Impact of Perioperative Absolute Neutrophil Count on Central Line-Associated Bloodstream Infection in Children With Acute Lymphoblastic and Myeloid Leukemia

Background There is lack of evidence concerning safety of placement of tunneled central venous catheters (TCVCs) in neutropenic children with acute leukemias. Here, we evaluate the impact of absolute neutrophil count (ANC) at the time of TCVC placement on development of central line-associated bloodstream infections (CLABSI) in children with lymphoblastic (ALL) or myeloid leukemia (AML). Materials and Methods A retrospective observational study of children undergoing TCVC placement at a tertiary referral hospital between January 2000 and December 2019 was performed. Traditional and competing-risks regression models were used to estimate the effect of perioperative ANC on development of CLABSI. Results A total of 350 children (median age 6.4 [IQR: 3.1–10.9] years) underwent 498 consecutive TCVC implantations in neutropenic (n = 172, 34.5%) and non-neutropenic conditions (n = 326, 65.5%). The median length of observation per TCVC was 217.1 (IQR: 116.1–260.5) days with a total of 99,681 catheter days (CD). There were no differences in early (within first 30 days after TCVC placement) and overall CLABSI rates between neutropenic and non-neutropenic patients (HR 1.250, p = 0.502; HR 1.633, p = 0.143). We identified female sex (HR 2.640, p = 0.006) and the use of TCVC for treatment of relapsed leukemia (HR 4.347, p < 0.0001) as risk factors for early CLABSI and the use of double-lumen catheters (HR 2.607, p = 0.003) and use of TCVCs during leukemia relapse (HR 2.004, p = 0.005) for overall study period. Conclusion The placement of TCVC in children with neutropenia undergoing anticancer therapy for acute leukemia is safe and not associated with an elevated rate of CLABSI.


INTRODUCTION
Acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) are the two most common types of acute childhood leukemias accounting for approximately 30% of all pediatric malignancies (1). The intensive treatment of both ALL and AML requires long-term central venous access, e.g., a tunneled central venous catheter (TCVC). This facilitates the administration of chemotherapeutics, antibiotics, fluids, blood products, and parenteral nutrients. However, TCVCs are often associated with infective events, including local catheter exit-site infections (CESI) and central line-associated bloodstream infections (CLABSI). Both may delay or interrupt anticancer therapy, resulting in substantial morbidity and mortality (2)(3)(4). Furthermore, CLABSI were found to be the most costly healthcare-associated infections (5). As pediatric patients with hematooncological diseases often present with neutropenia (absolute neutrophil count [ANC] <0.5 × 10 9 /L or <1.0 × 10 9 / L) and are therefore a priori at a higher risk of infection, some studies have attempted to identify the effect of neutrophil levels at the time of catheter insertion on development of CESI and CLABSI. Several studies in pediatric patients with cancer have found an association between neutropenia at the time of catheter insertion and increased incidence of CESI/CLABSI, proposing low perioperative ANC as a potential risk factor for both (6)(7)(8). Consequently, some centers recommend TCVC placement only with ANC higher than 0.5 × 10 9 /L or even 1.0 × 10 9 /L. However, the literature is inconsistent regarding the association of perioperative neutropenia and increased incidence of CESI/ CLABSI, with other studies showing no significant differences of CESI/CLABSI rates in neutropenic compared to nonneutropenic children at the time of catheter placement, indicating the safety of TCVC implantation also in the setting of neutropenia (9)(10)(11). Finally, most of the previous studies investigate the association between perioperative ANC and TCVC-associated infective complications in heterogeneous patient populations, which differ at baseline risk for CLABSI (12). Therefore, we aimed to investigate the impact of perioperative neutropenia on development of CESI and CLABSI in a large and well-defined cohort of highly vulnerable pediatric ALL and AML patients undergoing placement of TCVC.

