Breast MRI phenotype → cancer treatment exposure → ECG change → cardiovascular outcome

From the breast MRI or its report, collect:

• Left- versus right-sided cancer
• Inner, central, or outer breast location
• Distance from the tumor or surgical bed to the chest wall
• Pectoralis or chest wall involvement
• Internal mammary nodal involvement
• Tumor size, multifocality, and distribution
• Type and extent of surgery
• Need for regional nodal irradiation

Then connect these variables with the radiation plan:

• Mean heart dose
• Left ventricular dose
• Right ventricular dose
• Left anterior descending coronary artery dose
• Left atrial and right atrial dose
• Dose to cardiac base, septum, and conduction system regions when segmentation is feasible
• Internal mammary chain treatment
• Deep inspiration breath hold use
• Fractionation and total dose

Study question. Do breast tumor location and treatment geometry identify patients who experience measurable ECG changes following breast radiotherapy? A prospective breast radiotherapy study found that anterior lead ECG changes, particularly T wave changes, occurred after treatment and were associated with mean heart radiation dose and echocardiographic strain changes, supporting serial ECG as a plausible low cost marker for radiation associated cardiac effects.

When the original images remain available, measurements can be performed manually on standardized slices:

• Pectoralis major and minor muscle area
• Muscle signal intensity or quality, with careful sequence standardization
• Subcutaneous chest wall adipose thickness
• Breast adipose and fibroglandular composition
• Sarcopenic or low muscle phenotype
• Obesity phenotype when linked to BMI
• Change in body composition when more than one clinical breast MRI already exists

Scientific question. Do baseline body composition characteristics modify the relationship between anthracycline, HER2 directed therapy, radiation exposure, and subsequent ECG or cardiac outcomes? This is especially valuable where the breast MRI patients overlap with the BMI cohort.

A blinded radiologist or cardiologist could review the breast MRI for prespecified findings:

• Pericardial effusion
• Pleural effusion
• Gross cardiac enlargement when adequately visualized
• Gross atrial or ventricular asymmetry
• Ascending aortic or pulmonary artery enlargement when included
• Intracardiac, pericardial, or mediastinal masses
• Chest wall and pericardial relationships

Breast MRI can contain clinically relevant extramammary cardiovascular findings, although field of view and motion influence ascertainment. The design would use two independent readers, a structured reading form, blinded adjudication, and interreader reliability, and would remain an opportunistic thoracic imaging study rather than a ventricular function study.

Even with reports alone, the MRI establishes the cancer phenotype that determines treatment:

• HER2 positive disease and exposure to HER2 directed therapy
• Tumor burden and anthracycline use
• Neoadjuvant treatment and MRI response
• Residual disease and escalation of therapy
• Left-sided disease and radiation exposure
• Regional nodal disease and broader radiation fields

The meaningful analysis focuses on the treatment pathway connecting cancer phenotype to cardiovascular effects, rather than attributing the cardiovascular effect directly to the tumor image.

Collect the ECG closest to treatment initiation, preferably before anthracycline, HER2 directed therapy, or chest radiation. A baseline ECG is already part of recommended cardiovascular risk assessment before potentially cardiotoxic therapy.

• Heart rate and rhythm
• PR interval
• QRS duration
• QT and Fridericia-corrected QT
• P wave duration and axis
• QRS and T wave axes
• Spatial QRS-T angle when digital signals permit it
• Low voltage
• Left ventricular hypertrophy criteria
• Bundle branch or fascicular block
• Atrioventricular conduction abnormalities
• Fragmented QRS
• ST segment and T wave abnormalities
• Premature atrial or ventricular complexes
• Atrial fibrillation or flutter

Principal question. Which baseline ECG features identify patients who subsequently develop echocardiographic cardiac dysfunction, heart failure, arrhythmia, treatment interruption, or cardiovascular hospitalization?

