Abstract
Background:
Infections after orthotopic heart transplantation (OHT) cause significant morbidity and mortality. Concurrent with increased pre-OHT temporary mechanical support (MCS), there have been recent concerns of a perceived increase in infections post-OHT. We examined the association between pre-OHT temporary vs. durable MCS and post-OHT infection.
Methods:
We performed a single-center retrospective review of patients who received OHT at Tufts Medical Center between January 2014 and April 2022. Information was collected on MCS device(s) in place during the admission in which the OHT occurred. Our composite outcome was the occurrence of bacteremia, invasive fungal infections, opportunistic infections, or skin/soft tissue infections of device sites within one-year post-OHT. We used Cox proportional hazards models to assess the relationship between type of pre-OHT MCS and time to first infection, treating death from other causes as a competing risk. We addressed confounding with two statistical methods: propensity score (PS) with inverse probability weighting (IPW) and an instrumental variable (IV) analysis.
Results:
Of the 320 OHT recipients during the study period, 268 required MCS prior to OHT; 192 were managed with durable MCS and 76 with temporary MCS before transplant. Patients receiving pre-OHT temporary MCS had no difference in time to first infection (unadjusted HR 0.77, 95% CI 0.41–1.44), compared to durable MCS. Results were similar in the model employing PS with IPW (HR 0.61, 95% CI 0.29–1.27) and the IV analysis (HR 0.28, 95% CI 0.26–2.36).
Conclusion:
Pre-OHT temporary MCS was not associated with the composite outcome of bacteremia, invasive fungal infections, opportunistic infections, or skin/device site infections post-transplant compared to durable MCS in this single-center cohort.
Introduction
Infections that occur after heart transplantation have serious implications for the transplant recipient‘s morbidity and mortality, with 42–60% of deaths in the first six months after transplant being attributed to infection.1,2 Prior to orthotopic heart transplant (OHT) many patients require mechanical circulatory support (MCS) to maintain their hemodynamics while they are waiting for an organ to become available. These can include durable devices such as left- or right ventricular assist devices (LVAD, RVAD), or temporary devices such as intra-aortic balloon pumps (IABPs), percutaneous assist devices (Impella), biventricular assist devices (BiVAD), and extracorporeal membrane oxygenation (ECMO). These devices are associated with their own infectious complications with bacteremia occurring in one third of patients within one year of LVAD placement.3 Limited data are available describing the frequency of bacteremia in patients on temporary MCS though one study reported a rate of 20%.4 ECMO specifically has been associated with rates of cannula related infection of 17.1% with nearly 60% of these being associated with a concomitant bacteremia.5
In 2018, the United Network for Organ Sharing (UNOS) made changes to the allocation system for heart transplants to prioritize sicker patients for more urgent listing status with the intention of decreasing waitlist mortality in the United States. These changes included prioritizing patients receiving temporary MCS for more urgent listing status over stable patients with durable MCS.6,7 As a consequence, heart transplant centers in the U.S. have been maintaining more patients on temporary MCS in the intensive care unit (ICU).8 Since these changes were made, there have been anecdotal reports of an increased incidence of infections following OHT in patients who received temporary MCS prior to transplant compared to those who received durable MCS. To the best of our knowledge, no studies have examined whether there is a differential risk of infection post-OHT between those receiving durable or temporary MCS. Therefore, the aim of our study was to examine the association of MCS type with the incidence of infection one year after OHT, using multiple methods to adjust for confounding.
Methods
This retrospective cohort study was performed at Tufts Medical Center (TMC). The preexisting Transplant Registry was used to identify 320 patients who underwent OHT at TMC between January 2014 and April 2022. Patients were included if they were aged eighteen years or over, required MCS prior to transplant, underwent OHT between January 1, 2014, and April 30, 2022, and survived at least 72 hours post-transplant. Patients were followed for one-year post-transplant or until time of death. Patients who were lost to follow-up or who transferred to another institution were censored at the time of their last visit at TMC. The Tufts Health Sciences Institutional Review Board approved this study. Informed consent was not required due to the retrospective study design and minimal risk posed to study participants. This research was conducted in compliance with the International Society for Heart and Lung Transplantation (ISHLT) Ethics Statement.
Data Collection and Covariates
Chart review was performed by two physicians (CAT and AM). Information was abstracted on the incidence of bloodstream infections, opportunistic infections, device site infections (including mediastinitis), and invasive fungal infections. Additional information was also obtained on factors that may influence a patient‘s risk for infection post-transplant including but not limited to recipient’s age, sex, medical comorbidities, pre-transplant MCS, need for dialysis pre- and post-transplant, and surgical complications. For patients lost to follow-up prior to one-year post-transplant, mortality information was obtained from the Transplant Registry.
Definitions
Bloodstream infection (BSI) was defined as a positive blood culture in accordance with previous definitions.9 Bacteremia caused by common skin contaminants was considered significant if the same organism was isolated from two separate blood cultures in the presence of clinical signs of infection.9 Opportunistic infections included Nocardia, Pneumocystis jirovecii, and non-tuberculous mycobacterial infections.10, 11 Invasive fungal infection (IFI) was defined as identification of fungal or yeast species by culture or histological examination from a normally sterile site.12 Solitary sputum, urine or non-sterile site cultures were not considered an IFI. Skin and soft tissue infection of the cannula site(s) were defined as erythema, swelling, warmth, and pain of the insertion site13 with positive culture of either tissue or drainage. Any positive tissue cultures obtained during OHT were considered infections that occurred prior to transplant. Mediastinitis was defined as the presence of clinical (i.e., incisional erythema with or without purulence, unstable sternum) or radiographic findings suggestive of infection of the mediastinum with positive tissue culture from the surgical site.14 Patients were considered to have had a chronic VAD infection if they were on suppressive antimicrobial therapy at time of transplant.
