Article Text
Abstract
Objective To assess the prognostic value of cardiac iodine-123 metaiodobenzylguanidine (123I-MIBG) scintigraphy to predict ventricular arrhythmias in patients with heart failure (HF) listed for implantable cardioverter-defibrillator (ICD) devices as primary prevention.
Design, setting and patients A prospective cohort study in 27 patients with HF referred for ICD implantation (alone or in combination with cardiac resynchronisation therapy) at a tertiary cardiac centre.
Methods Cardiac 123I-MIBG scintigraphy was performed with calculation of early and late heart-to-mediastinum (H:M) ratios, washout rate, and summed defect score from single photon emission computed tomography (SPECT) acquisition. Resting myocardial perfusion SPECT using 99mTc-tetrofosmin was also performed and a summed score calculated. Innervation-perfusion mismatch was evaluated by comparing SPECT scores.
Main outcome measure Ventricular arrhythmia requiring ICD therapy.
Results At 16 months median follow-up, 10 (37%) patients experienced a significant arrhythmic event. Compared with patients who suffered no event, these individuals had lower early and late H:M ratio and higher 123I-MIBG SPECT defect scores: 1.83±0.43 versus 2.34±0.33 (p<0.001); 1.54±0.38 versus 1.96±0.38 (p=0.005); 37.0±9.4 versus 25.5±7.7 (p=0.001). Mismatch scores were also higher: 18.5±8.5 versus 8.4±5.0 (p<0.01). Optimal thresholds for predicting arrhythmia were <1.94 for early H:M ratio (sensitivity 70%, specificity 88%); <1.54 for late H:M ratio (sensitivity 60%, specificity 88%); 123I-MIBG SPECT defect score ≥31 (sensitivity 78%, specificity 77%).
Conclusions In HF patients without prior ventricular arrhythmia, 123I-MIBG imaging strongly predicts future arrhythmic risk. This may inform the process of case selection for ICD therapy on an individual basis, although no single measurement provides sufficient reassurance to obviate device implantation if otherwise clinically indicated.
- Metaiodobenzylguanidine
- implantable cardioverter-defibrillator
- heart failure
- ventricular arrhythmia
- imaging and diagnostics
- nuclear cardiology
- echocardiography
- cardiac resynchronisation therapy
- arrhythmias
- allied specialities
- defibrillation
- pacemakers
- cardiac function
Statistics from Altmetric.com
- Metaiodobenzylguanidine
- implantable cardioverter-defibrillator
- heart failure
- ventricular arrhythmia
- imaging and diagnostics
- nuclear cardiology
- echocardiography
- cardiac resynchronisation therapy
- arrhythmias
- allied specialities
- defibrillation
- pacemakers
- cardiac function
Introduction
Sudden cardiac death due to ventricular arrhythmia is the most common mode of death in patients with moderate heart failure (HF).1 Implantable cardioverter-defibrillator (ICD) devices have become established as a means of reducing the risk of death from arrhythmia in this setting. Multiple trials have demonstrated the prognostic value of ICD implantation following a previous episode of ventricular arrhythmia.2–4 Evidence also suggests prognostic benefit for selected patients without prior arrhythmia, but deemed to be at high risk. This applies both in ischaemic heart disease (IHD) and non-ischaemic cardiomyopathy.5–7
Notwithstanding the potential benefit, there are considerable drawbacks to device therapy. ICDs are expensive, with further costs related to pacemaker clinic follow-up and the need for subsequent generator renewal. The implant procedure carries a risk of complications (particularly in inexperienced hands) and there are significant ongoing medical and psychological hazards associated with ICD therapy.8 Foremost among these is the risk of inappropriate ICD shocks, while conversely, many individuals never suffer an arrhythmia requiring appropriate ICD therapy.9 There is, accordingly, much interest in refining the risk stratification process for patients undergoing ICD implantation, to identify those individuals who are most likely to benefit, and to avoid unnecessary implants where possible.
