Influence of Ile655Val polymorphism on trastuzumab-induced cardiotoxicity in early-stage HER2 positive breast cancer

Trastuzumab has improved the prognosis of HER2 positive breast cancer, but cardiotoxicity remains a concern. We aimed to identify risk factors for trastuzumab-induced cardiotoxicity, with an emphasis on the HER2 Ile655Val single nucleotide polymorphism. This single-center case–control study included 1056 patients with early-stage HER2 positive breast cancer that received adjuvant trastuzumab. Cardiotoxicity was defined as a decline in left ventricular ejection fraction (LVEF) > 15% in patients without previous cardiomyopathy, or > 10% in patients with baseline LVEF of < 50%. Patient characteristics and cardiac parameters were compared in 78 (7.38%) cases and 99 randomly assigned controls, and the polymorphism was genotyped using real-time polymerase chain reaction. Cardiotoxicity was independently associated with advanced age (P = 0.024), lower body mass index (P = 0.023), left breast involvement (P = 0.001), N3 status (P = 0.004), diabetes (P = 0.016), and a family history of coronary artery disease (P = 0.019). Genotype distribution was as follows: A/A (Ile/Ile) was found in 111 (62.7%) patients, A/G (Ile/Val) in 60 (33.9%) patients, and G/G (Val/Val) in 6 (3.4%) patients. The genotype was not associated with cardiotoxicity or the severity of heart failure, reversibility, and recovery time. We found no association between the HER2 Ile655Val polymorphism and trastuzumab-induced cardiotoxicity; therefore, we do not recommend routine cardiotoxicity-risk stratification using this polymorphism.

The promise of personalized medicine has prompted increasing research in chemotherapy pharmacogenomics, and there is evidence of a genetic component to explain the variable susceptibility to chemotherapy-induced cardiotoxicity 9 . Single nucleotide polymorphisms (SNPs) of the HER2/neu oncogene have been explored as potential markers of trastuzumab-induced cardiotoxicity, with the codon 655 [Ile655Val] polymorphism the most extensively examined. This polymorphism is a change from adenine to guanine which causes an amino acid change from isoleucine to valine.
Identifying SNPs associated with trastuzumab cardiotoxicity could help risk-stratify patients and minimize unnecessary avoidance of a proven life-prolonging therapy. Furthermore, identifying high-risk patients would ensure the timely introduction of cardiac medications known to decrease cardiotoxicity, and identify who would benefit from more frequent cardiac monitoring. Despite these potential benefits, studies examining the role of HER2 polymorphisms on trastuzumab cardiotoxicity are conflicted [10][11][12][13] . Therefore, the aim of this study was to identify risk factors for trastuzumab-induced cardiotoxicity in patients with early-stage HER2 positive breast cancer, with an emphasis on the HER2 Ile655Val polymorphism, and describe the characteristics of trastuzumab cardiotoxicity.

Patients and methods
This single-center, case-control study was conducted at a tertiary care hospital in Zagreb, Croatia between January 2007 and December 2016. During this time, 1056 patients with HER2 positive breast cancer were treated with adjuvant trastuzumab. The studied population was predominately Caucasian. Cardiotoxicity was defined as a decline in left ventricular ejection fraction (LVEF) by > 15% in patients without previous cardiomyopathy, or > 10% in patients with a baseline LVEF of < 50%. Using these criteria, 78 (7.38%) cases were identified, and 99 randomly assigned patients without cardiotoxicity were used as controls. Patients with metastases were excluded, as these patients were more likely to receive complex chemotherapy regimens prior to trastuzumab that could increase cardiotoxicity, and indications for trastuzumab discontinuation in this group are less clear.
Trastuzumab was indicated in patients with HER2 positive breast cancer, identified using immunohistochemistry (IHC) or fluorescent in situ hybridization (FISH) in accordance with the European Society for Medical Oncology Guidelines 14 . HER2 status in tumor biopsies was determined using IHC (Hercept test), where positive results were marked as 3 + . When the Hercept test was marked as 2 + , FISH was used to confirm positivity (FISH amplification ratio ≥ 2). Adjuvant trastuzumab was given to all patients with confirmed HER2 positive breast cancer. It was given for one year, with 17 applications, divided in 3-week intervals, and was administered in a short intravenous infusion (163 patients, 92.09%) or subcutaneously (14 patients, 7.91%).
