Retrograde Coronary Venous Ethanol Infusion for Ablation of Refractory Ventricular Tachycardia

Abstract

Background

Radiofrequency ablation (RFA) of ventricular tachycardia (VT) can fail due to inaccessibility to the VT substrate. Trans-arterial coronary ethanol ablation (TCEA) can be effective, but entails arterial instrumentation risk. We hypothesized that retrograde coronary venous ethanol ablation (RCVEA) can be an alternative bail-out approach to failed VT RFA.

Methods and Results

Out of 334 consecutive patients undergoing VT/PVC ablation, seven patients underwent RCVEA. Six of seven patients had failed RFA attempts (including epicardial in 3). Coronary venogram-guided venous mapping was performed using a 4F quadripolar catheter or an alligator-clip-connected angioplasty wire. Targeted veins included those with early pre-systolic potentials and pace-maps matching VT/PVC. An angioplasty balloon (1.5-2 × 6 mm) was used to deliver 1-4 cc of 98% ethanol into a septal branch of the anterior interventricular vein (AIV) in 5 patients with LV summit VT, a septal branch of the middle cardiac vein, and a postero-lateral coronary vein (n=1 each). The clinical VT was successfully ablated acutely in all patients. There were no complications of RCVEA, but one patient developed pericardial and pleural effusion attributed to pericardial instrumentation. On follow-up of 590 ±722 days, VT recurred in 4/7 patients, three of whom were successfully re-ablated with RFA.

Conclusions

RCVEA is safe and feasible as a bail-out approach to failed VT RFA, particularly those originating from the LV summit.

Introduction

Radiofrequency catheter ablation (RFA) is the standard of care for ablation of drug-refractory ventricular tachycardia (VT), particularly in the setting of ischemic heart disease.^1-3 Ablation success is far from uniform, in part due to technical challenges reaching the VT substrate with current available technologies, since all require some degree of tissue contact with the targeted tissue. Epicardial instrumentation⁴ may allow reaching VT substrates refractory to endovascular approaches,^{5, 6} but remains difficult in patients who have undergone prior cardiac surgery,⁷ and it is not always helpful, since many ventricular arrhythmias derive from deep intramural origins or in proximity to coronary vessels^8-10. This is particularly true for VT originating from the LV summit, where an intramural origin, proximity to coronary vessels and inaccessibility to the epicardial approach limit radiofrequency ablation success.¹¹

Transarterial coronary ethanol ablation (TCEA) as an alternative treatment modality has been reported extensively in the literature.^12-16 TCEA has become a modality of last resort in the treatment of VT not amenable to alternate contact-based ablation.^{12, 16} TCEA is limited by the risks of arterial instrumentation, the dependence on a feasible arterial anatomy –commonly affected by the ischemic disease that led to the VT substrate, the risk of unintended collateral damage and the logistic challenges of requiring interventional cardiology support. To overcome these limitations, retrograde coronary venous ethanol ablation (RCVEA) has been described as an alternative to the arterial approach. Its initial application in canines showed feasibility and effective myocardial ablation.¹⁷ We have reported feasibility in humans and acute procedural success of RCVEA in two cases¹⁸. The venous approach is uniquely suited for detailed activation mapping using an angioplasty wire. Here we report its combination with venous wire mapping in a variety of VT substrates, particularly the LV summit, and report its chronic outcomes.

Methods

Data collection

A total of 7 patients were included. All were counseled about the unconventional nature of venous ethanol ablation, and gave informed consent to the intervention. Clinical data collection was performed under an IRB-approved protocol. Initial data relating two procedures carried out in 2011 (n=2) were collected retrospectively from patient charts. Acute outcomes pertaining to the latter have previously been reported.¹⁸ The subsequent 5 patients were included from a cohort of consecutive patients undergoing VT/PVC ablation studied prospectively from January 2012 through February 2016. This data included medical history, procedural reports and major post-procedural and follow-up events.