Iodine-123 metaiodobenzylguanidine (123I-MIBG) is an analogue of the sympathetic neurotransmitter norepinephrine, which can be used to assess adrenergic nerve function. Cardiac 123I-MIBG scintigraphy allows non-invasive evaluation of myocardial sympathetic activity, with values for 123I-MIBG uptake found to correlate with tissue norepinephrine content on myocardial biopsy specimens.10 Furthermore, there is increasing clinical evidence to suggest that abnormal 123I-MIBG imaging predicts adverse clinical events including death and ventricular arrhythmia in high-risk populations, for example, patients with HF, or following ST-elevation myocardial infarction.11–16
Study hypothesis
Despite a considerable body of evidence suggesting overall prognostic value, there is limited evidence for the use of cardiac 123I-MIBG imaging to predict arrhythmias in the setting of ICD therapy for HF. The main aim of this study was to investigate the extent to which well-preserved cardiac sympathetic function—assessed by global and regional uptake/washout of 123I-MIBG—predicts low risk of future ventricular arrhythmia in HF patients receiving ICD therapy for primary prevention. A secondary aim was to evaluate whether any cut-off values could identify a subgroup for whom ICD therapy could be safely avoided or deferred.
Methods
Adult patients with known HF were recruited prospectively at the point of initial referral for device therapy. None of the subjects had previously experienced a clinically significant ventricular arrhythmia (causing symptomatic haemodynamic compromise or cardiac arrest). All individuals had a conventional indication for receiving an ICD as primary prevention, either as a stand-alone device or in combination with cardiac resynchronisation therapy (CRT-D), based on current UK and international guidance.17 ,18 Mandatory inclusion criteria included left ventricular ejection fraction (LVEF) <30%, or <35% with QRS duration >120 ms on ECG. Patients with any severity of HF symptoms were considered for inclusion if they were otherwise appropriate ICD candidates according to the above guidance.
Patients were excluded if they had suffered a myocardial infarction or undergone a revascularisation procedure in the preceding three months. 123I-MIBG imaging was carried out prior to the implant procedure if possible: in those cases where imaging could only be performed after implantation, subjects were excluded if any ICD therapy had been delivered in the interim (excluding defibrillation testing at the time of implant). Clinicians were blinded to the results of 123I-MIBG imaging.
Imaging protocol
Following thyroid blockade with 400 mg oral potassium perchlorate, 400 MBq (±10%) 123I-MIBG (AdreView, GE Healthcare) was injected intravenously at rest. Planar imaging was carried out 10–15 min after injection, and then repeated 3–4 h later. A single non-gated single photon emission computed tomography (SPECT) acquisition was also performed immediately following initial planar imaging. Planar images were acquired in an anterior projection using a dual-headed Philips Forte gamma camera fitted with medium-energy collimators. Counts were acquired for 900 s into a 256×256×16 pixel matrix. The main energy window was set to 159 keV (±10%) with simultaneous acquisition of a 194 keV (±10%) window to allow subsequent scatter correction. SPECT acquisition was carried out using the same camera: 32 projections were acquired over a 180° contoured orbit, into a 64×64×16 pixel matrix with 60 s per projection. Images were processed on a Philips Pegasys workstation, using Cedars-Sinai AutoQuant software for the SPECT data. Using early and late planar images, heart-to-mediastinum (H:M) ratios were derived by manually drawing separate regions of interest (ROI) around the entire heart and upper mediastinum, and measuring counts per pixel within these regions. Washout Rate (WR) calculation, incorporating correction for radioactive decay, was performed using the formula given below (adapted from Ogita et al13):
Pixel count values were scatter-corrected by subtracting counts obtained from the 194 KeV energy window from those obtained at the main 159 KeV window. This correction factor was used in the calculation of both H:M ratios and WR. SPECT images were analysed using a 17-segment model with 123I-MIBG uptake graded on a scale of 0–4, where 0=normal; 1=mildly reduced; 2=moderately reduced; 3=severely reduced; 4=absent. Summed defect score was calculated by summing the score for all segments.
