Pharmacogenomic predictors of anthracycline-induced cardiotoxicity in breast cancer patients: a systematic review and meta-analysis

Abstract

Background

Anthracyclines significantly improve overall survival and play a vital role in the treatment of breast cancer. However, they are associated with cardiac dysfunction, which is often irreversible. Although anthracycline-induced cardiotoxicity (AIC) is dose-dependent, the difference in susceptibility patterns suggests the role of pharmacogenomics. Several studies explored the role of genetic variants in AIC. Integrating pharmacogenomic testing with routine anthracycline surveillance will help to predict the individuals who are at risk of developing AIC. Therefore, this current systematic review aims to evaluate and synthesize the evidence on the pharmacogenomics association of cardiotoxicity in individuals receiving anthracyclines for breast cancer treatment.

Methods

PubMed, Embase, and Scopus databases are systematically searched for the literature. After the initial search, 842 records have been identified. Following screening, 18 studies investigating genetic associations with AIC in breast cancer patients were found to be eligible for inclusion in the study. The quality of studies is assessed with the Q-Genie tool. The data is extracted and summarized with odds ratios and corresponding confidence intervals were reported.

Results

A total of 18 candidate gene association studies involving 4,703 breast cancer patients were included in the qualitative synthesis. Out of 57 genetic variants reported, 18 genetic variants (31.5%) are associated with an increased risk, while 3 genetic variants (5.3%) have demonstrated a risk-reducing tendency.

Discussion

Characterizing genetic variants in biological pathways of anthracyclines could inform precision therapy and the development of targeted interventions. The limitations of the synthesized evidence include the inadequate sample size, methodological bias within the included studies, inconsistent findings from different studies, and imprecise effect estimates.

Conclusion

The genetic variants influence susceptibility to AIC in breast cancer patients. Further large-scale studies with longer follow-up are warranted to validate these associations and facilitate their translation into clinical practice.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40959-026-00449-3.

Introduction

Breast cancer remains the second most diagnosed cancer across the world, with roughly 2.3 million new cases every year. In India, it affects approximately 0.2 million individuals annually, constituting 26.6% of cancers among females and 13.6% across both sexes [1]. Anthracyclines were introduced into clinical practice for breast cancer treatment following the NSABP B-11 trial [2]. Since then, anthracycline-based regimens have improved overall survival (OS) in breast cancer patients. A pooled analysis demonstrated that the anthracycline reduces the recurrence risk of breast cancer by one-third in 10 years and mortality by 20–25% [3].

However, they are among the most frequently used cytotoxic drugs that are responsible for anthracycline-induced cardiotoxicity (AIC) [4]. Long-term follow-up data of breast cancer patients revealed that cardiovascular complications contribute to about 15.9% of deaths in breast cancer survivors [5]. A pooled analysis by Smith et al. reported that anthracycline-based regimens increase the risk of clinical cardiotoxicity by about five times and subclinical cardiotoxicity by about six times compared to a non-anthracycline regimen [6]. The cumulative administered dose of anthracyclines is a key determinant of type 1 cardiotoxicity [7]. For instance, about 28% of survivors who received a combined dose of anthracyclines of more than 300 mg/m² developed signs of AIC, compared to about 7% of survivors who received a total dose of less than 250 mg/m² [8–10]. In addition to cumulative dose, clinical risk factors such as hypertension, diabetes mellitus, and obesity increase the susceptibility to AIC [11].

In the clinical settings, it is also observed that some patients experience AIC despite receiving similar anthracycline doses and having comparable baseline clinical factors, indicating possible genetic predisposition [12]. Several candidate genes have been identified with genetic variants involved in anthracycline transport, metabolism, oxidative stress, DNA repair, cardiac signalling, and autophagy-related pathways, which are associated with AIC in breast cancer patients. The variants in genes such as ABC, SLC, CBR, UGT, GST, NCF4, NOS3, NADPH, RAC, ETFB, TRPC6, HFE, TP53, HLA, RAAS, and ATG could be the cause of variations in patients’ cardiac tolerance to anthracycline chemotherapy [13–21]. Several predictive tools are also being developed for the early detection of cardiotoxicity due to interindividual variability [22, 23]. Current methods for assessing cardiotoxicity primarily depend on functional parameters, which may not identify early myocardial injury [24]. Combining genetic predisposition with cardiac monitoring may help identify at-risk patients earlier, enabling timely intervention. Therefore, we aim to systematically evaluate and synthesize the evidence for genetic association and AIC in the breast cancer cohort.

Methods

Search strategy

The present systematic review was carried out following the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) [25]. The protocol has been registered in PROSPERO (registration number: CRD420251040386). A systematic search of PubMed, Embase, and Scopus for articles published in the English language from inception till June 2025 was performed to identify relevant articles assessing the pharmacogenomics association of AIC in breast cancer cohorts. Only peer-reviewed full-text publications were included and grey literature and conference abstracts were excluded. The detailed search strategy is presented in in Supplementary Table 1 (S1) and Supplementary Table 2 (S2).

Study selection criteria

Two independent reviewers (VMB) and (AC) performed a literature search, title and abstract screening, identification, and selection based on the inclusion and exclusion criteria. Any disagreements were resolved by an independent reviewer (MM). Inclusion criteria were prospective or retrospective human studies in which breast cancer patients received anthracyclines. The studies that have evaluated the effect of genetic polymorphisms and AIC reported clinical or subclinical cardiotoxicity outcomes. The excluded studies include case series, case reports, letters, editorials, commentaries, animal studies, non-English literature studies, review articles, cell line studies, abstracts, summaries, conference reports, studies conducted in non-breast cancer patients, studies in which anthracyclines were not the primary chemotherapeutic drugs, and studies that did not report pharmacogenomic associations with anthracycline-induced cardiotoxicity.

A total of 46, 149, and 647 articles were identified from PubMed, Embase, and Scopus, respectively. Following deduplication, 137 articles were excluded. A total of 438 full-text articles were available after screening titles and abstracts. Out of which, 420 articles were excluded for the following reasons: Case reports, Case series, Book chapters, Editorials, Conference reports/Posters = 84, Review articles = 175, Cell Line studies = 27, Animal Studies = 15, Irrelevant to PICO = 111, Full text not available = 3, GWAS Studies = 5 Finally,18 studies are included for qualitative synthesis. The detailed selection criteria of studies are represented in Fig. 1.

Fig. 1.

PRISMA Flow chart (Created with RevMan Version 5.4)

Data extraction

Two reviewers (VMB) and (AC) extracted the data independently from the 18 included studies. Data discrepancies were solved by MM. For all selected articles, the first author, publication year, total sample size, study site, study design, participant characteristics, cumulative anthracycline dose, method of genotyping, cardiotoxicity endpoints, and pharmacogenomics assessed in the study were extracted. We reported the definitions of cardiotoxicity as provided by the authors, which were based on the standards applied in each study. We extracted odds ratios when reported, and we prioritised adjusted odds ratios over unadjusted estimates when they were available. An illustration of the PRISMA flow chart was created using RevMan [26].

Quality assessment

The selected studies are evaluated for methodological quality by two independent reviewers (VMB), (MM) using the quality of genetic association studies (Q-Genie) tool [27]. Overall, 11 studies were rated as good quality [13–19, 24, 28–30], and 8 were rated as moderate quality [20, 31–36], with no studies categorized as poor. Common strengths of the studies include a clear biological rationale, mentioning the plausible mechanisms for AIC. Technical classification of exposure and outcome measures has been well reported across the studies. Some studies are limited in the categories of sample size [33, 36] statistical analyses [16, 36] and indirectness in the measurement of outcome [35] The individual risk of bias assessment for the included studies is summarized in Table 1.

Table 1.

