Cardiac sympathetic hyperinnervation after myocardial infarction: a systematic review and qualitative analysis

Abstract

Background

Cardiac sympathetic hyperinnervation after myocardial infarction (MI) is associated with arrhythmogenesis and sudden cardiac death. The characteristics of cardiac sympathetic hyperinnervation remain underexposed.

Objective

To provide a systematic review on cardiac sympathetic hyperinnervation after MI, taking into account: (1) definition, experimental model and quantification method and (2) location, amount and timing, in order to obtain an overview of current knowledge and to expose gaps in literature.

Methods

References on cardiac sympathetic hyperinnervation were screened for inclusion. The included studies received a full-text review and quality appraisal. Relevant data on hyperinnervation were collected and qualitatively analysed.

Results

Our literature search identified 60 eligible studies performed between 2000 and 2022. Cardiac hyperinnervation is generally defined as an increased sympathetic nerve density or increased number of nerves compared to another control group (100%). Studies were performed in a multitude of experimental models, but most commonly in male rats with permanent left anterior descending (LAD) artery ligation (male: 63%, rat: 68%, permanent ligation: 93%, LAD: 97%). Hyperinnervation seems to occur mainly in the borderzone. Quantification after MI was performed in regions of interest in µm²/mm² (41%) or in percentage of nerve fibres (46%) and the reported amount showed a great variation ranging from 439 to 126,718 µm²/mm². Hyperinnervation seems to start from three days onwards to >3 months without an evident peak, although studies on structural evaluation over time and in the chronic phase were scarce.

Conclusions

Cardiac sympathetic hyperinnervation after MI occurs mainly in the borderzone from three days onwards and remains present at later timepoints, for at least 3 months. It is most commonly studied in male rats with permanent LAD ligation. The amount of hyperinnervation differs greatly between studies, possibly due to differential quantification methods. Further studies are required that evaluate cardiac sympathetic hyperinnervation over time and in the chronic phase, in transmural sections, in the female sex, and in MI with reperfusion.

ABSTRACT

KEY MESSAGES

• Cardiac sympathetic hyperinnervation occurs three days after MI mainly in the borderzone and remains present at all timepoints.

• It is most commonly studied in male rats with permanent LAD ligation.

• The amount of hyperinnervation differs greatly between studies, possibly due to the differential quantification methods.

Keywords:

• Autonomic nervous system
• cardiac innervation
• myocardial infarction
• reperfusion therapy
• hyperinnervation
• systematic review

Introduction

Myocardial infarction (MI) remains the leading cause of sudden cardiac death in the Western world, despite advancements in prophylactic, percutaneous and surgical therapies [1]. In patients with MI, the cumulative incidence of sudden cardiac death has been estimated at 5.4% and accounted for 30.5% of all deaths due to cardiovascular causes [2]. Sudden cardiac death can be mainly attributed to ventricular arrhythmias, including ventricular tachycardia and fibrillation [3]. Although this is generally attributed to heterogeneous conduction in the MI border zone [4], cardiac innervation also plays an important role in its pathophysiology [5].

The cardiac autonomic nervous system with its sympathetic and parasympathetic branches is essential in maintaining cardiac homeostasis [6]. MI may lead to cardiac autonomic dysfunction contributing to ventricular arrhythmias and sudden cardiac death [7–9]. Changes in cardiac innervation after MI comprise acute sympathetic denervation followed by sympathetic hyperinnervation, which is often regarded as an early reaction, though the exact extent and timeframe remain ambiguous [10–13].

According to the location and severity of ventricular scars that are caused by MI, the myocardium can roughly be classified into different zones, that encompass the infarct core, borderzone (also known as peri-infarct zone) and the remote zone [14]. Nerve fibres in the myocardium seem to regenerate after ischemic injuries creating a greater nerve density in the left ventricle [15]. This observation has often been reported on the borderzone [16–18], whereas other reports describe this phenomenon in the infarct core or remote zone [11,13,19,20]. The fact that nerves can regain their capacity for growth after cardiac damage is by itself intriguing. Likely, factors secreted by the myocardium or epicardium provide a trigger for reinitiation of neural outgrowth [21,22]. To solve the mechanism of sympathetic hyperinnervation after cardiac damage, experimental models in a variety of species have been developed in recent years.

MI in animal models can be induced permanently, such as permanent ligation of a coronary artery, but can also include reperfusion therapy (MI-R) using a temporary ligation or balloon catheter [23]. Sympathetic hyperinnervation has been described in both permanent MI as well as in MI-R [20]. The latter is considered to be of high relevance as currently most MI patients receive acute reperfusion therapy. In the past years, increasing attention has been gained for the role of sex differences in the outcome after MI [24–27]. In addition, differences in autonomic function between men and women have been established [28]. This raises the awareness that sex should be taken into account in experimental models; however, the extent to which this currently occurs is unclear.

From a clinical point of view, neuromodulation seems a promising additional treatment option in cardiac arrhythmia management, but significant knowledge gaps in basic cardiac neurophysiology and the need for more and larger patient studies limit the success of these therapies [1,29]. Although numerous reports have observed sympathetic hyperinnervation after MI, a detailed overview focusing on the specific characteristics of sympathetic hyperinnervation, such as the timing and the relevance of sex, is lacking.

Here, we aim to provide a state-of-the art systematic literature review of cardiac sympathetic hyperinnervation after MI, focusing on the identification of consensus as well as gaps in knowledge on this intriguing topic.

Methods

Research questions

In order to systematically review the current literature on cardiac sympathetic hyperinnervation after MI(-R), our research questions were defined as follows:

1. How is cardiac sympathetic hyperinnervation defined in the literature?

2. In which experimental models and species is hyperinnervation studied? Is equal data present in males and females?

3. In what location(s) of the heart does sympathetic hyperinnervation occur?

4. How much cardiac sympathetic hyperinnervation occurs? How is innervation quantified?

5. What is the timeline of occurrence of cardiac hyperinnervation after MI?

Search strategy

This systematic review was conducted in the databases PubMed, Embase and Web of Science up to 19 July 2022 and was in adherence with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines [30], using the same workflow as in previous systematic reviews from our group [6,31,32]. The systematic review was registered before analysis and data extraction at the Center for Open Science (https://osf.io/2yex6/). The search strategy was conducted by using keywords for sympathetic innervation and MI; the full search strategy for each respective database can be viewed in the Supplementary Materials (Appendix A).

Selection criteria

After removal of duplicates, papers were considered eligible for inclusion when cardiac sympathetic hyperinnervation after MI or MI-R was studied. Histological studies performed in the myocardium of the left ventricle in humans and all other species in postnatal stages were included. In humans, ischemic cardiomyopathy was also considered eligible for review. Studies reporting results only from the healthy heart, exclusively from prenatal humans/animals, or in vitro experiments were excluded. In addition, reviews, editorials and articles of which the full-text could not be retrieved were excluded. Finally, non-ischemic denervation methods and MI induced by substance injection and non-English articles were excluded. Reference lists of reviews were searched for eligible articles, which were not identified in the literature search.

Quality appraisal

All records were screened for eligibility by two independent authors (H.S.C. and L.V.R.), based on the titles and abstracts. Discrepancies between the observer’s judgements were resolved by discussion and consensus. Selected papers underwent full-text review. The methodological quality of the papers was scored (H.S.C. and L.V.R.) using the ARRIVE guidelines 2.0 (Animal Research: Reporting In Vivo Experiments), a checklist of recommendations to improve the reporting of research involving animals [33]. After comparison and discussion of the individual scores, a consensus score was achieved for each paper included in this review.

Data analysis

Relevant data from the included studies were extracted (H.S.C.) in piloted forms when available. Outcome measures included: staining information (staining, type, location), section handling (fixation, embedding, thickness), quantification method, amount of hyperinnervation, number of animals and the experimental model (coronary artery, timing, species, sex, age/weight). In order to assess the amount of hyperinnervation in various studies in a uniform way, only studies reporting the nerve density were presented in scatterplots in µm²/mm². Nerve density values reported in percentage of stained nerves over the total tissue area could be transformed in µm²/mm² and were included in the scatterplots. Different groups within the studies are separately represented in the scatterplots. Data from all other outcome measures were presented in barcharts. Figures were created with RStudio (Boston, MA, Version 1.4.1717).