Medical records were reviewed for prior chemotherapy regimens and chest radiation, presence of cardiovascular disease, arterial hypertension, dyslipidemia, and diabetes, and family history of cardiovascular disease.
Prior to enrollment, patients were informed about the nature and aim of the investigation, and written informed consent was obtained. The study was approved by the Clinical Hospital Center "Sestre Milosrdnice" Ethics committee in accordance with the tenets of the Declaration of Helsinki (reference number: 380-59-10106-15-168/185).

Assessment of cardiac function.
A transthoracic echocardiogram (ECHO) was used to assess cardiac function. ECHO reports before and during trastuzumab treatment were retrieved from medical records. In accordance with current guidelines, baseline ECHOs were performed prior to trastuzumab therapy, with repeat ECHOs at 3-month intervals, and additional ECHOs if patients developed signs or symptoms of cardiotoxicity. If trastuzumab was discontinued due to cardiotoxicity, control ECHOs were performed every 3 to 5 weeks to assess reversibility and whether trastuzumab could be reinitiated. Complete reversibility was defined as a return of normal LVEF (LVEF ≥ 50%), and partial reversibility was defined as a 10% increase in LVEF in pateints with LVEF ≤ 50%. Diastolic dysfunction was assessed using pulsed Doppler and the transmitral flow velocity curve. Other methods were not used due to the technical limitations of devices at that time. Patients had a final ECHO after completing trastuzumab therapy.
HER-2/neu genotyping. All patients had blood samples drawn for HER-2/neu genotyping using standard venipuncture. Deoxyribonucleic acid (DNA) was extracted from 6 ml of whole blood and ethylenediaminetetraacetic acid (EDTA) using a commercial isolation kit, the High Pure PCR Template Kit Roche (Roche diagnostics, Mannheim Germany) according to the manufacturer's instructions. The sample of whole blood was incubated with proteinase K and lysis buffer for protein denaturation and cell membrane lysis. The denatured proteins and cellular degradation products were discarded, and the supernatant, which contained nucleic acid, was passed through columns from the test packet. The columns had a semi-permeable membrane, which retained nucleic acid, while the rest of the cellular content was eluted and discarded. The bound nucleic acid was washed twice using a wash buffer. In the final step, DNA bound to the column's membrane was eluted in a sterile test tube using an elution buffer. The isolated DNA was stored at + 4 °C until sample analysis. The concentration of DNA isolated was around 50 ng/µL.
The HER-2/neu genotype was determined using real-time PCR (qPCR) and melting curve analysis. The PCR reaction was performed on a LightCycler 480 analyzer (Roche Diagnostics, Mannheim, Germany) using commercially available reagents, LightCycler DNA Master HybProbe (Roche Diagnostics, Mannheim, Germany), Statistical analysis. Patient characteristics were assessed using descriptive statistics. Continuous variables were compared with one-way analysis of variance (ANOVA) and the Bonferroni method was used for post-hoc analyses. Categorical variables were analyzed using the Chi square test. Dependent continuous variables were analyzed with ANOVA for repeated measurements (RM ANOVA) and this method was used to adjust for confounding factors. Multivariate Cox regression analysis was used to analyze the association between all baseline variables and the development of cardiotoxicity and to analyze variables independently associated with recovery of cardiac function. Power analysis was performed using data obtained from the meta-analysis published by Gómez Peña et al 13 . In this meta-analysis, cardiotoxicity was observed in 14/217 (6.45%) patients with the AA genotype and 29/113 (25.6%) patients with the AG genotype. These rates were used as input parameters for sample size calculation because the distribution of the AA and AG genotype had the allocation ratio 2:1. When considering type 1 error (α) to be 5% and type 2 error (β) to be 10%, the minimum sample size was 150 (75 patients with cardiotoxicity and 75 patients without cardiotoxicity). Two-sided p values < 0.05 were considered significant. The statistical analysis was done using IBM SPSS Version 23.0.
