ICD in Cardiac Sarcoidosis: Variables Associated with Appropriate Therapy, Inappropriate Therapy, and Device Complications

: Introduction: Those with cardiac sarcoidosis (CS) are at risk of sudden cardiac death (SCD), which may be prevented using an implantable cardioverter–defibrillator (ICD). There are limited data available that follow the post-procedural outcomes of patients with cardiac sarcoidosis (CS) who have had an ICD implanted. Areas Covered: This review highlights studies that focused on both appropriate and inappropriate therapies in those with an ICD, as well as device complications in this group. There were several variables, including age, sex, ventricular characteristics, and findings on cardiac imaging that were investigated and discussed as influencing factors in predicting appropriate and inappropriate therapies. Conclusions: Adverse events in those with an ICD and CS have been minimally reported in the literature. Individuals diagnosed with CS are at high risk of ventricular arrhythmia, with comparable rates of appropriate therapy but with a higher incidence of side effects and inappropriate therapy. The younger average age of CS patients in comparison to other ICD cohorts warrants the need for further, large-scale, prospective trials with periodic interim follow-ups focused on those with this condition.


Introduction
Sarcoidosis is a multi-system disease that most commonly affects the lungs but can also affect almost any other organ including the skin, liver, and heart.It is characterised by the infiltration of non-necrotising granulomas and may result in permanent architectural changes to the pulmonary system.Cardiac involvement is an increasingly recognised manifestation of sarcoidosis.The potentially fatal arrhythmias secondary to cardiac sarcoidosis (CS) have spurred research into the use of an implantable cardioverter-defibrillator (ICD) in this condition.Appropriate ICD therapy refers to either an appropriate shock or a correct device-based rhythm analysis and the recognition of anti-tachycardia pacing for ventricular tachyarrhythmias such as ventricular tachycardiac (VT) or ventricular fibrillation (VF).Ellenbogen et al.'s article helped frame the utility of "appropriate therapy" as an important surrogate marker of sudden cardiac death (SCD) in non-ischaemic cardiomyopathies, albeit one that will overestimate the benefits of ICD shocks [1].Inappropriate therapy, however, is a shock that is delivered in the absence of VT or VF, the consequence of which has been previously shown to lead to myocardial dysfunction, advancement of heart failure, and increased mortality [2][3][4].As we have previously discussed in other works, clinicians are starting to recognise the importance of ICDs in CS, even in those with a potentially normal ejection fraction [5][6][7].Consequently, as the number of patients who receive ICDs increases, so does the importance of recognising those patient groups who will most benefit from its insertion, and, crucially, those who may not.We sought to understand the patient groups that receive appropriate therapies and inappropriate therapies and identify those in whom device complications are more prevalent in order to help guide future guidelines and recommendations that will result in better patient care.

Factors Associated with Appropriate Therapy
Several retrospective studies from across North America, India, and Japan sought to investigate an association between appropriate therapy and patient characteristics in those with CS [8][9][10][11][12][13][14].An overview of this research is provided in Table 1.Many of these single-centre studies were pulled together in a meta-analysis by Azoulay et al. [15], showing that of the 464 participants, 39% of CS patients received appropriate therapy-a value significantly higher when compared to other studies focusing on non-sarcoid-related heart disease, and a similar value to that of a recent UK-based single-centre study where of appropriate therapy were shown to be 32.7% [16].The variables that were identified in this study, alongside other presently discussed studies, fit into three broad categories that will be discussed in greater detail below:
It is worth noting that the burden of CS differs between ethnic groups, with previous studies showing varied rates of cardiac sarcoid-related death between them [7,[21][22][23].Many of the above studies did not include this dataset, and as such are unable to draw conclusions around the impact of appropriate therapies in differing populations.

