Incidence of Invasive Infections Among Hemato-Oncology Patients with Significant Burden of Candida in Stool

Candidemia is a serious infection associated with increased mortality. It is unclear whether a high concentration of Candida in stool in patients with hematologic malignancies is associated with a higher risk for developing candidemia. In this observational historical study in patients hospitalized in hemato-oncology departments, we describe the association between gastrointestinal Candida colonization and the risk for candidemia and other severe outcomes. Data from 166 patients with heavy burden of Candida in stool were collected and compared to a control group of 309 patients with minimal or no Candida in stool, from 2005 to 2020. Severe immunosuppression and recent use of antibiotics were more common in heavily colonized patients. Outcomes of heavily colonized patients were worse as compared to the control group with statistical significance in 1-year mortality (53% vs. 37.5%, p = 0.001) and borderline statistical significance in candidemia rate (12.6% vs. 7.1%, p = 0.07). Risk factors for 1-year mortality were significant colonization of Candida in stool, older age and recent use of antibiotics. In conclusion, significant stool burden of Candida among hospitalized hemato-oncology patients may pose a risk for 1-year mortality and increased candidemia rate.


Introduction
Invasive candidiasis, mainly candidemia, is the most common fungal disease among hospitalized patients, most commonly among intensive care unit (ICU) patients and immunocompromised hosts [1]. It is estimated that candidemia affects more than 250,000 people worldwide annually, with more than 50,000 deaths [1,2].
Traditionally, Candida albicans was the dominant pathogen, but over the past few decades the incidence of non-albicans strains is increasing and an increasing number of studies world-wide are describing nonalbicans Candida as the cause of at least half of candidemia cases [3]. This change in species distribution is worrisome due to increased resistance to antifungal therapy, mainly azoles, among non-albicans strains, and has implications for therapeutic strategies [2].
The origin of candidemia is mainly the gastrointestinal (GI) tract, and occasionally the skin, via cutaneous colonization near the insertion site of venous catheters. It has been postulated that stool colonization with Candida spp. (mainly Candida parapsilosis) precedes colonization of the skin [4,5].
Risk factors for candidemia include intensive care unit (ICU) stay, central venous catheters, total parenteral nutrition (TPN), history of abdominal surgery and use of broad-spectrum antibiotics and immunosuppression (mainly hematologic malignancies, solid organ or hematopoietic transplantation). Candida colonization in multiple sites is strongly associated with candidemia in critically-ill ICU patients [6,7].
The GI origin of candidemia in patients with hematologic malignancies is commonly chemotherapy induced-mucosal damage, immunosuppression (e.g. neutropenia), frequent treatment with broadspectrum antibiotics, known to suppress the gut normal microbiota and frequent admissions in ICU [2,8]. It was never clearly demonstrated that a high concentration of Candida in stool in these patients, is associated with a higher risk for developing candidemia, though several relatively small studies have shown this connection among patients with selected hematologic malignancies, such as leukemia [9][10][11].
This study aimed to assess whether the presence of significant Candida burden in stool of patients with hematologic malignancies, is a risk factor for developing candidemia, severe infections and mortality and to identify risk factors for these outcomes.

Materials and Methods
This study was an observational historical analysis of data retrieved from medical records from 1 January 2005 to 31 December 2020. The study was performed at the hemato-oncology departments (hematology, oncology and bone marrow transplantation) at the Hadassah Medical Center in Jerusalem, a 900-bed inpatient tertiary hospital, serving a population of over one million people. The study was approved by the local IRB (HMO-12-0460).

Patients
The Study included patients of all ages who were admitted to hemato-oncology departments, were immunosuppressed secondary to malignancy and/or immunosuppressive therapy, and had their stool examined microscopically in the microbiology laboratory (mostly because of diarrhea), according to a routine laboratory protocol, with reported significant burden of Candida (significant colonization). Controls were similar patients in whom examination of stool did not identify significant amount of Candida.
Study subjects were Israeli citizens and non-Israelis from the Palestinian authority and Eastern Europe (former USSR region).

Laboratory Work-up
Fecal specimens were examined using light microscopy (Carl Zeiss NTS Ltd., Germany) by a single experienced laboratory technician. Significant burden of Candida was reported when yeast forms were observed throughout the field at X100 magnification.
Other observations were considered as non-significant colonization, as per the laboratory protocol.

Data
Demographic, clinical and laboratory data were collected from patients' medical files. Parameters were compared between patients with significant burden of Candida in stool and those without. The date of stool sample collection was the reference date.

