Colonization with multidrug-resistant organisms is associated with in increased mortality in liver transplant candidates

Objectives Rising prevalence of multidrug-resistant organisms (MDRO) is a major health problem in patients with liver cirrhosis. The impact of MDRO colonization in liver transplantation (LT) candidates and recipients on mortality has not been determined in detail. Methods Patients consecutively evaluated and listed for LT in a tertiary German liver transplant center from 2008 to 2018 underwent screening for MDRO colonization including methicillin-resistant Staphylococcus aureus (MRSA), multidrug-resistant gram-negative bacteria (MDRGN), and vancomycin-resistant enterococci (VRE). MDRO colonization and infection status were obtained at LT evaluation, planned and unplanned hospitalization, three months upon graft allocation, or at last follow-up on the waiting list. Results In total, 351 patients were listed for LT, of whom 164 (47%) underwent LT after a median of 249 (range 0–1662) days. Incidence of MDRO colonization increased during waiting time for LT, and MRDO colonization was associated with increased mortality on the waiting list (HR = 2.57, p<0.0001. One patients was colonized with a carbapenem-resistant strain at listing, 9 patients acquired carbapenem-resistant gram-negative bacteria (CRGN) on the waiting list, and 4 more after LT. In total, 10 of these 14 patients died. Conclusions Colonization with MDRO is associated with increased mortality on the waiting list, but not in short-term follow-up after LT. Moreover, colonization with CRGN seems associated with high mortality in liver transplant candidates and recipients.

Introduction Liver transplantation (LT) is an established treatment in patients with acute liver failure and advanced liver disease with and without hepatocellular carcinoma [1,2]. Bacterial infections are a major cause of short-term mortality after LT, and unresolved infections are considered a contraindication against liver transplantation [3]. Moreover, infections are a major trigger of acute-on-chronic liver failure in patients both with compensated and decompensated cirrhosis [4][5][6].
Rising prevalence of multidrug-resistant organisms (MDRO) and especially colonization with multidrug-resistant gram-negative bacteria (MDRGN) is considered one of the most crucial and yet unresolved problems in public healthcare [7,8]. In a study of 475 liver graft recipients from Asia, MDRGN were identified as the dominant pathogens in liver transplant recipients [9]. Moreover, infections with MDRO in patients with cirrhosis have been associated with increased mortality [10,11] and liver transplant recipients have a substantial risk for infections with MDRO [12,13]. Furthermore, infections with MDRGN are a serious complication after LT [13,14]. The situation is less clear for colonization with MDRO, especially in LT candidates or recipients with vancomycin-resistant enterococci (VRE) colonization. So far, data are scarce with respect to current prevalence of MDRO colonization in LT candidates and even more the impact of MDRO colonization on the mortality prior and after LT.
In Germany, organ shortage results in a high model of end stage liver disease (MELD) score at LT and a prolonged waiting time between listing for LT and graft allocation [15,16]. It is accepted that an uncontrolled infection regardless of MDRO status remains a contraindication against LT in common. However, there is uncertainty about the prospects of MDRO-colonized patients awaiting LT. The aim of the current study was to determine the prevalence of MDRO colonization at our center at the time point of listing, and the incidence of new MDRO colonization during the waiting time for LT. Moreover, we addressed whether MDRO colonization is associated with higher mortality prior and after LT.

Study design
This retrospective study was conducted at a German tertiary liver care unit in patients with indication for LT. The standard LT evaluation protocol included a screening protocol for MDRO colonization comprising of nasopharyngeal/throat, rectal, and optional cutaneous smear samples/swabs. Following the recommendations of the German Commission for Hospital Hygiene and Infection Prevention (KRINKO), the local hospital's infection control strategy requires MDRO screening in case of hospitalization and/or admission to the intermediate and intensive care unit [17]. Inclusion criteria contained evaluation and formal registration for LT and age of at least 18 years. Study approval was obtained by the local Ethics Committee for Medical Research of the Medical Faculty, University Hospital, Goethe University, Frankfurt am Main, in accordance with the 1975 Declaration of Helsinki prior to research (file number 268/13). Informed consent was obtained upon study inclusion, and the database was pseudoanonymized. Patients who had been listed for LT, but died before informed consent could be obtained, were included into the analysis in accordance with the Ethics Committee vote. Screening for MDRO colonization had been repeated at planned and emergency hospitalization events including time point of LT. Moreover, laboratory and clinical data were collected and analyzed retrospectively including infectious complications. Follow-up period was at least one year after evaluation for LT, or at least three months in patients who received a liver graft. Data acquisition was closed by July 2019, and thus six patients without LT were followed up for less than one year (286, 319, 321, 354, 357 and 361 days). Fatal outcome was defined as primary end point, and patients who did not reach this end point were censored from the last day of follow-up. Fatal outcome due to severe infection, tumor progression, and other causes were defined as secondary endpoints. Survival status of patients who were lost to follow-up was surveyed by telephone interview, in correspondence with hospitals of referral, or via public registration office inquiry. Data from individual patients may have been reported previously with respect to different topics [18][19][20].