Risk of bias assessment

S.no	Study author, Year	Q1	Q2	Q3	Q4	Q5	Q6	Q7	Q8	Q9	Q10	Q11	Overall Quality
1	Advani et al. (2023) [17]	6	5	-	7	4	5	5	6	4	3	6	51 (Good)
2	Domas Vaitiekus et al. (2025) [14]	5	6	5	6	4	4	3	4	5	3	3	48 (Good)
3	Ebaid N et al. (2024) [32]	5	5	4	4	3	3	4	3	4	4	3	42 (Moderate)
4	Gintare Muckiene et al. (2023) [13]	5	6	5	6	5	4	3	4	5	4	3	50 (Good)
5	Grakova et al. (2021) [31]	5	5	4	5	3	3	4	3	4	3	2	41 (Moderate)
6	Hertz et al. (2016) [24]	6	5	5	6	4	5	4	5	5	4	2	51 (Good)
7	Kopeva et al. (2022) [20]	5	5	5	5	3	3	5	3	4	4	2	44 (Moderate)
8	Lang et al. (2021) [15]	6	6	-	6	5	4	4	5	6	4	3	49 (Good)
9	Li et al. (2019) [30]	5	5	4	6	5	4	4	5	6	4	3	51 (Good)
10	Liu et al. (2018) [16]	5	5	4	5	4	5	4	3	4	3	4	46 (Good)
11	Ruiz-Pinto et al. (2018) [29]	5	5	-	5	5	4	4	5	6	3	3	45 (Good)
12	Nyangwara V et al. (2022) [33]	5	5	5	3	4	4	2	4	3	3	3	41 (Moderate)
13	Norton et al. (2020) [19]	6	5	5	7	4	5	5	6	4	3	6	56 (Good)
14	Todorova et al. (2017) [36]	5	5	4	5	4	3	2	3	4	3	4	42 (Moderate)
15	Vaitiekus et al. (2021) [18]	5	6	5	6	4	4	3	4	5	3	3	48 (Good)
16	Vivenza et al. (2013) [35]	4	5	4	5	3	4	4	3	4	3	2	41 (Moderate)
17	Volkan Salanci et al. [34]	5	5	4	5	3	3	4	3	4	3	2	41 (Moderate)
18	Vulsteke et al. (2015) [28]	6	6	5	6	5	4	4	5	6	4	4	55 (Good)

Domain legend

Question 1: Adequacy of the presented hypothesis and rationale

Question 2: Classification of the outcome (e.g. disease status or quantitative trait)

Question 3: Description of comparison groups (e.g. cases and controls)

Question 4: Technical classification of the exposure (i.e., the genetic variant)

Question 5: Non-technical classification of the exposure (i.e. the genetic variant)

Question 6: Disclosure and discussion of sources of bias

Question 7: Adequate power of the study

Question 8: Description of planned analyses

Question 9: Statistical methods

Question 10: Description and test of all assumptions and inferences

Question 11: Conclusions drawn by the authors were supported by the results and appropriate methods

The following cut-points to designate low, moderate, and high quality studies : For studies with case/control status as the outcome of interest: scores ≤ 35 on the Q-Genie tool indicate poor quality studies, > 35 and ≤ 45 indicate studies of moderate quality, and > 45 indicate good quality studies similarly, cut-points for studies without control groups were created by excluding question 3 from the calculation of the total score on Q-Genie: scores ≤ 32 on the Q-Genie tool indicate poor quality studies, > 32 and ≤ 40 indicate studies of moderate quality, and > 40 indicate good quality studies

Results

A total of 4,703 breast cancer patients were involved in 18 candidate gene studies reporting the association of 57 genetic variants with AIC. The qualitative characteristics of the included studies are presented in Table 2. Out of 57 genetic variants, the odds ratios and corresponding confidence intervals as effect estimates have been reported for 31 genes. Out of which, 18 genetic variants (31.5%) are associated with increased risk, while 3 genetic variants (10%) are associated with a protective effect. The associations between genetic variants and the risk of anthracycline-induced cardiotoxicity, expressed as odds ratios (ORs), are summarized in Table 3. Protective variants are presented in Table 4.

Table 2.

