Illustrated are the cumulative 2-year incidences of cardiac events for the patients assigned to receive candesartan and those assigned to the placebo group.
Mean LVEF values plotted by time according to the schedule of follow-up. Error bars represent standard deviation.
eTable 1. Patient baseline characteristics
eTable 2. Treatment characteristics of evaluable patients
eTable 3. Patient baseline and treatment characteristics of evaluable patients by whether an event occurred
eTable 4. Potential Prognostic Factors for occurrence of Cardiac event
eTable 5. Adverse events according to the National Cancer Institute’s Common Terminology Criteria for Adverse Events ≥ grade 3
eTable 6. Number of evaluable patients with LVEF values and worst NYHA score by follow-up time
eTable 7. HER2-genetic variability
eFigure 1. Algorithm (dis)continuation trastuzumab treatment
eFigure 2. Schedule of follow-up visits
eFigure 3. Mean and standard deviation of NT-proBNP levels by follow-up time
eFigure 4. Mean and standard deviation of hs-TnT levels by follow-up time
Customize your JAMA Network experience by selecting one or more topics from the list below.
Boekhout AH, Gietema JA, Milojkovic Kerklaan B, et al. Angiotensin II–Receptor Inhibition With Candesartan to Prevent Trastuzumab-Related Cardiotoxic Effects in Patients With Early Breast Cancer: A Randomized Clinical Trial. JAMA Oncol. 2016;2(8):1030–1037. doi:10.1001/jamaoncol.2016.1726
This is the first randomized placebo-controlled evaluation of a medical intervention for the prevention of trastuzumab-related cardiotoxic effects.
To determine as the primary end point whether angiotensin II antagonist treatment with candesartan can prevent or ameliorate trastuzumab-related cardiotoxic effects, defined as a decline in left ventricular ejection fraction (LVEF) of more than 15% or a decrease below the absolute value 45%.
This randomized, placebo-controlled clinical study was conducted between October 2007 and October 2011 in 19 hospitals in the Netherlands, enrolling 210 women with early breast cancer testing positive for human epidermal growth factor receptor 2 (HER2) who were being considered for adjuvant systemic treatment with anthracycline-containing chemotherapy followed by trastuzumab.
A total of 78 weeks of candesartan (32 mg/d) or placebo treatment; study treatment started at the same day as the first trastuzumab administration and continued until 26 weeks after completion of trastuzumab treatment.
Main Outcomes and Measures
The primary outcome was LVEF. Secondary end points included whether the N-terminal of the prohormone brain natriuretic peptide (NT-proBNP) and high-sensitivity troponin T (hs-TnT) can be used as surrogate markers and whether genetic variability in germline ERBB2 (formerly HER2 or HER2/neu) correlates with trastuzumab-related cardiotoxic effects.
A total of 206 participants were evaluable (mean age, 49 years; age range, 25-69 years) 103 in the candesartan group (mean age, 50 years; age range, 25-69 years) and 103 in the placebo group (mean age, 50 years; age range, 30-67 years). Of these, 36 manifested at least 1 of the 2 primary cardiac end points. There were 3.8% more cardiac events in the candesartan group than in the placebo group (95% CI, −7% to 15%; P = .58): 20 events (19%) and 16 events (16%), respectively. The 2-year cumulative incidence of cardiac events was 0.28 (95% CI, 0.13-0.40) in the candesartan group and 0.16 (95% CI, 0.08-0.22) in the placebo group (P = .56). Candesartan did not affect changes in NT-proBNP and hs-TnT values, and these biomarkers were not associated with significant changes in LVEF. The Ala1170Pro homozygous ERBB2 genotype was associated with a lower likelihood of the occurrence of a cardiac event compared with Pro/Pro + Ala/Pro genotypes in multivariate analysis (odds ratio, 0.09; 95% CI, 0.02-0.45; P = .003).