Ventricular Characteristics
A large meta-analysis found that having a low left ventricular ejection fraction (LVEF) (−10.5, 95% CI: −18.23 to −2.78, p = 0.008), receiving ventricular pacing (OR: 6.44 95% CI: 2.57 to 16.16, p < 0.0001), and having a history of complete heart block (CHB) (OR: 2.19, 95% CI: 1.20 to 3.99, p = 0.01) predict appropriate therapy [15].Regarding ejection fraction, the mean LVEF is generally shown to be lower in those who experienced appropriate therapy when compared to those who did not receive appropriate therapy, and furthermore, no patients within a group with a higher ejection fraction have required transplantation or died.These findings emphasise the highly arrhythmogenic nature of those with CS and a reduced LVEF, supporting the recommendations in the 2014 HRS Consensus, 2017 AHA/ACC/HRS and 2022 ESC Guidelines of implanting an ICD in order to prevent sudden cardiac death (SCD) [24][25][26].
A recent 2022 study found that second/third degree atrioventricular block (AVB) and prior ventricular arrhythmias are indicative of patients receiving appropriate therapy [20].Interestingly, in contrast with the aforementioned meta-analysis, no significant difference has been observed in the LVEF between those who received appropriate therapy and those who did not.As discussed by researchers, this may be as a contribution from the greater prevalence of high-degree AVB in their population, and it was previously shown to be associated with VA, even in patients with a preserved LVEF [27].Prior to this study, AVB had been shown to be predictive of appropriate ICD therapy, a finding which likely corresponds to the greater severity and extent of disease in order for AVB to manifest, which, as such, will predispose individuals to an increased risk of arrhythmias [18].
This increased risk of arrythmia, and as such appropriate therapy, in those with CS was further explored in a different meta-analysis from Franke et al.In studies that analysed individuals who received multiple shocks, nearly one in five patients received more than five therapies.They found that the combined overall rates of appropriate ICD therapy or SCD were 29.0% and that across the entire cohort, 39.0% of patients received an ICD.As expected, the results showed higher rates of appropriate therapy in those who met secondary prevention indications when compared with primary prevention (22.7% vs. 58.4%)[19].

Imaging Findings
Advances in cardiac imaging, particularly cardiovascular magnetic resonance (CMR) scanning, has given clinicians the ability to improve the quality of care for patients with CS.This has proven important, particularly given the poor sensitivities previously seen in CS patients undergoing endomyocardial biopsy (EMB), for achieving a histological confirmation of a diagnosis [28].However, there remain conflicting data about the clinical impact of a histological diagnosis on prognosis [29,30].An update in this knowledge gap is anticipated.
A recent 2022 study concluded that it was patients with late gadolinium enhancement (LGE) on CMR who most frequently received appropriate therapy and, moreover, that all 10 patients without LGE did not receive appropriate ICD therapy.This association between appropriate therapy and LGE distribution was seen most strongly in those with uptake in the area of the right ventricle (RV), but also in the anterior and inferior walls.The involvement of the RV in CS predicts both a poor outcome as well as high rates of arrhythmia [31][32][33].It is this scarring pattern that likely accounts for the high levels of appropriate therapy seen in this cohort-a theory evidenced in a study from Schuller et al. (OR: 6.73, 95% CI: 2.69-16.8,p < 0.01) [8].
Although there is an association between 18 F-flourodeoxyglucose positron emission tomography (FDG-PET) and Major Adverse Cardiovascular Events (MACEs) [16], to our knowledge there is no research available that connects a positive finding on this imaging modality and appropriate therapy.This is perhaps a gap in the literature that future research should address.

Non-Predictors of Appropriate Therapy
Whilst there were many positive predictive variables associated with appropriate therapy in the above works, Azoulay et al. also identified other factors that were not associated with receiving appropriate therapy in their meta-analysis [15].These were LBBB, RBBB, a positive cardiovascular magnetic resonance (CMR), and syncope.Two of the studies included in this meta-analysis had separately analysed factors that were non-predictors of appropriate therapy.The conclusions of these agreed with those from the above meta-analysis, with the exception of having a lower mean LVEF.This was discussed in Taha et al.'s review [17]: "One found that extent of cardiac involvement on CT-PET and pre-procedure ventricular arrhythmia (VA) burden were non-predictors [13].The other described being >60 years old, New York Heart Association (NYHA) class III/IV, LVEF < 35%, non-sustained VT, paroxysmal atrial fibrillation (AF), QRS interval > 150 ms, QTc interval > 470 ms and concurrent amiodarone therapy as non-predictors of appropriate therapy [14]".

