Clinical Subphenotypes of Staphylococcus aureus Bacteremia

Abstract

Background: Staphylococcus aureus bacteremia (SAB) is a clinically heterogeneous disease. The ability to identify subgroups of patients with shared traits (subphenotypes) is an unmet need to allow patient stratification for clinical management and research. We aimed to test the hypothesis that clinically relevant subphenotypes can be reproducibly identified among patients with SAB.

Methods: We studied 3 cohorts of adults with monomicrobial SAB: a UK retrospective observational study (Edinburgh cohort, n = 458), the UK ARREST trial (n = 758), and the Spanish SAFO trial (n = 214). Latent class analysis was used to identify subphenotypes using routinely collected clinical data without considering outcomes. Mortality and microbiologic outcomes were then compared between subphenotypes.

Results: Included patients had predominantly methicillin-susceptible SAB (1366 of 1430, 95.5%). We identified 5 distinct, reproducible clinical subphenotypes: (A) SAB associated with older age and comorbidity, (B) nosocomial intravenous catheter-associated SAB in younger people without comorbidity, (C) community-acquired metastatic SAB, (D) SAB associated with chronic kidney disease, and (E) SAB associated with injection drug use. Survival and microbiologic outcomes differed between the subphenotypes. Mortality was highest in subphenotype A and lowest in subphenotypes B and E. Microbiologic outcomes were worse in subphenotype C. In a secondary analysis of the ARREST trial, adjunctive rifampicin was associated with increased mortality in subphenotype B and improved microbiologic outcomes in subphenotype C.

Conclusions: We have identified reproducible and clinically relevant subphenotypes within SAB and provide proof of principle of differential treatment effects. Through clinical trial enrichment and patient stratification, these subphenotypes could contribute to a personalized medicine approach to SAB.

Keywords: Staphylococcus aureus; adjunctive treatment; bacteraemia; patient stratification; subphenotypes.

Figure 1.

Comparison of class-defining variables for SAB subphenotypes. Vertical bars show the number of patients assigned to each subphenotype. Cells are shaded according to row z score (ie, comparing subphenotypes) except “Acquisition” and “Source of SAB,” where shading is by column z score (ie, comparing within each subphenotype). Intensity of red shading reflects a more positive z score (ie, above the mean), and intensity of blue shading reflects a more negative z score (ie, below the mean). In the SAFO trial, people with Child-Pugh C liver cirrhosis or MRSA infection and people who inject drugs were not recruited. Abbreviations: IV, intravenous; MRSA, methicillin-resistant Staphylococcus aureus; SAB, Staphylococcus aureus bacteremia; SSTI, skin or soft tissue infection.

Figure 2.

Comparison of outcomes between Staphylococcus aureus bacteremia (SAB) subphenotypes. Comparison of all-cause 84-day mortality (A), persistent or recurrent bacteremia in the Edinburgh retrospective observational cohort (B), and composite microbiologic failure in the ARREST trial placebo arm (C). Bars represent z scores comparing the outcome between subphenotypes within the same cohort. Differences in the proportion of patients with each outcome between subphenotypes were compared using the Fisher exact test or χ² test.

Figure 3.

Effect of adjunctive rifampicin in Staphylococcus aureus bacteremia (SAB) subphenotypes. Comparison of outcomes of patients randomized to placebo or adjunctive rifampicin when SAB subphenotypes were considered separately. Treatment outcomes within each subphenotype were compared using the Fisher exact test. Because 2 comparisons were made within each subphenotype, the significance level was set at 0.025 (α=0.05, n = 2).