leukemia diagnosis (14 months and > 34 months) had received only epirubicin and tamoxifen as adjuvant treatment for their early breast cancer.
Risk Factors Evaluation: Incidence Rates and Multivariate Analysis
The median reported age for the 9,796 assessable patients at start of adjuvant treatment was 49 years (range, 23 to 93 years)
age was not reported for 73 patients. Incidence rates for AML/MDS were similar in the age groups of 49 or older than 49 years (Table 3).
After surgery, 7,216 patients (81%) received local radiotherapy (details of dose, field, and modality were not requested), whereas 1,664 (19%) patients did not receive radiotherapy
for 916 patients, no information on radiotherapy was reported. Twenty-seven cases of AML/MDS were reported in patients who received radiotherapy, and three cases were reported in patients who did not receive radiotherapy
incidence rates were 0.666 and 0.329/1,000 PY, respectively, which do not differ significantly (P = .2354, Table 3). No cases were reported in the patients without information on radiotherapy.
Treatment with G-CSF was required by protocol in four study arms with intensive chemotherapy (GETIS02, arm I
EORTC10921, arm I
SBG9401, arms I and II). Although in some of the other studies G-CSF was to be administered as needed, G-CSF administration other than that required by protocol was rarely reported. Thirteen cases of AML/MDS occurred in the 796 patients mandated to receive G-CSF by protocol, giving a significantly (P = .0001, Table 3) higher incidence rate (5.453/1,000 PY) versus patients without G-CSF administration (0.336/1,000 PY). The patients mandated to receive G-CSF according to protocol received median cumulative doses of 634 mg/m2 of epirubicin and 8,267 mg/m2 of cyclophosphamide, both of which were significantly (P = .0001) higher than the corresponding median cumulative doses (304 mg/m2 and 4,547 mg/m2, respectively) in patients not mandated to receive G-CSF.
In the 3,230 patients who actually received tamoxifen, regardless of their study protocol (median age, 54 years
range, 22 to 93 years), there were 19 cases of AML/MDS. Eleven cases occurred in the 6,567 patients who did not receive tamoxifen (median age, 46 years
range, 23 to 85 years)
the corresponding AML/MDS rates were significantly different (P = .0001
1.260 and 0.289/1,000 PY, respectively
Table 3). Median epirubicin and cyclophosphamide cumulative doses were 516 and 4,775 mg/m2, respectively, in the group of tamoxifen-treated patients and were significantly higher (P = .0001) than the corresponding doses (300 and 4,684 mg/m2, respectively) in the group of patients who did not receive tamoxifen.
At Poisson regression analysis, assigned cumulative doses of epirubicin and cyclophosphamide were found to be the only independent risk factors. Age, reported radiotherapy, mandated use of G-CSF, and reported treatment with tamoxifen were not found to be independent risk factors
patients who received these treatments also tended to receive high cumulative doses of epirubicin and cyclophosphamide.
AML/MDS Incidence Rates and Hazard Function
The distribution of assessable patients and follow-up across the five assigned treatment groups is shown in Table 4, which also lists the average patient age at start of adjuvant treatment, the percentage of patients who were mandated to receive G-CSF according to the original study protocol, the number of reported AML/MDS cases, and the rate/1,000 PY of the AML/MDS event together with its 8-year cumulative probability. There were no significant differences in incidence rates between patients treated with epirubicin at a planned dose 100 mg/m2, patients receiving chemotherapy not including epirubicin, and patients treated with hormone therapy. The incidence rate was 3.620/1,000 PY (95% CI, 2.260 to 5.798/1,000 PY) in patients treated with epirubicin administered in various combinations at a planned dose per cycle of more than 100 mg/m2, which was significantly (P = .0001) higher than in all other treatment groups (rates ranging from 0.117 to 0.345/1,000 PY).
The hazard function for AML/MDS events in all epirubicin-treated patients is shown in Figure 1. A first peak (0.15 x 10–3) occurs in the second year
thereafter, the hazard function seems to plateau, increasing slowly to a less pronounced peak (0.09 x 10–3) in the fifth year. The third peak (0.07 x 10–3) in the seventh year is a result of two cases occurring when only a limited number of patients remained at risk.
