Author + information
- Received January 2, 2017
- Revision received January 31, 2017
- Accepted February 9, 2017
- Published online April 26, 2017.
- Miguel Valderrábano, MD∗ (, )
- Percy Francisco Morales, MD,
- Moisés Rodríguez-Mañero, MD,
- Candela Lloves, MD,
- Paul A. Schurmann, MD and
- Amish S. Dave, MD, PhD
- Division of Cardiac Electrophysiology, Methodist DeBakey Heart and Vascular Center and Houston Methodist Hospital, Houston, Texas
- ↵∗Address for correspondence:
Dr. Miguel Valderrábano, Division of Cardiac Electrophysiology, Houston Methodist Hospital, 6550 Fannin Street, Suite 190, Houston, Texas 77030.
Objectives This study catalogued the human venous left atrium (LA) circulation system and the ablative effects of ethanol in different branches.
Background Vascular routes to target the LA could have significant therapeutic potential. Beyond the vein of Marshall (VOM), the fluoroscopic LA venous anatomy has not been described.
Methods Patients undergoing ethanol infusion in the VOM as adjunctive therapy to atrial fibrillation (AF) catheter ablation were included in this study. Balloon occlusion venograms of the VOM and other LA veins were obtained in 218 patients.
Results Sequentially from the coronary sinus (CS) ostium, LA veins included: 1) proximal septal vein draining the inferior septum; 2) inferior LA vein in the annular inferior LA; 3) VOM; 4) LA appendage vein; and 4) anterior LA vein. Additionally, venous sinuses not connected to the CS included roof veins and posterior wall veins, which drained into the right and left atria, respectively. Venous connections between LA veins through capillaries and with pulmonary veins were abundant. Extracardiac collateral vessels were present in 38 patients (17.4%). Ethanol infusion in LA veins led to tissue ablation in their corresponding regions.
Conclusions The atrial venous anatomy is amenable to selective cannulation. Consistent anatomical patterns are present. Targeting atrial tissues through atrial veins can be used for therapeutic purposes.
Vascular targeting of atrial tissue could be attractive for therapeutic purposes as novel pharmacological or cellular therapies develop for cardiac disease. The left atrial (LA) coronary artery anatomy has been described in detail (1), but the inconsistent anatomy and small size preclude reliable atrial artery cannulation. Venous approaches appear attractive given the ease of cannulation through the coronary sinus (CS).
The LA vein described best is the oblique vein of the left atrium, the vein of Marshall (VOM), which was described in 1850 (2). Embryologically, the VOM is a remnant of the left superior vena cava (3). The VOM descends obliquely, posterior to the LA appendage (LAA) over the epicardial aspect of the LA lateral ridge, running along the posterolateral LA toward the CS (4,5). Little is known about the anatomical variations of the VOM, its in vivo drainage patterns, its function as a vein, or its connection to other components of LA venous circulation.
Prior anatomical descriptions of the VOM and other veins have involved pathological necropsy specimens (5–7). We have previously described the technique of retrograde ethanol infusion in the VOM as a potential therapeutic tool in atrial fibrillation (AF) (8–12). Cannulation and ethanol infusion of the VOM and other additional LA veins have allowed us to compile a complete in vivo angiographic catalogue of the human LA venous drainage. Additionally, ethanol-ablated tissue revealed the myocardial drainage territory of the infused veins.
The purpose of this study was to describe the anatomical variations of the LA venous drainage as a framework for venous endovascular therapies targeting the LA.
