Nephrotoxicity of Polymyxins: Is There Any Difference between Colistimethate and Polymyxin B?

ABSTRACT

Nephrotoxicity is a common adverse effect of the clinically used polymyxins, colistin and polymyxin B. This adverse effect is dose limiting for both polymyxins, as the plasma polymyxin concentrations associated with renal damage overlap those required for antibacterial effect. Since development of acute kidney injury (AKI) during therapy is highly undesirable, it is extremely important to know whether there is any difference between the nephrotoxic potential of colistin (administered as its inefficient prodrug, colistimethate) and polymyxin B (administered as the active form). Both polymyxins are cytotoxic to renal tubular cells and are prone to cause nephrotoxicity in vivo because of the renal handling mechanisms that facilitate accumulation of these compounds in these cells, processes that are reviewed in this article. Also reviewed are the emerging data that strongly suggest significantly higher rates of AKI in patients treated with colistimethate compared to patients treated with polymyxin B. This finding may be due to differences in pharmacokinetics and renal handling mechanisms of colistimethate and formed colistin versus polymyxin B, and consequently the relative amount of polymyxin material delivered to tubular cells. A lower risk of AKI with polymyxin B is one of several potential advantages over colistimethate. The relative safety and efficacy of the two agents require closer examination in well-designed clinical studies.

INTRODUCTION

The polymyxins (colistin and polymyxin B [PMB]) are old antibiotics that are a key part of the therapeutic armamentarium against multidrug-resistant Gram-negative pathogens. Colistin and PMB have very similar chemical structures and antibacterial activity in vitro (1–3). Whereas PMB is administered to patients in its active form, colistin is used in the form of its inactive prodrug colistimethate (CMS), and conversion to colistin is required in vivo (1, 4, 5). This difference in the form administered results in a number of potential clinical pharmacological advantages for PMB (1). Acute kidney injury (AKI) occurs in a substantial proportion, up to 50 to 60%, of patients receiving CMS or PMB and is the major dose-limiting adverse effect of the polymyxins (1, 6). The plasma polymyxin concentrations associated with increased risk of AKI overlap those required for antibacterial effect, and therefore, the polymyxins have a narrow therapeutic window (7–10). In some parts of the world, only parenteral products of CMS are available, while in the United States and many other countries, pharmaceutical formulations of both CMS and PMB are approved for clinical use (11). There is considerable interest in how CMS and PMB, both of which are “last-line” antibiotics, compare in regard to their potential to cause AKI. If a difference exists, it would be an essential component in the deliberation around which polymyxin to choose—CMS or PMB. Here we review key aspects of the nephrotoxicity of the polymyxins and emerging clinical data on the relative rates of AKI of the two polymyxins. We also discuss mechanisms that may underpin any difference in AKI risk and consider the associated clinical implications of such a difference.

WHY DO POLYMYXINS HAVE THE PROPENSITY TO CAUSE NEPHROTOXICITY?

From studies conducted in cell lines and in vivo preclinical models, it is clear that the polymyxins have the potential to be toxic to renal tubular cells. The cellular mechanisms involved include oxidative stress, apoptosis (via mitochondrial, death receptor, and endoplasmic reticulum pathways), cell cycle arrest, and autophagy (12–14). However, to fully understand the propensity for CMS and PMB to cause AKI, it is necessary to be aware of their respective dispositions in the body, in particular their handling within the kidneys.

CMS is predominantly cleared by renal excretion, involving glomerular filtration, and there is potentially secretion through tubular cells from peritubular capillary blood into tubular urine also involved (15). In animals and humans with good renal function, ∼70% of administered CMS is excreted into urine, and only a small fraction of each dose is converted systemically to colistin (Fig. 1) (15, 16). Due to the chemical instability of CMS in an aqueous environment, a substantial amount of the CMS excreted via the kidneys is recovered as colistin in urine with the likelihood that at least some of the chemical conversion occurs in tubular urine and within tubular cells (15, 16). Colistin formed systemically within the body is delivered to the kidneys via the circulation system, but only a very small percentage of the colistin that is filtered into tubular urine undergoes renal excretion (Fig. 1) (16, 17). This is because colistin is subjected to extremely avid reabsorption through tubular cells (17). Indeed, the fraction of filtered colistin that undergoes reabsorption is comparable to, or greater than, the fractional reabsorption of filtered water, consistent with the reabsorption of colistin involving a carrier-mediated process (17, 18).

FIG 1.

