Incidence and Management of Access Site Infection Following Trans‐Axillary Implant of a Microaxial Left Ventricular Assist Device

ABSTRACT

Background

The Impella 5.5 is microaxial left ventricular assist device (LVAD) implanted most commonly into the axillary artery (AxA). This study aims to investigate the incidence and management of vascular graft infection (VGI) following circulatory support with the Impella 5.5.

Methods

This is a single‐center retrospective analysis of 92 patients who were supported by Impella 5.5 between June 2020 and August 2024. Since November 2023, Dacron graft has been tunneled through the pectoralis major muscle via a separate incision in patients expected to require prolonged support. The graft was trimmed and buried beneath the pectoralis major muscle during explant. Support period was classified as short (≤ 14 days, Group SS) or extended (> 14 days, Group ES).

Results

Of the 92 patients, 26 (28.3%) died while on support, Impella was explanted in 66 patients (71.7%): Bridge‐to‐recovery in 36 (39.1%), bridge‐to‐durable LVAD in 8 (8.7%), and bridge‐to‐transplant in 22 (23.4%) including 5 heart and kidney transplant (HKT). Median support duration was 7 [5, 8] days in Group SS (N = 54) and 27.5 [18, 38] days in Group ES (N = 38). No infections were reported in Group SS. Four patients in Group ES who underwent HKT experienced VGI following 39, 47, 51, and 85 days of support. Two patients with tunneled grafts required patch repair of AxA, while the remaining two patients with non‐tunneled grafts required extensive extra‐anatomical bypass.

Conclusion

Approximately 10% of patients requiring extended support developed VGI. Heart transplant recipients with a high risk of infection such as extended support or HKT will benefit from a comprehensive approach for early diagnosis and therapeutic intervention.

Vascular graft infection following trans‐axillary Impella 5.5 implant is rare overall but occurs with notable frequency in patients requiring prolonged support, especially immunosuppressed transplant recipients. Vigilant surveillance and multidisciplinary care are critical to improving outcomes in this high‐risk population.

1. Background

Temporary mechanical circulatory support (tMCS) devices are being increasingly used to manage cardiogenic shock, decompensated heart failure, and postcardiotomy shock [1]. Their application has expanded into high‐risk cardiac surgery, percutaneous coronary interventions, transcatheter valve therapies, and catheter ablations. These devices are also used as a bridge to decision, recovery, or definitive therapies such as durable left ventricular assist device (LVAD) implant or heart transplantation (HT).

The Impella 5.5 with Smart Assist System (Abiomed, Danvers, MA) is an advanced tMCS device featuring a microaxial flow pump capable of delivering up to 5.5 L/min, effectively unloading the left ventricle [2, 3, 4]. The device is commonly implanted via a Dacron graft anastomosed end‐to‐side to the axillary artery (AxA) [2, 5, 6]. Prolonged support longer than FDA‐approved 14 days is increasingly common, particularly for patients awaiting transplantation or with delayed myocardial recovery due to its safety profile [7, 8].

Complications of Impella 5.5 implant include upper‐extremity neuropathy, ipsilateral ischemia, hematoma, thromboembolism represented by stroke, hemolysis, device malfunction, and wound infection including vascular graft infection (VGI) [9, 10, 11, 12]. VGI is a serious complication that may result in critical bleeding, pseudoaneurysm formation, thromboembolism, sepsis, or death [13, 14]. The mechanisms of VGI are multifactorial: (i) the close proximity of the Dacron graft to the skin, (ii) prolonged support duration, (iii) multiple manipulations of the wound during repositioning or removal, and (iv) contamination of the remaining Dacron graft after Impella removal. This risk is especially relevant in heart transplant recipients, who are vulnerable to infection due to low‐output status prior to transplantation and immunosuppressive therapy post‐transplant [15, 16]. Despite the increasing use of Impella 5.5, data on the incidence, risk factors, and management of VGI remain limited, particularly among patients undergoing heart transplantation after extended device support [13, 17, 18]. This information is especially essential where waiting times for patients with support on Impella have been increasing in the current era [19, 20, 21].

This study aims to evaluate the incidence, risk factors, clinical presentation, and surgical management of VGI following trans‐axillary Impella 5.5 implant, with particular emphasis on extended‐term support and post‐transplant outcomes.