Inappropriate Therapy and Device Complications
Franke et al.'s meta-analysis drew conclusions from 19 studies and showed the rates of inappropriate therapy to be 17.9% [19].Other studies showed the rates of inappropriate therapy to vary widely, anywhere from 2.9% to as high as 30% [8,12,[18][19][20].The lowest of these was from a retrospective cohort analysis of 105 patients where three subjects (2.9%) received inappropriate shocks (having previously received appropriate therapy).All three received an implant for secondary prevention, and were all triggered for atrial fibrillation [20].Kron et al.'s study demonstrated one of the highest rates of inappropriate therapy in 24.3% of their patients, which is again most commonly caused by supraventricular arrhythmias-a theme reflected across many of the abovementioned papers [12].Mohsen et al. was another study in the literature that showed that, although 36.7% patients received appropriate therapy, 63.3% received no appropriate therapy, and 30% of individuals received inappropriate therapy [9].
Although inappropriate therapy is a well-recognised drawback of ICD implantation, there are few studies that have thoroughly investigated inappropriate therapy.Those papers that have done so are listed in Table 2.The most expansive study in the collection described above had a relatively short follow-up period and, as previously highlighted, the small sample size limits and underpowers its ability in statistical analysis.We were unable to find significant data that identify patients who are at higher risk of inappropriate therapy based on their demographic information.

The Association of ICD Device Complications and Risk Factors in Patients with Cardiac Sarcoidosis
With sudden cardiac death accounting for up to 80% of all fatalities in cardiac sarcoidosis, ICD implantation should always be considered in those selected groups in whom it is appropriate [34].However, the implantation of an ICD is a high-cost procedure with its own risk profile.A Danish study in 2014 on nearly 6000 patients with cardiac implantable electronic devices (CIEDs) for a variety of indications (not necessarily CS) estimated that approximately 10% of their patients experienced device complications [35].
The first publication to analyse the association between ICD complications in CS patients was Kron et al. in 2012, showing adverse events in 17.4% of their 235 patients at a median follow-up of 4.2 ± 4.0 years.Over half of these complications were related to lead dislodgement or fracture [12].A later retrospective study on 105 patients showed devicerelated complications at a comparable rate of ~18% at a shorter median follow-up time of 2.8 years [20].Again, complications were most commonly caused by lead malfunction in nearly 1/3 of patients.Both of these papers link this specific complication to the young age and increased level of activity of CS patients when compared to the average ICD patient-a difference of 10 years at the point of implantation [15,36].As commented on by the authors, combining this variable with the high failure rates of the 'Medtronic Sprint Fidelis' leads (particularly in the younger population), which were prominent at the time of the study, may account for these particularly high rates of adverse effects [37].
Many patients have also been treated with immunosuppressive drugs such as prednisolone and methotrexate.We expected this to have an impact on infection rates in the CS cohort.In both of Kron et al.'s and Mathijssen et al.'s studies, infection was the second most common adverse event after lead-related complications at rates of 2.6% and 4.8%, respectively [12,20].A CIED analysis from the previously referenced Danish study showed infection rates of 0.83% in their population [35].

Strengths and Limitations
The researchers in many of the above studies have recognised the need for a global approach for the management of CS, with results being amalgamated from studies across multiple continents.It is worth noting, however, that the lack of multivariable analyses in some of these studies from the meta-analysis prevented independent predictors of ICD appropriate/inappropriate therapy to be elicited.As discussed earlier, there is little data available to reflect the wide-ranging consortium of patients that can be affected by CS, or to further determine which ethnic groups may have a lesser or greater chance of receiving appropriate or inappropriate therapies.Future studies should collect data that span across a broad array of ethnic backgrounds to capture a true reflection of the disease.Another limitation common to several CS meta-analyses is the risk of data overlap.Patientidentifying features have been removed due to lack of consent, and as such, there will certainly be inaccuracies in the results, particularly given the relatively small sample sizes.In addition to this, the retrospective study design lends itself to objectively poor result quality.Finally, although previously discussed as an effective proxy in light of having no better alternative, the use of "appropriate therapy" as a surrogate endpoint is not a true representation of SCD and is generally accepted to be an imperfect tool that likely over-estimates risk.

