Clinical Characteristics and Outcomes of S. Aureus Bacteremia in Patients Receiving Total Parenteral Nutrition

Background: Patients on total parenteral nutrition (TPN) are at risk of developing central line-associated infections. Specifically, Staphylococcus aureus bacteremia (SAB) is feared for its high complication rates. This prospective cohort study compares characteristics, clinical course and outcome of SAB in patients with and without TPN support. Methods: Clinical and microbiological data from all patients with positive blood cultures for S. aureus from two facilities, including our referral center for TPN support, were retrieved (period 2013–2020). Primary outcome was overall mortality, and included survival analysis using a multivariate Cox regression model. Secondary outcomes comprised a comparison of clinical characteristics and outcomes between both patient groups and analysis of factors associated with complicated outcome (e.g., endocarditis, deep-seated foci, relapse and death) in patients on TPN specifically. Results: A total of 620 SAB cases were analyzed, of which 53 cases received TPN at the moment the blood culture was taken. Patients in the TPN group were more frequently female, younger and had less comorbidity (p < 0.001). In-hospital death and overall mortality were significantly lower in TPN patients (4% vs. 18%, p = 0.004 and 10% vs. 34%, p < 0.001, respectively). Positive follow-up blood cultures, delayed onset of therapy and previous catheter problems were associated with a higher incidence of complicated SAB outcome in patients on TPN. Conclusion: Our data show that patients on TPN have a milder course of SAB with lower mortality rates compared to non-TPN SAB patients.

S. aureus can cause a broad range of community-acquired, hospital-acquired and/or healthcare-associated infections [10]. Among these, S. aureus bacteremia (SAB) presents as a life-threatening infection that is associated with an all-cause mortality ranging from 20 to 30% in the general population [11][12][13][14]. The most common co-existent factors for SAB are the presence

Data Collection
We assessed the medical records of all patients and gathered electronic data on patient demographic characteristics. The data were retrieved from the infectious disease (ID) consultation documentation and results of diagnostic studies (including laboratory, microbiologic and imaging data). We determined onset of bacteremia, presence of foreign body material and intravascular catheters, clinical parameters at SAB onset, diagnostic tests, antimicrobial therapy and survival outcomes during and after hospital stay and at 3 and 6 months. Preexisting underlying disease and comorbidity were calculated according to the Charlson comorbidity index (CCI) [25] adjusted for age. Data were entered in a secure database (Castor EDC, Amsterdam, The Netherlands) and processed anonymously for further analysis. Follow up time was 6 months, since the majority of patients (>80%) had regular follow-up appointments at the infectious diseases outpatient clinic during this period.

Ethical Consideration
This study was, according to Dutch law, cleared from the requirement of approval by an ethics committee, because of the observational character of this study and the anonymous processing of data. The regional institutional review board approved this study (2015-2257) [26].

Management of Central Line Sepsis and S. aureus Bacteremia in Patients with TPN Support
Prior to the start of parenteral nutrition administration in the home situation (HPN), patients are trained in aseptic handling of their CVC by specialized nurses during a training period of 1-2 weeks in our referral center. This training is implemented according to a standardized protocol and is in line with recent European Society for Clinical Nutrition and Metabolism (ESPEN) guidelines [27]. Extra emphasis is put on the awareness of potential central line sepsis by instructing the patients to Nutrients 2020, 12, 3131 3 of 14 monitor their body temperature regularly and whenever signs of infection are present. In addition, they are strongly advised to contact our referral center directly in case of fever (>37.5 • C) or other signs of infection (e.g., inflammation at the insertion site of the CVC) arise. More details about the CVC care and central line sepsis management protocol can be found elsewhere [28].
All patients with S. aureus bacteremia (including patients with TPN) were managed according to the advice of the national antimicrobial stewardship program, which included the performance of echocardiography and consultation by an ID specialist. Patients were treated according to the national guideline for SAB, which is in accordance with the Infectious Diseases Society of America guideline [29]. Yet, our institutional guideline recommends antimicrobial therapy for 2 weeks instead of 4-6 weeks for patients with risk factors for complicated outcome (e.g., presence of CVC), but without endocarditis or signs of metastatic infection on 18-fluor-FDG positron emission tomography/computed tomography ( 18 F-FDG PET/CT); these patients were considered as uncomplicated SAB cases [26].