Results

Study selection

A total of 486 records were identified (Figure 1). Two hundred and ninety records were duplicates and the number of excluded studies by abstract screening was 139; most of these reported results lacked histological quantification of sympathetic nerves in the left ventricle. The remaining 57 articles were screened for eligibility based on the full-text. After full-text assessment, three records were excluded and six additional records were found through cited references. A total of 60 studies, published between 2000 and 2022, were included in this systematic review (Table 1).

Figure 1. PRISMA flow diagram.

Table 1. List of included studies.

Author (ref.)	Title
Cao et al. [34]	Nerve sprouting and sudden cardiac death
Kaye et al. [35]	Reduced myocardial nerve growth factor expression in human and experimental heart failure
Lai et al. [36]	Colocalization of tenascin and sympathetic nerves in a canine model of nerve sprouting and sudden cardiac death
Pak et al. [16]	Mesenchymal stem cell injection induces cardiac nerve sprouting and increased tenascin expression in a Swine model of myocardial infarction
Lin et al. [37]	Autologous transplantation of bone marrow mononuclear cells improved heart function after myocardial infarction
Swissa et al. [38]	Long-term subthreshold electrical stimulation of the left stellate ganglion and a canine model of sudden cardiac death
Zhou et al. [11]	Mechanisms of cardiac nerve sprouting after myocardial infarction in dogs
Hasan et al. [13]	Sympathetic hyperinnervation and inflammatory cell NGF synthesis following myocardial infarction in rats
Jiang et al. [39]	Relationship between sympathetic nerve sprouting and repolarization dispersion at peri-infarct zone after myocardial infarction
Jiang et al. [40]	Effects of metoprolol on sympathetic remodeling and electrical remodeling at infarcted borderzone after myocardial infarction in rabbits
Lee et al. [41]	Effect of pravastatin on sympathetic reinnervation in postinfarcted rats
Lee et al. [19]	Physiological concentration of 17beta-estradiol on sympathetic reinnervation in ovariectomized infarcted rats
Kang et al. [42]	Effect of ATP-sensitive potassium channel agonists on sympathetic hyperinnervation in postinfarcted rat hearts
Wernli et al. [43]	Macrophage depletion suppresses sympathetic hyperinnervation following myocardial infarction
Yuan et al. [44]	A novel peptide ghrelin inhibits neural remodeling after myocardial infarction in rats
Lee et al. [45]	Effect of N-acetylcysteine on sympathetic hyperinnervation in post-infarcted rat hearts
Wen et al. [18]	Carvedilol ameliorates sympathetic nerve sprouting and electrical remodeling after myocardial infarction in rats
Xin et al. [17]	Favourable effects of resveratrol on sympathetic neural remodeling in rats following myocardial infarction
Zhang et al. [46]	Alteration of parasympathetic/sympathetic ratio in the infarcted myocardium after Schwann cell transplantation modified electrophysiological function of heart
Lee et al. [47]	Differential effects of NADPH oxidase and xanthine oxidase inhibition on sympathetic reinnervation in postinfarct rat hearts
Gu et al. [48]	Artemisinin suppresses sympathetic hyperinnervation following myocardial infarction via anti-inflammatory effects
Lee et al. [49]	Effect of sildenafil on ventricular arrhythmias in post-infarcted rat hearts
Wang et al. [50]	Risk of ventricular arrhythmias after myocardial infarction with diabetes associated with sympathetic neural remodeling in rabbits
Wu et al. [51]	Desipramine pretreatment improves sympathetic remodeling and ventricular fibrillation threshold after myocardial ischemia
Yang et al. [52]	Transplantation of adipose tissue-derived stem cells overexpressing heme oxygenase-1 improves functions and remodeling of infarcted myocardium in rabbits
Ajijola et al. [53]	Focal myocardial infarction induces global remodeling of cardiac sympathetic innervation: neural remodeling in a spatial context
Drobysheva et al. [54]	Cardiac sympathetic innervation and PGP9.5 expression by cardiomyocytes after myocardial infarction: effects of central MR blockade
Gardner and Habecker [20]	Infarct-derived chondroitin sulfate proteoglycans prevent sympathetic reinnervation after cardiac ischemia–reperfusion injury
Jia et al. [55]	Aliskiren ameliorates sympathetic nerve sprouting and suppresses the inducibility of ventricular tachyarrhythmia in postinfarcted rat heart
Lorentz et al. [12]	Sympathetic denervation of peri-infarct myocardium requires the p75 neurotrophin receptor
Mias et al. [56]	Cardiac fibroblasts regulate sympathetic nerve sprouting and neurocardiac synapse stability
Pellegrino and Habecker [57]	STAT3 integrates cytokine and neurotrophin signals to promote sympathetic axon regeneration
Wang et al. [58]	Metoprolol-mediated amelioration of sympathetic nerve sprouting after myocardial infarction
Wang et al. [59]	Myocardial infarction induces sympathetic hyperinnervation via a nuclear factor-κB-dependent pathway in rabbit hearts
Chen et al. [60]	Aerobic exercise inhibits sympathetic nerve sprouting and restores β-adrenergic receptor balance in rats with myocardial infarction
Chen et al. [61]	Mesenchymal stem cells enhanced cardiac nerve sprouting via nerve growth factor in a rat model of myocardial infarction
Hu et al. [62]	Targeted NGF siRNA delivery attenuates sympathetic nerve sprouting and deteriorates cardiac dysfunction in rats with myocardial infarction
Lee et al. [63]	Differential effect of phosphodiesterase-3 inhibitors on sympathetic hyperinnervation in healed rat infarcts
Lee et al. [64]	Dipeptidyl peptidase-4 inhibition attenuates arrhythmias via a protein kinase A-dependent pathway in infarcted hearts
Lee et al. [65]	Inhibition of glycogen synthase kinase-3β prevents sympathetic hyperinnervation in infarcted rats
Lee et al. [66]	Sitagliptin attenuates sympathetic innervation via modulating reactive oxygen species and interstitial adenosine in infarcted rat hearts
Hu et al. [67]	Semaphorin 3A attenuates cardiac autonomic disorders and reduces inducible ventricular arrhythmias in rats with experimental myocardial infarction
Lee et al. [68]	Sitagliptin decreases ventricular arrhythmias by attenuated glucose-dependent insulinotropic polypeptide (GIP)-dependent resistin signalling in infarcted rats
Lee et al. [69]	Effects of urate-lowering agents on arrhythmia vulnerability in post-infarcted rat hearts
Tu et al. [70]	Long noncoding NONRATT021972 siRNA normalized abnormal sympathetic activity mediated by the upregulation of P2X7 receptor in superior cervical ganglia after myocardial ischemia
Yang et al. [71]	Semaphorin 3a transfection into the left stellate ganglion reduces susceptibility to ventricular arrhythmias after myocardial infarction in rats
Yang et al. [72]	Atorvastatin attenuates sympathetic hyperinnervation together with the augmentation of M2 macrophages in rats postmyocardial infarction
Yin et al. [73]	Inhibition of Notch signaling pathway attenuates sympathetic hyperinnervation together with the augmentation of M2 macrophages in rats post-myocardial infarction
Gao et al. [74]	Targeted P2X7 R shRNA delivery attenuates sympathetic nerve sprouting and ameliorates cardiac dysfunction in rats with myocardial infarction
Lee et al. [75]	Inhibition of infarction-induced sympathetic innervation with endothelin receptor antagonism via a PI3K/GSK-3β-dependent pathway
Lee et al. [76]	Targeting the pathway of GSK-3β/nerve growth factor to attenuate post-infarction arrhythmias by preconditioned adipose-derived stem cells
Yin et al. [77]	P2X7 receptor inhibition attenuated sympathetic nerve sprouting after myocardial infarction via the NLRP3/IL-1β pathway
Yokoyama et al. [78]	Quantification of sympathetic hyperinnervation and denervation after myocardial infarction by three-dimensional assessment of the cardiac sympathetic network in cleared transparent murine hearts
Lee et al. [79]	Effect of proton pump inhibitors on sympathetic hyperinnervation in infarcted rats: Role of magnesium
Yu et al. [80]	Evaluation of specific neural marker GAP43 and TH combined with Masson-trichrome staining for forensic autopsy cases with old myocardial infarction
Wang et al. [81]	Targeting blockade of nuclear factor-κB in the hypothalamus paraventricular nucleus to prevent cardiac sympathetic hyperinnervation post myocardial infarction
Lee et al. [82]	Taurine alleviates sympathetic innervation by inhibiting NLRP3 inflammasome in postinfarcted rats
Zhao et al. [83]	Vagus nerve stimulation in early stage of acute myocardial infarction prevent ventricular arrhythmias and cardiac remodeling
Lee et al. [84]	17β-Oestradiol facilitates M2 macrophage skewing and ameliorates arrhythmias in ovariectomized female infarcted rats
Wang et al. [85]	EphrinB2-RhoA upregulation attenuates sympathetic hyperinnervation and decreases the incidence of ventricular arrhythmia after myocardial infarction

ATP: adenosine triphosphate; GAP43: growth-associated protein 43; GSK-3β: glycogen synthase kinase-3 beta; NADPH: nicotinamide adenine dinucleotide phosphate; NGF: nerve growth factor; PGP: protein gene product; shRNA: short hairpin RNA; siRNA: small interfering RNA; STAT3: signal transducer and activator of transcription 3; TH: tyrosine hydroxylase.