2. Methods

We conducted a single‐center, retrospective study of patients who underwent trans‐axillary Impella 5.5 implant with subsequent explant at the time of cardiac recovery, transplantation, or durable LVAD placement between July 2020 and August 2024. Institutional review board approval was obtained, with informed consent waived (protocol number: 12413).

Patients were identified through surgical logs and electronic medical records. Demographics, comorbidities, device indication, support duration, operative details, and clinical outcomes after explant were collected. Cultures, histopathology, and microbiology reports were reviewed in cases with VGI.

Duration of support was classified as short (≤ 14 days) or extended (> 14 days) in accordance with FDA recommendations. Standard post‐transplant immunosuppression includes triple therapy with tacrolimus (target level: 8–12 ng/mL), mycophenolate mofetil, and corticosteroid (typically 20–30 mg range at the time of discharge).

2.1. Impella Implant

Surgical implantation was performed under general anesthesia via an infraclavicular incision to expose the second portion of the AxA. The pectoralis major muscle was divided by electrocautery, but the pectoralis minor muscle was spared. An 8‐ or 10‐mm Dacron graft was anastomosed end‐to‐side to the artery.

In the early study period, the graft was externalized through the same incision. After we had experienced two VGI patients who required extra‐anatomical bypass and extensive vascular repair, we transitioned to the tunneling technique; the Dacron graft was tunneled through the pectoralis major muscle and routed to the skin, laterally and caudally. The graft was cut as short as possible outside the skin. The primary infraclavicular wound was completely closed. This technique was selectively employed for patients expected to require prolonged support or transplantation. Occlusive dressing was applied to cover the main incision and the exit site of the Dacron graft. The dressing was assessed daily and changed twice a week. Postoperative antibiotics were discontinued after 24 h.

2.2. Impella Explant

Explant was performed in the operating room for those bridged to LVAD or transplantation and either in the operating room or at bedside under local anesthesia in patients with myocardial recovery. For non‐tunneled grafts, the previous incision was reopened, the graft clamped and divided, ligated as short as possible, and buried under the pectoralis muscle. For tunneled grafts, two techniques were utilized: (i) the graft was ligated at the exit site, and the stump was buried under two pectoralis muscles, (ii) the main incision was partially opened, the graft was radically shortened and ligated, and the distal segment of the graft was removed from the exit site (Figure 1). The first technique was utilized especially following extended support. Complete graft excision was not routinely performed unless gross infection was present.

FIGURE 1.

Surgical steps of Impella 5.5 explant. (A) Impella 5.5 was implanted with graft tunneling. (B) The main incision was partially opened, and the graft was exposed. (C) The Impella 5.5 was removed. (D) The graft was ligated and cut as short as possible and was removed from the exit site. [Color figure can be viewed at wileyonlinelibrary.com]

2.3. Infection Definition and Management

VGI was defined according to Management of Aortic Graft Infection Collaboration (MAGIC) criteria [13, 14], including local signs (erythema, drainage, wound dehiscence, exposure of the Dacron graft), systemic findings (fever, leukocytosis), culture results or radiographic evidence (fluid collection, pseudoaneurysm) on either ultrasound or computed tomography with or without contrast. Emergent surgical exploration was performed whenever infection was diagnosed or suspected. The primary wound was opened, and the old Dacron graft was exposed. Depending on the quality of arterial tissue, in situ repair or extra‐anatomical vascular reconstruction was performed. For in situ repair, a bovine pericardial patch was used to fill the defect after explant of the Dacron graft (Figure 2). For extra‐anatomical bypass, the right common carotid to brachial artery bypass was performed with cadaveric vein graft and the axillary artery was ligated proximally and distally (Figure 3).

FIGURE 2.

In situ repair of the axillary artery. (A) Proximal and distal control of the axillary artery, dissection of the graft conduit. (B) Excision of the graft and debridement of the arterial edge. (C) Patch repair of the artery with a bovine pericardial patch. [Color figure can be viewed at wileyonlinelibrary.com]

FIGURE 3.