Qualitative study characteristics of the included studies

S.no	Study: author (year)	Study site	Study design, number of participants	Age (in yrs)	Anthracycline dose used/cumulative dose	Method used for genotyping	Definition of cardiotoxicity	Genes studied
1.	Advani et al. (2023). [17]	Multicentre clinical trial, USA.	N = 993, Retrospective replication study within NSABP B-31 RCT cohort	49.6 ± 9.9 years	Doxorubicin 60 mg/m² for 4 cycles cumulative dose:240 mg/m²	MALDI-TOF mass spectrometry	Symptomatic congestive heart failure confirmed by MUGA or echocardiogram, or Probable/definite cardiac death (CD).	TRPC6 (rs77679196), BRINP1 (rs62568637), LDB2 (rs55756123), RAB22A (rs707557), LINC01060 (rs7698718), CBR3 (rs1056892).
2.	Domas Vaitiekus et al. (2025 [14]	Lithuanian University of Health Sciences (LUHSH) Kaunas Clinics, Europe.	N = 81, Prospective study	54.11 ± 9.4 Cases: 54.8 ± 8.9 Controls: 52.9 ± 10.3	Doxorubicin median dose 129.000–303.200 mg/m²	TaqMan genotyping	LVEF Decline of ≥ 10% from baseline to a value ≤ 55% in the absence of clinical symptoms or signs	SULT2B1 (rs1042637), UGT1A6 (rs1786378), CBR1 (rs9024), CBR3 (rs1056892), NCF4 (rs1883112), CYBA (rs1049255).
3.	Ebaid et al. (2024) [32]	Menoufia University, Egypt.	N = 100, Prospective study	45.8 (9.50)	Doxorubicin 60mg/m2 for 4 cycles cumulative dose:240mg/m2	TaqMan genotyping	Common Terminology Criteria for Adverse Events (CTCAE), ejection fraction was assessed before treat- ment initiation and after 6 months of treatment.	CBR1 (rs20572), SLC22A16 (rs714368).
4.	Gintare Muckiene et al. (2023) [13]	Lithuanian University of Health Sciences Kauno Klinikos, Europe.	N = 71, Prospective study	53.76 ± 9.23 cases: 53.30 ± 11.25 Controls: 53.94 ± 8.42	Doxorubicin cumulative dose: 236.70 mg/m2	TaqMan genotyping	LVEF Decline to a value below the lower limit of normal LVEF < 53%	ABCB1 (rs1045642), ABCC1 (rs4148350), ABCC1 (rs3743527).
5.	Grakova et al. (2021) [31]	Russia.	N 176, Prospective study	45 (42; 47)	Doxorubicin cumulative dose:300–360 mg/m2	Polymerase Chain Reaction	LVEF Decline by > 10% at 12 months after chemotherapy and the development of HF with symptoms and clinical signs.	NADPH oxidase (rs4673), NOS3 (rs1799983), EDNRA (rs5335).
6.	Hertz et al. (2016). [24]	University of Michigan Comprehensive Cancer Center, USA.	N = 166, Cross-sectional study	50 (24–80)	Doxorubicin cumulative dose:240 mg/m2 (120–366)	MALDI-TOF mass spectrometry.	LVEF < 55%.	ABCB1 (rs1045642), CBR3 (rs1056892), RAC2 (rs13058338), NCF4 (rs1883112), SLC28A3 (rs7853758), TOP2B (rs10865801), UGT1A6 (rs17863783), CYBA (rs4673).
7.	Kopeva et al. (2022)[20]	Russia.	N = 176, Prospective study	Cases :45 (42; 47), Controls:45 (42; 50)	Doxorubicin cumulative dose: 360 mg/m2 (300–360)	Polymerase Chain Reaction	LVEF Decline by > 10% at 12 months after chemotherapy and the development of HF with symptoms and clinical signs	NOS3 (rs1799983), EDNRA (rs5335), NADPH oxidase (rs4673), p53 (rs1042522), NOS3 (rs1799983), Caspase 8 (rs3834129, rs1045485), Interleukin-1b gene (rs1143634), TNF-α (rs1800629), SOD2 (rs4880), GPX1 (rs1050450)
8.	Lang et al. (2021). [15]	Roswell Park Comprehensive Care Center, Buffalo, New York, USA.	N 155, Prospective study	52 ± 11	Doxorubicin cumulative dose: 240mg/m2	TaqMan genotyping	Reduction of LVEF ≥ 5 to < 55% with symptoms of heart failure or Asymptomatic reduction of LVEF ≥ 10 to < 55% defined by CREC.	CBR3 (rs1056892)
9.	Li et al. (2019). [30]	The Central Hospital of Wuhan, China	N = 427, Prospective cohort study.	45.3 ± 6.0	Epirubicin cumulative dose 302.0 mg/m2 (281.0–321.0)	Pyrosequencing	LVEF absolute decline of at least 10% from baseline to a value less than 53% on an echocardiogram, Heart failure, acute coronary artery syndrome, or fatal arrhythmia	UGT2B7 (rs7668258)
10.	Liu et al. (2018) [16]	Cancer Hospital, Chinese Academy of Medical Sciences, Beijing, China.	N = 147 Retrospective study	Cases − 50 (32–72), Controls − 51 (30–75)	Epirubicin 90 mg/m2 for 4 cycles, cumulative dose: 360mg/m2 Epirubicin 75 mg/m2 for 6 cycles, cumulative dose: 450 mg/m2	MALDI-TOF mass spectrometry	ST-T segment abnormalities, elevated myocardial enzymes, arrhythmia, and QRS pat tern or duration abnormalities	ATM (rs1003623, rs227060, rs228589, rs664143, rs664677), ATG 5 (rs473543 and rs3761796), ATG 7 (rs2594971, rs111595248 and rs4684789), ATG12 (rs1058600 and rs5870670), ATG13 (rs13448, rs10838611), MAP1LC3A (rs4911429 and rs6088521), MAP1LC-3B (rs9903, rs35227715, rs7865, and rs16944733), CASP3 (rs1049216, rs12108497, rs2720376), CRYAB (rs14133), and STMN1 (rs182455).
11.	Ruiz-Pinto et al. (2018). [29]	La Paz University Hospital, Madrid, Spain	N = 61 (cases = 18, Controls = 43)	Cases: median 59.5 (36–72), Controls: median 49 (27–73)	Doxorubicin cumulative dose in cases: 298.4 (200–588) mg/m2 Doxorubicin cumulative dose in controls: 298.6 (150–375) mg/m2	Illumina array	Cardiac failure grade 3–5, Asymptomatic decrease of left ventricular ejection fraction (LVEF) ≥10%.	ETFB rs79338777
12.	Nyangwara V et al. (2022) [33]	University of Zimbabwe, Zimbabwe.	N = 50, Prospective study	48.0 (44.5–59.0)	Doxorubicin cumulative dose 238.89 mg/m2	TaqMan genotyping,	LVEF threshold of greater than 10% reduction from the normal echocardiograms (normal ≥ 60%)	SLC28A3 (rs7853758), UGT1A6 (rs17863783) RARG (rs2229774)
13.	Norton et al. (2020). [19]	N9831 trial- Multicenter study (United States): Mayo Clinic Biobank sites in Jacksonville, Florida, and Rochester, Minnesota.	N = 1010, Retrospective nested case–control study	76.0 (61–81)	Doxorubicin 60 mg/m² every 3 weeks for 4 cycles cumulative dose:240 mg/m²	Sanger sequencing	Symptomatic CHF, definite cardiac death because of myocardial infarction, CHF or arrhythmia, or probable cardiac death without documented etiology	TRPC6 (rs77679196),
14.	Todorova et al. (2017). [36]	University of Arkansas for Medical Sciences, USA.	N = 30, case–control genetic study.	53.1 (35–76)	Doxorubicin cumulative dose: 240 mg/m2	Illumina array	A decline of LVEF by > 10% or below 55% was considered abnormal.	NFKBIL1, TNF-α, ATP6V1G2, MSH5, MICA, LTA, BAT1, NOTCH4,HLA
15.	Vaitiekus et al. (2021) [18]	Lithuanian University of Health Sciences (LUHSH), Kaunas, Europe	N = 81, Prospective study	Case: 52.9 ± 10.29, Controls: 54.8 ± 9.02	Doxorubicin cumulative dose in Cases : 241 (146–248) mg/m2 Doxorubicin cumulative dose in controls :239(150–300) mg/m2.	TaqMan genotyping	Reduction in LVEF is ≥ 10% from baseline to ≤ 55%, without accompanying signs or symptoms defined as Subclinical Cardiotoxicity.	HFE (rs1799945), HFE (rs1800562)
16.	Vivenza et al. (2013). [35]	Not Reported	N = 48, Prospective study	57.5 (28–73)	Epirubicin cumulative dose: 3.60 mg/m2	TaqMan genotyping	Development of overt CHF (grade III) or a decline of LVEF below 50% (grade II) at any time point during the 3-year follow-up	AGT (rs4762), AGT (rs699), ACE (rs4340), CYP11B2 (rs1799998), AGTR1(rs5186), GSTM1, GSTT1, GSTP1 (rs1695), TP53 (rs1042522)
17.	Volkan-Salanci et al. (2012). [34]	Not Reported	N = 70 (Breast cancer = 54, Lymphoma = 10), Prospective Study	49.1 ± 13.6	Doxorubicin dose cumulative dose: 317.1 ± 94.9 mg/m2	TaqMan genotyping	LVEF decrease > 10%; LVEF ≤ 50%	CBR3 (rs1056892), GSTP1
18.	Vulsteke et al. (2015). [28]	Leuven Multidisciplinary Breast Cancer Center, University Hospitals Leuven, Belgium.	N = 877, Retrospective cohort study	50.3 ± 9.5	Epirubicin 100 mg/m2 3–6 Cycles cumulative dose: 300–600 mg/m2	MALDI-TOF mass spectrometry	Asymptomatic decrease of Left Ventricular ejection fraction (LVEF) > 10% and cardiac failure grade 3–5 (CTCAE 4.0)	ABCC1 (rs3743527, rs4148350, rs4551140, rs246221), ABCC2 (rs280440, rs8187710), CYBA (rs4673), NCF4 (rs4673), RAC2 (rs13058338), SLC28A3 (rs7853758)

Table 3.

Summary of the significant genetic variant associations and AIC risk with Odds Ratios (OR) and 95% Confidence Intervals (CI) reported

