Changing Characteristics of Staphylococcus aureus Bacteremia: Results From a 21-Year, Prospective, Longitudinal Study

Background. We conducted a longitudinal study to evaluate changes in the clinical presentation and epidemiology of Staphylococcus aureus bacteremia (SAB) in an academic, US medical center. Methods. Consecutive patients with monomicrobial SAB were enrolled from January 1995 to December 2015. Each person’s initial bloodstream S. aureus isolate was genotyped using spa typing. Clonal complexes (CCs) were assigned using Ridom StaphType software. Changes over time in both the patient and bacterial characteristics were estimated with linear regression. Associations between genotypes or clinical characteristics and complications were estimated using multivariable regression models. Results. Among the 2348 eligible participants, 54.2% had an implantable, foreign body of some type. This proportion increased significantly during the 21-year study period, by 0.96% annually ( P = .002), as did comorbid conditions and acquisition outside of the hospital. Rates of any metastatic complication also significantly increased, by 0.94% annually ( P = .019). Among the corresponding bloodstream S. aureus isolates, spa -CC012 (multi-locus sequence type [MLST] CC30), -CC004 (MLST CC45), -CC189 (MLST CC1), and -CC084 (MLST CC15) all significantly declined during the study period, while spa -CC008 (MLST CC8) significantly increased. Patients with SAB due to spa -CC008 were significantly more likely to develop metastatic complications in general, and abscesses, septic emboli, and persistent bacteremia in particular. After adjusting for demographic, racial, and clinical variables, the USA300 var-iant of spa -CC008 was independently associated with metastatic complications (odds ratio 1.42; 95% confidence interval 1.02–1.99). Conclusions. Systematic approaches for monitoring complications of SAB and genotyping the corresponding bloodstream isolates will help identify the emergence of hypervirulent clones and likely improve clinical management of this syndrome.

Patients were excluded from the study if they were: <18 years of age; neutropenic (defined as an absolute neutrophil count ≤1 × 10 9 /L); not admitted to the hospital; previously enrolled for another SAB episode; had no signs or symptoms of infection; an additional clinically significant bacterial pathogen was isolated from their blood culture; or they declined informed consent. Within the SABG-PCS cohort, only patients whose initial bloodstream S. aureus isolate was viable and available for further testing met the inclusion criteria for the present study and were further analyzed.
Only the initial presentation for eligible patients was included in the study. Clinical data were collected on a standardized case report form and entered into an electronic database (Microsoft Access), including demographics; past medical history; history of surgery within the previous 30 days; site of acquisition of SAB; follow-up blood culture results; initial source of SAB; Acute Physiology and Chronic Health Evaluation (APACHE) II score [10], calculated on the day of the index positive blood culture; metastatic complications; and outcome at 90 days. The outcomes of SAB, per study protocol, recorded deaths (attributable to S. aureus or other causes), cures, or recurrent bacteremia within the 90-day follow-up period.
The study was approved by the Duke Institutional Review Board. Written informed consent was obtained from patients or their legal representatives. If a patient expired prior to the notification of their blood culture results, the subjects were included using an Institutional Review Board-approved Notification of Decedent Research.

Definitions
Instances of SAB were categorized either as hospital-or community-acquired, according to criteria previously defined [11]. The latter category was further delineated as either communityacquired, healthcare-associated (CA-HCA) or communityacquired, nonhealthcare associated (CA-NHCA) [11].
A metastatic infection was defined as a site of S. aureus infection resulting from the spread from an initial site (via bloodstream seeding or direct extension). Patients were considered to have metastatic infections if they developed any of the following conditions: infective endocarditis, vertebral osteomyelitis, septic arthritis, septic emboli, septic thrombophlebitis, a metastatic abscess, or another deep tissue abscess (ie, epidural or psoas abscess). Distinguishing a metastatic infection from the source of SAB is often difficult, and was out of the scope of this study.
Clinical diagnoses for endovascular, osteoarticular, pulmonary, or other metastatic sites of S. aureus infections were based on clinical findings and laboratory, microbiology, radiology, or pathology data. Infective endocarditis was diagnosed according to the modified Duke criteria [12]. Persistent bacteremia was defined as ≥5 days of positive blood cultures after an appropriate treatment was initiated. Appropriate antimicrobial therapy was defined as therapy that included at least 1 antimicrobial, active in vitro against the SAB isolate. Recurrent SAB was defined as another episode of SAB within the 90-day follow-up period, after the resolution of the first episode.

