Is screening for abnormal ECG patterns justified in long‐term follow‐up of childhood cancer survivors treated with anthracyclines?

ECG and echocardiography are noninvasive screening tools to detect subclinical cardiotoxicity in childhood cancer survivors (CCSs). Our aims were as follows: (1) assess the prevalence of abnormal ECG patterns, (2) determine the agreement between abnormal ECG patterns and echocardiographic abnormalities; and (3) determine whether ECG screening for subclinical cardiotoxicity in CCSs is justified.


INTRODUCTION
Survival of children with cancer has improved over the last decades to almost 80%, 1-3 resulting in a growing number of childhood cancer survivors (CCSs). These survivors are at a risk of (life-threatening) late effects of their cancer treatment, including secondary malignant OR, odds ratio; SCD, sudden cardiac death; TBI, total body irradiation; TdP, torsades de pointes; VCFc, rate corrected velocity of circumferential fiber shortening neoplasms, metabolic syndrome, hypertension, and cardiovascular diseases. 4,5 Several studies reported a significant increase in mortality risk due to cardiac causes compared with the general population. [6][7][8] CCSs are more likely than their siblings to experience congestive heart failure, myocardial infarction, pericardial disease, and valvular abnormalities. The cumulative incidence of these adverse cardiac outcomes continues to increase up to 30 years after diagnosis. 9 Anthracyclines and radiotherapy are the main causes of cardiotoxicity in CCSs.
Anthracycline-induced clinical heart failure is reported in 1-5% of survivors and subclinical heart failure in up to 65%. [10][11][12] While echocardiography is the most common test for screening and monitoring of late-onset (subclinical) cardiotoxicity, there is an ongoing discussion about the additional value of electrocardiography (ECG). In the recently published international recommendations for cardiomyopathy surveillance, screening for conduction abnormalities (where ECG is essential) was mentioned as a topic that will be addressed by future evidence-based international guideline development. 13 In the Dutch and American guidelines, evidence of ECG abnormalities during long-term follow-up is limited to conduction disorders. 14,15 Previous literature reported a higher incidence of prolonged QT dispersion intervals or prolonged QTc. 16,17 Conduction abnormalities and specifically QT prolongation are a major risk factor for sudden cardiac events and death.
Supportive data show that ECG screening identifies healthy young people (age <35 years) and athletes who are harboring potentially serious cardiac diseases. 18 In Italy, screening young athletes with an ECG has been suggested to be effective in decreasing the annual rate of sudden cardiac death (SCD). 19 There is an ongoing debate on the feasibility and interpretation of large-scale ECG screening in young adults, 18,20 since the majority of SCDs in the young affect the general population and not only athletes. 21 The American Heart Association (AHA) recently published a scientific statement on the topic of screening with ECG for detection of cardiovascular disease in the general young population (12-25 years of age). Their recommendation is to consider screening with ECG in selected groups (in order to find cardiovascular abnormalities) combined with comprehensive history taking and physical examination. 22 Long-term CCSs treated with anthracyclines could be classified as such a selected group.
In this study, we aim to assess the prevalence of abnormal ECG patterns and determine the agreement between abnormal ECG patterns and echocardiographic abnormalities in asymptomatic longterm CCSs. In addition, we investigate whether ECG screening for subclinical cardiotoxicity in this population is justified.

Study population
Participants were CCSs enrolled in the Late Effects Clinic of the Radboud University Medical Center in the Netherlands, from May 2006 through May 2010 with the aim to provide comprehensive long-term follow-up care. Eligibility for inclusion in the current analysis was diagnosis of pediatric cancer at age 18 years or less and evaluation 5 years or more after anthracycline treatment. Survivors were evaluated according to the national guidelines for follow-up of CCSs ranging from once every 2-5 years depending on cumulative anthracycline dosage and/or mediastinal irradiation. 14 Exclusion criteria were abnormal ECG and/or echocardiogram at the start of therapy, incomplete data (ECG and/or echocardiographic) at follow-up, clinical heart failure, cardiac medication, and congenital heart disease. Clinical parameters such as age (at diagnosis and at follow-up), body mass index (BMI), diagnosis, previous cancer treatment, follow-up duration/years after anthracycline treatment, and irradiation to the heart region (including mediastinal, thoracic, spinal, left or whole upper abdominal, or total body irradiation [TBI]) were obtained by medical record review. We calculated the cumulative anthracycline dose as the sum of the doxorubicin and daunorubicin doses. Survivors underwent a complete physical examination, a clinical history was obtained, and a 12-lead ECG and echocardiography were performed. Clinical heart failure was defined by the New York Heart Association classification (NYHA). 23

