Critical illness occurs frequently in pediatric oncology patients with nearly 40% requiring admission to the PICU during treatment of their cancer (1). Despite improved outcomes for these patients over the last 3 decades, PICU mortality rates in this population still range from 16% to 73%, which is significantly higher than the 3–5% mortality rate of the general PICU population (1–5). Risk factors for mortality in these patients include the known contributors to poor outcomes in critically ill, noncancer patients such as need for mechanical ventilation (MV), the presence of infection, and fluid overload (FO) (3, 8). Additional factors specific to cancer patients include the effects of malignancy on organ function as well as antineoplastic therapy-associated systemic toxicity (2, 6–8).
The toxicities associated with antineoplastic therapies have been extensively studied. In particular, the effects of anthracyclines, a commonly used class of chemotherapeutic agents, on cardiac function are well described. Anthracyclines are known to cause cardiac toxicity through free radical-mediated oxidative damage to myocytes (9–15). Anthracycline-associated cardiac toxicity increases with cumulative anthracycline dose, especially at doses 250–500 mg/m2, and affects approximately 10% of exposed patients resulting in increased both short- and long-term morbidity and mortality (16–23).
Although oncology patients admitted to the PICU have higher mortality than nononcology patients, there is little data describing the risk of critical illness in this population or the outcomes of critically ill, anthracycline-exposed, pediatric oncology patients (1–5). Due to the importance of cardiac output in maintaining all organ function, it would be intuitive to hypothesize that anthracycline-exposed patients would do poorly, especially if cardiac dysfunction (CD) was present. However, specific risk factors for ICU admission and poor outcomes in these patients have yet to be identified. In addition, there are no studies evaluating whether anthracycline exposure alone, without clinical evidence of CD, is a risk factor for ICU admission or poor outcomes.
Therefore, the purpose of this study was to determine if anthracycline-exposed pediatric oncology patients with CD were more likely to experience critical illness and suffer greater morbidity and mortality than their anthracycline-exposed counterparts without CD. We also aimed to identify noncardiac organ dysfunction at the time of admission as well as other demographic data (including cumulative anthracycline dose), which may impact the need for intensive care and could serve as specific risk factors for poor outcomes in this population.
This is a retrospective, institutional review board approved, parallel cohort study examining the medical records of all anthracycline-exposed pediatric oncology patients admitted to Ann & Robert H. Lurie Children’s Hospital of Chicago (formerly Children’s Memorial Hospital) between January 1, 2006, and December 31, 2014. The observation period extended from the time of first anthracycline dose through 12 months following the last anthracycline dose. For children with more than one PICU admission during the timeframe, the first admission was reviewed. For children admitted to a non-ICU location, the first nonroutine chemotherapy admission was reviewed.
The patients included in this study were those with an oncologic diagnosis who received anthracyclines within the specified time period, birth to 18 years old, and were admitted to the hospital for any reason other than routine chemotherapy. We excluded patients who were less than term gestation or greater than 18 years old, patients without cumulative anthracycline dose data or without data regarding cardiac function (echocardiograms), any chronically ventilated patients (automatic criteria for PICU admission), or patients admitted with a preexisting do not attempt resuscitation order.
The primary aim of this study was to determine if anthracycline-exposed patients with CD were more likely to be admitted to the PICU than those anthracycline-exposed patients without CD. Secondarily, we sought to determine if anthracycline-exposed patients with CD admitted to the PICU have higher mortality (both in hospital and within 1 yr of admission) and longer length of PICU and hospital stay than those anthracycline-exposed PICU patients without CD. Additional outcomes measured included respiratory and cardiovascular support (duration of MV, ventilator-free days [VFD], need for noninvasive positive pressure ventilation [NIPPV], and duration and amount of vasoactive medications), other medications and infectious data during the admission, presence of FO, and requirements for renal replacement therapy (RRT). A Kaplan-Meier curve was generated to identify time to death following ICU admission.
CD was defined as an ejection fraction (EF) of less than or equal to 55%, a shortening fraction (SF) less than or equal to 28%, ventricular dilation with a z score exceeding 2, or the notation of at least moderate dysfunction on the most recent echocardiogram (12, 24). Oncologic Pediatric Risk of Mortality (O-PRISM) scores were calculated for each patient (25). Percent FO was calculated as net fluid status during admission divided by baseline weight in kg. This was also treated as a dichotomous variable (> or < 10% FO). Insensible losses were not included in the FO calculation. Vasoactive Inotrope Score, based on dosing of inotropes, was calculated hourly during the admission and the highest score was recorded (26).