Patients undergoing catheter ablation of atrial fibrillation (AF) gave written consent to participate in pilot mechanistic studies (8–12) or the on-going VENUS-AF (Vein of marshall EthaNol for Unablated perSistent atrial fibrillation) or MARS-AF (vein of Marshall ethanol for Recurrent perSistent atrial fibrillation) clinical trials (NCT01898221), or they underwent VOM ethanol infusion as part of their clinical treatment. The protocols were approved by the Institutional Review Board. Table 1 shows patient population data. The procedure was aimed at cannulating the VOM for ethanol infusion, but occlusion VOM venograms allowed us to visualize the atrial venous anatomy described herein. It was attempted in 235 patients. We inserted a 9-F sheath through the right internal jugular vein (n = 232), the left subclavian vein (n = 2), or the right femoral vein (n = 1), through which the CS was engaged, using commercially available sheaths for left ventricle lead delivery, most commonly a CPS (St. Jude Medical, St. Paul, Minnesota) or a Preface sheath (Biosense Webster, Diamond Bar, California). Then, a left internal mammary angioplasty guide catheter was inserted through the CS sheath and used for contrast injection. To aim for the VOM, the left internal mammary guide tip was directed superiorly and posteriorly in the CS while contrast was injected to identify the VOM or other branches. The VOM was identified as a branch of the CS directed posteriorly and superiorly, immediately distal (toward the CS ostium) to the valve of Vieussens. When the VOM was not amenable to cannulation, other atrial veins were attempted. Once a vein engagement was obtained, an angioplasty wire (Balance Middle Weight [BMW]; Abbott, Abbott Park, Illinois) was advanced. A pre-loaded angioplasty balloon (8-mm length, 2-mm nominal diameter, or 1.5 mm by 6 mmm) was then advanced into the vein for selective venograms. Vein size, branching, and collateral circulation and other opacified atrial veins were catalogued in all patients included in this study.
Measurements and digital subtractions were made using OsiriX DICOM (digital imaging and communications in medicine) format viewer (version 8.0.1, Pixmeo Sarl, Geneva, Switzerland), calibrated to known interelectrode distances on the CS decapolar catheter.
We have previously described the technique (8–12). Once a vein was cannulated, the angioplasty balloon was advanced as distally as possible. The balloon was then inflated from 2 to 4 atm at its most distal location, and 1 cc of ethanol was infused over 2 min, followed by saline flush. The balloon was then deflated and retracted ∼1 cm and a repeated 1-cc ethanol inflation was performed. This procedure was repeated until the balloon was at the ostium of the vein. Depending on the length of the cannulated vein, up to 4 ethanol infusions were delivered in total. For quantification of ethanol-ablated tissue, pre- and post-ethanol bipolar voltage maps were created using the Carto mapping system (Biosense Webster, Diamond Bar, California) and a circular or 5-spline duodecapolar catheter. For consistency, maps were color-scaled between 0.1 and 0.5 mV, and scar tissue was defined as areas of bipolar voltage <0.1 mV.
Gaussian continuous variables are reported as mean ± SD and non-Gaussian variables as median (minimum–maximum). Qualitative findings were described as numbers and percentages. All statistics were performed with the use of SPSS software (SPSS version 19; IBM, Armonk, New York). All authors had full access to the data, and all take responsibility for its integrity. They have read and agreed to the manuscript as written.
Patient population and procedural parameters
Table 1 summarizes patient and procedure characteristics. Overall success at cannulating an atrial vein was 93% (218 of 235 patients). Table 2 shows angiographic results. The most commonly cannulated vein was the VOM in 188 of 235 patients (80%). Other cannulated veins included LAA veins in 16 of 235 patients (7%), inferior veins in 7 of 235 patients (3%), septal veins in 5 of 235 patients (2%), and a terminal anterior atrial vein in 2 of 235 patients (1%).
Left atrial venous circulation anatomy: overall arrangement
Even though the VOM was the most commonly cannulated vein, other atrial veins were variably opacified through collateral flow, which enabled us to compile an atlas of the LA venous circulation. We found a consistent pattern of atrial branches arising from the CS ostium. Starting from the CS ostium, the LA veins included: 1) a septal vein; 2) a second, inferior atrial vein; 3) the VOM; 3) LAA veins; and 4) an anterior roof vein. We also found LA veins that were not connected to the CS, including: 5) roof veins and posterior wall veins, which were commonly connected with 6) extracardiac collaterals.
