Introduction
The ageing of the population and the improved medical treatment of cardiac patients has led to an increased prevalence of heart failure (HF) [
1]. Despite the therapeutic innovations, the mortality rate of patients with HF remains high with an estimated 5-year mortality rate of 54 % in men and 40 % in women [
2].
Increased myocardial sympathetic activity is a prominent feature of HF by which the failing heart tries to compensate for the reduced cardiac output [
3]. In the chronic state of HF these compensatory mechanisms become deleterious, causing myocardial hypertrophy and fibrosis, leading to cardiac remodelling and restructuring [
4]. At the cellular level, the increased sympathetic activity causes increased neuronal release of norepinephrine (NE), that leads to a significant reduction in presynaptic NE uptake due to posttranscriptional downregulation of the cardiac NE transporter [
5]. The decrease in the NE reuptake mechanism can be assessed noninvasively by radionuclide imaging with the
123I-labelled NE analogue meta-iodobenzylguanidine (MIBG) [
6]. MIBG is taken up into the presynaptic cardiac sympathetic nerves by the NE uptake-1 transporter, and the amount of MIBG retention over several hours after administration reflects neuronal integrity [
7]. The most commonly used quantitative measurements of myocardial MIBG uptake are the calculated heart-to-mediastinum (H/M) ratio and washout ratio (WR) determined from planar MIBG images. A low delayed H/M ratio has been shown to be an independent predictor of ventricular tachyarrhythmia [
8,
9], appropriate ICD therapy [
10] and sudden cardiac death [
11] in HF patients. In addition, an increased myocardial washout has been associated with an adverse prognosis [
12]. Subsequently, planar MIBG myocardial scintigraphy has been demonstrated to have strong prognostic value [
13,
14].
Despite the proven prognostic value of MIBG imaging in patients with HF, there are still several limitations that prevent this technique from being implemented as a clinical management tool in patients with HF [
15]. Although almost all reports include the H/M ratio and WR as the measure of myocardial uptake, the methods used to obtain these parameters show substantial variation [
16]. This variation can be caused by the influence of collimator choice, acquisition time and duration, and the location and size of the cardiac and mediastinal regions of interest (ROIs) [
17‐
19]. MIBG uptake can be very low, particularly in patients with HF, hampering the assessment of the cardiac ROI on planar MIBG images. Therefore, the level of experience in drawing the cardiac ROI on the planar MIBG images might influence the measured H/M ratio and WR. Moreover, because of this difficulty in determining the correct contours of the cardiac ROI in HF patients, the assessment of the location and size of the cardiac ROI might affect the measured H/M ratio and WR more in patients with HF as compared to healthy subjects. Furthermore, different methods are used to calculate the WR from early and delayed planar MIBG images [
16].
To our knowledge there are no studies that have compared the use of these different mathematical methods for the assessment of WR. Therefore, the effects of the method on the clinical interpretation of WR remain unclear, and it is uncertain which method would provide the more accurate and reproducible estimates. Consequently, the reproducibility of the clinically important parameters of planar MIBG myocardial scintigraphy remains unidentified in HF patients. Therefore, the purpose of this study was to assess the reproducibility of the H/M ratio and WR in HF patients. We evaluated the influence of two different methods to assess the WR, the effect of cardiac ROI size and position on the assessment of the H/M ratio, as well as the impact of the level of postprocessing experience on the reproducibility of planar MIBG images.
Discussion
The present study showed excellent reproducibility of the assessment of the most reported clinical measurements on planar MIBG images in HF patients. In particularly, when using a manually drawn polygonal cardiac ROI the highest level of agreement was observed between the measured H/M ratios and WR. However, a fixed-size cardiac ROI can also be used for the assessment of delayed H/M ratios, since high agreement was seen between the measured delayed H/M ratios obtained using a traditional manually drawn cardiac ROI and those obtained using a fixed-size cardiac ROI. In addition, for the calculation of WR, the formula using only the cardiac ROI from the early and delayed planar MIBG images (WR = (H
e − H
1)/H
e × 100) provided more reliable values than the formula including background correction. Excellent agreement was also observed between the measurements of the H/M ratios and WR performed by an experienced and an inexperienced observer. A short period of training on the postprocessing technique was enough to accurately assess both the mediastinal ROI and the cardiac ROI, showing that a limited amount of experience in combination with basic anatomical knowledge is sufficient to successfully reproduce the results of an experienced observer. Subsequently, the results for the intra- and interobserver analyses could be reproduced in a subpopulation of patients with a very low H/M ratio, showing that the reproducibility remained excellent over the total range of delayed H/M ratios. These results imply that the measurements on planar MIBG myocardial scintigraphy are reliable even in patients with severe myocardial sympathetic denervation patterns.