Conclusions and Relevance
The findings do not support the hypothesis that concomitant use of candesartan protects against a decrease in left ventricular ejection fraction during or shortly after trastuzumab treatment in early breast cancer. The ERBB2 germline Ala1170Pro single nucleotide polymorphism may be used to identify patients who are at increased risk of trastuzumab-related cardiotoxic effects.
clinicaltrials.gov Identifier: NCT00459771
Trastuzumab, a humanized monoclonal antibody against the extracellular domain of human epidermal growth factor receptor 2 (HER2), improves overall survival in patients with HER2-positive metastatic breast cancer and disease-free and overall survival in HER2-positive early stage breast cancer.1-4 Although trastuzumab is generally well tolerated, cardiac dysfunction is an important adverse effect, especially in women who were previously or concomitantly treated with anthracycline-based chemotherapy.5-11
The most frequent clinical manifestation of trastuzumab-related cardiotoxic effects is diminished systolic left ventricular function measured with left ventricular ejection fraction (LVEF). The development of cardiac dysfunction due to cancer treatment negatively affects patients’ cardiac outcomes and seriously limits their therapeutic opportunities, especially continuation or reintroduction of trastuzumab treatment and thereby the overall exposure of the tumor to this drug. Therefore, effective interventions to prevent or limit the development of treatment-related cardiotoxic effects are warranted. Since certain pharmacological interventions can attenuate or reverse left ventricular remodeling in patients with heart failure,12,13 several approaches have been proposed with regard to anthracycline-related toxic effects. For example, it has been reported that angiotensin II–receptor blockers (ARBs), angiotensin-converting enzyme (ACE) inhibitors, and β-blockers may be effective in preventing anthracycline related cardiotoxic effects.14-17 To date, no prospective clinical data have been published on the protective properties of these types of drugs for the prevention or amelioration of trastuzumab-related cardiotoxic effects.
Preliminary data suggest that cardiac markers such as N-terminal pro-B-type natriuretic peptide (NT-proBNP) and cardiac troponins are sensitive and specific markers to detect myocardial injury and to predict treatment-related decreases in LVEF and the development of left ventricular dysfunction and heart failure.18-21 Genetic susceptibility may also play a role, and the effect of several single nucleotide polymorphisms (SNPs) in the transmembrane, extracellular, and intracellular region of HER2 on trastuzumab-related cardiotoxic effects has been examined, although the results have been contradictory so far.22,23
To examine the effect of the ARB candesartan, we conducted a double-blind placebo-controlled multicenter trial in patients with early-stage breast cancer who were treated with anthracyclines followed by trastuzumab. The objective of this study was to determine whether ARB treatment prevents or ameliorates trastuzumab related cardiotoxic effects and to examine whether cardiac biomarkers and SNPs can detect and predict trastuzumab-related cardiotoxic effects.
Question Does treatment with angiotensin II antagonist candesartan prevent trastuzumab-related cardiotoxic effects?
Findings Efficacy analysis was performed in 206 women with early-stage breast cancer who were treated with trastuzumab and concomitantly with either candesartan or placebo. Trastuzumab-related decline in left ventricular ejection fraction (LVEF) of more than 15% or an LVEF decrease below the absolute value of 45% occurred in 20 participants in the candesartan group and in 16 in the placebo group, a nonsignificant difference.
Meaning The findings do not support the concomitant use of candesartan as a protective agent against a decrease in LVEF during or shortly after trastuzumab in early-stage breast cancer.