Subject characteristics and outcomes were summarized with mean values and sd for normally distributed continuous variables and with median values and interquartile ranges (IQRs) for continuous variables that were not normally distributed. Categorical variables are reported as counts and percentages. We used t test (or nonparametric equivalent) for analysis of continuous variables, and chi-square test for categorical variables. Univariate conditional logistic regression models were created to estimate the strength of the association between potential predictors and outcome measures. If a variable had greater than 10% missing data, it was not included for analysis. Any variables with a p value of less than or equal to 0.1 were included for analysis in the final models. Stepwise multivariable logistic regression was used to identify independent risk factors for death. Stepwise multiple linear regression was used to identify risk factors associated with PICU length of stay (LOS). Two-sided p values of less than 0.05 are considered to be statistically significant in all final models.
Seven-hundred thirty-four patients received anthracyclines at our institution between January 1, 2006, and December 31, 2014 (Fig. 1). One-hundred eighty-nine were excluded. Two-hundred nineteen patients had at least one PICU admission, and 326 patients had at least one non-PICU admission within 1 year of anthracycline therapy and were included in analysis.
Patient Demographic Information
Table 1 outlines the demographic characteristics of analyzed patients. Children admitted to the PICU were older on average and more commonly admitted with shock states and multiple organ dysfunction. They were also more likely to have undergone stem cell transplant and had higher cumulative anthracycline dose, lower SF, and higher O-PRISM scores.
Patients with CD were more likely to be admitted to the PICU (87% vs 37% rate of PICU admission for patients with and without CD, respectively). Fifty-five percent of echocardiograms on which CD was determined were obtained within 1 month prior to admission to the hospital, and 94 percent were within 6 months prior to admission.
Outcomes of critically ill anthracycline-exposed children with and without CD were also compared and are summarized in Table 2. Patients with CD had significantly higher cumulative anthracycline doses at the time of admission than non-CD patients, (216 mg/m2; IQR, 98–340 vs 123 mg/m2; IQR, 88.5–230; p < 0.001) and patients with CD had significantly higher O-PRISM scores at admission (7.66 ± 5.3 vs 4.28 ± 4.5; p < 0.001).
Eighteen patients (8%) died during their PICU stay. PICU mortality was significantly higher in patients with CD (26% vs 6%; p < 0.001). When comparing PICU and hospital LOS, both were significantly longer in critically ill patients with CD (7 d; IQR, 2.5–13 vs 2 d; IQR, 1–4; p < 0.001 for the PICU and 13 d; IQR, 7–25 vs 7 d; IQR, 1–12; p = 0.002 for total hospital LOS).
The majority of mortality in this study was seen during PICU admission. However, when comparing mortality within 1 year of PICU admission, patients with CD persisted with higher mortality rates (demonstrated in Fig. 2).
Significantly more patients with CD required MV during their PICU admission (59% vs 18%; p < 0.001) (Table 3). Once intubated, patients with CD had significantly fewer VFD at 28 days (0 d; IQR, 0–23 d) versus non-CD (24 d; IQR, 8–27 d) (p = 0.008). Those children with CD also received NIPPV more often than non-CD children (40% vs 10%; p < 0.001).
There was no difference in vasoactive medication use in CD versus non-CD patients. Forty-three percent of all patients received at least one vasoactive medication (most commonly norepinephrine). However, patients with CD received vasoactive medications for a significantly longer period of time (7 d; IQR, 3–10 vs 2 d; IQR, 1–3; p < 0.001) than non-CD patients.
FO greater than 10% was present in 20% of the cohort but developed more commonly in patients with CD (41% vs 17%; p = 0.013). Patients with CD were also more likely to receive diuretics (69% vs 18%; p < 0.001). RRT was used infrequently in this cohort, only 5% of patients, with no significant difference between patients with and without CD (15% vs 3%; p = 0.256).
Multivariable logistic regression (Table 3) demonstrated that death within 1 year of ICU admission was associated with cumulative anthracycline dose (odds ratio [OR], 1.17; CI, 1.01–1.37; p = 0.037), O-PRISM score (OR, 1.11; CI, 1.03–1.2; p = 0.002), and receipt of RRT (OR, 22.4; CI, 2.51–199.9; p < 0.001).