Laboratory Studies
The identification and methicillin susceptibility testing are described in the Supplementary Material.

Molecular Typing
The initial bloodstream S. aureus isolate from each eligible patient underwent spa typing using an in-house, high-throughput, DNA isolation method (Supplementary Material) and polymerase chain reaction for the spa gene, with specific primers [13].
The assignment of spa types was performed using Ridom StaphType (Ridom GmbH, Würzburg, Germany). The spa types were clustered into spa Clonal Complexes (spa-CCs) using the based upon repeat pattern algorithm, at a cost setting of ≤4 and excluding spa types with <5 repeats. Genotypic characterization was repeated using the eGenomics software (http://tools.egenomics.com/spaTypeTool.aspx). Throughout this manuscript, the spa-CC nomenclature will be accompanied by the corresponding multi-locus sequence typing (MLST) CC nomenclature, for easy reference.
Isolates that failed to produce amplicons twice or that produced amplicons of such low quality that reading was unreliable were characterized as nontypeable for the statistical analysis.

Confirmation of USA300 Genotype
All isolates assigned to spa-CC008 (MLST CC8) were analyzed for the presence of both Panton-Valentine Leukocidin (PVL)and Arginine Catabolic Mobile Element (ACME)-encoding genes [14]. For the statistical analysis, isolates that were methicillin resistant, PVL positive, and ACME positive and belonged to spa-CC008 (MLST CC8) were characterized as consistent with the USA300 type [15][16][17].

Statistical Analysis
Each variable for demographics, patient clinical characteristics, spa-CC, and USA300 was summarized using descriptive statistics and calculated separately for each calendar year and overall. An additional analysis further stratified the spa-CC by methicillin resistance. Frequencies and percentages were used for categorical variables and medians and quartiles were generated for continuous measures. Plots of proportions by calendar year present these data for the clinical measures and bacterial genotypic characteristics. To test for secular trends in these parameters, linear regression models were fit for the calendar year means and proportions, with year as the independent variable. Multivariable logistic regression models assessed the associations of clinical characteristics and genotypes with any metastatic complications and abscesses, embolizations, and persistent bacteremia separately. Analyses were performed with SAS 9.4 (SAS Institute, Cary, NC) and Spotfire SPlus 8.1 (TIBCO, Palo Alto, CA).

Clinical Ascertainment Bias
To examine whether any apparent increases in the rates of metastatic infections were due to an increasing use of imaging studies during the study period, we compared rates of use of imaging studies in a randomly selected 10% subgroup of study patients for each year, from 1995 to 2015 (Supplementary Material).

Study Population
Between 1995 and 2015, 2423 unique patients were enrolled in SABG-PCS. We excluded 53 from the study because the initial bloodstream isolate could not be retrieved. Another 22 were excluded from the analysis due to inconsistent data in the outcome variable. A mean of 112 eligible patients were enrolled per year (range, 72-135 patients). The annual distribution of the 2348 eligible patients is shown in Supplementary Figure 1.
Baseline demographic and clinical characteristics of the study population are summarized in Table 1. The median age of patients was 60 years, 56.6% were male, and 61% were Caucasian. The most common comorbidities were diabetes mellitus (38.6%), end-stage renal disease (ESRD; 21.1%), malignancy (hematologic or solid tumor; 19.8 %), and recent corticosteroid treatment (19.4%). The study cohort exhibited significant trends during the study period, including increases in rates of diabetes (annual increase 0.7%; P = .017), malignancy (0.6%; P = .012), transplantation (0.3%; P = .015), corticosteroid use (0.5%; P = .032), and rheumatoid arthritis (0.1%; P = .03) and decreases in rates of hemodialysis dependence (annual decline 0.5%; P = .023). The majority of patients (1273; 54.2%) carried some type of implantable device, such as a central vascular catheter, cardiac device, vascular graft, or orthopedic implant. The rate of patients with an implantable foreign body increased significantly, from 40% in 1995 to 54.7% in 2015 (P = .002; Figure 1). This increase was mainly driven by an increase in the use of cardiac devices and orthopedic implants. Patients with prosthetic devices were more likely to develop persistent SAB (P = .0002).