Quality assessment

An appraisal of methodological quality was performed for all included studies. Seven studies scored lower than 60%, 40 studies scored between 60 and 69% and 13 studies scored between 70 and 79% (Figure 2(A)). Assessment of the quality across the different domains revealed predominantly high scores for study design, outcome measures, experimental procedures, abstract, background, objectives and ethical statement (Figure 2(B)). Incomplete reports accounted for the moderate scores in the domains of sample size, inclusion and exclusion criteria, randomization and results. Results were usually interpreted in the light of current theory and relevant studies in the literature, but study limitations were missing not only as a separate section, but in the entire text in nearly half of the studies. The domain ‘animal care and monitoring’ received low scores due to the use of sedative treatment without sufficient analgesic effect during surgery. Data on blinding, housing and husbandry, protocol registration and data access were lacking completely in many papers.

Figure 2. Quality appraisal. (A) Detailed quality assessment according to the 21 ARRIVE Guidelines 2.0 items. (B) Summary of the quality appraisal according to the 21 ARRIVE Guidelines 2.0 item.

Definition of cardiac sympathetic hyperinnervation

In all included papers, cardiac sympathetic hyperinnervation was defined as an ‘increased sympathetic nerve density or increased number of sympathetic nerves compared to another group’. This other group was a sham-operated control group in the majority of studies (49 out of 60 studies; 82%). Of the 11 studies without a sham-operated control group, seven studies had a regular control group. Four of the included studies did not have a control group, as they were the control group themselves for other experimental therapies [34,40,52,61]. The most commonly used types of staining were diaminobenzidine (50%) and immunofluorescence (48%) (Figure 3(A)). The remaining study used histofluorescence (2%, method using sucrose–potassium phosphate–glyoxylic acid). In nearly all studies, sympathetic nerve staining was performed using antibodies against tyrosine hydroxylase (55/56 studies) or dopamine β-hydroxylase (2/56 studies). Non-sympathetic markers used in the included records often consisted of growth-associated protein 43 (GAP43, nerve sprouting) and neurofilament (general nerve staining) (Supplementary Table 1). A detailed overview per article is presented in Table 2.

Figure 3. Barplots. *Acute and chronic MI [80] classified as 1 week and >3 months, respectively.

Table 2. General information.

Author (ref.)	Control group	Additional information control group	Sympathetic nerve staining	Type of staining	Coronary artery	Induction method	Species	Sex	Additional information experimental group	Tissue sample location
Cao et al. [34]	None	–	TH	Diaminobenzidine	Mid LAD	MI	Dog	NR	MI with complete AV block	NR
Kaye et al. [35]	SHAM	–	Catecholamines	Histofluorescencea	Proximal LAD	MI	Rat	NR	–	RZ
Lai et al. [36]	NC	–	TH	Diaminobenzidine	Mid LAD	MI	Dog	NR	–	IC, BZ, RZ
Pak et al. [16]	NC	–	TH	Diaminobenzidine	Distal LAD	MI	Swine	NR	Intramyocardial culture medium injection	BZ
Lin et al. [37]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI	Rabbit	NR	–	BZ, RZ
Swissa et al. [38]	NC and SHAM	Electrical stimulation of the LSG	TH	Diaminobenzidine	Mid LAD	MI	Dog	Female	MI with complete AV block, electrical stimulation of the left stellate ganglion or NGF infusion into the LSG	NR
Zhou et al. [11]	NC	–	TH	Diaminobenzidine	Mid LAD	MI	Dog	NR	MI induction with intracoronary balloon inflation or epicardial ligation	RZ
Hasan et al. [13]	NC and SHAM	–	TH, DBH	Immunofluorescence	Proximal LAD	MI	Rat	Female	–	BZ
Jiang et al. [39]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI	Rabbit	Both	–	BZ, RZ
Jiang et al. [40]	None	Second thoracotomy for electrophysiological measurements during sympathetic nerve stimulation	TH	Diaminobenzidine	Proximal LAD	MI	Rabbit	Both	Second thoracotomy for electrophysiological measurements during sympathetic nerve stimulation	BZ
Lee et al. [41]	SHAM	–		Diaminobenzidine	Proximal LAD	MI	Rat	Male	Vehicle via gastric gavage	RZ
Lee et al. [19]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI	Rat	Female	–	RZ
Kang et al. [42]	SHAM	Saline via gastric gavage	TH	Diaminobenzidine	LAD, not specified	MI	Rat	Male	Saline via gastric gavage	RZ
Wernli et al. [43]	SHAM	Untied suture around artery, saline or PBS saline tail vein injections	TH, DBH	Immunofluorescence	Proximal LAD	MI	Rat	Female	Saline or PBS saline tail vein injections	BZ
Yuan et al. [44]	SHAM	Saline via subcutaneous injection	TH	Diaminobenzidine	LCA, not specified	MI	Rat	Male	Saline via subcutaneous injection	BZ, RZ
Lee et al. [45]	SHAM	Saline via oral gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline via oral gavage	RZ
Wen et al. [18]	SHAM	–	TH	Diaminobenzidine	Mid LAD	MI	Rat	Male	Saline via oral gavage	BZ
Xin et al. [17]	SHAM	Intraperitoneal injection of a vehicle	TH	Diaminobenzidine	Mid LAD	MI	Rat	Male	Intraperitoneal injection of a vehicle	BZ
Zhang et al. [46]	SHAM	–	TH	Immunofluorescence	Proximal LAD	MI	Rat	NR	Intramyocardial serum-free medium injection	NR
Lee et al. [47]	SHAM	Saline via oral gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline via oral gavage	RZ
Gu et al. [48]	SHAM	Vehicle via oral gavage	TH	Diaminobenzidine	Proximal LAD	MI	Rat	Male	Vehicle via oral gavage	BZ
Lee et al. [49]	SHAM	Vehicle via oral gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Vehicle via oral gavage	RZ
Wang et al. [50]	NC	–	TH	Diaminobenzidine	Mid LAD	MI	Rabbit	NR	Saline injection in the lateral ear vein	BZ, RZ
Wu et al. [51]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI-R	Rat	Male	–	BZ
Yang et al. [52]	None	–	TH	Diaminobenzidine	Mid LCX	MI	Rabbit	Male	Intramyocardial PBS injection	BZ
Ajijola et al. [53]	NC	–	TH	Diaminobenzidine	Mid LAD	MI	Swine	NR	Microspheres infusion during balloon catheter angioplasty	ICb, BZb, RZ
Drobysheva et al. [54]	SHAM	Intracerebroventricular cannulation surgery	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Intracerebroventricular vehicle infusion	BZ
Gardner and Habecker [20]	SHAM	–	TH	Immunofluorescence	LAD, not specified	Both	Mouse	Both	–	IC, BZ
Jia et al. [55]	SHAM	–	TH	Diaminobenzidine	LAD, not specified	MI	Rat	Male	–	BZ
Lorentz et al. [12]	SHAM	–	TH	Immunofluorescence	LAD, not specified	MI-R	Mouse	Both	–	IC, BZ
Mias et al. [56]	SHAM	–	TH	Immunofluorescence	LAD, not specified	MI	Rat	Male	–	BZ
Pellegrino and Habecker [57]	SHAM	–	TH	Immunofluorescence	Proximal LAD	MI-R	Mouse	Both	–	BZ
Wang et al. [58]	SHAM	Saline via oral gavage	TH	Diaminobenzidine	Mid LAD	MI	Rabbit	Male	Saline via oral gavage	BZ, RZ
Wang et al. [59]	SHAM	–	TH	Diaminobenzidine	Mid LAD	MI	Rabbit	Male	–	BZ, RZ
Chen et al. [60]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI	Rat	Male	–	BZ
Chen et al. [61]	None	–	TH	Diaminobenzidine	Proximal LAD	MI	Rat	Male	Intramyocardial PBS injection	BZ
Hu et al. [62]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI	Rat	Male	Intramyocardial 0.9% NaCl injection	BZ, RZ
Lee et al. [63]	SHAM	Vehicle administration (not specified)	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Vehicle administration (not specified)	RZ
Lee et al. [64]	SHAM	Saline via oral gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline via oral gavage	RZ
Lee et al. [65]	SHAM	Vehicle administration (not specified)	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Vehicle administration (not specified)	RZ
Lee et al. [66]	SHAM	Vehicle administration (not specified)	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Vehicle administration (not specified)	RZ
Hu et al. [67]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI	Rat	Male	–	BZ
Lee et al. [68]	SHAM	Saline via oral gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline via oral gavage	RZ
Lee et al. [69]	SHAM	Saline via oral gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline via oral gavage	RZ
Tu et al. [70]	NC and SHAM	–	TH	Diaminobenzidine	LAD, not specified	MI	Rat	NR	–	BZ
Yang et al. [71]	SHAM	–	TH	Diaminobenzidine	Proximal LAD	MI	Rat	Male	PBS injection into the left stellate ganglion	BZ, RZ
Yang et al. [72]	SHAM	–	TH	Immunofluorescence	LAD, not specified	MI	Rat	Male	PBS via gastric gavage	BZ
Yin et al. [73]	SHAM	DMSO tail vein injection	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	DMSO tail vein injection	BZ
Gao et al. [74]	SHAM	Intramyocardial DMSO injection	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Intramyocardial vehicle injection	BZ
Lee et al. [75]	SHAM	Saline via oral gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline via oral gavage	RZ
Lee et al. [76]	SHAM	–	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Intramyocardial PBS injection	RZ
Yin et al. [77]	SHAM	–	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Intraperitoneal vehicle injection	BZ
Yokoyama et al. [78]	NC	–	TH	Immunofluorescence	Mid LAD	MI	Mouse	Male	–	IC, BZ
Lee et al. [79]	SHAM	Saline administration (not specified)	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline administration (not specified)	BZ
Yu et al. [80]	NCc	–	TH	Diaminobenzidine	LAD, not specified	MI	Human	Both	–	BZ
Wang et al. [81]	SHAM	PBS injection into the hypothalamus	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	PBS injection into the hypothalamus	BZ
Lee et al. [82]	SHAM	Saline via gastric gavage	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Saline via gastric gavage	RZ
Zhao et al. [83]	SHAM	ICD implantation	TH	Diaminobenzidine	Mid LAD	MI	Dog	Both	ICD implantation, cervical vagal nerve stimulation	BZ
Lee et al. [84]	SHAM	–	TH	Immunofluorescence	Proximal LAD	MI	Rat	Female	–	RZ
Wang et al. [85]	SHAM	–	TH	Immunofluorescence	Proximal LAD	MI	Rat	Male	Intramyocardial vehicle injection (only 1 MI group)	BZ

