Molecular concordance of methicillin-resistant Staphylococcus aureus isolates from healthcare workers and patients

Abstract

Background:

Methicillin-resistant Staphylococcus aureus (MRSA) is a significant nosocomial pathogen in the ICU. MRSA contamination of healthcare personnel (HCP) gloves and gowns after providing care to patients with MRSA occurs at a rate of 14%–16% in the ICU setting. Little is known about whether the MRSA isolates identified on HCP gown and gloves following patient care activities are the same as MRSA isolates identified as colonizing or infecting the patient.

Methods:

From a multisite cohort of 388 independent patient MRSA isolates and their corresponding HCP gown and glove isolates, we selected 91 isolates pairs using a probability to proportion size (PPS) sampling method. To determine whether the patient and HCP gown or gloves isolates were genetically similar, we used 5 comparative genomic typing methods: phylogenetic analysis, spa typing, multilocus sequence typing (MLST), large-scale BLAST score ratio (LSBSR), and single-nucleotide variant (SNV) analysis.

Results:

We identified that 56 (61.5%) of isolate pairs were genetically similar at least by 4 of the methods. Comparably, the spa typing and the LSBSR analyses revealed that >75% of the examined isolate pairs were concordant, with the thresholds established for each analysis.

Conclusions:

Many of the patient MRSA isolates were genetically similar to those on the HCP gown or gloves following a patient care activity. This finding indicates that the patient is often the primary source of the MRSA isolates transmitted to the HCP, which can potentially be spread to other patients or hospital settings through HCP vectors. These results have important implications because they provide additional evidence for hospitals considering ending the use of contact precautions (gloves and gowns) for MRSA patients.

Staphylococcus aureus, including methicillin-resistant S. aureus (MRSA), is a common cause of healthcare-associated infections that increase patient morbidity, length of stay, and mortality. ^1,2 Transmission of MRSA from patient to patient in healthcare settings is often an indirect transmission due to limited, or no, direct patient-to-patient contact; however, transmission in the healthcare setting is thought to most often occur via environmental or healthcare personnel (HCP) vectors. ³ Particularly, MRSA transmission from patient to HCP has been demonstrated to occur at a rate of 14%–20%, often through contamination of HCP gown or gloves after performing patient care activity on a patient with confirmed MRSA colonization and/or infection. ^3–5 These studies surmised that the isolates identified on the gown or gloves of the HCP are the same as those found on the patient; however, this was not directly demonstrated in these previous studies using genomic epidemiology.

Prior studies have used pulsed field-gel electrophoresis to determine MRSA relatedness between patients who were part of the hospital and outbreak investigations. ^6–8 However, this traditional typing method has limited ability to discriminate between closely related isolates when compared to newer, more comprehensive genomic methods, such as whole-genome sequencing (WGS), which has been demonstrated in recent studies characterizing MRSA transmission in the healthcare setting. ^9–11

Although genomic epidemiology approaches have been used to study the interactions among patients, healthcare workers, and the environment, no study to our knowledge has reported whether isolates on gloves or gowns of the HCP acquired after patient care activity are genetically similar or identical to isolates from the patient.

In this study, we sought to determine whether MRSA isolates on HCP gown or gloves after patient care are genetically similar to the MRSA isolates recovered from patients. We used multiple molecular typing schema to demonstrate the genetic relatedness between HCP gown and glove isolates and patient isolates.

