Efficiency of polymyxin B treatment against nosocomial infection: a systematic review and meta-analysis

Background Some cohort studies have explored the effects and safety of polymyxin B (PMB) in comparison to other antibiotics for the treatment of nosocomial infections, yielding inconsistent results. This systematic review aims to explore the effectiveness and safety of PMB and compared it with other antibiotics. Methods A systematic literature search was conducted in PubMed, Embase, the Cochrane Library, and Web of Science, searching specific terms to identify quantitative cohort studies or RCTs that compared the effects of PMB with other antibiotics in terms of their efficacy and safety. The Newcastle–Ottawa Scale (NOS) was conducted to evaluate the risk of bias of observational studies. Odds ratios with 95% confidence intervals were used for outcome assessment. We evaluated heterogeneity using the I2 test. Results A total of 22 observational trials were included in the analysis. The PMB group had a higher mortality rate compared to the control group (odds ratio: 1.84, 95% CI: 1.36–2.50, p<0.00001, I2 = 73%). while, the ceftazidime-avibactam group demonstrated a distinct advantage with lower mortality rates, despite still exhibiting high heterogeneity (odds ratio 2.73, 95% confidence interval 1.59–4.69; p = 0.0003; I2 = 53%). Additionally, the PMB group had a lower nephrotoxicity rate compared to the colistin group but exhibited high heterogeneity in the results (odds ratio 0.58, 95% CI 0.36–0.93; p = 0.02; I2 = 73%). Conclusion In patients with nosocomial infections, PMB is not superior to other antibiotics in terms of mortality, specifically when compared to ceftazidime-avibactam. However, PMB demonstrated an advantage in terms of nephrotoxicity compared to colistin.


Background
In the past decade, nosocomial infections caused by multidrugresistant (MDR) bacteria have emerged as a significant cause of mortality and morbidity worldwide, particularly among critically ill patients (1)(2)(3)(4)(5)(6)(7)(8).Notably, limitation in antibiotics options against MDR bacterial infections became an extreme challenge in nosocomial settings in recent decades.Estimates suggest that in the United States alone between 5 and 10% of hospitalized patients admitted to acute-care hospitals may acquire nosocomial infections (3,4).
Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacteriaceae are the most common MDR bacteria causing fatal nosocomial infections, especially the strains producing extendedspectrum β-lactamases (ESBLs), New Delhi Metallo-beta-lactamase (NDM) and Carbapenemases as well.With a growing threat of MDR bacteria, however, there have been only a few new drugs developed in recent decades, such as tigecycline, ceftazidime-avibactam and eravacycline.As instead, some old antibiotics were reassessed for their potential effectiveness against MDR Gram-negative bacteria.Of them, colistin and polymyxin B (PMB) have gained a great of attention in the last decade (9,10).Significantly, they were reintroduced as salvage therapy for infections caused by MDR Gram-negative bacteria that did not respond to other treatments (11).Meanwhile, a controversy remains in physician's options for their clinical uses owing to the adverse events, especially with regard to the nephrotoxicity and the neurotoxicity.For instance, PMB demonstrated a lower renal toxicity than the comparable dose of colistin in several trials (12)(13)(14)(15)(16). Additionally, pharmacologically experimental data demonstrated that PMB achieved effective plasma concentrations more rapidly than colistin while administrating intravenously (17,18).Moreover, previous researches suggested that PMB showed a better efficacy in treatment of nosocomial infections caused by MDR Gram-negative bacteria including Acinetobacter baumannii, Pseudomonas aeruginosa, and CRO (12, 19).On the other hand, a few cohort studies showed that Colistin exhibited a lower mortality rate compared to PMB (12).In this systematic review and meta-analysis, therefore, we sought to study the effectiveness and safety of PMB in treatment of nosocomial infection caused by MDR bacteria comparing with colistin and other antibiotics as well.

Method
We conducted our systematic review and meta-analysis by using predefined protocol according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) reporting guideline (20).The review protocol was registered in PROSPERO (CRD42023446418).

