Cardiovascular magnetic resonance (CMR) is valuable for early cardiotoxicity detection by assessing cardiac anatomy, tissue composition, and function [9, 10]. CMR can detect subtle myocardial changes before a decline in left ventricular ejection fraction (LVEF) and accurately quantifies left ventricular mass changes that correlate with anthracycline dose and cardiac event risk [9]. Hybrid PET/MRI integrates PET’s functional and metabolic capacity with MRI’s enhanced anatomical detail and soft-tissue contrast [11]. This combination allows for enhanced lesion characterisation and precise image co-registration in a single session, providing incremental information beyond individual modalities. MRI additionally offers parameters such as dynamic contrast enhancement (DCE) for vascularisation assessment and diffusion-weighted imaging (DWI) for cellularity quantification, aiding in tumour characterisation [11, 12]. In cardio-oncology, ¹⁸F-FDG PET/MRI can identify early anthracycline-induced myocardial damage by assessing glucose uptake, and shows promise for cardiac sarcoidosis by combining late gadolinium enhancement (LGE) MRI for myocardial injury with ¹⁸F-FDG PET for active inflammation, offering stronger prognostic value than either modality alone [12, 13].

AI, in particular machine learning (ML) and deep learning (DL), is transforming cardio-oncology by improving risk stratification, optimising treatment, and enabling individualised care. These techniques are also instrumental for early detection of cardiac dysfunction, enabling timely and personalised interventions to prevent irreversible damage [14]. AI models analyse electrocardiogram (ECG), echocardiographic, CT, and MRI data to identify patterns indicative of cardiotoxicity [15]. In echocardiography, AI addresses limitations arising from user-dependency and measurement variability by automating complex assessments, enhancing diagnostic precision, and increasing accessibility [16]. Specific applications include improving the accuracy and efficiency of LVEF assessment and automating GLS measurement, a validated early marker for cardiotoxicity, thereby improving reproducibility and predictive accuracy [16, 17].

Novel radiotracers are designed with high specificity for molecular targets and optimal kinetics for signal detection, enabling paired diagnostic and therapeutic agents to target identical mechanisms, which is the core principle of theranostics [18]. Table 1 (see below) summarises key radiotracers with potential applications in cardio-oncology. Clinically, ⁶⁸Ga-DOTATATE and [⁶⁸Ga]pentixafor are most suited for imaging myocardial inflammation in the context of immune checkpoint inhibitor (ICI)-associated myocarditis, where CXCR4-positive immune cell infiltration can be quantified. The ¹⁸F-DHMT tracer targets superoxide production and is most applicable to early anthracycline cardiotoxicity surveillance, while ⁶⁸Ga-FAPI provides direct imaging of fibroblast activation protein, facilitating the characterisation of myocardial fibrosis in post-therapy cardiomyopathy. The advantage of each tracer lies in its mechanistic specificity; the disadvantage is the limited availability of radiopharmaceutical infrastructure outside specialist centres.

Anthracyclines exert their cardiotoxic effect primarily through topoisomerase IIβ inhibition, triggering mitochondrial reactive oxygen species (ROS) generation, double-strand DNA breaks within cardiomyocytes, and subsequent apoptosis. Trastuzumab, by contrast, impairs cardiomyocyte survival through ErbB2 receptor blockade, disrupting downstream survival signalling pathways including neuregulin-1/ErbB2 and PI3K/Akt. Cardiac troponins (cTnI and cTnT) are specific proteins released upon cardiomyocyte necrosis, providing a sensitive indicator of myocardial damage from these mechanisms [19]. High-sensitivity troponin (hs-Tn) assays enable earlier detection of subtle injury and predict subsequent irreversible CTRCD when elevated [20]. Natriuretic peptides (NPs), specifically B-type natriuretic peptide (BNP) and NT-proBNP, are released in response to myocardial stress and wall stretch from increased intracardiac pressures associated with ventricular dysfunction. Both troponins and NPs are recommended at baseline for patients receiving potentially cardiotoxic agents [20]. Limitations include their non-specificity to the underlying cause of damage and a lack of robust comparative analyses between different hs-Tn assays, hindering standardisation [20].