AV: atrioventricular; BZ: borderzone; IC: infarct zone; ICD: implantable cardioverter-defibrillator; LAD: left anterior descending artery; LCX: circumflex artery; LSG: left stellate ganglion; MI: myocardial infarction; MI-R: myocardial infarction with reperfusion; NC: normal control; NR: not reported; PBS: phosphate-buffered saline; RZ: remote zone.

^a Method using sucrose-potassium phosphate-glyoxylic acid.

^b Infarct core and borderzone biopsies pooled.

^c Death due to non-cardiac cause.

Experimental model

Cardiac sympathetic hyperinnervation after MI was mostly performed in the rat (68%), followed by rabbit (12%), dog (8%) and mouse (7%) (Figure 3(B)). Larger species such as swine (3%) and human (2%) have only been sporadically studied. Male-only subjects were investigated in 63% of included studies (Figure 3(C)), 12% of studies had subjects from both sexes and only 8% had female-only subjects. The sex in 17% of studies remained unspecified. MI was induced exclusively in branches of the left coronary artery (LCA). The left anterior descending (LAD) artery was the preferred branch in the included studies (97%; 62% proximal LAD, 20% mid LAD, 1.5% distal LAD, 13.5% LAD unspecified) (Figure 3(D)). Another coronary artery branch used, was the mid left circumflex (LCX, 1.5%) artery. In 1.5% of the studies, the exact location of MI induction in the LCA was unspecified. The majority of MI inductions was permanent (93%), whereas in only 5% MI was followed by reperfusion and 2% of the included studies used both models (Figure 3(E)). In case of MI-R, reperfusion followed after 30 min.

A detailed overview per article is presented in Table 2. Additional information regarding the experimental model can be found in the Supplemental Material (Supplementary Table 1).

Location of cardiac sympathetic hyperinnervation

Cardiac sympathetic hyperinnervation was mainly determined in biopsies from the borderzone (43%) or remote zone (30%) (Figure 3(F), Table 2). In 13.5% of studies, biopsies from both the borderzone and remote zone were taken. Biopsies from the infarct core combined with borderzone were present in 5% of the studies. All different locations (infarct core, borderzone and remote zone) were investigated in 3.5% of the studies and in 5% the location was not described. Although the studies agreed on nomenclature of the regions in infarct core, borderzone and remote zone, no exact definitions of these areas were provided and the exact location of the tissue sample was not described (Supplementary Table 1).

Quantification methods

Quantification of innervation was performed in many cases using conventional software, such as Image-Pro Plus (67%) and ImageJ (15%) (Figure 3(G), Table 3). Tissue Studio, Imaris, Bioanalysis Image and AnalySIS were also used (1.5% each). Twelve percent of the studies did not specify which software was used. The majority of included studies did not specify the thickness of the sections (30%) (Figure 3(H), Table 3). The most common section thicknesses mentioned were 5 µm (44%) and 10 µm (15%), and other thicknesses mentioned fall more or less within this range. The amount of innervation was mostly defined by the percentage of stained nerves divided by the total area (46%). Instead of a percentage, some authors used µm²/mm² to calculate the amount of innervation (41%). The absolute number of nerves was used in 8% of studies and in 5% other ways of quantifying innervation were applied (Figure 3(I), Table 3). Of note, all quantifications were performed in regions of interest and none of the studies used transmural biopsies. In addition, quantification details differed greatly. In particular, the number of (blinded) investigators, the (random) selection of regions of interest and the number of selected fields varied between included studies (Table 3).

Table 3. Specification of quantification methods.