Table 4 compares children admitted to the PICU and those admitted to non-PICU locations. Children admitted to the PICU had longer LOS and higher rates of mortality both during admission and at 1 year following admission. Those patients admitted to the PICU were also more likely to have an identified infection and had higher rates of FO.
Multivariate logistic regression model demonstrated an association between cumulative anthracycline dose and PICU admission (OR, 1.063 per 10 unit change in dose; 95% CI, 1.036–1.090; p < 0.0001), SF (OR, 0.6 per 5 unit change in SF; 95% CI, 0.65–0.98; p 0.03), receipt of steroids (OR, 5.16; 95% CI, 3.16–8.44; p < 0.0001), and oxygen by nasal cannula (OR, 14; 95% CI, 6.9–28.7; p < 0.0001).
Our study demonstrates that more than 30% of anthracycline-exposed children required a PICU admission and of those admitted to the PICU, 8% died. However, of anthracycline-exposed children with CD, 87% required PICU admission and had a mortality rate of 26%. Although children with oncologic diagnoses who receive any amount of anthracyclines are at high risk for critical illness, our data suggest that with increasing anthracycline dose or development of CD, the risk of PICU admission and mortality are significantly increased.
Previous studies indicate that children with cancer admitted to the PICU do poorly. We suggest that it is actually the anthracycline-exposed children with cancer and CD who do poorly. Although we cannot draw any conclusions about nonanthracycline-exposed children, our data indicate that children with cancer who are exposed to anthracyclines, develop CD, and become critically ill have high rates of mortality, longer ICU and hospital LOS, and higher rates of multiple organ dysfunction than anthracycline-exposed children without CD. These data suggest that CD is associated with poor organ function in general and ultimately mortality. The higher frequency of respiratory failure requiring MV, hemodynamic instability requiring higher vasoactive dosing, and FO in the CD patients reinforce the critical physiologic concept that adequate cardiac output and oxygen delivery are necessary for organ recovery. The more than five-fold increase in mortality demonstrated in our study is striking. It is possible (and plausible) that the cause of death in these patients was not solely cardiac failure but more likely due to the complex interactions between multiple organ systems resulting in poor end-organ perfusion and unrecoverable organ function, although addressing this hypothesis was beyond the scope of this study. Further, there are likely additional modifiable risk factors for mortality are likely also present not studied here.
One of our more interesting findings is that cumulative anthracycline dose, regardless of presence of diagnosed CD, is independently associated with PICU admission and mortality. This implies that there may be undetected CD impacting organ function and patient outcomes. Further, the median cumulative anthracycline dose in our patients with CD was lower than the 250 mg/m2 threshold often cited at higher risk for toxicity. Although this study did not account for how anthracyclines were administered (which may play a role in development of CD regardless of total dose), the association between dose alone and death poses the question of whether our current standard of detecting and diagnosing CD in this population is sufficient. Children who receive anthracyclines are routinely monitored with transthoracic echocardiograms in an attempt to detect signs of CD including left ventricular dysfunction (27, 28). However, there is growing evidence indicating that assessment of left ventricular dysfunction by standard echocardiographic metrics is insensitive and unable to detect early, subclinical signs of toxicity (29). The best mode of screening and surveillance for CD in this population remains unclear. Evaluation of global longitudinal strain may be a more sensitive marker but has not been evaluated in pediatrics and is not a standard part of echocardiographic surveillance, although warrants further study (30). Should we be screening with other imaging modalities or looking at things other than EF and SF? Should we use tissue Doppler imaging, global longitudinal strain (30), stress echo, or cardiac MRI (29, 32, 33) in all anthracycline-exposed children, in children deemed to be at higher risk for anthracycline-associated cardiomyopathy or not at all? Research to date has demonstrated little utility in serum biomarkers, although this remains an important area for further investigation and may augment screening practices (23, 34–36). Our inability to recognize the proposed subclinical effects of anthracyclines on the heart precludes clinicians from considering the implication of undetected organ toxicity and, thus, may represent missed opportunities for intervention when treating anthracycline-exposed children admitted to the hospital with acute illness.