S.no	Gene	SNP (rsID)	Genetic Model	Genotype Comparison	OR (95% CI)	p-value	Reference allele	References
1.	ABCC1	rs4148350	Dominant model	TG vs. GG	9.661 (1.418–65.82)	0.0021	G	Gintare Muckiene et al. (2023) [13].
		rs246221	Heterozygous comparison	TC vs. TT	1.59 (1.07–2.35)	0.021	T	Vulsteke et al. (2015) [28].
2.	ATP6V1G	rs2071594	Allelic model	C/G	6.83	0.02	G	Todorova et al. (2017) [36].
		rs3130059	Allelic model	G/C	6.83	0.02	C
		rs11796	Allelic model	T/A	4.12	0.03	A
3.	BAT1	rs2239527	Allelic model	G/C	4.13	0.05	C	Todorova et al. (2017) [36].
4.	C6orf10	rs2050190	Allelic model	G/A	3.89	0.02	A	Todorova et al. (2017) [36].
5.	CBR3	rs1056892	Additive model	GG = 0, GA = 1, AA = 2	2.50 (1.22–5.11)	0.012	G	Hertz et al. 2016 [24]
6.	ETFB	rs79338777	-	-	14.1 (1.60–124)	0.0079	C	Ruiz-Pinto et al. (2018) [29].
7.	GPX1	rs1050450	Recessive model	CC vs. (CT + TT)	2.345	0.007	C	Kopeva et al. (2022) [20].
8.	HLA-C	rs9264942	Allelic model	G/A	8.61	0.01	A	Todorova et al. (2017) [36].
		rs2523619		G/A	6.56	0.01	A
		rs10484554		A/G	5.41	0.04	G
9.	HFE (H63D)	rs1799945	Dominant model	GG + GC vs. CC	3.57 (1.06–8.59)	0.014	C	Vaitiekus et al. (2021) [18]
10.	LTA	rs909253	Allelic model	G/A	6.83	0.02	A	Todorova et al. (2017) [36].
		rs1041981		A/C	6.83	0.02	C
11.	MICA	rs2523451	Allelic model	A/G	4.50	0.04	G	Todorova et al. (2017) [36].
12.	MSH5	rs3131379	Allelic model	A/G	11.58	0.04	G A	Todorova et al. (2017) [36].
		rs3131378		G/A
13.	NADPH oxidase	rs4673	Homozygote comparison	TT	2.7529 (1.30–5.80)	0.0077	C	Grakova et al. (2021) [31].
		rs4673	Homozygote comparison	TT	2.753	0.008	C	Kopeva et al. (2022) [20].
14.	NFKBIL1	rs2071591	Allelic model	A/G	6.83	0.02	G	Todorova et al. (2017) [36].
		rs3093949		A/G	8.87	0.01	G
		rs2071592		A/T	7.99	0.01	T
15.	NOTCH4	rs3134931	Allelic model	G/A	0.22	0.05	A	Todorova et al. (2017) [36].
16.	NOS3	rs1799983	Homozygote comparison	TT	3.0585 (1.20–7.73)	0.0182	G	Grakova et al. (2021) [31].
17.	TNF- α	rs1800629	Allelic model	A/G	5.67	0.03	G	Todorova et al. (2017) [36].
18.	TRPC6	rs77679196	Allelic Model	Frequency of “A” allele	12.84	0.032	A	Advani et al. (2023) [17].
					2.01	0.002	A	Norton et al. (2020) [19].

Table 4.

Protective genetic variants associated with reduced risk of Anthracycline-Induced Cardiotoxicity (AIC)

S.no	Gene	SNP (rsID)	Genetic Model	Genotype Comparison	OR (95% CI)	p-value	Reference allele	References
1.	ABCB1	rs1045642	Additive model	CC = 0, CT = 1, TT = 2	0.48 (0.23–1.00)	0.049	C	Hertz et al. (2016) [24].
2.	NCF4	rs1883112	Dominant model	AA + AG vs. GG	0.486 (0.272–0.867)	0.015	A	Domas Vaitiekus et al. (2025) [14].
3.	UGT2B7	rs7668258	Additive model	CC-0, CT-1, TT-2	0.259 (0.103–0.651)	0.0004	C	Li et al. (2019) [30]

Study characteristics

A total of 18 candidate gene association studies were found to be eligible for qualitative synthesis, involving 4703 breast cancer patients (Fig. 1). Race/Ethnicity of the study participants has been reported in four studies [15, 17, 24, 36]. Of the identified 18 studies, two studies were conducted in the Asian continent, both of which were from China [16, 30]. Five studies were conducted in North America, of which four were conducted in the United States [17–19, 24] and one study was conducted in New York [15] Five studies were conducted in Europe, of which three were from Lithuania [13, 14, 18] ,one from Belgium [28], and one from Spain [29]. Two studies were conducted in the African region, one in Zimbabwe [33] and one in Egypt [32]. Two studies were conducted in Russia [20, 31]. While the other two studies did not report the location of the study [34, 35].

Anthracycline dose

Doxorubicin (n = 14) and Epirubicin (n = 4) are the common anthracyclines reported in the included studies. The cumulative doxorubicin doses range from approximately 120 mg/m² to 366 mg/m². Median values for cases were reported between 239.5 mg/m² and 238 mg/m², while median values for controls ranged from 236 mg/m² to 239 mg/m². Several studies reported a fixed regimen of 240 mg/m² (60 mg/m² every 3 weeks for 4 cycles) [15, 17, 19, 24, 36], whereas others reported higher cumulative exposures, with mean values up to 317 ± 94.9 mg/m² [34], and a median of 360 mg/m² [31]. The reported cumulative Epirubicin dose ranges from 300 to 600 mg/m² for 3–6 cycles, with a median of 302.0 mg/m² (IQR: 281.0–321.0) across the regimens.

Outcome measures

Across the 18 included studies, cardiotoxicity has been defined with both objective measures and subjective clinical outcomes. Objective criteria most commonly involved echocardiography or MUGA-based left ventricular ejection fraction (LVEF) decline, with a threshold such as an absolute decline of ≥ 10% [14, 18, 20, 28–31, 33, 34, 36] from baseline to values below 55% [14, 15, 24, 36] or ≥ 5% with concurrent symptoms or ≥ 10% without symptoms as per Cardiac Review and Evaluation Committee (CREC) criteria [15]. The lower cut-offs of LVEF vary across studies, like the decline to < 53% [13, 30] or ≤ 50% [34]. Additional objective markers included ST–T sgment changes, QRS abnormalities, arrhythmia on ECG, and elevated myocardial enzymes [16].

Cardiotoxicity was defined based on subjective outcomes in some studies, such as symptomatic congestive heart failure (CHF), acute coronary syndrome, fatal arrhythmia, and probable or definite cardiac death [17, 19, 30] Several studies distinguished subclinical cardiotoxicity from overt events [35], while others incorporated grading criteria from the National Cancer Institute Common Toxicity Criteria (NCI CTC)(version 2.0) [35], Common Terminology Criteria for Adverse Events (CTCAE) [28, 29, 32] American Society of Echocardiographers Guidelines [24].

Genotyping technique

Various genotyping techniques have been reported across the studies, such as the TaqMan genotyping assay (n = 8) [13–15, 18, 32–35], MALDI-TOF mass spectrometry (n = 4) [16, 17, 24, 28], pyrosequencing (n = 1) [30], Sanger sequencing (n = 1) [19], Polymerase Chain reaction (n = 2) [20, 31] and Illumina genotyping arrays (n = 1) [29, 36].

Results of the individual studies

The genetic variations involved in various biological pathways influence AIC in various ways. The detailed mechanism is illustrated in Fig. 2. The findings from individual studies were synthesized and grouped based on the mechanism.

Fig. 2.

Candidate gene determinants involved in biological pathways of AIC. A. CBR3, GSTP1, GSTM1, and GSTT1 modulate the enzymatic reduction and detoxification of anthracyclines. B. Genes such as ABCB1, ABCC1, SLC28A3, and SLC22A16 regulate the intracellular accumulation of anthracyclines. Variations in these genes can alter drug efflux and uptake. C. Genes such as NOS, NADPH RAC, NCF4, ROS regulators, and ETFB are involved in the generation and detoxification of reactive oxygen species. D. HFE gene variants influence iron homeostasis, promoting iron-mediated free radical generation. E. Genes such as TOP2B and p53 mediate anthracycline-induced DNA strand breaks and apoptotic signalling, contributing tocardiomyocyte death. F. RARG, TRPC6, EDNRA, NFAT, and PKS regulate hypertrophic signalling that remodels cardiac tissue following injury. G. Variants in HLA, AT1R, ANG II, CYP11B2, and AGT rs699 modulate inflammatory and RAAS, influencing immune-mediated injury and cardiac remodelling 

Drug transporters

ATP-binding cassette transporters

Muckienė et al. reported that the ABCC1 rs4148350 variant leads to AIC in breast cancer patients, with TG carriers showing a higher risk compared to GG carriers (OR 9.661, 95% CI 1.418–65.824, p = 0.0021) [13]. In contrast, Vulsteke et al. did not observe a statistically significant association for this variant [28]. However, in the same study, it was found that the heterozygous T-allele carriers of the ABCC1 rs246221 were prone to be at a higher risk of LVEF decline by about 10% compared to homozygous TT carriers (OR 1.59, 95% CI 1.1–2.3, p = 0.02) [28].