Overview of the pharmacokinetic pathways for CMS (colistimethate) and colistin (left) and polymyxin B (right). The thickness of the arrows indicates the relative magnitude of the respective clearance pathways when kidney function is normal. After administration of CMS, extensive renal excretion of the prodrug occurs with some of the excreted CMS converting to colistin within the kidneys and bladder. As a result, only a relatively small fraction of each CMS dose is converted to colistin within the body. This colistin formed systemically when delivered to the kidneys undergoes extensive tubular reabsorption; the same renal disposition applies to polymyxin B administered as such.

As with colistin formed from CMS, PMB administered as such undergoes very extensive tubular reabsorption following glomerular filtration. This was first demonstrated in critically ill patients wherein it was calculated that 90 to 95% of filtered PMB underwent reabsorption through tubular cells (19). Indeed, in that study (19), it was estimated that the extent of tubular reabsorption of PMB was so large that an amount of drug greater than the daily dose can be trafficked through tubular cells each day. As depicted in Fig. 1, the result of this extensive recycling process is that an average of <5% of each PMB dose is excreted into urine, while the tubular cells are exposed to a very large amount of the drug (19, 20).

Carrier-mediated systems involved in the renal handling of CMS have not been identified. In regard to the tubular reabsorption of colistin and PMB, it appears that megalin, an endocytic receptor expressed in the apical membrane domain of proximal tubular cells and involved in the reabsorption of proteins, peptides, and cationic compounds, is a key player in the reabsorption of the polymyxins (21–23). The affinity of polymyxin B for megalin appeared to be substantially greater than that of gentamicin (21), which is also a substrate of megalin (24). Given that the polymyxins are polypeptides, it is not surprising that renal PEPT2 (peptide transporter 2), a proximal tubule peptide transporter involved in reabsorption of endogenous and exogenous peptides (25), also plays a role in the renal reabsorption of colistin and PMB (18, 26). The affinity of a polymyxin analog for PEPT2 was comparable to that of glycylsarcosine, a dipeptide substrate of the transporter (26). Thus, carrier-mediated transporters are involved in the renal handling and recycling of the polymyxins, and this serves to expose the tubular cells to a very large amount of these potentially nephrotoxic agents (Fig. 1).

TO WHAT EXTENT DO POLYMYXINS ACCUMULATE IN THE KIDNEY AND WHAT IS THE LOCATION?

In rats administered colistin (sulfate) intravenously twice daily for 7 days, the ratio of the average concentration of colistin in kidney homogenate to that in plasma after the final dose was 65.7 (27). For PMB, after even a single intravenous dose, the mean kidney homogenate-to-serum concentration ratios were 7.45 and 19.6 at 3 h and 6 h postdose, respectively (28). The ratios for kidney were substantially higher than those for heart, lung, liver, spleen, and muscle. Because the ratios in these studies were for kidney homogenate (27, 28), they would underestimate the concentration in the region of the greatest polymyxin accumulation. Immunostaining studies of sections of rodent kidneys from animals treated with PMB have shown that accumulation occurs predominantly in the renal cortex, primarily within proximal tubular cells (28, 29). A synchrotron X-ray fluorescence microscopy study revealed that the concentration of a polymyxin analog in single rat (NRK-52E) and human (HK-2) kidney tubular cells in culture was several thousandfold higher than the extracellular concentration (30). Clearly, facilitated by the renal handling mechanisms and recycling processes discussed above (Fig. 1), the polymyxins accumulate extensively within proximal tubular cells of the kidneys, this being the primary site at which megalin and PEPT2 are expressed in the kidney (31, 32) and where the polymyxins induce the damage that may lead to AKI in patients (27, 33).

ARE COLISTIN AND POLYMYXIN B EQUALLY NEPHROTOXIC?

Preclinical studies.

In kidney cell lines, PMB appears to be slightly more toxic than colistin (34, 35). However, in a study of mice administered PMB or colistin (the latter administered directly as a salt, not as CMS), there was no apparent difference in the degree of nephrotoxicity (35). However, the most important comparison is for intravenous CMS and PMB in patients.

Clinical studies.

There have been five clinical observational studies, most retrospective, comparing nephrotoxicity rates between CMS and PMB (34, 36–39). AKI was defined according to RIFLE (Risk, Injury, Failure, Loss, and End-stage renal disease) criteria in three studies (34, 37, 39), Acute Kidney Injury Network (AKIN) criteria in one study (38), and the remaining study used a criterion approximating the “Injury” class of RIFLE or stage 2 of AKIN (36). With the exception of the latter study, the distinguishing findings were the higher overall AKI rates observed among patients treated with CMS compared to those receiving PMB (Fig. 2). It is noteworthy that in three studies, the greatest differences were observed in the rates of “Failure” class of RIFLE or stage 3 of AKIN, which were 17.9% and 9.0% (37), 21.4% and 5.0% (38), and 38.3% and 12.7% (39), in patients treated with CMS and PMB, respectively. In the study of Phe et al. (34), the “Failure” incidences in the cohort without cystic fibrosis patients were similar for CMS (12.0%) and PMB (11.5%), and this was not evaluated in the study of Oliveira et al. (36). Taken together, these data suggest that the differences in the overall AKI rates (Fig. 2) were mainly driven by higher rates of more-severe renal injury in CMS-treated patients.