Procedural approach: vein mapping

In all procedures, efforts were initially made to localize VT substrate within an area amenable to RFA. Electroanatomical maps (EAM) were constructed by using 3D mapping systems (NavX, St Jude Medical, St Paul, MN, n=3) or Carto3 (Biosense-Webster, Diamond Bar, CA, n=4)). Mapping strategies included substrate maps to localize low-bipolar voltage areas in the presence of structural heart disease, activation maps and pace-maps. Access to the epicardial space via a subxiphoid anterior puncture was undertaken when suitable for epicardial mapping and ablation (n=3).

Coronary vein mapping was performed by advancing a sheath in the coronary sinus via the right femoral vein (Preface, Biosense-Webster, Diamond Bar, CA) or via the right internal jugular vein (CPS sheath, St Jude Medical, Sylmar, CA). Coronary venograms were performed. A multipolar catheter was inserted in the coronary sinus and selected ventricular branches (4F quadripolar IBI, St Jude, or Deca-Nav, Biosense-Webster). Mapping and pacing from small coronary veins was performed by advancing an angioplasty wire (BMW 0.014”, Abbott), connected to an alligator clip in a unipolar configuration with reference electrode as a needle inserted in the thigh skin. This approach led to significantly reduced noise compared to using Wilson's central terminal or an indifferent electrode in the inferior vena cava,¹⁹ and provided exclusively local signals, compared to using a neighboring electrode. Selective wire cannulation of different targeted veins was achieved by introducing a LIMA angioplasty guide catheter and torqueing it in the desired direction, or simply by guiding the angioplasty wire with the help of a torqueing device. In order to obtain unipolar signals from selective portions of the targeted vein, the angioplasty balloon was advanced over the wire in order to cover it except for the most distal 3-5 mm, which acted as the active electrode. RCVEA was considered when: 1) RFA failed at the best endocardial sites as guided by the earliest activation or best pace-mapping (PM); 2) when feasible, epicardial RFA failed or was deemed not indicated due to proximity to coronary arteries or due to presence of the earliest activation site at a broad area; and 3) when optimal PM and/or earliest activation was obtained from within a coronary vein.

The bulk of our experience targeted LV summit VT, which we targeted by mapping septal branches of the anterior interventricular vein (AIV; n=5). Other targets included a postero-lateral coronary vein (n=1) or the middle cardiac vein (n=1).

Prior to ethanol infusion, the presence of vein-to-vein collaterals was assessed by inflating the balloon (1.5 - 2 mm by 6 mm, depending on vein caliber) to achieve venous occlusion and injecting contrast. Ethanol infusion was injected only in the absence of vein-to-vein collaterals in order to deliver ethanol to the capillaries and the myocardium and avoid vein-to-vein shunting. Once a vein was selected, the balloon was inflated to 4-8 ATM (aiming to occlude the selected vein). In the first 4 cases 1 cc of 98% ethanol were infused over two minutes. In the subsequent 3 cases, 2-4 injections were administered. Flushing with normal saline was performed after ethanol. Balloon occlusion time ranged 2-8 minutes. An infusion of cold saline was attempted in one case, but it failed to terminate the VT –despite an eventual ethanol success- and cold saline was abandoned for subsequent cases.

Procedural success

In PVC ablation (n=3), ablation success was defined as the cessation of spontaneous ectopy after 20 seconds of ethanol infusion. In VT ablation (n=4), attempts were made to induce the clinical tachycardia before and after ablation, and the procedure was considered successful if induction failed post-ablation. The EP testing protocol used 400-ms and 600-ms drive trains followed by one to three ventricular extrastimuli that were 2 ms in duration at twice the diastolic threshold. Extrastimuli were decremented down to a coupling interval no shorter than 200 ms. If this protocol proved ineffective, rapid burst pacing up to a cycle length of 280 ms was utilized instead.

Clinical Follow-Up

All patients were hospitalized following the ablation procedure. They underwent continuous telemetry monitoring, 12-lead electrocardiography and 24-hour ambulatory electrocardiography to document absence of VT/PVC recurrence prior to discharge. Echocardiograms were obtained if clinically indicated. Patients were later seen in the clinic at one week and one month post-procedure. They were followed on a biannual basis thereafter. Four patients had an implanted implantable cardioverter defibrillator (ICD), and so interrogation and shock history were retrieved during follow-up. For the remaining patients, 12-lead electrocardiography and 24-hour ambulatory electrocardiographic monitoring were performed upon follow-up.

