Discussion
Using a sample of cancer patients that had both MUGA and CMR performed within 30 days, we found that MUGA LVEFs were only modestly accurate when compared with reference LVEFs by CMR, the gold standard technique for the assessment of LVEF. MUGA LVEFs were systemically lower by 1.5%, suggesting only a small mean discordance between the two methods. However, the limits of agreement between MUGA and CMR LVEFs were wide (−19 to 16%). This suggests large individual discordance, or in other words, low accuracy for MUGA LVEFs when compared to CMR LVEFs. Furthermore, using LVEF thresholds of ≥50% and ≥55% to define normal, there was misclassification between MUGA and CMR LVEFs in categorizing 35 and 20% of patients respectively.
Schwartz et al. in 1987 demonstrated in patients receiving doxorubicin that LVEF estimates by MUGA used per their proposed guidelines reduced the incidence and severity of clinical congestive heart failure [
16]. Normal LVEF was defined as ≥50%, and cardiotoxicity was defined as an absolute decrease in LVEF of ≥10% with a final LVEF of ≤50% [
16]. These data, along with data demonstrating high reproducibility [
17] and low variability [
18], established MUGA as the modality of choice for serial testing of LVEF in patients with cancer. However, validation of the accuracy of MUGA in these studies was done with comparisons to contrast left ventriculography [
19], which has significant variability [
20], and is arguably a poor reference standard. In a phantom study comparing CMR, MUGA, and left ventriculography, left ventriculography was the least accurate, and MUGA was less accurate than CMR [
21]. Of note, CMR in this study was performed using a gradient-echo sequence, which has lower blood-to-myocardium contrast [
22], accuracy and reproducibility than the currently used SSFP sequence [
23]. Furthermore, in an in vitro model, LV volumes by CMR have been shown to be highly accurate when compared to volumes obtained using latex casts of excised human LVs [
24].
The systematic bias between CMR and MUGA LVEFs has been variable in the literature [
5,
25,
26]. The discrepancies are likely due to differences between institutions in imaging and analysis techniques, and software. Audits of MUGA LVEFs in the United Kingdom, Australia and New Zealand have demonstrated significant variability between centers, mainly due to differences in the software used for LVEF analyses [
27,
28]. The systemic bias between mean MUGA and CMR LVEFs in this study is overshadowed by the substantial random error when comparing the two modalities. Possible sources of inaccuracy associated with MUGA include suboptimal patient positioning with current gamma cameras, limited spatial resolution, the need for background correction, errors from overlapping structures, and gating inaccuracies due to arrhythmias.
Our findings are in contrast with those of Walker et al. [
29] who studied 50 consecutive patients with breast cancer prior to adjuvant trastuzumab, at 6 and 12 months and found a strong correlation (
r of 0.88, 0.97 and 0.87 at baseline, 6 and 12 months respectively) between MUGA and CMR. It could be argued that this study was not reflective of real-life practice as demonstrated by the exclusion of patients with a history of atrial fibrillation or intraventricular conduction delay. Although consecutive patients were enrolled in the study, a majority had LVEF <55% at baseline (prior to trastuzumab therapy) and proceeded to receive trastuzumab. Additionally, whether MUGA and CMR LVEF analyses were blinded to the results of the other imaging technique was not stated.
Early MUGA studies in the 1970s and 1980s were performed using small-field-of-view, single-headed gamma cameras that allowed optimal positioning of the patient to obtain the best separation between the left and the right ventricles. Current gamma cameras are predominantly large-field-of-view, dual-headed systems that do not permit this degree of patient positioning [
3].
We attempted to identify potential associations for misclassification between MUGA and CMR LVEFs in our cohort. Smaller LV size was a significant predictor of misclassification, which we speculate is a consequence of differences in spatial resolution between the two modalities. Smaller hearts may have less accurate LVEF measurements by MUGA due to its relatively lower spatial resolution. The presence of arrhythmias may have also contributed to discrepancies between the two imaging modalities, as a history of atrial fibrillation was a significant predictor of misclassification at the higher LVEF threshold.
