Despite developments in conventional therapy, heart failure remains an important cause of morbidity and mortality . Although there have been slight increases in the survival rates, the incidence of heart failure remains on the rise . At 40 years of age, the lifetime risk of developing heart failure for both men and women is one in five .
Over the last decade, there has been great progress in understanding the various intracellular molecular mechanisms that play key roles in the development and progression of heart failure. These advancements have enabled the translation to new therapeutic approaches (i.e. gene therapy). Gene therapy for the treatment of heart failure is emerging as a multidisciplinary field demonstrating advances with respect to identifying key signaling pathways, modernized vector creation and delivery technologies. Although these discoveries offer significant progress, selecting optimal methods for the vector delivery remains a key component for clinically relevant and efficient cardiac gene therapy. Based on these emerging technologies, this review focuses specifically on the delivery methods. Comprehensive knowledge of the numerous methods of gene delivery with their advantages, disadvantages and relative risks is key to the appropriate selection of patient-specific optimal methods for each vector, as well for the carried genes. In the decision-making process for the delivery method, the proportion of target tissue has to be considered. For example, in cases of myocardial infarction (MI), focal transduction surrounding the lesion can be sufficient, whereas global myocardial dysfunction usually requires broad and homogenous gene transfer. In addition, maximal expression levels require more complicated and invasive procedures at elevated risk levels. Therefore, safety and feasibility profiles will be required to balance risks and evaluate clinical outcomes.
Below, we discuss the various methods of gene delivery.
Peripheral venous injection
Peripheral intravenous gene transfer has the advantage of being a non-invasive and simple delivery method and, recently, promising results have been obtained in rodents [3, 4]. However, this approach is limited by the immediate dilution of the injected agent in the blood volume with a wide systemic distribution of the vector. Therefore, especially in large animals or humans, high doses of vector would be necessary to achieve adequate gene expression in the target cardiac tissue. In addition, expression in nontargeted organs can be significant . Several approaches to deal with these issues have emerged. One attempt to overcome immediate dilution and the delivery to nontargeted organs is to deliver the agent through the distal port of a wedged Swan–Ganz catheter. However, efficacy was not increased, nor was collateral expression reduced . Employing more highly cardiotropic vectors can reduce collateral transduction of nontarget organs. Indeed, some adeno-associated virus serotypes show a clear tropism to cardiac cells, and a tissue-specific promoter can further increase the efficacy of this approach . Recently, DNA shuffling using directed evolution technologies has been employed to develop cell type-specific vectors . Müller et al.  modified an adeno-associated virus 2 vector and showed 50-fold higher luciferase activities in the heart compared to skeletal muscle and liver. Another approach is ultrasound-enhanced microbubble gene delivery that can enhance site-specific expression in cardiac tissue [8, 9]. Perturbation of the cell membrane and vessel walls by the ultrasound energy increases permeability to gene or vector carrying microbubbles and thereby interstitial delivery . Although representing an interesting approach for improving the efficacy of intravenous gene transfer, future studies conducted in large animal models to demonstrate safety and efficacy are required for its comparison with other delivery methods. In conclusion, intravenous gene therapy targeting the failing heart can be a successful method in rodents, although it has not yielded favourable results in large animal models. Currently, other methods of gene transfer show more promise. (Table 1)
Table 1. Advantages and disadvantages of the delivery methods
Peripheral intravenous injection
High expression in nontargeted organs
Low myocardial transduction rate
Antegrade coronary injection
Lower transduction efficiency
Limited delivery to the ischemic area
Retrograde coronary injection
Higher transduction efficiency
Requires blockage of antegrade flow
Higher transduction efficiency
Direct intramyocardial injection
No endothelial barrier
Surgical delivery being the most invasive delivery method
No first pass-effect of liver and spleen
Damage along the needle track
Reduced T-cell triggered inflammation
Region of interest might be not completely covered as a result of a restricted area of injection
Independent of neutralizing antibodies
Limitation of vector delivery as a result of leakage at the injection site
Endomyocardial approach being less invasive
Transduction mostly in pericardial cells and only very limited in myocardial cells
High concentration and long duration of vector possible
Little systemic vector distribution
Possible leakage of injected agent into thorax cavity with transduction of nontargeted organs
In future clinical use, difficult to apply after pericarditis or previous cardiovascular surgery
Antegrade coronary artery injection
Antegrade coronary injection (Figure 1A) allows the simple and minimally invasive delivery of vectors with cardiac selectivity. As a result of the severe cardiac dysfunction and the delicate hemodyanamic balance in most candidates for cardiac gene therapy, the minimally invasive nature of this delivery method has a great advantage over other methods. This delivery method has been widely used in both clinical and experimental gene therapy studies. Although the injection of the vectors into all of the coronary arteries can achieve homogeneous distribution, vector delivery can be fine-tuned to specific coronary artery branches with this method. However, the desired distribution may be impaired in patients with coronary artery disease, especially when there is a total coronary occlusion. In addition, antegrade coronary artery injection of vectors results in relatively low transduction efficiency. The endothelium acts as a barrier for the vectors to reach myocardium. Thus, various agents have been utilized to increase the permeability of the vascular bed. Adenosine, vascular endothelial growth factor , histamine  and nitroglycerine  have been shown to successfully improve the transduction efficiency.