Materials and methods

Isolate selection

Our cohort contained clinical or surveillance MRSA isolates from 388 independent patients from the intensive care unit (ICU) and the paired isolates from the corresponding HCP’s gown or gloves, as part of a previously described study. ^5,12,13 In the parent study by O’Hara et al, ⁵ the MRSA isolates from each patient were defined as high, mid-level, or low transmitters, with the low transmitters having no transmission events that occurred from the patient to HCP. The low-transmitting isolates were not included in this analysis. ⁵ This study was conducted across 4 hospitals, 2 in Maryland and 1 each in New York and California. Clinical cultures were defined as cultures ordered by HCP to determine whether patients had an active infection, and in comparison, surveillance cultures were cultures used to screen patients for colonization with MRSA and were taken at the time of admission and, depending on the unit, weekly until discharged. These patients were on contact precautions for MRSA; thus, HCP were required to don a new pair of gloves and a gown prior to entering a patient’s room. After an HCP entered the patient’s room and performed patient-care activities, the HCP gown and gloves were swabbed. Swabs were cultured onto a CHROMagar MRSA (Becton Dickinson, Sparks, MD) and incubated overnight. ^5,12,13

From the 388 independent patient isolates, we selected 96 paired isolates using stratified sampling. We selected the patient isolates in proportion to the number of the isolates that were identified as part of the different genomic clades identified in our previous study. ¹⁴ We then selected a paired HCP sample at random, either a glove or gown isolate from an HCP who provided care to the patient whom we selected previously. Figure 1 outlines how isolates were selected through the various steps of the current study.

Fig. 1. Study flow diagram for the paired isolates used in the study.

Genome sequencing

The genome sequencing and assembly used to analyze the patient isolates and HCP glove and gown isolates was described in Adediran et al. ^12,13 After 5 pair of isolates were removed from the analysis due to failing quality control metrics after sequencing or molecular typing issues, 91 pairs of isolates remained. Thus, 182 total isolates were included in the genomic epidemiology studies. All genome assembly metrics and accession numbers for isolates included in the comparative analysis are included in Supplementary Table 1 (online).

Comparative genomics

Phylogenetic analysis

The In Silico Genotyper (ISG) was used to infer the whole-genome phylogeny. ¹⁵ Sequence data from the patient isolates and HCP gown and glove isolates were aligned to the USA300-ISMMS reference genome (GenBank Assembly Accession: GCA_000568455.1). ¹⁶ Gaps in 1 or more genomes were removed to create the core alignments for the isolates. ¹⁷ A phylogenetic tree was created using FastTree as previously described ^18–20 and visualized with FigTree version 1.4.0 software (http://tree.bio.ed.ac.uk/software/figtree/). Genetic concordance was defined as paired isolates within the same phylogenetic group (Fig. 2).

Fig. 2. Phylogenetic analysis of newly sequenced methicillin-resistant Staphylococcus aureus (MRSA) paired isolates. Genomes were aligned to one another, and 102,599 single-nucleotide polymorphisms (SNPs) were identified using ISG. ¹⁵ RAxML ⁴⁰ was used to create the phylogenetic tree using 100 bootstrap replicates, and FigTree (http://tree.bio.ed.ac.uk/software/figtree/) was used for visualizations. ^15,40 Black brackets represent paired isolates neighboring each other on the tree and are within the same group. Green brackets represent paired isolates that are within the same phylogenetic group. Red brackets represent paired isolates that are not within the same group and do not neighbor each other.

MLST analysis

The 7 conserved housekeeping loci (arcC, aroE, glpF, gmk, pta, tpi, and yqiL) of the MLST scheme previously developed were identified in each of the genomes. ²¹ The allele numbers of each locus and the sequence types (STs) of each genome were determined using BIGSdb software (https://pubmlst.org/saureus/). ²² We identified the STs and clonal complex (CC) for each patient isolate and HCP gown or glove isolate. We defined genetically similar isolates as patient isolates with the same ST and CC as the HCP gown or glove isolates.

spa typing analysis

A spa-typing analysis was performed on the 182 MRSA isolates of interest using spaTyper version 1.0 software (Center for Genomic Epidemiology, Denmark, https://cge.cbs.dtu.dk/services/spatyper/) with default parameters. ²³ Genomes were examined to identify the spa types for each patient and HCP gown or glove isolates. ²³ We defined the isolates as genetically similar when the patient isolate exhibited the same spa type as the corresponding paired HCP gown or glove isolate.