Statistical analysis

Gaussian continuous variables are reported as mean ± standard deviation and non-Gaussian variables as Median [minimum-maximum]. Qualitative findings were described as numbers and percentages. Analyses were performed using Sigmastat (version 3.11) and Stata software (version 13). Statistical significance could not be attributed owing to the inherent limitation imposed by the small sample size.

Results

Baseline Characteristics

A total of 334 consecutive patients underwent VT/PVC ablation in a single tertiary center between January 2011 and February 2016. Of these, seven cases were considered suitable for RCVEA, and underwent the intervention accordingly (Table 1). Three patients underwent ablation for PVC, and four for VT. Six patients were male, and mean age at procedure time was 62.2±12.6 years. Patient clinical characteristics and comorbidities are detailed in Table 1. Of particular relevance, six of the seven patients had undergone previous VT ablation with recurrence, including epicardial ablation in 3 cases (1.8 ±1.0 previous ablations). In one patient with LV summit VT, an attempt to advance an ablation catheter into the AIV failed due to the small vein caliber, which made it impossible to reach the targeted site. Two patients had previously been implanted with a left ventricular assist device (LVAD) as well (Figures 1 and 2), which precluded epicardial access. One LVAD patient had recurrent VT in the setting of ischemic cardiomyopathy. The second LVAD patient had incessant VT in the setting of a nonischemic cardiomyopathy, which recurred after LVAD implant despite maximum tolerated doses of pharmacologic therapy including amiodarone and mexiletine. Four patients were diagnosed with non-ischemic cardiomyopathy (NICM) and had an ICD implanted, while one was diagnosed with tachycardia-induced cardiomyopathy (TICM). Mean left ventricular ejection fraction was 41.1 ±20.6%.

Table 1.

Patient demographics and clinical outcomes

Patient	Age	Comorbidities	LVEF (%)	Previous Ablations	AICD	Acute success	Follow-up interval (days)	Procedure time (hours)	Fluoroscopic Exposure time (mins)	Recurrence interval (days)	Complications
1	70	*HTN, DL, AF, CHF, mechanical aortic valve	35-39	3	Yes	Yes	1552	5:11	70.5	169	None
2	65	*HTN, DL, DM II	57	3	Yes	Yes	1705	5:51	74.1	258	Pericarial effusion; Ethanol-induced myocardial injury
3	72	*HTN, DM II, CHF s/p LVAD	<20	0	Yes	Yes	206	4:24	0.55	N/A	None
4	58	*HTN, DL, AF, DMII, hypothyroid, CHF s/p LVAD	10	1	Yes	Yes	207	3:37	27.82	N/A	None
5	42	*HTN, DL, DM II, CHF	40	2	No	Yes	404	5:26	47.03	224	None
6	52	*HTN, DL, GERD, COPD, OSA, CAD	60-65	1	No	Yes	40	3:34	42.15	N/A	None
7	77	HTN, CAD	60-65	1	No	Yes	19	3:49	68.27	1	None

^*HTN: hypertension, DL: dyslipidemia, DM II: diabetes mellitus II, CHF: congestive heart failure, AF: atrial fibrillation, CAD: coronary artery disease, COPD: chronic obstructive pulmonary disease, GERD: gastro-esophageal reflux disease, OSA: obstructive sleep apnea, LVAD: left ventricular assist device.

Figure 1.

Mapping of septal branches of the anterior interventricular vein (AIV) in a patient with LV summit PVCs, and successful venous ethanol ablation. A. Three-dimensional activation map (Carto, Biosense-Webster, Sylmar, CA) of the right ventricle (RV), coronary sinus (CS, using a decapolar –deca- catheter), and AIV. The aortic cusps and LV outflow activation map is included in B. C. Venogram of the distal CS demonstrating 3 septal branches denoted s1, s2 and s3, at the origin of the AIV. D. Surface ECG of the PVC, including signals recorded from the decapolar catheter, and pace-maps (% matching with PVC) obtained with unipolar pacing via an angioplasty wire in each of the 3 septal branches. S1 has a perfect pace-map match. E through G, angioplasty wire position and unipolar signals obtained from septal veins s1 through s3, respectively. Signal from s1 precedes QRS by 36 ms. H, balloon occlusion of s1 and selective s1 venogram showing myocardial staining (white arrows). I and J, intracardiac echocardiogram images of the LV outflow tract before and after ethanol, showing increased echogenicity of the ethanol-injected myocardium (white arrows). K, ECG before and after ethanol, showing elimination of PVCs.

Figure 2.

Septal vein mapping and ethanol ablation of LV summit PVCs. A, Activation maps (Carto) of the RV, LV, CS, and AIV, showing earliest site in the AIV. B, ECG and Pace-maps of the PVCs (including AIV signals, recorded from the distal CS, CSd), obtained from 3 different septal branches: s1, s2, s3. A near-perfect pace-map is obtained from s2. C, D, and E show angioplasty wire cannulation of s1, s2, and s3 and their respective unipolar signals, which were earliest in s2. F and G, myocardial staining during selective contrast injection in s2, shown in right (F) and left (G) anterior oblique (AO) projections. H, increased echo-density in the basal LV septum after ethanol injection. I, ECG before and after ethanol injection, with elimination of extrasystoles. This case had not received any previous ablation and RFA of the AIV was not feasible due to its small size.

LV summit VT: Ethanol infusion in septal veins

Six cases had incessant PVC/VT ablated via ethanol infusion in septal veins. Five had PVCs arising from the LV summit (Figures 1 through 4) –one has been previously reported.¹⁸ In all cases, extensive endocardial mapping in both sides of the septum was initially performed, as well as mapping from the great cardiac vein and AIV with a multipolar catheter. In 3/7 cases, RFA had been previously delivered via the RV endocardium and LV endocardium without success. Epicardial mapping –performed in 2 cases with LV summit PVC/VT (one each) revealed earliest activation over a broad area of the LV anterior base. In one case, no suitable ablation targets were found in either RV or LV and the ablation was performed exclusively with ethanol (Figure 2).

Figure 4.

Mapping and ablation of LV summit VT in the context of non-ischemic cardiomyopathy. A and B, activation maps of VT, obtained with a PentaRay catheter (PR, Biosense Webster) in the LV and a decapolar catheter in the CS. The distal CS signals (CSd in panel D) are earliest. C, absence of endocardial low voltage areas in either LV or RV. E, CS cannulation with a decapolar. F, CS venogram. G, septal vein wire cannulation and unipolar wire signals during paced rhythm, showing late, post QRS activations. H, Pace-map from the angioplasty wire from a septal vein branch. I and J, septal vein balloon cannulation for ethanol infusion showing perfect QRS match.

After mapping the great cardiac vein and AIV with a multipolar catheter showed a promising signal from the AIV, selective AIV venograms were performed to identify septal branches. Then, detailed unipolar septal vein mapping was performed with the angioplasty wire, which could be used to selectively obtain unipolar signals and pace-maps from the different septal branches (Figures 1-3). The septal branch with perfect pacemap match and earliest unipolar signals was selected for ethanol infusion. Figure 1 shows septal vein mapping in a patient with LV summit PVCs in whom the first septal branch was chosen to deliver ethanol after mapping and pacing with the angioplasty wire in other veins. In Figure 2, three septal branches were mapped as well, and the second one was shown to have earliest unipolar signals with perfect pace-maps. In both Figure 1 and 2, the septal branches were relatively short and the angioplasty wire only advanced ~2 cm deep into the septum. Figure 3 illustrates a case in which a large septal vein was identified and cannulated (panels c and d), and, after mapping with the angioplasty in different locations within the vein (panels e-j), the most proximal region of the septal vein (panel g) offered the best signal and was used for ethanol infusion with successful elimination of PVC.

Figure 3.

Septal venous mapping and ethanol ablation of LV summit VT. A, NavX (St Jude, St Paul, MN) map of the activation time in the RV, and CS. The wire tip was localized by NavX using an alligator clip. The LV and epicardium were also mapped (not shown) but showed later activation. The unipolar signal from the septal AIV branch precedes all other signals. B, ECG and unipolar wire signal in sinus rhythm and during a PVC. Wire signal obtained from the proximal septal branch (panel g) preceded QRS by 43 ms. E through G, angioplasty wire mapping of different branches of the septal vein (RAO). H through J, LAO views of the wire location relative to earliest sites in the RV outflow tract (RVOT, H), LVOT (I) and septal vein (J). K and L, intracardiac echocardiography images before and after ethanol infusion, showing echogenicity of the LV basal septal septum (arrows). M, ECG before and after ethanol infusion. Abbreviations as in Figure 1.

In one patient with non-ischemic cardiomyopathy and prior LVAD implant, endocardial activation mapping of incessant VT revealed a large component of the VT cycle length contained in the area of the LV summit (Figure 4ab), and AIV mapping with a decapolar catheter showed earliest activations (Figure 4abde). Venograms of the AIV were used to navigate an angioplasty wire into the veins and obtain optimal pace-maps (Figure 4gh). A septal vein was then selected for ethanol infusion (Figure 4ij).

In one patient, the third septal branch of the middle cardiac vein was cannulated and subjected to ethanol infusion with elimination of VT originating from an inferoseptal myocardial infarction.¹⁸

Lateral LV vein ethanol

In a patient with ischemic cardiomyopathy, severe heart failure and prior LVAD implant, endocardial mapping revealed a patchy basal inferolateral scar, from which the exit site of the VT was mapped (Figure 5abc). Mapping of a left ventricular vein with a decapolar catheter revealed earlier signals on the epicardial aspect of the scar (Figure 5d), which prompted more detailed mapping with the angioplasty wire. Venograms demonstrated ample post-capillary collaterals (Figure 5f). In order to deliver ethanol to the myocardium, small venules without such collaterals were selected. Recording from the angioplasty wire revealed mid-diastolic signals and optimal pace-maps (Figure 5ei). Ethanol was infused in two such venules (Figure 5f-k).

Figure 5.

Postero-lateral vein mapping and ethanol infusion in ischemic cardiomyopathy-related VT in a patient with an LVAD. A, activation map showing earliest site arising from the LV postero-lateral base, adjacent to a scar shown in B. RFA failed to terminate the VT at those sites (pink dots). D, A postero-lateral vein on the epicardial side of the scar was mapped with a decapolar catheter and showed signals earlier than the earliest endocardial site. E, angioplasty wire mapping the LV vein shows pace-map match (left) and mid-diastolic signals (right), obtained from the wire position shown in F, which also shows the vein anatomy (with an LV pacing lead in situ) and vein-to-CS collaterals. G, Selective contrast injection in a vein branch without collaterals (RAO view). Myocardial staining without collateral flow shunting is seen (white arrows). Inset shows signals from the wire that vein. H, LAO view. This vein was injected with ethanol. I, additional targeted vein included a small branch without collateral shunting (red arrow). J and K, selective contrast injection in that same small branch showing no collateral flow and myocardial staining (white arrows, RAO and LAO, respectively). This vein was also targeted with ethanol. C, post ethanol endocardial scar map.

Capillary and vein obliteration by ethanol

Targeted veins were selected based on the electrogram timing and/or pace-mapping. Once a targeted vein was cannulated with the angioplasty balloon, selective contrast injection through the inflated balloon was performed. In 6/8 targeted veins, myocardial staining was obtained. In one case, the optimal signal was obtained from the proximal aspect of a septal vein (Figure 3), whereas mapping from more distal in the vein yielded suboptimal signals and pace-maps. Ethanol was delivered in the proximal aspect of the vein, which led to PVC elimination as well as septal vein obliteration (Figure 6ab). In one case, selective septal vein contrast injection opacified a collateral vein returning to the coronary sinus, and upon ethanol injection, all contrast was directed to the myocardium (Figure 6bc). In two cases, initial pre-ethanol selective venograms led to myocardial staining, whereas post-ethanol contrast injection demonstrated enhanced collateral flow as well (Figure 6d-i).

Figure 6.

Coronary venograms before and after ethanol injection. A and B, septal vein obliteration after ethanol. A large septal vein is shown in A. Wire mapping showed best signals from its proximal portion (Figure 3). Ethanol was delivered proximally at a slow rate to allow retrograde flow keep ethanol close to the injection site, achieving PVC elimination. Post ethanol (B) the septal vein was obliterated and contrast injection showed myocardial staining (white arrows). C and D, elimination of post-capillary collaterals with ethanol. A septal vein was selectively cannulated (s2 in Figure 2). Selective contrast injection opacified a collateral vein shunting blood back to the CS (arrowheads in C). After ethanol infusion, contrast injection lead to myocardial staining (arrows in D) and no opacification of the collateral. D through F, new collateral shown after ethanol infusion in a septal branch of the middle cardiac vein (MCV). MCV venogram showed multiple septal branches (D), one of which was selectively cannulated (E) and contrast injection led to myocardial staining (white arrows). After ethanol infusion, contrast injection showed new collateral flow aside from myocardial staining. G through I, new collateral flow after AIV septal ethanol injection. G, non selective AIV venogram. A first septal branch was targeted (white arrow). A collateral vein is shown (black arrowheads). H, selective septal vein venogram shows myocardial staining (arrows). I, after ethanol injection, repeated venogram shows enhanced flow to the collateral (black arrowheads).

Procedural parameters and outcomes

Overall, mean procedure time was 273 ± 56 minutes, 47 ± 26 minutes of which were spent under fluoroscopic exposure. Three out of 7 procedures included some degree of intraprocedural failed RFA prior to RCVEA, averaging 10 ± 8 applications over a span of 1062 ± 827 seconds. (An additional 3 patients had prior failed RFA procedures and had no further RFA attempts during RCVEA, and one patient only had RCVEA as the sole ablative technique –Figure 2).

Acute procedural success was documented in all patients immediately following venous ethanol ablation. In the three cases of LV summit PVC ablation, ectopy was completely abolished after the first 30 seconds of ethanol infusion (Figures 1 through 3). In one of them, PVCs recurred during the patient's hospitalization. In two LV summit VT ablations (Figure 4 and the previously reported¹⁸), there was acute success since attempts to induce the arrhythmia failed following ethanol ablation. In the LV summit VT patient with an implanted LVAD, after ethanol infusion a separate VT was inducible, but not the clinical VT. No major complications occurred during the course of intervention in any patient. One patient developed a small pericardial effusion –attributed to concomitant pericardial access- and 250 ml of bloody fluid was drained from the pericardium at the end of the procedure from the preexistent pericardial access.¹⁸ The patient underwent CMR imaging 48 hours post-procedure. It revealed significant expansion of basal anteroseptal hyperenhancement, indicative of ethanol-induced injury. CMR images demonstrating this outcome are previously published¹⁸. One patient developed a coronary sinus dissection that resolved spontaneously and did not prevent AIV cannulation. There were no other adverse events attributable to the venous ethanol infusion protocol.

Chronic outcomes

Overall, follow-up interval was 590±722 days. The first three patients to undergo the intervention 1113±711 days ago have all suffered recurrence of VT/PVC 217 ±45 days following the ethanol ablation –albeit with a subtly different QRS morphology in all 3. All 3 underwent repeat ablation, this time using RFA – two LV summit PVCs were successfully treated with RFA delivered at the left pulmonary artery, and one had infero-septal RFA. Following RFA, all three patients remain free of VT recurrence to date. One patient had recurrence of his PVC's 1 day following the procedure. All recurrences occurred in cases in which only 1 cc of ethanol was infused: none of the patients in which more than 1 cc ethanol was delivered recurred. No patients suffered any adverse events related to the procedure during the follow-up interval.

Discussion

The salient results of our experience are: 1) coronary venous ethanol is useful as a bail-out strategy in refractory PVC/VT, particularly when originating from the LV summit; 2) detailed coronary venous mapping is helpful in delineating the origin of RFA-refractory PVC/VT and can be achieved by unipolar mapping with an angioplasty wire; 3) capillary obliteration occurs after venous ethanol infusion; 4) despite uniform acute success, recurrence can occur that appears to decrease with repeated –up to 4 – injections.

