- © 2001 by American Society of Clinical Oncology
Anthracycline-Induced Clinical Heart Failure in a Cohort of 607 Children: Long-Term Follow-Up Study
- From the Departments of Pediatric Oncology, Pediatric Cardiology, and General Pediatrics, Academic Medical Center, Emma Kinder Ziekenhuis, University of Amsterdam, Amsterdam, the Netherlands.
- Address reprint requests to L.C.M. Kremer, MD, Department of Pediatrics, Emma Kinder Ziekenhuis, Academic Medical Center, University of Amsterdam, G8-259, Meibergdreef 9, 1105 AZ Amsterdam, the Netherlands; email L.C.Kremer{at}AMC.UVA.NL
Abstract
PURPOSE: To determine the early and late cumulative incidence of anthracycline-induced clinical heart failure (A-CHF) after anthracycline therapy in childhood and to identify associated risk factors.
PATIENTS AND METHODS: The cumulative incidence of A-CHF and the risk factors of A-CHF were assessed in a cohort of 607 children who had been treated with anthracyclines between 1976 and 1996. For 96% of the cohort, we obtained the clinical status up to at least January 1997. The mean follow-up time was 6.3 years.
RESULTS: The cumulative incidence of A-CHF was 2.8%, after a mean follow-up time of 6.3 years and a mean cumulative dose of anthracyclines of 301 mg/m2. A cumulative dose of anthracycline higher than 300 mg/m2 was associated with an increased risk of A-CHF (relative risk, 11.8; 95% confidence interval, 1.6 to 59.5) compared with a cumulative dose lower than 300 mg/m2. The estimated risk of A-CHF increased with time after the start of anthracycline chemotherapy to 2% after 2 years and 5% after 15 years.
CONCLUSION: Up to 5% of patients will develop A-CHF 15 years after treatment, and patients treated with a cumulative dose of anthracyclines higher than 300 mg/m2 are at highest risk for A-CHF. This is thus a considerable and serious problem among these young patients. The findings reinforce the need for strategies for early detection of patients at risk for A-CHF and for the evaluation of other chemotherapeutic possibilities or cardioprotective agents in relation to the survival.
ANTHRACYCLINES HAVE gained widespread use in the treatment of childhood leukemia and solid tumors. Clinically, the most important side effect of anthracycline chemotherapy is a dose-dependent cardiotoxicity.1
Heart damage after anthracycline chemotherapy can be divided into early and late cardiotoxicity. By definition, early cardiotoxicity refers to cardiotoxicity that develops during chemotherapy or in the first year after its completion; whereas late cardiotoxicity occurs at least 1 year after completion of therapy.2
Cardiotoxicity can become manifest in patients as subclinical heart failure and as clinical heart failure. Several studies have evaluated the incidence and risk of early clinical cardiotoxicity and late subclinical cardiotoxicity.3-15 The risk for both types of cardiotoxicity increases with a higher cumulative dose of anthracyclines, female sex, younger age at the start of chemotherapy, type of tumor, black race, presence of trisomy 21, radiation therapy involving the heart, and exposure to cyclophosphamide, ifosfamide, or amsacrine.3-15
Only a few studies have evaluated the incidence of late anthracycline-induced clinical heart failure (A-CHF) in long-term survivors.10-18 However, these studies have serious limitations in that the study populations were small, only subgroups were described, and most studies had a short follow-up period. Because children should have a long life expectancy after a successful antineoplastic treatment, it is important to evaluate the frequency and consequences of late A-CHF. In this study, we evaluated the cumulative incidence of early and late A-CHF and related risk factors in a large cohort of children treated with anthracyclines between 1976 and 1996.
PATIENTS AND METHODS
Patients
Since 1966, all children who are treated for childhood cancer in the Emma Kinder Ziekenhuis/Academic Medical Center are registered at the Hospital Tumor Registry. This registry maintains data on diagnosis, treatment, last known medical status, and follow-up of all admitted patients and served as the basis for the selection of patients in the current study. We included only those patients who received their first treatment with anthracyclines after 1976, because the chemotherapeutic treatment was not specified in the early years of the registration. According to the computer database of the Hospital Tumor Registry, 609 children had been treated in the Emma Kinder Ziekenhuis/Academic Medical Center with anthracycline chemotherapy between 1976 and 1996.
