1 INTRODUCTION

Patients with hematologic malignancies frequently require nontunneled central venous catheters (CVC) for the administration of chemotherapy, fluid infusions, blood transfusions, antimicrobial therapy, parenteral nutrition, and blood sampling over several weeks. However, CVC carry a major risk of catheter-related infections, leading to prolonged hospitalization and increased morbidity and mortality.1-3 Particularly in the neutropenic phase, the risk for local infection at the insertion site and central line-associated blood stream infections (CLABSI) and catheter-related blood stream infections (CRBSI) is high.2, 4 The selection of the insertion site, either the internal jugular vein (IJV) or the subclavian vein (SCV), may contribute to the emergence of catheter-related complications. Analyses of CVC-related complications by insertion site are published for patients in the intensive care unit (ICU).5-10 Recently, we and others presented data on patients undergoing allogeneic hematopoietic cell transplantation.4, 11 However, there are scarce data comparing both insertion sites concerning catheter-related complications in hematologic patients with severe and prolonged neutropenia after intensive chemotherapy. Thus, the aim of this study was to systematically analyze CVC-related complications at both insertion sites for patients receiving induction or consolidation chemotherapy for acute leukemia or high-dose-chemotherapy followed by autologous hematopoietic cell transplantation (HCT).

2 MATERIAL AND METHODS 2.1 Patients and data collection

In this retrospective study, all consecutive adult patients were included obtaining a CVC either in course of high-dose-chemotherapy for autologous HCT or induction or consolidation chemotherapy for acute leukemia at our institution between January 2011 and June 2013. All procedures were performed in accordance with the ethical standards of the responsible committee on human experimentation and with the Helsinki Declaration of 1964, as revised in 2013.

2.2 CVC insertion and nursing care

After patient's informed consent has been obtained, a 3-lumen 7-French standard nontunneled CVC with a length of 30 cm (Arrow by Teleflex, Morrisville, NC) was placed either via the IJV or the SCV. CVC were nonimpregnated and made of polyurethane. The catheter was inserted by one of five different experienced physicians who had been trained in placing CVC at both insertion sites according to the center's applicable standard operating procedures. CVC were placed after local disinfection using Softasept™ N (ethanol 74%/2-propanol 10%, B. Braun, Melsungen, Germany) under full barrier precautions according to the institutional guideline. The insertion site was chosen at the responsible physician's own discretion. Before placing the CVC in the IJV, demonstration of the IJV on both sides was performed by ultrasound. CVC were covered by chlorhexidine gluconate impregnated (CGI) dressings, either as transparent polyurethane (Tegaderm™ CHG, 3M Medica, Neuss, Germany) or as sponge dressing (Biopatch™, Ethicon, Somerville, NJ). The transparent dressings were routinely replaced after 7 days. The insertion site was inspected for signs of local inflammation (LI) such as erythema, swelling or pain on a daily routine by the responsible nurse. In case of the appearance of LI signs, the finding was reviewed by the responsible physician. The degree of the first occurrence of a LI as well as the time point was documented in the patient's chart. Henceforth, the insertion site was reviewed daily with respect to the decision to remove the CVC. Reasons for premature CVC removal were (i) suspected or proven CVC-related BSI, (ii) progression of LI at the insertion site, or (iii) any other CVC-related severe adverse event. All CVC-related procedures were performed according to the center's standard operating procedures.

2.3 Antimicrobial prophylaxis and routine surveillance

Cotrimoxazole was administered as Pneumocystis jiroveci prophylaxis three times per week. Patients neither received a fluoroquinolone prophylaxis nor any other additional systemic antibiotic prophylaxis, routinely. According to the institutional guidelines, all patients who underwent induction chemotherapy for acute leukemia obtained an antifungal prophylaxis either with posaconazole or micafungin. Blood cultures were obtained from the CVC and concomitantly from a peripheral site whenever an infection was clinically suspected, especially in the case of fever (temperature ≥ 38.0°C). Low-dose nadroparin was used as thrombosis prophylaxis in all patients with a platelet count >30.000/μl.