Exposure and Composite Outcome
We analyzed the binary exposure of temporary vs. durable MCS. We excluded patients who did not require any mechanical circulatory support prior to transplant as they were felt to represent a fundamentally different population with different risk profiles. Patients were included in the temporary MCS group if they required IABP, Impella, BiVAD, or ECMO prior to OHT during the hospitalization in which the transplant occurred. Patients in the durable MCS group were those maintained with durable left- and/or right- ventricular assist devices before transplant (LVAD, RVAD). Only devices that were in place during the hospitalization for heart transplant were considered. Patients who had both device types in place during the admission in which heart transplantation occurred were included in the temporary MCS group for the main analysis.
Our primary outcome was a composite of the first instance of bloodstream infection, opportunistic infections, invasive fungal infection, or skin or soft tissue infection of a device site (including mediastinitis) within one year of OHT. Patients who died due to infection were included in the composite outcome. Death due to noninfectious etiologies was a secondary outcome.
Statistical Analysis
Baseline characteristics were compared between those who received temporary MCS and those who received durable MCS. Continuous variables are summarized as mean +/− standard deviation (SD) while categorical variables are summarized by the frequency and the proportion of the total in that group. Differences between exposure groups were assessed using chi-squared and Student’s T-tests. The log-rank test was used to evaluate the differences in time to infection between the temporary and durable MCS groups. The time-to-event analysis was performed with death from noninfectious etiology considered as a competing risk creating cumulative incidence curves for both outcomes and device types. A series of cause-specific Cox proportional hazards (PH) regression models assessed the relationship between temporary and durable MCS and development of infection by one-year post-transplant with mortality from noninfectious causes treated as a competing risk and using different methods to address confounding. The assumptions of the Cox PH model were assessed graphically via a log[−log(S(t))] plot and Schoenfeld residual plots.
Propensity Score Analysis
Propensity scores allow the analysis of an observational study to more closely mimic the results of a randomized controlled trial by creating a conditional probability of being in the exposure group based on prespecified baseline characteristics.15 A logistic regression was fit predicting the exposure (temporary vs. durable device preceding transplant) using the following covariates: age at device placement, sex, body mass index (BMI), ischemic cardiomyopathy, chronic kidney disease (CKD), and diabetes mellitus. These covariates were chosen as they were felt to be potential confounders of the relationship between pre-transplant MCS device type and post-transplant infection.
A Cox PH model was fit applying the PS using inverse probability weighting (IPW) and adjusting for the variables included in the PS. This model was then repeated examining the relationship between temporary MCS and three of the components of the composite outcome individually (opportunistic infections were not analyzed separately due to small numbers).
Instrumental Variable Analysis
The change in UNOS policy was used as an instrumental variable (IV), which is a way to identify causal associations in the absence of a randomized controlled trial.16 The advantage of an IV analysis over a PS analysis is that it addresses both measured and unmeasured confounding. To be considered valid, an IV must meet three conditions: it must be predictive of the exposure of interest, it must be independent of the outcome, and it must affect the outcome only through the exposure.16 In this case, the policy change served as an exogenous source of variation, allowing for a quasi-random assignment of patients to different types of MCS.
We performed a two-stage least-squares (2SLS) design. First, a logistic regression predicting the exposure (temporary vs. durable device preceding transplant) based on the IV was conducted, which was also used to assess the strength of the IV. Then, these predicted values were used as the exposure variable in a Cox PH model with the composite outcome adjusting for age at device placement, sex, BMI, ischemic cardiomyopathy, CKD, and diabetes mellitus. This model was then repeated examining the relationship between temporary MCS and three of the components of the composite outcome individually (as above, opportunistic infections were not analyzed separately due to small numbers [N=3]). The C-statistic for the association between the UNOS policy change and the type of MCS pre-transplant was 0.81 which is indicative of a strong predictive relationship.
In total four models were conducted: (1) unadjusted Cox PH model, (2) adjusted Cox PH model, (3) Cox PH model applying PS with IPW, and (4) Cox PH model with IV analysis (Table 1).
Table 1:
Summary of the models used to assess the relationship between temporary MCS and post-transplant infectious complications using Propensity Score with IPW or Instrumental Variable Analysis to Control for Confounding.