Several studies have shown the excellent prognostic value of delayed H/M ratios and WR independent of other commonly used clinical parameters such as LVEF [
12‐
14]. In addition, patients with decreased delayed H/M or increased myocardial MIBG washout show a higher risk of ventricular arrhythmias and sudden cardiac death [
8‐
11]. Therefore, MIBG myocardial scintigraphy could assist in a more individualized treatment strategy for HF patients. Moreover, changes in adrenergic cardiac activity in response to pharmacological treatment in patients with HF can be tracked with the use of MIBG imaging [
23‐
25]. However, even though multiple studies support the usefulness of this technique, MIBG myocardial scintigraphy is not yet widely clinically implemented. One factor hampering the clinical implementation of MIBG myocardial scintigraphy is the lack of standardization of the acquisition and postprocessing parameters [
18]. Improved standardization of imaging protocols could contribute to increased applicability of MIBG imaging in HF patients [
16].
Although the H/M ratio obtained from planar images has been used as a quantitative parameter in MIBG myocardial scintigraphy, H/M ratios may vary markedly according to the imaging equipment, especially with different types of collimator [
18,
26]. However, Nakajima et al. have recently shown that standardization of H/M ratios by a heart chest calibration phantom method is feasible among different collimator types, allowing practical multicentre comparisons of H/M ratios [
27]. While the effect of collimator selection on H/M ratios has been studied, similar attention should be paid to the variation caused by differences in setting the mediastinal and cardiac ROIs. Many independent reports from centres in different continents have shown that MIBG myocardial scintigraphy provides valuable prognostic information. However, among these different investigations there is no consensus on the shape, size and positioning of the ROIs for the heart and the mediastinum.
The mediastinal ROI on planar cardiac
123I-MIBG images reflects nonspecific mediastinal tissue activity and is a good reference site for the qualification of the cardiac sympathetic innervation pattern, because of the small amount of scatter and the low sympathetic activity in the mediastinum [
28]. Different sizes for the mediastinal ROI have been reported, ranging from 7 × 7 pixels to 9 × 9 and 40 pixels with a 128 × 128 matrix, and from 13 × 13 pixels to 20 × 20 and 100 pixels with a 256 matrix [
16]. However, mostly the use of a rectangular mediastinal ROI has been reported with unspecified size [
16]. Since a rectangular mediastinal ROI is recommended in the proposal for standardization by Flotats et al. [
21], we chose to use a rectangular mediastinal ROI with a size of 13 × 20 pixels (with a 256 × 256 matrix) placed in the upper part of the mediastinum. A standardized shape and size of the mediastinal ROI is preferable, because small differences in size can cause variations in the measured H/M ratio [
17]. Furthermore, on the determination of WR there is little consensus in the literature. Over the different prognostic studies, more than six different methods are described for the calculation of the WR from the early and delayed planar MIBG images [
16]. In addition, the normal ranges and cut-off values for the WR remain unclear, while the normal values are method-dependent. Importantly, the discrepancies that arise from the use of the different mathematical methods are clinically relevant [
16]. In the current analysis, the most reliable results for the calculation of the WR in HF patients were provided by calculation without background correction and without the use of
123I time decay correction. Therefore, this method might be the most appropriate for the calculation of WR in HF patients.