Large-scale BLAST score ratio (LSBSR)

LSBSR analyses were performed on the isolates as previously described. ¹⁷ The LSBSR uses predicted coding sequences from all query genomes to align each coding sequence to each genome. Each alignment generates a query bit score. ²⁴ The query bit score is divided by the reference bit score to obtain a final BSR value. We completed a gene-by-gene pairwise comparison of the genomic content of the paired isolates (ie, patient isolates and HCP gown or glove isolates). We defined overall genetic similarity as the paired isolates having genomic content that was 90% similar, which was calculated by the number of genes that had the same LSBSR value divided by the total number of genes within the genomes. ²⁵

SNV analysis

We conducted a single-nucleotide variant (SNV) analysis using ParSNP (https://github.com/marbl/parsnp). We conducted pairwise comparisons for each pair of isolates with the patient isolate as the reference. We determined the number of SNVs between each of the paired isolates. Isolates were defined to be the same if they differed by <40 SNVs, a threshold previously utilized when examining genetic similarities of MRSA isolates. ^9,11,27,28 Bee-swarm plots were created to visually examine the threshold for the defining number of SNVs. ^9–11 We calculated the summary statistics using R version 4.02 software (R Foundation for Statistical Computing, Vienna, Austria). ²⁹

Results

Phylogenetic analysis of MRSA paired isolates

Phylogenetic analysis was performed on the 91 paired MRSA isolates. We identified 4 main phylogenetic groups among the paired isolates, which corresponded to the 4 main phylogenetic groups identified in the parental genomic study. ¹⁴ The most frequent transmission type among the patient isolates was midlevel transmitters (n = 64 of 91, 70%), which were defined as MRSA transmitted to the HCP at rates between 1% and 49%, based on the examination of 10 HCP–patient interactions. The remaining 27 patient isolates (30%) were considered high transmitters, defined as a transmission rate >50%, based on the examination of 10 HCP-patient interactions. Of the 91 patient isolates, 47 (52%) were obtained from clinical cultures; the remaining isolates were obtained from surveillance cultures. We detected no statistical difference between these groups and phenotypic transmission type. Also, 71 (81%) of the examined isolates came from Maryland. Comparing the transmission and isolate type (ie, clinical vs surveillance or high vs midlevel transmitters) by genomic group, we identified no significant association between these groups (P = .34 and .33, respectively). However, we identified geographic location to be the most significant association with the genomic groups (P < .001). We identified 76 (83.5%) paired isolates that were genetically similar (Table 1).

Table 1. Typing Schema Among Paired Patient and HCP Gown or Glove Isolates (N=91)

^a Concordance was defined as paired isolates with same spa type.

^b Concordance was defined as paired isolates with same CC type.

^c Concordance was defined as paired isolates with a genetically similar ≥90%.

^d Concordance was defined as paired isolates with <40 SNVs.

^e Concordance was defined by the phylogenomic similarity in Figure 2.

MLST typing

In total, 10 MLST types were identified from the 91 paired isolates. Among the typable isolate pairs, the MLST sequence types were the same between the patient and HCP gown and glove isolates in 54 (59.3%) of the 91 isolate sets. Additionally, 57 (62.6%) of the 91 paired isolates shared the same clonal complex. (Table 1)

spa typing

We identified a total of 18 different spa types among 91 paired isolates. Among both the patient isolates and HCP gown or glove isolates, the most common spa types were t008 (n = 33 of 91, (36.8%) and t002 (n = 17 of 91, 18.7%). We defined genetic concordance as paired isolates with the same spa type. Based on our definition by this analysis, 71 (78%) of 91 paired isolates were genetically similar. (Table 1)

LSBSR

The genome content of the patient and HCP gown or glove isolates were analyzed using LSBSR. ¹⁷ The LSBSR matrix is composed of 8,523 potential coding sequences. Of the 91 paired isolates, 77 (84.6%) were considered genetically similar based on our definition. Among the discordant pairs, the range of gene content concordance was 34%–53.4% (Table 1).