Patients with the G/G genotype had a somewhat greater decrease in LVEF during trastuzumab treatment, but this did not reach statistical significance (P = 0.323). There was no statistically significant difference in the severity of heart failure, reversibility, and recovery time in regard to the HER-2/neu genotype (Table 2). www.nature.com/scientificreports/ A subgroup analysis based on the lowest EF. Patients with cardiotoxicity were divided into two subgroups based on their lowest EF being < 50% or ≥ 50%. Like the primary analysis, there was a significant difference in terms of age, BMI, left breast involvement, N status, diabetes, and family history of coronary artery disease. Patients with lowest EF < 50% were more likely to be symptomatic and have worse NYHA functional status, have lower rates of cardiac recovery, and have longer recovery time. Trastuzumab was reinitiated in fewer patients with lowest EF < 50% and only a minority of patients completed trastuzumab treatment (Table 3). We found no differences in the HER2 polymorphism or chemotherapy protocols between the two subgroups (Table 3).

Discussion
This study examined the characteristics of trastuzumab-induced cardiotoxicity, and the role HER2 Ile655Val polymorphism has on its development. We found no significant association between this polymorphism and cardiotoxicity (P = 0.934). Although there was a greater decline in LVEF in patients with the G/G genotype, it was not statistically significant (P = 0.323). Distribution of the polymorphism was comparable to previously published results in Caucasians 10 . Studies examining the role of this polymorphism on trastuzumab-induced cardiotoxicity  www.nature.com/scientificreports/     www.nature.com/scientificreports/ are conflicted [10][11][12][13][15][16][17] . The first study to examine this polymorphism included 61 patients with metastatic breast cancer, and all five patients with cardiotoxicity had the Ile/Val genotype (P = 0.0058) 10 . Another study published as an abstract included 40 cases and 84 controls and found no difference in genotype distribution (P = 0.91) 15 . Lemieux et al. genotyped 73 patients with non-metastatic breast cancer and found that along with heavy alcohol intake, patients with the A/G genotype (Ile/Val carriers) were more likely to develop cardiotoxicity compared to the A/A genotype (Ile/Ile carriers) (OR = 6.00, P = 0.01) 11 . Similarly, a study of 78 patients also found that compared to the A/A genotype, those with the A/G genotype were more likely to develop cardiotoxicity (P = 0.012), and a meta-analysis combining these data with previous studies confirmed the association 13 . These differences may emerge from small sample sizes (N ranging 61-140 patients) 10,12,13,18 , cardiotoxicity definition, or ethnicity 12 . A recent, large genome-wide association study of cardiotoxicity in the N9831 trial, did not show any association with the Ile655Val polymorphism and decline in LVEF in 800 patients treated with doxorubicin, paclitaxel and trastuzumab (Arms BC) 17 . The authors explained that the lack of replication could be due to cardiotoxicity definition. Whereas a quantative model was used in the N9831 trial, previous studies used a binary definition of cardiotoxicity or no cardiotoxicity, with varying definitions. However, genotypic analyses of the AG + GG genotypes versus the AA genotype on a case control basis, using the definition of cardiotoxicity proposed by Gomez et al. (decrease in LVEF > 10% points to < 50% or any decrease > 15% or any decrease resulting in < 45% or diagnosis of CHF) 13 showed no association (P < 0.05). We agree with the authors that the lack of negative studies may be a result of publication bias towards positive associations and small sample size 17 . Our study had sufficient power and a homogenous population, and we did not find this polymorphism to be a predictor of trastuzumab cardiotoxicity. However, the burden of preexisting cardiovascular comorbidities in our studied population was high, which could explain why this polymorphism did not emerge has a predictive factor. HER2 receptor expression depends on HER2 gene amplification. This is important in tumor tissues marked as 2+ after IHC analysis of HER2 receptor expression. This is why it is mandatory to use FISH to confirm HER2 receptor positivity (FISH amplification ratio ≥ 2). Tumors labeled as HER2 2+ after immunostaining may be recategorized into HER 3+ if FISH shows an appropriate amplification ratio. Our study is the first to examine HER2 receptor intensity and HER2 genotype. We analyzed HER2 genotype separately in patients with HER 3+ after ICH analysis and those that were initially labelled as HER 2+ but were recategorized as HER 3+ after FISH. We did not find a significant association between HER2 intensity (HER2 3+ vs HER2 2+) and genotype, which suggests that HER2 intensity cannot predict genotype. This was also the first study to compare HER2 genotype and tumor grade, size, lymph node involvement, and hormone status, and we did not find any associations. Furthermore, distribution of the codon 655 polymorphism was not associated with patient age, cardiovascular risk factors (arterial hypertension, diabetes, and dyslipidemia), degree of left ventricular diastolic dysfunction, NYHA class, time to develop cardiotoxicity, or reversibility of cardiac dysfunction. A decrease in left ventricular diastolic function was observed during trastuzumab treatment (predominantly type I and II dysfunction), which has not been previously reported.
