To our knowledge, this is the largest study of enterococcal CLABSI in cancer patients and the first to have fully defined comparator groups. The predominance of E. faecalis isolates in G2 (patients with documented CRBSI and CLABSI without MBI) could be attributable to the superior capability of E. faecalis to form biofilms . From the data at hand it is difficult to make a definitive determination of the value of early CVC removal, however there was a trend for a better success rates in G1 and G2 with early CVC removal (< 3 days) compared to CVC removal between 3 and 7 days. However, surprisingly, CVC retention was associated with a high success rate in all three groups. CVC could have been retained in more clinically stable cases which could explain the high success rate with catheter retention. Unfortunately, the rationale behind the decision to remove or retain the CVC was not available. Although CVC has been often removed unnecessarily in the setting of non-CLABSIs , the clinical practice guideline by the Infectious Diseases Society of America recommends CVC removal for most pathogens when the CVC is the likely source of the bloodstream infection. In this setting, if the catheter is retained, an antibiotic lock could be used particularly for bloodstream infections caused by enterococcus, coagulase negative staphylococcus, and gram-negative bacilli . In cancer patients, early removal of the CVC within the first 3 days of Staphylococcus aureus-CLABSI has been associated with better outcome and a lower probability for relapse . Similarly, in patients with gram-negative bacilli, CVC removal within 48 to 72 h was associated with a better infectious outcome and a lower rate of mortality [9, 10]. Likewise, in patients with commensal organisms such as coagulase-negative staphylococcal or bacillus causing CLABSI, CVC removal was associated with a lower rate of relapse compared to CVC retention [11, 12]. Potential benefit of removal of CVC in the context of enterococcal CLABSI can be inferred from available literature, albeit with limited data and small sample sizes.
Sandoe et al.  evaluated treatment outcomes in 61 cases of enterococcal CRBSI. Cure was achieved in 40 of 48 (83%) episodes managed with CVC removal but only 5 of 13 (38%) episodes in which the CVC was retained (and patients received combined antimicrobial therapy including an active cell wall-acting agent and an aminoglycoside). The study did not address the timing of CVC removal. Despite the study’s small sample size, the authors concluded that CVC removal resulted in higher cure rates and that combination therapy is needed if the CVC is to be retained .
Reigadas et al.  retrospectively examined 75 episodes (in 73 patients) of enterococcal CRBSI, focusing on patient characteristics and risk factors. They concluded that the high mortality rate observed in patients with enterococcal CRBSI required a better therapeutic approach .
In a retrospective review, Marschall et al.  compared outcomes of patients with retained CVCs to those who underwent CVC removal in a cohort of 111 patients with enterococcal CLABSI. They found that in-hospital crude mortality, 30-day mortality, and 90-day mortality were all associated with CVC retention, although they did not specify the time interval in which the CVC was removed. In addition, that study lacked a comparator group .
Enterococcus species are generally considered to be of low virulence, so the previously reported association of enterococcal infections with higher mortality and a poor prognosis could be due to the association of these infections with malignancy and [“other”?] chronic comorbid conditions [16, 17]. In our current study, harboring an isolate displaying vancomycin resistance was associated with higher infection-related and all-cause mortality, attributed by multivariate analysis to isolation of E. faecium and hospitalization in a critical-care setting. The poor outcomes of VRE BSIs were reported in a retrospective review of 7128 adult and pediatric patients who had received their first HSCT. Multivariable models showed that VRE-BSI was associated with higher non-relapse mortality and lower overall survival . In our present study, most VRE isolates were found in G1 (CLABSI with MBI), which also happened to have a higher VRE colonization rate (Table 1). The majority of our VRE isolates speciated into E. faecium (90%). While the emergence of E. faecium as a pathogen with poorer outcomes than E. faecalis has been well reported in the literature , in our cohort, the poor outcomes may have been associated with its resistance to vancomycin.
Our study was limited by its retrospective nature, particularly in that the indications for CVC management were not consistently documented. The decision to remove the CVC was also based on the decision of the treating physician, with no clear pattern or time frame and often without a documented rationale and irrespective of clinical status at point of removal. Another limitation is that given the retrospective design of the study, patients were not followed on a defined prospective clinical protocol. Therefore, not all patients had follow-up daily blood cultures, although it is a standard practice of care to repeat blood cultures, and some patients had missing data for some variable. Hence, these patients were excluded in the final analysis. This reduced the number of cases and may have impacted the statistical significance of our data. We also depended on phenotypic susceptibility pattern to identify recurrent isolates, use of more accurate methods was not possible given the retrospective nature of the study (physical samples no longer available).