Reproducibility
In order to confirm reproducibility of the 123I-MIBG imaging technique, the first 22 scans performed were evaluated by two separate readers; with the average value for each parameter taken for the main analysis. Having established that the technique was highly reproducible, the remaining scans were assessed by a single reader. Reproducibility was excellent: mean difference between the two readers for early H:M ratio was 0.01 (SD 0.07), R2=0.98; for late H:M ratio difference was 0.00 (SD 0.06), R2=0.98; for WR mean difference was 0.7% (SD 3.68%), R2=0.95.
Other investigations
On a separate day, participants underwent gated SPECT myocardial perfusion scintigraphy using 750MBq 99mTc-tetrofosmin (MyoView, GE Healthcare) injected at rest. Imaging was carried out on the same γ camera with a conventional myocardial perfusion imaging protocol. Image processing was performed using the same workstation and software as for 123I-MIBG SPECT imaging, including calculation of a summed rest score to quantify resting myocardial perfusion/viability and permit assessment of innervation-perfusion mismatch. Once again, a 17-segment model was used for SPECT image analysis. Mismatch was evaluated on a segment-by-segment basis, subtracting the perfusion score from the 123I-MIBG SPECT defect score for each segment.
Prior to device implantation, all participants underwent transthoracic echocardiography to evaluate LV function, with LVEF calculated using Simpson's biplane method. Most subjects also had a venous blood sample taken for B-type natriuretic peptide (BNP) level.
Device implantation
ICD/CRT-D devices were implanted as per local protocol if clinically indicated. Not all study subjects were initially deemed suitable candidates for device therapy, although one individual did proceed to ICD implantation subsequently. Nevertheless, all subjects were followed-up.
Follow-up
Subjects were reviewed at least 6 months after device implantation. Each individual was contacted in person to ascertain whether any symptomatic clinical events had occurred; where necessary, clinical records were also reviewed. ICD interrogation was performed in all subjects to clarify the exact nature of symptomatic events, and to detect additional asymptomatic arrhythmias requiring device therapy. Standard device interrogation algorithms were used to distinguish different arrhythmias. The primary outcome measure was the occurrence of potentially life-threatening arrhythmia—either ventricular fibrillation (VF) or sustained rapid ventricular tachycardia (VT) requiring ICD therapy either by anti-tachycardia pacing (ATP) or defibrillation—or of sudden death likely to have been arrhythmic in nature. Non-sustained VT, neither requiring therapy nor causing haemodynamic compromise, did not qualify as an arrhythmic event; neither did ‘inappropriate’ ATP or defibrillation, for example, for rapid atrial fibrillation. For the one individual who did not have a device implanted, outcome was based on symptoms alone.
Statistical analysis
Subjects were grouped according to the pre-specified primary outcome described above, and statistical analysis carried out between the two groups. A significance level of p<0.05 was used for all analyses. Categorical variables were analysed using the two-tailed Fisher's exact test, with unpaired t-test to analyse continuous variables. Nonparametric variables were analysed using the Mann-Whitney U test. Receiver-operating characteristic (ROC) curve analysis was performed to determine optimal cut-off points for predicting outcome.
Results
A total of 27 patients were recruited into the study: 25 males, 2 females; mean age 60.7±16.6 years. HF was due to IHD in 12; idiopathic dilated cardiomyopathy in 10; burnt-out valve disease in three; familial DCM in one; and anthracycline cardiotoxicity in one. Mean LVEF was 23.7%±9.3%. Baseline clinical characteristics and investigation findings are shown in table 1 for all study subjects, and stratified according to outcome. There were no important differences in terms of medical therapy or HF aetiology between the two outcome groups.