Several approaches are proposed to increase both transduction efficiency and the cardiac specificity. Low-dose continuous infusion minimized the spillover to the systemic circulation, at the same time as increasing the duration of vector exposure to the coronary arteries. Although this method does not result in the highest transduction efficacy, simple and effective gene transfer has been demonstrated [14, 15]. Coronary artery blockade using the occlusion balloon enables the injection of vectors without any dilution and controls the infusion rate (Figure 1C). Although high-pressure infusion with coronary artery blockade is demonstrated to achieve higher transduction efficiency , it elevates the risk of coronary injury. Furthermore, coronary occlusion may not be well tolerated, even for a short period of time, in a severely impaired heart. The V-focus system is another method of restricting vector delivery to the heart by isolating coronary circulation from systemic circulation with an extracorporeal device (Figure 1B). Repeat circulation of the vectors in this closed circuit results in longer vector exposure to the heart and thus provides high transduction efficiency [17, 18].
Retrograde coronary artery injection
The coronary sinus is another conduit that allows distribution to most parts of the heart. Retrograde infusion using the coronary sinus (Figure 2A) enables access to the ischemic myocardium, regardless of the occluded coronary artery , thus allowing homogeneous distribution even in patients with severe coronary artery disease. Pressure-regulated retro-infusion by coronary sinus blockade has shown to be more effective compared to direct myocardial injection  and antegrade infusion . However, this method requires the transient interruption of coronary flow to eliminate the antegrade flow effect on pressure. Although the concurrent delivery of vectors via both antegrade and the retrograde has been shown to achieve the highest reported efficiency , these methods are more invasive because of the need for antegrade coronary flow interruption.
Retrograde access can be also applied to the direct injection method using a composite catheter system (TransAccess; TransVascular Inc., Menlo Park, CA, USA)  (Figure 2B). This novel catheter based delivery technique utilizes an intravascular ultrasound device to aid in selection of the precise injection site. A snug fitting catheter is held by the coronary sinus lumen and contributes to the accurate injection, even in a beating heart. This method shares most of the advantages and disadvantages of other direct injection methods; however, vector deliverable area may be limited by coronary sinus accessibility.
Local intravessel wall delivery
Advances in new catheter technologies have enabled local coronary gene transfer. Although the use of some catheters or balloons enables efficient gene delivery to endothelial and superficial smooth muscle cells [22-25], infiltrator catheters  achieve high transduction with penetrance to the adventitia . Vector-eluting stents are alternative options that allow local delivery of the vectors  (Figure 3A). Longer attachment duration to a specific site may increase the transduction efficiency. Because of their site specificity, these delivery methods are best suited for local conditions such as restenosis after coronary interventions.
Aortic cross clamping
Transient aortic cross clamping during the vector delivery provides homogeneous gene transfer throughout the left and the right ventricles [15, 29, 30] (Figure 3B). With the aortic cross clamp in place, a closed circuit involving the coronary system is maintained at the same time as high perfusion pressure opens the capillaries. By contrast to the brief clamping of the aorta in the beating heart, cardiopulmonary bypass with cardioplegic arrest was performed with aortic clamping for 30 min during vector delivery . This approach provided efficient cardiac uptake without detectable gene expression to nontarget organs. In addition to the systemic cardiopulmonary bypass, Bridges et al. [29, 30] employed a second cardiopulmonary bypass for the coronary circulation. This method allowed vectors to re-circulate multiple times through the coronary circuit, and resulted in a more homogeneous transduction. However, because of their highly invasive nature, these methods are associated with inherently higher risk and remain experimental.
In summary, additional approaches or devices to simple antegrade coronary injection offer improved transduction efficiency. However, they are often accompanied by more invasive and sometimes very complicated procedures. Further development in both pharmacological and vector specificity may provide better transduction efficiency by overcoming the endothelial barrier.
