Clinical features of anthracycline‐induced cardiotoxicity in patients with malignant lymphoma who received a CHOP regimen with or without rituximab: A single‐center, retrospective observational study

Abstract We investigated the incidence of cardiotoxicity, its risk factors, and the clinical course of cardiac function in patients with malignant lymphoma (ML) who received a cyclophosphamide, doxorubicin, vincristine, and prednisolone (CHOP) regimen. Among all ML patients who received a CHOP regimen with or without rituximab from January 2008 to December 2017 in Nagoya City University hospital, 229 patients who underwent both baseline and follow‐up echocardiography and had baseline left ventricular ejection fraction (LVEF) ≥50% were analyzed, retrospectively. Cardiotoxicity was defined as a ≥10% decline in LVEF and LVEF < 50%; recovery from cardiotoxicity was defined as a ≥5% increase in LVEF and LVEF ≥50%. Re‐cardiotoxicity was defined as meeting the criteria of cardiotoxicity again. With a median follow‐up of 1132 days, cardiotoxicity, symptomatic heart failure, and cardiovascular death were observed in 48 (21%), 30 (13%), and 5 (2%) patients, respectively. Multivariate analysis demonstrated that history of ischemic heart disease (hazard ratio (HR), 3.15; 95% CI, 1.17‐8.47, P = .023) and decreased baseline LVEF (HR per 10% increase, 2.55; 95% CI, 1.49‐4.06; P < .001) were independent risk factors for cardiotoxicity. Recovery from cardiotoxicity and re‐cardiotoxicity were observed in 21 of 48, and six of 21, respectively. Cardiac condition before chemotherapy seemed to be most relevant for developing cardiotoxicity. Furthermore, Continuous management must be required in patients with cardiotoxicity, even after LVEF recovery.


INTRODUCTION
Anthracycline is recognized as an effective chemotherapeutic agent for several kinds of tumor. Therefore, its cardiotoxicity, which can disturb chemotherapy, is currently the subject of crucial discussion [1][2][3][4][5][6][7][8][9][10][11][12]. On the basis of convincing evidence proving anthracycline's benefits for malignant lymphoma (ML) and other cancers [7,13,14], many clinicians continue to use this agent, so an adequate understanding of this adverse effect and a plan to counter it are needed. After several risk factors for cardiotoxicity such as high cumulative dose of anthracycline and general cardiovascular risk factors were confirmed [3,12], clinicians have tended to avoid using over 400 mg/m 2 , and recently over 250 mg/m 2 , of anthracycline [15], and have begun total cardiovascular care for patients who receive anthracycline-containing chemotherapy. Furthermore, strategies to prevent cardiotoxicity have been studied frequently and have shown convincingly that angiotensin-converting enzyme inhibitors, beta blockers, mineralocorticoid receptor antagonists, and statins are relevant to the prevention of cardiotoxicity [16][17][18][19]. According to a recent study, anthracyclineinduced cardiotoxicity appears within 1 year after chemotherapy starts, and half or more of these cardiotoxicities are reversible to the normal range of left ventricular ejection fraction (LVEF) [20]. These findings are very important, as the number of cancer survivors is growing. However, the detailed clinical features and risk factors related to anthracycline-induced cardiotoxicity for each disease or chemotherapy regimen have not been clarified. Particularly, there is only limited evidence regarding anthracycline-induced cardiotoxicity in patients with ML who receive cyclophosphamide, doxorubicin, vincristine, and prednisolone (CHOP) as a uniform chemotherapy regimen. The aim of the current study was to clarify the risk factors, clinical features, and prognosis of anthracycline-induced cardiotoxicity in patients with ML who received a CHOP regimen with or without anti-CD20 antibody, rituximab.

Study population and design
All 443 patients with ML who received a CHOP regimen with or without rituximab from January 2008 to December 2017 in Nagoya City University hospital were retrospectively reviewed. Of these, 229 patients with LVEF ≥ 50% at baseline who underwent baseline and at least one or more follow-up echocardiography studies were enrolled ( Figure 1). Cardiotoxicity was defined as a 10% or greater decrease in LVEF and absolute LVEF < 50% by follow-up echocardiography after chemotherapy [20]. Partial recovery was defined as a 5% or more increase in LVEF and LVEF > 50% after cardiotoxicity occurred, and full recovery was defined as improving to 95% or more of baseline LVEF. Recardiotoxicity was defined as meeting the criteria of cardiotoxicity again once partial recovery or full recovery was achieved.