Other less surprising findings in this study include that the need for respiratory support; via nasal cannula or NIPPV, is associated with ICU admission and increased mortality. Respiratory failure is already a known poor prognostic indicator in this population. However, even respiratory insufficiency, requiring nasal cannula, seems to be associated with increased critical illness and mortality.
Due to the immunosuppression associated with chemotherapy as well as cancer diagnoses themselves, pediatric oncology patients are at high risk for infection. In these children, the threshold to treat for serious bacterial infection is low. As a result, the majority of children in this study received antibiotics for presumed infection. However, documentation of true infection, by culture or polymerase chain reaction, was much more common in the children admitted to the PICU. This could be a sampling bias, given that critically ill children are more likely to have frequent cultures obtained. This could also be attributed to the general severity of illness in this population. Due to the longer turn-around time for studies of infection, this finding would be difficult to use as a predictor for critical illness but could generate further investigation.
FO has been associated with increased mortality in critically children, typically at a threshold of 10–20% (31). Our data demonstrated a significant effect on 1-year mortality with FO, but median levels were less than 10% in both groups. This could be interpreted in several ways. FO may simply be an indicator of more severe illness—lower urine output secondary to renal hypoperfusion or aggressive fluid management in the setting of hemodynamic instability. However, these data may suggest that FO even at levels of greater than 5% may confer worse prognosis.
Another aspect of care in this population that could not be assessed for in this study is prevention of cardiomyopathy. Current practices include primary prevention with dexrazoxane (37) or secondarily with angiotensin-converting enzyme (ACE) inhibitors (38, 39). This study did not explicitly look at who was on ACE inhibitors at the time of admission, but these data could have an impact of outcomes of these patients. Most patients in this study did not receive dexrazoxane, and it was not part of the chemotherapy protocols for the entire time period studied, so its effect is difficult to determine with this cohort.
Clinically, our goal is to understand if there is a way to reliably predict PICU admission and mortality in these patients. Development of a model to predict need for PICU admission and/or risk of death at the time of presentation to the emergency department (ED) or clinic could provide valuable information to providers with the goal of improving patient outcomes with early identification. Additional research could develop predictive models to stratify risk of critical illness and mortality in these patients; to initiate more frequent screening of cardiac function in some patient or flag high-risk patients in the medical record at the time of presentation to the ED. Another area for further investigation is the use of adjunctive imaging modalities to screen and survey some high-risk patients.
There are several limitations to this study. First, these data were collected at a single center with institution-specific practices regarding oncology care and indications for PICU admission. This may limit its generalizability to other institutions. Second, the retrospective design means there can be no assumptions of causality and only associations can be observed. Additionally, we only examined the records of children exposed to anthracyclines and therefore cannot make comparisons to nonexposed children. We did not assess the potential impact of other chemotherapeutic agents in this population, which may have contributed to differences between patients. Further, although we have demonstrated an association between anthracycline exposure and mortality, there are likely other contributing factors (such as sepsis) that may precipitate or worsen existing CD and impact other organ systems that are not able to be quantified and therefore not evaluable. Last, our definition of CD is based on clinical echocardiogram criteria which, as noted, may not detect subtler evidence of dysfunction (29). Therefore, there may be children with anthracycline exposure and CD without clinical signs or symptoms, so they are not being screened; representing a population that could be missed and potentially negating some of these data. Due to the retrospective study design and lack of standardization on when echocardiograms are obtained, the technicians obtaining images and cardiologists interpreting the data, we are unable to discern the timing of onset of CD and may not have captured all patients with CD. Additional studies are needed to determine the best method of diagnosing CD in order to facilitate earlier and more targeted therapies for patients at higher risk for poor outcomes.
Children with cancer who receive anthracyclines, especially at higher doses, and develop CD are more likely to require ICU care and have higher rates of mortality. In addition to cardiac function, respiratory insufficiency, infection, and FO are associated with critical illness in this population. Once admitted to the PICU, anthracycline-exposed children with CD have higher rates of mortality and longer PICU and hospital LOS than those critically ill, anthracycline-exposed children without CD. Data also indicate that children with CD have higher rates of multiple organ dysfunction represented by fewer VFD, increased requirement for MV, receipt of higher doses of vasoactive medications, and higher rates of FO. These children may represent a subset of oncology patients who should be further studied and potentially approached more aggressively during times of critical illness.
