Systematic Review of Outcomes after Non-Cardiac Surgery in Patients with Implanted Left Ventricular Assist Devices

Abstract

The number of patients supported with left ventricular assist devices (LVADs) is rising rapidly, and non-cardiac surgery (NCS) in these patients presents unique challenges. Given the controversy regarding the safety and timing of elective NCS, we performed a systematic review examining the perioperative morbidity and mortality of NCS in stable patients with LVADs. The published literature was searched using strategies created by a medical librarian. All reports involving 5 or more patients with implanted LVAD undergoing NCS were eligible for inclusion. 161 patients who underwent 252 surgeries were included from 7 studies. Cohort size ranged from 8 to 47 patients undergoing 12 to 67 NCS. Median age ranged from 50.1 to 68 years and 75 to 100% were male. Thirty-day postoperative mortality ranged from 6.4 to 16.7%, although 4 studies reported no deaths. Due to the small number of included studies with relative few patients and widely heterogeneous reporting of outcomes a formal quantitative meta-analysis was not performed. NCS in patients with LVADs appears to be safe and feasible in select patients. Future studies should use standard study design and reporting parameters to facilitate the systematic examination of safety and outcomes for elective NCS in LVAD patients.

INTRODUCTION

The use of implantable Left Ventricular Assist Devices (LVADs) is becoming increasingly prevalent since the landmark trial by Slaughter et al demonstrating their effectiveness in the management of select patients with end-stage heart failure.¹ The number of LVADs implanted yearly in the United States has risen from 206 in 2006 to over 2,261 in 2012,^2,3 and this rapid increase in the number of patients supported with mechanical assist devices has presented new challenges to both patients and clinicians. According to the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) the current 2-year survival rate after LVAD implant is 70%.³ Moreover, the approval of LVADs for destination therapy (DT) in non-transplant candidates has further increased both the number of HF patients receiving LVADs and their duration of support. Thus, as device therapies and supportive management continues to improve, patients on LVAD support will increasingly have needs for non-LVAD related medical care.

The need for non-cardiac surgical procedures in patients with LVADs presents unique challenges, and there is controversy regarding when and if elective operations should be performed. To date, there have been only small retrospective case series and cohort studies of LVAD patient outcomes after non-cardiac surgeries. Individually, these reports provide little evidence or rationale for surgeons or cardiologists to have informed conversations with patients regarding the risks and benefits of surgery. In accordance with the PRISMA statement,⁴ we sought to conduct a systematic review and meta-analysis of the literature to assess outcomes of non-cardiac surgery in patients on LVAD support. With this study we sought to assess the available data to help guide clinical practice for LVAD teams and to identify areas of need for future research. Specifically, the objective of this analysis was to examine the perioperative morbidity and mortality in stable patients with LVADs following non-cardiac surgery. Additionally, we planned to characterize the types of NCS being performed and perioperative management including use of anticoagulation and anti-platelet medications.

METHODS

Literature Search Strategy

The published literature was searched using strategies created by a medical librarian for the concepts of left ventricular assist devices, surgery, morbidity, and mortality. The concept of “non-cardiac surgery” was included by using a Boolean operator to exclude terms like “cardiac surgery.” These strategies were established using a combination of standardized terms and key words, and were implemented in PubMed 1946-, Embase 1947-, Scopus 1823-, the Cochrane Database of Systematic Reviews, Cochrane Database of Abstracts of Review Effects (DARE), Cochrane Central Register of Controlled Trials (CENTRAL), clinicaltrials.gov, Proquest Dissertations and Theses 1743-, and FirstSearch Proceedings. Searches were limited to humans using the Human filter for PubMed recommended in the Cochrane Handbook for Systematic Reviews of Interventions.⁵ This was also used as a model to create similar “human” filters for the other databases searched. Searches were completed in February 2014. All results were exported to EndNote and 975 duplicates were assumed to be accurately identified and removed for a total of 4,248 unique citations. Complete search strategies are shown in Supplementary Material 1.

Study Entry Criteria

All case reports, case series, and cohort studies involving 5 or more stable patients with implanted LVADs undergoing non-cardiac surgery were eligible for inclusion. Perioperative procedures and those arising from complications during the index implant hospitalization were excluded. Studies were also excluded if the non-cardiac surgeries were not described or if they were related to routine LVAD care. Such studies included those that described procedures for LVAD-related complications (e.g., driveline revision, pump-pocket exploration, device revision or exchange, sternal wound debridement, use of muscle flap to treat device-related infection), percutaneous cardiac procedures (cardiac catheterization, intra-cardiac structural procedures), electrophysiology studies or device placement or revision, or simple dental extractions. Abstracts or articles with fewer than 5 patients were excluded.

Endpoints

The primary endpoint was 30-day postoperative mortality. Secondary endpoints were postoperative morbidity and complications, specifically bleeding and device malfunction, as well as perioperative anticoagulant and anti-platelet strategies. A list of all parameters collected in the Data Extraction Sheet can be found in Supplementary Material 2.