Treatment and Follow-Up Data
Data of 567 of 609 children were collected directly from the medical records by one of the authors (E.C.vD.). For 40 patients whose medical records were missing, we obtained information by means of the registry charts kept by the Hospital Tumor Registry. No data were available for two children. Therefore, the study population consisted of 607 patients.
For each patient, the following information was recorded: (1) date of birth, (2) sex, (3) type of malignant lesion and tumor status, (4) chemotherapeutic protocol, including the cumulative doses of administered anthracyclines (doxorubicin, daunorubicin, idarubicin, and/or epirubicin), cyclophosphamide, and ifosfamide, (5) date of start of anthracycline treatment, (6) concurrent radiotherapy involving the heart, including left abdominal irradiation, (7) last follow-up date, (8) date and cause of death, (9) signs and symptoms of clinical heart failure and, if that was the case, etiology, treatment, and clinical outcome.
Attempts were made to establish the clinical status of patients lost to follow-up by sending a questionnaire to their general practitioners. We finally succeeded in obtaining information on the clinical status up to at least January 1997 (or date of death) for 580 patients (96% of the cohort).
Definition of Clinical Cardiotoxicity
A case of A-CHF was defined as congestive heart failure not attributable to other known causes, such as direct medical effects of the tumor, septic shock, or renal failure. We defined congestive heart failure as the presence of the following clinical signs and symptoms: dyspnea, pulmonary edema, peripheral edema, and/or exercise intolerance that were treated with anticongestive therapy. A pediatric cardiologist (J.O.) confirmed the diagnosis in patients with cardiac events that may or may not have met this definition of clinical cardiotoxicity. The clinical outcome of A-CHF was either “death,” “alive with anticongestive treatment,” or “clinical recovery without current requirement for anticongestive therapy, but anticongestive treatment previously.” Depending on the time of onset, the A-CHF was classified as early A-CHF, ie, during anthracycline chemotherapy or within the first year after the end of treatment, or as late A-CHF, ie, CHF more than 1 year after the completion of anthracycline chemotherapy.
Statistical Analysis
The main outcome event was defined as the occurrence of A-CHF. Event-free survival was defined as the time from the start of first administration of anthracycline until the development of A-CHF, or until the latest follow-up evaluation, or until death. The risk factors for A-CHF were a high cumulative dose of anthracyclines (> 300 mg/m2), treatment with ifosfamide or with cyclophosphamide, irradiation involving the heart, female sex, and age at diagnosis younger than 4 years. The risk factors for A-CHF were first compared with the log-rank test. The relative risk for each risk factor was estimated by the Cox regression model.19 Statistical significance (P < .05) was determined with a two-sided test. Multivariate analysis was performed with a stepwise Cox regression model. The P value of the coefficient of the risk factor had to be less than .10 for the risk factor to remain in the model.
The cumulative risk of A-CHF was estimated as a function of the follow-up time from the start of anthracycline treatment and of the cumulative anthracycline dose by the Kaplan-Meier method.20 Survival curves were constructed and the confidence intervals (CIs) were calculated. Analyses were performed using the statistical software SPSS for Windows 8.0 (release 1997; SPSS, Inc, Chicago, IL).
RESULTS
Patients
The clinical characteristics of the study population are listed in Table 1. The mean age at the time of diagnosis was 9.1 years (range, 0.1 to 17.9 years). The mean cumulative dose of anthracyclines was 301 mg/m2 (range, 14 to 960 mg/m2): 37 children received only daunorubicin, 352 children received only doxorubicin, 95 children received only epirubicin, and 118 children received a combination of doxorubicin, daunorubicin, and epirubicin. The exact cumulative dose of the different anthracyclines of 13 patients and the exact chemotherapeutic combinations and doses for five patients are unknown. In addition to anthracyclines, 307 children were treated with cyclophosphamide, and 138 children were treated with ifosfamide.
