Background
Cancer patients are highly susceptible to bloodstream infection (BSI) due to frequent hospital admissions, cytotoxic chemotherapy, use of invasive procedures, and exposure to broad-spectrum antibiotics [
1,
2]. Accordingly, they witnessed a more significant increase in the incidence of BSI, and a higher mortality rate than noncancer patients in recent years [
3], with prevalence ranging from 11 to 38% and the mortality rate around 40% [
4]. In a study investigating nearly 14 million patients with BSI in the US from 2006 to 2014, Gram-positive bacteria are found to be the leading causative pathogens (27.38%) in cancer patients [
3]. despite a shift from Gram-positive to Gram-negative organisms has been documented in recent years [
5].
Staphylococcus aureus, a common Gram-positive bacterium colonizing the skin, the nares, and the perineum, frequently causes skin, soft tissue, and bloodstream infections in human beings [
6].
S. aureus Bacteremia (SAB) is one of the most serious situations in
S. aureus infections and is related to mortality rates of 15–60% [
7,
8]. Underlying malignancy has been reported to be a risk factor for mortality in patients with SAB in several studies [
9‐
12]. In turn, the presence of SAB also indicated an increased risk of death in cancer patients [
13,
14].
S. aureus is known to be frequently antibiotic-resistant, and methicillin-resistant
S. aureus (MRSA) infections have been the main cause of mortality and the immense economic burden attributed to
S. aureus infections worldwide [
15,
16]. Methicillin-resistance adversely affected the outcome of patients with SAB whether in the general population or cancer patients [
9,
14,
17‐
19], and appropriate empirical antibiotic treatment significantly improved the outcome [
20]. The recommended first-line therapy for MRSA bacteremia is appropriately dosed vancomycin, with daptomycin an effective alternative [
21]. However, the NCCN Guidelines strongly recommended vancomycin not be routinely included in the empiric therapy alone for cancer patients [
22], which could impair the prognosis of patients with MRSA BSI. Thus, there is an urgent need for data on the prevalence, and the risk factors for the development and mortality of MRSA BSIs in cancer patients for better management of MRSA in this population. In this systemic review, we performed a meta-analysis on the global MRSA prevalence among bacteremia in patients with malignancy and further summarized the limited information on the predictors and mortality of cancer patients with MRSA BSI.
Methods
This study was conducted and reported based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [
23].
Literature search
We searched the PubMed and EMBASE databases for studies published up to 11 Mar 2020. The search strategy consisted of #1 (tumor* OR tumour* OR cancer* OR carcinoma* OR malignanc* OR neoplasia* OR oncolog* OR sarcoma* OR hematolog* OR haematolog* OR leukemia* OR leukaemia* OR lymphoma*), #2 (“bloodstream infection*” OR “blood stream infection*” OR bacteremia OR bacteraemia), #3 (“Methicillin-Resistant
Staphylococcus aureus” OR MRSA OR “
Staphylococcus aureus” OR “
S. aureus”). The search was conducted as #1 AND #2 AND #3 limited to the title and abstract with the filters: human, English, and recent 20 years. Additional articles were identified by manually searching reference lists of the identified articles and relevant reviews [
24,
25]. Further articles citing the included studies were retrieved using Web of Science.
Study selection
Studies were included if they met either inclusion criterion, and did not meet any of the exclusion criteria. The inclusion criteria were: 1. studies containing primary data on the total number of blood isolates from BSI cases and the number of MRSA isolates in 10 or more cancer patients; 2. studies that provided data on the predictors or mortality of cancer patients with MRSA BSI. The exclusion criteria were: 1. no differentiation between laboratory-confirmed infections and contamination; 2. studies that focused on microbial isolates with no relation to clinical BSI cases; 3. MRSA outbreak; 4. non-English publications; 5. not available in full-text, including conference abstracts. Title and abstract screening and full-text assessment were conducted independently by two authors after duplicates were removed, with discrepancies resolved by consensus.
Data extraction and quality assessment
Data collected included study characteristics (primary author, year of publication, country, study design, study period, and study setting), prevalence (the total number of BSI isolates and the number of isolates of Gram-positive bacteria, S. aureus, and MRSA). The total number of microbial isolates from cancer patients with BSI was collected to calculate the prevalence of MRSA, because the number of patients, as well as the BSI episodes, was not available in all studies. MRSA rates in S. aureus were calculated in a study when the number of S. aureus was more than 10. Furthermore, demographic and clinical factors associated with MRSA predictors and mortality, as well as the mortality rate of patients with MRSA BSIs, were collected.
The methodological quality of all identified studies was evaluated by the Newcastle-Ottawa scale [
26]. Based on this scale, studies received stars across three categories: selection and comparability of study groups, and outcome ascertainment. Generally, the rating criteria were: Low quality = 0–5; medium quality = 6–7; and high quality = 8–9. The studies on prevalence could score five stars at most and were considered of high quality with four or five stars [
24], because the non-exposed cohort and the outcome at the beginning were not applicable to these studies.
Statistical analysis
A random-effects meta-analysis was performed to estimate the pooled prevalence of MRSA with 95% confidence intervals (95% CIs) [
27]. World Health Organization (WHO) geographical stratification was applied to study the geographic variation of MRSA prevalence. Temporal trends in MRSA prevalence was assessed by linear regression. The proportions of MRSA isolates in two subgroups were compared with Chi-square test. The meta-analysis, linear regression, and Chi-square test were conducted by MedCalc statistical software, version 15.2 (MedCalc Software, Ostend, Belgium). Publication bias was tested by funnel plot and Egger’s test, and the data were analyzed by the statistical software R (Version 3.6.3, R Foundation for Statistical Computing, Vienna, Austria) with the meta package (Version 4.11–0). Rates in each study were stabilized using the Freeman-Tukey double-arcsine transformation before analysis to minimize the effect of extremely small rates [
28]. Study heterogeneity was assessed with the I
2 test and interpreted as relevant if > 50%. Generally, a
P-value of < 0.05 was considered statistically significant, but statistically significant publication bias was suggested when
P-value was less than 0.1 on Egger’s test.