In our study, cardiotoxicity was independently associated with advanced age, lower BMI, left-sided breast involvement, N3 status, history of diabetes, and family history of coronary artery disease. Previous studies examining risk factors for trastuzumab-induced cardiotoxicity have reported conflicting results 8 , which may be due to an overall low cardiotoxicity rate, limiting sample size 8 , and different inclusion/exclusion criteria. A recent metanalysis of 17 studies showed that hypertension, diabetes, previous anthracycline use, and advanced age increased the risk of cardiotoxicity. However, because advanced age is usually accompanied by other cardiovascular comorbidities and an unfavorable lifestyle, it is unclear whether age is an independent risk factor 8 . The metanalysis did not show any significant association between trastuzumab toxicity and race, BMI, dyslipidemia, or history of coronary artery disease, and too few articles described cumulative anthracycline dosage, or left/right breast involvement to allow for statistical analysis 8 . Advanced age, diabetes, and family history of coronary artery disease are known risk factors for adverse cardiac events 19 , so it is not surprising that this was associated with cardiotoxicity in our study. We observed higher rates of left-sided breast involvement, which might be attributed to prior left chest wall radiotherapy, although studies have reported conflicting results 20,21 . Whether the association between N3 status and cardiotoxicity is due to larger radiation doses and subsequent cardiac damage is unclear.
Trastuzumab cardiotoxicity commonly manifests as an asymptomatic fall in LVEF, and in contrast to anthracyclines, does not depend on cumulative dose, is frequently reversible, and is generally well tolerated when Table 3. Subgroup analysis of patients with cardiotoxicity based on lowest ejection fraction. BMI Body Mass Index, HER2 Human epidermal growth factor receptor 2, A Adenin, G Guanine, T Tumor, N Node, ER Estrogen Receptors, PR Progesterone Receptors, ISH In situ Hybridization, IHC Immunohistochemistry, EF Ejection Fraction, NYHA New York Heart Association, P value was calculated with one-way analysis of variance (ANOVA). www.nature.com/scientificreports/ reinitiated following cardiac recovery 22 . Previous studies have reported overt heart failure in only 1% to 4% of patients treated with trastuzumab 23-25 , with around 10% experiencing an asymptomatic fall in LVEF 24,26,27 . We observed higher rates of symptomatic heart failure (61.5%) and severe heart failure (15.4%) compared to other studies 28 . The majority of our patients received anthracycline-based protocols, which are known to increase cardiotoxicity 29 and could explain these differences. Cases and controls had similar baseline LVEF, and LVEF dynamics during trastuzumab treatment and were comparable to previous reports 30,31 . Similarly to previous reports 29 , cardiac recovery was observed in 83.3% of patients, with a median time to recovery of 30 days. There was a strong correlation between the lowest LVEF and time to recovery, so patients with EF below < 50% can expect longer recovery times before trastuzumab continuation. In multivariate analysis, recovery of cardiomyopathy was independently associated with NYHA status, grade II diastolic dysfunction, and the presence of arterial hypertension.
The major limitation of the study is the small sample size. Our study included a relatively homogenous Caucasian population and genetic homogeneity is needed for this type of study. Although this a strength of our study, our results may be limited to Caucasians because SNP frequencies vary by ethnicity 12 .
In conclusion, we did not find that the Ile655Val HER2 polymorphism was associated with trastuzumabinduced cardiotoxicity. Therefore, a personalized approach using this polymorphism will unlikely help riskstratify patients. For now, the decision to use anthracycline-based regimens in patients with HER2 positive breast cancer should be carefully weighed, especially in older patients with cardiovascular risk factors, knowing that the added risk of adding such regimens to trastuzumab candidates may rob them of a proven life-prolonging therapy. Basic research is needed to better understand the mechanism of trastuzumab toxicity, which could help identify specific biomarkers of trastuzumab toxicity.

Data availability
The majority of data used and analyzed in the current study is included in the manuscript. The remainder is available from the corresponding author upon request.