Clinical Characteristics of 607 Children Treated With Anthracyclines
The mean follow-up time after the first dose of anthracycline treatment for the whole group was 6.3 years (range, 0.01 to 21.7 years). The mean age of the patients at the end of the follow-up was 15.4 years (range, 0.3 to 37.6 years). For patients alive at the most recent follow-up date, 64.3% of the cohort, the mean follow-up period was 8.6 years (range, 0.23 to 21.7 years). For 23.4% of the patients, the follow-up was longer than 10 years, and for 7.6%, it was longer than 15 years. At last contact, 217 patients (35.7%) had died. Four patients died of A-CHF ( Table 2).
Characteristics, Treatment, and Follow-Up of 17 Patients With A-CHF
Incidence and Outcome of A-CHF
The cumulative incidence of A-CHF at a mean follow-up time of 6.3 years was 2.8% (17 patients). The risk of developing A-CHF as a function of the follow-up time based on Kaplan-Meier estimates is shown in Fig 1. The risk of A-CHF after 2 years was 2% (95% CI, 0.8% to 3.2%); after 5 years, it was 2.8% (95% CI, 1.3% to 4.3%); after 10 years, it was 3.3% (95% CI, 1.6% to 5.0%); and after 15 years, it was 4.8% (95% CI, 1.8% to 8.3%).
Fig 1. Kaplan-Meier plot of the estimated risk of A-CHF as a function of the follow-up time (in years) after the start of anthracycline therapy.
Thirteen cases (2.1%) of A-CHF occurred during or within the first year of therapy. The median time between the first dose of anthracycline and A-CHF was 3.6 years. The characteristics of the patients with A-CHF are listed in Table 2. The mean age of the patients with A-CHF at the time of anthracycline chemotherapy was 9.6 years (range, 1.3 to 15.9 years). The mean cumulative dose of anthracycline at the onset of A-CHF was 461 mg/m2 (range, 225 to 803 mg/m2). Seven children were also treated with cyclophosphamide, and four children were also treated with ifosfamide. Of these children, the differences in the chemotherapy doses between patients with and without A-CHF are listed in Table 3. Four patients (0.7% of the study group) died from A-CHF within 0 to 5.5 years after the onset of symptoms. One patient with A-CHF was still being treated with anticongestive treatment, whereas the other five were not. Of the five patients, one patient developed A-CHF during pregnancy. Follow-up echocardiographic studies of this last patient are not known. The left ventricular shortening fraction, measured by echocardiography, of the other four patients was 28%, 32%, 37%, and 42% at 7, 8.5, 6, and 7.5 years, respectively, after the start of A-CHF.
Cumulative Doses of Chemotherapy for Patients With and Without A-CHF
Risk Factors for A-CHF
Univariate analyses of the risk factors for the occurrence of A-CHF showed a statistically significant increase in the risk of A-CHF associated with a cumulative dose of anthracyclines higher than 300 mg/m2. The results of univariate analysis of the different risk factors are listed in Table 4. The other possible risk factors for A-CHF (ie, treatment with ifosfamide or with cyclophosphamide, radiation therapy involving the heart, female sex, and age at diagnosis younger than 4 years) were not associated with an increased risk. In multivariate Cox analysis, a cumulative dose of anthracyclines higher than 300 mg/m2 was the only independent risk factor.
Relative Risks for A-CHF (univariate analyses)
The risk of A-CHF as a function of the cumulative anthracycline dose is shown in Fig 2. The risk of A-CHF at 300 mg/m2 was 1.1% (95% CI, 0.1% to 2.5%); at 450 mg/m2, it was 4.5% (95% CI, 1.5% to 7.5%); and at 600 mg/m2, it was 17.8% (95% CI, 5.5% to 30.1%).
Fig 2. Kaplan-Meier plot of the estimated risk of A-CHF as a function of the cumulative dose of anthracyclines.
DISCUSSION
Our study is the first to investigate the incidence of both early and late A-CHF in a large cohort of children treated with anthracyclines with such a long and complete follow-up. We estimated the cumulative incidence and risk and possible risk factors for A-CHF in a cohort patients treated with anthracycline therapy for childhood malignancy. In this study, a marked number of patients was found to have developed A-CHF at a young age.
