Comparing right- and left sided injection-drug related infective endocarditis

The aim of the study was to compare background characteristics, microbiology and outcome of patients with right-sided and left-sided intravenous drug use (IDU) associated infective endocarditis (IE). A nationwide retrospective study using the Swedish Registry on Infective Endocarditis between 2008 and 2019 was conducted. A total of 586 people with IDU-IE were identified and divided into left-sided (n = 204) and right-sided (n = 382) IE. Descriptive statistics, Cox-regression and Kaplan–Meier survival estimates were used. The mean age of patients in the left-sided group was 46 years compared to 35 years in the right-sided group, p < 0.001. Left-sided IE had a higher proportion of females. Staphylococcus aureus was the causative pathogen in 48% of cases in the left-sided group compared to 88% in the right-sided group. Unadjusted and adjusted long-term survival was better in right-sided IE compared to left-sided IE. Independent predictors of long-term mortality were increasing age, end-stage renal disease, nosocomial infection, brain emboli and left-sided IE. Left-sided IE was common in people with IDU but the proportion of females with left-sided IE was low. S. aureus was twice as common in right-sided IE compared to left-sided IE, and the long-term prognosis of right sided IDU-associated IE was better compared to left-sided IE despite the fact that few were operated.

ists. The study period was from 2008 to 2019. During this period a total of 5969 episodes of IE were entered into the registry. IE was defined as definitive IE based on the modified Duke's criteria 14 . Patients who did not have a Duke's criteria classification in the database were classified within the present study based on the various clinical variables available in the database that constitute the Duke's major and minor criteria. IDU-status was based on the physician's assessment using either direct information from the patients or medical records. As seen in Fig. 1, following the exclusion of non-definitive IE and non-IDU IE a total of 656 cases of definitive IDU-IE were identified. In order to compare right-sided versus left-sided IE, we excluded patients with IE affecting both sides of the heart (n = 29), those with an unspecified or where no valve was affected (n = 41). The study population of 586 patients was divided into right-sided (n = 382) and left-sided (n = 204) IE. The primary endpoint was long-term mortality, short-term mortality was defined as ≤ 30 days. Other variables of interest were background characteristics, microbiology, IE complications including surgery and short-term mortality.
The study protocol was approved by the Swedish Ethical Review Authority (Registration number 2019-03549). As this was a retrospective study, individual consent was waivered. All methods were performed in accordance with the relevant guidelines and regulations.
Follow-up. Follow-up on survival was done through the National Population Registry. Follow-up time was defined as the period between hospital admission and the current or latest status of the patient. Follow-up was completed on 8th of October 2019. The median follow-up time was 4.1 years (Interquartile range 1.9-6.8 years) and included a total of 2629 patient years. Statistical analysis. Categorical variables were presented as number and percentages. Continuous variables were normally presented as mean ± standard deviation if they were normally distributed while non-normally distributed variables were presented as median with interquartile range. Categorical variables were compared using the Chi-square test or the Fisher's exact test if at least one of the cells had an expected count of less than five. Continuous normally distributed variables were compared with Student's t-test while non-normally distributed variables were compared with Mann-Whitney U test. Predictors of long-term mortality were analyzed using Cox-regression. Univariable cox-regression was first calculated on variables that are known risk factors and those that were deemed important in this study. Variables that had a p-value < 0.10 were then included in the multivariate Cox-regression. Unadjusted survival was estimated using the Kaplan-Meier method Log-  Abscess formation occurred in 15% of the left-sided IE cases and 0.8% of the right-sided IE cases (p < 0.001). The total number of episodes with abscess that occurred in left-sided IE was 30, whereof 13 also had prosthetic valve IE. Emboli to the brain were more common in left-sided IE whereas emboli to the lungs were more common in right-sided IE. Patients with left-sided IE underwent surgery more often compared to right-sided IE (42% and 5.5%, respectively).
Microbiology. The causative pathogen for both left-sided and right-sided IDU-IE is summarized in Table 2 and Fig. 2. In right-sided IE, Staphylococcus aureus compromised 85% of the cases compared to left-sided IE where it compromised 46% of the cases (p < 0.001). Alpha-hemolytic streptococci and Enterococci were more common in left-sided IE than in right-sided IE, 14% and 3.7% vs. 24% and 3.1% respectively (p < 0.001). No difference was found for other bacteria which caused relatively few IE episodes.
Long-term survival and predictors of long-term mortality. The unadjusted survival rates after 1-year and 5-year follow-up were 85% (95% confidence interval [CI] 80% to 90%) and 55% (95% CI 47% to 63%) for left-sided IE, respectively. The survival rates for right-sided IE in the same follow-up time were 97% (95% CI 95% to 99%) and 84% (95% CI 80% to 88%), respectively. The long-term survival was thus significantly worse in   (Fig. 3a, log rank p < 0.001). Predictors of long-term mortality in people with IDU-IE are presented in Table 3. Age, end-stage renal disease, nosocomial infection, emboli to the brain and left-sided IE were independent predictors of long-term mortality in IDU-IE. The estimated survival adjusted for age, end-stage renal disease, nosocomial infection, and brain emboli is shown in Fig. 3b.