Seven patients received stand-alone ICD implants; 19 received combined CRT-D devices; one did not undergo device implantation. Patients receiving CRT-D devices had lower LVEF than those receiving ICD devices (22.1% vs 29.4%, p=0.03), but there were no other significant differences between the groups in terms of demographics or medical therapy. During a median follow-up period of 15.9 months (range 5.6–33.6 months), 10 (37.0%) subjects suffered an arrhythmic event fulfilling the primary outcome criteria: three VF requiring defibrillation, six sustained VT (four treated by defibrillation and two by ATP) and one sudden arrhythmic cardiac death. The proportion with an arrhythmic event was similar for ICD and CRT-D recipients (43% vs 37%, p=NS).
For scintigraphic parameters, both early and late H:M ratios on planar imaging were significantly lower in the arrhythmia group (1.83 vs 2.34, p<0.001, and 1.54 vs 1.96, p=0.005, respectively). There was a trend towards higher WR, though this did not reach statistical significance. Figure 1 shows a scatter plot of early and late H:M ratio for each patient, grouped according to outcome.
Turning to SPECT imaging, summed 123I-MIBG SPECT defect scores were higher in the arrhythmia group (37.0 vs 25.5, p=0.002), as were ‘mismatch’ scores comparing innervation and perfusion data (18.5±8.5 vs 8.4±5.0, p=0.007). There was a trend to better myocardial viability (lower summed rest perfusion score) in the group without arrhythmia, but this failed to reach statistical significance. These findings are displayed in figure 2, with summed 123I-MIBG SPECT defect score, summed resting perfusion score and innervation-perfusion mismatch score shown for all individuals according to outcome group.
Non-scintigraphic indices were not strong predictors of arrhythmia, with no significant difference in mean LVEF, 6 min walk distance or New York Heart Association (NYHA) class between outcome groups, though BNP levels were significantly higher in the arrhythmia group. Although these indices were not available for all patients, there was no significant difference in baseline variables, including HF aetiology and medical therapy, or any 123I-MIBG-imaging parameter between those individuals with complete data compared with those without (p=NS for all comparisons).
Receiver-operator characteristic (ROC) analysis was used to compare the performance of individual 123I-MIBG-imaging parameters at predicting arrhythmias. All demonstrated good predictive values, with the greatest area under the curve being 0.83 for SPECT defect score. An optimal derived cut-off score of ≥31 yielded sensitivity of 78% and specificity of 77% for predicting arrhythmic events. With a 37% observed incidence of such events in the study group, this equates to a 67% positive predictive value (PPV) and 86% negative predictive value (NPV). Optimal cut-off for early H:M ratio was 1.94 (sensitivity 70% and specificity 88%) and for late H:M ratio 1.54 (sensitivity 60%; specificity 88%), with better PPV than for SPECT defect score, (77% and 75%, respectively), but inferior NPV (83% and 79%) (figure 3, panels A–C).
Discussion
Abnormal cardiac sympathetic function assessed by 123I-MIBG scintigraphy strongly predicts the likelihood of future arrhythmic events in HF patients undergoing ICD implantation for primary prevention. Both early and late H:M ratios were significantly lower in patients who suffered an arrhythmic event than in those who did not, with a clear trend towards higher WR as well. Furthermore, 123I-MIBG SPECT data revealed significantly higher defect scores, and greater innervation-perfusion mismatch in patients with an arrhythmic event. There was a significant association between raised BNP level and arrhythmia risk, though no difference in other clinical parameters, including LVEF, NYHA class or 6 min walk distance. Using ROC analysis, a SPECT defect score cut-off of ≥31 was found to predict arrhythmia with sensitivity and specificity both above 75%; a score of <31 had high NPV (86%) for excluding subsequent ventricular arrhythmia.
