RESUMEN
BACKGROUND: We aimed to test the hypothesis that development of metastatic infection represents a distinct clinical endpoint from death due to SAB. METHODS: We conducted a retrospective observational study of adults with SAB between 20/12/2019 and 23/08/2022 (n=464). Simple logistic regression, odds ratios, and z-scores were used to compare host, clinical and microbiologic features. RESULTS: Co-occurrence of attributable mortality and metastatic infection was infrequent. Charlson Comorbidity Index and age were strongly associated with attributable mortality, but not metastatic infection. We compared patients with fatal SAB (without clinically-apparent metastatic complications, 14·4% of cohort), metastatic SAB (without attributable mortality, 22·2%), neither complication (56·7%), and overlapping fatal/metastatic SAB (6·7%). Compared to SAB without complications, fatal SAB was specifically associated with older age and multi-morbidity. Metastatic SAB was specifically associated with community acquisition, persistent fever, persistent bacteraemia, and recurrence. Endocarditis was over-represented in the fatal/metastatic SAB overlap group, which shared patient characteristics with fatal SAB. In contrast to other (predominantly musculoskeletal) metastatic complications, endocarditis was associated with increased mortality, with death occurring in older multi-morbid patients later after SAB onset. CONCLUSIONS: Patients with SAB experience distinct clinical endpoints: (i) early death, associated with multi-morbidity and age; (ii) metastatic (predominantly musculoskeletal) SAB; (iii) endocarditis, associated with late death occurring in older people with multi-morbidity, and (iv) bacteraemia without complications. These distinctions could be important for selecting appropriate outcomes in clinical trials: different interventions might be required to reduce mortality vs. improve clinical response in patients with metastatic SAB.
RESUMEN
BACKGROUND: Staphylococcus aureus bacteraemia (SAB) is a clinically heterogeneous disease. The ability to identify sub-groups of patients with shared traits (sub-phenotypes) is an unmet need that could allow patient stratification for clinical management and research. We aimed to test the hypothesis that clinically-relevant sub-phenotypes can be reproducibly identified amongst patients with SAB. METHODS: We studied three cohorts of hospitalised adults with monomicrobial SAB: a UK retrospective observational study (Edinburgh cohort, n=458), the UK ARREST randomised trial (n=758), and the Spanish SAFO randomised trial (n=214). Latent class analysis was used to identify sub-phenotypes using routinely-collected clinical data, without considering outcomes. Mortality and microbiologic outcomes were then compared between sub-phenotypes. RESULTS: Included patients had predominantly methicillin-susceptible SAB (1366/1430,95.5%). We identified five distinct, reproducible clinical sub-phenotypes: (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 sub-phenotypes. 84-day mortality was highest in sub-phenotype A, and lowest in B and E. Microbiologic outcomes were worse in sub-phenotype C. In a secondary analysis of the ARREST trial, adjunctive rifampicin was associated with increased 84-day mortality in sub-phenotype B and improved microbiologic outcomes in sub-phenotype C. CONCLUSIONS: We have identified reproducible and clinically-relevant sub-phenotypes within SAB, and provide proof-of-principle of differential treatment effects. Through clinical trial enrichment and patient stratification, these sub-phenotypes could contribute to a personalised medicine approach to SAB.