SNV analysis

The minimum number of SNVs between the paired isolates was zero, and the maximum number of SNVs was 62,464, with a median value of 48.5 SNVs between the paired isolates. Among the 91 paired isolates, 45 (48%) were genetically similar by this metric (Fig. 3).

Fig. 3. Single nucleotide variant differences within paired isolates paired methicillin-resistant Staphylococcus aureus (MRSA) isolates using Parsnp. ²⁶ A bee-swarm plot was used to plot single-nucleotide variant (SNV) differences and was generated using R version 4.02. ²⁹ Genetic concordance was defined as paired isolates differing by <40 SNVs as previously defined in the literature. ^9,10

Fig. 4. A heatmap of the frequency of genetic concordance among the paired isolates using the 5 comparative genomic techniques in the study. The line on the figure is the line of concordance. Paired isolates below the line are considered discordant based on the 4 of the typing methods.

Summary of genomic epidemiology results

We examined the frequency of paired isolates being considered genetically similar based on all the typing methods used. Only 28 (30.7%) of the 91 paired isolates were considered to be genetically similar using all 5 typing mechanisms, followed by 28 paired isolates (30.7%) that were genetically similar in 4 of 5 typing schemas (Fig. 4). The most frequent discordant typing schema was SNV, with 49 samples being discordant.

Discussion

The objective of this study was to determine whether MRSA isolates identified from HCP gown or gloves were genetically similar to MRSA isolates from the patient. Our phylogenetic analysis identified 83% of the paired isolates as genetically similar. Similarly, the spa typing and the LSBSR analysis indicated that >75% of the examined isolate pairs were concordant. Among the 5 typing schemes, 56 pairs (61.5%) were considered concordant based on criteria of being concordant on 4 typing schemes. We utilized several typing methods of varying discriminatory power to convey genomic differences between the paired isolates. This is the first study to our knowledge that has employed genomic epidemiology to understand patient-to-HCP transmission in multiple-ICU setting.

Few previous studies have used WGS to determine whether MRSA transmission occurred in the healthcare setting ^9–11 ; however, each of the previous studies differs significantly from our study. Stine et al ¹⁰ focused on direct acute patient-to-patient transmission rather than patient–HCP transmission, and they used an SNV-based analysis that identified 3 transmission clusters in nursing home over a 12-week period. Two additional WGS-based studies focused on MRSA transmission by examining patient, HCP, and environmental surfaces such as computers and mobile devices in the ICU setting. ^9,11 Price et al ⁹ and Popovich et al ¹¹ each examined how the HCP or environment could be potential vectors of transmission to patients in the ICU setting using a longitudinal cohort. Both studies identified transmission events between patient and the HCP; acquisition occurred 7 of 25 times in the study by Price et al and 4 of 6 times in the study by Popovich et al. ^9,11 However, Price et al focused on HCP nasal carriage as a proxy of potential transmission, which is significantly different than our study, which used HCP gown or gloves as a measure of transmssion. Nasal carriage suggests potential colonization and does not consider transient contamination and short-term carriage that fails to result in colonization.

In contrast, our study focused on the acute transient transmission of MRSA from the patient to HCP gown and gloves. We obtained isolates from the gown and gloves of HCP immediately after patient-care activity, suggesting an acute transmission event directly or indirectly from the patient to the HCP. Due to the longitudinal focus and the time between patient contact and measurement of the HCP, Price et al may not have ascertained direct acute transient transmission, which has been demonstrated to be a frequent occurrence (16.2% of the time in MRSA) in the ICU setting among HCP- and MRSA-positive patients. ^5,6 Additionally, we are the first researchers, to our knowledge, to employ multiple genomic epidemiology techniques to ascertain transmission of MRSA from the patient to HCP.