Limitations of ICD therapy
Since the first landmark trials over 10 years ago demonstrated a prognostic benefit from ICD use following an episode of ventricular arrhythmia, indications for ICD therapy have progressively expanded.2–4 ,19 In particular, the MADIT-II trial suggested that ICDs reduced mortality in patients with LVEF <30% after myocardial infarction, regardless of any previous documented arrhythmia, greatly expanding the number of potential ICD recipients.5 This was followed by the COMPANION and SCD-HeFT trials which demonstrated similar prognostic benefit for selected patients in the setting of non-ischaemic cardiomyopathy as well.6 ,7
Nevertheless, despite proven benefit at the population level, a large proportion of individuals receiving ICD implants never experience an arrhythmia requiring therapy from the device. In one study, the annual rate of ICD shock was 7.1% (5.1% appropriate shock for sustained VT or VF), and even at 5 years only a fifth of ICD recipients had received appropriate device therapy.7 Furthermore, a large proportion of shocks delivered are found to be inappropriate: during MADIT-II follow-up 11.5% patients received at least one inappropriate shock, and 31% of all shocks delivered were inappropriate.20Aside from the associated physical discomfort and psychological distress, evidence suggests worse prognosis in individuals who receive inappropriate ICD shocks compared with those who never receive a shock.9 Additional stratification prior to implantation has the potential to identify both those individuals at highest risk of arrhythmia and, perhaps more importantly, those at low risk in whom the risks associated with ICD therapy may actually outweigh the likely benefit.
A number of alternative non-invasive markers have been proposed for arrhythmia risk stratification of patients receiving ICDs, particularly those with IHD: these include signal-averaged ECG and heart rate turbulence/variability on Holter monitoring, and microvolt T-wave alternans analysis. However, conflicting findings from different studies have limited their adoption into clinical guidelines or routine practice.21 ,22 Moreover, none of these markers are suitable for patients with AF, or with ventricular paced rhythm, so a significant proportion of the HF population would not be eligible for testing anyway. Nevertheless, use of these markers in combination with 123I-MIBG imaging might serve to improve overall predictive value, and is certainly worth exploring further in future studies.
123I-MIBG and arrhythmia risk
The use of cardiac 123I-MIBG imaging to evaluate arrhythmia risk has been investigated previously. In 2003, a small pilot study looked at patients with ICD devices already in situ, some of whom had experienced a recent ICD shock. Those individuals with ICD discharges had lower H:M ratio and more extensive SPECT 123I-MIBG defects; but also had a higher likelihood of previous HF and lower LVEF, than subjects without ICD shock, suggesting that the differences seen on 123I-MIBG imaging may simply have reflected a higher prevalence of HF per se.23
Subsequent research has investigated the role of 123I-MIBG imaging to predict arrhythmias in the setting of ICD therapy. In one study, ICD discharge was predicted by a combination of late H:M ratio <1.95 together with either LVEF <50% or elevated BNP level.24 A second, prospective study from the same group examined the value of simultaneous innervation and perfusion imaging to predict arrhythmia, quantifying 123I-MIBG and 99mTc-tetrofosmin uptake in 60 patients. Those patients with ICD shocks during follow-up had significantly lower H:M ratio and more abnormal perfusion scores than those without.25 However, both studies predominantly utilised subjects with prior ventricular arrhythmia, and only a small minority had HF (in the latter study, only a third of subjects had any symptoms, and mean LVEF was almost 50%).
Of more direct relevance is a recent prospective study in patients with HF undergoing ICD implantation.26 In a largely comparable population (90% primary prevention; mean NYHA Class 2.9; mean LVEF 28%), there was a significant association between late SPECT defect score and likelihood of appropriate ICD discharge. In fact, the optimal derived SPECT defect score cut-off was identical to that obtained in the present study. In contrast with most of the published literature, as well as our own results, none of the conventional planar MIBG parameters appeared to predict arrhythmia, although a number of important differences in image acquisition protocol and processing between the two studies may account for this.