Background Comorbidities
Charlson Comorbidity index (CSI) was used to asses overall medical condition in admission. A cut-off of 5 points and above was selected to define patients with high risk for mortality [12,13].
Chemotherapeutic treatment was defined as conventional cytotoxic medications administered in the 3 months prior to admission.
Corticosteroid therapy was defined as C 0.3 mg/kg of prednisone or equivalent for C 2 weeks in the preceding 3 months.
Biologic therapy was defined as therapy that was given in the 3 months prior to admission and included BCR-ABL inhibitors (eg Imatinib), monoclonal antibodies, BCL-2 inhibitors (e.g. Venetoclax), proteasome inhibitor (e.g. Bortezomib) and the drug lenalidomide. Anti CD20 and BTK inhibitor therapy if given in the preceding year were also included.
Previous antibiotic/antifungal therapy was defined as a beta-lactam with gram-negative activity or a carbapenem or an azole, which were administered for C 72 h in the preceding 3 months.

Hospitalization Parameters
Under hematologic disorders, we included hematologic malignancies and other disorders associated with neutropenia, such as aplastic anemia and hematopoietic stem cell transplantation (HSCT) due to marrow failure (e.g. osteopetrosis, aplastic anemia). Diagnoses were divided by risk of neutropenia: acute myeloid leukemia (AML) and acute lymphoblastic leukemia (ALL) were referred to as 'acute leukemia' together with myelodysplastic syndromes (MDS). Lymphoproliferative disorders and multiple myeloma were considered separate categories.

Outcomes: Infections and Mortality
Candidemia was defined as C 1 blood cultures growing Candida within 1 month before/after stool sample collection.
Bacteremia was defined as C 1 positive blood cultures, that were considered by the study investigators to represent true bacteremia, within 1 month before or after stool sample collection.
Gram-negative isolates that were resistant to 3rd generation cephalosporin were considered as extended-spectrum beta-lactamase (ESBL) producers.
Carbapenemase-producing Enterobacteriaceae (CPE) were bacterial isolates found to produce carbapenemase in the microbiological laboratory by PCR, as part of the hospital screening protocol.
Invasive fungal infection (IFI) was primarily defined as a clinical diagnosis by the attending physicians at the time of hospitalization, and their decision to treat it as such; Cases were diagnosed according to either a biopsy and pathology results, or clinical presentation, imaging studies and fungal biomarkers.
Mortality was divided into 4 categories: 1) infection (bacterial/fungal) associated in-hospital mortality 2) in-hospital mortality not related to infection 3) 30-day mortality 4) mortality at 1-year post stool sampling.

Statistical Analysis
For univariate analysis, categorical parameters were analyzed using chi-square test and Fisher's exact test and quantitative parameters were analyzed using Student's t-test. All the tests were two-tailed and P value of \ 0.05 was determined as significant. A logistic regression model was created for multivariate analysis with candidemia, bacteremia and mortality as the dependent variables. We included also nonsignificant parameters from the univariate analysis in effort to achieve a more comprehensive model. Software product used were JASP (v.0.14.1, JASP Team) and WINPEPI v.11.65.

Results
During the study period, there were 70,400 admissions of patients with hemato-oncologic disorders to our hospital. Of those, 475 (0.6%) stools were examined under a light microscope, 166 were found to have significant Candida burden and 309 did not.

Patients' Characteristics and Outcomes
Background and clinical characteristics of 475 patients included in the study are detailed in Table 1. Patients with significant burden of Candida in stool were younger compared to those without. Amongst patients with hematologic malignancies, acute leukemia and MDS were more frequent in patients with significant Candida burden. Patients with significant burden received more frequently chemotherapy in the preceding 3 months, and concordantly had significantly higher rates of severe (less than 500/lL) and prolonged ([ 1 week) neutropenia. Beta-lactam antibiotics, excluding carbapenem, that were administered for C 72 h in the previous 3 months, were prescribed more frequently in patients with significant burden, whereas carbapenem was prescribed more often in the control group. There was no difference in prior azole treatment (C 72 h in the preceding 3 months) between the groups.
Both groups had a similar rate of bacteremia, mainly with Enterobacterales, though the rate of extended spectrum b-lactamases (ESBL) producing Enterobacterales was lower in patients with significant burden of Candida in stool, compared to the control group, as well as the rate of carbapenemaseproducing Enterobacterales (CPE) fecal carriage ( Table 2).
There was a higher rate of candidemia amongst patients with significant burden of Candida in stool, with a statistical trend and borderline significance. Most candidemia patients had acute leukemia or MDS. There was no difference in other invasive fungal infections between the groups (Table 2).
There was no difference in 30-day mortality between patients with and without significant burden. However, at 1 year, the mortality rate was significantly higher among patients with significant burden ( Table 2). (Table 3 and Supplementary Table 1