Data sources and search strategy
A systematic literature search was conducted in PubMed, Embase, the Cochrane Library, and Web of Science from the beginning to August 1, 2023, to include clinical trials in patients of nosocomial infections associated with polymyxin B (PMB) and other antibiotics.The search only included English publications.The search terms were "Polymyxin B, " "Drug Resistance, Multiple, Bacterial, " "Aerosporin, " "mortality, " "Hospital-Acquired Pneumonia." In addition, the reference lists of relevant studies, systematic reviews, and meta-analysis were manually examined to identify any additional publications that could contribute to our analysis.For studies that did not report complete data on our dichotomous or continuous outcomes but were eligible for inclusion in the meta-analysis, we made efforts to contact the original investigators via email in order to acquire Supplementary Data.The search strategy was presented on Supplementary Table S1 (including PubMed database, Embase, Cochrane library and Web of science).

Study selection
Two of us (LP and XL) screened the titles and abstracts of all initially selected articles independently in accordance with the eligibility criteria.Then, the above two reviewers independently assessed full text of the selected articles from first step.Disagreements were resolved by reaching a consensus or seeking help from a third author (PM).We considered overall mortality as the primary outcome of our meta-analysis.Secondary outcomes were nephrotoxicity and microbiological eradiation.

Inclusion criteria
We considered studies to be following criteria:

Data extraction and study quality assessment
Data extraction was conducted by 2 reviewers independently (YH and XL) using a standardized form and discussed with another reviewer (PM).And the following date were extracted: the study characteristics (setting, sample size), participant characteristics (sex, age, duration of medication), specification of interventions (types of pathogenic bacteria, the control group).

Quality assessment
Two reviewers (LP and ZZ) independently conducted risk-ofbias assessments using the Newcastle-Ottawa Scale (NOS) to evaluate the observational studies included in the meta-analysis.The NOS tool assesses the risk of bias across three domains: selection, comparability, and outcome.The NOS criteria yield a quality score ranging from 0 to 9, with higher scores indicating better study quality (21,22).Disagreements between the two investigators were resolved through discussions involving the third investigator (LC).

Subgroup analysis
We performed subgroup analysis based on the risk of bias in the trials to optimize heterogeneity and obtain more reliable results.We also performed subgroup analysis to compare the effectiveness of PMB with different antibiotics, based on the type of antibiotic used in the control group.

Heterogeneity
Heterogeneity was assessed visually using forest plots and quantified with I 2 statistics, categorized as low (0-25%), moderate (26-50%), and high (>50%).If substantial heterogeneity was present and an adequate number of publications were available (n = 10), we aimed to investigate potential sources of heterogeneity through prespecified subgroup analysis outlined in the study protocol, such as restricting the analysis to high-quality studies.Additionally, a post hoc sensitivity analysis was conducted to assess heterogeneity after excluding studies that demonstrated a statistically significant benefit from placebo treatment.

Publication bias
We intended to assess publication bias by utilizing funnel plots for outcomes with data available from 10 or more studies.

Statistical analysis
All statistical analysis were performed with the use of RevMan software (version 5.3.3;The Cochrane Collaboration).The odds ratios with associated 95% confidence intervals were used to assess outcomes.Heterogeneity was assessed by the I 2 test.Random effects models are used for all analysis.We considered p values <0.05 statistically significant.Comparison adjusted funnel plots are used to evaluate the possibility of study.

Search results
We identified 832 citations and finally 22 studies met the inclusion criteria (Figure 1).The searcher strategies were shown in Supplementary Table S1.Nineteen studies reported data on the mortality, while 16 studies described nephrotoxicity.

Study characteristics
The study characteristics are displayed in Table 1.Colistin was reported as the control group antibiotic in 9 studies, tigecycline was reported in 7 studies, and ceftazidime-avibactam was reported in 7 studies.The pathogen species, the control group and the dosage and duration of antibiotic use among all studies were shown in Table 1.
According to NOS analysis, quality scores ranged from 6 to 9, and 7 trials were rated as low risk (equal 9 score), 7 trials were rated as moderate risk (equal 8 score), and 8 trials were rated as high risk (equal 7 score or less than 7 score).The risk bias of included studies was shown in Table 1 and Supplementary Table S2.Visual analysis of the funnel plots suggested no publication bias (Figure 2).

