Due to nephrotoxicity concerns, the clinical use of polymyxins was considerably reduced in the early 1970s (Keirstead et al., 2014). The nephrotoxicity rates of polymyxins reported range from 11.8 to 58.1% (John et al., 2017; Aggarwal and Dewan, 2018; Yarelis et al., 2019; Sisay et al., 2020). However, polymyxin was reintroduced in clinical practices worldwide because of the increasing incidence of infections caused by multi-drug resistant (MDR) Gram-negative microorganisms and the lack of new antibiotics. Multi-drug resistant (MDR) bacteria are defined as the bacteria that exhibited resistance to clinically used class 3 or more antimicrobial drugs simultaneously. PMB has been widely used as a kind of polymyxin in carbapenem-resistant Gram-negative bacterial (CR-GNB) infections (Vaara, 2019). Because PMB was developed before modern drug development trials, there is little data available on its pharmacokinetics and pharmacodynamic properties. Increasing pharmacokinetics and pharmacodynamics studies on PMB have been published, providing a more precise dosing strategy to maximize the efficacy and reduce nephrotoxicity (Quagliano et al., 2020; Wang et al., 2020; Xie et al., 2020).

Although polymyxins showed strong antimicrobial activity against Gram-negative bacteria, their adverse reactions (ADRs), such as nephrotoxicity and neurotoxicity, limit their clinical use (Soman et al., 2020). Pierson-Marchandise et al. (2016) analyzed 38,782 ADRs recorded in the French national pharmacovigilance database and found that 3.2% of ADRs were classified as cases of acute kidney injury (AKI). AKI increases the risk of death and serious morbidity in hospitalized patients (Hoste et al., 2015). Previous studies investigated the nephrotoxicity of PMB and colistin and the risk factors of polymyxin-associated AKI (Aggarwal and Dewan, 2018; Ozel et al., 2019; Almutairy et al., 2020; Cai et al., 2020; Zhang et al., 2021). Moreover, some system reviews and meta-analyses focused on the nephrotoxicity of polymyxins. They found that malignancy, co-infection with other microorganisms, being elderly, the high daily dose of PMB, having underlying diseases such as diabetes, and use of concomitant nephrotoxic drugs were independent predictors of nephrotoxicity (Oliota et al., 2019; Sisay et al., 2020; Wagenlehner et al., 2020). Ritesh et al. compared the nephrotoxicity of colistin with PMB administered in currently recommended doses and found that PMB is less nephrotoxic than colistin (Aggarwal and Dewan, 2018). However, there is only one amino acid difference between PMB and colistin, which had been documented to have well-known nephrotoxicity and neurotoxicity (Zavascki and Nation, 2017). The relative safety of the two agents requires closer examination in well-designed clinical studies (Zavascki and Nation, 2017).

Studies have demonstrated that PMB is substantially accumulated in proximal renal tubules and potentially toxic to renal tubular cells in in vitro and in vivo models (Zavascki and Nation, 2017; Azad et al., 2019). Cai et al. (2020) found that higher daily PMB doses (∼30,000 IU/kg/day) and the higher number of concurrent nephrotoxins were independently associated with AKI. Despite previous reports on the incidence of PMB-related nephrotoxicity ranging from 14.0 to 50.6% and the associated risk factors (Nelson et al., 2015; Soares et al., 2017; Cai et al., 2020; Sisay et al., 2020; Zhang et al., 2021), the association between PMB and the development of nephrotoxicity remain inconclusive, especially the dosing and concurrent nephrotoxins and underlying diseases such as kidney disease. Moreover, considering the racial differences in drug dose and PK/PD, it is imperative to explore clinical variables that affect the nephrotoxicity of PMB among Chinese patients.

In this study, we aim to evaluate the nephrotoxicity of Chinese patients using PMB and the possible risk factors for AKI caused by PMB to provide a reference for clinical use of PMB.

We conducted a retrospective cohort study about PMB-associated nephrotoxicity and its predictors. Our study found that the AKI incidence was 28.63% in 234 CR-GNB-infected patients with PMB-based regimens. Males, digestive system diseases, using furosemide (>20 mg/day), and high baseline serum creatinine were associated with a higher risk for AKI during hospitalization. Excluding patients with AKD or CKD before PMB use, digestive system diseases and furosemide (>20 mg/day) were associated with a high risk for AKI. The severity of acute kidney injury is related to the maintenance dose of PMB and high baseline serum creatinine. These risk factors should be taken into consideration when initiating PMB-based therapy.

The incidence rate of PMB-related nephrotoxicity ranges from 11.8 to 50.6% in previous studies (Nelson et al., 2015; John et al., 2017; Soares et al., 2017; Aggarwal and Dewan, 2018; Cai et al., 2020; Sisay et al., 2020; Zhang et al., 2021). Our study subjects comprise a Chinese population. Whether the ethnics contribute to the incidence rate of PMB-related nephrotoxicity is still not known. The AKI rate in our study was 28.63%, which was lower than the rate reported in previous reports on Caucasians and Asians (Nelson et al., 2015; John et al., 2017; Soares et al., 2017; Aggarwal and Dewan, 2018; Cai et al., 2020; Sisay et al., 2020; Zhang et al., 2021). This may be due to various reasons that lead to the high variation of AKI incidence in different studies, such as different population characteristics, renal function status, PMB doses, and the different proportions of drug use with nephrotoxicity. Patients with fluctuating renal function were not excluded, which might have increased the AKI rates (John et al., 2017). In addition, using different entry criteria may also be one of the influencing factors. The low rates of nephrotoxicity with PMB may be associated with the patient on any other nephrotoxic drug and who received significant doses of diuretics, for diuretics have shown to potentiate polymyxin nephrotoxicity (Deryke et al., 2010; Pogue et al., 2011). Similarly, in the study by Yarelis et al. (2019), the patients with a baseline creatinine clearance (CrCl) <10 ml/min and/or on renal replacement were also excluded. However, the overall rate of AKI in this cohort was 35% (Yarelis et al., 2019). In this cohort, the dosing for PMB ranged from 15,000 to 25,000 U/kg/day (Yarelis et al., 2019). In our study, 79.06% of patients used the low daily doses of PMB compared to the recommended doses, as Pooja et al. suggested that a higher daily dose of PMB was associated with a more rapid onset of nephrotoxicity (Manchandani et al., 2017). A total of 46% of patients developed any degree of AKI during PMB treatment with the high dose reported by John et al. (2017). In this study, patients received PMB at a dose of >3 mg/kg of body weight/day or a total dose of ≥250 mg/day, while the dose in our study was 1.68 (1.66–2.00) mg/kg/day. Moreover, after excluding the AKD or CKD patients before PMB use, the AKI rate in patients with normal baseline serum creatinine was 32.8%, similar to previously published data by Zhang et al. (2021) in which the AKI rate was higher with 38.7%. In their study, the daily dose of PMB is also less than 2 mg/kg. A recent study reported that 11.8% of patients’ incidence of nephrotoxicity was in the PMB group (Aggarwal and Dewan, 2018). The source of discrepancy used different criteria to define nephrotoxicity; the “risk I” category was excluded in their study, even though RIFLE criteria were used (Aggarwal and Dewan, 2018).

We also found that the maintenance dose of PMB in the AKI group is higher than in non-AKI groups (p = 0.048). The loading dose rate and loading dose have the trend of being higher in the AKI group [52.25% vs. 38.92%, p = 0.063; 80 (50–100) vs. 50 (50–100) mg, p = 0.057]. Therefore, the doses of PMB are associated with the incidence of AKI. Our study results are consistent with previous meta-analyses (Oliota et al., 2019; Sisay et al., 2020; Wagenlehner et al., 2020) and case–control studies (Cai et al., 2020; Zhang et al., 2021). A recent meta-analysis found that older age, the daily dose [OR = 1.46 (1.09–1.96)], underlying diabetes mellitus, and concomitant nephrotoxic drugs were independent risk factors for polymyxin-induced nephrotoxicity (Sisay et al., 2020).

Moreover, we first report that the maintenance dose of PMB is also associated with the severity of AKI. In the univariate analysis of the pairwise comparison of AKI severity, we found that between the R stage and I stage, the greater the maintenance dose of PMB, the greater the severity of AKI risk. However, the maintenance dose of PMB did not show a difference in stage F, which may indicate that PMB does not cause serious kidney damage. The maintenance dose of PMB in this study is lower than the standard dose of PMB (1.25–1.5 mg/kg q12 h) and does not cause severe AKI. Larger prospective studies on the dose of PMB to nephrotoxicity are needed.

Our data showed that the independent risk factors associated with AKI development included males, digestive system diseases, using furosemide, and high baseline serum creatinine. In the study by Sisay et al. (2020), though they were not statistically significant, they were 1.52 times more likely to induce nephrotoxicity. We observed that males were one risk factor of AKI. Sisay et al. (2020)also reported that the PMB-associated nephrotoxicity in male patients is 1.52 times more than that in female patients. The previous study also implied that the male sex is associated with an increased incidence of hospital-associated AKI requiring renal replacement therapy (Neugarten et al., 2018). They suggested that reno-protection is mediated by the effects of sex hormones on cellular processes instrumental in the pathogenesis of AKI, analogous to our suggestion that sex hormones mediate the beneficial effects of female sex on the course of chronic kidney disease (Neugarten et al., 2018). Sex-related differences in the generation of nitric oxide, in the synthesis and vascular response to endothelin-1, in the inflammatory, hemodynamic, and humoral responses to ischemia–reperfusion injury, and in the renal hemodynamic response to angiotensin II have been demonstrated in experimental models and human subjects (Takayama et al., 2007; Hutchens et al., 2008). Whether these sex-related differences influence the incidence of PMB-associated AKI needs further research.

Older age is a known risk factor for AKI that has also been found in previous studies with PMB (Balkan et al., 2014; Phe et al., 2014). Although the median age of the AKI group was higher than that of the non-AKI group, there was no difference in the older age in our study. Using furosemide is also associated with the onset of nephrotoxicity in our research. Diuretics have potentiated polymyxin-associated nephrotoxicity in previous studies (Deryke et al., 2010; Pogue et al., 2011). In our study, digestive system diseases were one of the risk factors for AKI. Alkhatib et al (2009) demonstrated that the prevalence of AKI is higher in elderly patients presenting with acute upper gastrointestinal bleeding (48.7%). Zhang et al. (2008) proposed that TNF-α acts directly on glomeruli and renal tubular capillaries, leading to ischemia and tubular necrosis. This may increase the platelet-activating factor, inducing and producing inflammatory mediators (Zhang et al., 2008). The secretion of phospholipase A2 (PLA₂), which originates from platelets, can produce a large quantity of thromboxane A2 (TXA₂) and prostacyclin (PGI₂). These two substances function in angiotasis regulation (Zhang et al., 2008). The proportional imbalance of TXA2/PGI2 can cause vasomotion disturbance, the formation of microthrombus, vascular occlusion, and other pathological changes, leading to abnormal contraction of blood vessels of the kidney, a decline in renal blood flow, and perfusion of kidney tissue (Zhang et al., 2008). These changes can cause serious injury to the kidney. Inflammatory mediators may increase mucosal permeability, causing endotoxins and bacteria to translocate from the colon. Endotoxin increases the level of endothelin, leading to vasoconstriction, renal blood flow reduction, and renal tubular necrosis, thereby promoting the development of AKI (Bose et al., 2002). Zhang et al. (2018) stated that during sepsis, the combined effect of erosion of the mucus barrier, a shift in the composition and virulence of intestinal microbes, and the inability of the host epithelium to regulate its proliferative and apoptotic responses might lead to a tipping point in gut function in which cascading inflammation drives AKI. The potent direct nephrotoxic effects of polymyxins include mechanisms that kill bacteria via interactions with lipid A, disrupting the Ca²⁺ and Mg²⁺ bridges, which destabilize the lipopolysaccharide molecules in the bacterial membrane, and the effects on D-amino content and fatty acid components that increase membrane permeability and the influx of cations (Falagas and Kasiakou, 2006; Velkov et al., 2013). In both LLC-PK1 cells and rat kidney models, PMB reduced creatinine clearance. It increased renal vascular resistance and oxidative damage, demonstrating that PMB nephrotoxicity is characterized by mitochondrial dysfunction and free radical generation (Vattimo Mde et al., 2016). These factors, such as males, digestive system diseases, using furosemide, and high baseline serum creatinine, increase the risk of PMB-associated AKI incidence and need further mechanistic research.

In the rate of AKI, the baseline serum creatinine is connected with AKI before we exclude the AKD or CKD patients. Moreover, the lower baseline serum creatinine has a higher risk of AKI. This is similar to the study by Aggarwal and Dewan (2018), which found that patients with low baseline creatinine clearance had a significantly lower incidence of renal failure than patients with high creatinine clearance.

One of the strengths of our study was that we included 234 patients, which is relatively large sample size data compared to recent studies about PMB-associated nephrotoxicity in the Chinese population. We strictly followed KDGIO’s definition of AKI and used the RIFLE classification to standardize the severity of AKI. Our study was limited by not eliminating all possible confounding factors contributing to renal failure. There is no general laboratory data such as complete blood count, liver function test, and infection markers such as C-reactive protein, which is also the limitation of our study. Moreover, due to the complexity of the causes of AKI, many possible related factors, such as other drugs used at the same time, other underlying diseases, and other factors that may affect the occurrence of AKI, were not considered in the study, which is also the limitation of our study. This project is a single-center retrospective cohort study, which requires further verification by a multi-center prospective cohort study with a larger scale size.

In conclusion, our study demonstrated that male, digestive system diseases, using furosemide (>20 mg/day), and high baseline serum creatinine were the independent risk factors of PMB-associated AKI development. The serum creatinine value should be closely monitored when using PMB. The maintenance dose of PMB may be related to the severity of AKI, and these risk factors should be considered when initiating PMB-based therapy.