Author (ref.)	Unit	Software	Section thickness (µm)	Quantification method
Cao et al. [34]	Number of nerves/mm^2	NR	5	Blinded investigator, 3 fields with the highest number of nerves, ratio between the total number of nerves and the total area
Kaye et al. [35]	Number of catecholaminergic profiles per field	NR	NR	Number of stained catecholamine profiles counted in 5 separate zones
Lai et al. [36]	Number of nerves	NR	5	Visual assessment, not specified
Pak et al. [16]	Number of nerves/mm^2, heterogeneity/mm^2	NR	5	Manually counted number of nerves in the 5 fields with highest nerve densities (with a blood vessel in the centre of the field) divided by the total area; for heterogeneity: standard deviation of nerve densities in the 5 fields
Lin et al. [37]	µm^2/mm^2	Image-Pro Plus 4.0	4	3 selected fields, nerve area divided by the total area examined
Swissa et al. [38]	Number of nerves/mm^2, µm^2/mm^2	Image-Pro Plus 4.0	5	Number of nerve fibres per unit area or total area of nerve fibres per unit area measured by a computer-assisted image analysis system (SHAM) or manually (MI rats with complete atrioventricular block)
Zhou et al. [11]	µm^2/mm^2	Image-Pro Plus	5*	Blinded investigators, 3 fields with the highest nerve density, nerve area divided by the total area
Hasan et al. [13]	µm^2/mm^2	AnalySIS	10	Blinded investigator, 10 stepped serial sections per heart, 3 sample fields (0.136 mm^2) containing maximum innervation, innervation density measured by threshold discrimination
Jiang et al. [39]	µm^2/mm^2	Image-Pro Plus 3.0	NR	3 fields with the highest density of nerves, nerve area divided by the total area
Jiang et al. [40]	µm^2/mm^2	Image-Pro Plus 3.0	NR	3 fields with the highest density of nerves were selected, the area occupied by the nerves in the field were automatically calculated, nerve area divided by the total area
Lee et al. [41]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Lee et al. [19]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Kang et al. [42]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area, experiments were replicated 3 times
Wernli et al. [43]	% of the total area	ImageJ	10	6 fields per section evenly distributed captured with a ×40 objective, nerve density was calculated by superimposing a stereological grid (8 µm line interval) over the captured image, counting grid intersections over nerves and dividing by the total number of intersections over the sampled tissue. Vascular innervation was excluded from quantification.
Yuan et al. [44]	µm^2/mm^2	Image-Pro Plus 3.0	NR	3 fields with the highest nerve density, nerve area divided by the total area
Lee et al. [45]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Wen et al. [18]	µm^2/mm^2	Image-Pro Plus 4.0	NR	5 selected fields with the highest density of nerves, nerve area divided by the total area
Xin et al. [17]	µm^2/mm^2	Image-Pro Plus 4.0	NR	Nerve area divided by the total area
Zhang et al. [46]	µm^2/mm^2	Image-Pro Plus 6.0	5	2 Blinded investigators, 5 randomly selected fields of each section, percentage of positively staining cells per field (×40 field, 0.13 mm^2)
Lee et al. [47]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Gu et al. [48]	µm^2/mm^2	NR	NR	3 selected fields, nerve area divided by the total area
Lee et al. [49]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area, experiments were replicated 3 times
Wang et al. [50]	µm^2/mm^2	Image-Pro Plus 6.0	5	3 sections per animal, at least 3 areas with the highest nerve density for each section, nerve area divided by the total area
Wu et al. [51]	µm^2/mm^2	Image-Pro Plus 3.0	NR	3 randomly selected fields, nerve area divided by the total area
Yang et al. [52]	% of the total area	Image-Pro Plus 4.0	NR	Nerves only identified when longer than 10 µm and stained brown, nerve area divided by the total area
Ajijola et al. [53]	µm^2/mm^2	Tissue Studio	NR	NR
Drobysheva et al. [54]	% of the total area	ImageJ	10	6 sections from each animal located at least 200 µm apart, nerve area divided by the total area
Gardner and Habecker [20]	% of the total area	ImageJ	10	3 sections obtained from a similar level of the base to apex axis per heart, nerve area divided by the total area
Jia et al. [55]	µm^2/mm^2	Image-Pro Plus 6.0	5	10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Lorentz et al. [12]	% of the total area	ImageJ	10	2 independent observers, innervation densities from two locations averaged together, one representative infarct picture from each section chosen and averaged across the four sections from one heart, nerve area divided by the total area
Mias et al. [56]	Other (arbitrary units)	NR	6	3 fields, computerized image analysis, not further specified
Pellegrino and Habecker [57]	% of the total area	ImageJ	10	2 independent observers, fibre density in 3 sections at least 150 µm apart, nerve area divided by the total area
Wang et al. [58]	µm^2/mm^2	Image-Pro Plus 5.0	10	3 fields with the highest nerve density, for each rabbit 5 slides, nerve area divided by the total area
Wang et al. [59]	µm^2/mm^2	Image-Pro Plus 5.0	10	3 fields with the highest nerve density, for each rabbit 5 slides, nerve area divided by the total area
Chen et al. [60]	µm^2/mm^2	Image-Pro Plus 6.0	5	10 fields per section, 6 sections from each group, nerve area divided by the total area
Chen et al. [61]	µm^2/mm^2	Image-Pro Plus 6.0	NR	5 randomly-selected microscopic fields, 1 section per heart, nerve area divided by the total area
Hu et al. [62]	µm^2/mm^2	Image-Pro Plus 5.0	NR	Blinded investigator, 3 fields of 1–2 slides per animal, nerve area divided by the total area
Lee et al. [63]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Lee et al. [64]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Lee et al. [65]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Lee et al. [66]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Hu et al. [67]	µm^2/mm^2	Image-Pro Plus 5.0	NR	Nerve area divided by the total area
Lee et al. [68]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Lee et al. [69]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Tu et al. [70]	Other (integrated optical density)	Image-Pro Plus	15	Changes in integrated optical density, not further specified
Yang et al. [71]	µm^2/mm^2	Image-Pro Plus 6.0	5	3 fields with the highest nerve density, nerve area divided by the total area
Yang et al. [72]	µm^2/mm^2	Image-Pro Plus 5.0	10	Blinded investigators, nerve area divided by the total area
Yin et al. [73]	% of the total area	ImageJ (version 1.38x)	NR	Blinded investigator, 10 randomly selected fields in a ×20 field (0.13 mm^2), nerve area divided by the total area
Gao et al. [74]	% of the total area	ImageJ	NR	Blinded investigator, 10 randomly selected fields in a ×20 field (0.13 mm^2), nerve area divided by the total area
Lee et al. [75]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Lee et al. [76]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Yin et al. [77]	% of the total area	ImageJ (version 1.38x)	NR	10 randomly selected fields, four sections adjacent to the Masson’s trichrome-stained set was analysed, first section taken was 1.5–2 mm apical to the ligation, nerve area divided by the total area
Yokoyama et al. [78]	Other (nerve volume and nerve surface area)	Imaris	Entire heart	3 selected regions of interest 300 µm × 300 µm × 300 µm with maximal innervation, nerve volume and surface area measured by contrast discrimination, size criteria used to discount any nonspecific staining of <100 voxels
Lee et al. [79]	% of the total area	Image-Pro Plus	5	Blinded investigator, 10 randomly selected fields at ×400 magnification, nerve area divided by the total area
Yu et al. [80]	µm^2/mm^2	ImageJ	NR	2 blinded pathologists, 5 selected fields at ×400 magnification with the highest density of nerves, nerve area divided by the total area
Wang et al. [81]	% of the total area	Image-Pro Plus 5.0	7	3 fields with the highest nerve density, 5 slides for each subject, nerve area divided by the total area
Lee et al. [82]	% of the total area	Image-Pro Plus	5	10 randomly selected fields at ×400 magnification
Zhao et al. [83]	µm^2/mm^2	Bioanalysis Image Software	3	10 randomly selected fields, nerve area divided by the total area
Lee et al. [84]	% of the total area	NR	NR	10 random scans per section at ×400 magnification, nerve area divided by the total area
Wang et al. [85]	% of the total area	Image-Pro Plus 5.0	7	3 fields with the highest nerve density, 5 slides for each subject, nerve area divided by the total area

*sections from puncture

Amount and timeline of cardiac sympathetic hyperinnervation

A major part of the studies investigated cardiac sympathetic hyperinnervation within 2 months after MI, in particular after exactly 1 month (35%) and 1 week (19%) (Figure 3(J), Table 4). Timepoints later than 2 months were scarce (2.5 months 1%, ≥3 months 4%). A great variation was reported in the amount of cardiac sympathetic innervation after MI(-R), ranging from 439 to 126,718 µm²/mm² (Table 4, Figure 4). In order to assess the amount of innervation in various studies in an uniform manner, only studies reporting the nerve density in µm²/mm² or percentage of stained nerves over the total tissue area (transformed to µm²/mm²) were included for comparison (52/60 studies). In nearly, all studies that included a control group, hyperinnervation was present in the experimental subjects as the nerve density was higher compared to control from three days onwards (Table 4). However, there was a significant overlap in nerve densities of the experimental and control groups between the different studies (Figure 4). This contributes to the fact that no evident peak could be distinguished in the timing of cardiac sympathetic hyperinnervation, but hyperinnervation remained present also at later timepoints. Hyperinnervation was observed mostly in the infarct core and borderzone from MI(-R) subjects throughout the entire timespan (Figure 5). Of interest, hyperinnervation seems to be unrestricted to the area surrounding the infarction, as many remote zone biopsies also show hyperinnervation at 28 days. Subanalyses regarding the experimental model (MI versus MI-R, species, sex) and section location showed inconclusive results due to its vast heterogeneity (Figure 6(A–D)).

Figure 4. Nerve densities over time in all groups. Acute and chronic MI [80] classified as 1 week and >3 months, respectively, the average was taken of all control groups. The average from all timepoints included in the study was taken for the SHAM group as the exact timepoint was not mentioned [85]. When a range of numbers was mentioned, the lowest number was taken.

