Improving Prediction of Cardiovascular Disease After Childhood Cancer

Reading time: 4 minutes

Clare Meernik

84% of children who are diagnosed with cancer today can expect to live for at least 5 years after their diagnosis. Though the prognosis is favorable for the majority of children–and much improved from a 5-year survival rate of 58% in the mid-1970s–the long-term adverse effects of some cancer treatments increase the risk for other health problems into adulthood. 

The risk of premature cardiovascular disease (CVD)–diseases like heart failure, coronary artery disease, and stroke–is particularly relevant for adult survivors of childhood cancer. Thirty years after treatment with anthracycline-based chemotherapy and chest radiation, 1 in 8 will have developed severe heart disease, contributing to a cardiac-related death rate that is 7 times as high as the general population. This makes CVD the leading cause of death among childhood cancer survivors. 

In the general U.S. adult population, heart disease is the leading cause of death. Public health campaigns such as American Heart Month seek to raise awareness of key risk factors for CVD that can often be controlled through lifestyle behaviors (for example, high blood pressure, high cholesterol, and smoking). User-friendly online CVD risk calculators have even been developed and are available from leading medical authorities, including the American College of Cardiology and the Mayo Clinic

However, these calculators do not account for the unique cardiotoxic treatments that childhood cancer survivors may have been exposed to and may underestimate this population’s true CVD risk. In an attempt to remedy this, Chen and colleagues developed risk prediction models that considered exposure to specific cancer treatments, along with the more traditional CVD risk factors of age, sex, hypertension, dyslipidemia, and diabetes. 

The study used data from participants of the Childhood Cancer Survivor Study–a cohort of nearly 25,000 individuals diagnosed with cancer between 1970 and 1999 as a child (<21 years of age) in the U.S. or Canada, and who survived at least five years after diagnosis. To assess how well their models predicted heart failure, ischemic heart disease, and stroke, this cohort was randomly split into a “discovery sample” where the prediction models were first developed, and a “replication sample” where the authors tested the prediction models in a new sample.

The ability of each model to distinguish individuals who experience CVD by age 50 from individuals who do not–also referred to as “model discrimination”–were evaluated using a statistical measure called ROC-AUC (Receiver Operating Characteristic – Area Under the Curve). A ROC curve is a plot of sensitivity vs. (1 – specificity), meaning, in short, that a model with an AUC value near 1 can correctly identify individuals with the outcome as diseased (a high true positive rate) and also correctly identify individuals without the outcome as non-diseased (a low false positive rate). An AUC of 0.5 means that the model is no better than a coin flip at distinguishing between diseased and non-diseased individuals, while an AUC of 0.7 or higher is considered reasonably accurate.

In both the discovery sample and the replication sample, Chen and colleagues observed AUCs that were 0.7 or higher in the majority of their prediction models for heart failure, ischemic heart disease, and stroke, with influential predictors of outcomes being anthracycline chemotherapy, chest and cranial radiation, diabetes, and hypertension. Though the models had important limitations (for instance, they did not account for lifestyle factors like obesity, smoking, and physical activity due to lack of data availability, and the cohort had limited racial and ethnic diversity), the moderate predictive ability observed emphasizes the importance of incorporating both traditional CVD risk factors and cardiotoxic cancer treatment exposures into strategies for surveillance and management of CVD in adult survivors of childhood cancer. 

To enhance the clinical utility of these models, an online cardiovascular risk calculator for childhood cancer survivors is available, providing estimated probabilities of heart failure, ischemic heart disease, and stroke by age 50, as well as relative risks compared to adults without a history of cancer. Future studies should continue to build upon and use these models with the goal of improving CVD risk prediction and management and lessening the burden of CVD after childhood cancer.

Edited by Jessica Desamero

Works Discussed

Chen Y, Chow EJ, Oeffinger KC, Border WL, Leisenring WM, Meacham LR, Mulrooney DA, Sklar CA, Stovall M, Robison LL, Armstrong GT. Traditional cardiovascular risk factors and individual prediction of cardiovascular events in childhood cancer survivors. JNCI: Journal of the National Cancer Institute. 2020 Mar 1;112(3):256-65.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create a website or blog at

Up ↑

%d bloggers like this: