Polymyxin has been used into clinical in the 1950s, but has been taken off the market attribute to its toxicity concerns. With the increasing severity of MDR and XDR strains of Pseudomonas aeruginosa, Acinetobacter baumannii, and Enterobacteriaceae , polymyxin as a salvage treatment has resurged. Although polymyxin was decades-old antibiotic, there still seem to have some confusions related to their proper use. A consensus therapeutic guidelines for their optimal use and dosing in adult patients were released .
Compared to the clear results of colistin [16–19], both the preclinical and clinical data were scarce for polymyxin B [19, 20–22]. However, given their similar molecular structures and in vitro activity [23, 24], to date a similar pharmacokinetics and pharmacodynamics (PK/PD) therapeutic target for colistin (AUCss,24 hr target of 50–100 mg. hour/L) was recommended for polymyxin B. But it was worth noting that a growing body of evidence indicated there were many differences between the two types of polymyxin, including formulations , toxicodynamic , and pharmacokinetic characteristics [27, 28]. Thus, any additional data, even observational study in the natural state, assessing polymyxin B application characteristic are urgently and meaningful.
Polymyxin B started being utilized with strict limitations since August, 2020 in our hospital. To limit the usage, we adopted a temporary procurement policy. The application for the use of the drug should be submitted by the doctor and then checked by the pharmacist. From our results we could see polymyxin B was mainly used in critically ill patients. All of them had already exposed to other antibiotics, especially β-lactams/β-lactamase inhibitors, Carbapenems, Tigecycline, and Glycopeptides antibiotics. The reasons to resort to polymyxin B mainly lied in previous treatment failure, changes of antimicrobial susceptibility, and drug intolerance (eg. Tigecycline related decrease of erythrocyte, leucocyte and platelet systems). The patients received the treatment of polymyxin B mainly suffered from lung infection and/or bloodstream infection. About 68.86% (57/84) of patients were given with a fixed doses of polymyxin B (50 MU, bid) without a loading dose. First it's worth pointing out that it was recommended all patients should have administered with a loading dose , priority should be given to critically ill patients with severe pneumonia or sepsis. It was also recommended that both loading dose (20,000–25,000 IU/kg equivalent to 2.0–2.5 mg/kg over 1 hour) and maintenance dose (12,500–15,000 IU/kg TBW equivalent to 1.25–1.5 mg/kg every 12 hours) should be determined based on total body weight (TBW). A dose of 500,000 IU might led to the problem of inadequate exposure. Moreover, data from the rat model of pulmonary infection with intravenous administration of polymyxin B suggested that  even higher drug exposures might still fall short of the concentration required to killing bacteria in lower respiratory infections. Although some studies reported the use of high doses (> 30,000 IU/kg equivalent to 3 mg/kg) [30, 31], more research should be done to further investigate the safety and efficacy of high dose regimens in different clinical scenarios.
Considering even if the upper limit dosage was used, blood concentration of polymyxin B was still not likely to reach the effective level in special populations (eg. critically ill patients) and specific site infection (lower respiratory tract infections) . The potential ADR concerns were also the main limitations for dosing [32–34]. To dissolve this problem, polymyxin B based combination regimen was recommended, which also could avoid the emergence of resistance induced by monotherapy [35–38]. Unfortunately, clinical studies aimed at assessing the effectiveness between the combination therapy and polymyxin B monotherapy was very controversial and not conclusive [39, 40]. For example, critically ill patients enrolled in the study often had complex preexisting comorbidities. These population themselves were prone to suffer from treatment failure and death, which were not associated with infection. These problem further challenged the interpretation of the all-cause mortality between the two groups. More clinical evidence no matter form randomized controlled trials (RCT) or the summary of experiences were need.
The potential risk of polymyxin B treatment was acute renal injury, which was the most common ADR (20%-60%)[41–43].In our study the nephrotoxicity was observed in only 5 cases (5.95%), which is much lower than the data reported in references. It was difficult for us to evaluate this ADR, for some of the patients had already undergone continuous renal replacement therapy. Anyway, to avoid the renal toxicity in clinical practice is very important. However, the data of exposure concentration and toxicity were still limited. The recommended maximal dose derived from a meta-analysis 
Therapeutic drug monitoring (TDM) could be used as an aid to clinical research as well as practical application, especially in the cases where higher doses were adopted. For the relationships between blood concentration of colistin and its effectiveness  / nephrotoxicity [45–47] had been confirmed by sufficient evidence. The data also showed that the breakpoint was 2 mg/L for of colistin [25, 29, 48, 49]. However, the target concentration of polymyxin B had not been established. TDM with polymyxin B had not been carried out in our hospital, which might be the biggest disadvantages of our study. A TDM during polymyxin B treatment might help us to build a relationship between blood concentrations of polymyxin B and its clinical outcome.