Author + information
- Received December 19, 2018
- Revision received March 19, 2019
- Accepted April 2, 2019
- Published online August 26, 2019.
- Shom Goel, MBBS, PhDa,b,c,d,∗∗ (, )
- Jia Liu, MD, PhDb,∗,
- Hao Guo, MSa,
- William Barry, PhDa,
- Richard Bell, MBBSe,
- Bronwyn Murray, RNb,
- Jodi Lynch, MBBSf,g,
- Patricia Bastick, MBBSf,
- Lorraine Chantrill, MBBS, PhDh,
- Belinda E. Kiely, MBBS, PhDi,
- Ehtesham Abdi, MBBSj,
- Josie Rutovitz, MBChBk,
- Ray Asghari, MBBSl,
- Anne Sullivan, MBBSi,
- Michelle Harrison, MBBSb,
- Maija Kohonen-Corish, MSc, PhDm and
- Jane Beith, MBBS, PhDb
- aDepartment of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts
- bDepartment of Medical Oncology, Chris O’Brien Lifehouse, Sydney, Australia
- cPeter MacCallum Cancer Centre, Melbourne, Australia
- dSir Peter MacCallum Department of Oncology, University of Melbourne, Melbourne, Australia
- eBarwon Health Cancer Services, Andrew Love Cancer Centre, Geelong, Australia
- fSt. George Cancer Care Centre, St. George Hospital, Sydney, Australia
- gDepartment of Medical Oncology, Sutherland Hospital, Sydney, Australia
- hMacarthur Cancer Therapy Centre, Liverpool Hospital, Sydney, Australia
- iConcord Cancer Centre, Concord Repatriation General Hospital, Sydney, Australia
- jThe Tweed Hospital, Tweed Heads & Griffith University, Gold Coast, Australia
- kNorthern Haematology and Oncology Group, San Integrated Cancer Centre, Sydney, Australia
- lBankstown Cancer Centre, Bankstown Hospital, Sydney, Australia
- mKinghorn Cancer Centre, Garvan Institute of Medical Research, Sydney, Australia
- ↵∗Address for correspondence:
Dr. Shom Goel, Peter MacCallum Cancer Centre and University of Melbourne, 305 Grattan Street, Melbourne, VIC 3124, Australia.
Objectives The aim of CATS (Cardiotoxicity of Adjuvant Trastuzumab Study) was to prospectively assess clinical, biochemical, and genomic predictors of trastuzumab-related cardiotoxicity (TRC).
Background Cardiac dysfunction is a common adverse effect of trastuzumab. Studies to identify predictive biomarkers for TRC have enrolled heterogeneous populations and yielded mixed results.
Methods A total of 222 patients with early-stage human epidermal growth factor receptor 2–positive breast cancer scheduled to receive adjuvant anthracyclines followed by 12 months of trastuzumab were prospectively recruited from 17 centers. Left ventricular ejection fraction (LVEF), troponin T, and N-terminal prohormone of brain natriuretic peptide were measured at baseline, post-anthracycline, and every 3 months during trastuzumab. Germline single-nucleotide polymorphisms in ERBB2, FCGR2A, and FCGR3A were analyzed. TRC was defined as symptomatic heart failure; cardiac death, arrhythmia, or infarction; a decrease in LVEF of >15% from baseline; or a decrease in LVEF of >10% to <50%.
Results TRC occurred in 18 of 217 subjects (8.3%). Lower pre-anthracycline LVEF and greater interval decline in LVEF from pre- to post-anthracycline were each associated with TRC on multivariate analyses (odds ratio: 3.9 [p = 0.0001] and 7.9 [p < 0.0001] for a 5% absolute change in LVEF). Higher post-anthracycline N-terminal prohormone of brain natriuretic peptide level was associated with TRC on univariate but not multivariate analyses. There were no associations between troponin T or ERBB2/FGCR polymorphisms and TRC. Baseline LVEF and LVEF change post-anthracycline were used to generate a “low-risk TRC score” to identify patients with low TRC incidence.