Definition of MDRO, colonization, invasive detection, infection, and severe infection
In the current study, MDRO were defined as MDRGN, VRE and methicillin-resistant Staphylococcus aureus (MRSA), as defined earlier [21]. In particular, MDRGN are defined as Enterobacterales with extended spectrum beta-lactamase (ESBL) phenotype as well as Enterobacterales, Pseudomonas aeruginosa, and Acinetobacter baumannii resistant against Piperacillin, any 3rd/4th generation cephalosporin, and fluoroquinolones [22]. CRGN are a subgroup of MDRGN that, beside ESBL phenotype (Enterobacterales) or resistance against piperacillin, ceftazidime and fluoroquinolones (P. aeruginosa), carry additional resistance to carbapenems [23]. Colonization with MDRO was defined as detection of MDRO in rectal, pharyngeal/throat, or cutaneous screening samples. Importantly, multiple different MDRGN strains may have been detected within single patients. Moreover, all MDRO strains isolated from ascites, blood, urine, bronchial or pleural secretion, bile, pus, wounds or surgical sites, on medical devices or indwelling catheters were defined as invasive MDRO detections. Diagnosis of infection was based on the combination of invasive detection and according clinical parameters, i.e. fever or laboratory signs of infection. The quick sequential organ failure assessment (qSOFA) score was applied to stratify severity of infection as recommended [24].

Detection of MDRO and molecular resistance analysis
For MDRGN screening CHROMagar™ ESBL plates (Mast Diagnostica, Paris, France) were used allowing the detection of drug-resistent Enterobacterales, P. aeruginosa, and Acinetobacter spp. Screening for VRE and MRSA colonization was performed using VRE selective agar (bioMérieux, Nuertingen, Germany) and Briliance MRSA 2 agar (Oxoid, Wesel, Germany), respectively. All recovered isolates were tested for their respective resistance profile. All laboratory procedures were performed under quality-controlled standards (laboratory accreditation according to ISO 15189:2007 standards; certificate number D-ML-13102-01-00, valid through January 25, 2021), as described earlier [25].

Statistical analysis
Baseline parameters. Statistical analyses were conducted with BiAS software (v11.06; Epsilon-Verlag, Darmstadt, Germany) and R (version 4.0.2, R Core Team (2020), Vienna, Austria, packages kmi and survival). Categorical variables were described as frequencies and percentages. Continuous variables were presented as medians or means, with ranges or quartiles, as appropriate. The Wilcoxon-Mann-Whitney-U-test was used for comparisons of quantitative and ordinal variables at baseline or at LT. All tests were two-sided and p-values �0.05 were considered statistically significant. Calculating the probability of undergoing LT, LT itself was considered as endpoint, and de-listing and death were defined as competing event.
Time-dependent endpoints and competing risks. Importantly, clinical endpoints were assessed depending on MDRO acquisition, which is a time-dependent variable. Regarding time-to-event analyses, the subsequent outcomes were analyzed with regard to the following details. i) Death on the waiting list as the event of interest up to one year after MDRO screening, considering de-listing and LT as competing events. ii) Death due to severe infection as the event of interest, with death due to other causes than severe infection being defined as additional competing risk. iii) We also analyzed the incidence of MDRO colonization on the waiting list by registering all new MDRO detections as a time-dependent variable. Finally, cumulative incidence of new MDRO colonization was calculated for patients who were MDRO-negative at study inclusion. Hereby death, LT, and de-listing were regarded as competing events. Results were depicted as log-hazard ratios (log-HR). Independent risk factors for events of interests were calculated using uni-and multivariate Cox regression (death), and a proportional sub-distribution hazards' regression model (all other end points). Hereby, MDRO colonization was included as time-dependent factor.
Endpoints in LT recipients with a "clock-reset" at LT. Since LT is a curative approach in cirrhosis, a different hazard for the abovementioned endpoints and competing risks was to be expected upon LT. In this particular analysis, LT was defined as time point zero, and we assessed the cumulative incidences of MDRO colonization up to 3 months after LT as a timedependent variable, as described above. Hereby death was regarded as competing event. We also conducted a time-to-event analysis of death and death due to severe infection up to 3 months after LT, where for the latter case death due to other causes than severe infection was defined as competing risk.

Clinical course in association with MDRO status including mortality analysis
Among 351 patients, 101 died on the waiting list (79.8% one-year survival rate), and evidence of MDRO was associated with increased risk for death (Fig 1, Table 2). The competing risk analysis showed a higher mortality in patients with MDRO colonization compared to noncolonized patients (HR = 2.57, p<0.0001; Table 2). Furthermore, in patients with MDRO, increased waitlist mortality was observed, and lethal severe infection occurred significantly more often (Fig 2, Table 2). Importantly, in patients receiving a liver graft, MDRO in general Uni-and multivariate analysis of potential confounders of MDRO colonization or survival showed that ICU, hepatic encephalopathy, MELD score, recent hospitalization, and recent antibiotic exposure were independent risk factors for death (univariate analysis). In the  multivariate analysis, MDRO status, MELD score and recent hospitalization proved to be predictors of death on the waiting list (Table 3). Among all 164 patients undergoing LT, 23 died within 3 months (90-days survival rate 85.9%). Thereof, 11 died from severe infection, and 12 died from other causes. Within the entire cohort, 146/351 patients died, with 41/146 fatalities from severe infection, 13/146 from tumor progression, and 68/146 from other non-infectious causes. In 24/146 patients, cause of death could not be determined; these fatalities had been reported to us through a public registration office inquiry.

