This investigation sought to identify and quantify any increased risk of long-term heart failure (HF) after thoracic radiotherapy (RT) for cancer and identify any population covariates that corresponded with increased risk. Electronic databases were systematically searched for studies reporting relative risk, odds ratio, and hazard ratio (HR) for symptomatic HF more than 5 years after RT administration. Clinical characteristics, study design, univariable effect sizes, and associated 95% CIs were extracted. Univariable effect size was pooled and computed in a meta-analysis using random-effects model weighted by inverse variance. Six studies (45,669 patients) with weighted median follow-up duration of 13.9 years were included, each data-linkage study that reported HRs for HF. Pooled HR for long-term HF was significant (HR 1.83 [1.09 to 3.08], p = 0.022), with significant between-study heterogeneity (Q 43.38, df = 5, p <0.001, I 2 88.47%). Statistical significance was lost when excluding studies of malignancies other than breast cancer or hematological malignancies and excluding studies with Newcastle-Ottawa scores <8, but the direction of effect and magnitude remained approximately the same. Subgroup and meta-regression analyses demonstrated that study differences in age at time of RT administration and duration of follow-up explained approximately 80% of observed heterogeneity. Earlier publication date was associated with increased HF risk. Other variables, including female proportion, proportion of adjuvant chemotherapy use, and sample size did not significantly impact the conclusions. In conclusion, RT approximately doubled the long-term risk of HF. This finding was associated with younger age at time of RT and longer follow-up duration, which explained approximately 80% of interstudy heterogeneity.
Heart failure (HF) affects approximately 1% to 2% of population in developed countries, with prevalence rising to >10% in subjects aged ≥70 years. Radiotherapy (RT)-treated patients represent a rare group with a clearly identifiable onset of risk for myocardial injury and as such may represent an attractive target for preventative screening and treatment. However, consensus guidelines by professional medical societies have been inconsistent in their recommendations. According to recent HF guidelines, previous chemotherapy represents an HF risk factor and confers a stage A HF status, but no specific guidance is given for patients treated with RT. The European Society of Medical Oncology 2010 guidelines states that “RT-induced risk is lifelong and requires long-term follow-up” but also note that “follow-up protocols are based on departmental or personal experience.” Expert consensus from the American College of Cardiology and European Association of Cardiovascular Imaging suggest echocardiography at 10 years after RT, with subsequent echocardiograms every 5 years, but further guidance regarding high-risk populations, disease manifestations, and prevalence are needed to optimally guide cardiac screening. Thus, we performed this systematic review and meta-analysis to quantify the long-term risk of HF after thoracic RT for malignancy.
Methods
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines for reporting the systematic review. The search strategy was defined prospectively and listed in the International Prospective Register of Systematic Reviews (PROSPERO) database (registration number CRD42015020508). Citations and details were stored in a database (EndNote X7.4; Thomson Reuters, New York, New York). A liberal search strategy was used to increase sensitivity. Two reviewers conducted a literature search of Medline/PubMed and EMBASE that investigated cohorts of patients treated with thoracic RT for HF for all years from inception to July 2015 (a copy of the search strategy is available as an Online Supplementary Appendix ). HF was defined as the clinical syndrome associated with insufficient cardiac function to meet the body’s demands.
Publications were limited to those published in English. References of publications and relevant studies were also searched for further reports. The search excluded studies of pediatric populations to prevent confounding of results, as a child’s heart appears to be more susceptible chronic RT adverse effects than an adult’s, as evidenced by recognition of younger age at RT being a risk factor for cardiotoxicity in addition to experimental evidence demonstrating greater biologic susceptibility to RT in developing hearts. Publications in peer-reviewed, English-language journals evaluating long-term HF were included in this study if they met the following inclusion criteria: (1) average patient age (either median or mean) of ≥18 years at time of RT administration; (2) administration of adjuvant RT therapy (either tangential or mediastinal) for cancer; (3) measure of risk of HF reported as a binary outcome, either as hazard ratio (HR), risk ratio, or odds ratio; and (4) presence of a interval of at least 5 years between RT and determination of HF status. Exclusion criteria included (1) studies not meeting all the inclusion criteria, (2) nonhuman studies, and (3) abstracts or conference proceedings. No restrictions were applied to the types of patients, study’s country of origin, or institution type where outcome was determined.
Relevant studies were selected from literature searches by 2 review investigators (M.N. and D.R.), who also extracted relevant data. Discrepancies between review investigators were resolved by consensus or, if necessary, by a third author (T.H.M.). Information on publication year, sample size, follow-up duration, average age (either median or mean), gender ratio, RT dose, and use of concurrent chemotherapy were extracted independently from each study.
Reported measures of risk, including odds ratio, HR, or risk ratio, were pooled and analyzed using a random-effects model weighted by inverse variance as described by DerSimonian and Laird. Assumption of heterogeneity was tested using Q, and between-study heterogeneity was quantified using the I 2 value. Sensitivity analyses were performed by removing studies with following characteristics; studies examining cancers other than breast cancer or hematological cancers, studies with Ottawa-Newcastle quality score <8, and studies with low sample size. Subgroup analyses were performed using mixed-effects model with pooled τ 2 estimates. Meta-regression analysis was performed using random-effects model. Publication bias was assessed visually by funnel plots of effect estimates and by Begg statistical test. If the number of studies assessed was <10, then further assessment of possible publication bias would be undertaken with Orwin’s fail-safe N and Duval and Tweedie’s Trim and Fill test. Statistical analysis was performed by Comprehensive Meta-Analysis software, version 2.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