Direct intramyocardial injection
There are two methods of direct injection into the myocardium: a percutaneous catheter-based approach  (Figure 4B) and a surgical approach  (Figure 4A). Although being less invasive, percutaneous gene delivery with an injection catheter requires a guidance system to identify the targeted injection area. These guidance modalities can include fluoroscopy , magnetic resonance imaging , echocardiography  and three-dimensional mapping systems . A study with microsphere injection reported an equivalent to better retention of injected material after catheter-based endomyocardial injection compared to the open chest epicardial injection . By contrast to intravascular gene transfer, both direct injection methods share the major advantages of being largely independent of the effects of neutralizing antibodies and T-cell triggered inflammation. Furthermore, these approaches avoid the challenge of vector penetrance through the endothelial barrier. Systemic distribution via the blood flow is minimized, although vector escape from the injection site is not completely prevented . Indeed, there is some collateral expression described, especially in lung and liver , which increases with rising injection volumes. In addition, expression is observed mainly within a few millimeters of the injection site. Thus, despite a defined area of injection, there can often be inhomogeneous and patchy expression patterns . However, direct intramyocardial injection remains highly attractive for achieving high transduction efficiency in a focused region of interest.
The pericardial cavity, comprising a closed space in between the parietal and visceral pericardial layers, has been accessed for electrophysiological procedures [40, 41] and the delivery of drugs  or genes  (Figure 4C). The pericardial space is accessible to both surgical  and percutaneous delivery. Because alternatives to cardiothoracic surgery are preferred for routine delivery, recent studies have focused on the percutaneous approach. Relative to more invasive approaches, it has proven to be a feasible and comparatively safe procedure . The pericardial sac is punctured via a substernal/xiphoidal approach most often, although transatrial [45, 46] access has also been demonstrated. The percutaneous puncture is less invasive, and causes less scar tissue and fibrosis relative to the transatrial approach. Thus, if indicated, repeated procedures are possible with the percutaneous approach. However, despite its minimally invasive nature, there still remains a risk for major complications, including hemothorax, cardiac perforation with cardiac tamponade, and injury to the abdominal organs .
In the concept of a closed space, the absence of circulatory blood flow not only allows high concentration delivery, but also longer durations of agent exposure within the pericardium. Indeed, the pericardial injection correlates with high transduction efficiency in epicardial tissue with low distribution to other nontargeted organs [6, 48]. Thus, the percutaneous pericardial approach is a minimally invasive and effective delivery method for therapeutic agents targeting the epicardium. However, despite close proximity to the myocardium, only minimal myocardial expression is observed after pericardial gene delivery [43, 49]. Injecting various pharmacological agents together with the vector can increase the transduction of cardiomyocytes. The addition of collagenase and hyaluronidase to the vector was shown to increase the rate of transfected myocardium by up to 40% in rodents .
Altogether, the percutaneous pericardial injection is an attractive approach, depending on the proportion of targeted cardiac tissue. Despite providing potent epicardial transduction and the potential of transduction-enhancing agents that increase the permeability towards the myo- and endocardium, a transmural gradient remains that requires consideration. In addition, percutaneous cannulation requires experience, especially in cases with no pericardial effusions.
Delivery methods used in clinical studies
A number of cardiovascular gene therapy trials have been conducted or are currently ongoing [22, 36, 50-55]. Although various delivery methods are tested in the preclinical studies, antegrade intracoronary injection [52-54] and direct intramyocardial injection [36, 50, 51, 55] are the most widely chosen techniques in the clinical studies. Although direct comparison of efficacy between the two methods is difficult as a result of the use of study specifics (i.e. vectors and doses), successful results are reported in both delivery methods [52-54, 56]. Initially, direct intramyocardial injection was performed through thoracotomy or during coronary artery bypass surgery [55, 57]. However, open chest surgery results in some adverse events related to the procedure  and is relatively invasive for patients with severe illness. To overcome these drawbacks, an endovascular approach has become more popular with successful injections [50, 51]. The direct injection procedure itself appears to be safe with no reported major adverse events other than peri-procedural premature ventricular contractions [51, 57]. Transient increases in serum transaminases are reported after intracoronary delivery of adenoviral particles containing a gene encoding fibroblast growth factor ; however, no other major complication from this injection method has been reported. As a result of the targeting of patients with severe cardiac diseases, identifying whether adverse events are related to the injection method remains challenging. However, the effect of injection methods on long-term outcome should be always taken into account . Unlike preclinical studies, detecting gene expression in remote organs is difficult; thus, carefully monitoring nontargeted organs and long-term monitoring for tumor growth are essential for safety evaluation.
One remaining central challenge in gene therapy for heart failure is to establish delivery methods that are safe and efficacious. Currently, every method of gene transfer presents a compromise where both features are balanced. However, safety has priority in the clinic. When conducting a cardiac gene therapy in clinical studies, patient safety, adequate delivery to the myocardium, and simplicity of the procedure in a catheterization laboratory setting need to be evaluated . Advancements in the bioengineering of new, more cardiotropic vectors, in combination with further developments in delivery methods, will provide successful cardiac gene therapies with clinical significance.
Conflicts of interest statement
Dr Roger Hajjar is scientific co-founder of Celladon, where he has a major interest.