Outcomes and Definitions
The primary outcome was overall mortality in the TPN-and non-TPN groups. Mortality was calculated from the date of first positive blood culture till the date of death or end of follow-up period (6 months). Attributable mortality (death due to SAB) was defined as clinical or microbiological evidence of infection with S. aureus at time of death and no other explanation for cause of death. Patients were considered to be cured in the case of the resolution of signs of infection, and no positive follow-up blood cultures were present after the discontinuation of antibiotic treatment. Relapse of SAB was defined as another episode of SAB within 3 months after the end of antibiotic treatment.
SAB outcome was considered complicated when patients had any of the following: infective endocarditis; metastatic infection; non-retainable infected foreign body; relapse of infection [14]. Metastatic infection was identified using 18 F-FDG-PET/CT scanning or with evidence from other relevant imaging studies. Infective endocarditis was diagnosed according to the modified Dukes criteria [30]. Predictors identified by the literature for complicated SAB were positive follow-up blood cultures, community acquisition, fever ≥72 h and skin abnormalities suggesting active systemic infection [14]. Patients received antibiotic treatment according to our national SAB guideline [31].
All used definitions were according to commonly used guidelines and are listed in the supplementary appendix (Appendix A, Table A1) [22,32]. The mode of acquisition was classified into community acquired, healthcare associated or hospital acquired according to Bishara et al. [33].
All patients who had a CVC that was present for at least 48 h and received parenteral nutrition at the time of positive blood culture were included in the TPN group and, among these patients, most (39/41) consisted of patients who received long-term (home) parenteral nutrition (HPN).

Statistical Methods
For the descriptive statistics, continuous variables were compared by use of the independent samples t-test or Mann-Whitney test when non-parametric testing was indicated. Categoric variables were compared by use of the chi-squared test or Fisher's exact test when values were small (<5). Differences were considered to be statistically significant at a two-sided p-value < 0.05. Sensitivity analysis was done for three specific variables to examine whether these influenced our data and outcomes, namely (1) patients treated in other hospitals with a significant amount of missing data, (2) cases with missing data on primary outcome and (3) short-term parenteral nutrition (e.g., <1 month). The primary endpoint of our analyses was a comparison of survival time between TPN and non-TPN groups. Survival was analyzed using Kaplan-Meier plotters with a log-rank test and multivariate Cox regression analysis. The patient-dependent variables entered in the multivariate analysis were those with at least 90% non-missing observations and a univariate p-value of <0.1. Because of the relatively low number of events (death) in our cohort, the inclusion of variables was limited to the ones most significant by univariate analysis. Furthermore, univariate logistic regression analysis was done to

Comparison of Patient Characteristics
During the study period, a total of 646 cases of S. aureus bacteremia were screened for inclusion ( Figure 1). Eventually, 620 cases in 604 patients could be included in the analysis. Fifty-three of these cases occurred in 41 patients receiving TPN. Patients receiving TPN were more frequently female (68.3% vs. 37.1%, p < 0.001), younger (mean age 53.4 vs. 63.2, p < = 0.001), had a lower Charlson comorbidity index score (mean 1.90 vs. 3.64, p < 0.001) and foreign body material (excluding CVC) was less frequently present (19.5% vs. 39.3%, p = 0.012). Risk factors for endocarditis and immune status did not differ between the two groups ( Table 1).

Comparison of Patient Characteristics
During the study period, a total of 646 cases of S. aureus bacteremia were screened for inclusion ( Figure 1). Eventually, 620 cases in 604 patients could be included in the analysis. Fifty-three of these cases occurred in 41 patients receiving TPN. Patients receiving TPN were more frequently female (68.3% vs. 37.1%, p < 0.001), younger (mean age 53.4 vs. 63.2, p < = 0.001), had a lower Charlson comorbidity index score (mean 1.90 vs. 3.64, p < 0.001) and foreign body material (excluding CVC) was less frequently present (19.5% vs. 39.3%, p = 0.012). Risk factors for endocarditis and immune status did not differ between the two groups ( Table 1).