Septal and inferior LA veins
Septal veins were detectable in 27 of 218 patients (13%) (Figure 1) and were selectively cannulated in 5 (Figure 1, Online Video 1). In the remaining 22 patients, septal veins were opacified through collaterals from the VOM or the inferior LA vein. Septal veins ran along the inferior interatrial septum toward the right inferior pulmonary vein (RIPV) and frequently drained into it, as shown by contrast dissipating along the RIPV ostium. Variable connections to atrial veins in the posterior wall and roof veins were present, reaching up to the VOM through collateral flow (Online Video 1).
A second inferior atrial vein running along the inferior LA wall toward the posterior LA wall was visible in 56 patients, either as a branch of the septal vein (19 of 27 septal branches) (Figure 1) or as a separate vein visible in 16 patients (7 of which were cannulated) (Figure 2), or a branch of the VOM (in 21 of 188 cannulated VOMs) (Figure 3). This inferior vein ran toward the posterior wall, where it communicated with posterior wall veins, draining into the RIPV, with roof veins and extracardiac collaterals and back toward the VOM (Figure 3, Online Video 2). In other cases, the inferior vein could communicate with posterior wall veins, into both the RIPV and the right superior PV and with a septal vein (Online Video 3).
Vein of marshall: anatomical variants
The VOM was the LA vein most consistently found. It was identified as the vein arising at the level of the valve of Vieussens (ostial to it). A total of 188 selective VOM venograms are included in this study. Additionally, when not cannulated, the VOM was visible through collateral flow in 14 of 30 venograms obtained from selective cannulation of other atrial veins (septal veins and LAA veins). For the purpose of VOM morphology characterization, we used only the selective VOM venograms. Distance from the CS ostium to the VOM ostium was 4.25 ± 2.57 cm, with substantial variability (Online Figure 1).
The VOM was most commonly a true atrial vein, with branches and visible venules draining the neighboring atrial tissue (8). There were 18 of 188 patients (10.4%) (Table 2) who demonstrated a venous plexus at the origin of the VOM, defined as VOMs without a clear dominant vein and more than 3 branches, and were not included in measurement analysis (Table 2). Figure 4 shows examples of plexus VOM. There were also patients in whom the VOM was a stump with little or no atrial branches and an abrupt transition from a CS origin into a cul-de-sac (23 of 188 patients [12.2%]). In most patients, variable branching into small venules was present (147 of 188 patients [78.2%]). Online Figure 2 shows examples of stump and branching VOM. The VOM length before branching was 2.99 ± 1.82 cm. Branching patterns of the VOM are summarized in Table 2.
By definition, the VOM traveled from the CS along the lateral ridge toward the left PVs. When divided anatomically in relation to the left inferior PV, identified by inserting a circular multipolar catheter in it, 33 of 188 patients (17.6%) had a smaller VOM, which terminated before reaching the left inferior PV. In the majority of patients, the VOM was visible up to the left inferior pulmonary vein (LIPV) (137 of 188 patients [72.8%]). In the remaining 18 of 188 patients (9.6%), the VOM went past the LIPV, reaching the left superior pulmonary vein (LSPV). Communication between the VOM and left PVs was demonstrated by the presence of contrast drainage in the PVs during the VOM venogram, appearing to connect through the left pulmonary vein carina (Online Video 4). It was noted in 71 of 188 cases (37.7%).
Left atrial appendage vein
Veins draining the LAA were visible in 115 of 218 patients (53%). In 40 of 218 patients (18%), the LAA vein arose as a branch from the VOM (Figures 5A, 5B, and 5C). In 16 of 218 patients (14%), the LAA vein was selectively cannulated using the angioplasty balloon (Figures 5D, 5E, 5F and 8d). Additionally, in 59 of 218 patients (27%), a distinct, separate LAA vein distal to the VOM and located at the base of the LAA was opacified through collaterals during VOM contrast injection, and the LAA vein could be filled all the way to its origin in the CS (Figures 5G, 5H, and 5I).