Conclusions Low baseline LVEF and greater LVEF decline post-anthracycline were both independent predictors of TRC. The other biomarkers did not further improve the ability to predict TRC. (Cardiotoxicity of Adjuvant Trastuzumab [CATS]; NCT00858039)
Trastuzumab, a monoclonal antibody directed against the human epidermal growth factor receptor 2 (HER2), has revolutionized the management of HER2-overexpressing breast cancer (1). Although trastuzumab is generally well tolerated, its major adverse effect is a dose-independent, reversible cardiotoxicity. The pathophysiology of trastuzumab-related cardiotoxicity (TRC) is distinct from that of anthracycline-induced cardiotoxicity but likely relates to suppression of cardiomyocyte HER2 signaling (2). The fact that lapatinib, a more potent inhibitor of HER2 signaling, does not cause cardiotoxicity suggests that other mechanisms, such as trastuzumab-induced antibody-dependent cellular cytotoxicity (ADCC) against cardiomyocytes, might also be operative (3). Because of the high incidence of TRC in the early trials in patients with metastatic breast cancer (4), guidelines for routine on-treatment monitoring of cardiac function every 3 months were introduced for adjuvant trials and are currently recommended as standard of care (5). Notably, the prevalence of trastuzumab-associated major adverse cardiac events in patients receiving contemporary therapeutic regimens is lower than initially reported (6,7).
Risk prediction of TRC is an important clinical goal. High-risk patients might benefit from early intervention with cardioprotective agents to prevent declines in left ventricular ejection fraction (LVEF). Conversely, it is conceivable that the frequency of cardiac monitoring could be reduced for patients at very low risk for TRC. Because elevation of serum cardiac troponin has been reported to be a predictive marker for chemotherapy-induced cardiomyopathy (8), several studies have investigated the value of serum biomarkers of myocyte injury (troponin T [TnT] and troponin I) and left ventricular wall tension (N-terminal prohormone of brain natriuretic peptide [NT-proBNP]) as markers for preclinical TRC. Patient populations in these studies have often been heterogeneous or small, however, and their findings have been mixed, leaving uncertainty around their utility as predictors of TRC (5).
In addition, other groups have explored the predictive capacity of germline genomic polymorphisms as predictors of TRC, with mixed results. For example, single-nucleotide polymorphisms (SNPs) in ERBB2 (Val655Ile [9–11] and Pro1170Ala ) have been associated with the development of TRC in some, but not all, prior clinical studies. Moreover, given that ADCC plays a role in the pathogenesis of TRC, polymorphisms in Fc-gamma receptors (FCGRs), which can modulate the efficiency of ADCC, might also be implicated. However, 2 small studies failed to detect an association of FCGR2A His131Arg and FCGR3A Val158Phe polymorphisms with TRC (9,13), and the role of these SNPs in mediating TRC remains unclear.
The aims of CATS (Cardiotoxicity of Adjuvant Trastuzumab Study) were to determine if baseline LVEF, changes in LVEF pre- and post-anthracycline, levels of serum biomarkers (TnT and NT-proBNP), or polymorphisms in ERBB2 and FCGR genes are associated with the risk for subsequent TRC in patients receiving adjuvant anthracycline-based chemotherapy followed by trastuzumab for early-stage HER2-positive breast cancer.
CATS was a multicenter prospective cohort study that recruited women with early-stage HER2-positive breast cancer older than 18 years from 17 Australian institutions. All participants were Eastern Cooperative Oncology Group status 0 to 2 and scheduled to receive conventional adjuvant anthracycline-based chemotherapy followed by taxane chemotherapy given with trastuzumab, followed by trastuzumab alone to complete a total of 52 weeks (Online Figure 1). Exclusion criteria included baseline (pre-anthracycline) LVEF <50%, pregnancy, and prior systemic chemotherapy. Human research ethics committee approval was obtained at all sites and all participants provided written informed consent.
LVEF was measured using either transthoracic echocardiography or multigated acquisition scan at baseline, after the completion of anthracycline (visit 2), and every 3 months during trastuzumab (visits 3 to 6) (Online Figure 1). For each patient, the modality used to assess LVEF was the same throughout the study. TRC was defined using similar criteria to the major adjuvant trials (14,15) as any of the following occurring during trastuzumab treatment: 1) death due to cardiac failure, myocardial infarction, or arrhythmia; 2) grade III or IV cardiac arrhythmia or ischemia or infarction (National Cancer Institute Common Terminology Criteria version 3.0); 3) New York Heart Association functional class III or IV heart failure; 4) an asymptomatic decrease in LVEF of >15%; and 5) an asymptomatic decrease in LVEF of >10% to an absolute value of <50%.