Our findings have important implications for clinical investigations and the care of patients receiving potentially cardiotoxic cancer treatment. MUGA is frequently used in these patients; in a study of 2203 patients 66 years or older who received trastuzumab for adjuvant treatment of breast cancer, 28% had baseline and serial assessment of LVEF with MUGA alone, and 23% with a combination of MUGA and echocardiography [
30]. Important therapeutic decisions are often based on the LVEF in patients with cancer. Imprecise LVEFs leading to incorrect classification of patients as normal or abnormal may lead to erroneous decisions about the choice of standard-of-care treatment, or less cardiotoxic – but potentially less effective – alternatives such as reduced doses of standard chemotherapy regimens or nonstandard regimens. Similarly, they may influence incorrect decisions regarding the frequency of clinical follow up, screening by imaging to detect cardiotoxicity, and treatment with cardiac medications for the cardiomyopathy. Ultimately, erroneous classification of patients as normal or abnormal due to inaccurate LVEFs may result either in cardiomyopathy and heart failure that could potentially have been prevented, or lower treatment response and worse cancer outcomes from less effective cancer treatment used in response to unwarranted concerns for cardiotoxicity.
MUGA involves the use of ionizing radiation in a patient population that requires serial studies. Guidelines for cardiac monitoring after trastuzumab treatment recommend the use of the same imaging modality throughout the course of treatment [
31,
32]. A breast cancer patient receiving adjuvant trastuzumab is recommended to have LVEF assessment before starting treatment, every 3 months during, upon completion of treatment, and every 6 months for at least 2 years following completion of treatment [
33]. More frequent monitoring is recommended if trastuzumab is withheld for a significant drop in LVEF [
33]. With 12 months of adjuvant trastuzumab as the standard of care, this translates into a minimum of nine studies. With an average typical effective ionizing radiation dose of 8 mSv per MUGA [
34,
35], the use of MUGA would result in a significant dose of ionizing radiation with associated risks of radiation-related secondary cancers [
36,
37]. A recent publication highlighted this issue through the case of a patient with multiple myeloma who received 17 MUGAs, corresponding to an effective radiation dose of 113 mSv, over a span of 3 years [
38]. A scientific statement from the American Heart Association on approaches to enhancing radiation safety in cardiovascular imaging carries the recommendation that when a cardiac imaging study is indicated, a comparable test with similar accuracy, cost and convenience, which does not use ionizing radiation, should be preferred [
39].
Echocardiography and CMR are alternatives that do not involve ionizing radiation. However, two-dimensional echocardiography has been shown to have limited performance compared to CMR for the detection of cardiotoxicity in adult survivors of childhood cancer for cardiomyopathy [
40]. Unlike MUGA, CMR provides additional clinically valuable information including assessments of right ventricular (RV) size and function, atrial size and function, valvular disease, pericardial disease, intracardiac thrombus and extracardiac pathology. In our study, CMR revealed that 80% of patients had at least one additional abnormality: 52% had RV dysfunction (defined as RVEF <50%), 33% had an enlarged left atrium, 12% had an enlarged right atrium, 13% had significant valvular disease, 9% had a pericardial effusion, 11% had pleural effusions, 5% had an intracardiac thrombus, and 3% had cardiac tumors. Additionally, late gadolinium enhancement CMR has the ability to detect the presence and patterns of fibrosis, which would help identify the etiology for cardiomyopathy [
41] in cancer patients. T1 mapping is a newer CMR technique for the detection of diffuse myocardial disease, that holds significant promise in the prediction, early detection and prognostication of cardiotoxicity [
42]. Thus, findings on CMR other than the LVEF may have significant impact on management and clinical decision-making [
43].
Limitations
Our findings must be interpreted in the context of the study design. Rather than to perform a head-to-head comparison of two imaging modalities, our aim was to examine the accuracy of real world, clinical LVEFs by MUGA, which oncologists and cardiologists use every day in clinical practice to make important decisions. To achieve this, we compared clinically analyzed MUGA LVEFs with CMR LVEFs that were ascertained by a single blinded expert investigator for this study. This design allowed comparison of real world MUGA LVEFs to arguably the most accurate estimates possible, of the true LVEF (since even a necropsy cannot provide a LVEF).
Since MUGA and CMR studies were not performed on the same day in all cases except one, there is a possibility of true LVEF changes in the interim period. We limited this possibility by excluding patients with clinical events and potentially cardiotoxic treatment during or in the time interval between the two studies. Additionally, we did not find correlations between the time interval between the two techniques and the absolute difference in LVEFs, whether MUGA clinical LVEF was higher or lower than CMR reference LVEF, or whether there was misclassification between normal and abnormal categories.
While making clinical decisions on the management of cardiotoxicity, the change in LVEF is often used in conjunction with the absolute LVEF. We did not investigate changes in LVEF in this study. Finally, this is a relatively small, single-center study subject to referral bias.