Potential for Ascertainment Bias
No significant difference was noted in the number of imaging studies ordered on a 10%, randomly selected sample of study patients, with the exception of transthoracic echocardiograms. From 1995 to 2015, there was an annual increase of 1.6% in transthoracic echocardiogram use (P < .0001; Supplementary Table 2).

Phenotypic and Genotypic Studies of Staphylococcus aureus Isolates
Among 2348 unique S. aureus isolates, 1222 (52%) were methicillin-susceptible (MSSA). The annual distribution of methicillin-resistant (MRSA) and MSSA isolates is shown in Supplementary Figure 1.
Among spa-CC008 (MLST CC8) isolates, a total of 168 (42.5%) were methicillin resistant and positive for both PVL and ACME, and were designated as USA300. USA300 emerged as a cause of SAB in our institution in 2003, and has increased in frequency thereafter, constituting 20.4% of all episodes of SAB in 2015 (P < .0001; Figure 3A and 3D).
In a multivariable analysis (   Includes vascular stents, vascular clips, biliary stents, urinary stents, draining tubes, percutaneous feeding tubes, intrathecal catheters, vena cava filters, peritoneal dialysis catheters, esophageal stents, meshes, penile implants, breast implants, tracheostomy tubes, spinal cord stimulation implants, drug delivery implants, and intra-aortic balloon pumps. j Calculated on Day 1 of SAB. k Score change, annually. l Calculated as part of the APACHE II score: 0, no severe organ system insufficiency or immune compromise; 2, severe organ system insufficiency or immune compromise and elective postoperative; 5, severe organ system insufficiency or immune compromise and nonoperative or emergency postoperative [10]. m Some patients had more than 1 complication. n Defined as persistently positive blood cultures for ≥5 days after initiation of appropriate antimicrobial therapy. o Includes cases of pneumonia, empyema, pericarditis, mycotic aneurysm, endophthalmitis, myositis, fasciitis, and urinary tract infections.  association was limited to USA300 variants, as the spa-CC008 clonal lineage outside of the USA300 subtype was not associated with metastatic complications. When USA300 was compared to all other MRSA types, the associations with metastatic complications, emboli, or persistent bacteremia were not significant (data not shown). This may be due to the loss of statistical power associated with a significant reduction in the sample size.
Other associations between strain characteristics and clinical phenotypes of SAB are shown in Supplementary Tables 5 and 6. No association was found between endocarditis or vertebral osteomyelitis and S. aureus genotypes.