FIG 2.

Summary of the overall incidence of acute kidney injury (AKI) in patients treated with colistimethate (CMS) or polymyxin B (PMB). AKI was defined as a twofold increase in serum creatinine or an increase of at least 1 mg/dl if initial creatinine was higher than 1.4 mg/dl in the study of Oliveira et al. (36), according to RIFLE criteria in the studies of Akajagbor et al. (37), Phe et al. (34), and Rigatto et al. (39), and according to AKIN criteria in the study of Tuon et al. (38). The incidences of AKI in CMS- and PMB-treated patients, respectively, were 26.0% (10 of 39 patients) and 27.0% (8 of 30) in the study of Oliveira et al. (36), 60.4% (64 of 106) and 41.8% (28 of 67) in the study of Akajagbor et al. (37), 48.2% (40 of 83) and 23.1% (24 of 104) in the cohort without cystic fibrosis patients in the study of Phe et al. (34), 38.9% (14 of 36) and 20.8% (20 of 96) in the study of Tuon et al. (38), and 74.1% (60 of 81) and 46.1% (189 of 410) in the study of Rigatto et al. (39)

Among the four studies that found higher rates of AKI in CMS-treated patients, multivariate analyses were performed in three studies (37–39). Use of CMS was independently associated with AKI in the study of Akajagbor et al. (hazard ratio [HR], 2.27; 95% confidence interval [95% CI], 1.35 to 3.82; P = 0.002) (37) and with renal failure (“F” of RIFLE), which was the primary outcome in the study of Rigatto et al. (HR, 3.35; 95% CI, 2.05 to 5.48; P < 0.001) (39), while it was not independently associated in the study of Tuon et al. (HR, 1.74; 95% CI, 0.82 to 3.69; P = 0.15) when adjusted for concomitant vancomycin use (P = 0.058) and high-dose regimens, defined as >300 mg/day of colistin base activity (CBA) or ≥200 mg/day of PMB (P = 0.04) (38). Phe et al. (34) did not perform multivariate analysis, but they selected patients from the entire cohort excluding those with cystic fibrosis and those who received low daily doses (<2.5 mg/kg of body weight by ideal body weight of CBA and <1.5 mg/kg of PMB) and performed a 1:1 matching according to duration of therapy and age. The incidence of AKI in the matched cohort was strikingly higher in CMS-treated patients than in PMB-treated patients (55.3% and 21.1%, respectively; P = 0.003) (34).

The results of a recent meta-analysis including these five studies support the proposition that patients treated with CMS presented a higher risk of developing AKI than those administered PMB, with an unadjusted relative risk (RR) of 1.55 (95% CI, 1.36 to 1.78) (40). Analyzing only the four studies with multivariate analysis (36–39), the pooled CMS HR was 2.16 (95% CI, 1.43 to 3.27). The meta-analysis also found that AKI episodes occurred earlier with CMS (40).

As discussed above, the current clinical data suggest that CMS may be more nephrotoxic than PMB in patients. However, it must be recognized that a major consideration in clinical studies comparing both polymyxins regarding either nephrotoxicity or efficacy is the lack of correlation between CMS and PMB doses, especially across the renal function spectrum. While a reduction in daily dose is appropriate for CMS in patients with decreased renal function (10, 41), this is not the case for PMB, as it will result in a decrease in plasma PMB concentration, which may impact antibacterial effect (19). Since PMB dose reduction in patients with renal function impairment has been a common practice based on old empirical recommendations, it is possible that patients in the five aforementioned studies (34, 36–39) who presented lower creatinine clearances at baseline received low doses of PMB with consequent low plasma concentrations and reduced renal exposure to the antibiotic. In contrast, owing to the effect of renal function on the disposition of CMS and formed colistin (1, 41), plasma colistin concentrations in patients with substantially impaired kidney function (creatinine clearance of <30 ml/min) receiving appropriately adjusted CMS doses (42) are expected to be similar to those in patients with creatinine clearance of 50 to 80 ml/min (10). Most patients with creatinine clearance of >80 ml/min would have substantially lower plasma colistin concentrations than those in the renally impaired patients, even with a daily CMS dose generally regarded as being at the upper limit (10). However, these considerations would not explain the results of Phe et al. (34) and Rigatto et al. (39), because in the former study, patients with impaired renal function and those receiving very low PMB doses (<1.5 mg/kg/day) were excluded from the analysis, while in the latter study, no PMB dose adjustment according to creatinine clearance was recommended. Nevertheless, in the study of Phe et al. (34), the mean PMB dose (∼104 mg/day) may be considered low, and this might have decreased the risk of AKI. Additionally, with the exception of the study of Rigatto et al. (39), in which PMB doses were not adjusted for renal function, no other study described how doses were managed according to creatinine clearance. Therefore, it is possible that patients who presented early or minor signs of nephrotoxicity during PMB therapy had their doses reduced, decreasing renal exposure to the drug and consequently reducing the probability of further progression of kidney injury. This might also help to explain why differences between CMS and PMB in the rates of more-severe kidney damage (“F” of RIFLE and stage 3 of AKIN) were higher than the differences in the rates of less-severe impairment. In contrast to this hypothesis, AKI rates were also significantly higher among patients treated with CMS compared with those administered nonadjusted doses of PMB in the study of Rigatto et al. (39), including when CMS-treated patients were compared only with patients receiving high doses of PMB (comparisons undertaken for two PMB dose clusters: ≥150 mg/day and ≥200 mg/day). Although the latter finding does not rule out the possibility of lower PMB kidney exposure in previous studies, it suggests that higher rates of AKI with CMS cannot be fully explained by this potential bias.