Illustration of extra‐anatomical bypass (Right common carotid to brachial artery). [Color figure can be viewed at wileyonlinelibrary.com]

3. Statistical Analysis

Continuous data are reported as mean ± standard deviation and as median and were compared with the Mann–Whitney U test. Categorical variables are reported as counts and percentages. p values less than 0.05 were considered statistically significant. Statistical analyses were performed using JMP 17 (JMP Statistical Discovery LLC, Cary, NC).

4. Results

A total of 92 patients underwent Impella implant during the study period (Figure 4). Twenty‐six patients (28.3%) died while on Impella support, and remaining 66 patients underwent Impella explant: 36 (39.1%) were weaned, 8 (8.7%) were bridged to durable LVAD, 22 patients (23.4%) were bridged to transplant including heart transplant (HT, 17 [18.5%]), and 5 (5.4%) to combined heart and kidney transplant (HKT, 5 [5.4%]). Fifty‐four patients (58.6%) had short support (median 7 [5, 9] days), while the remaining 38 patients (41.4%) required extended support (median 27.5 [18,38] days).

FIGURE 4.

Flow diagram of patients undergoing Impella 5.5 implantation. [Color figure can be viewed at wileyonlinelibrary.com]

The median age at implant was 62 years [48, 69] in the short‐term support group and 58 years [49,65] in the extended support group (p = 0.53) (Table 1). The mean body mass index was 27.7 [24.8,32.5] in the short‐term group and 28.9 [25.0, 30.8] in the extended support group (p = 0.54). There were no statistically significant differences between groups in baseline characteristics, including rates of diabetes and chronic kidney disease. Preoperative use of intra‐aortic balloon pump (IABP) was more common in the extended support group.

TABLE 1.

Baseline characteristics of the patients.

Variables	Total (n = 92)	Short (n = 54)	Extended (n = 38)	p
Age	61 [48.5,68]	62 [48,69]	58 [49,65]	0.53
Female	22 (23.9)	13 (24.1)	9 (23.7)	0.96
Hypertension	48 (52.2)	28 (51.9)	20 (52.6)	0.94
Hyperlipidemia	44 (47.8)	25 (46.3)	19 (50.0)	0.72
Diabetes	30 (32.6)	16 (29.6)	14 (30.7)	0.46
Ischemic etiology	37 (40.2)	18 (44.4)	19 (50.0)	0.11
Peripheral vascular disease	13 (14.1)	7 (13.0)	6 (15.8)	0.70
Chronic kidney disease	28 (30.4)	15 (27.7)	13 (34.2)	0.51
Body mass index	28.0 [24.8,32.5]	27.7 [24.7,33.1]	28.9 [25.0, 30.8]	0.54
Preoperative MCS
IABP	41 (62.1)	25 (46.2)	28 (73.7)	0.01
ECMO	22 (33.3)	16 (29.6)	8 (21.1)	0.36
Indication of Impella
Acute decompensated heart failure	50 (62.1)	25 (46.2)	25 (65.8)
Acute coronary syndrome	25 (27.3)	18 (33.3)	7 (18.4)
Postcardiotomy shock	6 (6.1)	4 (7.4)	2 (5.2)
Ventricular tachycardia	4 (4.5)	2 (3.7)	2 (5.2)
Myocarditis	1 (1.5)	1 (1.9)	0 (0)
High risk surgery/procedure	6 (12.1)	4 (7.4)	2 (5.2)

Abbreviations: ECMO, extracorporeal membrane oxygenation; IABP, Intra‐aortic balloon pump; MCS, mechanical circulatory support.

Impella was implanted via the right AxA in 71 patients (77.2%), with the remainder via the left (Table 2). The median diameter of the AxA was 8.3 [7.5,9.1] mm. Laterality and size of AxA were similar between the two groups. VGI occurred in 4 patients (4.3% of the total population), exclusively in the extended support group, with a significantly higher incidence compared with the short‐term group (10.5% [4/38] vs. 0% [0/54], p = 0.01). No statistically significant differences were observed in other complications, including access site bleeding, neurologic events, arm ischemia, device malfunction, or need for repositioning.

TABLE 2.

Perioperative outcomes of the patients.