AML/MDS Cumulative Probability Over Time in Relation to Treatment
The cumulative probability of AML/MDS in the 7,110 patients treated with epirubicin was 0.27% (95% CI, 0.14% to 0.40%) at 3 years, 0.46% (95% CI, 0.28% to 0.65%) at 5 years, and 0.55% (95% CI, 0.33% to 0.78%) at 8 years. Figure 2 shows that the cumulative probability of AML/MDS was significantly higher in the patients assigned to the epirubicin regimens with a planned dose of more than 100 mg/m2/cycle (P = .0001) or more than 33.3 mg/m2/wk (P = .0001) or who were administered a cumulative dose of more than 720 mg/m2 (P = .0001) compared with patients assigned to epirubicin doses that were less than these amounts
the cumulative probabilities of AML/MDS were 2.3% v 0.27% at 8 years, 3.2% v 0.31% at 6 years (graph truncated at 6 years because, after this, too few patients remained at risk in the > 33.3 mg/m2/wk group), and 4.4% v 0.62% at 8 years, respectively. Only the epirubicin planned dose ( 100 or > 100 mg/m2/cycle) and the epirubicin administered cumulative dose ( 720 or > 720 mg/m2) remained as independent factors in the Cox model (both P = .0001)
the adjusted relative risks of having the event of AML/MDS were 5.76 (95% CI, 2.36 to 14.04) for the more than 100 mg/m2/cycle dose group versus the 100 mg/m2/cycle dose group and 6.80 (95% CI, 2.86 to 16.13) for the more than 720 mg/m2 dose group versus the 720 mg/m2 dose group. Epirubicin planned dose-intensity ( 33.3 v > 33.3 mg/m2/wk) was not found to be an independent factor.
AML/MDS Cumulative Probability in Relation to Administered Cumulative Doses of Epirubicin and Cyclophosphamide
Epirubicin and cyclophosphamide were administered in combination in most patients. Given that the Poisson distribution analysis demonstrated that the planned cumulative dose of each was an independent risk factor, the cumulative probability of AML/MDS was plotted in relation to the administered cumulative dose of each agent. The plot against administered epirubicin cumulative dose (Fig 3) shows a progressive increase in risk, with cumulative probabilities of 0.07% at 300 mg/m2, 0.34% at 600 mg/m2, and 1.53% at 720 mg/m2. Thereafter, the cumulative probability increases steeply, reaching 9.94% (95% CI, 4.46% to 15.41%) at 900 mg/m2. The AML/MDS cumulative probability also increases with administered cyclophosphamide oral or intravenous cumulative dose, with cumulative probabilities of 0.08% at 3,000 mg/m2, 0.29% at 6,000 mg/m2, and 0.98% at 9,000 mg/m2
thereafter, the cumulative probability increases steeply, reaching 6.4% (95% CI, 1.92% to 10.90%) at 12,000 mg/m2.
Table 5 shows the cumulative probability of AML/MDS over time for subgroups according to whether administered cumulative doses of either epirubicin or cyclophosphamide or both were within or above the standard doses. The CIs overlap for the two epirubicin subgroups (< 720 mg/m2 or > 720 mg/m2) when cyclophosphamide was administered within the standard cumulative dose (6,300 mg/m2), but when cyclophosphamide was administered above the standard cumulative dose, there was a significant increase in incidence at all three time points when epirubicin was administered above the standard cumulative dose. In the 4,760 patients whose administered cumulative doses of both epirubicin and cyclophosphamide did not exceed those used in standard regimens, the 8-year cumulative probability of developing AML/MDS was 0.37% (95% CI, 0.13% to 0.61%) compared with 4.97% (95% CI, 2.06% to 7. 87%) in the 261 patients who received higher cumulative doses of both epirubicin and cyclophosphamide.
DISCUSSION
The success of adjuvant chemotherapy in prolonging disease-free survival has resulted in the treatment of thousands of women each year. The risk of AML/MDS is an important consideration in a patient’s treatment decision. Secondary leukemia risk has been evaluated in many studies of commonly used alkylating agents and anthracyclines.
The Epirubicin Monitoring Plan is the most comprehensive analysis of secondary leukemia after treatment with adjuvant epirubicin completed to date, with 9,796 patients observed for a total of 53,080 PYs. The analysis confirms a small but significant risk of AML/MDS in patients treated with adjuvant epirubicin and cyclophosphamide for early breast cancer. In all of the 7,110 patients treated with epirubicin-containing regimens, the 8-year cumulative probability of AML/MDS was 0.55%. The risk of developing AML/MDS increased in relation to total dose of epirubicin and total dose of cyclophosphamide, with a steep increase at cumulative doses above those used in recommended regimens for adjuvant therapy (epirubicin up to 720 mg/m2 and cyclophosphamide up to 6,300 mg/m2). The highest risk (4.97% cumulative probability at 8 years) was seen in patients who had been administered greater than standard doses of both epirubicin and cyclophosphamide.