In the event of TRC, decisions to discontinue trastuzumab and/or institute cardiac therapy were made at the discretion of the treating oncologist, cardiologist, and patient. Once a TRC event occurred, data collected from that subject in subsequent visits (including LVEF and biomarkers) were not included in the analyses.
Biomarker and polymorphism assessment
Plasma TnT and NT-proBNP were measured at visits 1 to 4 inclusive. All samples were batched and analyzed centrally using the Elecsys electrochemiluminescence immunoassay (Roche Cobas 6000, Roche Diagnostics, Basel, Switzerland). TnT assay range was 3 to 10,000 ng/l, and values below the detection range were reported as <3 ng/l. Elevated TnT was defined as ≥3 ng/l. The absolute NT-proBNP value was recorded in picomoles per liter.
Genomic deoxyribonucleic acid was extracted from the buffy coat of peripheral blood samples collected at baseline using the ISOLATE II Blood DNA Kit (Bioline, Eveleigh, Australia). TaqMan SNP assays for FCGR3A 158V/F (rs396991), FCGR2A 131H/R (rs1801274), ERBB2 V655I (rs1136201), and ERBB2 P1170A (rs1058808) of the ERBB2 gene were purchased from Life Technologies (Carlsbad, California). Sequencing reactions were performed centrally on the ABI7900 (Life Technologies) and analyzed in triplicate. Where allelic discrimination was ambiguous, restriction fragment length polymorphism was used to cross-validate.
Logistic regression was used to examine whether baseline demographics and clinical factors were associated with subsequent development of TRC. Univariate logistic regressions were performed first. Factors with p values <0.20 in the univariate model were included in multivariate models. A backward regression procedure was used for model selection. The mixed-effect model was used to assess changes in LVEF and NT-proBNP over time whereby time of assessment and whether a patient developed TRC were considered as fixed effects, and patients were entered as a random effect. The parameters that demonstrated low incidence were evaluated using the Fisher exact test. We logarithmically transformed NT-proBNP values to achieve normality. Receiver-operating characteristic (ROC) curves were used to assess the performance of linear predictors of risk for TRC, using factors significantly associated with TRC.
We analyzed germline SNPs using both dominant and recessive models. For each gene, we defined the variant or minor allele as follows: valine for ERBB2 V655I, proline for ERBB2 P1170A, arginine for FCGRA2 H131R, and phenylalanine for FCGR3A V158F. All tests were conducted with 2-sided α value of 0.05 using SAS version 9.4 (SAS Institute, Cary, North Carolina).
A total of 222 participants were recruited, and 217 were eligible for inclusion (Online Figure 2). Reasons for exclusion included loss to follow-up (n = 3) and incomplete baseline and follow-up data (n = 2). Table 1 shows the baseline characteristics of all participants, those with TRC, and those without. Eighty participants received 4 cycles of doxorubicin, and 137 patients received 3 cycles of epirubicin.
The incidence of TRC was 8.3% (n = 18). There were no significant differences in the baseline characteristics of the participants who developed TRC compared with those who did not (Table 1). TRC events included a >10% decrease of LVEF to an absolute level <50% (n = 14), New York Heart Association functional class III or IV heart failure symptoms (n = 1), and cardiac arrhythmia or infarction (n = 3). There were no deaths due to cardiac failure, myocardial infarction, or arrhythmia. Of the 14 patients with TRC due to LVEF decrease, 8 (67%) had LVEF recovery (2 or more subsequent LVEF assessments of 50% or greater), 3 patients did not have recovery of LVEF, and 3 patients did not have adequate LVEF follow-up to assess for recovery, because TRC occurred late during treatment in 2, and trastuzumab was permanently ceased in 1 patient because of progression of the breast cancer. Clinical recovery occurred in the 3 patients with arrhythmia or infarction and 1 patient with New York Heart Association functional class III or IV heart failure symptoms with institution of cardioprotective medications.
Seven participants had trastuzumab ceased permanently, and 11 had trastuzumab withheld temporarily. The timing of the first TRC event and type of event is shown in Online Figure 3. Notably, 10 of 18 first TRC events (56%) occurred by visit 3 (3 months after trastuzumab commencement). Given the small number of TRC events occurring after visit 3, we restricted our analysis of the predictive value of TnT and NT-proBNP levels to samples drawn prior to visit 3 (i.e., at visits 1 and 2).