Circulating miRNAs (cmiRNAs) show promise as early biomarkers for anthracycline-induced cardiotoxicity. Specific cmiRNAs (e.g., miR-34a, miR-1, miR-423, miR-126, miR-130a) may predict cardiotoxicity development [20]. Myeloperoxidase (MPO), implicated in inflammation and oxidative stress, is a promising early biomarker for CTRCD in breast cancer patients receiving anthracyclines and trastuzumab [20, 21]. Combining multiple biomarkers, such as MPO and troponin, may improve predictive accuracy.

Liquid biopsy, a minimally invasive approach that analyses circulating tumour cells (CTCs), circulating tumour DNA (ctDNA), and other blood components, is primarily used in oncology for monitoring cancer progression [21, 22]. It carries additional potential for monitoring cardiotoxicity [22]. Changes in cardiomyocyte-specific cell-free methylated DNA (cfDNA) may indicate cardiomyocyte death from cardiotoxic therapies including radiation [22–24]. Importantly, cfDNA elevation following anthracycline exposure does not correlate closely with high-sensitivity cardiac troponin I (hs-cTnI), suggesting that these markers reflect distinct pathological processes and provide complementary mechanistic information. When cardiomyocyte cfDNA rises in the context of active cancer therapy, this signal may serve as a trigger for escalating cardioprotective measures, including the initiation of nanoparticle-based cardioprotective strategies, thereby operationalising the theranostic loop. Liquid biopsy enables dynamic tracking of both cancer progression and therapeutic cardiac responses, facilitating timely adjustments to therapy [23, 25].

Exosomes are extracellular vesicles that mediate intercellular communication by carrying heterogeneous cargo, including miRNAs [26]. Exosomal miRNAs (ex-miRs) are stable in circulation and carry potential as non-invasive biomarkers for heart failure and cardiotoxicity [24]. Specific ex-miRs, including miR-107, miR-146a, and miR-502, are potential markers of doxorubicin-induced cardiotoxicity (DIC) [24, 26]. Exosomal miR-34a-5p, derived from PD-1 inhibitor-treated macrophages, is mechanistically linked to ICI-related cardiac dysfunction through induction of cardiac senescence. Beyond diagnostic potential, ex-miRs signatures are emerging as functional guides for therapeutic decisions: elevated miR-34a-5p may indicate a patient phenotype more likely to respond to senolytic or mitochondria-targeted interventions, while miR-107 upregulation may support selection of liposomal doxorubicin formulations with optimised cardiac protection profiles. This diagnostic-to-therapeutic translation exemplifies the operational theranostic loop in biomarker-guided cardio-oncology [24, 26].

cfDNA originates from dead cells releasing genomic DNA into the bloodstream. Its tissue specificity allows identification of cardiomyocyte-derived cfDNA. Elevated cardiomyocyte cfDNA levels following anthracycline chemotherapy are significantly associated with an increased risk of subsequent CTRCD [26]. While hs-cTnI also rises in this context, its poor correlation with cardiomyocyte cfDNA indicates distinct underlying pathological processes, and the two biomarkers therefore provide complementary clinical information.

Encapsulation alters doxorubicin biodistribution, reducing accumulation in the myocardium while promoting tumour accumulation via the enhanced permeability and retention (EPR) effect [28]. This significantly lowers myocardial drug concentrations, reducing the topoisomerase IIβ-mediated oxidative injury that underlies anthracycline cardiotoxicity. Liposomal doxorubicin also induces interferon-stimulated genes (ISGs) in the heart, promoting cell survival and reducing apoptosis [28, 29].

pH-sensitive nanoparticles are engineered to respond to the acidic tumour microenvironment (TME), selectively releasing their payload at the disease site while maintaining stability in healthy tissues. Thermally responsive biopolymers respond to mild local hyperthermia (approximately 41°C) at the tumour site. These biopolymers promote preferential accumulation and drug release within the heated tumour, significantly reducing doxorubicin accumulation in the myocardium and preserving cardiac function [28, 29].

Immunomodulatory nanoparticles are designed to interact specifically with the immune system, enhancing or suppressing its activity for therapeutic outcomes [30]. They address challenges of conventional immunotherapy by delivering tumour antigens and adjuvants to antigen-presenting cells (APCs) and modulating the TME [30]. Nanotheranostics combine diagnostic and therapeutic capabilities, permitting direct delivery of agents to damaged cardiac tissues for targeted, localised treatment with minimal systemic adverse effects. This approach can improve the safety profile of ICIs, which can cause immune-related adverse events including myocarditis, by achieving specific and localised immune response modulation in tumour tissue while reducing systemic side effects [30]. Figure 1 provides an infographic summary of biomarkers of theranostic interest in cardio-oncology [19, 21, 24, 30].