Variables	Total (n = 92)	Short (n = 54)	Extended (n = 38)	p
Technical success	92 (100)	54 (100)	38 (100)
Usage of right axillary artery	71 (77.2)	40 (74.1)	31 (81.6)	0.40
Axillary artery diameter (mm)	8.3 [7.5,9.1]	8.5 [7.6,9.4]	8.0 [7.0,9.8]	0.68
Support duration (day)	12 [8,26]	7 [5,10.0]	27.5 [17.0,33.8]	0.01
Destination of management
Bridged to LVAD	8 (8.7)	5 (9.3)	3 (7.9)	0.82
Bridged to HT	17 (18.5)	7 (13.0)	10 (26.3)	0.10
Bridged to HKT	5 (5.4)	0 (0)	5 (13.2)	0.01
Recovery	36 (39.1)	24 (44.4)	12 (31.5)	0.21
Death on support	26 (28.3)	19 (35.2)	7 (18.4)	0.08
Complication
Graft infection	4 (4.3)	0 (0)	4 (10.5)	0.01
Access site bleeding	4 (4.3)	2 (3.7)	2 (5.3)	0.74
Neurologic events	4 (4.3)	2 (3.7)	2 (5.3)	0.74
Arm ischemia	1 (2.2)	1 (1.9)	0 (0)	0.40
Device malfunction	4 (4.3)	1 (1.4)	3 (7.9)	0.12
Reposition	6 (6.5)	2 (3.7)	4 (10.5)	0.14

Abbreviations: HKT, heart and kidney transplant; HT, heart transplant; LVAD, left ventricular assist device.

All 4 patients who developed VGI were male, with a mean age of 47.5 years (range 37–56) and mean BMI of 28.7 kg/m² (range 22.6–32.4) (Table 3). All had chronic kidney disease, three were on chronic hemodialysis, and two had type 2 diabetes mellitus. Two patients required peripheral venoarterial extracorporeal membrane oxygenation support prior to Impella implant. All patients underwent Impella placement via the right AxA, and one patient had technically challenging access due to an AxA diameter < 7 mm. The mean duration of Impella support was 55 days (range 39–85). All four patients were ultimately bridged to combined HKT, accounting for 18.2% of all patients who underwent heart or combined HKT (n = 22).

TABLE 3.

Patient information with graft infection.

	Patient 1	Patient 2	Patient 3	Patient 4
Sex	Male	Male	Male	Male
Age	37	44	56	53
Body mass index (kg/m2)	22.6	30.8	32.4	28.9
Blood type	O (+)	O (+)	O (+)	O (+)
Indication	ADHF/NICM	ADHF/NICM	ADHF/NICM	ADHF/NICM
Diabetes	No	No	yes	Yes
Dialysis	Yes	Yes	No	Yes
Preoperative ECMO	Yes	Yes	No	No
Graft side/size	Right, 10 mm	Right, 10 mm	Right, 10 mm	Right, 10 mm
Challenging delivery	Yes (small diameter)	No	No	No
Graft tunneling	No	No	Yes	Yes
Postoperative bleeding	No	No	No	No
Device exchange	Twice	No	No	Once
Bridging	HKT	Reop HKT	HKT	HKT
Graft explant location	Operating room	Operating room	Operating room	Operating room
Duration of Impella Support (days)	39	51	85	47
Timing of graft infection (days after explant)	32	48	0	67
Symptom at diagnosis	Purulent discharge	Abscess formation	Purulent discharge	Protrusion of graft stump
Timing of graft infection (days after implant)	71	99	85	114
Culture results	S. aureus , VRE, K. pneumoniae	P. aeruginosa	P. acnes , P. mirabilis	P. aeruginosa
Drug resistance	Yes	No	No	No
Number of vascular procedures	2	3	1	1
Initial vascular procedure	Debridement, patch repair	Debridement, patch repair	Debridement, patch repair	Debridement, patch repair
Additional procedure	Reinfection, Extra‐anatomical bypass	Reinfection, Extra‐anatomical bypass Graft stenosis, Angioplasty and stenting	None	None
Follow‐up	Death 24 months (non‐compliance)	Alive 24 months	Alive 22 months	Alive 14 months
Recurrent infection	None	None	None	None

Abbreviations: ADHF, acute decompensated heart failure; ECMO, extracorporeal membrane oxygenation; HKT, heart and kidney transplant; K. pneumoniae , Klebsiella pneumoniae ; NICM, non‐ischemic cardiomyopathy; P. acnes , Propionibacterium acnes ; P. aeruginosa , Pseudomonas aeruginosa ; P. mirabilis , Proteus mirabilis ; S. aureus , Staphylococcus aureus ; VRE, vancomycin resistant enterococcus.