Likely the most valuable first study. Obtain every ECG available before, during, and after therapy, then align each ECG with treatment dates and study changes in:

• QTc
• PR and QRS duration
• T wave amplitude, direction, and axis
• ST-T abnormalities
• Fragmented QRS
• New conduction disease
• New atrial fibrillation
• New ectopy
• Resting heart rate
• Low voltage development
• QRS-T angle

Treatment-specific trajectories can be examined after:

• Anthracycline exposure
• Trastuzumab or other HER2 directed therapy
• Left- versus right-sided radiation
• Combined anthracycline and HER2 therapy
• CDK4/6 inhibitors or other QT-active agents
• Endocrine therapy
• Immune checkpoint therapy, when present

When serial echocardiograms are available, compare the timing of:

1. ECG change
2. GLS decline
3. LVEF decline
4. Clinical heart failure or treatment modification

Question. Do conventional ECG changes emerge before measurable echocardiographic cancer therapy–related cardiac dysfunction? The ECG serves as a predictor or intermediate phenotype; echocardiography, cardiac biomarkers, or adjudicated clinical events define the cardiac outcome.

Compare exposure measures:

• Left- versus right-sided disease
• Mean heart dose
• Ventricular and atrial doses
• LAD dose
• Dose to the interventricular septum
• Dose to the cardiac base and conduction system regions
• Internal mammary nodal treatment
• Deep inspiration breath hold use

against electrical response measures:

• Anterior lead T wave changes
• ST segment changes
• QTc changes
• QRS duration
• Fragmented QRS
• New conduction abnormalities
• Atrial arrhythmias

A dose-response design is stronger than a simple left-versus-right comparison, because patients with the same laterality can receive substantially different cardiac doses.

Possibly the most distinctive multimodal project: does medial or chest wall adjacent breast cancer lead to greater cardiac radiation exposure, followed by greater longitudinal ECG change? The pathway is analyzed in stages — MRI-defined tumor location → treatment field and cardiac dose → serial ECG phenotype — with radiation plan measurements verifying the relationship between prone breast MRI anatomy and actual treatment geometry.

Whether pectoralis muscle area, subcutaneous adiposity, BMI, and sarcopenic obesity are associated with:

• Baseline QTc or QRS-T angle
• Atrial electrical abnormalities
• Low voltage
• Treatment-related ECG change
• Later atrial fibrillation
• Cancer therapy–related cardiac dysfunction
• Cardiovascular hospitalization

This connects the breast MRI dataset with the separate high-BMI cohort.

With Holter monitoring, telemetry, ambulatory ECG, or sufficiently long digital rhythm recordings, the study can evaluate:

• Heart rate variability
• Deceleration capacity
• Mean heart rate
• Circadian heart rate patterns
• Ectopic burden
• Nonsustained atrial or ventricular arrhythmias

A prospective study in HER2 positive breast cancer found that reduced heart rate deceleration capacity was associated with subsequent trastuzumab-related cardiotoxicity. A routine ten second ECG supports standard intervals and morphology; longer recordings provide the appropriate foundation for heart rate variability and deceleration capacity analyses.

• A baseline breast MRI or structured breast MRI report
• An ECG before treatment
• At least one ECG during or after treatment
• Complete systemic therapy information
• Radiation treatment information when applicable
• Follow-up echocardiography, biomarkers, or cardiovascular outcomes

When serial echocardiography is available: cancer therapy–related cardiac dysfunction defined using prespecified LVEF and GLS criteria.

When echocardiography is sparse: a clinically adjudicated cardiovascular outcome, such as heart failure, cardiovascular hospitalization, new atrial fibrillation, clinically significant arrhythmia, or cardiovascular treatment initiation.

• Baseline ECG phenotype
• Change in QTc, QRS, and ST-T morphology
• Anthracycline cumulative dose
• HER2 directed therapy
• Radiation laterality and cardiac dose
• MRI-defined tumor location
• BMI and MRI-derived body composition

• Paired pretreatment and posttreatment ECG comparisons
• Linear mixed-effects models for repeated ECG measurements
• Time-to-event models for cardiac outcomes
• Treatment-specific subgroup analyses
• Interaction analyses for obesity, body composition, and baseline cardiovascular risk
• Internal validation through bootstrapping
• Blinded cardiologist ECG adjudication on a representative subset

No deep learning model is required.