This study was performed in 19 Dutch hospitals. Random assignment to 1 of the 2 groups (candesartan or placebo) was performed within 3 weeks after day 1 of the last anthracycline chemotherapy cycle. All participants were women, and eligibility criteria were age 18 years or older; early-stage HER2-positive breast cancer; completion of anthracycline-based adjuvant treatment; performance status 2 or lower; LVEF at least 50% as measured by echocardiography or multiple gate acquisition radionuclide imaging; creatinine clearance greater than 50 mL/min (by Cockcroft-Gault formula); thyroid stimulating hormone level between 0.5 and 3.9 MU/L or thyroid hormone free thyroxine between 8 and 26 pmol/L; systolic blood pressure between 100 and 180 mm Hg and diastolic blood pressure between 60 and 100 mm Hg; and first trastuzumab infusion received at least 3 weeks after day 1 of the last anthracycline infusion. Exclusion criteria were a history of hypersensitivity to the study medication; previous malignant condition requiring anthracycline-based chemotherapy, prior biological therapy, immunotherapy, or mediastinal radiotherapy; uncontrolled serious concurrent illness; New York Heart Association (NYHA) class II, III, or IV chronic heart failure; myocardial infarction less than 6 months before enrolment; treatment with an ACE inhibitor, ARB, or lithium; pregnancy or breast feeding. The study protocol (Supplement 1) was approved by the medical ethics committees of all participating centers, and all patients gave written informed consent.
Eligible patients were randomly assigned (1:1) to receive either candesartan, 32 mg/d, or placebo daily. The pharmacy supplied and labeled identical tablets and kept a complete drug accountability record for patients enrolled on this trial. Computer-generated randomization was performed centrally; only the study statistician and the investigational pharmacist had the randomization assignments. The remaining study personnel were masked to randomization. Only in case of emergencies could the sealed envelopes with treatment allocations data be opened.
Blood samples for hematological and serum biochemical monitoring were analyzed locally by the centers. Analyses of the cardiac markers, high sensitivity troponin T (hs-TnT), NT-proBNP, and SNP genotyping were performed centrally. The cardiac markers were measured in plasma samples using a sandwich immunoassay (Modular E system, Roche Diagnostics). Genomic DNA was used to assess genetic variability within ERBB2 (formerly HER2 or HER2/neu) (Val654IIe, Val655IIe, and Ala1170Pro) and fragment, crystallizable (Fc)-gamma receptors (FcgR3A Val158Phe, FcgR2A His166Arg and FcgR2B IIe232Thr). The assay details are outlined in eMethods in Supplement 2.
Cardiac examination included recording of symptoms, findings on clinical examination, adverse effects (graded according to the National Cancer Institute Common Terminology Criteria of Adverse Events [CTCAE], version 3.0), an assessment of the LVEF and NYHA classification, with hematologic and chemical analyses at baseline (before start of trastuzumab therapy), and in weeks 12, 24, 36, 52, 78, and 92. At each visit, cardiac questionnaires were used to estimate the presence of signs or symptoms of congestive heart failure. An electrocardiogram was performed at baseline and in weeks 52 and 78.
Toxic effect of the cancer treatment to the heart resulting in a cardiac event was defined as a decrease in LVEF of more than 15% compared with baseline or a decrease to an absolute value of LVEF below 45%. The decision to (dis)continue trastuzumab and/or the study treatment was based on the algorithm depicted in eFigure 1 in Supplement 2. All cardiac events, suspected unexpected serious adverse reactions, and serious adverse events were reported from start until the end of the study period. A safety monitoring committee, consisting of independent physicians and statisticians with access to unblinded data, monitored the safety of the study. All patients adhered to the same schedule of follow-up visits (eFigure 2 in Supplement 2).
Approved anthracycline-based (neo)adjuvant chemotherapy and radiotherapy (as indicated) was completed before randomization. Trastuzumab was given for 52 weeks in a (3-)weekly schedule and was commonly used as a single agent or in combination with taxane-based chemotherapy followed by trastuzumab as single agent. Endocrine therapy was given after chemotherapy to women with hormone receptor–positive disease. Patients started the trial on the same day as the first trastuzumab administration with oral candesartan, 16 mg/d, or oral placebo daily for 1 week. From week 2 until 26 weeks after completion of trastuzumab treatment, patients took 32 mg of candesartan or placebo daily. If CTCAE grade 1 toxic effects occurred, the dose was reduced to 16 mg, and treatment was discontinued if the patient developed CTCAE grade 2, 3, or 4 toxic effects.