Planned Analysis

Independent review of articles identified by title, then abstract, and finally full text was conducted by the authors. The screening results of two independent searches were compared and any discrepancies were resolved through a review of entry and exclusion criteria. We planned to perform a systematic review and meta-analysis using STATA software to compare events (mortality or morbidity/complications) to non-events in all studies; however the quality of the data was inadequate (see results). Variables of interest included number of patients, number and type of noncardiac procedures, surgical mortality, surgical complications, and perioperative anticoagulation and anti-platelet strategies. The Newcastle-Ottawa Scale, which is based on selection, comparability, and exposure, was used to assess study quality.⁶

RESULTS

Systematic review of Embase, PubMed, Scopus, the Cochrane Library, Clinicaltrials.gov, ProQuest Dissertations and Theses, and FirstSearch Proceedings yielded 5,223 records, 950 of which were duplicates. Initial manual title screening excluded 4,234 off-topic citations. Of the remaining 39 records assessed for eligibility, both screeners identified 39 as meeting inclusion criteria. Of these, 32 were excluded for the following reasons: 13 were case-reports of <5 non-cardiac procedures, 5 reported structural percutaneous intra-cardiac procedures, 5 reported on electrophysiology procedures, 3 were abstracts from conference presentations, 1 record was a duplicate, 1 was a series of muscle flap procedures done for device-related complications, 1 was a series of only dental procedures, 1 was a study that did not list the non-cardiac surgeries performed, 1 was a series in post-heart transplant patients, and 1 was a review article. Seven studies were included for systematic review and meta-analysis (Figure 1).^7–13

Figure 1.

Article screening process (adopted from Shamseer et al 2015⁶)

A total of 161 patients who underwent 252 surgeries were in the included studies. Cohort size ranged from 8 to 47 patients undergoing 12 to 67 NCS. Median age ranged from 50.1 to 68 years and 75 to 100% were male. When reported, 25.5 to 83.3% had ischemic cardiomyopathy, 18 to 100% were implanted as bridge to transplant and 0 to 82% were DT. Complete study baseline characteristics are shown in Table 1. The type of NCS included abdominal (range 25 to 50.7%), thoracic (range 0 to 10.5%), vascular (range 1.5 to 18.2%), orthopedic (range 1.5 to 16.7%), ENT (range 3 to 7.5%), urologic (range 3 to 16.7%), plastic (1 series, 5%), neurologic (range 3 to 9.1%), and oral (range 3 to 10%). Complete characteristics of the non-surgical procedures are given in Table 2. Information regarding the perioperative management of anticoagulation and anti-platelet therapies was available in 3 of 7 studies. Complications of NCS were inconsistently reported; 2 studies failed to report complications and 1 series reported no complications. Thirtyday postoperative mortality ranged from 6.4 to 16.7%, although 4 studies reported no deaths. However, in studies that included more than 20 patients at least 3 perioperative deaths occurred, with intracerebral hemorrhage and multisystem organ failure being the most common causes. Complete characteristics of NCS management and outcomes are given in Supplementary Table 2

Table 1:

Baseline Study Characteristics

Author	Year Published	No. of Patients Undergoing NCS	No. of NCS	Age	Male, No. (%)	Ischemic CM, No. (%)	BTT, No. (%)	DT, No. (%)
Barbara, et al.	2013	33	67	65.4 +/− 9.5 (median 68, range 31–79)	82 (52/67 procedures)	25 (37%)	12 (18%)	55 (82%)
Bhat, et al.	2012	110 total, 36 with operations, 64 without	63	61.4 ± 11.4	81%	NR	NR	NR
Ahmed, et al.	2012	6	6	median 60.8 (range 25–77)	100%	5 (83.3%)	NR	NR
Arnaoutakis, et al.	2013	173 total, 47 (27%) with operations, 126 without	67	50.2 ± 13.5	37 (78.7%) [pts]	12 (25.5%)	31 (66%)	16 (34%)
Morgan, et al.	2012	20	25	50.1+/− 12.7	13 (65%) [pts]	6 (30%)	12 (60%)	8 (40%)
Garatti, et al.	2009	11	12	mean 52 +/−10	9 (81.8%)	3 (27.3%)	8 (100%)	0
Goldstein, et al.	1995	8	12	52.7 +/−14.3	6 (75%)	4 (50%)	NR	NR

NR = Not reported. BTT = Bridge to Transplant. DT = Destination Therapy.

