Long-term trends in the incidence of peritoneal dialysis-related peritonitis disclose an increasing relevance of streptococcal infections: A longitudinal study

Background The selective impact of strategies for prevention of PD-related peritonitis (PDrP) may have modified, in the long term, the causal spectrum, clinical presentation and outcomes of these infections. Objectives To compare trends in the incidence of PDrP by different microorganisms during a 30-year period, with a particular focus on streptococcal infections. To analyze the clinical presentation and outcomes of these infections. Secondarily, to investigate how the isolation of different species of streptococci can influence the clinical course of PDrP by this genus of bacteria. Method Following a retrospective, observational design we investigated 1061 PDrP (1990–2019). We used joinpoint regression analysis to explore trends in the incidence of PDrP by different microorganisms, and compared the risk profile (Cox), clinical presentation and outcomes (logistic regression) of these infections. Main results Our data showed a progressive decline in the incidence of PDrP by staphylococci and Gram negative bacteria, while the absolute rates of streptococcal (average annual percent change +1.6%, 95% CI -0.1/+3.2) and polymicrobial (+1.8%, +0.1/+3.5) infections tended to increase, during the same period. Remarkably, streptococci were isolated in 58.6% of polymicrobial infections, and patients who suffered a streptococcal PDrP had a 35.8% chance of presenting at least one other infection by the same genus. The risk profile for streptococcal infections was comparable to that observed for PDrP overall. Streptococcal PDrP were associated with a severe initial inflammatory response, but their clinical course was generally nonaggressive thereafter. We did not observe a differential effect of different groups of streptococci on the clinical presentation or outcome of PDrP. Conclusions Time trends in the incidence of PDrP by different microorganisms have granted streptococci an increasing relevance as causative agents of these infections, during the last three decades. This behaviour suggests that current measures of prevention of PDrP may not be sufficiently effective, in the case of this genus of microorganisms.

Enter a financial disclosure statement that  This study was evaluated and approved by the joint Ethical Committee of A Coruña and Ferrol University Hospitals (code 2020/190).
Oral informed consent was requested and obtained from all the participants active at the time of this retrospectrive analysis . Data bases were fully anonymized for data processing. . We used joinpoint regression analysis to explore trends in the incidence of PDrP by different microorganisms, and compared the risk profile (Cox), clinical presentation and outcomes (logistic regression) of these infections.

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Main results: We disclosed significant trends to a decrease in the incidences of PDrP by staphylococci and Gram negative bacteria, while the absolute rates of streptococcal (average annual percent change +1,6%, 95% CI -0,1/+3,2) and polymicrobial (+1,8%, +0,1/+3,5) infections tended to increase, during the same period. Remarkably, streptococci were isolated in 58,6% of polymicrobial infections, and patients who suffered a streptococcal PDrP had a 35,8% chance of presenting at least one other infection by the same genus. The risk profile for streptococcal infections was comparable to that observed for PDrP overall. Streptococcal PDrP associated a severe initial inflammatory response, but their clinical course was generally nonaggressive thereafter. We did not observe a differential effect of different groups of Introduction Peritonitis (PDrP) represents one of the most feared complications of chronic Peritoneal Dialysis (PD), associating significant rates of mortality [1,2] and PD technique failure [3][4][5][6]. The incidence of PDrP declined markedly between 1985 and 1995, after the introduction of Y-set, double bag systems, but improvements have been slower thereafter [5,7], and rates as high as one episode every two patient-years are still considered acceptable by current standards [8]. Singular advances, including the introduction of low glucose degradation products-based (low-GDP) solutions, may have been beneficial, but their impact has been disappointing, in clinical terms [9]. On the contrary, comprehensive approaches to prevention, including continuous quality improvement strategies, have provided a renewed hope of progress [10], and may mark the route to an optimized control of these infections.
However, a selective impact of preventive measures, preferentially oriented to reduce the consequences of touch contamination and catheter-related infections, may have influenced the etiologic spectrum and the strategies of management of PDrP [8], because these measures do not protect evenly from infections by different microorganisms. In particular, streptococcal infections have been a subject of limited attention in the past, due to the perceptions that these infections are relatively infrequent (with reported relative incidences of 5-12% of all PDrP) and follow a relatively benign clinical course [13][14][15][16]. On the other hand, the taxonomy and nomenclature of the ubiquitous genus Streptococcus has been reviewed in the last years [17] and, aside from single case reports, information on the compared aggressiveness of PDrP by different members of this family of microorganisms is scarce [18]. Finally, streptococci may be a common component of polymicrobial PDrP, but the real incidence of this circumstance and its clinical significance are also unclear.
Following an observational, retrospective design, we have investigated time trends in the incidence of PDrP in our centre during a period of 30 years. Our main interest was to 4 pursue a clinical perception that advances in the prevention and management of PDrP could have modified the clinical scenery of PD-related peritonitis, conferring streptococci a progressively dominant role, with potential consequences for the clinical presentation, management and outcomes of these infections.