ICIs trigger fulminant myocarditis via CD8+ T-cell-driven inflammation [30]. Emerging approaches combine [⁶⁸Ga]pentixafor PET, targeting CXCR4-positive inflammatory cells, with cardiac MRI to detect subclinical inflammation [13]. Biomarker-guided strategies using exosomal miR-34a-5p provide mechanistic insight into cardiac senescence, enabling risk stratification [24, 26].

Cytokine release syndrome (CRS) affects 50 to 90% of CAR-T recipients and causes haemodynamic compromise [32]. Theranostic monitoring combines liquid biopsy biomarkers, including endothelial-derived cfDNA and interleukin-6, with echocardiographic strain imaging to guide tiered interventions [25, 26]. Engineering advances incorporating on/off safety switches aim to minimise systemic toxicity [32].

The STOP-CA trial utilised baseline high-sensitivity troponin T (hs-TnT) and NT-proBNP to identify lymphoma patients likely to benefit from atorvastatin prophylaxis [31]. Statin recipients demonstrated reduced CTRCD rates (8% vs. 22% placebo), particularly those with elevated biomarkers [31]. Ongoing research is exploring polygenic risk scores to refine patient selection [20].

Gene editing carries significant potential for cardio-oncology. In vivo CRISPR-Cas9 strategies have successfully corrected pathogenic sarcomere gene variants in animal models of hypertrophic cardiomyopathy, reducing cardiac hypertrophy and fibrosis [33]. High-throughput CRISPR screens in human iPSC-derived cardiomyocytes are also identifying genes, including SLCO1A2 and SLCO1B3, that influence doxorubicin sensitivity [32]. Future CRISPR-based strategies may also be applicable to the repair of chemotherapy-induced DNA damage or patient-specific cardiomyopathy mutations, enabling truly personalised cardioprotection [33, 34].

ML constitutes another transformative tool. Models integrating patient data, including demographics, cancer type, and treatment regimens, can identify patients at high risk for cardiotoxicity [35]. DL applied to routine ECGs can detect subclinical declines in ejection fraction following anthracyclines, with reported AUC values of approximately 0.93 [36]. AI algorithms applied to cardiac imaging, including echocardiography, MRI, and CT, improve quantification of cardiac volumes and function, highlighting subtle changes before symptoms emerge [37]. These tools facilitate earlier intervention by identifying at-risk patients and streamlining surveillance.

Mitochondria-targeted antioxidants are a promising pharmacological advance. Compounds such as Mito-TEMPO and MitoQ accumulate in cardiomyocyte mitochondria and scavenge ROS. Preclinical studies demonstrate that Mito-TEMPO administered during chemotherapy significantly reduces doxorubicin- and 5-fluorouracil-induced cardiac injury by preserving mitochondrial function [38]. Mitochondria-targeted coenzyme Q analogues (e.g., MitoQ, SKQ1) have similarly protected animal hearts from drug-induced oxidative damage [37]. Clinical trials are required to confirm the safety and efficacy of these agents as co-administered cardioprotectants.

CAR-T therapy is expanding beyond haematological malignancies into solid tumours, with new designs improving cardiac safety. Recent engineering advances include affinity tuning and logic-gated receptors requiring two tumour signals to activate, thereby sparing healthy tissues. On/off safety switches and methods for local delivery of CAR-T cells aim to confine activity to tumour sites, retaining anti-tumour potency while minimising on-target, off-tumour cardiotoxicity of the type observed in early HER2-directed CAR trials [39].

Hybrid imaging modalities combined with AI analysis represent a further frontier. PET/CT or MRI/PET scans can measure myocardial metabolism and fibrosis in cancer patients, and AI-based analysis can detect early cardiotoxic changes. AI-enhanced hybrid imaging substantially improves the accuracy and efficiency of cardiotoxicity assessment. In practice, such pipelines could allow real-time monitoring of cardiac function during therapy, so that oncologists and cardiologists can adapt treatment, for example by pausing chemotherapy or initiating cardioprotective agents, at the earliest detectable signal [40].