• © 2008 by American Society of Clinical Oncology

Left Ventricular Ejection Fraction and Cardiotoxicity: Is Our Ear Really to the Ground?

The cardiotoxicity of anthracyclines has been studied extensively, yet many uncertainties persist. A vast number of reports address not only the clinical features of anthracycline-related cardiac toxicity, but also mechanistic considerations, ultrastructural changes, risk factors, prevention, treatment options, and, more recently, interactions with other potentially cardiotoxic agents. Despite extensive information, the ability to make firm conclusions about the overall prognosis and management of patients who have been exposed to anthracyclines is problematic. There are several reasons for the ongoing interest in treatment-related cardiotoxicity: (1) anthracyclines, even after three decades, continue to play a prominent role in the treatment of a wide variety of both hematologic and solid tumors; (2) cardiotoxicity can be a devastating and sometimes fatal event; (3) given that the purported mechanisms for oncologic effect and cardiotoxicity are different in the case of anthracyclines but conceptually linked in the case of trastuzumab, it remains intriguing to attempt to maintain oncologic benefits while mitigating the potential devastation of cardiotoxicity; and (4) the effect of anthracycline on the heart is a cumulative dose-related phenomenon, making this entity more predictable than some other forms of cancer treatment-related cardiotoxicity.

It is now well established that anthracycline cardiotoxicity is a cumulative dose-related effect, suggesting that each administration constitutes additive or sequential damage. Once damage has progressed beyond a threshold level, myocyte injury is probably irreversible and additional exposure constitutes added stress and further damage. The heart, however, has substantial reserves, making it virtually impossible to appreciate subclinical damage with the usual methods employed to evaluate these patients. These are all facts regarding anthracyclines that have been appreciated for decades. So why, then, does this subject remain so much in the forefront? Why are there still enigmas and confusion regarding anthracycline cardiotoxicity?

The answers to these questions, at least in part, are related to the broad group of newer agents often administered in combination with anthracyclines that have resulted in novel interactions that are now being investigated and only partially understood. Many recent examples of cancer therapy are highly influenced by cardiotoxicity, and certainly cardiac concerns may dictate future choices. The enigmas and confusion surrounding anthracyclines are, to a large extent, related to inherent weaknesses in how we define and measure cardiotoxicity. Both in the management of individual patients and in virtually all of the clinical trials that have considered cardiotoxicity, the predominant parameter for cardiac dysfunction has been an abnormality or a serial decrease in left ventricular (LV) contractility. This is determined principally by estimating the chamber dimensions or volume and calculating the left ventricular ejection fraction (LVEF). The measurements are obtained by either cardiac ultrasound or nuclear imaging techniques. The derived value is compared with institutional norms, or in the case of follow-up studies, with the patient's previous, pretreatment, or serial determinations. Decreases in LVEF usually are deemed to be related to the offending agent and, equally problematic, unchanged values are often equated to a lack of cardiotoxicity. Neither of these assumptions regarding LVEF is necessarily valid. It is not adequately appreciated that the normal heart has tremendous recruitable contractile ability, and that for the LV to exhibit a decrease in ejection fraction (EF), the myocardium must have undergone sufficient damage so as to exceed its ability to compensate. Thus, a decreased EF after treatment is a marker for advanced myocyte damage; it is an imperfect marker that is influenced by nonanthracycline cardiac stressors and volume status on one hand, and significant interpretative variation on the other.

There is convincing evidence that myocyte damage occurs with anthracyclines at much lower cumulative dosages than can be appreciated by declines in the LVEF values. The most compelling is that cardiac biopsy changes occur at cumulative dosages far below those usually associated with EF decreases.¹ Anthracycline damage may well start at the time of initial exposure in susceptible patients, and each subsequent dose represents an incremental stress.^2,3 Considerable efforts have been expended in identifying risk factors for anthracycline toxicity and in seeking ways to protect the myocardium from anthracycline damage. These toxicity risk factors can be thought of as stressors that impair myocardial reserve, or additive factors that augment the effects of prior damage, or both. These include advanced age and prior cardiac disease, such as hypertension, on which initial LV myocardial damage builds. As a result, the threshold for clinically recognizable cardiac damage is met at a lower anthracycline cumulative dose. There are some individuals whose risks for toxicity have not yet been recognized, and there is a strong probability that genetic factors may predispose some to exhibit a higher degree of sensitivity to the actual cardiotoxic insult than others. All of this leads us to appreciate that anthracycline cardiac toxicity, distinct mechanistically from some newer agents associated with cardiac dysfunction, gradually erodes the extensive but finite reserves of the myocardium to compensate.⁴ Ultimately a threshold is reached that is detectable. Unfortunately, the problem does not end when anthracyclines are discontinued; other myocardial stressors may be encountered, including any typical cardiac risk factor that promotes LV dysfunction. In the cancer patient, sepsis or other complications of cancer therapy may contribute. Overt evidence of this damage may not become apparent for years or decades after anthracycline exposure.^5,6

Another major consideration leading to confusion regarding both anthracycline-related and newer forms of cardiac toxicity is how cardiac damage is defined. The substantial limitations of using only changes in LVEF are compromised further by our knowledge that approximately half of all heart failure diagnosed in the United States occurs in patients who maintain a normal LVEF; their overall cardiac outcomes are similar to those who exhibit a low LVEF.⁷ As a result, biomarkers such as B-type natriuretic peptide and troponins (I and T) are increasingly being used to stratify patients into higher and lower risk categories. This process is well established in the cardiology literature and recently has been reported in oncology patients. In fact, an elevated troponin during chemotherapy seems to correlate with increased risk for the development of cardiac toxicity.⁸ In addition, an elevated B-type natriuretic peptide level before any cycle of anthracycline chemotherapy also indicates a high risk of cardiac toxicity; LVEF changes, when looked at by themselves, do not.⁹ On the basis of this and other data, it seems paramount that biomarkers become incorporated into the subsequent definitions for cardiac toxicity, especially when anthracyclines have been used. Biomarker monitoring of patients during anthracycline exposure may provide crucial evidence of early damage that has not heretofore been appreciated.