The monitoring plan covered a period of clinical experience extending over more than 15 years. During this period, the therapeutic approach to early breast cancer adjuvant treatment became more aggressive, with the epirubicin dose increasing from approximately 50 mg/m2 every 3 weeks in the earlier studies up to 120 mg/m2 every 3 or 4 weeks in the later studies. In the six-cycle regimens that are approved for adjuvant therapy by the US Food and Drug Administration, epirubicin (in combination with cyclophosphamide and fluorouracil) is administered at 100 mg/m2 on day 1 every 3 weeks or at 60 mg/m2 on day 1 and on day 8 every 4 weeks, for total cumulative doses of 600 and 720 mg/m2, respectively. To evaluate the effect of the epirubicin dose on the risk of AML/MDS, cutoff points for dose per cycle, dose per week, and total cumulative dose were chosen to correspond to upper limits of dosing in these regimens
the cutoff point for dose per cycle was chosen for the first of these regimens, which is the more widely used regimen in Europe.
The incidence rate (3.62/1,000 PY) in the group of patients treated with epirubicin administered at a planned dose per cycle of more than 100 mg/m2 was significantly (P = .001) higher than the incidence rate (0.345/1,000 PY) in the group of patients treated with epirubicin at a planned dose of 100 mg/m2, suggesting an appreciable dose effect. At a planned dose of 100 mg/m2, the low risk of secondary leukemia does not differ significantly from the risk in patients not treated with epirubicin or who received hormonal therapy only.
Given that the risk of AML/MDS is not constant over time, as demonstrated by the hazard function (Fig 1), the AML/MDS cumulative probabilities were calculated and compared. Significant epirubicin dose effects were seen when the groups treated with a high planned epirubicin dose per cycle (> 100 mg/m2) or per week (> 33.3 mg/m2) or with a high administered cumulative dose (> 720 mg/m2) were compared with the corresponding groups treated at a lower dose (Fig 2). However, in the Cox model analysis, only planned dose per cycle and administered cumulative dose remained as independent factors, indicating that, in this analysis, dose-intensity does not have a significant independent association with AML/MDS risk.
Calculation of the AML/MDS cumulative probability as a function of the administered cumulative dose of epirubicin provides a method for estimating the risk in the individual patient according to the total dose received, similar to the risk estimation for anthracycline cardiotoxicity. If the cumulative dose remains within the maximum recommended for this indication (720 mg/m2), the AML/MDS cumulative probability increases slowly, up to 1.53% at 720 mg/m2, whereas above this dose, the incidence increases rapidly to reach 9.94% at a cumulative dose of 900 mg/m2 (Fig 3).
The monitoring plan was not designed specifically to evaluate the role of cyclophosphamide because trial selection was determined by inclusion of epirubicin in a study arm and subgroup analysis was determined by epirubicin dose schedule. However, 92% of patients treated with epirubicin also received cyclophosphamide, and the multivariate analysis identified epirubicin and cyclophosphamide as the only independent determinants of the leukemia risk. Secondary leukemias seen in patients treated with alkylating agents, such as cyclophosphamide, typically differ from those seen with DNA topoisomerase II inhibitors, such as the anthracyclines. For example, a myelodysplastic preleukemic phase has been observed in patients treated with alkylating agents but is not common in patients treated with DNA topoisomerase II inhibitors.25 In addition, AML in patients treated with topoisomerase II inhibitors typically has an early onset and an M4/M5 subtype, whereas AML after alkylating agents has a later onset and a M1/M2 subtype. Chromosomal changes also differ
chromosomes 5 and 7 are typically altered, usually with unbalanced changes, in AML/MDS after treatment with alkylating agents, whereas abnormalities of chromosomes 11 and 21, especially the balanced translocations 11q23 and 21q22, are more typical of AML after treatment with topoisomerase II inhibitors.
The features typical of AML/MDS caused by alkylating agents were seen in a number of patients in the present study, suggesting a major contribution by cyclophosphamide to the development of AML/MDS. For instance, four patients with MDS for whom cytogenetic information was available had unbalanced abnormalities of chromosomes 5 or 7. The only AML patient who did not received epirubicin but who did receive a large cumulative dose of cyclophosphamide had a long preleukemic myelodysplastic phase and also had an unbalanced chromosome 5 deletion. Six patients had M2 FAB subtype
of these six patients, five had onset of disease more than 2 years after start of adjuvant therapy, although only one patient had a typical unbalanced chromosome abnormality (deletions of both chromosomes 5 and 7).
Consistent with a role of cyclophosphamide, the plot of AML/MDS probability against administered cumulative dose of cyclophosphamide (Fig 3) shows that, if the cumulative dose remains within the maximum recommended dose in the combination regimens with epirubicin (6,300 mg/m2), the AML/MDS cumulative probability increases slowly up to 0.33%, whereas, above this dose, the cumulative probability increases more rapidly to reach 6.4% at a cumulative dose of 12,000 mg/m2.