2.4 Statistical analysis

Primary study end points were (i) time to CRBSI, (ii) time to CLABSI and (iii) time to local inflammation at the insertion site (LI). Further end points were (a) catheter lumen obstruction, (b) deep-vein thrombosis, (c) pneumothorax, and (d) catheter-related death. Statistical analysis was performed by using Graph Pad Prism™ (La Jolla, CA) and SPSS Statistics™ (Armonk, NY), respectively. Time to event estimation was calculated by using the Kaplan Meier method. Curves were compared by Gehan-Breslow-Wilcoxon test. Since early events within the first 14 days of observation are crucial for the significance of catheter-related complications, the Gehan-Breslow-Wilcoxon test was chosen instead of the log rank test which is more sensitive to late differences.12 Cox analysis was used to identify risk factors for the occurrence of CVC complications. Multivariable risk factors were identified using stepwise Cox analysis with a variable entry criterion of p ≤ .10 and a variable retention criterion of p ≤ .05. Cox results are presented as hazard ratio (HR) with 95% CI and corresponding p value.

2.5 Definitions

The event “LI” was defined as visible erythema or swelling at the insertion site of the catheter as described by Hentrich et al.13

“Time to LI” was defined as the interval from CVC insertion until the occurrence of a LI.

“Blood stream infection” (BSI) was characterized by the detection of a microbiological pathogen in a blood culture. By definition, a BSI by coagulase-negative Staphylococci needed the detection of these bacteria in two independent blood cultures.

A “CLABSI” was defined as a blood stream infection in a patient with a CVC in place at least 48 h before development of the BSI and no association to any infection at another site.14

The diagnosis of “CRBSI” required either the isolation of the same microorganism in the blood culture and the catheter-tip or a differential time to positivity. The differential time to positivity was characterized by an at least 2 h earlier positive central line blood culture compared to the simultaneously drawn peripheral blood culture.14, 15

“Time to CLABSI” and “Time to CRBSI,” both were defined as the interval from CVC insertion until the occurrence of CLABSI or CRBSI.

“Neutropenic days” were defined as days with neutropenia grade IV according WHO during hospitalization.

“Catheter lumen obstruction” was diagnosed if at least one lumen of the CVC was obstructed.

Deep-vein thrombosis in the IJV or SCV was diagnosed either by sonography or radiography.

Pneumothorax was diagnosed by chest X-ray, which was performed routinely after the placement of a CVC.

“CVC-related death” was defined as death caused by a catheter-related complication such as CRBSI, pneumothorax, or deep-vein thrombosis.

3 RESULTS 3.1 Patient characteristics

In total, 87 adult patients (male n = 53, female n = 34) receiving either induction or consolidation chemotherapy for acute leukemia or high-dose chemotherapy followed by autologous HCT between January 2011 and June 2013 were included in this analysis. All these were provided with a CVC, in total 153 catheters. The number of CVC is higher than the number of patients, since several patients received more than one CVC either during the same inpatient stay or due to a subsequent consolidation chemotherapy. Out of 153 CVC, 94 (61%) were located in the IJV and 59 CVC (39%) were placed in the SCV, respectively. Median age at time of CVC-placement was 61 years (range 19–79). A total of 120 CVC were placed in 57 patients for the need of induction (n = 105) or consolidation (n = 15) chemotherapy for acute leukemia (AML, n = 114; ALL, n = 6). Further 33 CVC were placed in 30 patients to carry out high-dose chemotherapy followed by autologous HCT (plasma cell neoplasm, n = 18; lymphoma n = 15). Table 1 shows patient characteristics and corresponding CVC data. Age, sex, underlying disease, and type of therapy were well balanced between the IJV and SCV-group.

TABLE 1. Patient characteristics by type of CVC Characteristic Patients (n = 87) CVC (n = 153) p value IJV (n = 94) SCV (n = 59) Age, yr, median (range) 61 (19–79) 63 (19–79) 59 (19–76) .165 Males/females, n 53/34 56/38 35/24 .976 Underlying disease, n Acute myeloid leukemia 53 70 44 .988 (AML vs. non-AML) Acute lymphoblastic leukemia 4 1 5 Lymphoma 14 14 1 Plasma cell neoplasm 16 9 9 Therapy, n Induction chemotherapy 57 63 42 .274 (induction or consolidation vs. autologous HCT) Consolidation chemotherapy 11a 8 7 Autologous HCT 30 23 10 3.2 CVC complications

The median duration of catheterization per CVC was 17 days (range 3–37) in the IJV-group and 12 days (range 6–36 days) in the SCV-group (p = .047), respectively. Focusing on the primary applied CVC, only, the median duration of catheterization per CVC was 18 days (range 8–34) compared to 16 days (range 6–36) for the IJV- and SCV-group, respectively (p = .234).