Anterior atrial vein
The last atrial vein branch of the great cardiac vein (prior to becoming the anterior interventricular vein) was identified as an anterior atrial vein. It was opacified in 42 of 218 patients. In 2 patients, it was selectively cannulated using the angioplasty balloon. The anterior atrial vein arose distal to the LAA vein and medial to it, collecting venous drainage from the anterior LA roof (Figures 6 and 8F, Online Video 5).
Noncoronary sinus (thebesian) venous atrial drainage
Selective cannulation of atrial veins through the CS often filled other components of the atrial venous drainage that were not connected to the CS. These veins were opacified through post-capillary collaterals. There were 2 main non-CS venous systems. The first, opacified during contrast injection in proximal veins such as septal, inferior, or VOM (in cases of relatively proximal VOMs), was a venous system in the posterior wall of the LA, visualized in 31 cases. This venous system connected through capillaries to the septal or inferior vein. Posterior veins were visualized in 2 of 5 septal vein venograms, in 3 of 7 inferior vein venograms, and in 26 VOM venograms (through the inferior vein branch of the VOM). It appeared to communicate with the right inferior pulmonary vein and it could itself connect to the roof veins (see below) (Online Videos 2 and 3, Figures 3A, 3E, 5G, 5H, and 5I).
A second Thebesian venous system was identified in the LA roof in a total of 94 of 218 patients (43%). Most commonly opacified during VOM venograms and seen to various extents in 78 of 188 VOM venograms (42%) (Figure 7, Online Video 6), it could also be detected in LAA vein venograms (11 of 16 cases) or in injections of the anterior left atrial vein (5 of 12 cases) (Figures 6D, 6E, and 6F, Online Video 5). Left anterior roof veins ran diagonally from the lateral anterior roof (where they communicated with the VOM through capillaries) toward the posterior medial aspect of the LA roof, communicating with the LA close to the right superior PV (Online Video 4) or the right anterior-superior vena cave (RA-SVC) junction (Online Video 5).
There were also 38 of 218 patients (17%) in whom extracardiac veins were opacified (Online Figure 3). In these patients, a small collateral vein was observed traveling superiorly in the mediastinum to an undetermined destination, traveling outside the cardiac silhouette. These extracardiac collaterals may represent remnants of a left SVC (Online Figures 3B and 3C), in which case they appeared simply as a continuation of the VOM superior to the heart silhouette, commonly connecting with roof veins. In one case (Figures 3D and 3H, Online Video 2) an extracardiac vein appeared to connect to LAA veins.
Ethanol-induced low-voltage scar distribution
Figure 8 shows representative examples of ethanol-induced scar for each vein, except for septal veins, which were not injected to avoid potential AV node damage. The extent of the ethanol-induced scar depended not only on the overall size of the vein cannulated but also on the capillary network attached to it. Stump veins with little or no capillaries could lead to minimal endocardial scarring, compared to those with more extensive capillary networks (Online Figure 4). Upon ethanol injection, capillary obliteration eliminated collateral connections with all other noncannulated veins, which would not be visible on post-ethanol contrast injections.
The novel findings of our work include anatomical demonstration of the in vivo entire venous circulation of the LA in humans. Besides the known anatomical locations of atrial veins, abundant interconnections are unveiled, for example, delineation of a percutaneous technique to deliver therapeutic agents to the LA in different regions beyond the VOM and delineation of the regional ablative effects of ethanol in each venous territory.