VGI was diagnosed intraoperatively at the time of heart transplant in one patient and at 32‐, 48‐, and 67‐days post‐transplant in the other three. Two patients presented with purulent wound drainage, while the remaining two presented with protrusion of graft stamp with abscess formation. Wound cultures revealed Pseudomonas aeruginosa (n = 2), Staphylococcus aureus with Klebsiella pneumoniae and vancomycin‐resistant Enterococcus (n = 1), and Propionibacterium acnes with Proteus mirabilis (n = 1). Two patients had undergone Impella exchange (once and twice, respectively) due to device malfunction.

5. Outcomes of VGI

All four patients underwent surgical wound debridement, complete excision of the remnant graft, and bovine pericardial patch repair of the AxA. Two patients developed recurrent infection and required additional resection of the infected segment of the AxA with extra‐anatomical bypass from the right common carotid artery to the proximal brachial artery. Both patients requiring bypass did not have tunneling of the Dacron graft at the time of initial implant. No infection‐related mortality occurred. One patient developed graft stenosis of the extra‐anatomical bypass 12 months later for which angioplasty and stenting were performed (Figure 5). There were no associated complications such as pseudoaneurysm formation or ipsilateral limb ischemia after vascular repair. At mean follow‐up of months, all wounds were fully healed without evidence of infection.

FIGURE 5.

Angioplasty and stenting following extra‐anatomical bypass (Right common carotid to brachial artery). (A) Pre‐procedural angiogram revealed stenosis of the cadaveric vein graft. (B) Post‐procedure angiogram following balloon angioplasty and stenting of the vein graft. [Color figure can be viewed at wileyonlinelibrary.com]

6. Discussion

In this single‐center retrospective study, we found that VGI following trans‐axillary Impella 5.5 implant was uncommon overall (4.3%), but was significantly more frequent among patients requiring prolonged support (> 14 days) with an incidence of 10.5% in this subgroup. VGI exclusively occurred in patients bridged to heart and kidney transplant, and this accounted for 18.2% of all patients who underwent heart or heart and kidney transplant, and in 26.7% of all transplanted patients requiring extended support with Impella. Notably, no infections were observed in patients supported short term regardless of bridging strategies following Impella explant.

The pathogenesis of vascular graft infection (VGI) following Impella implantation is multifactorial, driven by mechanical, anatomical, and immunological vulnerabilities. First, the transcutaneous configuration of the device predisposes the access site to early bacterial colonization. Repetitive micro‐movement of the catheter during early patient ambulation causes continuous micro‐trauma at the exit site. This disrupts local healing, prevents complete epithelialization, and facilitates the direct downward tracking of normal skin flora along the catheter conduit. Second, the creation of surgical dead space, combined with the requisite systemic anticoagulation, frequently results in subpectoral hematoma or seroma formation, which provides an ideal, nutrient‐rich culture medium for pathogen proliferation. Third, the profound cardiogenic shock state preceding implantation typically induces a severe systemic inflammatory response syndrome. When coupled with splanchnic hypoperfusion, this state promotes bacterial translocation into the systemic circulation, establishing a vector for the hematogenous seeding of the Dacron graft. Finally, in the bridge‐to‐transplant population, the transition to post‐transplant immunosuppression and the use of broad‐spectrum antibiotics radically alter the host microbiome. This selects for highly virulent, opportunistic pathogens. If the synthetic graft is retained, it acts as an intractable nidus for multidrug‐resistant organisms that are largely refractory to systemic antimicrobial therapy.

7. Comparison With Prior Studies

The incidence of access‐site complications after Impella 5.5 implant varies widely in the literature, with infection rates ranging from 0% to 10.9% [2, 9, 11]. In the largest reported series, Lewin et al. described early infection in 2.9% and late infection in 10.9% of patients, all managed with graft excision and patch angioplasty [11] Our overall rate (4.3%) falls within this range, but patients requiring prolonged support had a distinctly higher risk. Unlike Lewin et al., who found no clear association between duration of support and infection, our findings suggest that prolonged support increases susceptibility, particularly in immunosuppressed transplant candidates. Differences in patient selection, implant technique, and institutional practices may explain these discrepancies.

