Usefulness of surveillance Cultures for Carbapenem-resistant Enterobacteriaceae, Carbapenem-resistant Pseudomonas aeruginosa and Vancomycin-resistant Enterococci in Hematopoietic Stem Cell Transplant Unit


 Surveillance strategies to detect colonization is an important tool to prevent and control the spread of microorganisms especially among Hematopoietic Stem Cell Transplant (HSCT) patients. Colonization by Multidrug-resistant organisms (MDRO) has been evaluated as a risk factor for blood stream infection (BSI) in HSCT patients. The aim of this study was to evaluate the use of routine surveillance culture to screening colonization and infection by carbapenem-resistant Enterobacteriaceae (CRE), carbapenem-resistant Pseudomonas aeruginosa (CRPa) and vancomycin-resistant enterococci (VRE) in a HSCT unit. Methods Surveillance cultures were collected from patients admitted to the HSCT unit over one-year, with swabs for cultures on admission and then weekly until discharge. We compared surveillance culture positivity for each site and agent, also clinical and epidemiological data according to the colonization status. Results 200 HSCT patients underwent surveillance, with 1.323 samples collected. Infection due to MDRO occurred in 52 (21.5%) patients, among them 45 (86.5%) were blood stream infection (BSI) and 12 (23%) had positive surveillance culture before infection. 554 (41.8%) surveillance cultures were performed for CRPa, 413 (31.2%) for VRE, and 356 (27%) for CRE. Of these, 179 (13.5%) surveillance culture were positive, with greater positivity for oropharynx (6, 35.3%) CRPa, and rectal samples (16, 20.7%) for CRE. Being colonized by any MDRO, CRE (p <0.001) and CRPa (p = 0.027) was associated with a higher risk of infection in the bivariate analysis but being colonized was not associated with risk of death. Conclusion Previous colonization by MDRO was a significant risk factor for infection by these pathogens, mainly colonization by CRE. Overall, rectal swab was the best site with the higher positivity, and the oropharynx was also an option for CRPa investigation. Feces culture showed low positivity and should be avoided. Although the impact of the strategy on the mortality of patients undergoing HSCT is not clear, VRE surveillance should be questioned in auto-HSCT patients as it has an additional cost and little impact on survival.

Conclusion Previous colonization by MDRO was a signi cant risk factor for infection by these pathogens, mainly colonization by CRE. Overall, rectal swab was the best site with the higher positivity, and the oropharynx was also an option for CRPa investigation. Feces culture showed low positivity and should be avoided. Although the impact of the strategy on the mortality of patients undergoing HSCT is not clear, VRE surveillance should be questioned in auto-HSCT patients as it has an additional cost and little impact on survival.

Background
Infections are the major cause of death in Hematopoietic Stem Cell Transplant (HSCT) patients [1]. These patients are at high-risk for acquiring health care infections and the use of antibiotics during febrile neutropenia leads to a higher prevalence of multi drug resistant organisms (MDRO) in this population, [2] in addition to the risk of dissemination in the transplant unit.
A previous study in our hospital has identi ed previous gut colonization by MDRO, particularly by Gram Negative bacteria, as associated with blood stream infection (BSI) in patients undergoing HSCT [3].
Surveillance strategies to detect colonization have been considered important tools for preventing and controlling the spread of MDRO in the hospital setting [4,5]. However, the cost effectiveness of this strategy in HSCT units and its impacts in patient's outcome is still controversy [6][7][8].
The aim of this study was to evaluate the use of routine surveillance culture to track colonization and infection by carbapenem-resistant Enterobacteriaceae (CRE), carbapenem-resistant Pseudomonas aeruginosa (CRPa) and vancomycin-resistant enterococci (VRE) in a HSCT unit.
The Hospital das Clínicas of Faculdade de Medicina da Universidade de São Paulo (HCFMUSP) is a tertiary hospital, in São Paulo, Brazil, with 2,200 beds. The HSCT unit is a 23-bed unit with 4 beds designated to allogeneic and 18 to autologous transplant. The beds designated for the allogeneic transplant are individual bedrooms, with positive pressure air circulation with high-e ciency particle air lter, and the beds designated for the autologous transplant are shared bedrooms, with the maximum number of two beds per room.
Antibacterial prophylaxis was administered with levo oxacin on the rst day of stem cell infusion and discontinued when recovery of neutropenia or if the patients developed febrile neutropenia. Antiviral prophylaxis, antifungal and anti-Pneumocystis were administered according to guidelines [1].
Surveillance cultures were collected from patients admitted to the HSCT unit over one-year (2012). Swabs were collected for cultures on admission and then weekly until discharge, from multiple sites: axilla, feces, oropharynx and/or rectum.
The culture swabs for CRE and CRPa were incubated overnight in liquid media and then plated in Agar MacConkey medium with a meropenem disk (Ferreira et al 2018), and surveillance cultures of feces samples for VRE, in medium with vancomycin 6 mg/L [9].
Colonization was de ned as the presence of at least one positive surveillance culture for one of the studied microorganisms.
Clinical and epidemiological data were collected regarding sex, age, length of stay in BTM unit, diagnosis, use of antibiotics, infection, blood stream infection (BSI), and intra-hospital death. Infection was de ned as CDC guidelines [10].