| Model | Type of Control for Confounding | Model type | Covariates adjusted for in the model |
|---|---|---|---|
| 1 | Unadjusted Cox Model | Cox PH | -- |
| 2 | Adjusted Cox Model | Cox PH | Age at device placement (years, continuous) Body Mass Index (kg/m2, continuous) Sex (binary) DM (binary) CKD (binary) Type of Cardiomyopathy (binary) |
| 3 | Propensity Score with Inverse Probability Weighting | Cox PH | Age at device placement (years, continuous) Body Mass Index (kg/m2, continuous) Sex (binary) DM (binary) CKD (binary) Type of Cardiomyopathy (binary) |
| 4 | Instrumental Variable Analysis | Cox PH | Age at device placement (years, continuous) Body Mass Index (kg/m2, continuous) Sex (binary) DM (binary) CKD (binary) Type of Cardiomyopathy (binary) |
CKD: Chronic Kidney Disease; DM: Diabetes Mellitus; MCS: Mechanical Circulatory Support; PH: Proportional Hazards; RRT: Renal Replacement Therapy
Supplementary Analyses
Supplementary analyses were performed including adding duration of mechanical circulatory support as a covariate in the models. Additional analyses were performed examining factors associated with multiple episodes of infections in the first-year post-transplant using chi-squared and t-tests. The results of these are included in the Supplementary appendix. Multiple sensitivity analyses were performed including models: (1) excluding the six patients who had both device types, (2) excluding the sixteen patients that underwent combined heart/kidney transplantation, (3) including the year of transplant as a variable, (4) including whether transplant was conducted before or after the UNOS change, and (5) using weights stabilized by the marginal probability of receiving the given device type for the PS with IPW model to reduce the impact of extreme weights due to outliers (data not shown).
Results
Of the 320 OHT performed during study period, 268 total patients required MCS before OHT, of whom 192 were managed with only durable MCS, 70 with only temporary MCS before transplant, and six with both temporary and durable (Figure 1).
Figure 1. Flow Chart of Cohort.
MCS: Mechanical Circulatory Support, OHT: Orthotopic Heart Transplant. Patients were included in the temporary MCS group if they required IABP, Impella, BiVAD, or ECMO prior to OHT during the hospitalization in which the transplant occurred. Patients in the durable MCS group were those maintained with durable left- and/or right- ventricular assist devices at home before transplant (LVAD, RVAD).
Most of the cohort was male (78%), white (80%), with non-ischemic cardiomyopathy as a cause of heart failure (66.7%). On average, the patients in the temporary support group were younger at time of transplant (52.3 vs 56.5 years, p-value 0.006), had a lower BMI (25.6 vs 28.4 kg/m2, p-value <0.001), and were more likely to require pre-transplant dialysis (17% vs 1%, p-value <0.001) and undergo combined heart-kidney transplant (11.8% vs 3.7%, p-value 0.02) (Table 2).
Table 2.
Pre- and Post-OHT Characteristics stratified by Temporary vs Durable MCS
| Durable Device (n=192) | Temporary Device (n=76) | p-value | |
|---|---|---|---|
| Pre-Device Placement | |||
| Age (years) at Device Placement (median, IQR) | 57.8 [48.2, 63.6] | 54.2 [49.0, 59.9] | 0.08 |
| Male (n, %) | 150 (78.1) | 59 (77.6) | >0.99 |
| White (n, %) | 151 (79.9) | 61 (80.3) | >0.99 |
| Ischemic Cardiomyopathy (n, %) | 75 (39.1) | 21 (27.6) | 0.11 |
| CKD (n, %) | 80 (41.7) | 35 (46.1) | 0.61 |
| DM (n, %) | 74 (38.5) | 31 (40.8) | 0.84 |
| Prior transplant (SOT or HSCT) (n, %) | 0 (0.0) | 2 (2.6) | 0.14 |
| Pre-transplant | |||
| Pre-OHT Dialysis (n, %) | 2 (1.0) | 13 (17.1) | <0.001 |
| Infection in the 30 days prior to transplant (n, %) | 34 (17.8) | 11 (14.5) | 0.64 |
| Chronic Driveline infection in patients with VAD (n, %) | 106 (55.2) | 2 (33.3) | 0.52 |
| Duration of support | |||
| Durable Device(days) (median, IQR) | 427.5 [262.8, 835.5] | 53.5 [2.5, 223.0] | 0.002 |
| Any Temporary Device (days) (median, IQR) | 34.0 [15.0, 56.0] | ||
| Blood Type (n, %) | 0.02 | ||
| A | 83 (43.2) | 23 (30.3) | |
| B | 24 (12.5) | 16 (21.1) | |
| AB | 7 (3.6) | 8 (10.5) | |
| O | 78 (40.6) | 29 (38.2) | |
| Listing Status | |||
| Prior to 2018 (n, %) | 0.49 | ||
| HR 1A | 135 (89.4) | 12 (100) | |
| HR 1B | 16 (10.6) | 0 (0.0) | |
| After 2018 (n, %) | <0.001 | ||
| Adult 1 | 1 (2.4) | 24 (37.5) | |
| Adult 2 | 11 (26.8) | 37 (57.8) | |
| Adult 3 | 23 (56.1) | 3 (4.7) | |
| Adult 4 | 6 (14.6) | 0 (0.0) | |
| Intraoperative | |||
| Age (years) at | 59.1 [50.1, 65.1] | 54.4 [49.0, 60.0] | <0.01 |
| Transplant (median, IQR) | |||
| BMI (kg/m2) (mean, SD) | 28.4 (5.1) | 25.65 (5.3) | <0.001 |
| Ischemic Time (minutes) (mean, SD) | 196.25 (45.09) | 200.21 (33.77) | 0.49 |
| Additional Induction (n, %) | |||
| Anti-IL2 | 37 (19.3) | 21 (27.6) | 0.18 |
| ATG | 3 (1.6) | 1 (1.3) | >0.99 |
| Combined Heart-Kidney transplant (n, %) | 7 (3.7) | 9 (11.8) | 0.02 |
| CMV Serostatus (n, %) | 0.55 | ||
| D+/R+ | 48 (25.0) | 16 (21.1) | |
| D+/R− | 64 (33.3) | 23 (30.3) | |
| D−/R+ | 39 (20.3) | 16 (21.1) | |
| D−/R− | 41 (21.4) | 21 (27.6) | |
| Open Chest (n, %) | 40 (20.8) | 15 (19.7) | 0.97 |
| Blood Transfused during OHT (units) (mean, SD) | 2.36 (2.53) | 2.67 (2.10) | 0.35 |
| Transplant Prior to UNOS change (n, %) | 151 (78.6) | 12 (15.8) | <0.001 |
| Post-transplant | |||
| Post-OHT Dialysis (n, %) | 33 (17.6) | 16 (21.9) | 0.53 |
| Post-OHT Temporary Device (n, %) | 19 (9.9) | 6 (7.9) | 0.78 |
| Rejection event treated with intravenous therapy within first year post-OHT (n, %) | 42 (21.9) | 15 (19.7) | 0.83 |
Temporary Device cohort includes six patients with both device types in the admission during which heart transplantation occurred.