Furthermore, for the assessment of the cardiac ROI in the present study, a polygonal manually drawn cardiac ROI was used including the myocardium and the left ventricular cavity, which resulted in excellent intra- and interobserver agreement of the early and delayed H/M ratios obtained. However, a fixed-size cardiac ROI may be advantageous for future development of standardized and automatically traced cardiac ROIs. Therefore, we evaluated the reproducibility of delayed H/M ratios using two types of fixed-size cardiac ROIs. Both the oval and the circular fixed-size cardiac ROIs showed excellent correlations with the manually drawn cardiac ROI (ICC 0.95 and 0.86, respectively). However, slightly more variation was observed when the circular cardiac ROI was used. This variation could be explained by the fact that the circular ROI covers only a part of the heart. To prevent scatter from the lungs or liver, a circular ROI with a radius of 21 pixels was chosen, fitting into the apex of the myocardium without covering surrounding lung or liver tissue. However, because in most HF patients the left ventricle is enlarged, a circular cardiac ROI with a radius of 21 pixels (approximately 30 mm) often covers only a small part of the myocardium in these patients. This might influence the accuracy of the measured delayed H/M ratios in these patients. Okuda et al. used a cardiac ROI with a radius of 30 mm in the development of a semiautomated algorithm for the calculation of the H/M ratio, showing high reproducibility (ICC 0.97 in the interobserver analysis of the delayed H/M ratio) [
19]. However, this study was performed in 37 patients of whom only 5 had HF. For future development of a semiautomated algorithm for the calculation of the H/M ratio in HF patients, the use of a fixed oval cardiac ROI is probably more suitable, because a larger amount of the myocardium is covered using this method.
When routine clinical application of MIBG myocardial scintigraphy is encouraged by the effort to standardize the imaging procedure, knowledge about the acquired level of experience needed to evaluate the images would be valuable. This study showed that the early and delayed H/M ratios and WR measured by an inexperienced observer were in excellent agreement with the measurements of an experienced observer. A 2-h period of training examining five example cases was enough for accurate setting of the mediastinal and cardiac ROI. This indicates that the data analysis and the interpretation of planar MIBG images are feasible. Therefore, MIBG myocardial scintigraphy could be implemented easily in the clinical work-up of HF patients. Moreover, the subanalysis performed in patients with very low delayed H/M ratios showed that the reliability remains excellent over the total range of observed myocardial sympathetic activity patterns detected on MIBG myocardial scintigraphy in HF patients.
Limitations
The current study had several limitations. First, only the reproducibility was evaluated in this study, while the precision of planar MIBG myocardial scintigraphy also depends on the validity of the measurements performed. Second, the reproducibility of MIBG myocardial scintigraphy within subjects could not be evaluated, because MIBG myocardial scintigraphy was not repeated in time within the same patients. In the current study existing data were used of one-time MIBG myocardial scintigraphy evaluations. For the within-subject reproducibility of planar MIBG myocardial scintigraphy, a prospective study with repeated MIBG myocardial scintigraphy over time within the same patient is required. However, repeated MIBG myocardial scintigraphy within the same patient would involve an increased radiation dose. Third, we did not evaluate the influence of the size and location of the mediastinal ROI on the measured H/M ratio. However, the size of the mediastinal ROI was fixed (13 × 20 pixels) and the location was specified so that the variation in the mediastinal ROI was minimal in the current analysis (as shown in the Bland-Altman plot in Fig.
4). Still, an automatic method to set the mediastinal ROI as “default” in order to correct any variation due to the operator may be advisable. Fourth, analyses for the assessment of regional myocardial sympathetic activity on MIBG SPECT images were not performed in the current study. Instead, this study focused mainly on measurements on planar MIBG images. Further studies are needed to investigate the reliability of cardiac MIBG SPECT imaging.
Conclusion
The present study showed excellent reproducibility of planar MIBG myocardial scintigraphy in HF patients. Moreover, for the calculation of WR, the formula without background correction showed high reliability and might therefore be preferable in HF patients. Furthermore, we observed only a small influence of the level of experience on the assessment of H/M ratios and WR, indicating that the data analysis and interpretation of planar MIBG images is feasible. In addition, the use of a fixed-size oval cardiac ROI could make the image analysis procedure even more standardized. Consequently, these results could contribute to the standardization of the data analysis and interpretation of planar MIBG images and therefore might provide extra validation for the reliability of planar MIBG myocardial scintigraphy in HF patients.