Assessing the Predictive Value of Methicillin-Resistant Staphylococcus aureus Nares Colonization Among Transplant Recipients and Patients With Neutropenia

Abstract Background Methicillin-resistant Staphylococcus aureus (MRSA) nares screening has been shown to be a powerful antibiotic stewardship tool for MRSA infections within 7 days of screening across a variety of anatomical locations given the high negative predictive value (NPV). However, the utility outside of 7 days and among transplant recipients and patients with neutropenia is less clear. Methods This was a retrospective cohort study across Veterans Affairs medical centers in the United States from 1 January 2007 to 1 January 2023 of patients tested for MRSA colonization and who had a subsequent positive bacterial culture within 28 days of MRSA screening. Sensitivity, specificity, positive predictive value, and NPV were calculated across different time points and anatomical culture locations. Results The cohort consisted of 686 174 patients, 6 277 437 MRSA nares tests, and 2 446 766 positive bacterial cultures within 28 days of MRSA testing. The NPV of MRSA nares screening for ruling out a MRSA infection within 28 days was 95.8% across all anatomical culture sites. The NPV was 97.9% among patients with neutropenia and 97.5% in solid organ and hemopoietic stem cell transplant recipients. Conclusions MRSA nares screening can reliably be used for de-escalation of anti-MRSA therapy within 28 days of bacterial culture for all patients, including solid organ and hematopoietic transplant recipients and patients with neutropenia.

Methicillin-resistant Staphylococcus aureus (MRSA) nares polymerase chain reaction (PCR) screening has proven to have a high negative predictive value (NPV) for MRSA infections across a variety of anatomical sites within 7 days of microbiological culture [1,2].This has allowed for the confident de-escalation and avoidance of empiric anti-MRSA treatment before the finalization of cultures [3,4].However, MRSA nares screening is not well validated in high-risk individuals, including those with neutropenia, hematopoietic stem cell transplant (HSCT) recipients, and solid organ transplant (SOT) recipients [5,6].
Febrile neutropenia is a common and potentially lifethreatening condition in cancer patients, with mortality rates approaching 10% [7].Infectious Diseases Society of America (IDSA) guidelines recommend empiric anti-Pseudomonas aeruginosa (Pa) coverage for all patients [8].In patients with a suspected catheter-related skin or soft tissue infection, pneumonia, or hemodynamic instability, the National Comprehensive Cancer Network and IDSA recommend adding anti-MRSA treatment [8,9].In SOT recipients, the rates of opportunistic infections are high due to the immunosuppressive medications required to prevent graft rejection [10].MRSA infections in SOT recipients are associated with longer hospital stays, increased hospital costs, and higher rates of mortality [11,12].However, the overuse of vancomycin is associated with increased adverse events and hospital costs [13].Therefore, timely and precise microbiological diagnosis is crucial to direct treatment and reduce unnecessary drug therapy [14].The primary goal of this study is to assess the predictive value of MRSA nares among high-risk individuals, including patients with neutropenia, HSCT recipients, and SOT recipients.
Risk factors for multidrug-resistant organisms (MDROs) are similar; thus, empiric broad-spectrum antibiotics constitute the backbone of therapy in critically ill patients [15][16][17].However, the IDSA guidelines recommend de-escalation of empiric therapy at 48 hours if microbial cultures do not identify an MDRO and the patient is clinically improving [18].While de-escalation aims to reduce antibiotic-related toxicity and limit the development of MDRO, this process is often delayed [19].Thus, the second goal of this study is to assess the predictive value of MRSA nares colonization for MDRO.