LI occurred in 69% (105/153) of CVC, thereof in 56% (53/94) of IJV- and 88% (52/59) of SCV-CVCs (p < .0001, Table 2), respectively. Analyzing primary CVC, only, data for both groups were similar (63% vs. 88%, p = .005, data not shown). The median time to LI was significantly shorter for SCV- compared to IJV-CVC (9 vs. 14 days, p < .0001; HR 2.079; Figure 1); correspondingly, 9 days vs. 13 days for primary applied CVC (p = .003, data not shown).

TABLE 2. Pairwise comparison of CVC outcome parameters Outcome All CVC (n = 153) IJV (n = 94) SCV (n = 59) p value CVC days per CVC, median (range) 15 (3–37) 17 (3–37) 12 (3–36) .047 CVC days per primary CVC, median (range) 17 (6–36) 18 (8–34) 16 (6–36) .234 Symptomatic deep-vein thrombosis, n 0 0 0 Pneumothorax, n 1 1 0 Catheter lumen obstruction, n (%) 27 (18) 14 (15) 13 (22) .262 Local inflammation, n (%) 105 (69) 53 (56) 52 (88) <.0001 CLABSI, n (%) 29 (19) 24 (26) 5 (8) .009 Identified pathogens in CLABSI, n 35 26 9 Gram positive bacteria, n 24 19 5 Gram negative bacteria, n 11 7 4 CLABSI/1000 CVC days, n 11.9 15.2 5.8 CLABSI/1000 neutropenic CVC days, n 18.3 23.8 8.7 CRBSI, n (%) 10 (7) 9 (10) 1 (2) .055 Identified pathogens in CRBSI, n 11 10 1 Gram positive bacteria, n 9 8 1 Gram negative bacteria, n 2 2 0 CRBSI/1000 CVC days, n 4.1 5.7 1.2 CRBSI/1000 neutropenic CVC days, n 6.3 8.9 1.7 Catheter-related deaths, n 0 0 0

Cumulative incidence of LI (A), CLABSI (B), and CRBSI (C) according to CVC insertion site

CLABSI occurred in 29 of 153 CVC (19%), in 24 cases (26%) with IJV-CVC and in 5 cases (8%) with SCV-CVC in place (p = .009). The rate of CLABSI referred to 1000 neutropenic CVC days was higher for the IJV-group compared to the SCV-group (23.8 vs. 8.7, Table 2). Additionally, CLABSI occurred significantly earlier in the IJV- compared to the SCV-group (p = .03; HR 2.667, Figure 1). The incidence rate of CLABSI at day 15 accounted for 27% in the IJV-group compared to 10% in the SCV-group.

CRBSI were diagnosed in 10 of 153 CVC (6.5%), that is, nine cases (10%) with IJV-CVC and a single case (2%) with SCV-CVC (p = .055), respectively. The rate of CRBSI referred to 1000 neutropenic CVC days was substantially higher in the IJV-group compared to the SCV-group (8.9 vs. 1.7, Table 2). Moreover, CRBSI were detected significantly earlier in the IJV-group in comparison to the SCV-group (p = .04; HR 5.405, Figure 1). The incidence rate of CRBSI at day 15 were 10% and 0% in the IJV- and SCV-group, respectively.

In CRBSI defining episodes, altogether 11 pathogens were detected in blood cultures, more often gram positive (n = 9) than gram negative (n = 2) bacteria. Out of these 11 pathogens 10 were detected in patients with IJV-CVC, thereof as gram positive bacteria (n = 8) Staphylococcus epidermidis (n = 5), Streptococcus mitis/oralis (n = 1), Enterococcus faecium (n = 2), and as gram negative pathogens Escherichia coli (n = 2). In patients with CVC placed in the SCV, only one blood culture was positive with the detection of Enterococcus faecium.

Symptomatic deep-vein thrombosis occurred in none of the patients, whereas iatrogenic pneumothorax was diagnosed once after placement of a CVC in the IJV. Catheter lumen obstruction was observed in 27 of 153 CVC (18%), in 15% (14/94) of IJV- and in 22% (13/59) of SCV-CVC, respectively. No patient died due to a catheter-related infection or complication.

3.3 Risk factors for the occurrence of CVC complications

To identify risk factors for the occurrence of catheter-related complications, the impact of the variables (a) age, (b) gender, (c) insertion site, (d) the consecutive number of the CVC, (e) kind of therapy, (f) neutropenia during hospitalization, and (g) neutropenia at time of CVC insertion on the endpoints (i) time to LI, (ii) time to CLABSI, (iii) time to CRBSI were analyzed (Table 3).