Of the 76 patients with pre-OHT Temporary MCS: 63 had intra-aortic balloon pumps, 32 had impellas, 24 had extracorporeal membrane oxygenation, and 22 had biventricular assist devices.
Chronic Driveline Infections and Durable Device Duration in the Temporary MCS group represents that number of chronic driveline infections among the six patients with both device types.
ATG: Antithymocyte globulin; BMI: Body Mass Index; CKD: Chronic Kidney Disease; CMV: Cytomegalovirus; D:Donor; DM: Diabetes Mellitus; HSCT: Hematopoietic Stem Cell Transplant; IQR: Interquartile Range; MCS: Mechanical Circulatory Support; OHT: Orthotopic Heart Transplant; R: Recipient; RRT: Renal Replacement Therapy; SD: Standard Deviation; SOT: Solid Organ Transplant; UNOS: United Network for Organ Sharing; VAD Ventricular Assist Device
Dialysis includes hemodialysis, peritoneal dialysis, and continuous renal replacement therapy.
Duration of support row excludes the 6 patients with both device types.
Infection in the 30 days prior to transplant includes culture proven bacteremia, invasive fungal infection, opportunistic infection, and skin/soft tissue infections of the device site including mediastinitis.
Additional induction is not mutually exclusive and not all patients received additional induction outside of our institution's protocol for intravenous steroids.
Intravenous therapy for rejection included: intravenous steroids, antithymocyte globulin, Muromonab-CD3, and Basiliximab.
A total of 53 patients in the cohort experienced the composite outcome (median time to event 34 days) with a rate of 20.8% (40/192) in the durable MCS group and 17.1% (13/76) in the temporary MCS group (p-value 0.60). There were 27 patients who had at least one episode of bacteremia, 19 who had at least one episode of IFI, 24 who had at least one episode of skin/soft tissue infection of a device site including mediastinitis, and 3 who had opportunistic infections (Table 3). Of the 53 patients who had experienced the composite outcome, 23 had multiple episodes of infection, 91% of whom were supported with durable MCS prior to transplant (21/23). There was a total of 16 deaths with 12 in the durable MCS group (6.3%) and 4 in the temporary MCS group (5.3%); of these, 58.3% (7/12) and 75% (3/4) were attributed to infection, respectively. One death in the durable support group had an unknown cause of death and thus attribution could not be determined.
Table 3.
Outcomes at One Year by stratified by Temporary vs Durable MCS
| Durable Device (n=192) | Temporary Device (n=76) | p-value | |
|---|---|---|---|
| Death from any cause (n, %) | 12 (6.2) | 4 (5.3) | 0.98 |
| Composite (n, %) | 40 (20.8) | 13 (17.1) | 0.60 |
| Bacteremia (n, %) | 22 (11.5) | 5 (6.6) | 0.33 |
| Invasive Fungal Infection (n, %) | 13 (6.8) | 6 (7.9) | 0.95 |
| Skin/Soft Tissue Infection of Device Site (including Mediastinitis) (n, %) | 20 (10.4) | 4 (5.3) | 0.27 |
Temporary Device cohort includes six patients with both device types in the admission during which heart transplantation occurred.
Of the 76 patients with pre-OHT Temporary MCS: 63 had intra-aortic balloon pumps, 32 had impellas, 24 had extracorporeal membrane oxygenation, and 22 had biventricular assist devices.
MCS: Mechanical Circulatory Support
The proportion of patients who experienced the composite outcome was plotted by the year in which they underwent OHT (Figure 2). Cumulative incidence curves for each exposure group with death from noninfectious etiologies as a competing risk is shown in Figure 3. These curves show a higher cumulative incidence of the composite outcome among the durable MCS patients, but this was not statistically significant (p-value 0.42). The competing risk of death from non-infectious causes was similar between the two MCS groups (p-value 0.87).
Figure 2. Proportion of patients who required pre-OHT MCS that developed the composite outcome per transplant year compared to the proportion that receiving temporary MCS.
This graph only includes transplant years 2014–2021 for which we had complete one-year follow-up (our study only included the first 6 months of 2022) and does not account for censoring. The temporary MCS line represents the number of recipients for each year that required temporary MCS divided by the total number of heart transplants performed that year. MCS: Mechanical Circulatory Support.
Figure 3. Cumulative Incidence of the Composite Outcome for Temporary vs. Durable Mechanical Circulatory Support.
The graph above shows the cumulative incidence for the composite outcome and the competing risk of death stratified by the type of device the patient received prior to transplant. “Events” in the tables refers to both combined and can represent either the composite outcome or death without infection. MCS: Mechanical Circulatory Support. The shaded areas in the figure represent the respective confidence intervals.