Preoperative Evaluation, Device Characteristics, and Placement Technique
Preoperative workup generally consisted of laboratory studies including complete blood count and coagulation tests. We did not consider ANC less than 0.5 × 10 9 /L as a contraindication for TCVC placement. A single-shot prophylactic perioperative antibiotic (e.g., cefuroxime) was administered to all patients irrespective of their ANC status. The TCVC placement took place in the operation theater under general anesthesia. All procedures were performed or supervised by surgical attendings. We did not use any in-line filters or anti-infective lock prophylaxis at the time of TCVC insertion. We used anatomical landmarks and the adapted Seldinger technique for catheter cannulation (13). All catheters were tunneled subcutaneously. The tissue ingrowth cuff was positioned at the distal end of the subcutaneous tunnel. In almost all cases, we used the right internal jugular vein for cannulation. The size of the catheter was determined by the age and weight of the patient. Correct positioning of the distal catheter tip between the superior vena cava and right atrium junction was verified by fluoroscopy. All operative procedures and applied techniques remained constant during the overall study period. Postoperative TCVC maintenance procedures were performed according to the standard operating procedure manual of our institution.

Definitions
A CLABSI was defined according to the Centers for Disease Control and Prevention (LCBI 1 criteria, [laboratory confirmed bloodstream infection]) and the National Healthcare Safety Network criteria as a laboratory-confirmed bacteremia or fungemia originating from one or both lumens of the implanted device developing at least 48 h after TCVC placement and not related to an infection at another site (14,15). Therefore, children who developed CLABSI within 48 h after TCVC placement were not considered as event of interest for the primary endpoint of the study. A recurrent CLABSI was defined as a catheter-associated infection after a time frame of 14 days following prior CLABSI. Consequently, positive blood cultures originating from catheter lumens occurring within the first 2 weeks after previous CLABSI were not considered as new CLABSI episodes. Neutropenia was defined as ANC < 0.5 × 10 9 cells/L.

Endpoints
The primary endpoint of our study was the effect of perioperativeANC (ANC < 0.5 × 10 9 cells/L vs. ANC > 0.5 × 10 9 cells/L) on development of (first) CLABSI (censored at premature catheter removal, death or last follow-up) within the first 30 days after TCVC placement. Analysis for differences in the primary endpoint was performed using the univariable (unadjusted) and multivariable (adjusted for the following co-variables: patient´s age at time of TCVC implantation, underlying malignancy, sex, purpose of catheter use [primary therapy or therapy for relapsed leukemia], type of TCVC, number of lumens, and size of catheter tube) survival models. Secondary endpoints included CLABSI incidence during overall study period, CESI incidence within 30 days period after TCVC implantation, and description of microorganisms detected during first CLABSI stratified by children's neutropenic status peri-operatively and at the onset of the infection.

Statistics
Categorical variables (e.g., sex and catheter type) were described using frequencies and percentages. Continuous variables (e.g., age and duration of catheter use) were reported as median and corresponding interquartile range (IQR). Fisher's exact test was used for analysis of categorical variables and Mann-Whitney U test was used for continuous variables. The total number of catheter days (CD) per case was calculated as the total number of days from TCVC insertion to last follow-up (e.g., elective catheter removal, premature catheter explantation, and death). The complication rate (CR) per 1,000 CD was calculated as 1,000 times the number of complications divided by the total number of CD (16,17). The risk of developing CLABSI during short-term (within first 30 days after TCVC placement) and overall followup period was estimated using the traditional Kaplan-Meier method (univariable/unadjusted analysis, crude HR) and Cox regression survival analysis (multivariable/adjusted analysis). Due to the presence of competing risks (catheter removal [elective vs. non-elective] and death), the risk of developing CLABSI during the short-term period and during the overall follow-up period was additionally estimated using competingrisks regression model (Fine & Gray [cumulative incidence function [CIF]) (18). The endpoint of interest in survival analysis was the time to first CLABSI. All consecutive CLABSI episodes were ignored in the analysis. All statistical tests were two-tailed and the tests were considered significant with p < 0.05.

