We present the largest cohort of HIV patients with CA-MRSA bacteremia and endocarditis documented to date. These data suggest a high prevalence of CA-MRSA among HIV patients presenting with S. aureus bacteremia at our institution and that CA-MRSA was significantly associated with the development of endocarditis compared to non-CA-MRSA. Furthermore, although not significant, patients with CA-MRSA also appeared to be at increased risk of death compared to patients with other S. aureus bloodstream infections.
Kempker et al. presented a large cohort of CA-MRSA bacteremia cases from eight hospitals over 3 years in the Atlanta, GA metropolitan area.(10) They had 414 episodes of CA-MRSA bacteremia of which 63 (16.2%) occurred in HIV positive patients and only 17 (4.1%) were associated with endocarditis. They observed greater in-hospital mortality (17%) compared to 8% in our study, however similar to the patients in our cohort, they observed more deep abscesses in their CA-MRSA group compared to the non-CA-MRSA group . Wang et al. performed a study of adult patients with S. aureus bacteremia in Taiwan, however, only 3 (10%) of their CA-MRSA bacteremia episodes had endocarditis, and none were HIV positive. On their univariate analysis, women appear to have a two-fold increase of dying with CA-MRSA bacteremia, a similar finding to our study . Several other papers on CA-MRSA bacteremia have had very few cases of endocarditis most of which appear to be associated with deep abscesses and pulmonary disease [31–34]. Kreisel et al evaluated CA-MRSA bacteremia and severe sepsis at 4 Veteran Affairs (VA) centers, which includes the Baltimore VA medical center and demonstrated that CA-MRSA was associated with more severe sepsis (adjusted relative risk 1.82, confidence intervals 1.16-2.87, p = 0.01) . They also noted that there was a greater propensity to ISU and African American patients having these severe infections, which is consistent with our population [35, 36].
Another major difference between our study and previously published work is the antibiotic susceptibility profile of our CA-MRSA strains compared to other studies with high levels of trimethoprim-sulfamethoxazole (TMP-STX) resistance, which was 32% of all isolates and 48% of the CA-MRSA isolates. This compares to ranges of 0-13% in other published studies [30, 32–34]. TMP-STX resistance could have several possible explanations given the patient population examined; 1) As the majority of the patients had a history of intravenous drug usage, it is possible they had more abscesses in the past and had been treated with TMP-STX several times or 2) the majority of the patients had CD4 < 200 cells/mm3 and were most likely taking TMP-STX for prevention of pneumocystis pneumonia. The second explanation appears more likely as our clindamycin resistance was similar to other studies [30, 32–34].
Our data compare similarly to data published from Detroit, which also has high rates of CA-MRSA. CA-MRSA was responsible for 35% of bloodstream infections in their series, but there were only 3% HIV infected patients in their cohort . Lalani et al. during a multicenter clinical trial that the reported a prevalence of CA-MRSA was 12% among 230 S. aureus isolates and 31% among the 88 typed MRSA isolates in a multicenter clinical trial.(28) Incidence of endocarditis was 23%. However, in contrast to our study, the Lalani et al. study population had a low prevalence of HIV/AIDS of approximately 3% and thus a thorough comparison to the patients in our study is not possible. Gebo et al. assessed risk factors and outcomes of 58 cases of infective endocarditis among HIV-infected patients and identified intravenous drug use, low (<50 cells/mm3) CD4 counts, and high (HIV RNA >100,000 copies/ML) viral load as independent risk factors for endocarditis. (3) Although S. aureus was the most prevalent organism (69% of cases) in their study, only 28% of S. aureus isolates were MRSA, which also complicates comparison to our data, which focused on differences in risk factors and outcomes between CA-MRSA and non-CA-MRSA.
We observed relatively low mortality (~8%) despite that greater than 20% of patients had a CD4 count of <200 cells/mm3 and less than 30% were receiving ART. The low rate of ART use in our population with high ISU reflects the difficulty in obtaining compliance in this population. There have been several studies on the difficulties in getting the HIV infected ISU patients to establish and maintain antiretroviral usage in the city of Baltimore. It has been suggested that ISU infected patients with HIV who were not receiving ART tended to be active drug users without clinical disease . It may explain why there was higher mortality in women in our study with the sex for illicit drug trade delaying the presentation for care .
Our results contrast previous studies, which observed mortality from 18% to 52% [3, 5]. However, the associations between ISU, right-sided endocarditis and low relative mortality have been previously described, even among HIV patients . Recall that more than 80% of our patients with endocarditis had right-sided disease only. Furthermore, as has been previously reported, the criteria used to measure mortality often vary in studies of infectious disease outcomes . For example, the 52% mortality reported by Gebo et al. included all deaths within one year, some of which were likely not attributable to the endocarditis episode . The low mortality in this study relative to the potential seriousness of the underlying infection, suggests that although CA-MRSA was significantly associated with endocarditis, this may be more related to risk associated with intravenous drug use and other outcomes such as mortality may still not be as severe as among other endocarditis patients [42, 43]. Furthermore, other studies have suggested that HIV infection was not significantly associated with mortality among patients with infective endocarditis in which 65% were caused by S. aureus, 38% of patients were HIV-positive and 16% mortality was observed .
The retrospective nature of the study limited which data were able to be collected and the availability of S. aureus isolates. Furthermore, we used a molecular definition of USA300 MRSA to define CA-MRSA. While USA300 remains the predominant CA-MRSA, we likely underestimated the prevalence of CA-MRSA in this study. Lastly, given the high prevalence of CA-MRSA, ISU, and HIV in the study population and in Baltimore City in general, the generalizability of these data to other patient populations outside of urban, tertiary-care settings should be performed with caution.