The primary end point of the study was the occurrence of a cardiac event (defined as decline in LVEF of >15% or an absolute value <45%) during trastuzumab treatment and 40 weeks after discontinuation of trastuzumab. Secondary end points were safety of the combination of trastuzumab and candesartan treatment; an evaluation of the predictive value of baseline cardiac markers NT-proBNP and hs-TnT; and detected candidate SNPs in the germline ERBB2 gene in trastuzumab-related cardiotoxic effects.
The sample size was calculated to detect a decrease in the proportion of patients with cardiac events from 30% in the placebo group to 13% in the candesartan group.2,24,25 To obtain 80% power, we enrolled 100 patients in each treatment group at a significance level of α = .05. Three prespecified interim analyses were performed after 10, 20, and 30 cardiac primary end points, with the final analysis at 200 evaluable patients. Error spending functions resembling the O’Brien-Flemming and the Pocock boundaries were used for finding the efficacy and safety stopping boundaries, respectively,26 based on cumulative incidence curves. All analyses were conducted according to the intention-to-treat principle, but patients with no LVEF measurement after randomization were nonevaluable for the primary end point. Safety analysis was performed on all patients who received treatment. Cumulative incidence of cardiac events was calculated using the Kaplan-Meier method by censoring patients without failure at the time of the last LVEF measurement and compared with the log-rank test. The Fisher exact test was used for absolute incidence of cardiac events and serious adverse events.
Prognostic factors at baseline for the occurrence of cardiac events (age, LVEF, NT-proBNP, hs-TnT, performance status, family history, and medical history) were investigated with univariable logistic regression models. A recessive model was used to test the prognostic value of ERBB2 genetic variability,27 SNPs with effects significant at the α = .01 level were analyzed in a multivariable model adjusting for baseline prognostic factors and randomized treatment. Associations between LVEF and NT-proBNP and hs-TnT marker levels were assessed with linear mixed-effects models with LVEF at up to 1.5 years after randomization as outcome. Time was entered as a random effect with random slope and intercept. Randomized treatment, time since randomization, and log marker levels were included as fixed-effect explanatory variables. In addition, Spearman correlation was calculated for the difference since baseline.
All statistical analyses were performed with SAS software (SAS Institute Inc), version 9.2 and R (The R Foundation), version 3.1.1.
A total of 210 women with early-stage breast cancer were enrolled in the study between October 2007 and October 2011. The trial was stopped according to the study protocol. At baseline, the patient characteristics were balanced across both arms (see eTable 1 in Supplement 2). A total of 206 women were evaluable for the primary end point (mean age, 49 years; age range, 25-69 years), of whom 103 were treated with candesartan (mean age, 50 years; age range, 25-69 years) and 103 with placebo (mean age, 50 years; age range 30-67 years) (Figure 1). In total, 48 patients discontinued study treatment prematurely (23.3%). The median follow-up (time from randomization to the last LVEF measurement) was 21 months in the candesartan group (range, 6-27 months) and 21 months in the placebo group (range, 5-25 months). Baseline and treatment characteristics of evaluable patients overall and by cardiac event are reported in eTable 2 and eTable 3, respectively, in Supplement 2.