Ethics
The study protocol was approved by the Institutional Review Board of the University Hospital of Freiburg on December 8, 2020 (IRB, protocol number 20-1257). Written or verbal informed consent was not required.

Organisms Isolated at First CLABSI in Neutropenic and Non-Neutropenic Patients
Overall, 28 different bacterial/fungal strains were isolated from 127 blood cultures including 140 isolates (Supplemental Table S1).

DISCUSSION
In this observational study, we retrospectively evaluated the association between perioperative neutropenia and other  potential risk factors on development of CLABSI among a large and homogeneous cohort of pediatric patients with ALL and AML undergoing placement of TCVC. We demonstrated that ANC < 0.5 × 10 9 /L at the time of catheter placement was not associated with increased CLABSI incidence in either shorttermor the overall study follow-up periods. We also showed that in the overall cohort of neutropenic and non-neutropenic at implantation patients, female sex and TCVC use in children with relapsed leukemia contributed to the development of CLABSI during the first 30 days after TCVC insertion. Investigating risk factors contributing to CLABSI during the overall study period, the use of double-lumen catheters and TCVC during leukemia relapse was found to increase the hazard for CLABSI. Our finding that periprocedural neutropenia did not influence the risk for CLABSI is consistent with numerous previous reports (10,(19)(20)(21)(22). Conversely, other studies evaluating the impact of perioperative neutropenia on catheter-associated infections have yielded a higher rate of CLABSI episodes in neutropenic children, thus proposing exclusion of these patients from TCVC placement and insertion of an alternative central venous access device, such as peripherally inserted central venous catheter (PICC) or percutaneous CVCs as a bridging strategy (7,8,23). However, comparison of our study findings to previously published reports is problematic because of a large heterogeneity within included study populations (e.g., different hematooncological diseases, lymphoma, and solid tumors).
In our study, we found early CLABSI events in 7.4% (37/498) and overall CLABSI events in 33.1% (165/498) of TCVC. This is very similar to a previous study investigating the incidence rate of late CLABSIs in children with cancer, reporting an overall rate of 29% for infectious episodes (24).
The CR for CLABSI during the overall study period was 1.65 per 1,000 CD [99,681 days]. This is in line with previous studies reporting pooled CR rates for different hematological malignancies and solid tumors ranging between 1.6 and 3.1 per 1,000 CD (25)(26)(27)(28)(29) and even below the published rate of 2.3 of the American National Healthcare Safety Network for the years 2006-2008 (30). In particular, children with acute leukemias were shown to have a 3.7-to 14.9-fold increased risk for catheter infections compared to patients with solid malignancies (12,25). Thus, in our study, we observed a remarkably low CLABSI rate in children with acute leukemias.
We additionally studied other factors potentially contributing to early CLABSI development and found that female sex (HR 2.640, 95% CI 1.328-5.250, p = 0.006) and the use of TCVC for therapy of relapsed leukemia (HR 4.347, 95% CI 2.283-8.264, p < 0.0001) increase the hazard for TCVCassociated infections. The unexpected finding that girls showed a higher risk for early CLABSI was not previously observed by other authors (31). The finding that children with relapsed malignancies face an elevated risk for CLABSI is consistent with the results of a prospective multicenter study involving pediatric cancer patients (32). This higher risk may be related to the treatment intensity in patients who are vulnerable to the side effects of antineoplastic therapy due to the accumulating effect of organ toxicities after receiving first-line therapy (33,34).
We also found that the use of dual-lumen TCVCs was associated with an increased probability for CLABSI development during the overall study period (HR 2.607, 95% CI 1.392-4.880, p = 0.003). These findings are comparable with the HR ranging from 2.4 (95% CI, 1.6-3.5) to 3.13 (95% CI 2.11-4.65) for double-lumen catheter compared to single-lumen devices found in the literature (35,36).