We anticipated that many paired isolates would be genetically similar; however, we identified several isolate pairs that were not genetically similar depending on the molecular typing schema used (20%–48%). Several hypotheses may explain these results. First, HCP may have picked up isolates from the patient room environment when performing healthcare activities; thus, the identified isolate may not be directly from the current patient but rather from other patients or sources, such as the HCP themselves or equipment within the ICU. ^6,7,9,11

Another possible explanation of why HCP gown and glove isolates differed from the patients isolate following patient-care activity is that the patient may harbor multiple MRSA strains that were not detected in the clinical sample. We did not capture the genomic diversity among the patient isolates because we examined only a single MRSA isolate per patient for WGS; however, patients may have multiple MRSA isolates from a single swab. ^10,30 Previous studies have demonstrated that some patients have >1 MRSA isolate, with the prevalence of multiple isolates in patient samples being as high as 38%. ^10,30 Additional studies that examine multiple diverse isolates per sample with WGS may be required for a complete understanding of the diversity of the patient and HCP samples.

Third, isolates identified on gowns and gloves of HCP could be from HCP nasal or hand carriage. The prevalence of MRSA carriage among HCP has been previously measured at 4.6%. ^31,32 Studies have demonstrated the HCP as a possible source of MRSA transmission through possible shedding from HCP nasal carriage. ^33,34 HCP may have unknowingly contaminated their gown and gloves with MRSA while performing routine daily duties, which might have facilitated spread to the patient environment and, subsequently, the patient.

Lastly, HCP gowns and gloves can be contaminated in the common areas where gowns or gloves are housed. HCP don new gowns and gloves from the communal supply area before entering the patient’s room. Diaz et al ³⁵ identified that 75% of gloves tested from examination rooms were positive for bacterial pathogens including coagulase-negative Staphylococcus, Bacillus spp, Pseudomonas aeruginosa, and Acinetobacter baumannii. However, gloves from a newly opened box were not contaminated, suggesting that contamination occurred after opening. ³⁵ However, additional studies have demonstrated that there is little contamination found in glove boxes. ³⁶ Further studies are needed to demonstrate the risk of contamination of glove boxes to determine whether this hypothesis explains some of the observed discordance between the paired isolates.

Despite its novelty, this study had several limitations. First, neither the HCP gown nor gloves were fully cultured to examine the total genomic diversity of the MRSA. We examined a sample from the gown and gloves using a standardized technique described in previous studies, which were the most likely areas that came into contact with the patient. ^4,5,36,37 Additionally, we did not find an association of the genotypes isolated or diversity observed with the origin of the HCP sample (glove or gown). Second, we did not culture the patient environment; therefore, we did not determine whether the isolates found on the gown and gloves of HCP were also common in the environment. Distinguishing between environmental and patient isolates may be difficult because patient-care activities require interaction with the environment (eg, blood pressure cuffs, IV tubing) as well as the patient. Third, we did not swab HCP hands and nasal carriage before patient-care activities to determine the MRSA burden and genomic diversity on the HCP. Finally, we did not attempt to assess the possible transmission from the HCP to secondary patients. Although it is an important aspect of organismal transmission, this study was not designed to assess secondary transmission; we examined the primary transmission events. Establishing secondary transmission patterns from the primary HCP would be interesting, but it was beyond the scope of analysis.

Overall, our results demonstrate that transmission of MRSA from the patient to HCP does occur when HCP care for patients, and most paired isolates were genetically similar. Comparative genomics has increased our understanding of the isolates identified on the gown and gloves of HCP. These findings strengthens our knowledge regarding the extent to which MRSA patients contaminate the HCP gown and gloves following HCP–patient interaction. These data suggest that if healthcare workers were not wearing gloves and gowns, their hands and clothing would frequently become contaminated with MRSA, resulting in subsequent transmission to other patients and the hospital environment. Our results provide important data related to the debate about the pros and cons of glove and gown use (ie, contact precautions) as part of hospital MRSA control programs. ^38,39