Unadjusted and Adjusted Cox PH models
Results of all four models are summarized in Table 4. The unadjusted Cox proportional hazards model showed no difference (p-value 0.41) in the hazard of developing the composite outcome between patients receiving temporary support compared to those who received durable support, (HR: 0.77; 95% CI 0.41–1.44). The HR for death from other causes within one year of OHT among individuals with temporary MCS was 0.81 (95% CI 0.08 – 7.77) compared to individuals with durable MCS. When this model was adjusted for age at device placement, sex, BMI, history of ischemic cardiomyopathy, history of CKD, and history of diabetes mellitus, no differences were found for the composite outcome (HR 0.68, 95% CI 0.34 – 1.34, p-value 0.26) nor death from other causes (HR 0.88, 95% CI 0.07 – 10.63, p-value 0.92). Similarly, no relationships were found for any of the components of the composite outcome with either the unadjusted or the adjusted Cox PH models.
Table 4.
Unadjusted and adjusted Cox Proportional Hazards models examining the relationship between temporary mechanical circulatory support and posttransplant infectious complications
| Cause-specific HR for outcome (95% CI) | p-value | Cause-specific HR for death (95% CI) | p-value | |
|---|---|---|---|---|
| Composite Outcome | ||||
| Unadjusted Cox PH | 0.77 (0.41 – 1.45) | 0.42 | 0.81 (0.08 – 7.77) | 0.85 |
| Adjusted Cox PH | 0.68 (0.34 – 1.34) | 0.26 | 0.88 (0.07 – 10.63) | 0.92 |
| Cox PH applying PS with IPW | 0.66 (0.32 – 1.34) | 0.25 | 1.02 (0.05 – 21.63) | 0.99 |
| Cox PH with IV analysis | 0.78 (0.26 – 2.37) | 0.67 | 0.73 (0.01 – 69.36) | 0.89 |
| Bacteremia | ||||
| Unadjusted Cox PH | 0.55 (0.21– 1.46) | 0.23 | 1.04 (0.27 – 4.01) | 0.96 |
| Adjusted Cox PH | 0.51 (0.18 – 1.43) | 0.20 | 1.09 (0.24 – 4.88) | 0.92 |
| Cox PH applying PS with IPW | 0.41 (0.15 – 1.12) | 0.08 | 1.07 (0.24 – 4.85) | 0.93 |
| Cox PH with IV analysis | 0.93 (0.21 – 4.18) | 0.92 | 6.98 (0.51 – 95.47) | 0.15 |
| Invasive Fungal Infections | ||||
| Unadjusted Cox PH | 1.12 (0.42 – 2.93) | 0.83 | 0.69 (0.14 – 3.34) | 0.65 |
| Adjusted Cox PH | 0.92 (0.31 – 2.72) | 0.88 | 0.55 (0.10 – 3.18) | 0.50 |
| Cox PH applying PS with IPW | 0.86 (0.30 – 2.49) | 0.78 | 0.59 (0.08 – 4.35) | 0.60 |
| Cox PH with IV analysis | 3.50 (0.57 – 21.60) | 0.18 | 0.44 (0.02 – 7.81) | 0.58 |
| Skin and Soft Tissue Device Site Infections (including Mediastinitis) | ||||
| Unadjusted Cox PH | 0.47 (0.16 – 1.38) | 0.17 | 1.01 (0.26 – 3.91) | 0.99 |
| Adjusted Cox PH | 0.46 (0.15 – 1.46) | 0.19 | 1.36 (0.29 – 6.39) | 0.69 |
| Cox PH applying PS with IPW | 0.53 (0.15 – 1.87) | 0.32 | 1.23 (0.21 – 7.13) | 0.82 |
| Cox PH with IV analysis | 0.38 (0.07 – 2.23) | 0.29 | 2.30 (0.16 – 33.29) | 0.54 |
All models EXCEPT the Unadjusted Cox PH adjusted for: Age at device placement (years), sex, body mass index (kg/m2), chronic kidney disease, diabetes mellitus, and ischemic cardiomyopathy.
HR: Hazard Ratio; IPW: Inverse Probability Weighting; IV: Instrumental Variable; PH:
Proportional Hazards; PS: Propensity Score.
Cox PH model applying the propensity score using IPW
Several statistical assessments were performed to assess the strength of the PS. The C-statistic of the score was 0.71, which indicates acceptable discriminatory power. Standardized mean differences between device types were compared for these covariates before and after applying the PS and all were <0.1 after IPW, which suggests a well-balanced model. The PS model did not show a relationship between type of pre-transplant MCS and the composite outcome (HR 0.66, 95% CI 0.32 – 1.34, p-value 0.25) or death from other causes (HR 1.02, 95% CI 0.05 – 21.63, p-value 0.99), nor with any of the three separate components of the outcome that were assessed (Table 4). There was a trend towards a decreased incidence of bacteremia among patients with temporary MCS as compared to durable, though not statistically significant (HR 0.41, 95% CI 0.15 – 1.12, p-value 0.08).
Cox PH model applying the Instrumental Variable
The results of this model did not show a relationship between type of pre-OHT MCS and the composite outcome (HR 0.78, 95% CI 0.26 – 2.37, p-value 0.67) or death from other causes (HR 0.73, 95% CI 0.01 – 69.36, p-value 0.89), nor with any of the three separate components of the outcome that were assessed (Table 4).