Current pitfalls in standard treatment

The most commonly utilized modality for VT ablation remains RFA. It has proven to be 81% effective in the acute abolition of VT²⁰, but only about 49% of patients will remain free of disease during medium-term follow-up¹. Other modalities that have been gaining popularity in this regard include cryoablation, laser and high intensity focused ultrasound^21-28. All of these interventions, as is the case of RFA, are limited by a need for adequate tissue contact. VT substrate often presents in the epicardium⁷. Although pericardial access through a subxiphoid technique is useful, this approach remains difficult in patients who have undergone prior cardiac surgery,⁷ particularly in post LVAD patients. Furthermore, many ventricular arrhythmias derive from deep intramural origins that are difficult to target with contact-based instrumentation^{9, 29, 30}. PVC/VT originating from the LV summit remain particularly challenging since they combine both an intramural origin and inaccessibility to epicardial approaches.¹¹ Proximity to coronary vessels is another factor that limits local application of radiofrequency^{8, 10}. Surgical cryoablation or TCEA have been reported.³¹ Strategies such as ablation in the AIV³², or bipolar ablation³³ have been proposed, but catheter size, proximity to coronary arteries or impedance rises limit the former³⁴ and the latter is not universally available. Other approaches have been described, such as hot saline or direct ethanol injections.^{35, 36} Novel approaches such needle RF irrigation,³⁷ facilitated RF with gadolinium or selected irrigants,^{38, 39} are not available or fully developed. Septal coronary venous mapping and ethanol ablation seems uniquely suited for LV summit VTs.

Coronary arterial ethanol ablation

The use of alcohol as an ablative solution preceded the development of RFA, and has been reported extensively in the literature^12-15. Intra-arterial injection of ethanol classically delivered cytotoxic injury through the vasculature supplying target myocardial tissue. In 1989, Brugada et al performed the intervention on three patients with incessant ventricular tachycardia post-myocardial infarction⁴⁰. They described acute procedural success in all three, with recurrence in one patient after collateral flow developed, as well as temporary complete atrioventricular (AV) block requiring pacemaker implantation in one. Similarly, Kay et al reported TCEA in ten patients, with 90% acute success and 50% recurrence¹⁵. Adverse events were common, however, including complete AV block in 40% and pericarditis in 10% of patients. More recently, intracoronary ethanol injection has become limited to arrhythmogenic foci that are not amenable to contact-based ablation^{12, 41}. Non-inducibility of VT following intracoronary ethanol ablation is reported between 56 and 84%, with a recurrence rate of around 33- 64%^{1, 31, 41, 42}. Unfortunately, adverse events are not uncommon, most notably coronary arterial dissection, thrombosis and myocardial infarction^13-15. Furthermore, ethanol infusion may spill over to non-targeted myocardium resulting in infarction or conduction block¹⁵. Additionally, in patients with CAD, the technique is limited by difficulty in localizing a terminal arrhythmia-related vessel, especially in the presence of coronary stenosis¹². Another limitation stems from the variability of vessel size and flow rate, which influence the cytotoxicity of ethanol and make its outcomes difficult to predict.

Retrograde venous ethanol ablation

To overcome some of these limitations, retrograde venous infusion of ethanol has been described as an alternative to the arterial approach. Its initial application in canines showed promise in circumventing arterial damage, thereby avoiding the risk of coronary arterial dissection and myocardial infarction¹⁷. Furthermore, off-target myocardial injury created by ethanol leak is not a concern of RCVEA, as retrograde flow allows dilution of the ethanol into the coronary sinus rendering it harmless. Still, an occlusive balloon is necessary, since given the retrograde direction of ethanol infusion, the normal blood flow direction naturally tends to wash out ethanol from the targeted tissue. In our experience, RCVEA has proven to be acutely effective in suppressing ventricular arrhythmias,¹⁸ including one case in which long-term elimination of VT was achieved by RCVEA alone without the use of any other ablative technology (Figure 2).