TABLE 3. Univariate and multivariate analysis of risk factors for the occurrence of CVC complications Characteristic Time to local inflammation (LI) Time to CLABSI Time to CRBSI p value p value p value HR (95% CI) Days, median HR (95% CI) %CLABSI at day 15a HR (95% CI) %CRBSI at day 15a Sex .694 .029 .154 Male 0.880 (0.60–1.30) 12 2.257 (0.96–5.28) 29.23 1.561 (0.40–6.05) 9.54 Female Reference 10 Reference 9.59 Reference 1.69 Sexb .058b Male 2.279 (0.97–5.34)b Female Reference Age .552 .275 .277 ≤60 years Reference 11 Reference 16.88 Reference 2.76 >60 years 0.846 (0.58–1.25) 12 1.437 (0.69–3.01) 26.03 2.263 (0.59–8.76) 9.96 Insertion site <.0001 .025 .036 Internal jugular vein Reference 14 2.667 (1.01–7.14) 27.04 5.405 (0.68–50) 9.67 Subclavian vein 2.079 (1.41–3.07) 9 Reference 9.99 Reference 0 Insertion siteb .001b .044b Internal jugular vein Reference 2.695 (1.02–7.14)b Subclavian vein 1.969 (1.33–2.91)b Reference Therapy .025 .822 .417 Induction/consolidation therapy 1.962 (1.15–3.36) 11 0.751 (0.33–1.71) 18.12 2.242 (0.28–17.97) 7.13 Autologous HCT Reference 16 Reference 31.13 Reference 4.17 Therapyb .035b Induction/consolidation therapy 1.790 (1.04–3.07)b Autologous HCT Reference Consecutive number of the CVC .664 .354 .633 1st CVC 0.890 (0.58–1.37) 12 1.690 (0.64–4.44) 23.20 0.772 (0.20–3.02) 5.29 >1st CVC Reference 11 Reference 16.79 Reference 11.12 Neutropenic days during hospitalization .036 .762 .421 ≤25 days Reference 14 Reference 23.0 Reference 5.25 >25 days 1.575 (1.07–2.33) 11 0.899 (0.43–1.87) 19.99 1.264 (0.35–4.59) 7.72 Neutropenic days during hospitalizationb .284b ≤25 days Reference >25 days 1.274 (0.82–1.99)b Neutropenia at time of CVC insertion .343 .644 .727 Yes 1.245 (0.84–1.84) 11 1.043 (0.49–2.22) 21.73 1.012 (0.28–3.57) 8.61 No Reference 12 Reference 21.18 Reference 5.20

By univariate analysis, the insertion of the CVC in the SCV (p < .0001, HR: 2.1), induction/consolidation chemotherapy (p = .025, HR: 2.0) and long-lasting neutropenia >25 days during hospitalization (p = .036, HR: 1.6) were identified as risk factors for a shorter time to LI. For the endpoint time to CLABSI, the IJV insertion site (p = .025, HR: 2.7) and male gender (p = .029, HR: 2.3) were identified as risk factors. For the endpoint time to CRBSI, only the insertion in the IJV (p = .036, HR: 5.4) was identified as the potential risk factor.

After multivariate analysis, the insertion of the CVC in the SCV (p = .001, HR: 2.0) and induction/consolidation chemotherapy compared to autologous HCT (p = .035, HR: 1.8) retained statistical significance as risk factors for the endpoint “time to LI.” Exclusively, the IJV insertion site was confirmed as risk factor for the endpoint time to CLABSI (p = .044, HR: 2.7). Since the insertion site was the only significant risk factor for time to CRBSI in the univariate analysis, a multivariate analysis was not performed.

4 DISCUSSION

In this single-center analysis on catheter-related complications, it was demonstrated that the insertion of CVC in the IJV is the main risk factor for the occurrence of CRBSIs in patients with induction or consolidation chemotherapy for acute leukemia or autologous HCT for lymphoma or plasma cell neoplasms. To our knowledge this is the first study to identify the insertion site of the CVC as a risk factor for CRBSI and CLABSI in patients with hematologic malignancies experiencing neutropenia IV°. In a hallmark study on CVC complications Parienti and colleagues demonstrated in 2015 for ICU patients that SCV-CVC were associated with a lower risk for blood stream infections in comparison to the insertion in the IJV or femoral vein.