Hertz et al. found that the ABCB1 rs1045642 protects against cardiac injury (OR 0.48, 95% CI 0.23–1.00, p = 0.049) [24]. However, this variant did not reach statistical significance in the study by Muckiene et al. [13]. Other ABC transporters, such as ABCC1 rs4551101 [28], ABCC1 rs3743527 [24] did not reach statistical significance.

Solute carriers

SLC28A3 rs7853758 did not demonstrate a statistically significant correlation with AIC in three studies [24, 28, 33]. An odds ratio with a broad confidence interval but a consistent trend (additive model: OR 0.55, 95% CI 0.16–1.9, p = 0.43) was reported by Hertz et al. [24]. Similarly, SLC22A16 rs714368, rs6907567, rs723685, rs12210538 variants are not associated with AIC [24].

Drug metabolizing enzymes

Carbonyl Reductases (CBR)

Hertz et al. reported that the CBR3 rs1056892 variant increases the risk of decline in LVEF to less than 55% (additive model: OR 2.50, 95% CI 1.22–5.11, p = 0.012), with greater risk after considering covariates (p = 0.008–0.048) [24]. Lang et al. found that doxorubicin treatment disrupts cardiac systolic function across all CBR3 genotype groups (F [1,89] = 50.33, p < 0.001). with greater LVEF decline in GG carriers, moderate LVEF decline in AG carriers, and a non-significant LVEF decline in AA carriers (p = 0.072) [15]. A similar trend of genotype-dependent decline in LVEF has been reported by Volkan-Salanci et al. (GG vs. AA: p = 0.039) [34]. However, the association did not reach statistical significance in the study conducted by Domas Vaitiekus et al. (AA vs. GG, OR 0.597, 95% CI 0.068–5.212, p = 0.640) [14]. Similarly, in NSABP B-31, the CBR3 rs1056892 variant was more frequently present in cases than in controls, but did not reach statistical significance (0.75 vs. 0.62; OR 1.82, 95% CI 0.65–5.26, p = 0.259) [17].

In the Egyptian breast cancer cohort, CBR1 rs20572 was associated with significantly higher exposure levels of doxorubicin [32]. While CBR1 rs9024 revealed no significant association with AIC (AG vs. GG: OR 1.49, 95% CI 0.30–7.40, p = 0.624) in a study by Vaitiekus et al. [14].

Uridine 5′-diphospho-GlucuronosylTransferases (UGT)

Li et al. reported the function of UGT2B7-161 rs7668258 in breast cancer patients undergoing epirubicin chemotherapy. Multivariate logistic regression showed that the − 161 T allele was independently associated with lower risk of cardiotoxicity (additive model: CC-0, CT-1, TT-2, OR 0.259, 95% CI 0.103–0.651, p = 0.0004) [30]. Trastuzumab administration, higher cumulative doses of epirubicin, and elevated cardiac biomarker concentrations were identified as independent risk factors for cardiotoxicity [30]. UGT1A6 rs17863783 did not reach statistical significance in three studies [14, 24, 33].

Glutathione S-Transferases (GST)

Volkan-Salanci et al. reported no initial variation in cardiac parameters across GSTP1 genotypes in the initial follow-up. However, at a 1-year follow-up, GSTP1 G-allele carriers (AG + GG) had a higher risk of AIC compared to AA carriers, with lower fractional shortening (−11.3 ± 12.3% vs. −1.7 ± 10.7%; P = 0.024) [34], higher end systolic diameter (7.3 ± 12.1 vs. −0.9 ± 8.7%; p = 0.018) [34], and more frequent peak filling rate decline (85% vs. 15.4% of AA; p = 0.007) [34]. Vivenza et al. reported no significant correlation of GSTT1 or GSTP1. But, elevated cardiotoxicity risk (p = 0.147) has been identified in the GSTM1 null genotype, which did not reach conventional levels of statistical significance [35].

Oxidative stress & antioxidant pathways

Neutrophil Cytosolic Factor 4 (NCF4)

A study by Vaitiekus et al. reported a risk reduction tendency for NCF4 rs1883112 (Dominant model: OR = 0.49, 95% CI 0.27–0.87, p = 0.015) [14]. In contrast, Hertz et al. and Vulsteke et al. reported no association of NCF4 rs1883112 with AIC [24, 28].

Nitric Oxide Synthase 3 (NOS3)

The NOS3 rs1799983 TT carriers are associated with higher risk of AIC (OR = 3.06, 95% CI: 1.21–7.73, p = 0.018) [20, 31].

Nicotinamide Adenine Dinucleotide Phosphate (NADPH)

The TT genotype of the NADPH oxidase rs4673 is associated with AIC (OR = 2.75, 95% CI: 1.31–5.80, p = 0.0077) [20, 31].

Ras-related C3 botulinum toxin substrate (RAC)

Hertz et al. reported the effect of RAC2 rs13058338 on systolic dysfunction, which did not reach statistical significance in the additive model (OR 0.75, 95% CI: 0.31–1.82, p = 0.38) [24]. Similarly, Vulsteke et al. did not observe any significant association for this variant [28].

Glutathione Peroxidase 1 (GPX1)

Kopeva et al. identified that the GPX1 rs1050450 CC genotype was associated with AIC (OR = 2.345, p = 0.007) and suggested it as a potential genetic marker for risk assessment before chemotherapy [20].

Electron transfer Flavoprotein subunit Beta (ETFB)

The ETFB rs79338777 minor T allele carriers are at risk of chronic AIC in breast cancer cohort (discovery cohort) which later replicated in childhood cancer cohort (replication cohort) when analysed separately (discovery cohort: OR 14.1, 95% CI 1.60–124, p = 0.017; replication: OR 6.17, 95% CI 1.61–27.7, p = 7.97 × 9 × 10^{− 3}) [29].

The other explored variants of Oxidative Stress & Antioxidant Pathways, such as CYBA rs4673 [24, 28], CYBA rs1049255 [14] and SOD2 rs4880 [20], PON1 rs662 (AG vs. AA, OR 0.724, 95% CI: 0.246–4.946, p = 0.80) [14] did not reach statistical significance in breast cancer patients.

Cardiac remodelling & signalling

Transient receptor potential cation channel subfamily C (TRPC6)

Advani et al. identified a significant association of TRPC6 rs77679196 with anthracycline-induced cardiotoxicity (AIC) in the NSABP B-31 trial (OR = 12.84, 95% CI 1.24–133.2; p = 0.032). The massive effect estimate was accompanied by wide confidence intervals. This can be attributed to the small number of cardiac events in the analysed NSABP B-31 chemotherapy sub-cohort (N = 10). In an independent cohort from the Mayo Clinic Biobank, Norton et al. reported that the TRPC6 rs36111323 variant was significantly more frequent among heart failure cases than among anthracycline/trastuzumab-exposed controls (OR = 2.01; p = 0.002), after adjustment for sex and hypertension.

RARG rs2229774 is not associated with AIC risk (P = 0.471) in a study by Nyangwara V et al. [33]. Similarly, Hertz et al. demonstrated a non-significant association of TOP2B rs10865801 with AIC (Additive model: OR 1.32, 95% CI 0.67–2.61, P = 0.47) [24]. Advani et al. evaluated variants in BRINP1 rs62568637, LDB2 rs55756123 (OR 1.93, 95% CI 0.25–14.7, p value = 0.525), and RAB22A rs707557 (OR 1.28, 95% CI 0.51–3.12, p value = 0.598), but none of these demonstrated significant associations [17]. Investigations of EDNRA rs5335 did not reach statistical significance [20, 31].

Iron metabolism genes

Homeostatic iron regulator (HFE)

The G allele carriers of HFE rs1799945 have an increased risk of developing AIC (OR = 3.44, 95% CI 1.40–8.47, p = 0.005). In contrast, the HFE rs1800562 variant is not statistically significant [18].