Supplementary Analyses Results
Multiple sensitivity analyses were performed including models: (1) excluding the six patients who had both device types, (2) excluding the sixteen patient that underwent combined heart/kidney transplantation, (3) including the year of transplant as a variable, (4) including whether transplant was conducted before or after the UNOS change, and (5) using stabilized weights for the PS with IPW model to address outliers and the results did not differ from the main analyses (data not shown).
Discussion
Based on our analysis, it does not appear that pre-OHT temporary MCS is associated with infectious complications after heart transplantation compared to pre-OHT durable MCS. Rates of infection in our cohort were lower than previously reported,1 though this is likely due to our study not including urinary tract infections and pneumonia with non-opportunistic pathogens in our composite outcome. We chose to exclude these infection types because they are not as clearly defined as the outcomes in our study and often require clinical evaluation which is not possible in a retrospective chart review. The proportion of deaths attributable to infections was consistent with previously published data.2
There have been anecdotal reports among the transplant cardiology community of an increased rate of infection post-OHT occurring after the UNOS policy change, possibly driven by the increased use of temporary MCS. Based on multiple statistical analyses, it does not appear that pre-OHT temporary MCS is associated with increased rates of infection within one-year post-OHT when compared to pre-OHT durable MCS. It is possible that, because the UNOS policy was designed to prioritize sicker patients, the current population receiving heart transplants is more critically ill, making them more prone to infection for other reasons independent of device type. Additional studies are needed to examine rates of infection over more recent years to draw more concrete conclusions about this trend.
The main strength of this study lies in its novelty as, to our knowledge, no other study has examined the relationship between the type of MCS prior to transplant and rates of infection post-transplant. Additionally, given our institutions long history of performing heart transplantations and the history of clinical and research collaboration between the cardiology and infectious diseases departments, we were able to build a large and comprehensive dataset. Another strength of our study is the use of multiple methods to control for confounding, as well as a variety of sensitivity analyses, all of which yielded similar results. Finally, to minimize overcounting infection rates in this population, we chose to only include patients with culture-proven infection. While this should decrease misclassification, it is possible that some of the cultures, particularly those of the skin/soft tissue/device sites, represent colonization rather than true infection. This was accounted for by including clinical signs/symptoms of infection in our definitions.
There are a few study limitations worth considering; this is a single center study examining patients who have undergone cardiac transplantation and follow-up at TMC which may limit the study generalizability. However, TMC is a large academic medical center that has performed more than 700 heart transplants since its inception in 1988, which allowed us to have a relatively large sample size. Similarly, because we only have information from our hospital system, if a patient had an infection that was treated outside of the TMC system, it would not be captured in this dataset. This was relatively rare as most organ recipients are transferred to their transplant center if they are hospitalized elsewhere within the first year of transplant. One other limitation to generalizability is the homogeneity of the population which consists mostly of older, white males. We tried to overcome this limitation by using the PS with IPW analysis, but there will be some remaining bias as the propensity score was modeled from the same population.
This study was conducted prior to the release of the consensus definitions of device infections by the ISHLT earlier this year.17 Though pre-transplant MCS infections were not included in the model, future studies could benefit from using these definitions when accounting for risk factors for post-transplant infection.
Additionally, because of our relatively small sample size, we chose to group all types of temporary support together rather than to compare them separately. Different temporary MCS devices may be associated with different rates of infection. Unfortunately, due to our small sample size, we could not examine this level of granularity. This represents an area for additional investigation in the future.
For the purposes of this study, we chose to focus on infections that occurred post-OHT. Patients on temporary MCS prior to OHT are often sicker and are, by the nature of the devices, required to stay in the ICU. As such, patients on temporary MCS may be at higher risk for infection in the pre-transplant period, which may impact their listing status. This would not be captured by this study but warrants separate investigation.
The retrospective nature of the study also limits our sample size and creates the potential for some missing data. Fortunately, we were able to obtain most of the information we were seeking with only minimal missingness, <3% for each covariate included in the models. There may be some confounding by treatment indication as it is possible that the type of MCS a patient receives is related to the overall severity of their illness, which may also directly relate to their risk for developing infection. Additionally, it is possible that there have been other changes over time to increase the risk for post-OHT infection, which would mean that the IV analysis is not truly unconfounded. Finally, there may have been an increase in infections in 2022 and beyond not captured by this dataset.
Conclusion
Based on our single center study, there does not appear to be a relationship between pre-transplant MCS type and incidence of bacteremia, opportunistic infections, IFI, and skin/device site infections post-transplant. Additional studies need to be undertaken to confirm this finding. While a randomized controlled trial would not be feasible given the multiple factors considered when choosing a patient‘s device type, a multicenter retrospective study could provide a larger dataset with more generalizability.
Supplementary Material
Table 5.