Figure 5. Nerve densities over time in MI(-R) subjects only categorized by tissue sample location. Acute-old and old MI [80] classified as 1 week and >3 months, respectively, the average was taken of all control groups. The average from all timepoints included in the study was taken for the SHAM group as the exact timepoint was not mentioned [85]. When a range of numbers was mentioned, the lowest number was taken.

Figure 6. Nerve densities per category. Subanalyses of nerve densities regarding: (A) the experimental model, (B) sex, (C) species and (D) tissue sample location.

Table 4. Nerve densities.

		Normal control		SHAM		MI(-R)
Author (ref.)	Timing	Amount of innervation	Number of subjects	Amount of innervation	Number of subjects	Amount of innervation	Number of subjects
Cao et al. [34]	3M	–	–	–	–	16.6 ± 5.3 nerves/mm^2	6
Kaye et al. [35]	8W	–	–	10.65 ± 2.53d,e profiles/field	8	3.16 ± 0.91^d^,^e profiles/field	8
Lai et al. [36]	35.7 ± 14.4D	40 nerves (visually normal area), 78 nerves (around blood vessels)	6	–	–	61 nerves (visually normal area) 2 nerves (infarct) 6 nerves (border) 35 (around blood vessels)	5
Pak et al. [16]	95.8 ± 3.5D	1.8 ± 1.1 nerves/mm^2, 1.3 ± 0.7 heterogeneity/mm^2	6	–	–	13.2 ± 16.4 nerves/mm^2 8.9 ± 12.4 heterogeneity/mm^2	6
Lin et al. [37]	2.5M	–	–	593 ± 352 µm^2/mm^2 (border) 650 ± 329 µm^2/mm^2 (remote)	5	587 ± 363 µm^2/mm^2 (border) 716 ± 438 µm^2/mm^2 (remote)	5
Swissa et al. [38]	41 ± 27D	26 ± 16 nerves/mm^2, 635 ± 339 µm/mm^2	6	93 ± 42 nerves/mm^2, 1657 ± 614 µm/mm^2	6	72.3 ± 22.6 nerves/mm^2, area of nerves NR	4
Zhou et al. [11]	3.5H, 3D, 1W, 1M	438.89 ± 109.72 µm^2/mm^2d	NR	X	X	551.29 ± 195.36 µm^2/mm^2 (3.5H)^d 620.87 ± 58.88 µm^2/mm^2 (3D)^d 845.67 ± 144.51 µm^2/mm^2 (1Wa)^d 861.73 ± 128.46 µm^2/mm^2 (1Wb)^d 829.62 ± 321.14 µm^2/mm^2 (1M)^d	3 (3,5H), 3 (3D^a), 2 (3D^b), 5 (1W^a), 5 (1W^b), 3 (1M)
Hasan et al. [13]	1D, 4D, 7D, 2W, 28D	255 ± 50 µm^2/0.136 mm^2 field^d^,f	5	255 ± 50 µm^2/0.136 mm^2 field^d^,^f	5	344 ± 60/0.136 mm^2 field^d^,^f (1D) 1253 ± 530/0.136 mm^2 field^d^,^f (4D) 1904 ± 617/0.136 mm^2 field^d^,^f (7D) 1167 ± 264/0.136 mm^2 field^d^,^f (2W) 524 ± 103/0.136 mm^2 field^d^,^f (28D)	4–5 per timepoint
Jiang et al. [39]	8W	–	–	1053 ± 256 µm^2/mm^2^d	10	2617 ± 526 µm^2/mm^2^d	20
Jiang et al. [40]	8W	–		–	–	2616 ± 528 µm/mm^2	20
Lee et al. [41]	4W	–	–	0.01 ± 0.02%^d	10	0.92 ± 0.23%	9
Lee et al. [19]	4W	–	–	0.02 ± 0.01%^d	10	0.12 ± 0.05%^d	9
Kang et al. [42]	4W	–	–	0.03 ± 0.02%^d	10	0.14 ± 0.07%	11
Wernli et al. [43]	7D	–	–	0.84 ± 0.07^d^,^e	4c,d	1.78 ± 0.21^d^,^e	9^d
Yuan et al. [44]	4W	–	–	738 ± 157 µm^2/mm^2 (BZ) 721 ± 151 µm^2/mm^2 (RZ)	12	2693 ± 464 µm^2/mm^2 (BZ) 2330 ± 458 µm^2/mm^2 (RZ)	15
Lee et al. [45]	4W	–	–	0.04 ± 0.03%^d	10	0.32 ± 0.14%	12
Wen et al. [18]	8W	–	–	2104 ± 305 µm^2/mm^2^d	20	7784 ± 483 µm^2/mm^2^d	21
Xin et al. [17]	4W	–	–	766 ± 130 µm^2/mm^2	8	6430 ± 455 µm^2/mm^2	12
Zhang et al. [46]	2W	–	–	996 ± 226 µm^2/mm^2	5	2529 ± 117 µm^2/mm^2	5
Lee et al. [47]	4W	–	–	0.05 ± 0.02%^d	10	0.28 ± 0.15%	10
Gu et al. [48]	4W	–	–	539.00 ± 62.72 µm^2/mm^2	10	3113.25 ± 402.61 µm^2/mm^2	20
Lee et al. [49]	4W	–	–	0.01 ± 0.01%^d	12	0.56 ± 0.17%	12g
Wang et al. [50]	8W	2485 ± 175 µm^2/mm^2 (BZ), 2465 ± 177 µm^2/mm^2 (RZ)	8	–	–	3453 ± 201 µm^2/mm^2 (BZ) 3205 ± 156 µm^2/mm^2 (RZ)	10
Wu et al. [51]	7D	–	–	12717.5 ± 4703.0 µm^2/mm^2	10	126717.5 ± 19900.3 µm^2/mm^2	10
Yang et al. [52]	42D	–	–	–	–	0.23 ± 0.27%	6
Ajijola et al. [53]	6W	8046 ± 1922 µm^2/mm^2^d^,^e	9	–	–	19,984 ± 4177 µm^2/mm^2^d^,^e (IC and BZ) 7571 ± 1828 µm^2/mm^2^d^,^e (RZ)	9
Drobysheva et al. [54]	10D	–	–	0.11 ± 0.01%e (endocardial) 0.82 ± 0.08%^e (epicardial)	6–7	1.63 ± 0.11%^e	6–7
Gardner and Habecker [20]	10D, 20D	–	–	2.21 ± 0.15%^d^,^e^,^f	4 (10D)	0.29 ± 0.20% (MI-R 10D BZ)^d^,^e 1.88 ± 0.09% (MI-R 10D IC)^d^,^e 1.84 ± 0.08% (MI-R 20D BZ)^d^,^e 0.13 ± 0.02% (MI-R 20D IC)^d^,^e 1.31 ± 0.19% (MI 10D BZ)^d^,^e 1.67 ± 0.06% (MI 10D IC)^d^,^e	4 (MI-R 10D), 4 (MI 10D), 4 (MI-R 20D)
Jia et al. [55]	6W	–	–	382.5 ± 23.5 µm^2/mm^2^e	9	747.8 ± 75.1 µm^2/mm^2^e	8
Lorentz et al. [12]	24H, 3D, 7D	–	–	1.26 ± 0.07%^d,e,h	4–8	0.88 ± 0.05% (24H proximal BZ)^d^,^e 1.25 ± 0.10% (24H distal BZ)^d^,^e 0.90 ± 0.12% (3D proximal BZ)^d^,^e 1.55 ± 0.11% (3D distal BZ)^d^,^e 0.87 ± 0.14% (1W proximal BZ)^d^,^e 0.96 ± 0.21% (1W distal BZ)^d^,^e	4–8
Mias et al. [56]	15D	–	–	17.53 ± 3.96 AU^d^,^e	3–6	200.79 ± 21.49 AU^d^,^e	3–6
Pellegrino and Habecker [57]	24H, 3D	–	–	0.71 ± 0.21%^d^,^e^,^h	4–6	0.65 ± 0.10% (24H)^d^,^e 1.30 ± 0.24% (3D)^d^,^e	4–6
Wang et al. [58]	7D	–	–	2401 ± 238 µm^2/mm^2 (BZ) 2378 ± 254 µm^2/mm^2 (RZ)	10	3492 ± 230 µm^2/mm^2 (BZ) 3175 ± 210 µm^2/mm^2 (RZ)	13
Wang et al. [59]	7D	–	–	2339 ± 135 µm^2/mm^2 (BZ) 2291 ± 143 µm^2/mm^2 (RZ)	10	4103 ± 120 µm^2/mm^2 (BZ) 3501 ± 142 µm^2/mm^2 (RZ)	20
Chen et al. [60]	8W	–	–	225 ± 30 µm^2/mm^2^d	6	1985 ± 191^d µm^2/mm^2^d	6
Chen et al. [61]	8W	–	–	–	–	1430.48 ± 952.50 µm^2/mm^2	10
Hu et al. [62]	5W	–	–	1318 ± 134 µm^2/mm^2 (BZ) 1339 ± 164 µm^2/mm^2 (RZ)	17	3458 ± 440 µm^2/mm^2 (BZ) 2591 ± 232 µm^2/mm^2 (RZ)	18
Lee et al. [63]	4W	–	–	0.09 ± 0.05%^d	10	0.21 ± 0.12%^d	12
Lee et al. [64]	4W	–	–	0.09 ± 0.03%^d	10	0.23 ± 0.11%	11
Lee et al. [65]	4W	–	–	0.07 ± 0.03%^d	10	0.39 ± 0.15%	11
Lee et al. [66]	4W	–	–	0.11 ± 0.04%^d	10	0.25 ± 0.10%	12
Hu et al. [67]	5W	–	–	1281 ± 75.1 µm^2/mm^2	15	3589 ± 278.1 µm^2/mm^2	9
Lee et al. [68]	4W	–	–	0.09 ± 0.10%^d	12	0.32 ± 0.13%	11
Lee et al. [69]	4W	–	–	0.05 ± 0.02%^d	10	0.48 ± 0.21%	11
Tu et al. [70]	30D	2218.75 ± 263.39 IOD^d^,^e	6	2571.43 ± 223.21 IOD^d^,^e	6	5473.21 ± 357.14 IOD^d^,^e	6
Yang et al. [71]	2W	–	–	612.16 ± 143.72 µm^2/mm^2^d (BZ) 556.27 ± 165.02 µm^2/mm^2^d (RZ)	5	2315.55 ± 519.00 µm^2/mm^2^d (BZ) 1929.63 ± 425.85 µm^2/mm^2^d (RZ)	5
Yang et al. [72]	7D	–	–	548.38 ± 32.16 µm^2/mm^2	10	3483.65 ± 108.23 µm^2/mm^2	10
Yin et al. [73]	7D	–	–	0.05 ± 0.02%^d	8–10	0.42 ± 0.14%	8–10
Gao et al. [74]	7D	–	–	0.06 ± 0.02%^d	8–10	0.39 ± 0.11%^d	8–10
Lee et al. [75]	1M	–	–	0.10 ± 0.01%^d^,^e	10	0.30 ± 0.05%^e	11
Lee et al. [76]	1M	–	–	0.10 ± 0.04%^d	5	^d0.52 ± 0.08%	5
Yin et al. [77]	7D	–	–	0.05 ± 0.02%^d	8–10	^d0.45 ± 0.09%	8–10
Yokoyama et al. [78]	2W	373.01 ± 72.78 × 10^3 mm^3^d^,^e (nerve volume) 343.64 ± 52.73 × 10^3 mm^2^d^,^e (nerve surface area)	5	–	–	0 × 10^3 mm^3^d^,^e (nerve volume, IC) 1103.87 ± 117.41 × 10^3 mm^3^d^,^e (nerve volume, BZ), 0 × 10^3 mm^2^d^,^e (nerve surface area, IC), 416.66 ± 63.85 × 10^3 mm^2^d^,^e (nerve surface area, BZ)	5
Lee et al. [79]	4W	–	–	0.14 ± 0.05%^d^,^e	10	1.42 ± 0.24%^e	10
Yu et al. [80]	Acute-old and old MIc	10,087 ± 2519 µm^2/mm^2 (infectious shock) 13,546 ± 3387 µm^2/mm^2 (amniotic fluid embolism) 7875 ± 4723 µm^2/mm^2 (bronchopneumonia) 12,100 ± 4773 µm^2/mm^2 (head injuries) 15,987 ± 5473 µm^2/mm^2 (death from hanging) 12,870 ± 2850 µm^2/mm^2 (death by fall from height)	4 4 9 9 4 4	–	–	76,450 ± 22,205 µm^2/mm^2 (old MI) 68,150 ± 19,392 µm^2/mm^2 (acute-old MI)	12 (8M/4F) 12 (8M, 4F)
Wang et al. [81]	7D	–	–	0.05 ± 0.03%^d	35	0.43 ± 0.05%	40
Lee et al. [82]	4W	–	–	0.08 ± 0.02%^d	10	0.42 ± 0.06%	10
Zhao et al. [83]	3W	–	–	2143 ± 612 µm^2/mm^2^d	5	38,180 ± 18,725 µm^2/mm^2	6
Lee et al. [84]	4W	–	–	0.13 ± 0.03%^d	10	1.29 ± 0.17%^d	10
Wang et al. [85]	3D, 7D, 14D, 28D	–	–	0.03 ± 0.01%^d^,^e^,c	4	0.21 ± 0.04%^d^,^e (MI, 3D) 0.43 ± 0.03%^d^,^e (MI, 7D) 0.39 ± 0.05%^d^,^e (MI, 14D) 0,29 ± 0,07%^d^,^e (MI, 28D) 0.42 ± 0.05%^d^,^e (MI, 14D) 0.24 ± 0.03%^d^,^e (MI + vehicle, 14D)	4 4 4 4 6 6