The majority of patients with AML responded to induction chemotherapy, and 11 patients with AML/MDS (MDS, n = 3
M2, n = 4
M3, n = 1
M4, n = 2
unclassified AML, n = 1) had survived more than 1 year from diagnosis at the date chosen for this analysis. In treatment-related AML after treatment with epipodophyllotoxins and anthracyclines, balanced translocations [t(8;21), t(15;17),] and inv(16) have been reported as favorable cytogenetic abnormalities, with response rates and survival similar to de novo adult AML.26,27 In contrast, for alkylating agent–related secondary leukemia, the results of chemotherapy have generally been disappointing.14 However, in the present survey, better survival was observed in the M2 subgroup (median latency, 44.5 months), two of whom showed t(8;21), than in the M4/M5 subgroup (median latency, 17.5 months). This is similar to the poor prognosis reported by Linassier et al28 in a group of 10 patients with AML (M1/M2, n = 3
M4/M5, n = 6
M3, n = 1) and early breast cancer treated with mitoxantrone, cyclophosphamide, and fluorouracil (latency, 16 months).
The effect of escalating doses of cyclophosphamide (600, 1,200, and 2,400 mg/m2/cycle for two or four cycles) associated with a fixed dose (60 mg/m2/cycle for four cycles) of another anthracycline, doxorubicin, on the risk of secondary leukemia in early breast cancer patients is being monitored in six National Surgical Adjuvant Breast and Bowel Project (NSABP) adjuvant trials. In a recent analysis of 8,563 patients,3 43 AML/MDS cases were reported
intensified cyclophosphamide dosing requiring filgrastim support was associated with an increased incidence of AML and MDS. Although a significant correlation between planned cyclophosphamide dose-intensity and the leukemia rate was reported, the authors did not find a significant relationship with planned cumulative dose of cyclophosphamide, even if there was a clear trend towards increased risk with the highest doses. However, the maximum cumulative dose of cyclophosphamide planned to be administered in the NSABP studies was 9,600 mg/m2, which, in the curve of AML/MDS probability against cyclophosphamide cumulative dose in the present analysis, is equivalent to a cumulative risk of 1.57% (95% CI, 0.19% to 2.95%), so that the range of incidence in terms of cyclophosphamide dose in the NSABP studies is likely to be narrow. Previous studies carried out in patients with early breast cancer suggested that the risk of leukemia after CMF regimens for adjuvant chemotherapy using cyclophosphamide at standard dose was not much higher than that in the general population.1,29,30
Age, radiotherapy, mandated use of G-CSF, and assigned treatment with tamoxifen were not found to be independent risk factors in the Poisson regression analysis. Lack of any age-related leukemia risk is at variance from the NSABP experience with doxorubicin, in which a significantly higher leukemia rate was observed in women older than 49 years. Other studies have reported an increased risk of AML/MDS after radiotherapy in patients treated with adjuvant chemotherapy for breast cancer.1,31,32 That no significant effect was seen in the present study is probably a result of the large proportion of patients who received radiotherapy, making the comparison unbalanced, and a result of the lack of homogeneity of radiotherapy administration between the individual studies.
Although 13 of the 30 AML/MDS patients in this study were mandated to receive G-CSF, the present study was not designed to determine whether G-CSF treatment was causally linked to AML/MDS in this population. The apparent effects of G-CSF may be explained by the high doses of epirubicin and cyclophosphamide received by the patients included in the treatment arms to which G-CSF was assigned. However, it has been suggested that growth factors may be leukemogenic,33 and a possible link between the use of supportive G-CSF in adjuvant doxorubicin treatment of breast cancer and secondary leukemia has been suggested. Future analyses might investigate the possible role of G-CSF cumulative dose and duration of exposure in development of leukemia in patients treated with epirubicin.
The AML/MDS rate was also significantly higher in the patient population who received tamoxifen, but most patients who received tamoxifen were also found to have received high doses of epirubicin and cyclophosphamide. An increase in estrogen-dependent secondary malignancies has been reported after treatment with tamoxifen,34,35 but the occurrence of secondary AML/MDS is an extremely rare event. Moreover, the extensive randomized direct comparison of tamoxifen with observation or placebo summarized in the Early Breast Cancer Trialists’ Collaborative Group overview36 has shown no excess of AML/MDS in the patients receiving tamoxifen. In the present study, one case (patient 16) of AML M4 occurred 6 years after enrollment in the group of 903 patients treated with hormone therapy (mostly tamoxifen) alone
this patient, however, had suffered a breast cancer relapse, which was treated with cytotoxic therapy, including epirubicin, cyclophosphamide, and mitoxantrone, 3 years before leukemia onset. Thus, it is likely that the increased rate of AML/MDS seen with tamoxifen therapy was a result of coadministration of epirubicin and cyclophosphamide.