Conclusions
In the current study, we determined the prevalence of MDRO colonization at the time point of listing for LT, and furthermore the incidence of new MDRO acquisition in initially not colonized patients. Our study demonstrates that MDRO colonization is common in LT candidates and therefore an important medical issue. Moreover, the cumulative incidence of MDRO increased considerably after listing until LT or last follow-up. While a survival benefit in patients undergoing LT is obvious in both in MDRO-positive and negative patients, our study showed that colonization with MDRO was associated with increased mortality in patients on the waiting list.
A detailed analysis of MDRO species showed that mortality was associated with all MDRO subtypes, while lethal infectious complications were mainly associated with VRE. Rising VRE incidence has been observed in LT candidates during the past 15 years [26,27]. It is not yet conclusively determined whether VRE colonization is a surrogate marker reflecting sicker patients receiving repeated antibiotic exposure that leads to intestinal dysbiosis, or whether VRE colonization has a pathophysiological influence on the clinical course of patients [28]. Indeed, a meta-analysis showed that VRE colonization may prompt infection in solid organ recipients, and amongst these especially in liver graft recipients [27]. A study reporting oneyear mortality rates as high as 60% in VRE carriers within a tertiary liver center in the United States did not discriminate between LT candidates and recipients, hence patients at high risk for death due to VRE were not unambiguously identified within this pooled cohort [29]. In our study, detection of VRE was associated with increased risk for death in LT candidates, but not recipients. Therefore, the data from our center support the approach that LT should swiftly be intended in VRE-positive patients.
The second important finding of our study was a noninferior clinical outcome in MDROcolonized patients who underwent LT, compared to non-MDRO LT recipients (Table 2). In this regard, the correlation between MDRO colonization and MDRO-associated infection must be addressed. Especially in patients with MDRGN positivity, the progression risk from colonization to infection has been well described [12,13,[30][31][32]. Since LT is a curative approach in cirrhotic patients, we assume that this also translates into a lesser impact of MDRO-associated lethal complications. However, the situation of CRGN colonization must be addressed separately, because detection of CRGN seems to be associated with deleterious outcome in LT candidates. This is in concordance with other studies indicating an unfavorable course in patients with CRGN colonization [13,25]. The data of our study do not conclusively clarify whether CRGN colonization may be considered a contraindication to LT. It seems reasonable to consider the given resistance pattern in CRGN-colonized LT candidates with respect to potential rescue antibacterial therapies like ceftazidim/avibactam, cefiderocol, meropenem/varbobactam in the decision process. The presented data of our study on liver transplantation in CRGN patients nevertheless underscores the importance of a high level of alertness and the need for an interdisciplinary approach including specialists in infectiology to address CRGN infections in patients undergoing LT.
While our study indicates that the incidence of MDRO colonization rises during waiting time for LT and that colonization itself is associated with increased mortality, some limitations have to be taken into account when interpreting the data. First, MDRO colonization may represent a surrogate parameter since they are more often detected in sicker patients; however, our data show that MELD score was equal in MDRO-positive patients at LT. Second, data analysis was done retrospectively, and records of infectious complications were not prospectively defined per protocol. Third, as a consequence of the longitudinal study design, the mean survival time could not be calculated reliably. Fourth, epidemiological data of this single-center study may vary from others since local microbiological patterns differ significantly in between regions worldwide [7].
On the other hand, some important strengths of our study may be highlighted. The infection control strategy of our center ensured a literally complete screening of LT candidates at evaluation and consistent MDRO surveillance thereafter. Moreover, extending over a period of nine years, this study covers one of the longest time spans in the field of MDRO bearers on the LT waiting list. Finally, prevalence and incidence data are reliable due to the structured MDRO infection control protocols at our center and our cohort appropriately reflects a realworld scenario in a high-MELD era. Thus, concise MDRO screening and surveillance seem essential in LT candidates. Moreover, any extension of waiting time puts LT candidates on risk for MDRO colonization and subsequently increased mortality.
In conclusion, MDRO colonization is common and an independent predictive factor for mortality in LT candidates. The occurrence of CRGN resembles a major event in these patients and must be addressed with particular concern, possibly warranting infectiologic stewardship. Moreover, our findings emphasize the need for strategies to reduce the waiting time for LT and overcome organ shortage. Future research should aim at decolonization strategies as well as disrupting the progression from MDRO colonization to infection in these patients.