APPLE Score in Atrial Fibrillation: A Simple Solution to Predict a Complex Process?

Ramanathan Parameswaran; Jonathan M. Kalman; Irina Savelieva


Europace. 2019;21(1):1-2. 

Left atrial fibrosis, characterized by low-voltage areas (LVA), suggests adverse atrial remodelling and negatively impacts ablation outcomes in atrial fibrillation (AF).[1]Targeting these LVA for ablation has been investigated with mixed results but a recent meta-analysis[2] showed that voltage-guided substrate modification by targeting LVA in addition to pulmonary vein isolation is more effective, safer, and holds a lower pro-arrhythmic potential than conventional ablation approaches. While there is a continued development of methods for characterizing the arrhythmia substrate, including atrial cardiac magnetic resonance imaging (MRI) and invasive voltage mapping, recent studies have also investigated a number of non-invasive scoring systems which include clinical variables and biomarkers to predict ablation outcomes. In the quest to achieve optimal ablation outcomes, these tools have the potential to inform patient risk assessment and help guide selection for catheter ablation on an increasingly individualistic basis.

The CHADS2 and the CHA2DS2VASc scores which are validated for risk-stratifying ischaemic stroke risk and vascular events in AF, have also been shown to modestly predict recurrence of AF after ablation particularly in patients with a score ≥2.[3] Several other clinical scores like the ALARMEc[4] and BASE-AF2[5] have also been proposed for predicting AF recurrences post-ablation but carry limitations in their definitions, including post-ablation variables like early recurrences and lack validation in external cohorts. More recently, the APPLE score[6] [1 point for Age > 65 years, Persistent AF, imPaired eGFR (<60 mL/min/1.73 m2), Left atrial diameter ≥43 mm, left ventricular Ejection fraction < 50%] was investigated and shown to be superior to the CHADS2 and CHA2DS2VASc scores in predicting AF recurrence post-ablation. The simplicity and inclusion of well-established parameters related to AF recurrence confers the practical advantages to implement this score in a wide variety of populations undergoing catheter ablation for AF.

Several biomarkers such as B-type natriuretic peptide (BNP) and its N-terminal prohormone (NT-proBNP),[7] C-reactive protein,[8] uric acid,[9] and high-sensitivity troponin T[10] have also been evaluated as predictors of post-ablation AF relapse. These serum biomarkers offer the potential for use as outcome stratifiers as they are easily accessible in every patient at baseline.

In this issue of EP-Europace, Kornej et al.[11] have investigated the association of specific biomarkers [atrial natriuretic peptide (ANP), BNP, vascular cell adhesion protein 1 (VCAM-1) and L-selectin] and the novel APPLE score to predict LVA in patients with AF undergoing catheter ablation. In the current study cohort of 214 consecutive patients undergoing first radiofrequency catheter ablation, 26% had LVA. Foreseeably, patients differed significantly in their median APPLE scores (3 vs. 2, P < 0.001), their biomarkers; NT-proANP (22 vs. 16 ng/mL, P = 0.018), NT-proBNP (2247 vs. 2007 pg/mL, P = 0.007), and VCAM-1 (1844 vs. 1478 ng/mL, P = 0.006) and were more frequently female (59% vs. 35%) in comparison to the patients who did not have LVA. Furthermore, the levels of these biomarkers increased in tandem with each APPLE point increment. Overall, there were 42% patients with APPLE ≥3. In patients with APPLE ≥3, the levels of NT-proANP (median 21 vs. 15 ng/mL, P < 0.001), NT-proBNP (median 2260 vs. 1991 pg/mL, P < 0.001), and VCAM-1 (median 1717 vs. 1478 ng/mL, P = 0.029) were significantly higher than in patients with APPLE < 3, while L-selectin levels were significantly lower in patients with APPLE ≥3 (median 429 vs. 471 ng/mL, P = 0.011). In the univariable analysis, the APPLE score [odds ratio (OR) 1.921, 95% confidence interval (95% CI) 1.453–2.538; P < 0.001], female gender (OR 2.283, 95% CI 1.280–4.071; P = 0.005), and NT-proANP (OR 1.031, 95% CI 1.008–1.054; P = 0.007) were significant predictors for LVA. On multivariate analysis, APPLE score (OR 1.807; P < 0.001) and female gender (OR 3.417, P = 0.009) were the only significant predictors of LVA. Indeed, addition of biomarkers to the APPLE score did not improve the predictive ability for LVA.