Discussion
This is a systematic review and meta-analysis of 22 clinical trials to evaluate the effectiveness and safety of PMB in patients with nosocomial infections.Our meta-analysis revealed that PMB is not superior to other antibiotics in terms of mortality, specifically when compared to ceftazidime-avibactam.However, PMB demonstrated an advantage in terms of nephrotoxicity compared to colistin.
Due to the rise of pathogen resistance, there is a need for the development of novel antibiotics and strategies.Despite the availability of new antimicrobial agents, the issue of their high cost persists.It is of paramount importance to reevaluate old antibiotics and investigate their clinical efficacy.Our meta-analysis revealed that PMB is not superior to other antibiotics in terms of mortality, particularly in comparison to ceftazidime-avibactam.Therefore, ceftazidimeavibactam could be a favorable option for nosocomial infections.However, the cost-effectiveness of the comparing antibiotics is not addressed in the present study.Moreover, PMB serves as an invaluable salvage therapy and should be contemplated for combination therapy.Naturally, additional evidence and further qualified studies are necessary to ascertain the cost-efficacy of PMB usage.
Both PMB and colistin are commonly administered in combination with other antibiotics to enhance effectiveness and reduce resistance development.In Maura S. Oliveira's study, both the PMB and colistin groups received combined antibiotics, including amphotericin, ampicillin, carbapenem, cephalosporin, ciprofloxacin, metronidazole, piperacillin-tazobactam, trimethoprimsulfamethoxazole, and vancomycin, with no statistically significant differences observed (23).Similar results were also seen in Ryan L. Crass's study (26).All the studies included were retrospective.The extensive range of antibiotics used in both groups limited further statistical analysis.The variety of antibiotics used in combination may potentially impact outcomes, underscoring the need for further research to investigate the efficacy of antibiotic combination therapy.
Our meta-analysis indicated that PMB demonstrated an advantage in terms of nephrotoxicity compared to colistin.Therefore, PMB may be a more favorable choice than colistin for treating nosocomial infections in patients with renal insufficiency.Additionally, the nebulized administration of both PMB and colistin has been investigated (11, 39-44).While we acknowledge the association between drug dosage and nephrotoxicity, the PMB dosages in different studies vary in presentation.Some articles specify dosages based on unit weight, while others mention total daily dosages.This inconsistency in reporting methods hinders the feasibility of performing regression analysis to explore the relationship between PMB dosage and nephrotoxicity.To explore new potential applications for old antibiotics, such as PMB and colistin, further research is warranted to investigate appropriate selection criteria, administration methods, and dosage recommendations.
The majority of the included articles are from Brazil and China.Subgroup analysis was also performed (Supplementary Figures S1, S2),  revealing that there was no significant difference in nephrotoxicity between the PMB group and the control group in Brazil and the USA.However, the PMB group exhibited a higher risk of developing nephrotoxicity when compared to the control group.The possible reason for this observation could be that many studies included from China in recent years predominantly used tigecycline and tigecycline as comparator groups.PMB shows stronger nephrotoxicity compared to these two drugs.In earlier studies, colistin was the primary comparator, and there was minimal difference in nephrotoxicity.
A few limitations were still present in our study.First, all the studies included in this analysis were cohort studies, and the level of evidence was very low.Second, there is variability in the timing of mortality among different studies.Most studies assessed 30-day mortality, while others focused on in-hospital mortality, 28-day mortality, and 14-day mortality.Such heterogeneity may potentially impair the quality of the synthesized evidence.Third, nosocomial infections often require the combined application of antibiotics.However, our study did not account for the concurrent use of Funnel plot of mortality.The mortality was compared between PMB and other antibiotics.The mortality was compared between PMB and other antibiotics (subgroup analysis based on the risk of bias).The mortality was compared between PMB and colistin.
medications in each group.Finally, there was significant heterogeneity in the dosage and duration of antibiotic between studies for many of our outcomes.Hence, our current findings were weakened, and our study was downgraded one level.

Conclusion
Our meta-analysis showed that PMB is not superior to other antibiotics in terms of mortality, specifically FIGURE 1Flow chart.

TABLE 1
Characteristics of selected clinical trials.
Peng et al.   10.3389/fmed.2024.1400757colistin.Additionally, there is a need for further high-quality studies to elucidate the appropriate antibiotic selection for nosocomial infections.