Secondary leukemia after adjuvant therapy for early breast cancer has also been investigated for regimens containing the anthracycline doxorubicin in combination with cyclophosphamide. Without direct access to the databases of these studies, interstudy comparisons are possible only from the published data, which are sometimes available only as abstracts, and using, as a common evaluation parameter, the leukemia frequency (number of AML/MDS cases divided by the number of patients treated in each study). Comparison of these frequencies with the results of the present analysis and between themselves is fraught with difficulty because of differences among and within other studies in the doses of doxorubicin and cyclophosphamide, the number of cycles, the duration of treatment, the additional treatment administered after completion of doxorubicin therapy, and the length of follow-up.
In the NSABP experience,3 doxorubicin has always been administered for four cycles only, to give a planned cumulative dose of 240 mg/m2. The frequency of AML/MDS increased in the three arms of the study in parallel with the cyclophosphamide dose-intensity. Including all three arms of the study, 43 AML/MDS cases in 8,563 treated patients were reported, corresponding to an overall frequency of 0.50%. Six-cycle regimens used by Albain et al37 and Haskell et al38 resulted in planned doxorubicin cumulative doses of 360 and 254 mg/m2, respectively, with a planned cyclophosphamide cumulative dose of 7,200 mg/m2 in both studies
the AML/MDS frequencies in the two six-cycle regimens were similar (0.77% and 0.81%, respectively) but higher than the frequency in the NSABP study (0.50%), in which the mean planned cumulative dose of cyclophosphamide was lower (4,500 mg/m2).
A risk evaluation of secondary leukemia in a number of adjuvant studies was carried out at the M.D. Anderson Cancer Center, and Diamandidou et al32 published results for the combination of doxorubicin plus cyclophosphamide in 1996. In 1,474 patients treated with a variable number of cycles and doses of doxorubicin and cyclophosphamide, an overall frequency of 0.95% was reported. The highest AML/MDS frequency (1.32%) was reported in a subgroup of patients treated with 24 cycles of chemotherapy, with a planned cumulative cyclophosphamide dose greater than 14,000 mg/m2 and a planned doxorubicin cumulative dose of 300 mg/m2. Finally, in a French adjuvant study, in which a low dose of doxorubicin per cycle (30 mg/m2) was combined with cyclophosphamide (1,200 mg/m2/cycle), fluorouracil, and vincristine, no cases of secondary leukemia were reported after a follow-up of 15 years.39
All these results are consistent with the suggested relationship between the risk of AML/MDS and increasing doses of doxorubicin and cyclophosphamide in combination. Overall, in these published studies, 91 AML/MDS cases were reported among 14,869 patients treated with various doxorubicin combinations, with an overall frequency of 0.61% and ranging from 0% to 1.32% according to doxorubicin study arm. In comparison, the overall frequency in published studies6,12,13,15,16,19,20 included in the Epirubicin Monitoring Plan was 0.49% (25 cases of AML/MDS in 5,085 patients treated with epirubicin in various doses and combinations), ranging from 0% to 3.58% (SBG9401 arm I: tailored FEC with dosing escalated to tolerance). These comparisons suggest that doxorubicin and epirubicin regimens normally used for adjuvant treatment of early breast cancer carry similar risks of secondary leukemia.
Anthracyclines and anthraquinones are biochemically related compounds, and therefore, it is not surprising that the risk of secondary leukemia was found by Chaplain et al40 to increase with cumulative doses of mitoxantrone. The 4-year cumulative rate of AML/MDS was reported to be 0.63% for mitoxantrone cumulative doses lower or equal to 12 mg/m2 and 3.54% for cumulative doses equal to or greater than 13 mg/m2.
For both epirubicin and cyclophosphamide, the analyses of cumulative probability in relation to cumulative dose indicate a low risk of AML/MDS at the standard cumulative doses used in recommended regimens. There were only 11 cases (0.23%) of AML/MDS in the 4,758 patients who were known to have received cumulative doses of epirubicin 720 mg/m2 and cyclophosphamide 6,300 mg/m2, in contrast to 15 cases of AML/MDS (1.2%) in the 1,263 patients known to have received cumulative doses (epirubicin > 720 mg/m2 and/or cyclophosphamide > 6,300 mg/m2) higher than those in currently recommended regimens. Although analysis of survival benefit was intentionally not a part of the Monitoring Plan, review of the available published reports from studies included in the Plan confirms the therapeutic advantage for patients treated with these agents, in both recommended and intensive regimens. The overall survival was significantly prolonged in the arms containing the intensive regimens of studies MA513 and GFEA05,9 whereas a significant advantage in relapse-free survival was reported for the tailored FEC regimen in the SBG9401 study19 and for the higher-dose regimen in the BE87001 study.6 When comparing the risk of AML/MDS between studies, it should be noted that the 95% CIs for studies MA5 and SBG9401 overlap.