There were 3.8% more cardiac events in the candesartan group than in the placebo group (95% CI, −7% to 15%; P = .58): 20 events (19%) and 16 events (16%), respectively. The 2-year cumulative incidence of cardiac events was 0.28 (95% CI, 0.13-0.40) in the candesartan group and 0.16 (95% CI, 0.08-0.22) in the placebo group (P = .56) (Figure 2). The LVEF values from baseline to week 92 in both treatment groups are shown in Figure 3. Baseline LVEF (≥55 vs <55%) value was a prognostic factor for the occurrence of a cardiac event (odds ratio [OR], 0.24; 95% CI, 0.11-0.53; P < .001) (see eTable 4 in Supplement 2. The incidence of adverse events was generally similar across treatment groups. eTable 5 in Supplement 2 shows the overall incidence of grade 3 or higher adverse events. Adverse events leading to withdrawal of study treatment were noted in 12 patients, 6 in each study arm, and included in the candesartan vs placebo arms, respectively, dizziness (n = 1, 1% vs n = 1, 1%), hypotension (n = 3, 3% vs n = 0), headache (n = 0 vs n = 1, 1%), myalgia (n = 0 vs n = 1, 1%), fever (n = 0 vs n = 1, 1%) thrombosis (n = 0 vs n = 1, 1%), psychological stress (n = 0 vs n = 1, 1%), and dyspnea (n = 2, 2% vs n = 0). After trastuzumab treatment, 13 patients developed conditions of NYHA class II or higher in the placebo group compared with 8 patients in the candesartan group (P = .36) (see eTable 6 in Supplement 2). A greater proportion of patients in the candesartan group (24.8%; 26 of 105) than in the placebo group (15.5%; 16 of 103) had serious adverse events, but the difference was not significant (P = .12). There were no suspected unexpected serious adverse reactions reported.
The Ala1170Pro homozygous genotype (Ala/Ala) was, for all study participants, associated with a lower likelihood of the development of a cardiac event than Pro/Pro + Ala/Pro genotypes (OR, 0.09; 95% CI, 0.02-0.45; P = .003) (Table). The other SNPs were not associated with cardiac events (eTable 7 in Supplement 2).
In the candesartan group, the median baseline NT-proBNP value was 62 (range 35-106) pmol/L, and the median hs-TnT value was 14 (range 7-20) µg/L. In the placebo group the median baseline NT-proBNP value was 70 (range 44-123) pmol/L and the median hs-TnT value was 13 (range 7-18) µg/L. No association was found between LVEF and either NT-proBNP (P = .51) or hs-TnT (P = .78) levels at the same time points (eg, before, during, or after completion of treatment (see eFigure 3 and eFigure 4 in Supplement 2). The risk of cardiac events was not associated with baseline NT-proBNP and hs-TnT levels. No association was found between the NYHA classification and NT-proBNP and hs-TnT levels (P = .13 and P = .43, respectively) during trastuzumab treatment.
The results of this trial do not support the hypothesis that the concurrent use of candesartan and trastuzumab in early-stage breast cancer after treatment with anthracyclines prevents or ameliorates the development of trastuzumab-related cardiotoxic effects, as measured by serial LVEF measurements during 92 weeks. Daily treatment with candesartan also did not affect plasma levels of the biomarkers NT-proBNP and hs-TnT.
This lack of effect could possibly be explained by the timing of ARB administration. Angiotensin II, an effector peptide of the renin-angiotensin system, plays a role in the pathophysiology of hypertension by mediating the development of cardiac hypertrophy and left ventricular remodelling.28,29 Preclinical and clinical trials on the ARB telmisartan showed evidence of protection against anthracycline-induced cardiotoxic effects.17,30 Anthracyclines can induce myocardial oxidative stress, which can lead to myocyte cell death and HER2 upregulation. Trastuzumab binds to the extracellular domain of HER2 expressed on the myocardial cells, which plays an important role in compensatory cardiac hypertrophy, leading to an insufficient compensatory mechanism. Although trastuzumab-related decrease in systolic cardiac function occurs without coadministered anthracyclines, in the present study we hypothesized that angiotensin II blockade by candesartan would prevent or ameliorate the effect of trastuzumab on the heart. The effect of this intervention would probably have been different if candesartan treatment had begun simultaneously with administration of anthracyclines and before trastuzumab. Data suggest that starting an intervention sooner after initiation of cardiac damage occurs may result in better outcome.31 In our study, patients with a history of congestive heart failure or baseline LVEF below 50% were excluded. Moreover, most patients (81%) started with a baseline LVEF value of at least 55%. It is therefore possible that AT1 receptor blockade as a preventive measure could have a more pronounced effect on trastuzumab-related damage in high-risk patient populations.