Echocardiography
LVEF was measured using the disk summation method, using a Cardiotoxicity (-) n = 181 F I G U R E 1 Study flow diagram including the steps from the screened patients to the finally analyzed patients, and the incidence of cardiotoxicity and other cardiac events. Two hundred twenty-nine patients who underwent baseline and at least one or more follow-up echocardiography studies were available for analysis, and those with baseline LVEF ≥50% were eligible for the current study. With a median follow-up period of 1132 days, forty-eight patients (22%) had cardiotoxicity. Of them, 30 patients (13%) experienced symptomatic heart failure and five patients (2%) died from cardiovascular causes to the guidelines of the American Heart Association [22]. To avoid a misdiagnosis of cardiotoxicity and recovery from diagnosis, we remeasured LVEF that was between 40% and 60% by the first measurer, by the second measurer (the other of the first measurer) to assess the accuracy of diagnosis.

Statistical analysis
Continuous variables were expressed as median (interquartile range, IQR

Risk factors for cardiotoxicity
Univariate Cox proportional hazards analysis showed that older age, history of ischemic heart disease, decreased estimated glomerular filtration ratio, higher log-BNP, and lower LVEF were significantly associated with the risk of cardiotoxicity (    Table 3. Two of nine patients (22%) who achieved full recovery and 4 of 12 patients (33%) who achieved partial recovery developed re-cardiotoxicity.

Incidence of recovery from cardiotoxicity and prognosis after recovery
One patient with full recovery had a sudden death, and there were 4 deaths due to heart failure in patients with no recovery (Table 3;   Table S1).

Impact of cardiotoxicity on survival
All death was observed in 78 patients during the observational period. Forty-eight patients (62%) died due to lymphoma progression. The remaining 30 patients died due to the following causes; chemotherapy-related (n = 9, 12%), non-lymphoma cancer (n = 10, 13%), and the other reasons (n = 11, 14%). All-cause mortality in patients with cardiotoxicity tended to be higher than in patients without cardiotoxicity, but the difference was not significant (log-rank P = .15; Figure S1). There was no significant difference in overall mortality between patients who developed symptomatic heart failure and those who did not. Individual details of patients who developed cardiotoxicity and died from cardiovascular causes are summarized in Table S1. Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin 2 receptor blocker; BB, beta blocker; MRA, mineralocorticoid receptor antagonist. Full recovery was defined as improving to 95% or more of baseline LVEF, and partial recovery was defined as a ≥5% increase in LVEF and LVEF > 50% after development of cardiotoxicity. Re-cardiotoxicity was defined as meeting the criteria of cardiotoxicity again after partial recovery or full recovery was achieved. The percentiles are for each recovery pattern.