ECG and echocardiography
A resting 12-lead ECG (Philips PageWriter Touch) was performed and evaluated by two independent cardiologists (LK and LB), who were unaware of the medical history. Disagreement between the cardiologists was resolved by consensus reached by subsequent joined evaluation of the electrocardiographic findings. The following ECG parameters were noted: heart rate, PQ-, QRS-, and QT-intervals, voltage parameters (total QRS voltage in limb leads), repolarization changes, arrhythmias, and other abnormalities. QTc interval was calculated using Bazett formula. 24 A QTc interval of more than 450 msec was defined as prolonged according to the Common Terminology Criteria for Adverse Events (CTCAE) of the National Cancer Institute (NCI). 25 Abnormal electrocardiograms were coded according to the Minnesota Code, see Supplementary Table S1. 26 If more than one abnormality according to the Minnesota Code was applicable to the ECG, the major abnormality code was used. Abnormal ECG patterns that could not be defined by the Minnesota Code were coded as "other." These other abnormal ECG patterns were coded independently by two experienced cardiologists.
The prevalence of abnormal ECG patterns in our group of longterm survivors was compared with the prevalence of abnormal ECG patterns in the general population, derived from the literature. 18,20 Transthoracic echocardiography at rest was performed by experienced An M-mode echocardiogram was performed in the parasternal long-and short-axis views to measure the left ventricle internal dimensions at end-systole and end-diastole (LVIDs, LVIDd) and the posterior and septal wall thickness at end-diastole (LVPWd, IVSd, respectively).

Measurements of left ventricular dimensions and left ventricular mass
(LVM) were indexed by body surface area (BSA). The left ventricle systolic function was indicated using fractional shortening (FS), ejection fraction (EF), rate-corrected velocity of circumferential fiber shortening (VCFc), and end-systolic wall stress (ESWS). FS (%) was obtained using the following formula: ([LVIDd -LVIDs] / LVIDd) x 100. EF was calculated with the modified Simpson's rule, while ESWS was calculated with the modified formula of Rowland. 28 The VCFc was assessed using the formula by Colan et al. 29 Left ventricular diastolic function was estimated using the ratio of peak early (E) to late atrial (A) mitral Doppler flow velocities (E/A ratio). 30 LVM was calculated with the formula of Devereux and Reichek. 31 Left ventricular diameter was measured in concordance with the guidelines of the ACC/AHA and classified as increased when LVIDd and/or LVIDs was above 2 standard deviations of the normal values. In addition to the above, all echocardiographic studies were assessed for valvular abnormalities (stenosis and/or insufficiency), hypo/dyskinesia of a myocardial wall, and contractile myocardial function. Valvular disease was evaluated according to the ACC/AHA guidelines. 32 As mild tricuspid and pulmonic regurgitation are common in the general population, these were considered as being physiological.
In case of normal cardiac evaluation, follow-up was advised according to the national guidelines for follow-up of CCSs. 14 In case of abnormalities on either ECG or echocardiogram, a survivor-tailored advice was formulated depending on the severity of the abnormality.

Statistical analysis
Characteristics of the study population, such as age (at diagnosis and at follow-up), BMI, follow-up duration, and cumulative anthracycline dose were summarized using mean and standard deviation or median and range after checking for normal distributions. The level of agreement between any abnormal ECG pattern and any echocardiographic abnormality was calculated by Cohen kappa. Comparisons were made between the group with normal ECGs and the group with abnormal ECG patterns in relation to gender, follow-up duration, age at diagnosis, anthracycline dosage (defined as cumulative anthracycline dose <300 and ≥300 mg/m 2 ), and irradiation to the heart. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated in logistic regression models. In the multivariable model, all variables were corrected for each other to obtain independent associations with abnormal ECG patterns for each variable. All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS), version 20.0.

Study characteristics
During the study period, 414 long-term CCSs visited our Late Effects Clinic. None were using anti-arrhythmic drugs. Survivors were  Table 1 shows the characteristics of the remaining 340 survivors. All were younger than 44 years of age, while 65% (221 of 340 survivors) were aged 24 years or less. None of the survivors included in our study had symptoms suggestive of cardiac disease and/or rhythm abnormalities during their visit. CCSs excluded in our study population did not use any medication affecting cardiac rhythm or QTc.

ECG and echocardiography
Abnormal ECG patterns were found in 73 survivors (21.5%) ( Table 2). Ventricular conduction disorders patterns, high amplitude R waves, and sinus bradycardia were the most common patterns found.
A prolonged QTc was found in two survivors; both of whom had a normal QTc at diagnosis.
Echocardiographic parameters are shown in Table 3. Only one survivor with a cumulative anthracycline dose of 300 mg/m 2 had an abnormal FS of 24% and an EF of 48% at 13.5 years of follow-up. Fortyfour (12.9%) survivors had an abnormal echocardiography, with mitral insufficiency and increased left ventricular dimension (LVIDd and/or LVIDs) being the most common (Table 3). These abnormalities were not seen in their previous echocardiograms. No survivor had more than one abnormal echo finding.