The cumulative incidence of early cardiotoxicity of 2.1% in this study is in agreement with the cumulative incidence of 1.6% mentioned before in two large cohort studies of children by von Hoff et al3 and Krischer et al.12 Two other cohort studies of more than 200 children showed cumulative incidences of 2.8% and 4.0%.4,6 The cumulative incidence of late A-CHF was 0.7% in our study with a mean follow-up time of 6.3 years and a mean cumulative dose of 300.9 mg/m2. Dearth et al13 and Krischer et al12 described cumulative incidences of late A-CHF of 0.9% and 0.1%, respectively, in a cohort of patients. However, the completeness and the length of follow-up were not mentioned. Other studies described an incidence of late A-CHF between 1.7% and 11%.10,11,14-18 A subgroup of patients was investigated in these studies, so selection bias could have caused this broad variation in frequency of A-CHF. In the present study, the rate of occurrence of A-CHF was highest in the first 2 years, ie, 0% to 2%, and 2% to 5% in the next 12 years. It is possible that the clinical condition of patients during chemotherapy, when they often suffer from anemia, acidosis, cachexia, fever, or overhydration from intravenous fluid, lowers the threshold for clinical heart failure.
It is known that A-CHF is a dose-dependent phenomenon. The cumulative risk of A-CHF at 550 mg/m2 of 10% in our study is higher than the 7% found by von Hoff et al3 in a study of both children and adults. The higher susceptibility of children with regard to adults and the longer follow-up of our study can explain these differences. A cumulative dose of anthracyclines higher than 300 mg/m2 was the only significant risk factor for A-CHF. In fact, A-CHF occurred in only one patient treated with a lower dose. The threshold of a cumulative dose of 300 mg/m2 as a risk factor for A-CHF, used in this analysis, is low in comparison with that used in other studies, which used a value of 550 mg/m2.12 Only Swain et al21 mentioned that A-CHF begins at a cumulative dose of 300 mg/m2 in adults. In the present study, 12 out of 17 patients with A-CHF were treated with a cumulative dose of anthracyclines lower than 550 mg/m2. The fact that A-CHF may occur in children at doses “as low as” 300 mg/m2 reinforces the need to re-evaluate the “safe” maximum cumulative dose of 450 to 550 mg/m2 of anthracyclines in children and the desirability of treating children who will receive a cumulative dose of anthracyclines of more than 300 mg/m2 with cardioprotective agents or other treatment possibilities. In contrast with other studies, which evaluated the risk factors related to subclinical cardiotoxicity or early A-CHF, we could not identify other risk factors for the development of A-CHF.3-15
A limitation of the present study is the lack of information on the course of action when subclinical heart damage was diagnosed by echocardiography during treatment. Echocardiography has been used in our hospital since 1985. It is our policy to stop or adjust treatment with anthracyclines if there is large decrease in the left ventricular shortening fraction or a left ventricular shortening fraction ≤ 28%. It is not clear whether this policy will diminish the frequency of early and late A-CHF. Lipshultz et al22 stated that the adjustment of the dose in asymptomatic patients may potentially harm patients if the efficacy of their anticancer therapy is reduced.
It is possible that we underestimate the real incidence of A-CHF. In this study, we used a very strict definition of A-CHF, ie, cardiac heart failure not attributable, as far as we know, to other known causes. Furthermore, not all the patients had had a check-up by a specialist in the hospital; for 96 patients, information was obtained from the general practitioner.
The risk of A-CHF 15 years after the start of anthracycline treatment was estimated to be approximately 5% in this study. It is unclear what the frequency of late A-CHF will be later on, when aging of the heart becomes important. That the percentage of subclinical cardiac abnormalities increases with longer follow-up times suggests a further increase in the incidence of late A-CHF with time. This reinforces the need for sufficiently long complete follow-up studies, strategies for early detection of patients at risk for A-CHF, and the evaluation of other chemotherapeutic possibilities or cardioprotective agents in relation to the survival.
- Received March 21, 2000.
- Accepted August 2, 2000.