Clinical Characteristics of S. aureus Bacteremia
A CVC was most likely the portal of entry in 91% of the patients on TPN and the mode of acquisition was healthcare associated in 83% and hospital acquired in the remaining 17% (Table 2). In non-TPN patients, skin was the most common portal of entry (34%) and onset was almost as frequent for community acquired (36%) as hospital acquired (37%). MRSA rates did not differ between the two groups, with an overall rate of 2.1%. At the onset of the bacteremia, patients receiving TPN showed a significantly lower level of C-reactive protein with a mean of 74.6 mg/L versus 169 mg/L (p < 0.001). Percentage of intensive care admissions did not differ between both groups with an occurrence of 23% in TPN patients vs. 29% in non-TPN patients (p = 0.40). On presentation, 85% of all patients experienced fever, which lasted for more than 72 h in 32% of all patients. Among patients on TPN, the percentage of persistent fever was remarkably lower (18% vs. 33%, p = 0.04). Although deep-seated foci were found with 18 F-FDG PET-CT in more than half of the TPN cases, this was significantly less than in non-TPN patients (54% vs. 73%, p = 0.02), with pulmonary and endovascular foci being the most prevalent (28% and 19%, respectively). In total, there were 20 patients with metastatic infection(s). In 80% of the cases, these were diagnosed with 18 F-FDG PET/CT scanning. Half of the TPN patients had signs of inflammation at the CVC insertion site prior to the SAB episode; most frequently, this was redness and tenderness (both 36%), followed by induration (16%) (Figure 2). In 44% (11/25 cases), an exit-site culture was performed shortly before the SAB episode, which was positive for S. aureus in 82% of the cases. Nasal S. aureus carriage was tested in 30 TPN cases and was positive in the majority of the patients (70%). The CVC was removed according to protocol in all but one TPN case, with a mean of 1.8 (SD 1.7) days from positive blood culture to removal. Half of the TPN patients had signs of inflammation at the CVC insertion site prior to the SAB episode; most frequently, this was redness and tenderness (both 36%), followed by induration (16%) (Figure 2). In 44% (11/25 cases), an exit-site culture was performed shortly before the SAB episode, which was positive for S. aureus in 82% of the cases. Nasal S. aureus carriage was tested in 30 TPN cases and was positive in the majority of the patients (70%). The CVC was removed according to protocol in all but one TPN case, with a mean of 1.8 (SD 1.7) days from positive blood culture to removal.

Antibiotic Treatment
Significantly more patients on TPN reported an allergy to antibiotics (26.4% vs. 12.0%, p = 0.003) ( Table 1), and fewer TPN patients were switched to oral antibiotics during treatment (10% vs. 34%, p < 0.001) ( Table 2). TPN patients less frequently received rifampicin in addition to conventional therapy and fewer patients underwent surgical drainage. No other significant differences in provided treatment were seen.

Mortality and Multivariate Survival Analyses
Follow up data on mortality were available for 604 (of 620) SAB cases. Mortality was substantially lower in patients on TPN at all time periods (Figure 3 and Table 3). Only two (4%) in-hospital deaths occurred in patients on TPN compared to 101 (18%) in non-TPN patients (p = 0.004). Forty percent (2/5) of the deaths in TPN patients occurred after hospital discharge in comparison to 43% (75/176) Nutrients 2020, 12, 3131 7 of 14 in those not receiving TPN. To analyze if TPN was associated with an increased survival probability, a multivariate Cox regression model was designed, including the significant (p < 0.1) variables from univariate analysis (CCI score adjusted for age and risk factors for endocarditis) (Figure 4 and Table 4). Age was excluded from multivariate analysis since a strong correlation with CCI score was present (Pearson R = 0.65, p < 0.001). According to the Cox model, receiving TPN resulted in a 38% higher survival probability (adjusted HR 0.38, p = 0.03). Both CCI score and risk factors for endocarditis were found to be independent risk factors for mortality (HR 1.32, p < 0.001 resp. HR 1.74, p < 0.01).