DNA damage response and cell cycle regulation

Kopeva et al. identified that AIC was significantly associated with the TP53 arginine genotype (OR = 2.97, p = 0.001), while the proline genotype appeared to have a protective effect (OR = 0.126, p = 0.028) [20]. Similarly, Vivenza et al. reported that the patients carrying the GC (p.Pro72Arg, 42%) and GG (p.Arg72Arg, 50%) genotype showed a higher prevalence of hypertension compared to those with the CC (p.Pro72Pro, 8%) genotype (p = 0.180). The association was borderline significant when GC and GG carriers were grouped and compared with CC patients (p = 0.077) [35].

In a study by Liu et al., the frequency of cardiac events did not reach statistical significance in ATM variants such as ATM rs1003623 (χ² = 1.57, p = 0.46) ATM rs227060 (χ² = 0.85, p = 0.65) ATM rs228589 (χ² = 0.82, p = 0.57) ATM rs664143 (χ² = 0.20, p = 0.90) and ATM rs664677 (χ² = 0.95, p = 0.62) [16].

Variants in CASP3, such as CASP3 rs1049216 (χ² = 1.75, p = 0.42), CASP3 rs12108497 (χ² = 0.97, p = 0.61), CASP3 rs2720376 (χ² = 0.46, p = 0.79), in breast cancer patients receiving anthracyclines did not reach significance [16].

Similarly, CRYAB rs14133 (χ² = 6.31, p = 0.10), and STMN1 rs182455 (χ² = 2.84, p = 0.24) did not demonstrated any statistical significance. Kopeva et al. reported insignificant association of AIC with CASP8 variants such as CASP8 rs3834129 (χ² = 0.002, p = 0.961) CASP8 rs1045485 (χ² = 0.073, p = 0.786) [20].

Autophagy & apoptosis genes

In a study by Liu et al., Chi-square analysis did not reveal any association between ATG gene variants and AIC. In ATG5, neither ATG5 rs473543 (χ² = 1.42, p = 0.49) nor ATG5 rs3761796 (χ² = 2.06, p = 0.36) indicated any significant association. Similarly, ATG7 rs2594971 (χ² = 0.37, p = 0.83), ATG7 rs111595248 (χ² = 0.33, p = 0.56), and ATG7 rs4684789 (χ² = 0.50, p = 0.78), were not correlated with cardiotoxicity [16].

Liu et al. observed that carriers of the G allele at ATG13 rs10838611 have abnormal electrocardiogram findings (OR = 2.26, 95% CI = 1.32–3.87; p < 0.01), indicating a higher probability of cardiac dysfunction. MAP1LC3A rs4911429 (χ² = 1.17, p = 0.56), MAP1LC3A rs6088521 (χ² = 0.67, p = 0.72) and MAP1LC3B rs9903 (χ² = 0.55, p = 0.76), MAP1LC3B rs35227715 (χ² = 0.53, p = 0.77), MAP1LC3B rs7865 (χ² = 0.57, p = 0.45), or MAP1LC3B rs16944733 (χ² = 1.58, p = 0.45) are not significantly associated with cardiac events [16].

Inflammatory and immune response genes

Todorova et al. carried out a candidate gene association study in breast cancer patients treated with doxorubicin and identified nine genes with eighteen genetic variants, including the human leucocyte antigen region, associated with abnormal decline in LVEF. The authors reported the estimated odds ratio for the development of cardiotoxicity for the minor allele, with the major allele as reference, although confidence intervals were not provided significant associations were reported for variants in HLA-C rs9264942 (OR = 8.61, p = 0.01), HLA- C rs2523619 (OR = 6.56, p = 0.01), HLA-C rs10484554 (OR = 5.41, p = 0.04) [36].

NFKBIL1 variants such as NFKBIL1 rs2071591 (OR = 6.83, p = 0.02), NFKBIL1 rs3093949 (OR = 8.87, p = 0.01), NFKBIL1 rs2071592 (OR = 7.99, p = 0.01), which have been associated with inflammatory and autoimmune disorders like Myocardial infarction [37] Graves’ disease [38], and rheumatoid arthritis [39] were also associated with AIC [36].

Additionally, AIC associations were reported for C6orf10 rs2050190 (OR = 3.89, p = 0.02), TNF-α rs1800629 (OR = 5.67, p = 0.03), MSH5 rs3131379 (OR = 11.58, p = 0.04), MSH5 rs3131378 (OR = 11.58, p = 0.04), MICA rs2523451 (OR = 4.50, p = 0.04), LTA rs909253 (OR = 6.83, p = 0.02) [36], LTA rs1041981 (OR = 6.83, p = 0.02), BAT1 rs2239527 (OR = 4.13, p = 0.05), NOTCH4 rs3134931 (OR = 0.22, p = 0.05), ATP6V1G rs2071594 (OR = 6.83, p = 0.02), ATP6V1G rs3130059 (OR = 6.83, p = 0.02) and ATP6V1G rs11796 (OR = 4.12, p = 0.03) [36].

Renin–Angiotensin–Aldosterone system (RAAS) genes

In a study by Vivenza et al., the baseline serum aldosterone levels showed a significant difference among CYP11B2 rs1799998 genotypes (p = 0.03), with the highest levels in CT, intermediate in TT, and lowest in CC carriers [35] The small cohort size limited the assessment of other cardiotoxicity outcomes in this study [35].

The ACE rs4340 variant was associated with higher diastolic blood pressure (DBP) and serum aldosterone levels (p = 0.04 and p = 0.003, respectively). Similarly, the AGT rs699 polymorphism is also correlated with serum aldosterone, with higher levels in CC carriers than CT or TT carriers (p = 0.045) [35].

No associations were observed for AGTR1 A1166C or AGT p.Thr174Met with aldosterone or blood pressure [35].

Discussion

The current systematic review explored the genetic association of breast cancer patients undergoing anthracycline chemotherapy. As the overall survival rate for patients with breast cancer has increased, it is crucial to preserve and improve their quality of life. One of the major side effects of treatment that can still have an impact on long-term results is cardiotoxicity.

Pharmacogenomic variants modulate anthracycline-induced cardiotoxicity (AIC) in breast cancer patients through gene-specific gain or loss-of-function effects that alter drug transport, drug metabolism, redox homeostasis, calcium signalling, and DNA damage response. For example, the ABC transporter gene, ABCC1, was first discovered in a doxorubicin-resistant small cell lung cancer cell line that did not overexpress P-glycoprotein [40]. This family predominantly exerts a gain-of-function effect, known to influence the pharmacokinetics of anthracyclines by enhancing efflux activity, which leads to drug resistance in cancer cells [41]. Similarly, Anthracycline metabolism plays a key role in AIC. The CBR3 rs1056892 variant makes doxorubicin more susceptible to anthracycline-induced cardiotoxicity by increasing its catalytic conversion to the cardiotoxic metabolite doxorubicinol [10]. Similarly, UGT2B7, a key enzyme responsible for the glucuronidation and elimination of anthracycline metabolites, exhibits a protective effect where the rs7668258 variant is associated with a reduced cardiotoxic risk due to improved detoxification efficiency [42].

The other widely studied mechanism of AIC includes the production of oxygen-free radicals that damage the heart. While RAC2, NADPH oxidases (NOX), uncoupled nitric oxide synthase, and mitochondria are major enzymatic sources for reactive oxygen species generation [43] the NCF4 rs1883112 polymorphism, due to its loss-of-function effect, inhibits oxidase activation and reduces the formation of reactive oxidant intermediates [14] and exhibits cardioprotective effect.

The iron metabolism genes, such as the HFE rs1799945 variant, that are involved in the intracellular metabolism of doxorubicin, are also associated with the generation of reactive oxygen intermediates through a gain-of-function effect [18]. Similarly, the increase in calcium ion influx due to the TRPC6 mutation leads to cardiac hypertrophy [17]. Genes such as ATM kinase play a crucial role in response to DNA damage and are associated with the response to oxidative stress [44]. Furthermore, the pharmacogenetic contributions of genes involved in inflammatory and immune responses, the RAAS system, and autophagy and apoptosis pathways require further exploration.