1. Piastra M, Fognani G, Franceschi A; ICARO Italian Network For Intensive Care In Pediatric Oncology
: Pediatric intensive care unit
admission criteria for haemato-oncological patients: A basis for clinical guidelines implementation. Pediatr Rep 2011; 3:e13
2. Hallahan AR, Shaw PJ, Rowell G, et al. Improved outcomes of children with malignancy admitted to a pediatric intensive care unit
. Crit Care Med 2000; 28:3718–3721
3. Zinter MS, Dvorak CC, Spicer A, et al. New insights into multicenter PICU mortality among pediatric
hematopoietic stem cell transplant patients. Crit Care Med 2015; 43:1986–1994
4. Duncan CN, Lehmann LE, Cheifetz IM, et al.; Pediatric
Acute Lung Injury and Sepsis (PALISI) Network: Clinical outcomes of children receiving intensive cardiopulmonary support during hematopoietic stem cell transplant. Pediatr Crit Care Med 2013; 14:261–267
5. Chima RS, Daniels RC, Kim MO, et al. Improved outcomes for stem cell transplant recipients requiring pediatric
intensive care. Pediatr Crit Care Med 2012; 13:e336–e342
6. Jacobe SJ, Hassan A, Veys P, et al. Outcome of children requiring admission to an intensive care unit
after bone marrow transplantation. Crit Care Med 2003; 31:1299–1305
7. Lombel RM, Kommareddi M, Mottes T, et al. Implications of different fluid overload definitions in pediatric
stem cell transplant patients requiring continuous renal replacement therapy. Intensive Care Med 2012; 38:663–669
8. Naeem N, Reed MD, Creger RJ, et al. Transfer of the hematopoietic stem cell transplant patient to the intensive care unit
: Does it really matter? Bone Marrow Transplant 2006; 37:119–133
9. Angsutararux P, Luanpitpong S, Issaragrisil S. Chemotherapy-induced cardiotoxicity: Overview of the roles of oxidative stress. Oxid Med Cell Longev 2015; 2015:795602
10. Chung WB, Youn HJ. Pathophysiology and preventive strategies of anthracycline
-induced cardiotoxicity. Korean J Intern Med 2016; 31:625–633
11. Damiani RM, Moura DJ, Viau CM, et al. Pathways of cardiac toxicity: Comparison between chemotherapeutic drugs doxorubicin and mitoxantrone. Arch Toxicol 2016; 90:2063–2076
12. Higgins AY, O’Halloran TD, Chang JD. Chemotherapy-induced cardiomyopathy. Heart Fail Rev 2015; 20:721–730
13. Mele D, Nardozza M, Spallarossa P, et al. Current views on anthracycline
cardiotoxicity. Heart Fail Rev 2016; 21:621–634
14. Mercurio V, Pirozzi F, Lazzarini E, et al. Models of heart failure based on the cardiotoxicity of anticancer drugs. J Card Fail 2016; 22:449–458
15. Simůnek T, Stérba M, Popelová O, et al. Anthracycline
-induced cardiotoxicity: Overview of studies examining the roles of oxidative stress and free cellular iron. Pharmacol Rep 2009; 61:154–171
16. Armstrong GT, Liu Q, Yasui Y, et al. Late mortality among 5-year survivors of childhood cancer: A summary from the Childhood Cancer Survivor Study. J Clin Oncol 2009; 27:2328–2338
17. Cardinale D, Biasillo G, Cipolla CM. Curing cancer, saving the heart: A challenge that cardioncology should not miss. Curr Cardiol Rep 2016; 18:51
18. Creutzig U, Diekamp S, Zimmermann M, et al. Longitudinal evaluation of early and late anthracycline
cardiotoxicity in children with AML. Pediatr Blood Cancer 2007; 48:651–662
19. Lipshultz SE, Colan SD, Gelber RD, et al. Late cardiac effects of doxorubicin therapy for acute lymphoblastic leukemia in childhood. N Engl J Med 1991; 324:808–815
20. Mulrooney DA, Yeazel MW, Kawashima T, et al. Cardiac outcomes in a cohort of adult survivors of childhood and adolescent cancer: Retrospective analysis of the Childhood Cancer Survivor Study cohort. BMJ 2009; 339:b4606
21. Paulides M, Kremers A, Stöhr W, et al.; German Late Effects Working Group in the Society of Pediatric Oncology
and Haematology (GPOH): Prospective longitudinal evaluation of doxorubicin-induced cardiomyopathy in sarcoma patients: A report of the late effects surveillance system (LESS). Pediatr Blood Cancer 2006; 46:489–495
22. Ryerson AB, Wasilewski-Masker K, Border WL, et al. Pediatric
quality of life in long-term survivors of childhood cancer treated with anthracyclines. Pediatr Blood Cancer 2016; 63:2205–2211
23. Sadurska E. Current views on anthracycline
cardiotoxicity in childhood cancer survivors. Pediatr Cardiol 2015; 36:1112–1119
24. van Dalen EC, van den Brug M, Caron HN, et al. Anthracycline
-induced cardiotoxicity: Comparison of recommendations for monitoring cardiac function during therapy in paediatric oncology
trials. Eur J Cancer 2006; 42:3199–3205
25. Schneider DT, Lemburg P, Sprock I, et al. Introduction of the oncological pediatric
risk of mortality score (O-PRISM) for ICU support following stem cell transplantation in children. Bone Marrow Transplant 2000; 25:1079–1086
26. Kumar M, Sharma R, Sethi SK, et al. Vasoactive Inotrope Score as a tool for clinical care in children post cardiac surgery. Indian J Crit Care Med 2014; 18:653–658
27. Spewak MB, Williamson RS, Mertens AC, et al. Yield of screening echocardiograms during pediatric
follow-up in survivors treated with anthracyclines and cardiotoxic radiation. Pediatr Blood Cancer 2017; 64:1–10
28. Watts RG, George M, Johnson WH Jr. Pretreatment and routine echocardiogram monitoring during chemotherapy for anthracycline
-induced cardiotoxicity rarely identifies significant cardiac dysfunction
or alters treatment decisions: A 5-year review at a single pediatric oncology
center. Cancer 2012; 118:1919–1924
29. Rose-Felker K, Border WL, Hong BJ, et al. Cardio-oncology
related to heart failure: Pediatric
considerations for cardiac dysfunction
. Heart Fail Clin 2017; 13:311–325
30. Charbonnel C, Convers-Domart R, Rigaudeau S, et al. Assessment of global longitudinal strain at low-dose anthracycline
-based chemotherapy, for the prediction of subsequent cardiotoxicity. Eur Heart J Cardiovasc Imaging 2017; 18:392–401
31. de Galasso L, Emma F, Picca S, et al. Continuous renal replacement therapy in children: Fluid overload does not always predict mortality. Pediatr Nephrol 2016; 31:651–659
32. Plana JC, Galderisi M, Barac A, et al. Expert consensus for multimodality imaging evaluation of adult patients during and after cancer therapy: A report from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. Eur Heart J Cardiovasc Imaging 2014; 15:1063–1093
33. Yu AF, Raikhelkar J, Zabor EC, et al. Two-dimensional speckle tracking echocardiography detects subclinical left ventricular systolic dysfunction among adult survivors of childhood, adolescent, and young adult cancer. Biomed Res Int 2016; 2016:9363951
34. Braunwald E. Biomarkers in heart failure. N Engl J Med 2008; 358:2148–2159
35. Lipshultz SE, Miller TL, Scully RE, et al. Changes in cardiac biomarkers during doxorubicin treatment of pediatric
patients with high-risk acute lymphoblastic leukemia: Associations with long-term echocardiographic outcomes. J Clin Oncol 2012; 30:1042–1049
36. Cardinale D, Biasillo G, Salvatici M, et al. Using biomarkers to predict and to prevent cardiotoxicity of cancer therapy. Expert Rev Mol Diagn 2017; 17:245–256
37. Wexler LH, Andrich MP, Venzon D, et al. Randomized trial of the cardioprotective agent ICRF-187 in pediatric
sarcoma patients treated with doxorubicin. J Clin Oncol 1996; 14:362–372
38. Sawyer DB, Peng X, Chen B, et al. Mechanisms of anthracycline
cardiac injury: Can we identify strategies for cardioprotection? Prog Cardiovasc Dis 2010; 53:105–113
39. Lipshultz SE, Lipsitz SR, Sallan SE, et al. Long-term enalapril therapy for left ventricular dysfunction in doxorubicin-treated survivors of childhood cancer. J Clin Oncol 2002; 20:4517–4522