Which Breast Cancer Patients Should Really Worry About Radiation-Induced Heart Disease—And How Much?

Abstract

The most difficult portion of my discussion with breast cancer patients about the potential adverse effects of radiation therapy (RT) is estimating their long-term chance of developing radiationinduced heart disease (RIHD). Very large studies combining tumor registry data with pre-existing regional or national hospital discharge diagnosis and death records and the Oxford metaanalyses of randomized trials found increased risks of coronary heart disease (CHD) for irradiated patients compared with nonirradiated ones, or for patients treated to the left breast or chest wall compared to those treated to the right side. However, most patients in these studies were treated with obsolete techniques that exposed much larger volumes of the heart to radiation than is done today, often with daily doses (fraction sizes) of 2.5 Gy or higher. Patients whose treatment was administered after 1975 to 1980 followed for median times of approximately 9 to 10 years had little or no increased risk of RIHD in these and additional registry and large clinical series (700 to 3,000 patients). For example, Giordano et al found that the 15-year risk of cardiac mortality for 8,652 patients treated from 1973 to 1979 was substantially higher when left-sided irradiation was administered than when right-sided irradiation was used (13.1% v 10.2%; P .02). This absolute 15-year rate decreased substantially for 6,495 patients treated from 1980 to 1984 (9.4% v 8.7%, respectively; P not significant), and there was further greater reduction for 12,136 patients treated from 1985 to 1989 (5.8% and 5.2%, respectively). Certainly, more patients in recent times have been treated to the intact breast without nodal irradiation, which eliminates cardiac irradiation entirely in most patients and, for those still exposed, reduces the irradiated volume. Improvements in treatment of CHD also no doubt played a major role in these absolute reductions in both laterality groups, a matter I will return to later. These data have shortcomings. Coding of whether patients actually received RT or had a cardiac event may be inaccurate, particularly in registry studies or when death certificates are the main source of information. For example, Vallis et al found that 16% of myocardial infarctions coded in Ontario (Canada) hospital registries were not verifiable according to standard criteria when individual medical records were reviewed. Further, RIHD can take at least 15 to 20 years to develop, and few of these more recently treated patients have been followed that long. Nonetheless, I am reasonably certain that patients treated today on average will have a very low risk of RIHD. My problem comes when trying to move beyond the average. For any one patient, the risk of RIHD seems likely to be a function of three variables: the cardiac “dose-volume histogram,” the use of potentially cardiotoxic systemic therapy, and the presence and treatment of independent risk factors for CHD (such as hypertension, diabetes, cholesterol and lipid disorders, and smoking). Unfortunately, I do not have an equation at hand with which to make this calculation reliably. Only very recently has the widespread availability of computed tomography (CT) -based treatment planning made it possible to measure accurately how much of the heart is included in or near the RT fields for individual patients. Prosnitz and Marks recently reviewed studies of radionuclide cardiac imaging which correlate (imperfectly) the development of perfusion abnormalities with increasing irradiated cardiac volume. However, it is simply too early to know whether this is a reliable surrogate marker for the long-term risk of RIHD. (Indeed, some of these defects disappear with time.) It is also not certain that this is the optimal parameter. Perhaps the critical target is really the coronary arteries, as suggested by a study of Hodgkin’s disease patients treated at Stanford University (Stanford, CA) in which increased left ventricular blocking by itself did not reduce the risk of myocardial infarction. Tracing the coronary arteries requires administering intravenous contrast, something not usually done in day-to-day clinical RT planning or in these perfusion imaging correlation studies. There is surprisingly little shortor long-term data on cardiac toxicity rates for patients treated with both anthracyclines and RT. Perhaps the most important information so far comes from Bonneterre et al, who studied cardiac function in surviving node-positive patients included in the French Adjuvant Study Group Trial 05, conducted from 1990 to 1993. Patients received six cycles of fluorouracil, cyclophosphamide, and epirubicin, the latter administered at either 50 or 100 mg/m/cycle (FEC50 or FEC100). Almost all patients had RT that included the internal mammary nodes (IMNs). With a median follow-up time of 102 months, the risk of left ventricular dysfunction in patients with right-sided cancers receiving FEC50 was 7% (two of 30), compared with none of 34 patients irradiated to the left side. For patients receiving JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 24 NUMBER 25 SEPTEMBER 1 2006