Association between pre-trastuzumab LVEF and development of TRC
In the entire population, LVEF decreased over the observation period (p < 0.0001), but there was no significant difference between LVEF measurements taken pre- and post-anthracycline (mean change 0%; range −18% to +18%). As expected, a mixed-effect model showed that estimates for the least square means for LVEF were lower in those who developed TRC compared with those who did not (p < 0.0001) (Figure 1A). Importantly, these differences in LVEF were apparent before the occurrence of TRC events (because LVEF data subsequent to the development of TRC were excluded in this model).
Mean baseline LVEF was 61.3% in the TRC group compared with 64.5% in participants without TRC. The absolute mean LVEF for patients with TRC compared with those without TRC is shown in Figure 1B and in the Central Illustration. On univariate analysis, a lower baseline LVEF was associated with an increased risk for TRC (odds ratio [OR]: 1.6; 95% confidence interval [CI]: 1.0 to 2.4; p = 0.04) (Table 2). A lower post-anthracycline LVEF was also associated with a greater risk for TRC (OR: 3.5; 95% CI: 2.0 to 6.3; p < 0.0001) (Table 2). Furthermore, a greater absolute decrease in LVEF from baseline to post-anthracycline was associated with an increased risk for TRC (OR: 2.7; 95% CI: 1.5 to 4.9; p = 0.0007). Importantly, the significant associations between both baseline LVEF and interval change in LVEF were upheld on multivariate analysis (Table 2). Post-anthracycline LVEF values were excluded from the multivariate analysis because they were highly correlated with both baseline LVEF and LVEF change and would thus bring multicollinearity into a multivariate analysis. Most notably, when baseline LVEF is fixed, the odds of TRC increased substantially for every 5% decrease of LVEF from baseline to post-anthracycline (OR: 7.9; 95% CI: 3.2 to 19.7; p < 0.0001).
Association between NT-proBNP levels and development of TRC
NT-proBNP levels increased over time across the cohort (p < 0.0001). The median NT-proBNP level was 7 pmol/l (range 1 to 122 pmol/l) at baseline and 10 pmol/l (range 1 to 66 pmol/l) post-anthracycline. The least squares means for baseline NT-proBNP were similar in participants with TRC compared with those without, but were higher in TRC subjects throughout the remainder of the study period (p < 0.0001) (Figure 1C). Baseline NT-proBNP was not associated with TRC (Table 3). However, a higher NT-proBNP level post-anthracycline (OR: 2.0; 95% CI: 1.0 to 3.7; p = 0.04) and a doubling in NT-proBNP from baseline to post-anthracycline (OR: 2.3; 95% CI: 1.1 to 4.8; p = 0.03) were each associated with TRC on univariate analyses (Table 3).
Association between TnT and development of TRC
The proportion of patients with positive TnT (≥3 ng/l) increased significantly from baseline (n = 57 [26%]) to post-anthracycline (n = 155 [71%]) (p < 0.001). Considering all results with detectable TnT, the median level was 7 ng/l (range 3 to 24 ng/l) at baseline and 11 ng/l (range 3 to 132 ng/l) post-anthracycline. However, an increase in TnT (from undetectable to detectable) between baseline and post-anthracycline levels was not associated with the subsequent development of TRC (p = 0.07) (Table 3).
Associations between SNPs in ERBB2 and FCGR genes and TRC
We extracted germline DNA from the peripheral blood of 192 subjects, 16 of whom developed TRC. All SNP genotypes were in Hardy-Weinberg equilibrium (ERBB2 V665I; p = 0.14; ERBB2 P1170A; p = 0.92; FCGR3A V159F; p = 0.66) except for FCGR2A H131R (p = 0.03). There were no significant relationships between SNP frequencies and the incidence of TRC (Figure 2).