AU: arbitrary units; BZ: borderzone; D: days; H: hours; IC: infarct core; IOD: integrated optical density; M: months; RZ: remote zone; W: weeks.

^a MI induction by ligation.

^b MI induction by intracoronary balloon catheter.

^c Timepoint unclear.

^d Estimated from figure.

^e Mean ± standard error.

^f Normal control and SHAM groups pooled for analysis.

^g Probably, 12 (as all groups consisted out of 12 in the supplemental material of the study).

^h Timepoints pooled.

Discussion

This systematic review found that cardiac sympathetic hyperinnervation in MI(-R):

1. is generally defined as an increased sympathetic nerve density or increased number of sympathetic nerves compared to control or sham;

2. is studied in a multitude of experimental models, but most commonly in male rats with permanent LAD ligation;

3. seems to occur mainly in the borderzone;

4. is usually quantified in regions of interest in µm²/mm² or percentage using conventional software;

5. varies greatly in the reported amount ranging from 439 to 126,718 µm²/mm²;

6. seems to be present after MI at all timepoints from three days onwards without an evident peak.

Standardization is required for the identification of regions of interest and definition of the tissue sample location

Many similarities exist in current literature in the method by which cardiac hyperinnervation is quantified (i.e. tyrosine hydroxylase as sympathetic marker, the focus on regions of interest and the determination of nerve density by dividing the sympathetic stained nerve area by the total area in µm²/mm² or percentage). However, it may be challenging to standardize the regions of interest for quantification. Some have tried to overcome this by either averaging multiple regions, taking multiple randomly chosen regions or the regions with highest nerve density (Table 3). In many studies, the investigators were additionally blinded to the origin of the histological sections, but due to the damaged cardiac tissue structure after MI(-R), it can be relatively easy by experienced investigators to distinguish experimental subjects from control or sham. Furthermore, a major factor that could account for the challenging comparison in nerve densities between studies are the definitions of the infarct core, borderzone and remote zone. These definitions, although in generally not specified in detail, are relevant, as in a MI-R mouse model, hyperinnervation was only observed in the distal borderzone after three days, but not in the proximal borderzone [12]. In a permanent MI rat model, Western blot analysis revealed that tyrosine hydroxylase expression was increased significantly at the base of the heart, but was decreased in the left ventricle below the MI area [86]. Despite the fact that it is unclear if tyrosine hydroxylase expression assessed with Western blot directly represents histological hyperinnervation, this finding suggests that differences might exist in various locations of the left ventricle after MI. While the borderzone is often described as the region adjacent to the myocardial scars and the remote zone as non-infarcted myocardium, in the majority of cases no location specifications are provided. When location specifications are described, these specifications are contradictory between studies. The borderzone has been defined as ‘within 3 mm of the marginal zone between infarct zone and non-infarct zone’ [39,55], ‘myocardium extending 0.5–1.0 mm from the infarct scar’ [44], ‘approximately 500–900 µm from the infarct’ [57], ‘450 or 520 µm from the edge of the infarct’ [12] and ‘approximately 2 mm around the infarcted area’ [74], irrespective of the type of species. The remote zone has been defined as ‘>2 mm outside of the infarct’ [19,42,45], but also ‘>2 cm away from the infarct site’ [50]. The margins have been defined by visual identification of discolouration [61]. However, it is technically difficult to identify the margins and take biopsies at exactly these sites. These differences between studies may account for the great variation in reported nerve densities and the contradictory reports of the presence of hyperinnervation in the infarct core and remote zone. In addition, as early reperfusion therapy after MI results in smaller and inhomogeneous fibrotic scars in the subendocardium extending towards the midmyocardium [87], classifications into infarct core, borderzone and remote zone may require other specifications for its definition.