Left atrial venous circulation
Descriptions of the human left atrial veins are scarce. von Lüdinghousen et al. (6) studied 100 post-mortem human hearts and their left atrial veins. They described 3 groups of veins. The first group included veins that drained directly into the right atrium. These included inferoseptal veins, with ostia in the neighborhood of the CS, and anteroseptal veins draining close to the SVC. The former had been originally described by Bochdalek (13), connecting to the right atrium as sinusoids separate from the CS. We show that selective cannulation of inferoseptal veins can be achieved from the CS itself and that they can connect with right inferior pulmonary veins. Additionally, septal veins can be opacified through collateral flow during injection in other veins. The anteroseptal veins draining into the SVC-RA junction described by von Lüdinghousen et al. (6) would correspond to the veins we define as “roof veins,” which, in our experience, were only visualized through collateral flow and did not have CS connections. In some cases, the roof veins did not drain in the RA but seemed to drain into the LA, close to the right superior pulmonary vein (Online Video 6).
A second, larger and more consistent venous system consisted of veins that were branches of the CS and great cardiac vein. These included inferior veins, lateral veins, the VOM, appendage veins, and anterior veins. We confirmed this overall pattern but show substantial variability in the individual vein size, shape, branching patterns, and connection to one another. The VOM is the most consistent LA vein. von Lüdinghousen et al. (6) found it in 99% of their specimens. In a study of 100 patients undergoing CS venograms, the VOM was present in 73% of patients (14). In that study, 75% of patients with a VOM had a poorly developed VOM as described by the authors (it did not reach the roof of the left atrium) (14). However, as shown here, selective venograms of the VOM, compared to nonselective venograms through the CS, may be able to better visualize the entire length of the VOM and the presence of collaterals veins.
In one of the larger descriptions of the VOM from a group of 275 anatomical specimens, the VOM was present in 84% of specimens, with a mean length of 2 to 3 cm and an average diameter of 1 mm (7). In other studies, the presence of a VOM has been variable, ranging from 34% in a study of 38 patients undergoing multislice cardiac computed tomography scanning (15) to 87% in a study of 23 pathologic specimens (16). Our own success at cannulating the VOM was 86.2%, but the VOM was visualized using collateral flow of other venograms. Despite its consistent presence, the VOM was significantly varied in size, branching patterns, and capillary content.
The third venous system included true thebesian veins, named “proper” atrial veins (6), that were located in the posterior wall, drained into the left atrium itself, and had frequent connections with mediastinal veins.
Our work confirms these anatomical features in vivo and describes the potential clinical utility of percutaneous cannulation of these veins.
Interconnections of the venous system
The value of in vivo cannulation and venograms, as opposed to post-mortem studies, is that we were able to show the abundant interconnections among different venous branches. Contrast injection in the septal veins commonly led to opacification of neighboring veins through capillaries, including inferior vein, posterior veins, and roof veins and even back to the VOM. Contrast injection in the VOM could opacify septal and inferior veins, but most commonly the VOM communicated with LAA veins and roof veins. Roof veins could only be demonstrated through collateral flow, from either the VOM, the LAA veins, or anterior veins. Communication between the atrial veins and the PVs was unexpected but probably represents the PVs’ own wall vasa vasorum venous drainage connecting to LA veins.
The VOM has been mechanistically related to the genesis of AF (for a review. see reference 17), both as a source of focal ectopic beats that trigger AF (18) and as an AF substrate linked to its abundant parasympathetic (19) and sympathetic (20) innervations that modulate electrical properties of atrial tissue and contribute to AF maintenance (21). These properties have made the VOM an attractive target during ablation of AF.