DISCUSSION
We chiefly demonstrated that 21% (48/229) of patients with ML who received a first-line CHOP regimen with or without rituximab developed cardiotoxicity; 13% had symptomatic heart failure, and 2% had cardiovascular death, with a median follow-up of 3.1 years. This report is the first to assess cardiotoxicity, recovery from cardiotoxicity, and recurrence of cardiotoxicity with a focus on patients with ML who received a CHOP regimen with or without rituximab. In previous studies, the incidence of cardiotoxicity has been reported to be 5.4% to 27% in patients with variable malignant disease who received anthracycline-containing chemotherapy [17,20,23,24] (Table S2).
Although cardiotoxicity is seen frequently after use of over 400 mg/m 2 of doxorubicin and the incidence of cardiotoxicity was dose dependent in previous studies, a recent prospective study showed that even 250 mg/m 2 of doxorubicin is a risk factor and that there may be no safe dosage without risk of doxorubicin-induced cardiotoxicity [12,15]. Here, we observed no significant relationship between car- In our study, baseline risk factors of cardiotoxicity were older age, history of ischemic heart disease, renal dysfunction, high BNP level, and lower LVEF within normal limit in the univariate Cox proportional hazards analysis. The analysis showed no significant correlation between general cardiovascular risk factors and cardiotoxicity, partly because our relatively recent participants might have already received adequate treatments for general cardiovascular risk factors. History of ischemic heart disease is a well-known risk factor for anthracyclineinduced cardiotoxicity in the previous study [25,26], thus our result supported and strengthened this knowledge. Although the main metabolic pathway of anthracycline is liver, renal dysfunction possibly causes elevated blood concentration of anthracycline, and this may result in cardiotoxicity. Renal dysfunction also results in decreased erythropoietin, which is reported to have a cardioprotective role [27,28]. In a recent large-scale prospective study that included some types of chemotherapy regimens for other cancers, lower LVEF at the end of chemotherapy, older age, female sex, family history of coronary artery disease, and higher cumulative anthracycline dose were identified as independent risk factors by multivariate Cox proportional hazard regression with forward stepwise selection [20]. A previous prospective study with a limited number of patients with ML indicated that older age, male sex, radiotherapy, higher cumulative dose of anthracycline, and overweight were the independent risk factors of cardiotoxicity by multivariate logistic regression analyses [24], however, in which the strict data of baseline LVEF was not available.
Change of contractility of LV can be influenced by lots of clinical factors, however only history of ischemic heart disease and LVEF at baseline were the independent risk factors for cardiotoxicity assessed with many factors for which analysis were desirable in our study, and we speculate that cardiac condition before chemotherapy is the most relevant influence on cardiotoxicity in our cohort, even in patients whose LVEF is within normal limits.
In that large-scale prospective study, 98% of anthracycline-induced cardiotoxicity was observed within 1 year after the start of chemotherapy in a heterogeneous patient population with variable primary diseases, in particular half or more of the patients had breast cancer, and chemotherapy regimens [20]. In our study, 67% of cardiotoxicity was observed within 1 year from the start of chemotherapy, with the remaining 33% observed after 1 year. Possible reasons for the difference in our results include the following: our participants were limited to patients with ML who received a CHOP regimen with or without rituximab, basic characteristics such as baseline age and renal function were different, and it might be associated with the varied timing of follow-up echocardiography in some patients.
We also analyzed changes in cardiac function over time after the development of cardiotoxicity. Partial recovery or full recovery were seen in 43% of patients with cardiotoxicity. We gathered serial data on cardiac contractions after recovery from cardiotoxicity. Two of nine patients (22%) with full recovery experienced a recurrence of cardiotoxicity, although four of 12 patients (33%) with partial recovery developed recurrent cardiotoxicity, even though some had not only discontinued chemotherapy but also received treatments for cardiotoxicity. Anthracycline-induced cardiomyopathy is recognized as type 1 myocardial damage that is considered cumulative and permanent [2,29,30]. Even if medical therapy for heart failure increases left ventricular contraction, myocardial tissue damage persists and we need watchful follow-up of patients with full or partial recovery. It is possible that the patients who did not experience a recurrence of cardiotoxicity had not only irreversible myocardial damage due to anthracycline, but also reversible changes such as those due to acute inflammation.
Previous research has described early detection and prompt administration of anti-remodeling therapy as crucially important to improving the cardiac systolic function of patients with anthracycline-induced cardiotoxicity [31][32][33]. Therefore, we must start therapy for cardiotoxicity as early as possible to limit irreversible myocardial injury to a minimum.
Most cardiovascular deaths in this study were observed in patients who never achieved any degree of recovery after cardiotoxicity. This finding indicates the necessity of maximizing efforts to improve LVEF, once LVEF decline is observed. Furthermore, it is very important to diagnose the cardiotoxicity at an early stage in high-risk patients.
Some biomarkers [34][35][36] and decreased global longitudinal strain [37,38] have been reported to be useful in the early diagnosis of cardiotoxicity. Furthermore, early diagnosis and therapy for cardiotoxicity have been reported to be associated with better outcome [31][32][33]. Importantly, primary preventive use of a beta blocker was examined in a recent study [39]. With appropriate consideration of cost-benefit ratios and risks of administration, further development of preventive strategies is expected, especially for patients at high risk for cardiotoxicity.
Our study has some limitations. First, it was a single-center, retrospective study and the number of enrolled patients was limited. Second, patients who did not undergo baseline or follow-up echocardiography were excluded, and the timing of follow-up echocardiography was relatively varied depending on the attending physician. The incidence of cardiotoxicity might be overestimated. Third, our study had no data on global longitudinal strain, which is considered to be an early indicator of cardiotoxicity.

CONCLUSION
In patients with ML who received a CHOP regimen with or without rituximab, 21% developed cardiotoxicity, 13% had symptomatic heart failure, and 2% died from cardiovascular causes. Our multivariate analysis showed that only baseline LVEF and history of ischemic heart disease were the independent risk factors for cardiotoxicity, indicating cardiac condition before chemotherapy is the most relevant for developing cardiotoxicity. Forty-four percent of patients who developed cardiotoxicity achieved LVEF recovery; however, 29% of patients with recovery from cardiotoxicity developed LVEF decline again.
Watchful follow-up and continuous management for LV dysfunction must be required in patients with cardiotoxicity, even after LVEF recovery.

ACKNOWLEDGMENTS
We would like to thank the patients who participated in this study and their families, as well as the research nurses, study coordinators, and operations staff. We also thank Dr. Hiroshi Inagaki and Dr.
Ayako Masaki (Nagoya City University Graduate School of Medical Sciences, Nagoya) for histologically diagnosing malignant lymphoma in this study.