In this current review, breast cancer patients with ABCC1 variants, such as the rs4148350 TG genotype [13] and the rs246221 heterozygous T-allele carriers [28], were at increased risk of AIC. Similarly, the GG carriers of the CBR3 rs1056892 variant had a higher risk of decline in LVEF following anthracycline therapy [15, 24, 34]. In a one-year follow-up, G-allele carriers of the GSTP1 polymorphism have lower fractional shortening, higher end-systolic diameter, and a decline in peak filling rate [34].

Similarly, it has been found that the TT carriers of NOS3 rs1799983 [20, 31], NADPH rs4673 [20, 31], are prone to cardiac dysfunction, while the GPX1 rs1050450 CC genotype represents a loss-of-function antioxidant variant increasing susceptibility to oxidative injury [20].The minor T allele of ETFB rs7938777 can also serve as a marker for cardiotoxic risk assessment [29]. The higher frequency of the TRPC6 variant has been observed in heart failure cases [19].

In iron metabolism genes, the G allele of HFE rs1799945 has been associated with AIC [14]. In the class of DNA damage response genes, carriers of the TP53 gene with the GG or GC nucleotide sequence are at risk of developing hypertension [35]. The G allele of the ATG13 gene, related to autophagy and apoptosis, has been linked to echocardiographic changes [16]. The CT carriers of CYP11B2 rs1799998 are associated with an increase in serum aldosterone levels, and the variant of ACE rs4340, AGT rs699, raises diastolic blood pressure [35]. Across nine genes, eighteen genetic variants in the HLA region have been associated with a fall in LVEF [36].

In contrast, genetic variants of UGT2B7−161 rs7668258 [30], NCF4 rs1883112 [14], and TP53 [20] have been associated with the risk reduction of AIC.

Understanding the genetic variants in anthracycline pathways may inform future therapeutic strategies. For instance, statins, due to their anti-inflammatory and antioxidant effects, have been investigated as potential cardioprotective agents for cardiomyopathy [45]. Similarly, dexrazoxane impairs the formation of iron compounds and oxygen-free radicals [46].

In addition to breast cancer, anthracyclines are also widely used in childhood cancers and haematological malignancies. Several studies have also reported the genetic predisposition of AIC in this cohort. Previous associations of ABCC2 rs8287710 that have been reported in haematological malignancies [47], and ABCC1 rs3743527, linked to decreased LVEF in childhood acute lymphoblastic leukaemia [48], did not reach significance in the breast cancer cohort. Similarly, SLC28A3 rs7853758 has been strongly associated with AIC in other cancer types (OR 0.35, p = 0.00008) [23] has not reached statistical significance in breast cancer patients [24, 28, 33]. For the CBR3 rs1056892 variant the similar findings were observed across both breast cancer [24, 34] and haematological malignancy cohorts (OR 1.79, p = 0.02) [10].

Following doxorubicin therapy in childhood leukaemia, heterozygous HFE rs1800562 carriers showed recurrent increases in cardiac troponin T and decreased left ventricular function [49]. In a Spanish cohort, anthracycline exposure and the HFE haplotypes were also associated with iron accumulation [50]. However, rs1800562 did not significantly correlate with AIC risk in breast cancer patients [18]. The other gene, RAC2, encodes a GTPase essential for NADPH oxidase activity [51, 52]. But, the RAC2 rs13058338 variant that is associated with congestive heart failure in haematological malignancies (OR 2.8, 95% CI: 1.4–5.6) [47], did not show statistically significant associations in breast cancer patients [24, 28]. While level B (moderate) evidence through a systematic evidence synthesis approach supports pharmacogenomic testing for RARG rs2229774, SLC28A3 rs7853758, and UGT1A6 rs17863783 in childhood cancer patients treated with doxorubicin or daunorubicin [53] did not reach statistical significance in patients with breast cancer.

This lack of replication observed in the breast cancer cohorts can be mainly attributed to population differences rather than the absence of a true genetic effect. Pharmacogenomic studies are widely influenced by minor allelic frequencies in the population. For instance, a study by Aminkeng et al. reported that the RARG rs2229774 variant has a minor allele frequency of approximately 25% in the European children’s cohort. It demonstrated a strong association with AIC, with an odds ratio of 4.7 (95% CI 2.7–8.3), suggesting sufficient power to detect a genetic effect [54]. However, the variant had a lower minor allele frequency of 14% in African ancestry populations, which may have limited the replication of this association in the breast cancer cohort [33].

Likewise, a protective effect is linked to the minor A allele of SLC28A3 rs7853758. The high allele frequency of 39% in Mexican paediatric anthracycline-treated cohorts provides sufficient power to identify a protective effect against cardiotoxicity [55]. The same allele occurs at a frequency of about 16–17% in Caucasian breast cancer cohorts, as reported in BCIRG 006 [56]. This could make it comparatively difficult to replicate protective associations in this cohort. Similarly, in a study by Visscher et al., Canadian paediatric cohorts reported a strong protective association. The A allele was more frequent in controls than in cases (20% vs. 7.7%), with a combined odds ratio of 0.31 (95% CI, 0.16–0.60). However, when replicated in an independent Dutch cohort, the effect estimate did not reach statistical significance, which may be linked to the low minor allelic frequency [23].

Hence, one of the major limitations of pharmacogenomic studies is that they cannot be broadly generalizable, as the associations are strongly influenced by ancestry-specific differences in minor allele frequencies and linkage disequilibrium patterns.

The discrepancies in genetic effects across distinct populations are influenced by various factors. In the dose-stratified analysis, the protective effect of the SLC28A3 variant is observed to be significant at higher cumulative anthracycline doses (> 250 mg/m²) (odds ratio, 0.43, P < 0.01), but not at lower doses (< 150 mg/m²) [57]. It implies that the genetic effect interacts with dosage exposure, and this protective association may not be statistically significant when the anthracycline dose is relatively low, which is frequently the case in many adult breast cancer protocols.

Cardiovascular comorbidities such as diabetes, hypertension, and pre-existing coronary disease are more common in adults. These risk factors can conceal subtle genetic influences that are noticeable in the paediatric population without comorbidities [58].

Additionally, anthracyclines are combined with other sequential cardiotoxic agents, such as trastuzumab, in breast cancer regimens [19], which makes it more difficult to isolate single gene-drug interactions and reduces the power to observe variant effects that are easier to detect in paediatric regimens. In retrospective analysis of breast cancer patients, the “healthy survivor” effect may decrease the apparent genetic effect, while paediatric research frequently tracks all treated patients prospectively and records a more complete incidence of cardiotoxicity.

The incidence of a clinically significant reduction in the LVEF is higher in patients receiving both trastuzumab and an anthracycline than in those receiving anthracyclines alone (36% vs. 9.5%, p = 0.001) [59].

Trastuzumab and radiotherapy are important modifiers of cardiotoxicity risk in breast cancer patients and can influence the interpretation of pharmacogenomic associations. Although several studies accounted for trastuzumab exposure, either through stratified analyses comparing chemotherapy alone versus chemotherapy plus trastuzumab [17, 19] or by adjusted odds ratios to trastuzumab exposure [15, 24, 34, 42]. This was not uniformly applied across all studies. In contrast, radiotherapy was considered in only a limited number of analyses [24, 28].The lack of uniform adjustment for concomitant cardiotoxic therapies likely contributes to heterogeneity and reduces the reproducibility of genetic associations in breast cancer studies.

Trastuzumab causes cardiotoxicity through a different mechanism than anthracyclines. While anthracyclines directly damage cardiac cells and generate reactive oxygen species through redox cycling and NADPH oxidase activity, trastuzumab interferes with HER2-mediated survival signalling in cardiomyocytes, leading to impaired protection against oxidative stress and increased mitochondrial vulnerability [60]. Because trastuzumab cardiotoxicity is primarily Type II rather than structural damage, the contribution of classic oxidative stress gene variants (such as NOS3 or NADPH oxidase genes) to trastuzumab-related cardiotoxicity is less well established [61].