Another potential cause of lack of effect is the rather insensitive assessment of only systolic cardiac function parameter LVEF, which is not sensitive enough to capture some intervention effects.32 In current clinical practice, multiple gate acquisition or echocardiography measurements are the most commonly used methods for assessing cardiac function by means of LVEF. More sensitive parameters that estimate subclinical changes such as tissue velocities, strain, and diastolic function33 were not assessed in this study, meaning that smaller toxic effects of trastuzumab may not have been detected by changes in LVEF. Several studies on real-time 3-dimensional echocardiography and strain rate imaging have shown promising results in detecting earlier subclinical changes in cardiac performance.34-36 Moreover, cardiovascular magnetic resonance imaging has detected significant functional changes in left and right ventricular function in anthracycline- and trastuzumab-treated patients.37 However, no parameters validated for trastuzumab-treated patients were available at the time this trial was conducted. Recent data suggest that trastuzumab treatment after anthracycline therapy has more subtle effects on cardiac damage parameters lasting for several years at least.25
A formal comparison of trastuzumab-related cardiotoxic effects among this study and previously published adjuvant trastuzumab trials is hampered by differences in the applied definition of cardiotoxic effects. The lack of a universally used definition of trastuzumab-related cardiotoxic effects is a methodological limitation in this field of research.
The sample size was calculated based on initial published data, which suggested an incidence of trastuzumab-related cardiotoxic effects during treatment of 30%.2,24 Later reports have suggested a much lower incidence.5,6,38 Although the relative incidence of decreases in LVEF (the primary cardiac end point) was higher in this study than in large adjuvant trials, the absolute number of cardiac end points was still rather low. But with 103 participants per group, this study had more than 80% power to detect a decrease of the event incidence from 16% to 4%. Hence, a smaller difference between both groups could have been detected. In the present study, the cardiac markers were not associated with decreases in LVEF and did not predict or detect cardiotoxic effects, even though clinical data suggest that increased NT-proBNP levels are predictive for the development of chemotherapy- cardiotoxic effects.19,39 However, to our knowledge, no association between changes in NT-proBNP and LVEF values during trastuzumab treatment has been reported to date.40,41 Several clinical studies have demonstrated an association between changes in troponin I levels and the risk of trastuzumab-related cardiotoxic effects.18,41,42 In contrast, 2 other studies43,44 detected no association between changes in troponin and LVEF values, consistent with our study results.
With the aim of identifying patients who are susceptible to trastuzumab-related cardiotoxic effects, we investigated several candidate SNPs. We observed an association between the germline proline allele of the Ala1170Pro SNP and lower risk of a cardiac event. This finding is consistent with the results of a study by Stanton et al45 but in contrast with a retrospective cohort study in 73 patients with breast cancer in whom no association of the germline Ala1170Pro SNP and trastuzumab-related cardiotoxic effects was found.23 However, a major limitation of that study was the small sample size. Furthermore, baseline LVEF values were not available in 35.9% of the patients. In our study, the consistent study-related assessments indicated a coherent relationship between trastuzumab treatment, predictive and detective measurements, and cardiac events. We found no association between the SNPs of ERBB2 Ile654Val and Ile655Val and FcgR2a His166Arg, FcgR2B Ile232Thr and FcgR3A Val158Phe and trastuzumab-related cardiotoxic effects. Two previous studies reported an association,22,23 and 2 published abstracts reported no association45-47 between Ile655Val polymorphism and trastuzumab-related cardiotoxic effects. In addition the results of gene polymorphisms of ERBB2 and their effect on trastuzumab-related cardiotoxic effects are contradictory. Therefore, future trials are warranted to explore relationships between these SNPs and trastuzumab-related cardiotoxic effects.
As evaluated by serial measurements of LVEF during 1 year of trastuzumab treatment and 40 weeks thereafter, treatment with candesartan did not prevent trastuzumab-related cardiotoxic effects, nor did it affect other markers of cardiac damage like NT-proBNP and hs-TnT. Consequently, our study does not support a role for the use of this drug as scheduled in conjunction with trastuzumab in this population of patients with breast cancer.