Discussion
S. aureus is one of the ‘ESKAPE’ organisms that are responsible for the majority of bacterial infections in patients with malignancy [
55]. Besides, bacteremia and multi-drug resistance has become a growing issue in cancer patients [
4]. In our study, MRSA is the causative pathogen in 3% of BSIs among patients with malignancy in general. This rate can be as high as 28% in an African country, rendering Africa being the second-highest region considering MRSA rates. However, three in five African countries had MRSA rates lower than the global pooled prevalence, indicating high heterogeneity among different regions. Statistically, high heterogeneity was present in each subgroup. Thus, The treatment of BSIs in patients with cancer should take the local microbiology and antibiotic-sensitivity patterns into account, besides referring to the established guidelines [
56]. It is notable that limited information is available in the literature on MRSA BSIs in South-East Asia, Eastern Mediterranean, and Americas, suggesting the need for further studies of high quality in these regions to better understand the overall burden of MRSA BSIs in cancer patients.
In recent years, a variety of effective measures, including improvements in preventing healthcare-associated infections and MRSA transmission interruption in hospitals, were taken to control infections [
57]. As a result, hospital-onset MRSA bacteremia rates substantially reduced between 2005 and 2012, and the rate of decrease has slowed since 2012 in the general population of the US [
57]. However, this decline was not detected in MRSA-BSIs among patients with malignancy, which might be explained by two possible reasons: firstly, although with the emergence of colony-stimulating factors, Immunomodulatory drugs, etc., cancer patients are still exposed to cytotoxic chemotherapy, broad-spectrum antibiotics, and undergo frequent invasive procedures, rendering them immunosuppressed and susceptible to MRSA; secondly, the rate of community-associated MRSA (CA-MRSA) remains stable, and there is a surge of MRSA in some regions [
16,
57]. CA-MRSA, unlike hospital-acquired MRSA which could be effectively controlled by vigorous hand hygiene, antimicrobial stewardship, and barrier precautions, lacks easily targeted prevention strategies [
58]. Strains of CA-MRSA are more likely to spread in densely populated regions, and therefore might be controlled by constant surveillance and early intervention [
58]. This could also partly explain the relatively high rate of MRSA in Western Pacific and South-east Asia in our analysis.
MRSA BSI predictors among cancer patients were reported in three studies. All of the studies compared MRSA to MSSA, but each study contained different clinical factors, which precluded a pooled analysis. The identified predictors were mainly related to healthcare-associated infections and immunosuppression. In a study exploring MRSA BSI predictors in HIV-infected patients, multivariate analysis revealed that frequent hospitalization, low numbers of CD4+ peripheral cells, and previous administration of beta-lactams were found to be independent risk factors of MRSA development [
59], which is partly in line with the findings in cancer patients. Prior antibiotic use was identified as the only independent predictor of MRSA bacteremia after analysis of data from two prospective multi-center studies in the general population [
60]. Unfortunately, this clinical event frequently experienced by cancer patients was not studied in our included studies. Further studies comparing MRSA BSIs with all-cause BSIs and involving more demographic and clinical factors are needed to establish a reliable prediction rule for MRSA BSIs in the cancer population.
As reported in the previous researches, the 30-day mortality of MRSA BSIs ranged from 16 to 44% in general hospital populations [
11,
20,
61‐
63]. This mortality was attributed to multiple complications of MRSA bacteremia, including infective endocarditis, deep tissue abscess, and septic shock [
64]. However, data on the mortality rate of MRSA-BSIs in cancer patients is scarce, and hard to make a comparison. When it comes to the risk factors for mortality among patients with MRSA bacteremia, community-onset infection, secondary BSI, and vancomycin MIC≥2 g/mL were significant in cancer patients [
53]. The relation of vancomycin MIC and mortality was also found in the non-cancer population [
61,
65]. Notably, several studies have identified the association of inappropriate empirical antibiotic treatment with increased mortality among patients with MRSA BSIs [
20,
62,
63,
66], and cancer patients were at an increased risk of receiving inappropriate therapy [
62]. Appropriate initial antibiotic treatment should, therefore, gain adequate attention among patients with malignancy.
Several limitations of our study deserve consideration. First, all the included studies are observational, and there is a possibility of selection and observational bias. Second, the data are limited on BSIs solely in patients with solid malignancies as well as on BSIs from some geographic regions. Thus, we could not depict a more comprehensive picture of MRSA BSIs in cancer patients. Third, a direct comparison of the MRSA prevalence between cancer patients and general patients could not be achieved, which is precluded by the huge population of the latter group. Fourth, we did not find any study that provided incidence data after the search, so we could not predict the incidence trend. Fifth, a comparison between MRSA and MSSA was not made in our study, because we aimed to study MRSA which was a greater threat to life and the economy. Finally, the scarce information on predictor and mortality of MRSA bacteremia in cancer patients did not allow us to perform risk difference analysis and further establish decent prediction rules for MRSA development and mortality. Nevertheless, this study does provide relevant information on the prevalence, predictors, and mortality of bloodstream infections due to MRSA in the cancer population.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.