Incidence of the composite outcome, bacteremia, invasive fungal infection, and skin and soft tissue infections of device site (including mediastinitis) by device type.
| Composite Outcome | |||
|---|---|---|---|
| No Composite (n=215) | Composite (n=53) | p-value | |
| Durable MCS | 157 (73.0) | 41 (77.4) | 0.64 |
| Any Temporary MCS | 63 (29.3) | 13 (24.5) | 0.603 |
| IABP | 54 (25.1) | 9 (17.0) | 0.29 |
| Impella | 26 (12.1) | 6 (11.3) | >0.99 |
| ECMO | 20 (9.3) | 4 (7.5) | 0.90 |
| BiVAD | 18 (8.4) | 4 (7.5) | >0.99 |
| Bacteremia | |||
| No Bacteremia (n=241) | Bacteremia (n=27) | p-value | |
| Durable MCS | 176 (73.0) | 22 (81.5) | 0.47 |
| Any Temporary MCS | 71 (29.5) | 5 (18.5) | 0.33 |
| IABP | 59 (24.5) | 4 (14.8) | 0.38 |
| Impella | 30 (12.4) | 2 (7.4) | 0.65 |
| ECMO | 23 (9.5) | 1 (3.7) | 0.51 |
| BiVAD | 21 (8.7) | 1 (3.7) | 0.60 |
| Invasive Fungal Infections (IFI) | |||
| No IFI (n=249) | IFI (n=19) | p-value | |
| Durable MCS | 184 (73.9) | 14 (73.7) | >0.99 |
| Any Temporary MCS | 70 (28.1) | 6 (31.6) | 0.95 |
| IABP | 60 (24.1) | 3 (15.8) | 0.59 |
| Impella | 28 (11.2) | 4 (21.1) | 0.37 |
| ECMO | 22 (8.8) | 2 (10.5) | >0.99 |
| BiVAD | 20 (8.0) | 2 (10.5) | >0.99 |
| Skin and Soft Tissue Device Site Infections (including Mediastinitis) (SSTI) | |||
| No SSTI (n=244) | SSTI (n=24) | p-value | |
| Durable MCS | 178 (73.0) | 20 (83.3) | 0.39 |
| Any Temporary MCS | 72 (29.5) | 4 (16.7) | 0.27 |
| IABP | 61 (25.0) | 2 (8.3) | 0.11 |
| Impella | 30 (12.3) | 2 (8.3) | 0.81 |
| ECMO | 22 (9.0) | 2 (8.3) | >0.99 |
| BiVAD | 20 (8.2) | 2 (8.3) | >0.99 |
Durable MCS includes patients with either Left- or Right- Ventricular Assist Devices.
Any Temporary MCS includes intra-aortic balloon pump (IABP), Impella, extracorporeal membrane oxygenation (ECMO), and biventricular assist devices (BiVAD). Patients may have had more than one device in place at any given time.
Acknowledgements:
The project described was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, Award Number TL1TR002546. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
This study was also supported in part by the Francis P. Tally, MD fellowship in Infectious Disease at Tufts Medical Center.
List of Abbreviations:
- 2SLS
Two-stage Least-squares
- BiVAD
Biventricular Assist Device
- BMI
Body Mass Index
- BSI
Blood Stream Infection
- CI
Confidence Interval
- CKD
Chronic Kidney Disease
- CMV
Cytomegalovirus
- CNS
Central Nervous System
- D(+/−)
Donor Positive / Donor Negative
- DM
Diabetes Mellitus
- ECMO
Extracorporeal Membrane Oxygenation
- HR
Hazard Ratio
- HSCT
Hematopoietic Stem Cell Transplant
- IABP
Intra-aortic Balloon Pump
- ICU
Intensive Care Unit
- IFI
Invasive Fungal Infection
- IPW
Inverse Probability Weighting
- ISHLT
International Society for Heart and Lung Transplantation
- IV
Instrumental Variable
- LVAD
Left Ventricular Assist Device
- MCS
Mechanical Circulatory Support
- OHT
Orthotopic Heart Transplant
- OR
Odds Ratio
- PH
Proportional Hazards
- PS
Propensity Score
- R(+/−)
Recipient Positive / Recipient Negative
- RRT
Renal Replacement Therapy
- RVAD
Right Ventricular Assist Device
- SD
Standard Deviation
- SOT
Solid Organ Transplant
- TMC
Tufts Medical Center
- UNOS
United Network for Organ Sharing
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Financial Conflicts of Interest: JKC: Merck, Kamada - research grants, Moderna - DSMB; AMS: Merk – research grant.; DRS: Merck, Seres Therapeutics, and Kamada – research grants; Moderna, Merck, Seres Therapeutics and Symbio – Consultant; ARV: Prior site PI for SHORE registry, sponsored by CareDx
References
- 1.Pons S, Sonneville R, Bouadma L, Styfalova L, Ruckly S, Neuville M, Radjou A, Lebut J, Dilly MP, Mourvillier B, Dorent R, Nataf P, Wolff M, Timsit JF. Infectious complications following heart transplantation in the era of high-priority allocation and extracorporeal membrane oxygenation. Ann Intensive Care. 2019. Jan 25;9(1):17. doi: 10.1186/s13613-019-0490-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lerman DT, Hamilton KW, Byrne D, Lee DF, Zeitler K, Claridge T, Gray J, Minamoto GY. The impact of infection among left ventricular assist device recipients on post-transplantation outcomes: A retrospective review. Transpl Infect Dis 2018. Dec;20(6):e12995. doi: 10.1111/tid.12995. Epub 2018 Oct 10. [DOI] [PubMed] [Google Scholar]
- 3.Angleitner P, Matic A, Kaider A, Dimitrov K, Sandner S, Wiedemann D, Riebandt J, Schlöglhofer T, Laufer G, Zimpfer D. Blood stream infection and outcomes in recipients of a left ventricular assist device. Eur J Cardiothorac Surg 2020. Nov 1;58(5):907–914. doi: 10.1093/ejcts/ezaa153. [DOI] [PubMed] [Google Scholar]