Thus, systematic follow-up of patients with early breast cancer treated in controlled clinical trials shows that the cumulative probability of AML/MDS may be correlated with the cumulative doses of epirubicin and cyclophosphamide. Risk is low at low cumulative doses of epirubicin and cyclophosphamide but increases sharply with increasing cumulative doses. The therapeutic benefit from adjuvant therapy with both agents in early breast cancer has been convincingly demonstrated and is of a different order of magnitude from the risk of secondary leukemia. In clinical practice, the described regimens have a strong benefit to risk ratio, and patients and clinicians can be reassured that the risk of leukemia is likely to be low if the cumulative doses of these regimens are not exceeded. Clinical trials attempting to improve therapeutic benefit by dose escalation need to take the resulting increased risk of leukemia into account in assessing potential benefit and risk.
Appendix
The following Epirubicin Plan Investigators participated in this study: GFEA01, GFEA02, GFEA03, GFEA04, GFEA05, GFEA06, GFEA07, GFEA08, GFEA09, GETIS02: Jacques Bonneterre (Centre Oscar Lambret, Lille), Pierre Fargeot (Centre Georges Franois Leclerc, Dijon), Pierre Fumoleau (Centre Rene Gauducheau, Nantes), Michel Hery (Monaco), Pierre Kerbrat (Centre Eugène Marquis, Rennes), Henri Roche (Institut Claudius Regaud, Toulouse), Moise Namer (Nice), Martine Bardonnet, Isabelle Chapelle-Marcillac (Pharmacia/Pfizer). EORTC10921, EORTC10902, BE 87001: Martine Piccart, Rosa Giuliani (Institut Jules Bordet, Brussels), Patrick Therasse (European Organisation for Research and Treatment of Cancer), Jean Pierre Lobelle (Pharmacia/Pfizer). NCIC/MA5: Mark Levine, Lois Shepherd (National Cancer Institute of Canada Clinical Trials Group, Kingston). ICCGC/4/87, ICCGC/2/84, ICCGC/6/89, ICCGC/9/91: Judith Bliss (Institute of Cancer Research, Sutton), R. Charles Coombes (Hammersmith Hospital, London), Michel Marty (Institut Gustave Roussy, Villejuif), Jacques Wils (Laurentius Hospital, Roermond), Jan Smeets (Pharmacia/Pfizer). SBG9401: Jonas Bergh, Annika Folin (Karolinska Hospital, Stockholm), Jonas Nilsson (University Hospital, Uppsala). Analysis and Writing: Claudio Praga (Former Senior Director, Pharmacia Global Drug Safety Europe, retired), Bruno Cesana (Università degli Studi, Brescia), Francesco Onida (Ospedale Maggiore di Milano, Istituto di Ricovero e Cura a Carattere Scientifico, Milan), David Rogers (Therapeutic Area Director, Oncology, Safety Surveillance and Reporting, Pharmacia, Gruppo Pfizer).
Authors' Disclosures of Potential Conflicts of Interest
The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Employment: David Rogers, Pfizer. Consultant/Advisory Role: Bruno Cesana, Pfizer
R. Charles Coombes, Argenta
Pierre Fumoleau, Pfizer
Francesco Onida, Pfizer. Stock Ownership: Claudio Praga, Pfizer
David Rogers, Pfizer. Honoraria: Jonas Bergh, Pfizer
R. Charles Coombes, Pfizer. Research Funding: Claudio Praga, Pfizer
Jonas Bergh, Pfizer
Judith Bliss, Pfizer
Bruno Cesana, Pfizer
R. Charles Coombes, AstraZeneca, Pfizer. For a detailed description of these categories, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration and Disclosures of Potential Conflicts of Interest found in Information for Contributors in the front of each issue.
Acknowledgment
We thank Maria Grazia Zurlo, Benjamin Winograd, Jan Smeets, Jean-Pierre Lobelle, Andrea Ashford, Rebecca Rosenstein, and Susan Pitman Lowenthal (all Pharmacia/Pfizer employees) for helpful advice and suggestions.
NOTES
Supported by Pharmacia, Gruppo Pfizer (previously Pharmacia & Upjohn).