The coronary venous approach offers logistical advantages to the operator as well. Particularly when combined with venous mapping –which entails coronary sinus access and selective vein branch cannulation-, adding an angioplasty balloon does not contribute increased logistical complexities to the case. Most centers using intra-arterial ethanol for VT recruit an interventional cardiologist for this procedure, which adds such logistical complexities. Most electrophysiologists are thoroughly familiar with the coronary venous anatomy and its instrumentation is aided by tools developed for LV lead placement in coronary veins.

The coronary venous anatomy is extremely redundant. Aside from venous return to the CS, Thebesian veins can drain directly in to the LV cavity.¹⁷ Within the epicardial venous system, collateral veins abound, communicating epicardial veins with one another and the CS (Figure 5). When targeting the myocardium with ethanol, it is important to use a vein with direct connection to capillaries in order to avoid ethanol shunting. Collateral veins may be present at baseline injection. In some cases, collateral veins disappeared after ethanol, whereas in others, they became more prominent. We hypothesize that ethanol obliterates capillaries. Collateral veins arising after the capillary territory obliterated by ethanol would disappear after ethanol (Figure 6 bc), whereas those collateral veins arising before the capillaries may become more prominent (Figure 6d-i). Figure 7 shows a schematic illustrating this concept.

Figure 7.

Schematic of coronary venous flow and proposed effects of ethanol. A, Redundant epicardial coronary veins show vein branches followed by venules leading to capillaries. Pre-capillary collaterals and post-capillary collaterals exist. Not shown are Thebesian veins draining into LV and RV cavity. Selective retrograde contrast injection shows flow into the capillaries (myocardial staining) and through postcapillary collaterals (as in Figure 6c). B, after ethanol infusion, capillaries are obliterated, and repeated retrograde contrast cannot reach post-capillary collaterals (as in Figure 6d). If pre-capillary collaterals exist, flow through them is enhanced after ethanol (as in Figure 6d-i).

Long-term outcomes remain challenging, since we observed recurrence in 4 out of 7 cases. The fact that the non-recurrent cases underwent repeated ethanol injections (up to 4 cc vs only 1 cc in the recurrent cases) suggests that successful myocardial ablation may require higher ethanol doses. Other potential reasons for failure include inaccurate localization, inadequate tissue destruction, or perhaps other mechanisms.

Limitations

This is a small, observational case series. As long as the intervention is considered a bail-out strategy of last resort, and in the presence of alternative bail-out modalities, recruitment of large numbers of clinically indicated patients will be difficult. Patients enrolled in this series host exceptionally drug and ablation-resilient arrhythmias, so recurrence outcomes cannot be generalized to all treatment-indicated LV summit cases. We cannot compare outcomes with other bail-out techniques, but the advantages are appealing. Furthermore, all procedures in this series were performed by a single operator who is experienced in ethanol ablation (MV). Hence, additional experience with a larger cohort of patients and by multiple operators will be necessary to assess the global clinical safety, efficacy and practicality of RCVEA. Moreover, the specific mechanism of capillary obliteration enforced by this technique remains ill defined and should ideally be explored through histologic data derived from animal models. Although the technique is promising, it is not without limitations, including the need for a suitable vein without collateral shunting, the technical challenges of vein cannulation, the uncertain mechanisms and dosing of ethanol.

Conclusion

RCVEA appears to be safe and feasible as a bail-out approach in the treatment of failed VT RFA, particularly for LV summit VT.

WHAT IS KNOWN

• Radiofrequency catheter ablation (RFA) of ventricular tachycardia (VT) fails when the VT substrate is inaccessible, as is common for those originating from the LV summit.

• Retrograde coronary venous ethanol ablation (RCVEA) has been shown to be feasible, but human experience is limited.

WHAT THE STUDY ADDS

• RCVEA was acutely successful in 7 patients with difficult ventricular arrhythmias, 5 of which arose from the LV summit. Although VT recurred in 4 it was then controlled with RFA.

• There were no major complications attributable to ethanol injection.

• RCVEA is reasonable to consider when RFA fails and an appropriate coronary venous target can be identified.