METHODS
A multicenter retrospective cohort study was conducted among the Veterans Affairs (VA) medical system from 1 January 2007 to 1 January 2023.All VA locations were queried, but results for Iowa and Nebraska are combined and Kansas is combined with Missouri due to overlapping station coding.Patients aged ≥18 years were queried for MRSA nares testing within the VA's Corporate Data Warehouse (CDW), which stores all VA patient medical data.MRSA colonization was assessed with PCR and standard culture techniques.Data were queried from CDW using structured query language (SQL) via the SQL Server Management Studio, and statistical analysis was completed with R programming language v.4.3.1 [20].The VA Informatics and Computing Infrastructure (VINCI)-a virtual machine architecture with dynamic memory-was used.Acquisition and analysis of the data was approved by the Edward Hines VA Institutional Review Board.
Date of birth, sex, race, hospital admission dates, absolute neutrophil count, International Classification of Diseases codes, and positive microbiological cultures within 28 days of MRSA colonization assessment were collected for all patients.The queried microbiological cultures of interest included MRSA, vancomycin-resistant Enterococcus, carbapenem-resistant Enterobacteriaceae, extended-spectrum β-lactamase (ESBL)producing gram-negative bacteria, Pa, Burkholderia, Acinetobacter, and Stenotrophomonas.
Neutropenia was defined as an absolute neutrophil count of <1500 cells/µL.Transplants included hematopoietic stem cells and solid organs, including the pancreas, lungs, liver, kidney, intestine, and heart.Culture sites were grouped by the anatomical collection site: wound, urinary, blood, intra-abdominal, pulmonary, reproductive, central nervous system, graft, device, drain, stool, and others.MRSA nares colonization was assessed using culture and PCR, but PCR results were only used to calculate the predictive performance given the rapid turnaround time and high accuracy.In the case of multiple MRSA tests being performed within 28 days of a positive culture, pseudorandom sampling was performed to eliminate bias from clustered data [21].Sensitivity, specificity, positive predictive value (PPV), and NPV were calculated.Proportional testing assessed for differences among proportions [22].Local polynomial regression fitting tested individual culture trends over time [23].

RESULTS
A total of 686 174 patients were screened for MRSA nares colonization, generating 6 277 437 total tests.The patient population was predominantly White (72.2%) and male (95.5%) with a median age of 68.7 years (Table 1).Of the MRSA nares screening tests, 59.1% were tested with PCR and 40.9% with standard culture.The percent positive of PCR and standard culture was 14.3% and 11.4%, respectively (P < .001; Figure 1A).
There were 2 446 766 positive culture results identified within 28 days of MRSA testing.The percent positive rate of MRSA colonization was 23.1% in individuals with a subsequent positive culture.Pa was the most common organism within our cohort at 10.6% while MRSA was second at 8.5% (Figure 1B).No other MDROs had an incidence of >2.5% in any of the study years.The incidence of ESBL-producing organisms, MRSA, and Pa increased throughout the study period (Figure 1B).Burkholderia and carbapenem-resistant Enterococcus had the lowest incidence among MDRO at 0.04% and 0.09%, respectively.Bacterial culture prevalence was stratified by state (Supplementary Table 1).The prevalence of MRSA infections was lowest in Maryland at 0.06% and highest in Alaska at 14.7%.The prevalence of Pa infections was lowest in Vermont at 6.5% and highest in Alaska at 20.7%.
MRSA nares screening provided for a high NPV among all MDROs (Table 2).Across all anatomical culture sites, the NPV of MRSA infections was 95.8% within 28 days of MRSA nares screening (Table 2).The NPV for devices and drains trended lower the farther screening was done from culture collection (Figure 1C).The NPV across all other anatomical sites remained consistent across the 28 days (Figure 1C).Interstate variability of MRSA nares predictive value was assessed for MRSA infections.The NPV was >91.8% for all states except for Montana, which had an NPV of 87.1% (Supplementary Table 2).The NPV for Pa decreased the most across the study period, starting at 88.8% and falling to 84.6% (Supplementary Figure 1).
There were 2149 patients with neutropenia; of those, 761 (35.4%) were patients with a hematologic malignancy.Among the patients with neutropenia, there were 3212 positive bacterial cultures identified within 4 weeks of MRSA nares screening.Of the positive cultures, there were 188 (5.9%) MRSA infections.The incidence of MRSA and Pa infections was significantly lower in the neutropenic cohort compared to the entire cohort (P < .001)(Supplementary Figure 2).Vancomycin-resistant Enterococcus (VRE) infections were significantly more common in patients with neutropenia compared to the complete cohort (P < .001;Supplementary Figure 2).The NPV of MRSA nares colonization was 97.9% for MRSA infections across all anatomical locations and >88% across all MDROs (Table 3).Among the entire cohort, 19 909 patients received an SOT or HSCT with 1658 patients receiving >1 transplant.There were 9294 kidney transplant recipients, 5226 liver transplant recipients, 3761 HSCT recipients, 2101 heart transplant recipients, 1051 lung transplant recipients, 233 pancreas transplant recipients, and 29 intestine transplant recipients.Among transplant recipients, the incidence of Pa infections was highest at 13.6%, MRSA at 6.8%, VRE at 2.8%, ESBL-producing organisms at 1.9%, Stenotrophomonas at 1.6%, Acinetobacter at 0.1%, and Burkholderia at 0.04%.The incidence of Pa and ESBL infections increased over the study period (Supplementary Figure 3).For MRSA infections, the NPV was 97.5% among HSCT and SOT recipients within 28 days of MRSA nares testing (Table 4).The cumulative NPV for Pa was 85.7% (Table 4).However, the NPV was lowest in lung transplant recipients at 71.1% (Supplementary Figure 4).