Discussion
At this relatively early point in our knowledge of CS, most implementable changes found from studies that this review analysed will be targeted at the "ground level".That is, aiding clinicians in their practice to filter patients deemed best to benefit from ICD implantation for CS.The review of the literature made it clear that balancing the higher rates of appropriate therapy with the largely unknown, but likely increased, likelihood of device complication in this younger patient cohort is challenging but important.The crux of ICD usage in CS hinges on scarring patterns and the consequential increased risk of arrhythmia.Although made more challenging by the high heterogeneity of the disease, identifying these high-risk patients and discussing treatment options to achieve a shared but informed decision are essential steps.
We have seen the influence that some of the discussed papers have had on the management of ventricular arrhythmias in those with CS in several iterations of guidance guidelines and recommendations, most recently in the ESC Guidelines from 2022 [26], but also in the 2014 HRS Consensus [24] and the 2017 AHA/ACC/HRS guidelines [25].These recommendations are summarised in Table 3.More widely speaking, this paper aimed to provide a stepping stone for further research to support the revision of these guidelines if necessary-especially relating to the information we amalgamated about the drawbacks of ICD implantation.Although a consequence of the innate rarity of the condition, the low cohort size seen in several of the previous studies lends itself to statistically underpowered analyses and therefore limits our ability to draw stronger conclusions.This highlights the need for larger-scale trials that are prospective in nature to truly characterise the utility of ICDs in CS.In particular, there is a need for more data surrounding both device complications and inappropriate therapy.These trials should compare a variety of different variables, examples of which may include analysing ICD therapies in those treated with and without immunosuppression, or perhaps looking at outcomes in patients with sarcoid isolated from the heart versus those with systemic manifestations of the disease.
Inappropriate therapy will likely remain an unavoidable complication of ICD therapy.However, through the improvement of VA therapy detection systems and supraventricular tachycardia (SVT) discrimination algorithms, the rates of inappropriate therapy will be reduced, and these will make ICD implantation a more viable option for a greater proportion of patients.We envisage machine learning to play a pivotal role in this through using AI-based pattern recognition on an ever-growing collection of patient datasets from devices and constantly refining the 'correct rhythm' used to deliver a shock.We have already seen some advancements made in this field recently [38], and we expect this to grow further and positively impact those with CS as well.It was evident that in many of the previously reviewed studies, those with CS experienced more frequent ICD therapies than other cohorts.In comparison to the results of the "Sudden Cardiac Death In Heart Failure Trial" (SCD-HeFT), the rates of ICD therapy were more than double those of the ~5% appropriate shocks delivered per annum [39].
Clinicians and cardiac physiologists should be wary of the potential flaws of ICD insertion, namely, inappropriate shocks from rhythms, such as atrial fibrillation being recognised as VT, and they should closely monitor the settings of ICDs in patients with CS.This is of particular importance in those with CS given the inflammatory nature of the condition and its high association with ventricular arrhythmia.As such, the settings (threshold and impedance) should be adequately adjusted so that inappropriate antitachycardia pacing and shock are avoided.
Increasing clinician awareness alongside more accurate and improved cardiovascular imaging and diagnostic testing has led to higher numbers of cases being diagnosed each year.We expect this trend to continue to increase in the coming years.We also envision the formulation of a more definitive diagnostic criteria that can be more specifically applied to a variety of ethnic groups.

Conclusions
We showed that those with an ICD implanted for CS receive comparable rates of appropriate therapy but are at higher risk of complication than the average ICD patient.Young age, male sex, and ventricular characteristics such as low LVEF have been frequently found to be significant factors in predicting appropriate therapy.There was little significant information available globally relating to device complications in those with CS.However, as discussed, the benefits of ICD therapy may be double-edged in nature, with potential for extensive side effect profiles and high rates of inappropriate therapy.
Although the conclusions drawn by the authors from the aforementioned studies are based largely off small datasets, the research groups should be commended on their contributions in such a rare condition and their help in improving the quality of care delivered.Although a niche field, cardiac sarcoidosis proves to be an interesting research area with a broad array of research opportunities for those in whom it piques interest.

Table 1 .
Overview of reviewed sources relating to appropriate therapy in CS.

Table 2 .
Overview of the reviewed sources relating to inappropriate therapy and device complications in CS.