Besides standardizing the regions of interest, another possibility to overcome these quantificational challenges could be the classification of biopsies based on the percentage of fibrosis and innervation quantification in the entire transmural tissue sections [88], which has been demonstrated in the porcine heart, but could easily be performed in other species. This would also allow to determine whether there is an ‘innervation gradient’ from epicardial towards endocardial areas, or vice versa. In addition, an issue on the assessment of the nerve density in the myocardium is that nerve fibres are very delicate, making it difficult to discriminate nerve fibres from artefacts. Some methodological adjustments could be included in future experiments to diminish the influence of similar looking structures, such as artefacts or iron deposits [88]. Examples are semi-automated quality controls or concealment by other stainings, e.g. Prussian Blue staining to conceal iron deposits in macrophages. With emerging technology, albeit with tissue clearing and in small-sized hearts, it is currently possible to visualize and analyse myocardial nerves in high-resolution in three dimensions [89]. However, standardization of regions of interest and specification of the tissue sample location are still relevant in the histological research of larger hearts, such as human hearts.

Although selective treatment of ventricular nerves have not been studied, epicardial nerves could be of interest as they are larger than myocardial nerves and would therefore be easier to analyse and better accessible for clinical treatment [90–92]. Moreover, it has recently been found that neurite outgrowth and density increases in the presence of epicardial derived cells by a paracrine effect, highlighting the importance of the epicardium [21].

Models of MI(-R) show several experimental gaps and in particular studies performed in females are underrepresented

The vast majority of studies investigated cardiac sympathetic hyperinnervation in male rats with permanent LAD ligation. Subanalyses regarding the experimental model (MI versus MI-R, species, sex) were inconclusive due to the small number of studies. Therefore, differences between experimental models may exist, warranting that conclusions need to be interpreted with caution and cannot per definition be extrapolated to other experimental models. When MI and MI-R models were included within the same study, at 10 days after permanent LAD occlusion, the nerve density in the infarct core and borderzone were the same, while 10 days after MI-R the nerve density was higher in the borderzone than in the denervated infarct core [20]. Permanent LAD ligation commonly produces large infarctions and therefore many studies included relatively large infarctions >30% of the left ventricle [73,79,82]. As currently most MI patients receive acute reperfusion therapy, future studies in MI-R models are mandatory. In addition, studies including female animals currently are largely underrepresented. Only six out of 56 (11%) of studies performed experiments on subjects from both sexes (Table 2). The number of each sex included was however not reported and the sexes were pooled for analysis. In the four studies performed in females only, bilateral ovariectomy was performed to minimize the influence of reproductive hormones [13,19,43]. This underrepresentation of female experimental models is problematic, as marked sex differences exist in autonomic innervation [28,93,94]. Numerous gene expression differences have been found between males and females, including genes encoding neuropeptides, and female hearts have been show to contain higher neurotransmitter content than male hearts [95]. Furthermore, it has been reported that women showed a significantly higher sympathetic activity in the left ventricular apex than male and that this sympathetic activity increases with age in women [96]. Besides differences in innervation, the responses to MI also seem to differ between the sexes [97,98]. The extent and localization of pain have been reported to vary between women and men [99,100]. Moreover, women have a lower risk of developing ventricular fibrillation during MI compared with men and are less likely to receive cardiac interventions than men [101].

Cardiac sympathetic hyperinnervation occurs after MI(-R) from three days onwards and remains present at later timepoints

No evident peak at a certain timepoint after MI(-R) could be identified in our qualitative analysis and cardiac sympathetic hyperinnervation seems to be present at all timepoints. It is important to highlight that this relies on data from different studies with a very heterogeneous group of MI(-R) subjects. Most studies that included multiple timepoints show hyperinnervation from three days onwards, although the findings of whether and when a peak occurs differ [12,13,57,80,85]. It has been reported that hyperinnervation peaks at seven days [13,85], while others report hyperinnervation without a transient development over time [11,80]. One study performed in MI-R mice differed in its findings [20] as compared to the other studies included, and reported that at 10 and 20 days (no other timepoints investigated) the nerve density in the borderzone was similar to sham.

Association with clinical arrhythmias

Ventricular arrhythmias, contributing to sudden cardiac death, may occur at any time after MI, starting from the onset of ischemia to the late chronic period [102,103]. Specific associations between type and timing of ventricular arrhythmias and applied treatment strategies are vague as multiple mechanisms are involved [102]. Myocardial scar constitutes a substrate for ventricular arrhythmias [4]. As the substrate is fixed and continually present, a trigger is needed that initiates the process culminating in arrhythmias, such as sympathetic activation. Interestingly, in a study performed in native hearts of transplant recipients, including coronary artery disease, the sympathetic cardiac nerve density directly correlated with the occurrence of life-threatening ventricular arrhythmia [104]. Therefore, studying sympathetic hyperinnervation and its timing according to standardized methods is important to clarify the dynamics of neural remodelling after MI. In addition to the anatomical analysis of innervation, functional insight into the autonomic nervous system is essential to fully understand the changes after MI. At a functional level, a balance between norepinephrine (NE) synthesis, release and reuptake exists in healthy sympathetic nerves fibres. MI disrupts this balance and results in a paradoxical elevation of cardiac extracellular NE [105]. These high levels of extracellular NE contribute to the risk of cardiac arrhythmias and increased mortality [106]. Besides sympathetic changes, MI can also lead to parasympathetic dysfunction, cholinergic transdifferentiation of cardiac sympathetic nerves and neural remodelling of the cardiac afferent pathway [107–111]. Therefore, research on anatomical and functional aspects of cardiac innervation could provide new insights into arrhythmogenesis after MI, and provide clues on which types of arrhythmias/arrhythmia mechanisms might be triggered by (changes in) autonomic state.

Clinical implications

Neuromodulation is an emerging therapeutic approach in the treatment of ventricular arrhythmias. These interventions include electrical vagal stimulation, spinal cord interventions, stellate ganglion modulatory therapies, interventions on the intrinsic cardiac nervous system and renal denervation [112]. An in-depth understanding of the characteristics of cardiac sympathetic hyperinnervation can aid in the optimization and timing of these neuromodulatory techniques. Currently, case selection for likely responders is limited by our lack of understanding of the cardiac autonomic nervous system [29]. Therefore, it is crucial to close the gaps in literature.

Neurotrophic factors have been proposed for clinical use in several organ systems [113–115]. A prediction model of patients at risk for life-threatening arrhythmias due to sympathetic hyperinnervation does not exist yet. Knowing the timing and location of hyperinnervation may eventually contribute to the identification of neurotrophic biomarkers that are involved in signalling cascades responsible for cardiac sympathetic hyperinnervation.

Limitations

In the analyses on nerve density, only values reported in µm²/mm² or percentage could be assessed. Additionally, as the timepoints and number of subjects were not always clear, an approximation was used in these cases. The heterogeneity of the included studies did not allow us to perform formal statistical analyses.

Conclusions and gaps in literature

Cardiac sympathetic hyperinnervation occurs after MI(-R) mainly in the borderzone and remains present at all timepoints from three days onwards. It is most commonly studied in male rats with permanent LAD ligation. The amount differs greatly between studies, possibly due to the various quantification methods. This review has exposed significant gaps in literature and therefore warrants: (1) studies in females; (2) studies in large animals, including human; (3) studies using transmural biopsies; (4) studies using MI-R as experimental model; and (5) studies evaluating the chronic phase (>2 months) after MI-R. Information from these studies will optimize our understanding of the phenomenon of sympathetic hyperinnervation, and its correlation to adverse arrhythmogenic events after MI-R, which is necessary for risk stratification, prevention and, ultimately, therapy.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Additional information

Funding

This work was supported by the Netherlands Organization for Scientific Research (NWO) (91719346 to MJ).