Two intriguing articles that add to our knowledge of the effects of cancer treatment on the heart appear in this issue of the Journal of Clinical Oncology. The first by Ganz et al¹⁰ looks at a group of patients treated with anthracyclines as part of the Southwest Oncology Group protocol S8897. There are two conclusions regarding cardiotoxicity in this study that at first glance seem to be in conflict. The authors report that in the doxorubicin-treated group of patients, the post-treatment LVEF measurements were 3.4 percentage points lower than in patients treated on the nondoxorubicin arm; this is very much in keeping with what would be expected (ie, some patients reached the threshold of myocyte damage that exceeded their ability to compensate). The second conclusion, namely that doxorubicin did not increase the likelihood of late adverse cardiac effects, although not proven, is especially intriguing.

As with any study that looks at a subgroup of patients who can be identified, contacted, and who are willing to participate in a follow-up subtrial, many data points are missing. In the initial study it was noted that there were 11 severe cardiac events in cyclophosphamide, doxorubicin, and fluorouracil arm, one of which was fatal, compared with four severe cardiac events in the cyclophosphamide, methotrexate, and fluorouracil arm. There were also a greater number of early deaths in the doxorubicin arm. These previously affected patients are not included in the later analysis. Those survivors who were not initially affected are less likely to have had impaired cardiac reserve and to demonstrate late cardiac toxicity. Here again, this demonstrates our inherent inability to see subtle changes in cardiac reserve when examining only LVEF measurements. Other factors such as possible sampling artifacts that are fully addressed in this important article may also have led to this conclusion. Ultimately, a combination of the nonprospective nature of the data, the possible elimination of the sickest patients, and our inability to establish a true picture of real cardiac damage by estimating LVEF from nuclear scans may all have contributed to a possible underappreciation of the true late effects of anthracyclines on the heart.

The second article, by Perez and colleagues,¹¹ describes the cardiac events in the North Central Cancer Treatment Group N9831 adjuvant trastuzumab breast cancer trial. It expands the information regarding cardiac effects that has been reported from the sister trials, namely the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-31 trial and the Herceptin Adjuvant (HERA) trial.^12,13 Although these trials varied as to inclusion criteria and pretrastuzumab therapy, they nevertheless draw our attention to some important unifying characteristics regarding cardiotoxicity; interesting and potentially important ideas emerge that can be expected to further our understanding. First, the NSABP B-31 trial and the HERA trial, and now the North Central Cancer Treatment Group trial, report reversibility of many of the cardiac events, strengthening the concept that some aspects of trastuzumab cardiotoxicity may be related to myocyte metabolic dysfunction rather than the more permanent cellular death typical of anthracyclines.¹⁴ Second, these trials clearly document that pretreatment with anthracyclines is a major factor leading to cardiotoxicity.

Another concept not yet adequately considered involves the timing of the various treatment components. The time interval between the anthracycline and the sequential insult of trastuzumab may be a crucial and underappreciated factor. Documented cardiotoxicity with the addition of trastuzumab was greatest in the pivotal trial where the drugs were administered concomitantly, and occurred much less frequently in the NSABP B-31 and in the North Central Cancer Treatment Group trial, and by far the least in the HERA trial, in which the median interval between anthracycline and trastuzumab was 89 days.¹³ The data reported by Perez et al¹¹ strengthens this hypothesis, but clearly does not establish a cause-and-effect relationship. However, in support of a time-interval association is the observation of increased binding of trastuzumab after anthracycline exposure and presumed injury.¹⁵ Although it will require additional study and confirmation, timing, in addition to the formulation of the anthracycline, infusion duration, and total cumulative anthracycline dose, may all be essential modifiable features that could affect cardiac toxicity associated with the sequential use of an anthracycline with trastuzumab.¹⁶

Although these articles provide important information, part of their strength lies in the integration of these new data into what others have shown before: (1) anthracyclines remain troublesome and late findings may be subtle; (2) early anthracycline damage leaves its myocardial cellular damage mark, only to be brought to the forefront when compensatory mechanisms are exhausted or later insults are added to the baseline injury; (3) the mechanism of trastuzumab damage is different and largely reversible; (4) timing of sequential cardiotoxic agents may be of greater importance than has heretofore been appreciated; and (5) cardiotoxicity should not be totally discounted even in the face of normal LVEF.

Neither of these studies fully addresses the vitally important issue of treatment for cardiac dysfunction during chemotherapy. It remains an unanswered question whether prophylactic therapy with typical heart failure medications may further mitigate or even eliminate cardiac toxicity during chemotherapy.^17,18 There is some evidence to suggest this may be an effective strategy, and that cardiac biomarkers seem necessary to detect cardiac toxicity at the earliest stages.¹⁹ Only if we collectively use our best tools for detection (biomarkers as well as others) and “keep our ears to the ground” will we hear the horses of cardiac toxicity galloping toward us before being trampled.