Method General design
Following an observational, retrospective design, we analyzed trends in the incidence and outcomes of PDrP diagnosed in our centre between January 1990 and December 2019.

Study population
We recruited for this study all patients starting PD in our unit between January 1, 1990 and October 1, 2019. Follow-up was closed by December 31, 2019. We excluded from analysis patients <18 years of age, those with a follow-up on PD <1 month, and those with inadequate clinical records. No patient denied consent for participation.

Study variables and strategy of analysis
Our interest was addressed to episodes of PDrP occurring during the study period. In particular, we focused on streptococcal infections, and used PDrP of other causes as control groups. For this purpose, we categorized the following study groups: 1) Streptococcus spp.

5) Polymicrobial
We used joinpoint regression analysis to disclose time trends in the overall and bycausative agent rates of incidence of PDrP, calculating the annual percent change (APC), and the average annual percent change (AAPC) for the whole study period [21]. The risk profile for We used the SPSS 19.0 software for general data management and statistic analyses.

Overview and time trends
The study population included 878 patients, after excluding 25 other individuals who did not meet the inclusion criteria. Their main baseline characteristics are presented in Table 1.   Table 2). Streptococci were less frequently identified in the latter group between 1990 and 1999 (38,9%), but kept consistently above 60% after 2000 (p=0,016). Table 2 Microorganisms isolated in polymicrobial peritonitis

Enterococcus spp 23
Enterobacteriaceae and nonfermenting Gram negative bacteria 83 Other 8 Table 3 displays the absolute and relative incidences of PDrP during the different study periods. Joinpoint analyses disclosed a progressive decrease in the incidence of PDrP overall.
This tendence was supported by a declining incidence of infections by staphylococci and GNB (Table 3, Figure 1), while an opposite trend was observed for polymicrobial and streptococcal infections (not significant in the latter case) (Figure 1)
On the other hand, the risk of suffering streptococcal PDrP was univariately linked to older age (p=0,003), lower plasma albumin (p=0,041), higher body mass index (p=0,021), lower blood haemoglobin (p=0,047) and automated PD rather than CAPD (p=0,050). PD vintage was a univariate predictor of the risk of streptococcal (p=0,004), but not overall PDrP (p=0,124)(other variables displayed in Table 1 not significant). Table 4 shows the multivariate (Cox) risk profile for PDrP overall, and streptococcal PDrP. Both models were quite similar, except for a clearly stronger impact of PD vintage on the risk of streptococcal infection, and a more marked association of baseline GFR with the later risk of PDrP overall.

Clinical presentation according to causative microorganism
The main clinical characteristics of PDrP by different causative microorganisms are displayed in Table 5. We did not record a single instance of streptococcal catheter-related PDrP between 1990 and 2019.

Clinical outcomes
The main clinical outcomes of PDrP are presented in Table 6. Remarkably, streptococcal PDrP carried the highest probability of uneventful continuation of PD. Catheter removal was less frequent than in any other type of PDrP, except culture-negative PDrP (p=0,04 versus CNSt). The risk of relapse, in opposition to the case of repeat infection (see above) was low, and treatment failure was less frequent than in any other group, although the difference with CNSt (p=0,072) and negative-culture PDrP (p=0,11) did not reach statistical significance.   Table 7 presents the results of multivariate analysis for the main outcomes.
Streptococcal PDrP carried a better prognosis than infections by SAu or GNB, and similar to CNSt and culture-negative infections.

Comparisons of PDrP by different species of streptococci
We compared the clinical presentation and outcomes of four different subgroups of streptococci, namely mitis/oralis/sanguis/gordoni (n=92), salivarius/vestibularis (n=85), other viridans (n=46) and non-viridans (n=12). We were unable to disclose any significant differences or trends, regarding the variables displayed in Tables 5 and 6 (data not presented).