Comparison between survivors with a normal ECG (n = 237) and survivors with an abnormal ECG pattern (n = 73)
All variables were analyzed with univariable and multivariable logistic regression and the results are shown in Table 4

DISCUSSION
Screening for conduction abnormalities was recently recommended as one of the topics of future research in CCSs. 13  In our study population, we found ECG abnormalities that were coded as "other" and could not be coded according to the Minnesota Code. Although the role of monitoring of these "other" ECG abnormalities in cancer survivors is not yet clear, we are of the opinion that these abnormalities may need further follow-up. These "other" ECG abnormalities were seen in 12% of the population with abnormal ECG patterns and in only 2.6% of the whole study population.
IRBBB is commonly found among healthy athletes and is perceived as a benign finding. 34 The same is true for heart rates between 50 and 60 beats/min without conduction abnormalities. A study by Liao et al. 35 showed an increased likelihood of development of a CRBBB when having an IRBBB. CRBBB, in contrast to IRBBB, has been associated with a higher risk of cardiovascular death. [35][36][37] High-amplitude R waves could be a result of physical training. 38 Conversely, a population study in Belgium showed a relative risk of In the present study, we found low agreement between abnormal ECG patterns and echocardiographic abnormalities. Landau et al. 44 also showed a poor correlation between ECG and echocardiographic abnormalities. The fact that a normal echocardiogram does not rule out an abnormal ECG, and vice versa, is not surprising.
Echocardiography is the most widely used bedside modality to diagnose subclinical anthracycline-induced cardiotoxicity, but it is a poor tool for arrhythmias and conduction abnormalities. The number of CCSs with abnormalities on ECG or echocardiography was too low to detect correlations between specific subgroups, for example, "highamplitude R waves" (MC 3-1) with the echocardiographic abnormality "increased LV dimension" (LVIDd and/or LVIDs). The ECG abnormalities were not associated with the cumulative anthracycline dose, duration of follow-up, and age at diagnosis, as described earlier by Postma et al. 45 We did find male gender to be a risk factor for abnormal ECG patterns. Previous literature showed female gender as a risk factor for anthracycline-induced cardiotoxicity using echocardiography. 46 In our study, we focused on abnormal ECG patterns. The prevalence of ECG abnormalities is also higher in men in the general population, indicating that this finding might be independent of anthracycline exposure. 47 In a recent study by Mulrooney et al., male gender was reported as a risk factor for cardiomyopathy (defined as an EF <50%).
An analysis with coronary heart disease and conduction abnormalities could not be done because of their low prevalence of conduction abnormalities (4.4%). 48 In our study, however, we included all ECG abnormalities (21.5% of survivors).
Although one in five survivors had an abnormal ECG, it is still unclear whether these abnormal ECG patterns are clinically relevant.
The ECG findings in our study are heterogeneous and not necessarily indicative of subclinical cardiomyopathy.
In the present Dutch guideline and the COG guideline for long-term follow-up of CCSs, screening with ECG is recommended only at baseline and when clinically indicated. 14,15 This study contributes to the knowledge that abnormal ECG patterns are common in asymptomatic CCSs, although it remains difficult to indicate the long-term clinical implications of our findings. However, the latter was not the aim of our present study.
We agree with others 13 that more research is needed to indicate if, and which, abnormal ECG patterns will lead to (clinical) cardiac disease in the future. A prospective follow-up study of longterm CCSs with abnormal ECG patterns should be conducted to identify those of clinical relevance. Only then can one answer the question whether screening with ECG in long-term follow-up of CCS is justified.
One of the limitations of the present study was its retrospective nature.
We had no data on whether survivors who did not visit the Late Effects Clinic upon invitation had clinical symptoms of cardiotoxicity. In addition, data on other possible risk factors for an abnormal ECG (such as smoking, hypercholesterolemia, and diabetes mellitus) were not available for our population. These data have now been incorporated in new national studies.
We conclude that abnormal ECG patterns were present in 21.5% of asymptomatic long-term CCSs visiting the Late Effects Clinic. The lack of agreement between abnormal ECG patterns and echocardiographic abnormalities may justify using both tests during follow-up. However, it is not clear whether the abnormal ECG patterns will be clinically relevant in the future. Our findings support the need for prospective studies on the relevance of ECG screening in the long-term follow-up of asymptomatic CCSs. In the meantime, we advise to perform an ECG at the first visit during LATER follow-up of CCSs as proposed in the current guidelines. 14, 15 We also advise to perform ECG following the initial ECG when there are symptoms suggestive of cardiotoxicity or conduction abnormalities.

ACKNOWLEDGMENTS
We kindly thank Imke Tomasouw-Janssen, Marian Zwartjes, and Romana d'Ancona for their help in obtaining the echocardiographic data.