Clinical significance of atrial LVAs

A growing number of studies have highlighted the prognostic significance of LVAs in AF. In a mixed cohort of AF patients undergoing ablation, areas of very low voltage (<0.05mV) were associated with a three-fold increased likelihood of recurrent AF during follow-up (78). Oshima and colleagues similarly reported a negative influence of LVAs in freedom from AF following PVI in patients with non-paroxysmal AF (79). In their study, LVAs were defined as regions with bipolar voltage <0.5mV, and determined LVAs covering more than 24% of the LA surface as a threshold for discriminating the likelihood of recurrent AF following ablation, albeit with a sensitivity of 52% and specificity of 78%. In a separate study also utilizing 0.5mV to define LVAs, the presence of LVAs was an independent predictor for AF recurrence following ablation (80). Interestingly, patients with PsAF and no LVAs had similar rates of freedom from AF as individuals with PAF. Furthermore, studies limited to subjects with PAF have similarly correlated LVAs with AF recurrence following ablation (81,82).
Mechanistic insights have further corroborated the potential role of LVAs in the pathogenesis of AF. Approximately three-quarters of high dominant-frequency sites, proposed as foci of rotational activation (83), were located within LVAs or their border zones in a cohort of 70 patients with non-paroxysmal AF (79). Left atrial bipolar voltage was also shown to correlate with conduction velocity (84). Miyamoto et al. reported conduction slowing in regions of the left atrial with bipolar voltage <0.5mV and these regions were significantly more likely to harbor CFAEs (85). Chauhan and colleagues reported co-localization of CFAEs to regions of low voltage, with the degree of fractionation inversely proportional to voltage (86). Potential left atrial non-pulmonary vein triggers for AF appear more common in PsAF and show a predilection for LVAs (87).
Temporal changes in the size of LVAs have been reported, mirroring the progressive fibrotic remodeling seen in animal models that underpin the increasing propensity for AF to sustain with time. AF type, either paroxysmal or persistent, was independently associated with LVAs on multivariable analysis in an early study of VGA (18). In the study by Hindricks’ group, LVAs were twice as likely to be observed in PsAF than in PAF (25). Yagishita et al. also reported higher prevalence of LVAs in patients with non-paroxysmal forms of AF compared to those with PAF, with LVAs covering a greater overall area of the atrium and being more diffusely spread in those with PsAF (88).
However, the frequency and burden of LVAs observed in a number of studies is at odds with these reports and raises several of important issues. For example, Huang et al. measured left atrial voltage in patients with either paroxysmal or persistent AF and found no difference in the size of LVAs between the groups (89). Additionally, the extent of LVAs at baseline did not predict AF recurrence following PVI. In patients with non-paroxysmal AF, the degree of LVAs did not correlate with AF duration (90). Birnie and colleagues used high-resolution mapping to characterize the burden of LVAs in patients with AF (91). No differences in the burden of LVAs were observed between patients with paroxysmal or persistent AF, with only age and LA size, being identified as predictors of LVAs on multivariable analysis. The study also reported significant variability in the presence of LVAs across the cohort, and such differences are also evident across the studies where left atrial voltage has been examined.
In an early study of VGA, LVAs were present in 10% of patients with PAF and 35% in those with PsAF (18). Data from the study by Kircher et al. were broadly similar with LVAs being detected in 29% overall, with 18% of patients with PAF and 41% of patients with PsAF (25). In contrast, other studies have reported substantially higher prevalence of LVAs, with nearly two-thirds of patients with PAF exhibiting LVAs (82,92) and over 80% of patients with PsAF (79,89).
Differences in study cohorts beyond AF classification may have contributed to such variation. However, reductions in atrial bipolar voltage have been purported to be a surrogate for the adverse atrial remodeling underpinning AF persistence and as such the lack of a difference in the burden of LVAs between paroxysmal and persistent AF within the same study, where the methods utilized for voltage mapping would be consistent, is intriguing. Recent analyses of AF burden using continuous ECG monitoring have highlighted that these clinical classifications of AF type only loosely associate with AF burden, with significant overlap in the time spent in AF between patients with paroxysmal and persistent AF (93,94). This variability in AF burden may be reflected in the extent of LVAs, and further endorse the need to more patient-specific approaches to ablation.
While a small proportion of studies report presence of LVAs in the majority of patients with PsAF, in most studies LVAs are recorded in less than half of the study cohort. The absence of LVAs in such cases may signify the absence of advanced remodeling where a standalone PVI strategy may suffice. However, the absence of LVAs in such a sizeable fraction of a cohort with sustained AF may conversely question the sensitivity of LVAs, and/or the techniques utilized in defining them, in characterizing the atrial substrate. Indeed, where the relationship of conduction and voltage have been studied, not all regions of low voltage were associated with conduction slowing and conversely some regions with abnormal conduction properties displayed normal bipolar voltages (84). Furthermore, approximately 30% of sites with high dominant frequency activation are found in regions of normal voltage (79).
These issues highlight the challenges in delineating the atrial substrate and perhaps representing this through voltage mapping in isolation, particularly with the approaches utilized up to now, may not do so with sufficient sensitivity and specificity. While ablation targeting LVAs as an adjunct to PVI gains traction, it is essential to note the significant paucity of data on precisely how voltage measurements relate to adverse atrial remodeling in AF. In particular, low voltage is generally considered to represent native atrial fibrosis, however at present there is no histological data to support this assertion. The myocardial fiber arrangement in the atrium is highly complex with regional heterogeneity in fiber orientation and tissue depth. Beyond this, the process of fibrosis has significant variability in the pattern of deposition and degree of transmurality, and how this affects voltage measurements is unclear. Added to this milieu is the multi-faceted nature of the atrial substrate as described above. It is unlikely that amongst the array of remodeling phenomena that have been documented in AF, none other than fibrosis alters the recorded tissue voltage. Therefore, while favorable outcomes have been reported in VGA studies, many questions remain unanswered in assessing atrial voltage.