The therapeutic validity of ethanol infusion in the VOM seems to be clear for difficult cases of perimitral flutter (11,22), or occasional cases of VOM atrial tachycardia (10). Only the aforementioned VENUS-AF or MARS-AF clinical trials (NCT01898221) will establish its role for the general AF ablation patient population. The use of ethanol in other veins is less established mechanistically. Because, by definition, CS-cannulated atrial veins are annular, the use of ethanol can be justified if seeking mitral annular conduction block. It is of interest that ethanol infusion in the anterior LA vein led to an ablation lesion strikingly similar to that of an anterior mitral line (Figure 8F) (23,24). Lesions obtained by ethanol injection in LAA veins lead to a low-voltage scar in the LAA base that can be used to obtain mitral annular conduction block as well. Other lesions created by inferior and septal lines can contribute to the larger lesion sets occasionally sought in long-standing persistent AF ablation. Generally, the extent of ethanol infusion reflected the size and capillary network of the cannulated vein and served here to illustrate the myocardial regions that can be targeted for cannulating each vein branch. Future therapies targeting atrial myocardium, for therapeutic purposes beyond mere chemical destruction with ethanol, can be based on the anatomical depictions and technical basis shown in our work.
Cannulating atrial veins other than the VOM can be difficult due to their small size. Thus, technical reproducibility may be limited. It is possible that using vasodilatory agents (e.g., nitroglycerin) could have helped expand the venous network to improve visualization. Although we do not believe anatomy should differ, the current data set was obtained in patients with AF and may not be generalizable beyond those patients.
COMPETENCY IN MEDICAL KNOWLEDGE: Venous approaches to targeting left atrial tissues for ablative purposes could be attractive. Our previous work has shown the feasibility and utility of vein of Marshall ethanol infusion for the treatment of atrial fibrillation and perimitral flutter. Herein, we expanded the approach to other atrial veins and describe the entire angiographic anatomy of the left atrial venous circulation, along with the ablative effects achieved by ethanol infusion. These approaches could be used for the treatment of left atrial arrhythmogenic substrates.
TRANSLATIONAL OUTLOOK: The current anatomical demonstration of the entire left atrial venous circulation, along with the myocardial regions drained by each vein, provides the procedural and anatomical foundation to use retrograde venous approaches to target atrial tissue. Although currently used for ablative purposes in the context of atrial fibrillation treatment, the approach presented herein could be used to selectively reach atrial myocardium for pharmacological or cellular therapies.
For supplemental figures and videos and their legends, please see the online version of this article.
This research was supported by U.S. National Institutes of Health/National Heart, Lung, and Blood Institute grants R21HL097305 and R01 HL115003 to Dr. Valderrábano and by Charles Burnett III Endowment and Antonio Pacifico, MD, fellowship support to Dr. Rodríguez-Mañero. Dr. Valderrábano is a consultant for and has received research support from BioSense Webster and speakers fees from Boston Scientific. Dr. Dave is a consultant to St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- coronary sinus
- left atrium
- left atrial appendage
- vein of Marshall
- Received January 2, 2017.
- Revision received January 31, 2017.
- Accepted February 9, 2017.
- 2017 American College of Cardiology Foundation
- James T.N.,
- Burch G.E.
- Marshall J.
- Irwin R.B.,
- Greaves M.,
- Schmitt M.
- Scherlag B.J.,
- Yeh B.K.,
- Robinson M.J.
- Cabrera J.A.,
- Ho S.Y.,
- Climent V.,
- Sanchez-Quintana D.
- Valderrabano M.,
- Chen H.R.,
- Sidhu J.,
- Rao L.,
- Ling Y.,
- Khoury D.S.
- Baez-Escudero J.L.,
- Keida T.,
- Dave A.S.,
- Okishige K.,
- Valderrabano M.
- Bochdalek V.A.
- Jongbloed M.R.,
- Lamb H.J.,
- Bax J.J.,
- et al.
- Rodriguez-Manero M.,
- Schurmann P.,
- Valderrabano M.
- Ulphani J.S.,
- Arora R.,
- Cain J.H.,
- et al.
- Kim D.T.,
- Lai A.C.,
- Hwang C.,
- Fan L.T.,
- Karagueuzian H.S.,
- Chen P.S.,
- Fishbein M.C.
- Ammar S.,
- Luik A.,
- Hessling G.,
- et al.