Discussion
The main findings in this study are that more than one-third of people with IDU had left-sided IE, the proportion of female patients with left-sided IE was much lower than the proportion of female patients with right-sided IE, and that the long-term prognosis of right sided IDU-associated IE was better compared to left-sided IE despite the fact that only 5.5% were operated. The current understanding of the pathophysiology of IE is that both bacteremia and endothelial damage are required for bacteria to attach to endothelium and multiply 15 . In right-sided IDU-associated IE, both requirements are met when injected particles damage the right-sided valves and when skin bacteria are introduced in the blood stream 16 . Particles injected into the blood are less likely to damage left-sided valves as they are more likely to get trapped in the lungs. Thus, left-sided IE is more likely to develop in the presence of underlying valve diseases that facilitate endothelial damage and the prevalence of degenerative valve disease increases with advanced age 17 . Epidemiological studies show that people with IDU in high-income countries are significantly older than those in low-income countries 18 . Furthermore, the mean age of people with IDU has been increasing in recent years 19 , which may further shift the trend towards left-sided IDU-associated IE. Our result show that more than a third of all IDU-associated IE reported to SRIE had a left-sided infection. The high frequency of leftsided involvement has previously been documented by Mathew et al., where the different sites had approximately equal frequency 20 . The mean age of the group with left-sided IE was 11 years higher than that of the right-sided group, which may reflect a higher burden of degenerative valve disease in the left-sided group. The presence of prosthetic valves is also an important substrate for IE. The significantly higher proportion of prosthetic valve IE in left-sided IE likely reflects that prosthetic valves are more common in the left side of the heart.
Our study showed that less than a quarter of left-sided IDU-associated IE were female while the proportion of right-sided patients who were female was 40%. The overall proportion of female patients with IDU-associated IE has been reported to be between 44 and 51% 21,22 . Previous studies on IDU-associated IE have not stratified patients according to left or right sided location. The reason for the difference in sex distribution in our cohort is unknown. However, the notion that left-sided IDU-associated IE is more similar to non IDU-associated IE than   23 and degenerative mitral valve disease 17,24 . Many of the valve lesions that are most commonly operated, are also potential substrates for endothelial damage that in conjunction with IDU-associated bacteremia may cause IE. In addition to this, male patients are more likely to have cardiovascular disease at a younger age compared to their female counterparts 25 . The protective effects estrogen active until menopause could further explain the male predominance in valve surgery. Surgery was almost eight times more likely in left-sided IE compared to right-sided IE. We believe that the reason for this is multifactorial. First, the indications for surgery are different in right-sided IE and left-sided IE 12,13 . Invasiveness and abscess formation, which constitute surgical indication 12 , primarily occur on the left side of the heart. Furthermore, tricuspid valve insufficiency is better tolerated than both aortic valve and mitral valve insufficiencies. Indeed, isolated tricuspid valve valvectomy has been reported to be used as a treatment modality in patients that are not deemed suitable for tricuspid valve replacement with favorable results 26 .
S. aureus was the predominant causative agent in right-sided IE and other pathogens accounted for a minority of episodes. S. aureus is very common in cutaneous portal of entry and is more likely to occur with poor injection hygiene. It is therefore not surprising that vast majority of right-sided IDU-associated IE are caused by this bacterium and that it is also the most common bacterium in IDU-associated IE on the left side. In left-sided IE the causative pathogens resembled more what could be expected in non-IDU-associated IE in left-sided IE, where a large proportion of the cases are caused by alpha streptococci and enterococci 27 . One explanation may be the different virulence factors expressed by the bacteria where some pathogens prefer certain conditions that may differ between the heart valves depending on their virulence factors. Pressure, flow, structure of the valves as well as calcification status differs between the left side and right side of the heart. Calcium deposits in the aortic valve and mitral annulus increase with age but are usually not seen on the right side of the heart. Thus, it does not surprise that these factors may predispose for certain microorganisms such as S. aureus and enterococci to attach to the surface of different heart valves 28,29 .
Left-sided IE was associated with worse long-term outcome than right-sided IE and at five years, 45% had died in the group with left-sided IE compared to 15% in the right-sided group. This was despite the fact that patients with left-sided IE were operated much more frequently than patients with right-sided IE (42% vs 5.5%). One explanation may be that left-sided infections more often is complicated by severe heart failure, systemic embolization and uncontrolled infection whereas invasiveness is uncommon in right-sided involvement 30 and lesions in the low-pressure right side are better tolerated than lesions on the left side 31 . Indeed, in multivariable cox regression, left-sided infection was an independent predictor of long-term mortality, suggesting that it is more favorable for people with IDU to suffer a right-sided infection. Surgery was not an independent predictor www.nature.com/scientificreports/ of survival in this cohort. A recent meta-analysis compared IDU-associated IE with non-IDU-associated IE that underwent surgery and showed shorter long-term survival in IDU-associated IE which may signal that surgery can be problematic in people with IDU 32 . It is worth to highlight the importance of preventing return to substance use to improve survival in IDU-IE treated with surgery 33 , as this is the predominant cause of death in people with IDU-IE undergoing surgery 34 .

Strengths and limitations.
The current study includes patients reported to a nationwide registry. The patients are reported from all hospitals that have an infectious disease department, which decreases the risk of a biased patient selection that is usually seen in reports from tertiary care facilities. However, as reporting is not mandatory, the registry does not contain all IE episodes in Sweden. The retrospective nature of the study has inherent risks of biases. The IE registry does not contain more data on the nature of IDU such as what type of substances were used, and whether the patients continued IDU following surgery. There was no data regarding socioeconomic and demographic factors. People who inject drugs may also get endocarditis from non-IDU related causes such as blood stream contamination following dental surgery, this was not specified in the registry. Thus, our findings are applied to IDU-association.

Conclusion
This study adds important information on IDU-associated IE and the differences seen in right-sided and leftsided IE in this patient group. IDU-associated IE is not limited to the right side of the heart as more than a third of the patients had left-sided IE. S. aureus occurred twice as much in right-sided IE compared to left-sided IE in IDU-associated IE. Survival in right-sided IE is relatively high and few patients in this group undergo surgery.