DISCUSSION
This study confirms that the clinical presentation of SAB in the early 21st century differs from that of SAB encountered in previous decades. The morbidity of the patients has increased as the clinical presentation has become more complicated. This could be the result of changes in the characteristics of the patient population, in the clones causing SAB, or both.
This study documents that the clones causing SAB at a single center differ significantly over time, and that these changes can have clinical consequences. Using a prospective cohort that spans 2 decades and includes over 2300 patients with SAB, the study demonstrated that spa-CC008 has progressively replaced multiple other previously predominant clones at our institution. After adjusting for patient comorbid conditions, USA300 was independently associated with metastatic complications in patients with SAB.
First identified in the late 1990s [18], USA300 rapidly emerged to become the dominant community-acquired MRSA in North America [19,20]. Although originally a cause of skin and soft issue infections, USA300 has progressively become endemic in nosocomial settings, and is now the most common MRSA type isolated from bloodstream infections in parts of the United States [19,21,22]. The success and high virulence of USA300 is thought to be due, in part, to its acquisition of various mobile genetic elements: namely, SCCmec type IVa, carrying the mecA gene; S. aureus pathogenicity island 5, containing sek (gene encoding enterotoxin K) and seq (gene encoding enterotoxin Q); phage phiSA2, carrying the PVL genes; and ACME type I, carrying the arginine deiminase cluster [arc] and speG gene [23][24][25][26]. USA300 has also been shown to exhibit increased expression of α-toxin and phenol-soluble modulins (cytolytic toxins), due to overexpression of the global transcriptional regulators agr and sae [27]. This extraordinary array of virulence factors and traits of enhanced invasiveness could explain the striking association of USA300 with metastatic complications of SAB shown in this study, and is consistent with the findings of other recent investigations [28].
Patient characteristics also changed during the study period. Patients had higher rates of comorbid conditions, including diabetes mellitus, cancer, transplantation, rheumatoid arthritis, and corticosteroid use. This may be the consequence of either the increasing prevalence of these conditions, the increasing life expectancy of patients suffering from these conditions [29], and/or the increasing use of the advances offered by modern medicine that predispose patients to SAB. Overall, patients with SAB were significantly sicker in 2015 than they were in 1995, by any measure. The only notable exception was in patients with ESRD, which was less commonly identified among patients with SAB in 2015 versus 1995. Plausible explanations could be the advances in vascular access methods, as permanent central venous catheters have increasingly been replaced by arteriovenous fistulas or vascular grafts; improvements in infection control practices in this population; or fewer patients with ESRD served in our institution over the 21-year study period.
A critical emerging characteristic of study patients during the study was the increased presence of prosthetic devices. By the end of the enrollment period, over half of study patients had at least 1 indwelling prosthetic device. The role of prosthetic devices-including pacemakers/defibrillators, prosthetic valves, endovascular prostheses, and arthoplasties-in the prevalence and changing epidemiology of SAB have been well documented [30][31][32].
Our study confirms that SAB remains primarily a consequence of health-care contact. Almost 88% of patients had CA-HCA infections in 2015. This percentage has significantly dropped, from ~97% in 1995. By contrast, 80% of patients developed their SAB outside the hospital in 2015, versus 50% in 1995. These shifts likely reflect the substantive changes in health-care delivery over the past 2 decades, with increasing outpatient care delivery and increasing use of indwelling prosthetic devices, and have been previously described among patients with S. aureus endocarditis [33]. In our study, true community-acquired infections significantly increased, from 3.2% in 1995 to 12% in 2015. This trend has been documented in the United States since the early 2000s [34], has been validated in all types of health-care settings, and has been linked to the emergence of community-associated MRSA: notably, the USA300 strain type [35][36][37][38].
This study has significant strengths. To perform this study, we used a unique resource, the SABG-PCS Biorepository, which has been maintained for 24 years to date. It contains robust clinical data and bacterial bloodstream isolates from more than 2700 prospectively ascertained patients, setting it among the largest resources of its kind in the world. To genotype these isolates, we developed and validated an in-house, high-throughput, DNA isolation method. Finally, we repeated our genotype-phenotype associations using 2 different software platforms for CC assignment: eGenomics and Ridom StaphType. Collectively, these design strengths allowed us to conclude that the results of the study are consistent, irrespective of the typing tool used.
Our study has limitations. It is possible that, as the abstractors have changed over the course of 2 decades, inconsistencies in data capture may have occurred. To minimize potential biases, the same standardized case report form, with the same data abstraction guidelines and definitions, was used throughout. In addition, random validation was performed by a single investigator (F. R.) for missing data or inconsistencies, and data were rectified as needed. Our study is subject to referral bias. It is possible that results deriving from 1 academic referral center may not be generalizable to other medical centers in the country. For example, a study of 5 geographically dispersed centers in the United States showed marked decreases in USA300 prevalence in some locations, like Los Angeles, and significant increases in others, like New York. It is unclear whether these patterns simply represent local trends or different phases in the expansion and contraction of USA300. In our center, which was 1 of the 5 academic medical centers of that study, USA300 accounted for 35% of the 80 MRSA isolates examined [35].
The management of SAB has changed over the past 2 decades that this study has enrolled patients, thereby introducing Results for each variable are adjusted for all other variables shown. ORs marked in bold are significant.
potential biases. However, we found no evidence for any biases in the diagnosis of abscesses, osteoarticular infections, or emboli when we evaluated the use of imaging practices in a 10% subset of study patients. Finally, we did not perform pulsed-field gel electrophoresis to unequivocally distinguish the USA300 strain type within our collection of isolates, but used a consensus definition (spa-CC008, methicillin resistance, PVL positivity, and ACME positivity) that has been previously validated in the literature to accurately identify this type [17,26] being aware that USA300 strains that are PVL and/or ACME negative have emerged but are still quite rare [38,39].
In conclusion, shifts in bacterial genetics can have significant consequences in the infections that they cause. Interestingly, most of these shifts in bacterial genotypes and severities of infections are likely to go unnoticed by the busy practitioner. Systematic approaches for bacterial genotyping and recording rates of specific complications will likely have a bidirectional benefit: improving clinical management and offering an effective means of detecting subtle shifts in clinical syndromes, which could indicate the emergence of a new, hypervirulent bacterial clone.

Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes
Disclaimer. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health (NIH).