To identify factors independently associated with the development of TRC, we conducted a multivariate analysis including all characteristics associated with TRC (defined as p ≤ 0.20 on univariate analyses), as shown in Tables 1, 2, and 3. The included characteristics were family history of ischemic heart disease, trastuzumab schedule, taxane schedule, LVEF method, baseline LVEF, post-anthracycline LVEF, absolute decrease in LVEF from baseline to post-anthracycline, post-anthracycline NT-proBNP, doubling from baseline to post-anthracycline NT-proBNP, and TnT. The final model revealed only baseline LVEF (OR: 3.9; 95% CI: 2.0 to 7.8; p = 0.0001) and the absolute decrease in LVEF from baseline to post-anthracycline (OR: 7.9; 95% CI: 3.2 to 19.7; p < 0.0001) as significant predictors of TRC. All other features that were significant on univariate analyses (including NT-proBNP level post-anthracycline and a doubling in NT-proBNP level from pre- to post-anthracycline measurements) were not significantly associated with TRC on multivariate analyses.
Development of a low TRC risk score
ROC analysis was performed to identify a cutoff threshold for the linear combination of variables that classifies a subset of patients with low TRC risk. Baseline LVEF and change in LVEF from baseline to post-anthracycline were used as predictors, with an area under the curve of 0.87 (95% CI: 0.77 to 0.96) (Figure 3). In an exploratory manner, we used this curve to identify a patient population at particularly low risk for TRC. We selected a reasonable cutoff defining high specificity for not developing TRC as 94% (Figure 3). At this cutoff point, the sensitivity is 42%. Using this ROC curve, CATS generated a “low TRC risk score”: 3 × (baseline LVEF) − 4.3 × (LVEF change) >201%, where baseline LVEF is LVEF pre-anthracycline, and LVEF change is baseline LVEF minus post-anthracycline LVEF.
If a patient’s LVEF at baseline and post-anthracycline fulfills this equation, he or she could be considered at “low TRC risk,” with a predicted probability of developing TRC of only 1.2%. In our cohort, 66 patients would be categorized as “low risk” using this score.
CATS analyzed early changes in LVEF, serum cardiac biomarkers, and germline SNPs as predictors of TRC in patients receiving adjuvant anthracyclines followed by trastuzumab for early-stage HER2-positive breast cancer. We found that both baseline LVEF and the magnitude of decline in LVEF between pre- and post-anthracycline measurements are each independently predictive of future TRC. Furthermore, these values might have the potential to be used together to identify patients with very low TRC risk, for whom the frequency of LVEF monitoring during adjuvant trastuzumab might be safely reduced. Although early changes in NT-proBNP levels were statistically associated with development of TRC, this association was not upheld on multivariate analyses. Finally, SNPs assessed in ERBB2, FCGR3A, and FCGR2A were not predictive of TRC.
Comparison with current published research
A number of groups have attempted to identify predictors of TRC in patients with HER2-positive breast cancer (8,13,16–23). The majority of these have been small (n = <100), conducted in heterogeneous populations (e.g., mixed adjuvant and metastatic patients, often with multiple comorbidities that could confound results), or acquired biomarker specimens at irregular time points. Indeed, with the exception of a substudy analysis of the HERA (Herceptin Adjuvant) trial (23), CATS is the largest such study conducted prospectively in an adjuvant population. As such, we believe that the CATS results help clarify some of the confusion that has arisen from interpretation of heterogeneous studies and that is reflected in differing recommendations for the use of biomarkers to monitor for TRC (5,24).
In agreement with previous reports, we found that TRC often occurs early, within the first 3 months of trastuzumab treatment and that a lower pre-anthracycline LVEF is a predictor of its occurrence (8,13,25). However, the finding that an asymptomatic decrease in LVEF from pre- to post-anthracycline predicts subsequent TRC is, to our knowledge, novel. Previous studies did not routinely capture pre-anthracycline LVEF and were thus were not designed to analyze this parameter (13,23). In CATS, multivariate analysis showed that a 5% decline in LVEF from the pre- to post-anthracycline measurements increases the risk for future TRC (OR: 7.9). On the basis of these data, we speculate that the combination of a patient’s baseline LVEF and the degree of decline in LVEF after anthracycline therapy could be useful tools to stratify their risk for TRC. Indeed, our ROC analysis suggests that these parameters may be used in a predictive “low-risk TRC score” to identify a group of subjects at particularly low risk for TRC (≤1.2%). We propose that the low-risk TRC score may potentially complement but not replace clinical judgement. For example, patients with higher baseline LVEFs (e.g., 75%) who develop small decreases in LVEF post-anthracycline (e.g., to 70%) would be classified as low risk. Alternatively, patients with moderate LVEFs at baseline (e.g., 65%) who are found to have increases in LVEF after anthracycline treatment (e.g., to 67%) would also be classified as low risk. Further evaluation of this predictive model in independent cohorts is being planned to validate this observation in an unbiased manner. If the model were upheld, it would help define a patient population in whom LVEF monitoring during adjuvant trastuzumab could be performed less frequently, potentially saving time and resources while reducing radiation exposure associated with multigated acquisition scans.