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Smith MA, Ungerleider RS: Therapy-related acute myeloid leukemia following treatment with epipodophyllotoxins: Estimating the risks. Med Pediatr Oncol 23:86-98, 1994
Smith MA, Rubinstein L, Anderson JR, et al: Secondary leukemia or myelodysplastic syndrome after treatment with epipodophyllotoxins. J Clin Oncol 17:569-577, 1999
Smith RE, Bryant J, DeCillis A, et al: Acute myeloid leukemia and myelodysplastic syndrome following doxorubicin-cyclophosphamide adjuvant therapy for operable breast cancer: The National Surgical Adjuvant Breast and Bowel Project Experience. J Clin Oncol 21:1195-1204, 2003
Pedersen-Bjergaard J, Sigsgaard TC, Nielsen D, et al: Acute monocytic or myelomonocytic leukemia with balanced chromosome translocations to band 11q23 after therapy with 4-epi-doxorubicin and cisplatin or cyclophosphamide for breast cancer. J Clin Oncol 10:1444-1515, 1992
Riggi M, Riva A: Therapy-related leukemia: What is the role of 4-epi-doxorubicin J Clin Oncol 11:1430-1431, 1993
Piccart MJ, Di Leo A, Beauduin M, et al: Phase III trial comparing two dose levels of epirubicin combined with cyclophosphamide with cyclophosphamide, methotrexate and fluorouracil in node-positive breast cancer. J Clin Oncol 19:3103-3110, 2001
Bernard-Marty C, Mano M, Paesmans M, et al: Second malignancies following adjuvant chemotherapy: 6-Year results from a Belgian randomized study comparing cyclophosphamide, methotrexate and 5-fluorouracil (CMF) with an anthracycline-based regimen in adjuvant treatment of node-positive breast cancer patients. Ann Oncol 14:693-698, 2003
Fumoleau P, Kerbrat P, Romestaing P, et al: Randomized trial comparing six versus three cycles of epirubicin-based adjuvant chemotherapy in premenopausal, node-positive breast cancer patients: 10-Year follow-up results of the French Adjuvant Study Group 01 trial. J Clin Oncol 21:298-305, 2003
French Adjuvant Study Group: Benefit of a high-dose epirubicin regimen in adjuvant chemotherapy for node-positive breast cancer patients with poor prognostic factors: 5-Year follow-up results of French Adjuvant Study Group 05 Randomized Trial. J Clin Oncol 19:602-611, 2001
Fargeot P, Roche H, Bonneterre J, et al: Disease-free survival advantage of weekly epirubicin plus tamoxifen vs tamoxifen alone as adjuvant treatment of operable, node positive elderly breast cancer: 5-year follow-up results of French Adjuvant Study Group, FASG-08 trial. Proc Am Soc Clin Oncol 21:37a, 2002 (abstr 145)
Kerbrat P, Roche H, Bonneterre J, et al: Can we substitute an epirubicin-vinorelbine regimen to 6 cycles of FEC 100 as adjuvant chemotherapy of operable, node positive breast cancer patients 5-Year follow-up results of French Adjuvant Study Group, FASG-09. Proc Am Soc Clin Oncol 21:48a, 2002 (abstr 189)
van der Hage JA, van de Velde CJH, Julien JP, et al: Preoperative chemotherapy in primary operable breast cancer: Results from the European Organization for Research and Treatment of Cancer Trial 10902. J Clin Oncol 19:4224-4237, 2001
Levine MN, Bramwell VH, Pritchard KI, et al: Randomized trial of intensive cyclophosphamide, epirubicin and fluorouracil chemotherapy compared with cyclophosphamide, methotrexate and fluorouracil in premenopausal women with node-positive breast cancer. J Clin Oncol 16:2651-2658, 1998
Crump M, Tu D, Shepherd L, et al: Risk of acute leukemia following epirubicin-base adjuvant chemotherapy: A report from the National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol 21:3066-3071, 2003
Wils JA, Bliss JM, Marty M, et al: Epirubicin plus tamoxifen versus tamoxifen alone in node-positive postmenopausal patients with breast cancer: A randomized trial of the International Collaborative Cancer Group. J Clin Oncol 17:1988-1998, 1999
Coombes RC, Bliss JM, Wils J, et al: Adjuvant cyclophosphamide, methotrexate, 5-fluorouracil (CMF) versus 5-fluorouracil, epirubicin, cyclophosphamide (FEC) chemotherapy in premenopausal women with axillary node positive operable breast cancer: Results of a randomised trial—The International Collaborative Cancer Group. J Clin Oncol 14:35-45, 1996
Marty M, Hall E, Wils J, et al: Evaluation of tolerability of CMF versus FEC in a randomised trial in node negative poor risk primary breast cancer patients. Proc Am Soc Clin Oncol 21:69a, 2002 (abstr 273)