Discussion
Given the high mortality rate of patients with candidemia and the increased rate of azole-resistant Candida glabrata surpassing C. albicans as a major causative agent [3,15] it is important to target patients at high risk among hemato-oncology patients. In this study we examined the role of significant Candida burden in stool as a risk factor for severe infections and outcomes in immunosuppressed hemato-oncology patients whose stool was examined for the presence of yeasts.
Patients with significant stool burden tended to have higher rates of candidemia compared to patients with no or modest burden, though this finding did not reach full statistical significance, probably due to a small sample size. In addition, the 1-year mortality was significantly higher among the GI colonized group. Since both groups had similar background characteristics, the higher rate of candidemia and mortality in colonized patients can be attributed to two main findings in the study (Table 1): 1) Severe immunosuppression: Patients with significant burden of Candida in stool had mainly acute leukemia and MDS, received more chemotherapy compared to the control group and had prolonged and severe neutropenia. 2) Frequent treatment with beta-lactam antibiotics in the significant-burden group, known to alter the gut normal microbiota; all considered known risk factors for candidemia [16][17][18][19][20] and mortality [18,21].
Previous studies have shown that 47.1% of ICU patients with candidemia were GI tract-colonized with yeasts within the last 7 days before documentation of bloodstream infection (BSI) [6]. The connection between Candida GI colonization and invasive disease in patients with hematologic malignancies is also described, however to a lesser extent [2,6,9,10] and mainly in acute leukemia patients, whose degrees of neutropenia and immunosuppression are marked [9][10][11]. In addition, studies involving microbiologic analysis have shown that hematologic patients with candidemia experienced a prior marked intestinal expansion of pathogenic Candida species and that fecal colonization with Candida may increase risk for graft failure and mortality in patients undergoing HSCT [22,23]. There are also reports that significant colonization of Candida in the gut may promote inflammation [24]. We argue that the finding that GVHD as a cause of diarrhea is more common among patients with significant colonization in stool (Table 1) may be related to these observations, as it could represent increased inflammation in the GI tract. Candidemia risk factors identified in logistic regression were known risk factors such as profound neutropenia and previous beta-lactam use [2]. Non-Israeli origin was also identified as a risk factor, possibly since such patients were often referred to our hospital following unsuccessful treatment and several courses of broad-spectrum antibiotics [25][26][27] (supplementary Table 1).
As expected, we found bacteremia rate was more common than fungal infections including candidemia, with an average rate of 34% among both groups. This rate is relatively higher than the previously reported * 20% [28]. Most isolates were Enterobacterales, with a similar rate in both groups ( Table 2). Bacteremia risk factors were severe and prolonged neutropenia, previous use of a beta-lactam and non-Israeli origin (supplementary Table 1). As noted above, we attribute the latter to extensive previous antibiotic use. Prolonged neutropenia has been well described as a major risk factor for serious infectious , solid tumor with bone marrow metastasis (n = 21), osteopetrosis (n = 14), thalassemia major (n = 10), genetic immunodeficiency syndromes (n = 5). j ANC \ 500/lL, C 1 week duration complications [29] and antibiotics, if used inappropriately, were formerly associated with bacteremia and with mortality [30]. The similarity in bacteremia rate between patients with significant Candida burden in stool and those without, is a surprising finding, given that the patients in the first group received more chemotherapy and had a higher rate of severe and prolonged neutropenia. A possible explanation is a protective role of significant Candida GI burden against bacteremia. Supporting this argument is the finding of decreased rate of ESBL bacteremia and CPE carriage among Candida GI colonized patients, compared to the control group (Table 2), as well as the protective role of significant Candida burden in stool  against ESBL bacteremia found in logistic regression (supplementary Table 1). Recent studies that used murine models and showed a potential benefit for gut colonization with Candida against infections caused by Clostridioides difficile and Acinetobacter baumannii, also support this speculation [31][32][33]. To further clarify this question, strong, prospective studies are required, particularly in hematologic patients.
The mortality rate at 1 year was significantly higher in stool colonized patients, as opposed to the 30-day mortality that did not differ between the groups ( Table 2) and was affected mainly by neutropenia and the presence of candidemia (Table 3). This finding  Table 1). Infection-associated mortality rate was * 13% in both groups (Table 2), compatible with previous reports of 5-40% among hemato-oncology patients with bloodstream infection. Our study has limitations. First, this was a retrospective study with a limited sample size. Therefore, additional studies are required to validate our results. Secondly, an estimation of stool volume and Candida quantification in stool samples were not used, as per our laboratory routine, though the estimation of Candida burden was performed by a skilled laboratory technician. Lastly, all samples were collected from patients suffering from diarrhea, that for itself may be a risk factor for GI mucosal disruption and secondary candidemia. Alternatively, although the presence of Candida in stool is not considered as an enteric pathogen, it is sometimes described as a cause of chronic secretory diarrhea, mainly in the elderly and malnourished patients [34].
In conclusion, significant burden of Candida in stool among hemato-oncology patients requiring hospitalization, is associated with severe morbidity status and may pose a risk for invasive candidiasis and 1-year mortality. Patients who are significantly GI colonized with Candida should be closely monitored for signs of infection and should receive timely antifungal treatment to reduce the risk of severe infection. Further studies are required to establish the efficacy of such approach.
Author Contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by RE and MK. The first draft of the manuscript was written by RE and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

Declarations
Conflict of interest The authors have no relevant financial or non-financial interests to disclose. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.