There is a wide heterogeneity in the definition of cardiotoxicity, ranging from an asymptomatic drop in LVEF to overt heart failure. The interpretation of genetic associations is affected by substantial variation. A drop in LVEF indicates early, subclinical myocardial damage. Symptomatic heart failure indicates a more advanced condition.

Studies that used strict clinical endpoints, such as symptomatic congestive heart failure or cardiac death [17, 19, 35] more consistently identified associations with TRPC6 and TP53 variants. TRPC6 plays a role in calcium signalling and cardiac hypertrophy, which supports its stronger association with overt heart failure rather than mild functional changes.

In contrast, studies defining cardiotoxicity mainly as asymptomatic LVEF decline of at least 10% [14, 18, 31] more often reported associations with genes involved in oxidative stress and drug metabolism, such as CBR3, NCF4, HFE, and NOS3. These genes are linked to early myocardial injury and subclinical cardiac remodelling, which may not immediately progress to clinical heart failure. AIC is more likely a spectrum that begins with subclinical myocardial injury and early asymptomatic declines in LVEF, progressing to clinical heart failure if untreated, which represents the notion of a pathophysiological continuum [60, 61].

A quantitative meta-analysis was not feasible due to substantial heterogeneity across studies. Many studies have reported genetic associations using different inheritance models and different effect estimates. These have limited the feasibility of pooling data. To address this, study-level effect sizes and heterogeneity are presented as forest plots in Figs. 3 and 4. The genetic associations investigated in various inheritance models are presented in Supplementary Table 3. We used qualitative methods to synthesize the evidence.

Fig. 3.

Forest plot showing odds ratios (ORs) and 95% confidence intervals (Cls) for the association between genetic variants and anthracycline-induced cardiotoxicity (AIC). ORs > 1 indicate increased risk, whereas ORs < 1 indicate a protective effect

Fig. 4.

Forest plot of odds ratios (ORs) and 95% confidence intervals (cls) associated with AIC reported under Additive, Dominant, and Recessive models. ORs > 1 indicate increased risk, whereas ORs < 1 indicate a protective effect

Despite using a rigorous and systematic approach, this study has some limitations. In pharmacogenomic research, studies with null results or small sample sizes are less likely to be published, which could lead to inflated effect estimates. Many of the included studies were underpowered and used inconsistent outcome definitions and methods to evaluate cardiotoxicity. Moreover, the genetic associations discussed represent only part of the complex biology underlying anthracycline-induced cardiotoxicity. Other relevant genetic variants, pathways, or gene-environment interactions, as well as clinical factors, may also influence AIC. Given the presence of multiple clinical risk factors for anthracycline-induced cardiotoxicity, isolating the effect of individual SNPs can be challenging. Future well-powered, multicentre prospective studies with standardized cardiotoxicity measures and comprehensive genomic analyses are needed to address these limitations.

For patients with breast cancer, a genomic perspective of AIC provides opportunities for individualised risk assessment and treatment. By identifying individuals at higher risk, clinicians can customise chemotherapy regimens or consider cardioprotective medications like dexrazoxane [4, 62]. The current anthracycline chemotherapy surveillance protocol involves baseline and serial echocardiography, and the measurement of cardiac biomarkers to identify early myocardial damage [63]. Predictive accuracy can be improved by combining the genetic risk factors with the cumulative dose of anthracycline and clinical risk factors.

Conclusion

Genetic variations associated with various cardiotoxicity-related pathways in breast cancer patients have been summarized in the current systematic review. These variations in the drug metabolizing pathway CBR3 rs1056892 have been consistently associated independently with AIC in four studies. Although a strong association between transporter gene variants and cardiotoxicity has been reported in breast cancer cohorts, these findings require validation in larger, ethnically diverse populations with extended follow-up. Similarly, genomic variants in different pathways explored in breast cancer patients require further investigation to reach strong evidence. While a major portion of the earlier evidence has come from childhood cancer cohorts, large-scale studies dedicated to the breast cancer population will be crucial for the validation of these findings and for translating these findings into clinical practice. While current evidence remains preliminary, the validation of genomic determinants may enable pharmacogenomic testing to guide anthracycline use in breast cancer patients.

Abbreviations

ABCB1

ATP Binding Cassette Subfamily B Member 1

ABCC1

ATP Binding Cassette Subfamily C Member 1

ABCC2

ATP Binding Cassette Subfamily C Member 2

ACE

Angiotensin-Converting Enzyme

AGT

Angiotensinogen

AGTR1

Angiotensin II Receptor Type 1

AIC

Anthracycline-Induced Cardiotoxicity

ATG12

Autophagy Related 12

ATG13

Autophagy Related 13

ATG5

Autophagy Related 5

ATG7

Autophagy Related 7

ATM

ATM Serine/Threonine Kinase

ATP6V1G2-DDX39B

ATPase H+ Transporting V1 Subunit G2 / DEAD-Box Helicase 39B

BAT1

HLA-B Associated Transcript 1

BRINP1

BMP/Retinoic Acid Inducible Neural Specific 1

CASP3

Caspase 3

CASP8

Caspase 8

CBR1

Carbonyl Reductase 1

CBR3

Carbonyl Reductase 3

CRYAB

Crystallin Alpha B

CTRCD

Cancer Therapy-Related Cardiac Dysfunction

CYBA

Cytochrome b-245 Alpha Chain

CYP11B2

Cytochrome P450 Family 11 Subfamily B Member 2

EDNRA

Endothelin Receptor Type A

ETFB

Electron Transfer Flavoprotein Subunit Beta

GPX1

Glutathione Peroxidase 1

GSTM1

Glutathione S-Transferase Mu 1

GSTP1

Glutathione S-Transferase Pi 1

GSTT1

Glutathione S-Transferase Theta 1

HFE

Homeostatic Iron Regulator

IL1B

Interleukin 1 Beta

LDB2

LIM Domain Binding 2

LINC01060

Long Intergenic Non-Protein Coding RNA 1060

LTA

Lymphotoxin Alpha

LVEF

Left Ventricular Ejection Fraction

MAP1LC3A

Microtubule Associated Protein 1 Light Chain 3 Alpha

MAP1LC3B

Microtubule Associated Protein 1 Light Chain 3 Beta

MICA

MHC Class I Polypeptide-Related Sequence A

MSH5

MutS Homolog 5

NADPH

Nicotinamide Adenine Dinucleotide Phosphate

NCF4

Neutrophil Cytosolic Factor 4

NFKBIL1

NFKB Inhibitor Like 1

NOS3

Nitric Oxide Synthase 3

NOTCH4

Notch Receptor 4

NR1I2

Nuclear Receptor Subfamily 1 Group I Member 2

RAB22A

RAB22A, Member RAS Oncogene Family

RAC2

Rac Family Small GTPase 2

RARG

Retinoic Acid Receptor Gamma

ROS

Reactive Oxygen Species

SNP

Single Nucleotide Polymorphism

SLC22A16

Solute Carrier Family 22 Member 16

SLC28A3

Solute Carrier Family 28 Member 3

SOD2

Superoxide Dismutase 2

STMN1

Stathmin 1

SULT2B1

Sulfotransferase Family 2B Member 1

TNF

Tumor Necrosis Factor

TOP2B

DNA Topoisomerase II Beta

TP53

Tumor Protein p53

TRPC6

Transient Receptor Potential Cation Channel Subfamily C Member 6

UGT1A6

UDP Glucuronosyltransferase Family 1 Member A6

UGT2B7

UDP Glucuronosyltransferase Family 2 Member B7

Authors’ contributions

All authors contributed to the study conception and design.Literature search was performed by V.B, N.A, A.K. Data analysis was performed by V.B, B.S, A.C, A.K, M.R and M.M. V.B, A.K, M.M wrote the manuscript text. V.B, M.M prepared the Figs. 1 and 2.S.S, S.D, S.J, S.M, K.U, J.S, R.K.K and M.M reviewed and edited the manuscript. All authors read and approved the final manuscript.

Funding

No, this research did not receive funding.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.