Statistical analysis
All data was stored in a database in Excel 97-2004 (Microsoft, Redmond, WA, United States). We compared surveillance culture positivity for each site and agent. Clinical and epidemiological data were analyzed according to the colonization status. All statistical analyses were performed using EpiInfo 7.0 (CDC, Atlanta, United States) Fisher's exact test or Chi-square test were used for categorical variables, as appropriate, and Mann-Whitney and Log-Rank testes for continuous variable. Univariate analysis and multivariate logistic regression analysis were performed (95% con dence interval). We considered a P value < 0.05 as statistically signi cant. A Kaplan-Meier curve was generated to compare survival among patients with and without BSI.

Results
A total of 200 HSCT patients underwent surveillance, with 1.323 samples collected. The mean age was 45 years, 107 (53.5%) males and 17.7% of hospital death, and 59% patients performed Auto-transplantation as shown in Table 1.   (Tables 3 and 4).  CRE infection occurred more signi cantly among CRE colonized (p = 0.004), and CRPa infection occurred more among CRPa colonized (p = 0.002). Having any MDRO was associated with a higher risk of infection in a bivariate analysis ( Table 2).
Chance of Survival was signi cantly lower among patients with BSI, as demonstrated in the Kaplan Meier curve (P = 0.012), as well as among those infected with CRPa (0.0053).

Discussion
The prevalence of colonization and type MDRO varies greatly according to the study centers, depending on epidemiology and local practices, such as the use of antibiotic prophylaxis, type of HSCT, among other factors [5][6][7][8]. In this study, the positivity of the surveillance culture was 13.5% lower than data from Europe [4,11]. CRPa and VRE were the main MDRO, both colonizing 42% of patients with MDRO. The average length of hospital stays until the rst positive culture varied according to the pathogen, the time for colonization with VRE was 10 days while for CRPa was much longer (20 days). Heidenreich et al diagnosed 27% of colonization by MDRO in the rst 100 days after transplantation, with a predominance of CRPa (26.9%), with a pre-induction prevalence of 16% in German (11). Sadowska-Klasa in Poland obtained 42% colonization by MDRO during hospitalization, with an important predominance of VRE [4].
In this cohort, previous colonization by MDRO was a signi cant risk factor for infection by these pathogens, mainly colonization by CRE. A study at our center, in the 2014-2015 period, also reported that colonization by previous MDRO was associated with BSI (p < 0.001), with 20% of patients colonized by GNB-MDR developed BSI by these agents [3]. Other studies have also found similar ndings [12,13] Therefore, strategies for selective decolonization of the gastrointestinal tract (SDD) have been evaluated. A single center study demonstrated costeffectiveness of SDD in CRE colonized patients in intensive care units [14]. However, a systematic review of 2019 still classi es the evidence indicating bene t in decolonization as limited and does not recommend this routine intervention, moreover. studies with immunocompromised patients are still extremely scarce in the literature [15].
Although, colonization is associated with a higher risk of infection, the impact of the strategy on the mortality of patients undergoing HSCT is not clear. We observed that infection by any MDRO, CRE and mainly by CRPa was associated with the risk of death; moreover, being colonized was not a risk factor for death. In other studies, a higher risk of death was observed in colonized patients. Sadowska-Klasa et al. [4] and Bilinski et al. [7] evaluated patients undergoing allogeneic HSCT and in both studies, colonization by MDRO had an impact on overall survival (OS) in 1 year. The relative high rate of Auto-HSCT in our series (59%) may explain our results.
In our study, we obtained a high prevalence of VRE colonization; which was the only MDRO that did not affect the risk of infection. Perhaps its lower virulence in relation to gram-negative bacillus (GNB) has in uenced the lower impact of colonization by MDRO on hospital mortality in our population [16]. In contrast, a study conducted at the Mayo Clinic, with a 10-year series evaluated the in uence of colonization by VRE on the prognosis of patients undergoing Allo-HSTC by AML. In multivariate analysis, colonization by VRE was an independent risk factor for VRE infection, but did not in uence any other post-transplant outcome [8].
Corroborating this point, in a multicenter study carried out in Italy (52 centers), being colonized by resistant gramnegative bacteria signi cantly reduced 4-month survival. In addition, colonization by CRE and CRPa increased the risk of infection with these pathogens (p < 0.001) [16]. On the other hand, Heidenreich and colleagues [11] also found a similar risk of death regardless of the status of colonization by MDRO, even though CRE was the main colonizer in his studied population [11].
Our data corroborate the fact that the rectal swab is more sensitive than the culture of feces [18,19], especially regarding VRE and CRE. Despite this, the use of rectal swab in HSCT patients should be used with care, to avoid skin or mucosal breakdown during severe neutropenia. In this scenario, the oropharyngeal swab may be an alternative, since it presented a high positivity for CRE and VRE. We observed that the type of surveillance culture site depends on the pathogen, feces culture showed low positivity and should be avoided; and VRE surveillance should be questioned in patients undergoing autologous transplantation because it has an additional cost and little impact on survival and development of bloodstream infection.
The present study, in addition to being carried out in a single center, has the disadvantage of being retrospective. However, it brings important re ections to the practice of screening MDR in TCTH patients. Firstly, being colonized by MDRO does not seem to be a su cient factor to interfere in post-HSCT survival, especially VRE colonization and in auto-HSCT patients.