Findings from CATS do not support a role for measuring serum cardiac biomarkers to predict TRC, because analyses showed that they do not provide additional predictive power beyond the LVEF parameters described earlier. Previous studies examining such markers have been inconsistent, with 5 of 11 reporting an association between TRC and elevated troponin (8,19,20,23,26), and 2 of 10 for NT-proBNP (16,23,27). The inconsistency likely stems from intertrial heterogeneity with respect to patient populations studied, sample sizes, uniformity of sample collection timing, definition of TRC, and the limits of detection of assays used (27,28). In addition, many studies did not specifically examine the ability of serum marker elevations to predict future TRC events but also considered changes detected synchronously with TRC as “predictive.” In analysis of CATS, we only considered marker elevations detected prior to TRC detection. For example, we found that elevated serum biomarkers (NT-proBNP) post-anthracycline was associated with TRC on univariate, but not multivariate, analyses. In contrast, the largest prospective study to date, the HERA cardiac substudy (n = 310 patients receiving adjuvant trastuzumab) (23), reported that elevated troponin detected prior to the introduction of adjuvant trastuzumab was associated with a 2.4- to 4.5-fold increased risk for cardiotoxicity on multivariate analysis, but this included TRC events that occurred synchronously with elevations in troponin levels.
We evaluated germline polymorphisms in ERBB2 because smaller prospective studies have reported an association between these SNPs and TRC. We analyzed FCGR gene polymorphisms because it is plausible that ADCC plays a role in the pathogenesis of TRC (29) and has been shown to predict the anticancer efficacy of trastuzumab. In our analysis, we found no association between ERBB2 or FCGR gene polymorphisms and TRC risk. The strongest evidence from other reports exists for the ERBB2 Ile655Val polymorphism, whereby the heterozygous genotype (Ile/Val) was found to be associated with a 5.4-fold increased risk for TRC compared with Ile/Ile patients (95% CI: 2.6 to 11.7; p < 0.0001) (10). Unfortunately, previous studies are limited in sample size, and we believe that our data support current guidelines that do not recommend routine genotyping of patients prior to trastuzumab.
First, the relatively small number of TRC events limits our statistical power to exclude an independent predictive role of TnT or NT-proBNP. Notably, our TRC event rate would have been higher (14%) if we had used the Cardiac Review and Evaluation Committee criteria rather than the HERA criteria to define TRC. It is unlikely that using different criteria would lead us to draw meaningfully different conclusions, however, and if anything this highlights the need for consistency in future biomarker studies.
Second, the early onset of TRC limited our ability to examine the effect of biomarker changes after patients had completed 3 months of adjuvant trastuzumab.
Third, 53% of participants had baseline LVEF measured using multigated acquisition, while 47% had LVEF measured using transthoracic echocardiography, reflecting contemporary clinical practice, and interassay variability and operator dependency of transthoracic echocardiography may have affected our findings.
The value of the CATS study is 2-fold. First, we present data detailing a novel, independent, and powerful predictor of TRC in the adjuvant setting: a decline in LVEF from pre- to post-anthracycline measurements. This finding suggests that anthracycline-induced cardiotoxicity induces or unmasks subclinical myocardial injury associated with LVEF reduction, which predisposes to subsequent TRC. Indeed, preclinical and clinical data support this notion (4,30) and emphasize the importance of monitoring LVEF at both the pre- and post-anthracycline visit. Furthermore, our data are consistent with the notion that multiple “hits” to the cardiovascular system during early breast cancer therapy can collectively reduce a patient’s cardiovascular reserve capacity and conspire to induce objective cardiac dysfunction (31). Specifically, the concept that anthracycline-induced cardiac dysfunction (measured through early declines in LVEF) increases the risk for subsequent TRC is well supported by the cardiovascular reserve capacity conceptual paradigm.
Second, by recruiting a relatively large population treated homogeneously and monitored prospectively, we believe that our data helps to clarify inconsistencies from previous studies and suggests that there is currently no role for routine serum biomarker or genotype screening to identify patients with increased susceptibility to TRC.