Clinical Characteristics of Complicated SAB in Patients Receiving TPN
When comparing the clinical characteristics of patients receiving total parenteral nutrition with and without complicated SAB it appeared that healthcare-associated onset, higher CRP at time of positive blood culture, recent problems with the central venous catheter, treatment onset >24 h and positive follow-up blood cultures were more prevalent in cases with a complicated SAB (Table 5).

Clinical Characteristics of Complicated SAB in Patients Receiving TPN
When comparing the clinical characteristics of patients receiving total parenteral nutrition with and without complicated SAB it appeared that healthcare-associated onset, higher CRP at time of positive blood culture, recent problems with the central venous catheter, treatment onset >24 h and positive follow-up blood cultures were more prevalent in cases with a complicated SAB (Table 5).

Clinical Characteristics of Complicated SAB in Patients Receiving TPN
When comparing the clinical characteristics of patients receiving total parenteral nutrition with and without complicated SAB it appeared that healthcare-associated onset, higher CRP at time of positive blood culture, recent problems with the central venous catheter, treatment onset >24 h and positive follow-up blood cultures were more prevalent in cases with a complicated SAB (Table 5).

Discussion
To our knowledge, this is the first study presenting a comprehensive overview of patient characteristics and clinical outcomes of S. aureus bacteremia in patients receiving total parenteral nutrition in the setting of chronic intestinal failure. In summary, our data show that patients on TPN have a milder course of SAB with lower mortality rates compared to non-TPN patients. Healthcare-associated onset and increased CRP at onset, together with recent catheter problems, positive follow-up blood cultures and delayed initiation of therapy, were more prevalent in the small subset of TPN patients with a complicated outcome.
Surprisingly, overall mortality (10%) was remarkably lower in SAB patients receiving TPN. An explanation for this may be the fact that these patients are well educated about their infection risk and do not hesitate to call and visit the hospital as soon as signs of infection become present [34]. Although onset of antibiotic treatment after the start of symptoms did not significantly differ between TPN and non-TPN patients, delayed treatment onset was associated with a complicated outcome of SAB in patients on TPN (p < 0.001). Next, patients with TPN support were found to be younger, more frequently female and with less comorbidity in comparison to non-TPN patients with S. aureus bacteremia. These are all factors that are associated with a lower mortality risk in the general population [12,15,20,35]. Nevertheless, receiving TPN was still independently associated with an increased survival probability in the multivariate Cox model. Lastly, our findings of relatively low mortality rates in TPN patients can in part be the successful outcome of a strict adherence to antimicrobial stewardship programs in the Netherlands (e.g., bedside ID consultation and performance of echocardiography in all patients).
Forty-two percent of the TPN patients developed a complicated SAB, which is in accordance with previous research conducted in the general population (36-42%) [16,44]. In contrast, the study performed in oncology patients with a CVC by Zakhem et al. [16] reported a substantially higher complicated SAB rate of 67%. It is important to note that in contrast to our study, persistent fever and bacteremia beyond 72 h were accounted for as complicated SAB as well. We consider these factors as risk factors for a complicated course, not necessarily being a complicated course or always resulting in it. Besides, some studies include SAB attributable mortality as an indicator for complicated SAB [14,43]. We decided to exclude SAB attributable mortality, since we believe this is a rather subjective outcome.
To establish metastatic infection, 18 F-FDG PET/CT seems to be a valuable technique in TPN patients as well, since 80% (16/20) of the metastatic infections were diagnosed with 18 F-FDG PET/CT [26,45]. Thus, only in four patients the metastatic infection was diagnosed with other diagnostic modalities. Moreover, the majority (53%) of the TPN patients received less than 4 weeks of treatment for their S. aureus bacteremia since metastatic infection was considered to be ruled out safely with the use of 18 F-FDG PET/CT. The total number of metastatic infectious foci was lower in TPN patients compared to non-TPN patients, which is in agreement with previous research in patients with a CVC [46]. Nevertheless, endovascular and pulmonary metastatic infections were more prevalent in patients receiving TPN. These are probably directly related to the presence of a catheter since the endovascular metastatic infections included mostly a septic thrombosis near the catheter. Additionally, previous research showed a higher rate of pulmonary septic emboli when (septic) thrombosis was present [18].
Our findings of clinical characteristics of SAB cases associated with complicated outcome are in agreement with previous findings [14,16]. These studies additionally performed multivariate regression analysis to identify independent risk factors for complicated SAB: the strongest predictor Fowler et al. [14] found was a positive follow-up blood culture at 48 to 96 h. The other study, by Zakhem et al. [16], specifically studied SAB in patients with a CVC and found that catheter site inflammation was an independent predictor of complicated course, which was not associated with complicated outcome in our study. One of the other identified predictors by Fowler et al. [14], fever > 72 h, was also not significantly more prevalent in our study, nor in the study of Zakhem et al. [16]. A new finding of our study was the incidence of recent catheter problems (e.g., recent infection, dislocation or thrombosis) being more prevalent in TPN patients with a complicated outcome. Whether the identified clinical characteristics of our study are independent risk factors in patients receiving TPN as well needs to be confirmed in a larger patient cohort, since we decided not to perform multivariate logistic regression analysis with this small sample size.
Several studies found that failure or delay of CVC removal is an independent risk factor for relapse or hematogenous complications of SAB in patients with a CVC [16,19,43,[47][48][49]. Since in our study, in almost all TPN patients, the CVC was removed early in their illness course (according to protocol), we cannot confirm nor contradict these findings.
Our study is subject to several limitations: (1) the study was conducted retrospectively, data collection, however, was performed mostly prospectively and almost all patients were diagnosed and treated according to a predefined protocol, leading to low amounts of missing data; (2) analysis had to be done with a rather small sample size of patients receiving TPN. Nevertheless, regarding the rare nature of the conditions leading to dependency on TPN, much higher patient numbers are difficult to gather, unless data are collected as a multicenter or merged into, for example, a meta-analysis with collected individual participant data; (3) some of the TPN patients might be incorrectly considered as uncomplicated SAB since they died early in their illness course, leading to an underestimation of the true incidence of complicated outcome as well. To overcome this, we performed an additional analysis excluding these patients from analysis: no differences in outcomes were seen.