The recent AdreView Myocardial Imaging for Risk Evaluation in Heart Failure (ADMIRE-HF) study provides further insight into the prognostic value of 123I-MIBG imaging in HF.27 In this large international study, over 950 patients in NYHA Class II/III with LVEF <35% underwent 123I-MIBG imaging, and were then followed to examine rates of adverse events, including ventricular arrhythmia. In total, 25% of individuals suffered an event; late H:M ratio <1.6 was a significant predictor, with an overall event rate of 37% in these individuals compared with 15% for those with H:M ratio >1.6. This study was not performed specifically in ICD recipients, but in an otherwise comparable population, and the H:M ratio threshold used in their analysis closely matches the optimal cut-off value derived in the present study.
Controversy remains over which is the most robust measurement for overall quantification of cardiac sympathetic activity using 123I-MIBG, although much recent literature has tended to favour late H:M ratio. A meta-analysis found that WR and late H:M ratio appeared to be the best predictors of prognosis in patients with HF.14 It should be noted, however, that 123I-MIBG SPECT defect score has been much less extensively studied, and it is certainly of interest that in the present study, this out-performed the planar indices in terms of overall predictive value for arrhythmia.
Study limitations
The relatively small study population meant that there was insufficient statistical power for multivariate analysis to be performed. It would otherwise have been desirable to investigate the added value of combining imaging results with other clinical findings—particularly BNP level—in predicting arrhythmia. Notwithstanding this limitation, there were several individuals who had preserved 123I-MIBG indices, but still suffered an arrhythmic event. Given the advanced LV impairment seen in the study population, even those patients with reasonable sympathetic innervation may be at relatively increased risk compared with the average ICD recipient seen in clinical practice. Perhaps for individuals with less severe systolic impairment, 123I-MIBG imaging may provide greater discriminatory power to predict arrhythmia risk.
Our study does not permit full economic analysis of the cost-effectiveness of 123I-MIBG imaging prior to ICD implantation. Clearly, however, a balance needs to be struck between the cost of the imaging test (around £1000, including reporting), and that of an ICD implant (£16 000 according to the 2006 NICE Guidance—as well as subsequent pacing of clinic follow-up). Thus, if we were to take these example figures as representative of real-life costs, 123I-MIBG imaging might be considered cost-effective if it helps avoid at least one ‘unnecessary’ device implant for every 16 scans performed; although this figure would, of course, vary substantially depending on other factors, including the required frequency of repeat imaging.
Conclusion
In patients with HF, cardiac 123I-MIBG imaging provides incremental prognostic information regarding the risk of future arrhythmia which may be helpful in informing the process of case selection for ICD therapy. Although no single value derived from 123I-MIBG imaging can definitively exclude arrhythmia, it is probably neither realistic nor appropriate to expect any isolated test to be able to do so. Nevertheless, the use of 123I-MIBG imaging in this setting may permit a more nuanced assessment of the balance between potential risk and benefit on an individual patient basis, with a decision to proceed to device implantation based on these considerations. Larger studies in future may be able to define clearer thresholds for different tiers of risk, and establish how frequently 123I-MIBG imaging might need to be repeated in order to maintain confidence regarding that risk level.
Acknowledgments
We thank all the staff in the Nuclear Medicine Department at Harefield Hospital for their help with the study; in particular Rommel Manlapig, Fabrice Ghiotto, Gabriella Apap-Bologna and Suzie Hinton-Taylor for technical support in cardiac 123I-MIBG and 99mTc-tetrofosmin scintigraphy acquisitions.
References
Footnotes
Funding The study was supported by GE Healthcare Ltd who provided all doses of the 123I-MIBG radiopharmaceutical used in the study. There was no additional financial support from any company or institution, and the study authors have no other conflicts of interest to declare.
Competing interests None to declare.
Ethics approval This study was approved by the local Research Ethics Committee (REC No. 07/Q0404/48) and Administration of Radioactive Substances Advisory Committee (ARSAC). Written informed consent was obtained from all participants. The study was carried out in accordance with the principles of the Declaration of Helsinki.
Provenance and peer review Not commissioned; externally peer reviewed.