COMPETENCY IN MEDICAL KNOWLEDGE: Previous studies of TRC biomarkers provide conflicting results. In this prospective cohort study of clinical, serum, and genomic biomarkers for TRC, the magnitude of LVEF decline from pre- to post-adjuvant anthracycline was the strongest independent predictor of TRC. Serum and genetic biomarkers did not add further predictive value and should not be measured in asymptomatic patients receiving trastuzumab in routine clinical practice.
TRANSLATIONAL OUTLOOK: There is a need to better risk-stratify for TRC. We propose a novel “risk score” to identify patients at low risk for TRC. This algorithm might be used to identify a subset of patients at such low risk for TRC that the frequency of LVEF monitoring during adjuvant trastuzumab could be safely reduced. Validation in independent cohorts is under way.
↵∗ Drs. Goel and Liu are joint first authors.
This project was funded by the National Breast Cancer Foundation of Australia (Novel Concept Award NC-08-03). Dr. Goel has received laboratory research funding from Eli Lilly; conducts clinical research sponsored by Eli Lilly and Novartis; and has served on advisory boards for Eli Lilly, Novartis, and G1 Therapeutics. Dr. Liu has received clinical research support from AstraZeneca. Dr. Chantrill serves on the advisory board for Merck, Sharp & Dohme; and has received travel reimbursement from Amgen Australia. Dr. Harrison serves on an advisory board for AstraZeneca; has received travel reimbursement from AstraZeneca; and has received paid honoraria from AstraZeneca and Roche. Dr. Beith serves on advisory boards for Roche, Novartis, Pfizer, and Lilly. Dr. Sullivan has passed away and is unable to provide a disclosure statement. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- antibody-dependent cellular cytotoxicity
- confidence interval
- Fc-gamma receptor
- human epidermal growth factor receptor 2
- left ventricular ejection fraction
- N-terminal prohormone of brain natriuretic peptide
- odds ratio
- receiver operating characteristic
- single-nucleotide polymorphism
- troponin T
- trastuzumab-related cardiotoxicity
- Received December 19, 2018.
- Revision received March 19, 2019.
- Accepted April 2, 2019.
- 2019 American College of Cardiology Foundation
- Mohan N.,
- Shen Y.,
- Endo Y.,
- ElZarrad M.K.,
- Wu W.J.
- Armenian S.H.,
- Lacchetti C.,
- Lenihan D.
- Moja L.,
- Tagliabue L.,
- Balduzzi S.,
- et al.
- Suter T.M.,
- Procter M.,
- van Veldhuisen D.J.,
- et al.
- Cardinale D.,
- Colombo A.,
- Torrisi R.,
- et al.
- Roca L.,
- Dieras V.,
- Roche H.,
- et al.
- Stanton S.E.,
- Ward M.M.,
- Christos P.,
- et al.
- Boekhout A.H.,
- Gietema J.A.,
- Milojkovic Kerklaan B.,
- et al.
- Perez E.A.,
- Suman V.J.,
- Davidson N.E.,
- et al.
- Perik P.J.,
- Lub-De Hooge M.N.,
- Gietema J.A.,
- et al.
- Fallah-Rad N.,
- Walker J.R.,
- Wassef A.,
- et al.
- Morris P.G.,
- Chen C.,
- Steingart R.,
- et al.
- Ky B.,
- Putt M.,
- Sawaya H.,
- et al.
- Matos E.,
- Jug B.,
- Blagus R.,
- Zakotnik B.
- Yu A.F.,
- Manrique C.,
- Pun S.,
- et al.
- Zardavas D.,
- Suter T.M.,
- Van Veldhuisen D.J.,
- et al.
- Zamorano J.L.,
- Lancellotti P.,
- Rodriguez Munoz D.,
- et al.
- Romond E.H.,
- Jeong J.H.,
- Rastogi P.,
- et al.
- Sawaya H.,
- Sebag I.A.,
- Plana J.C.,
- et al.
- Witteles R.M.
- Levis B.E.,
- Binkley P.F.,
- Shapiro C.L.
- Gordon L.I.,
- Burke M.A.,
- Singh A.T.,
- et al.
- Khouri M.G.,
- Douglas P.S.,
- Mackey J.R.,
- et al.