De Novo Versus Upgrade Cardiac Resynchronization Therapy: A Different Patient Population and Outcome?

This article refers to ‘Long-term outcome after upgrade to cardiac resynchronization therapy: A propensity score-matched analysis’ by S. Trenson et al., published in this issue on pages 511–520. Cardiac resynchronization therapy (CRT) is a well-established treatment, it reduces morbidity and mortality in patients with heart failure with reduced ejection fraction (HFrEF) and wide QRS.1 In de novo CRT patients, plenty of data confirmed the mid- and long-term beneficial outcome, whereas evidence for CRT upgrade in those who have been already implanted a conventional pacemaker (PM) or implantable cardioverter defibrillator (ICD) and have a higher proportion of right ventricular pacing (RVP) is scarce. The randomized controlled trial (RCT) BUDAPEST CRT Upgrade came out this year, which proved the efficacy of CRT upgrade as compared to ICD alone with a robust risk reduction on the primary endpoint in HFrEF patients with >20% of RVP during a median follow-up time of 12.4 months.2 This was the first RCT on the field to investigate the efficacy on hard outcomes, which served as a true control group without an activated left ventricular lead versus biventricular pacing after a CRT upgrade. Previously, CRT upgrade patients were compared with de novo CRTs in meta-analyses3 and observational studies,4-6 but long-term outcome data were still limited and controversial. In a previous meta-analysis with predominantly smaller single-centre observational studies, all-cause mortality and clinical benefit was comparable in de novo versus upgrade CRT patients.3 In contrast, in a prior multicentre prospective registry with a mean follow-up of 37 ± 28 months, the risk of mortality was 60% higher after upgrade as compared to de novo CRT with defibrillator implantations even after multivariate analysis or after propensity score matching (PSM).6 However, it is of primary importance to conduct such analyses using real-world data when RCTs are limited, especially because different outcomes were observed regardless of the extended populations or advanced statistical analyses. In this issue of the Journal, Trenson and colleagues aimed to investigate the long-term outcome of de novo CRT versus upgrade CRT patients on one of the largest populations, analysing 2276 patients (CRT upgrade n = 605 [26.6%]) retrospectively from a multicentre database.5 The primary endpoint was all-cause mortality, heart transplantation or ventricular assist device implantation. During a mean follow-up of 3.6 ± 2.7 years, the incidence of the primary endpoint was significantly higher in the CRT upgrade group with 10.8%, having a 35% increased risk for mortality in these patients. Also, this poorer outcome remained significant even after PSM. Interestingly, the other, high-scale—almost identical—observational study with 2524 subjects (CRT upgrade n = 547 [22%]) with the same primary composite endpoint, conducted by Schwertner et al.4 in 2021, described different findings on a similarly long, 3.7 year follow-up. Despite the similar inclusion period and advanced statistics, the univariate risk was 41% higher in the upgrade group, but these outcome differences diminished between the CRT upgrade versus de novo CRT patients after multivariate analysis or after PSM with more than 500 pair of patients. The long-term outcome differences are most likely associated with the characteristics of the investigated population, selection bias and residual imbalances or unknown confounders between the two groups even after using advanced statistical methods. In the analysis by Trenson, patient characteristics were pretty similar to the Schwertner's CRT patient cohort; however, a few differences should be highlighted. First, in the cohort of Trenson and colleagues the mean age of the two subgroups of patients did not differ significantly, having relatively older patients in the de novo CRT group with a median age of 70.2 (61.9–76.7) years, and younger ones in the upgrade group with a median age of 70.6 (61.8–76.8) years as compared to literature data,7 which reported an average age of 65 and 72 years in general, respectively. Of course, it can be explained by the elapsed time from PM or ICD implantation, as the development of pacing-induced cardiomyopathy (PICM) can be prolonged. Additionally, the proportion of patients with current or history of atrial fibrillation (AF) was very low, especially in CRT upgrade patients with 21%, whereas in the study by Schwertner or in the BUDAPEST CRT Upgrade trial, this rate reached 50%. Previously, the European Society of Cardiology (ESC) CRT Survey II7 and the BUDAPEST CRT Upgrade trial8 comparisons with previous de novo CRT studies also revealed that there are substantial differences in the clinical characteristics of CRT upgrade versus de novo subjects. HFrEF patients with a higher proportion of RVP and symptoms referred for a CRT upgrade are significantly older, the proportion of male patients reaches 80–90% and have a high burden of comorbidities (e.g. high rate of AF, frequent ischaemic heart disease, renal impairment), and an advanced stage heart failure. In the Trenson study, beside age and lower incidence of AF, the use of CRT-D devices was particularly frequent, especially in the CRT upgrade group, where more than two-thirds of patients got a CRT-D upgrade. In the ESC Survey II data, two-thirds of patients are referred with a PM and upgraded to a CRT-P system, therefore these rates may also influence the long-term outcome.7 Notably, CRT upgrade patients mainly show a substantial benefit from left ventricular lead implantation with echocardiographic reverse remodelling, but the subsequent, long-term need for a backup ICD should be clarified. However, based on the BUDAPEST CRT Upgrade trial, in those with remarkable echocardiographic improvement, the incidence of major arrhythmias decreased.2 The right ventricular pacing burden is another, one of the most relevant parameters, that should be considered, when mid- and long-term responses are investigated. Unfortunately, pacing rates were not available in this retrospective database. It is known that patients with higher RVP rate can have a better response.9 Moreover, the presence of mechanical dyssynchrony in PICM patients seems to be a predictive factor for a more beneficial outcome.10 In the Trenson study, despite this higher risk of long-term outcome in the CRT upgrade group, it did not associate with worse clinical response, which also supports the fact that the differences between the two groups may come from the characteristics of the investigated cohort and non-cardiovascular events derived from comorbidities. It needs to be recognized that a few other relevant factors can also influence the long-term outcome, such as the position of the right and left ventricular leads,11, 12 which were not reported in these studies. The potential differences between PM and ICD patients may be related to the fact that ICD leads are more frequently implanted in a lower septal-apical position, whereas RVP leads to mid/higher septal regions or outflow tract.12 Moreover, these questions must be re-evaluated when physiological pacing systems can be added and investigated. Additionally, whether the differences between the two observational studies by Trenson and Schwertner are also related to regional differences or statistical disparities, is unknown. Altogether such real-world, high-scale observational data are essential when long-term evidence is limited on hard outcomes from RCTs. Therefore, future investigations are warranted to clarify these questions and optimal device selection for CRT upgrade patients. However, the clinical decision-making will rely on the robust data provided from the RCT BUDAPEST CRT Upgrade, which proved the substantial benefit of CRT upgrade in HFrEF patients. Therefore, in everyday clinical practice, close follow-up and forced screening of PM/ICD patients with high RVP burden is crucial to identify future CRT upgrade candidates. Conflict of interest: none declared.