Quick summary: Standard cardiovascular risk scores were not built for cancer patients. Our new state-of-the-art review in JACC: CardioOncology (Sen et al., April 2026) lays out a practical framework for building risk prediction tools that work across the cancer journey, covering competing risks, dynamic prediction, machine learning, and clinical implementation. The review is open access.
Who is actually at risk, and when?
It is one of the most important questions in modern cancer care. And one we still cannot answer well.
Cancer therapies are advancing faster than ever. So is the need to protect the heart before, during, and long after treatment. Anthracyclines. HER2-targeted therapies. Immune checkpoint inhibitors. Tyrosine kinase inhibitors. Radiation. Each carries its own cardiac signature. Each demands a different lens.
Get it right, and patients get the cancer treatment they need with the cardiac protection they deserve. Get it wrong, and we either miss the patients who matter most, or burden those who do not.
Why standard cardiovascular risk scores fall short in cancer patients
Most of the cardiovascular risk tools used in routine practice, including the Framingham Risk Score, the Pooled Cohort Equations, and SCORE2, were developed in populations without cancer. They assume stable patients, stable risk profiles, and long horizons undisturbed by competing causes of death.
Cancer changes all of that.
A patient on active chemotherapy is not the same patient they were three months ago. A breast cancer survivor with decades ahead needs a different prediction horizon to a lung cancer patient with limited prognosis. Standard cardiovascular risk scores do not flex with the cancer journey. When applied without adaptation, they miss, in both directions.
What our cardio-oncology review covers
Our new state-of-the-art review in JACC: CardioOncology lays out a framework for building cardio-oncology risk prediction tools properly. The key themes:
Define the right question
A useful prediction tool starts with a clear purpose. Target population, outcome of interest, index date, and prediction horizon each shape everything downstream.
Competing risks
A patient who dies of cancer cannot go on to develop heart failure. Treating cancer deaths as simple censoring inflates cardiovascular risk estimates. Cause-specific hazard models and Fine-Gray subdistribution hazard models are two standard approaches..
Treatment as a moving target
Incorporating cancer therapy into a risk model introduces immortal time bias and confounding by indication. Both can distort risk estimates if not handled carefully.
Dynamic risk prediction
Cardiovascular risk evolves with every treatment cycle, biomarker, and scan. Landmark models, joint models, and multistate models allow risk estimates to update over time.
Machine learning, used well
Random forests, gradient boosting, and deep learning survival models can capture complex interactions. Challenges include class imbalance, interpretability, and the need for external validation.
Clinical implementation
Workflow integration, transparency, equity, and ongoing recalibration determine whether a model actually reaches the bedside.
Who the review is for
The review is written for clinicians and researchers across cardiology, oncology, primary care, and allied health. It is particularly relevant for:
- Cardiologists and cardio-oncologists assessing cardiovascular risk in cancer patients
- Medical oncologists and radiation oncologists weighing cardiotoxic treatment decisions
- General practitioners and family physicians managing cancer survivors long-term
- Researchers developing or validating cardiovascular risk prediction models
- Trainees in cardiology, oncology, or clinical epidemiology
Cardio-oncology: a field with room to grow
Here is the exciting part. The field is still in its early chapters. Most existing cardiovascular risk prediction tools were not built for the complexity of modern cancer care. The opportunity to build better ones is wide open.
And it needs more of us. More clinicians. More researchers. More collaboration across cardiology, oncology, and primary care.
The best cardio-oncology tools will not come from statisticians alone, or oncologists alone, or cardiologists alone. They will come from teams who understand the clinical decisions, the data, and the methods all at once.
A personal note
This is work I am genuinely excited to be part of. My goal is to help build clinical risk prediction tools that are rigorous, practical, and useful at the bedside. The kind clinicians actually reach for. The kind patients actually benefit from.
If that is something you care about too, I would love to connect. Clinician, researcher, trainee, or somewhere in between. There is meaningful work to do here.
Frequently asked questions
What is cardio-oncology? Cardio-oncology is a subspecialty that focuses on the cardiovascular care of patients with cancer, across the full treatment journey from pre-treatment assessment through active therapy to long-term survivorship.
Why do cancer patients need specialised cardiovascular risk prediction? Cancer and its treatments introduce risks that standard cardiovascular risk scores were not designed to capture, including therapy-specific cardiotoxicity, competing risks from cancer mortality, and evolving risk across the treatment journey.
What is competing risk in cardio-oncology? A competing risk is an event that prevents the outcome of interest from occurring. In cardio-oncology, death from cancer is a common competing risk because patients who die from cancer cannot subsequently develop heart failure or other cardiovascular events. Standard survival methods that ignore competing risks can overestimate cardiovascular risk.
What is dynamic risk prediction? Dynamic risk prediction updates a patient’s estimated risk over time as new clinical data become available, such as imaging results, biomarkers, or cumulative treatment exposure. Landmark models and joint models are two common approaches.
📄 Read the full paper: JACC: CardioOncology 2026
#CardioOncology #Cardiology #Oncology #PrimaryCare #ClinicalResearch #RiskPrediction #CancerSurvivorship #Cardiotoxicity
📞 Referrals: For urgent or routine cardio-oncology referrals, please contact Premier Cardiology at The Wesley Hospital. More information at https://jonathansen.com/refer/