- 4.Nair RM, Kumar S, Saleem T, Chawla S, Vural A, Abdelghaffar B, Lee R, Higgins A, Cremer P, Rampersad P, Menon V. Characteristics and Impact of Bloodstream Infections in Cardiogenic Shock Patients on Temporary Mechanical Circulatory Support. JACC Cardiovasc Interv 2022. Oct 24;15(20):2110–2112. doi: 10.1016/j.jcin.2022.07.017. Epub 2022 Sep 28. [DOI] [PubMed] [Google Scholar]
- 5.Allou N, Lo Pinto H, Persichini R, Bouchet B, Braunberger E, Lugagne N, Belmonte O, Martinet O, Delmas B, Dangers L, Allyn J. Cannula-Related Infection in Patients Supported by Peripheral ECMO: Clinical and Microbiological Characteristics. ASAIO J 2019. Feb;65(2):180–186. doi: 10.1097/MAT.0000000000000771. [DOI] [PubMed] [Google Scholar]
- 6.Stevenson LW, Kormos RL, Young JB, Kirklin JK, Hunt SA. Major advantages and critical challenge for the proposed United States heart allocation system. J Heart Lung Transplant. 2016. May;35(5):547–9. doi: 10.1016/j.healun.2016.04.001. Epub 2016 Apr 16. [DOI] [PubMed] [Google Scholar]
- 7.Liu J, Yang BQ, Itoh A, Masood MF, Hartupee JC, Schilling JD. Impact of New UNOS Allocation Criteria on Heart Transplant Practices and Outcomes. Transplant Direct. 2020. Dec 15;7(1):e642. doi: 10.1097/TXD.0000000000001088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Varshney AS, Berg DD, Katz JN, Baird-Zars VM, Bohula EA, Carnicelli AP, Chaudhry SP, Guo J, Lawler PR, Nativi-Nicolau J, Sinha SS, Teuteberg JJ, van Diepen S, Morrow DA; Critical Care Cardiology Trials Network Investigators. Use of Temporary Mechanical Circulatory Support for Management of Cardiogenic Shock Before and After the United Network for Organ Sharing Donor Heart Allocation System Changes. JAMA Cardiol 2020. Jun 1;5(6):703–708. doi: 10.1001/jamacardio.2020.0692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Munoz-Price LS, Slifkin M, Ruthazer R, Poutsiaka DD, Hadley S, Freeman R, Rohrer R, Angelis M, Cooper J, Fairchild R, Barefoot L, Bloom J, Fitzmaurice S, Snydman DR. The clinical impact of ganciclovir prophylaxis on the occurrence of bacteremia in orthotopic liver transplant recipients. Clin Infect Dis 2004. Nov 1;39(9):1293–9. doi: 10.1086/425002. Epub 2004 Oct 11. [DOI] [PubMed] [Google Scholar]
- 10.Kumar R, Ison MG. Opportunistic Infections in Transplant Patients. Infect Dis Clin North Am 2019. Dec;33(4):1143–1157. doi: 10.1016/j.idc.2019.05.008. [DOI] [PubMed] [Google Scholar]
- 11.Clauss HE, Bettiker RL, Samuel R, Truant AL, Jungkind D, Suh B. Infections in Heart and Lung Transplant Recipients. Clin Microbiol Newsl 2012. Feb 1;34(3):19–25. doi: 10.1016/j.clinmicnews.2012.01.001. Epub 2012 Jan 17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.George MJ, Snydman DR, Werner BG, Griffith J, Falagas ME, Dougherty NN, Rubin RH. The independent role of cytomegalovirus as a risk factor for invasive fungal disease in orthotopic liver transplant recipients. Boston Center for Liver Transplantation CMVIG-Study Group. Cytogam, MedImmune, Inc. Gaithersburg, Maryland. Am J Med 1997. Aug;103(2):106–13. doi: 10.1016/s0002-9343(97)80021-6. [DOI] [PubMed] [Google Scholar]
- 13.Ki V, Rotstein C. Bacterial skin and soft tissue infections in adults: A review of their epidemiology, pathogenesis, diagnosis, treatment and site of care. Can J Infect Dis Med Microbiol 2008. Mar;19(2):173–84. doi: 10.1155/2008/846453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Baldwin RT, Radovancevic B, Sweeney MS, Duncan JM, Frazier OH. Bacterial mediastinitis after heart transplantation. J Heart Lung Transplant. 1992. May-Jun;11(3 Pt 1):545–9. [PubMed] [Google Scholar]
- 15.Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res 2011. May;46(3):399–424. doi: 10.1080/00273171.2011.568786. Epub 2011 Jun 8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Widding-Havneraas T, Zachrisson HD. A Gentle Introduction to Instrumental Variables. J Clin Epidemiol 2022. Sep;149:203–205. doi: 10.1016/j.jclinepi.2022.06.022. Epub 2022 Jul 7. [DOI] [PubMed] [Google Scholar]
- 17.Aslam S, Cowger J, Shah P, Stosor V, Copeland H, Reed A, Morales D, Giblin G, Mathew J, Morrissey O, Morejon P, Nicoara A, Molina E. The International Society for Heart and Lung Transplantation (ISHLT): 2024 infection definitions for durable and acute mechanical circulatory support devices. J Heart Lung Transplant. 2024. Jul;43(7):1039–1050. doi: 10.1016/j.healun.2024.03.004. Epub 2024 May 1. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.