Conclusions
We conducted the first study of S. aureus bacteremia in chronic TPN patients, resulting in an extensive and robust data analysis with >90% complete follow-up data. It provides new insights about its presentation and outcomes in this vulnerable population, who are at continuous risk for developing severe bloodstream infections. We have showed that patients on TPN have a milder course of SAB with lower mortality rates compared to non-TPN SAB patients. Late onset of antibiotic therapy, previous catheter problems, and positive follow-up blood cultures seem to be associated with complicated outcome. Nevertheless, larger, preferably randomized and multicenter clinical trials are needed to further investigate and validate our findings.   According to the Centers for Disease control and Prevention (CDC) [29] Community acquisition of SAB SAB cases were community-acquired in all cases that did not meet the definition of healthcare associated SAB or hospital acquired SAB, thus infection in a patient who has had no contact with the health care system within the last 90 days.

Contamination
Contamination was defined as only one positive blood culture and no clinical signs of infection.

Exit site infection
An exit site infection is defined as an infection at the catheter exit site, with local erythema, induration and/or tenderness around the catheter exit site, and/or purulent discharge from the catheter exit site. When an exit site culture was performed and it was positive for S. aureus, it was labeled 'definite'. Without an exit site culture, it was labeled 'probable' [22]  Healthcare associated SAB SAB cases were healthcare-associated when patients were receiving home and/or ambulatory intravenous therapy, chemotherapy, hemodialysis, when patients were admitted in a hospital for at least 48 h within the last 90 days; and patients residing in a nursing home or long-term care facility [30].
Hospital acquired SAB SAB cases were labeled hospital-acquired if patients had been admitted for at least 48 h before the first positive blood culture.
Onset of SAB The first positive blood culture was defined as the onset of SAB.
Portal of entry for SAB The site that was most likely responsible for entry of S. aureus into the bloodstream, based on clinical signs, microbiological cultures and imaging results.
Staphylococcus aureus bacteremia (SAB) At least one positive blood culture for Staphylococcus aureus combined with clinical signs of infection.
Time of delay between onset of SAB and adequate treatment The onset of clinical symptoms (fever, chills) was used to define the time of delay between onset of SAB and starting of adequate treatment.