Table 2:

Characteristics of Non-Cardiac Surgical Management and Outcomes

Author	Anticoag. or anti-plt use within 5d of NCS	Reversal of anticoag. or anti-plt agent(s)	General Anesthesia	Local/Spinal/MAC	Complications	Deaths, No. (%)
Barbara, et al.	49/67 (73%); ASA + warfarin 23/49 (47%), heparin 13/49 (27%), ASA 5/49 (10%), ASA + heparin 5/49 (10%), ASA + warfarin + heparin 1/49 (2%), warfarin 1/49 (2%), clopidogrel 1/49 (2%)	32/49 (65%); stop heparin 14/49 (44%), hold warfarim 8/49 (25%), FFP + vit K 3/49 (9.5%), FFP 2/49 (6.5%), FFP + plts 1/49 (3%), vit K 1/49 (3%), LMWH bridge 1/49 (3%), LMWH, vit K, cryo, plts 1/49 (3%), stop ASA 1/49 (3%)	100%	0	11 patients (33%) after 18 NCS (28%). 0 AKI. 10/67 bleeding, 3 bleeds requiring return to OR.	3 (9.1%) 1 intracerebral hemorrhage, 2 multi-organ failure (<30 days)
Bhat, et al.	NR	NR	56	7	NR	6 (16.7%), All emergent operations. 3 intracerebral hemorrhage, 2 ischemic bowel, 1 multisystem organ failure (<30 days)
Ahmed, et al.	3 of 6	no bridging	100%	0	0	0
Arnaoutakis, et al.	NR	NR	NR	NR	NR	3 (6.4%), 1 intraop bleeding, 2 sepsis (< 30 days)
Morgan, et al.	15/25 kept ASA & warfarin; 6/25 kept ASA only; 4/25 stopped both	8 bridged with IV heparin	100		9 of 25 bled (36%) (all on warfarin)	0
Garatti, et al.	NR	NR	8 (73%)	3 (27%)	10 (90.9%) bled	0

NR = Not reported.

Due to the small number of included studies with relative few patients and widely heterogeneous reporting of outcomes we chose not to perform a formal meta-analysis or data synthesis. This heterogeneity between individual study effects could not be pooled across the studies. Study quality, as evaluated with the Newcastle-Ottawa Scale, was also an issue with inconsistent selection of the exposed cohort and varying reporting of management and outcomes. As multiple studies did not report any perioperative deaths or complications, we deemed it inappropriate to perform a formal meta-analysis with regards to mortality. The significant inconsistency of the variables reported, including the type of LVAD, the type of NCS and the management of perioperative anticoagulation, also reduced the validity of a formal analysis.

DISCUSSION

With more durable LVADs implanted into people who are living longer, it is inevitable that medical and surgical issues unrelated to the LVAD will arise. Therefore, it is essential that the outcomes for LVAD patients undergoing NCS be investigated in a standardized and transparent manner. As an example, understanding how to approach perioperative issues related to anticoagulation/ anti-platelet management and LVAD settings will be important. Moreover, developing risk predictions models to help assess post-operative morbidity and mortality will be necessary to inform physician and patient decisions. To achieve these objectives, future studies should include a detailed discussion of what medications were held before NCS and what strategies were used for peri-procedural anticoagulation (i.e. heparin bridging, goal PTT). In addition, the validity and usefulness of future studies would be enhanced by the use of standard methods and definitions for reporting complications, such as those published by the American College of Surgeons National Surgical Quality Improvement Program (NSQIP).¹⁴

The most important finding of this systematic review is that it sheds light on the paucity of high quality data regarding the appropriate management and anticipated risk of NCS procedures in patients with durable LVAD support. To our knowledge this is the first systematic review on this topic. Limitations of our study stem from the incomplete and inconsistently reported variables and outcomes in combination with the small size and number of studies. Although we recognize that the number of patients and sources are limited, this is a reflection of the available data. The studies were also heterogeneous with regards to the variability in outcome reporting, as few studies reported perioperative mortality. However, in studies which included more than 20 patients perioperative deaths occurred in a range of 6.4–16.7%. Given the paucity and heterogeneity of published data, we felt a formal meta-analysis was not appropriate, and thus completed the study as a systematic review.

With this analysis we sought to systematically assess the available data, not to provide the basis for management guidelines. We focused on common ambulatory surgeries that face LVAD teams and non-cardiac surgeons. For this reason we excluded complications of NCS occurring during the index implantation hospitalization, as well as non-cardiac procedures now considered either routine (GI procedures and dental extractions) or non-elective (neurosurgical procedures for intracrainial hemorrohage). In our opinion, it is critical that decisions regarding the need for NCS and the approach to peri-operative management should be made in the context of multidisciplinary LVAD team.

While no definitive conclusions can be made based on the available data, NCS in patients with LVAD devices appears to be safe and feasible in select patients. In the future, studies in this area should use standard study design and reporting parameters to facilitate the systematic examination of safety and outcomes for elective NCS in LVAD patients. We would recommend detailed, consistent reporting of LVAD type, all procedures performed, and standardization of quality measures including the incidence of significant bleeding and blood transfusions, frequency of LVAD device malfunction (particularly device infection and thrombosis), and the management of perioperative anti-coagulation and anti-platelet medications. A centralized formal prospective registry of NCS could provide the details necessary to guide clinical practice as the number of patients on LVAD support continues to grow. In summary, this systematic review highlights the growing need for additional study of NCS in patients with durable continuous flow LVAD support.