Clinical significance of the isolation of streptococci in polymicrobial PI
As previously stated, at least one streptococcal strain was isolated in 78 of the 133 polymicrobial PDrP included in the analysis. In 28 cases, polymicrobial PDrP was caused by 2 different strains of streptococci. Polymicrobial PDrP with presence of streptococci presented with marginally higher peritoneal cell counts (3501 vs 2764 cells/mm 3 , p=0,092) and % of neutrophils (81,9 vs 75,9%, p=0,031), than infections without participation of these bacteria.
We did not detect any other clinical difference, between these subsets (data not presented).

Discussion
Our results provide clues to understand how the causative spectrum of PDrP may have evolved, during the last three decades. The systematic implementation of measures of prevention, essentially oriented to reduce touch contamination and catheter-related PDrP [8,22] resulted in a progressive decline of the incidence of infections by staphylococci and GNB (  [14,15]. The oral cavity is a subject of particular concern as a source of streptococcal PDrP, due to a high degree of colonization by this family of bacteria [25] and to a potential for both hematogenous spread and direct contamination, particularly if face masks are used inappropriately during the PD exchange [8]. On the other hand, the high proportion of polymicrobial PDrP with isolation of streptococci [13,14,16] suggests that the lower gastrointestinal tract may be another significant source of infections by these bacteria. Unfortunately, we were unable to create a specific risk profile for streptococcal PDrP (Table 4), partly due to the nonavailability of potentially relevant variables, including bucodental care, adherence to the use of face masks during the PD changes or intestinal disorders. Older age, hypoalbuminemia and overweight associated a higher risk of streptococcal PDrP, but this predictive model lacked specificity, because these factors were also linked to the general risk of PDrP [26][27][28]. This inability to establish a specific predictive model for streptococcal PDrP was also observed in the best powered study so far [14].
Our results disclosed a higher relative incidence of streptococci in both monobacterial and polymicrobial PDrP than previous reports [11][12][13][14]16,[29][30][31][32]. However, most of the cited studies investigated patients started on PD in the 1990s', when the incidence of streptococcal 20 PDrP was also lower in our centre. In addition, local factors, including differences concerning PD practices and the global incidence of PDrP, could help to explain this discrepancy. For instance, irregular adherence to measures with a demonstrated efficacy to prevent staphylococcal PDrP [33] may associate a lower incidence of streptococcal infections [14,34,35]. The latter factor could also explain why only some studies [12,13,36] were able to detect the trend to an increase in the incidence of streptococcal PDrP, so clearly disclosed by our analysis. The exclusion by protocol of rampant enteric PDrP and the high proportion of polymicrobial peritonitis with isolation of streptococci may help to explain why, in our study, the observed outcomes of these infections were more benign than usually reported (Table   6) [29,31].
The present study agrees, for the most part, with previous reports on the clinical presentation and outcome of streptococcal PDrP (Tables 5 and 6). These infections have a relatively severe initial presentation, with a marked inflammatory response, as shown by the baseline peritoneal cell counts (Table 5), and a relatively slow resolution (Table 6) [13].
However, initial aggressiveness is usually followed by complete recovery with antibiotic therapy alone, with low rates of hospital admission, catheter removal, relapse, recurrence and, most importantly, hard outcomes (mortality, technique failure)( Table 7) [13][14][15][16]18]. Previous studies have shown that the time course of peritoneal cellularity, rather than the initial count, is the best marker of outcome of PDrP [37,38]. Our data also confirm that these bacteria are usually susceptible to common antibiotic regimes [39], as either vancomycin, betalactams or fluoroquinolones were able to control streptococcal PDrP (Table 5) [14,15]. Current ISPD recommendations endorse the use of ampicillin, for this purpose [8] but, in our unit, the success of the antimicrobials used for initial management of PDrP explains why ampicillin was not a usual choice.
Our results do not suggest particularities for any subgroup of streptococci, regarding clinical presentation or outcomes. Isolation of Streptococcus bovis has been linked to an 21 increased risk of colorectal cancer. We recorded 5 instances of this infection, and all followed an uneventful clinical course, in agreement with a recent report [18]. We neither observed any significant diference among polymicrobial PDrP with or without isolation of streptococci.
Previous studies have underlined that the isolation of GNB [31], intestinal anaerobics and Enterococcus faecium [40] may represent better markers of a complicated outcome of polymicrobial PDrP.
Our study suffers significant limitations, including a retrospective, single-centre design, which may overrate the effect of local factors. Some variables with a potential impact on the risk of streptococcal PDrP (e.g. oropharingeal disorders) could not be recorded. On the other side, significant strengths include the large time span, very appropriate to investigate time trends, and the high quality of our database, which permitted a complete analysis of the study population.
In  Conflict of interest statement: None to be reported