In the early study period, our technique involved externalizing the graft through the primary incision due to its simplicity; however, the proximity of the anastomosis site from the semi‐open wound raises a concern of susceptibility to infection. After the experience of VGI requiring extra‐anatomical bypass for severe infection, we started to utilize the tunneling technique. This technique may reduce direct exposure of the anastomosis to the skin flora while on prolonged Impella 5.5 support.

For explant of the Impella which was exteriorized through the same incision, Bernhardt et al. described (a) ligating the graft, then burying it under the skin/pectoralis muscle without opening the incision, (b) radical shortening with partial wound reopening, or (c) complete graft removal with patch repair of the AxA [8]. For tunneled grafts, options include (d) ligation at the exit site and burying the stump, or (e) partially reopening the main incision, shortening and ligating the graft. Both techniques may have benefits and potential downsides.

8. Clinical Implications

Our findings highlight several implications. First, the incidence of VGI appears rare in short‐term use but rises significantly following prolonged support, especially in heart transplant recipients. Second, infections were frequently polymicrobial. Third, the tunneling technique might prevent the spread of superficial infection to deeper vascular structures. This is likely attributable to the complete closure of the primary wound and presence of the pectoralis major muscle, which serves as a protective layer.

In the present study, all infected patients were HKT recipients receiving immunosuppressive therapy with preoperative renal failure, reinforcing their vulnerability to infection as well highlighting the longer wait times in these patients. VGI can also happen to isolated heart transplant recipients, with an underlying common risk factor being prolonged duration of Impella use. Careful surveillance, early recognition of subtle signs, and a multidisciplinary approach are therefore essential in the successful management of these infections.

Based on our experience, our current approaches are as follows:

• Graft tunneling through the pectoralis major muscle with minimal graft length outside the skin, particularly in patients expected to require prolonged support or heart transplantation.

• Avoidance of reopening the axillary wound at the time of explant, especially in patients with prolonged Impella 5.5 support.

• Once VGI is suspected, aggressive surgical management includes complete resection of the Dacron graft and patch repair using a bovine pericardium; extra‐anatomical bypass should be reserved for extensive arterial involvement.

• Heightened vigilance and individualized surveillance in transplant candidates on immunosuppression, including a low threshold for imaging or surgical exploration.

As wait times for transplantation lengthen under current allocation systems, prolonged Impella 5.5 support will become increasingly common [19, 20, 21]. Larger, multicenter studies are needed to refine risk stratification, optimize surgical strategies, and evaluate the role of prophylactic tunneling and extended antimicrobial regimens.

9. Limitations

The main limitations of this study include the small sample size and its single‐center, retrospective design, which may limit the generalizability of findings. Additionally, the low incidence of vascular graft infection may limit the ability to identify independent risk factors. In addition, long‐term outcomes were not uniformly available due to variability in timing of implant, explant, and follow‐up.

10. Conclusions

Vascular graft infection following trans‐axillary Impella 5.5 implant is rare overall but occurs with notable frequency in patients requiring prolonged support, especially immunosuppressed transplant recipients. Vigilant surveillance and multidisciplinary care are critical to improving outcomes in this high‐risk population.

Author Contributions

Concept/design: J.S., S.O.; Data analysis/interpretation: I.L., S.B., R.M., A.G., A.D.; Drafting article: J.S., N.J.; Critical revision of article: S.O., A.L.; Approval of article: J.S., N.J., G.A., I.L., S.B., R.M., A.G., S.L., A.L., A.D., D.S., S.O.; Statistics: J.S., N.J.; Data collection: NJ, G.A.

Funding

The authors have nothing to report.

Disclosure

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Shimamura J., Jaramillo N., Amar G., et al., “Incidence and Management of Access Site Infection Following Trans‐Axillary Implant of a Microaxial Left Ventricular Assist Device,” Artificial Organs 50, no. 7 (2026): 1067–1076, 10.1111/aor.70146.

Data Availability Statement

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