All the variables used in the APPLE score are commonly linked with AF and structural and functional remodelling of the atria. Studies of electrical remodelling in aging,[12] heart failure,[13] and hypertension[14] have all shown an increase in low-voltage zones associated with conduction slowing and complex fractionated signals. The final common pathway in the mechanistic role of these varied comorbidities is usually reactive or replacement fibrosis from chronic stretch and oxidative stress.[15] Although data evaluating the impact of female sex on remodelling have been mixed, a recent study using high-density mapping observed low-voltage zones and conduction slowing to be more prevalent in males compared with females.[16]

Sex differences in the clinical aspects of AF are well recognized. Among individuals with AF, women are more likely than men to experience symptoms associated with worse quality of life and have increased risk of complications such as stroke compared with men. It might be hypothesized that this increased stroke risk also reflects a more advanced underlying atriopathy. The mechanisms underlying sex-related differences in the AF substrate are poorly understood. Theories range from differences in K+ and Ca+ channels to a higher propensity for atrial fibrosis as a consequence of elevated plasma concentrations of the inflammatory marker C-reactive protein and fibroblast growth factor-23.[17] Both factors have been associated with increased risk of AF and might contribute to increased atrial fibrosis in women.[18]

The current study builds on prior work by the same group using the previously proposed DR-FLASH score[19] [based on diabetes mellitus, renal dysfunction, persistent form of AF, left atrial (LA) diameter >45 mm, age >65 years, female sex, and hypertension] for predicting LA arrhythmogenic substrate based on voltage analysis. The c-statistic was impressively high for both scores: 0.767 in the validation cohort for DR-FLASH and 0.711 for APPLE. Although the results are similar, this study also demonstrates that biomarkers increase with each increment in the APPLE score. However, the use of biomarkers in risk stratification, particularly in the context of a specific electrophysiological endpoint, is impeded by the absence of known cut-off points prompting the inclusion of median values in the risk score. While shown to be associated with a two-fold higher rates of LVA, female gender has not been included in the APPLE score. Whether incorporating this variable, if proven significant in a larger cohort, in the (modified) APPLE score may result in a greater predictive power compared with DR-FLASH remains uncertain.

One of the key issues raised by these studies is the uncertain gold standard for detection of abnormal atrial myocardium and regional fibrosis. In the current study, the authors used points collected both from the large antenna-like bipole of the ablation catheter distal pair and also from the smaller ring electrodes with 2 mm spacing of a circumferential catheter. These two methodologies have been shown to produce quite different results.[20] Cut-off points for abnormal tissue have never been adequately validated in the atrium and the values of 0.5 mV for low voltage and 0.2 mV for scar used in this study, while certainly not without precedent, may not adequately reflect subtle regional abnormalities. Recent work using ultra-high resolution mapping has demonstrated that atrial substrate abnormalities may be highly regionalized and therefore require this level of sophisticated mapping for accurate identification.[21] Furthermore, it is not clear that sinus rhythm maps can accurately identify early substrate given the impact of rate and directionality on voltage map results. In this context, the ability of the APPLE score to predict substrate observed on MRI is also worthy of evaluation.

The current study represents a significant advance in the search for non-invasive markers of more advanced atrial remodelling. As such it represents an important step towards identifying those persistent AF patients least likely to benefit from catheter ablation.