DISCUSSION
Although early active antibiotic therapy is associated with improved outcomes in patients with sepsis, appropriate and timely de-escalation is also important [24].Anti-MRSA therapy is associated with a higher risk of kidney injury, Clostridioides difficile infection, VRE infection, and secondary gram-negative rod identification [25].Unnecessarily broad antimicrobial treatment is also associated with higher mortality rates [26].Recently, MRSA nares screening has been shown to be a valuable tool in the de-escalation of anti-MRSA therapy, demonstrating a high predictive value for MRSA infections across all anatomical locations within 7 days of screening [1,2].In this retrospective cohort study, we assessed an expanded time from MRSA colonization testing to microbial culture, demonstrating a high NPV within 28 days of colonization testing.While smaller studies have reported a high NPV for respiratory cultures outside of the 7-day window, we assessed all anatomical culture locations [27].We found that for the majority of anatomical locations, the NPV remained high.This will allow for the confident avoidance of empiric anti-MRSA coverage, with fewer MRSA nares screening tests throughout a patient's hospitalization, decreasing testing burden and, potentially, healthcare costs.
The literature on high-risk patient populations and MRSA nares screening is limited to pneumonia.Recently, a small study (n = 98) investigated patients with acute myeloid leukemia and pneumonia, demonstrating a high PPV and NPV for MRSA nares screening [5].However, to our knowledge, no prior work has specifically investigated the predictive value of MRSA nares across other anatomical locations in patients with neutropenia or hematological malignancies.In febrile neutropenia, the most common source of bacteremia is from enteric bacteria translocation into the blood.Gram-negative rods, particularly Enterobacteriaceae, and Pa are the most common causative organisms and are associated with high rates of mortality [28].The IDSA recommends anti-Pa coverage with either a fluoroquinolone or a β-lactam agent [8,29].The IDSA also recommends against the use of agents targeting aerobic gram-positive cocci unless the suspected source is a catheter-related infection, a skin or soft tissue infection, or pneumonia or the patient demonstrates hemodynamic instability.Although hemodynamic stability was not included in our analysis, MRSA nares screening had a high NPV for catheterrelated, skin, soft tissue, and lung infections.The judicious use of vancomycin is particularly important in patients with neutropenia as antibiotic stewardship aims to limit the selection of resistant organisms, including VRE, which we found to be 5 times more prevalent in patients with neutropenia.SOT and HSCT recipients are 2 additional groups of patients who are at high risk for opportunistic infections, including MRSA infections [14,30].To our knowledge, there have been no published data on the predictive value of MRSA nares in these specific high-risk groups.Overall, the NPV remains consistent with other patient groups, highlighting the potential clinical utility for anti-MRSA de-escalation and avoidance.However, the consequences of untreated MRSA infections can be significant and antibiotic de-escalation must be considered closely [31].In this study, we did not attempt to assess the net benefit or harm that empirical treatment may provide.The overall NPV of this study is likely underestimated as the prevalence of MRSA colonization is higher in this patient cohort compared to the general population due to the older age of study participants [32].Prospective trials are needed to establish the safety and efficacy of MRSA nares screening to guide antibiotic therapy, especially in high-risk patient populations.
The IDSA recommends empiric MRSA and Pa coverage in severe infections.However, there are currently no well-validated PCR-based de-escalation tools for Pa [17].Therefore, anti-Pa treatment is guided by standard microbiological cultures.Given the similar risk factors for MRSA and Pa infections (ie, recent intravenous antibiotic use, hospital admission), we hypothesized that MRSA nares may be a useful tool for de-escalation of anti-Pa therapy [18].We found an NPV of 86.8% for Pa infections.However, the likelihood ratio and the incidence of Pa infections preclude the screening tool from any clinical utility.Future work will investigate alternative laboratory tests and clinical data for the effective de-escalation of empiric Pa coverage.

CONCLUSIONS
In the current study, we observed a high NPV of MRSA colonization for MRSA infections across various anatomical locations within 28 days of testing across a large patient cohort across the United States, including transplant recipients and patients with neutropenia.