Bliss JM, Wils J, Marty M, et al: Evaluation of the tolerability of FE50C versus FE75C in a prospective randomized trial in adjuvant breast cancer patients. Proc Am Soc Clin Oncol 21:51b, 2002 (abstr 2017)
Bergh J, Wiklund T, Erikstein B, et al: Tailored fluorouracil, epirubicin, and cyclophosphamide compared with marrow-supported high-dose chemotherapy as adjuvant treatment for high-risk breast cancer: A randomized trial. Lancet 356:1384-1391, 2000
Campone M, Roche H, Kerbrat P, et al: Secondary leukemias after epirubicin-based adjuvant chemotherapy for operable breast cancer patients: 15 years experience of French Adjuvant Study Group. Breast Cancer Res Treat 76:S157, 2002
Clayton D, Hills M: Statistical Models in Epidemiology. Oxford, United Kingdom, Oxford Science Publications, 1993, pp 96-109, 122-132
Stokes ME, Davis CS, Koch GG: Categorical Data Analysis Using the SAS System. Cary, NC, SAS Institute Inc, 1995, pp 463-481
Peto R, Pike MC, Armitage P, et al: Design and analysis of randomised clinical trials requiring prolonged observation of each patient: I. Introduction and design. Br J Cancer 34:585-612, 1976
Bennet JM, Catovsky D, Daniel MT, et al: Proposed revised criteria for the classification of the acute myeloid leukemia: A report of the French-American-British (FAB) Cooperative Group. Ann Intern Med 103:620-625, 1985
Pedersen-Bjergaard J, Philip P: Two different classes of therapy-related and de-novo acute myeloid leukemia Cancer Genet Cytogenet 55:119-124, 1991
Quesnel B, Kantarjian H, Pedersen-Bjergaard J, et al: Therapy-related acute myeloid leukemia with t(8;21), inv(16) and t(8;16): A report on 25 cases and review of the literature. J Clin Oncol 11:2370-2379, 1993
Hoffmann L, Mo P, Pedersen-Bjergaard J, et al: Therapy-related acute promyelocytic leukemia with t(15;17)(q22;q12) following chemotherapy with drugs targeting DNA topoisomerase-II: A report of two cases and a review of the literature. Ann Oncol 6:781-788, 1995
Linassier C, Barin C, Calais G, et al: Early secondary acute myelogenous leukemia in breast cancer patients after treatment with mitoxantrone, cyclophosphamide, fluorouracil and radiation therapy. Ann Oncol 11:1289-1294, 2000
Valagussa P, Moliterni A, Terenziani M, et al: Second malignancies following CMF-based adjuvant chemotherapy in resectable breast cancer. Ann Oncol 5:803-808, 1994
Tallman MS, Gray R, Bennet JM, et al: Leukemogenic potential of adjuvant chemotherapy for early-stage breast cancer: The Eastern Cooperative Oncology Group experience. J Clin Oncol 13:1557-1563, 1995
Curtis RE, Boice JD Jr, Stovall M, et al: Risk of leukemia after chemotherapy and radiation treatment for breast cancer. N Engl J Med 326:1745-1751, 1992
Diamandidou E, Buzdar AU, Smith TL, et al: Treatment-related leukemia in breast cancer patients treated with fluorouracil-doxorubicin-cyclophosphamide combination adjuvant chemotherapy: The University of Texas M.D. Anderson Cancer Center experience. J Clin Oncol 14:2722-2730, 1996
Brodsky RA, Bedi A, Jones RJ: Are growth factors leukemogenic Leukemia 10:175-177, 1996
Morgan RW: Risk of endometrial cancer after tamoxifen treatment. Oncology 11:25-33, 1997
Matsuyama Y, Tominaga T, Nomura Y, et al: Second cancers after adjuvant tamoxifen therapy for breast cancer in Japan. Ann Oncol 11:1537-1543, 2000
Early Breast Cancer Trialists' Collaborative Group: Tamoxifen for early breast cancer: An overview of the randomized trials. Lancet 351:1451-1467, 1998
Albain K, Green S, Ravdin P, et al: Overall survival after cyclophosphamide, Adriamycin, 5-FU, and tamoxifen (CAFT) is superior to T alone in postmenopausal, receptor(+), node(+) breast cancer: New Findings from Phase III South West Oncology Group Intergroup Trial S8814 (INT-0100). Proc Am Soc Clin Oncol 20:24a, 2001 (abstr 94)
Haskell CM, Green SJ, Sledge GW, et al: Phase III comparison of adjuvant high-dose doxorubicin plus cyclophosphamide (AC) versus sequential doxorubicin followed by cyclophosphamide (AC) in breast cancer patients with 0-3 positive nodes (INTERGROUP 0137). Proc Am Soc Clin Oncol 21:36a, 2002 (abstr 142)
Delaloge S, PalmaM D, Guerin S, et al: Adjuvant doxorubicin, vincristine, cyclophosphamide, fluorouracil (AVCF) for breast cancer (BC): No treatment related leukemia at 13 year follow-up. Proc Am Soc Clin Oncol 17:117a, 1998 (abstr 449)
Chaplain G, Milan C, Sgro C, et al: Increased risk of acute leukemia after adjuvant chemotherapy for breast cancer: A population-based study. J Clin Oncol 18:2836-2842, 2000(Claudio Praga, Jonas Berg)