This article is more than 5 years old. Information may no longer be current.
Researchers improve model to predict cardiac events after radiotherapy for breast cancer
A dose–effect relationship between radiotherapy dose and acute coronary events persisted for 9 years following adjuvant radiotherapy in women with breast cancer, according to a study published in Journal of Clinical Oncology.
Researchers confirmed the association between mean heart dose and acute coronary events, but they found that the volume of the left ventricle receiving 5 Gy (LV-V5) was the most important prognostic dose–volume parameter.
Adjuvant radiotherapy after surgery can be effective in the treatment of breast cancer. However, incidental radiation exposure of the heart increases the risk for radiation-induced cardiac toxicity, which, in turn, could impact health-related quality of life, Anne P.G. Crijns, MD, PhD, from the department of radiation oncology at University Medical Center Groningen in the Netherlands, and colleagues wrote.
A previous study by Darby and colleagues demonstrated a relative 16% increase in the rate of major acute coronary events in the first 9 years after radiation exposure per Gy of mean heart dose.
However, limitations of the study included its design, use of outdated radiotherapy technologies and use of reconstructed mean heart doses derived from two-dimensional data.
Crijns and colleagues aimed to validate the findings and to determine whether dose–distribution parameters, other than mean heart dose, could better predict acute coronary events.
Cumulative incidence of acute coronary events within 9 years of follow-up served as the primary outcome.
The researchers collected mean heart dose and various other dose–distribution parameters of the cardiac substructures from 3-D CT planning scans of 910 consecutive patients (median age, 59 years) with breast cancer. Patients received a median heart dose of 2.37 Gy.
Median follow-up was 7.6 years (range, 0.1-10.1).
Thirty patients developed an acute coronary event during follow-up, including 10 patients who died as a result of ischemic heart disease.
Cumulative incidence of acute coronary events was 1.9% (95% CI, 0.9-2.9) at 5 years and 3.9% (95% CI, 2.3-5.5) at 9 years.
The cumulative incidence of acute coronary event increased by 16.5% (95% CI, 0.6-35; P = .042) per Gy.
The researchers used the Hosmer–Lemeshow test for model performance and c-statistic for discrimination to validate the model of Darby and colleagues.
The Hosmer–Lemeshow test demonstrated no significant difference between expected and observed rates of acute coronary events, which indicated good calibration, the researchers wrote.
Model discrimination showed good results, based on a c-statistic of 0.79 (95% CI, 0.71-0.87).
To identify the most relevant dose–distribution parameters, researchers compared the mean dose parameters of the patients who experienced and acute coronary event with those who did not.
Results showed that LV-V5 was the most important prognostic dose–volume parameter (HR = 1.02; 95% CI, 1-1.03).
Replacement of mean heart dose with LV-V5 in the normal tissue complication probability model resulted in an improvement of the c-statistic of the to 0.8 (95% CI, 0.72-0.88).
The most optimal multivariable normal tissue complication probability model for acute coronary events consisted of LV-V5, age, and weighted acute coronary event risk score per patient (c-statistic = 0.83; 95% CI, 0.75-0.91), which was significantly better than the mean heart dose model (P = .042).
“Radiation dose to the heart is an important risk factor for acute coronary events in breast cancer survivors,” Crijns and colleagues wrote. “Model performance was significantly improved by replacing mean heart dose with LV-V5 and using the weighted acute coronary event risk score, but this optimized model requires further external validation in an independent data set.”
However, the limitations of this study — which include the small number of events, limited follow-up and the lack of power to determine whether there is a dose–volume threshold beneath which there is no increased cardiac risk from radiotherapy exposure — may be substantial, Abram Recht, MD, from the Beth Israel Deaconess Medical Center and Harvard Medical School, wrote in an accompanying editorial.
Recht noted other possible predictors should be validated by other groups.
“Computer programs can likely be created to perform calculations of an individual’s excess risk of cardiac events or death in relation to dose–volume and clinical parameters, although I suspect it will be some years before they provide sufficiently validated, narrow estimates of risk to be clinically useful,” he added. “Perhaps by then we will also have better ways to explain such sobering decisions to patients.” – by Kristie L. Kahl
Disclosure: Crijns reports no relevant financial disclosures. Other researchers report research funding from AbbVie, IBA, Mirada, Philips Healthcare, Siemens, RaySearch Laboratories and Roche, as well as a consultant/advisory or speakers bureau role with IBA. Recht reports consultant roles with CareCore and US Oncology and research funding from Genomic Health