The ERBB2-germline Ala1170Pro single nucleotide polymorphism may be used to identify patients who are at increased risk of trastuzumab-related cardiotoxic effects.
Corresponding Author: Jan H. M. Schellens, MD, PhD, Division of Clinical Pharmacology, Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX, Amsterdam, the Netherlands (firstname.lastname@example.org).
Accepted for Publication: January 27, 2016.
Published Online: June 23, 2016. doi:10.1001/jamaoncol.2016.1726.
Author Contributions: Dr Werkhoven had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Boekhout and Gietema contributed equally to this work.
Study concept and design: Boekhout, Gietema, Mandigers, Smilde, de Vries, Schellens.
Acquisition, analysis, or interpretation of data: Boekhout, Gietema, Milojkovic Kerklaan, van Werkhoven, Altena, Honkoop, Los, Smit, Nieboer, Smorenburg, Mandigers, van der Wouw, Kessels, van der Velden, Ottevanger, de Boer, van Veldhuisen, Kema, de Vries, Schellens.
Drafting of the manuscript: Boekhout, Gietema, van Werkhoven, Smilde, Schellens.
Critical revision of the manuscript for important intellectual content: Boekhout, Gietema, Milojkovic Kerklaan, van Werkhoven, Altena, Honkoop, Los, Smit, Nieboer, Smorenburg, Mandigers, van der Wouw, Kessels, van der Velden, Ottevanger, Smilde, de Boer, van Veldhuisen, Kema, de Vries, Schellens.
Statistical analysis: Boekhout, Gietema, van Werkhoven, Schellens.
Obtained funding: Schellens.
Administrative, technical, or material support: Boekhout, Milojkovic Kerklaan, Altena, Honkoop, Los, Smit, Mandigers, van der Wouw, van der Velden, Ottevanger, de Boer, van Veldhuisen, Kema, Schellens.
Study supervision: Boekhout, Gietema, Milojkovic Kerklaan, Mandigers, Smilde, de Boer, van Veldhuisen, de Vries, Schellens.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by unrestricted financial contributions from the Netherlands Cancer Institute and Roche; AstraZeneca provided the study drug and placebo.
Role of the Funder/Sponsor: The funding institution and corporations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We are very thankful for the substantial and uncompensated contributions to this trial in the following areas: statistical analysis, the late Dr O. Dalesio, MSc, Department of Biometrics, the Netherlands Cancer Institute; data collection, Edward Fiets, MD, PhD, Medical Center, Leeuwarden, the Netherlands, Marcel Soesan, MD, Slotervaart Hospital, the Netherlands, Pieter-Paul Schiphorst, MD, Beatrix Hospital, Winterswijk, the Netherlands, Bart de Valk, MD, PhD, Onze Lieve Vrouwe Gasthuis, Amsterdam, the Netherlands, Vera Lustig, MD, PhD, Flevo Hospital, Almere, the Netherlands, Brigitte Dufourny, PhD, Patricia Hagen, and Joline Claassen, Division of Clinical Pharmacology, Netherlands Cancer Institute, Amsterdam, the Netherlands, Artur Burylo, Department of Molecular Pathology, the Netherlands Cancer Institute, Amsterdam, the Netherlands; conception of design, analysis of the data, Dick Richel, MD, PhD, Academic Medical Center, Amsterdam, the Netherlands, Guus Hart, MSc, Biometrics Department, Netherlands Cancer Institute, Amsterdam, the Netherlands; obtaining funding: Sietse van Steenis and Astrid Timmerhuis, Roche, the Netherlands, Manfred Marang, AstraZeneca, the Netherlands, Anders Ljunggren, AstraZeneca, Sweden; pharmacy, Jos Beijnen, PhD, Kees Koks, PhD, and Alwin Huitema, PhD, Division of Pharmacy and Pharmacology, Netherlands Cancer Institute, Amsterdam, the Netherlands.