In contrast to several other studies reporting a higher incidence of TCVC-associated infections in younger patients (22,37), in children with AML (38) or in patients with tunneled externalized catheters compared to ports (27,39), we found no influence of patient´s age, underlying type of leukemia, or device type on early or overall CLABSI episodes in our study when analyzing using multivariable regression models. One possible explanation for the lack of age dependency on CLABSI risk observed in our study is that, in the regression models, we used age as continuous data and did not perform any categorization or dichotomizing of the patients. This would be arbitrary because the cutoffs are not defined, leading to loss of information or increasing bias effect (40,41). Regarding the lack of association between type of underlying disease and CLABSI rates, it is important to emphasize that intensity and duration of anticancer therapy vary significantly between different biologically distinct subtypes of ALL and AML (42)(43)(44), thus potentially contributing to the variation of infectious complications within the subgroups.
In the present study, the predominant organisms detected in neutropenic at CLABSI patients were Viridans group streptococci (VGS), which represent a part of the normal flora of the gastrointestinal tract. In neutropenic conditions, these organisms are able to translocate through compromised mucosal barriers of the gastrointestinal tract into the bloodstream. Our results of VGS in the setting of neutropenia at CLABSI onset are consistent with earlier studies demonstrating that VGS are the most common isolates in neutropenic adult and pediatric oncological study populations (45,46).
We are cognizant of the several limitations of this study. First, the retrospective nature and limitation to a single center involving institutional-specific treatment preferences and practices could have an effect on the generalizability of the study results. Second, although we used well-defined criteria of CLABSI (LCBI 1 criteria), the clinical characteristics of the patients (LCBI 2 criteria) were not considered, as no sign or symptom such as fever are needed to meet LCBI 1 criteria. Moreover, important clinical information such as severity of CLABSI is lacking. Nevertheless, we believe that in a setting of retrospective data collection, CLABSI definition according to LCBI 1 criteria is more suitable and robust. From an epidemiological point of view, because of the application of LCBI 1 criteria (eligible central line is a central line that has been in place for more than two consecutive calendar days), none of the study participants was in effect at risk for the study outcome during the first 48 h after TCVC placement (immortal person-time), which could result in an underestimated cumulative rate of the first CLABSI reported by the current study. Third, due to variations in frequency of blood sample collection during the further treatment course, we were not able to sufficiently track the ANC values from the time point of TCVC insertion until CLABSI onset. Consequently, it was not possible to adjust the CLABSI incidence to duration of neutropenia. However, we adjusted CLABSI incidence to 1,000 utilization days for TCVC. Finally, given that the purpose of this study was to determine the risk factors for CLABSI, we did not consider the treatment options for CLABSI. Notwithstanding, we believe that our study provides robust evidence on risk factors contributing to CLABSI in hematooncological patients with ALL and AML thanks to application of a strict CLABSI definition, inclusion of a large and homogeneous patient series, and conduction of a univariable and multivariable analysis minimizing interactions between clinical and device characteristics.
In conclusion, we showed that a perioperative ANC < 0.5 × 10 9 /L does not impact the incidence of early and overall CLABSI episodes in pediatric patients with acute leukemia. We also showed that the use of double-lumen catheters and the use of TCVC for treatment of relapsed leukemia were independent risk factors for CLABSI development. Efforts should be made to reduce CLABSI incidence through more intensive surveillance by implementing catheter care programs.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The study protocol was approved by the Institutional Review Board of the University Hospital of Freiburg on December 8, 2020 (IRB, protocol number 20-1257). Written or verbal informed consent was not required. Written informed consent for participation was not provided by the participants' legal guardians/next of kin because written or verbal informed consent was not required.

AUTHOR CONTRIBUTIONS
JS and IM conceptualized and designed the study. JS collected data. IM carried out the statistical analyses and drafted the initial manuscript. JS and IM reviewed and revised the manuscript. Both authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.