WHY MIGHT CMS BE MORE NEPHROTOXIC THAN PMB IN PATIENTS?

The results of studies of cell lines (34, 35) and an animal model of polymyxin-induced nephrotoxicity (35), all of which involved direct use of PMB and colistin (i.e., CMS was not used), cannot explain the clinical observations reviewed above. It is possible that differences in the shape of the plasma concentration-versus-time profiles for formed colistin (relatively “flat” profile) following administration of CMS (41) and PMB (larger “peak-to-trough” fluctuation) (19) may influence the extent of accumulation in tubular cells and consequently the relative risk of AKI in patients. In apparent support of this suggestion is that the rate of uptake of polymyxin B1 by porcine kidney proximal tubular (LLC-PK1) cells grown in vitro as a monolayer was saturable (23). However, the extent of nonlinearity in the rate of uptake was relatively small across the range of PMB concentrations likely to be present in the proximal tubular urine over a dosage interval in patients, i.e., associated with the plasma “peak-to-trough” fluctuations arising from clinical dosage regimens (19). In a study conducted to investigate the influence of dosage regimen on uptake of PMB into kidney tissue in rats, two groups of animals received the same daily dose (20 mg/kg) subcutaneously for 24 h, but one group was administered the daily dose in four divided doses, while the other group received the entire daily dose in one administration (20). The concentrations of PMB components in kidney tissue homogenate at 24 h in the animals receiving once-daily dosing were approximately half those in the animals that received the same daily dose divided into four administration events (P < 0.05). The investigators proposed that saturable carrier-mediated reabsorption of PMB resulting from the relatively high peak plasma concentrations with once-daily dosing may have been the explanation for the apparent lower extent of accumulation with that dosage regimen (20). However, it is very important to note that the kidney tissue samples were collected at the time of a “trough” PMB concentration for both groups. Even in the absence of any saturable (i.e., nonlinear) process, pharmacokinetic principles dictate that animals receiving once-daily dosing would have substantially lower “trough” concentrations than those receiving the daily dose in a divided fashion at 6-h intervals. Moreover, the urinary recovery over 24 h of the total dose as unchanged PMB was <5% in both groups (20), but it was not mentioned whether there was higher recovery of the total dose in the once-daily dosing group, which may be expected if a saturable tubular reabsorption process was involved. Thus, overall, there is insufficient evidence to support differing degrees of saturability of tubular reabsorption of PMB versus formed colistin associated with the different shapes of the respective plasma concentration-versus-time profiles as the explanation for differential nephrotoxicity of PMB and CMS in patients. Nevertheless, the different shapes of the plasma concentration-versus-time profiles for formed colistin (relatively “flat” profile), following administration of CMS, and PMB (larger “peak-to-trough” fluctuation) may conceivably contribute to the lower rate of AKI in patients receiving PMB via mechanisms yet to be identified.

It is also possible that the finding that CMS is more nephrotoxic than PMB in patients is the result of the inefficiency of CMS as a prodrug and of the role of the kidneys in processing these different compounds (CMS and formed colistin versus PMB). In considering this possibility, it is important to recognize that the CBA unit of dosing is based upon a measure of microbiological activity from an in vitro incubation of CMS and does not indicate the amount of “colistin-related material” (i.e., CMS, partially methanesulfonated derivatives [1] and colistin) presented to the body. Consider a patient with good renal function administered a typical CMS daily dose of 300 mg CBA. After considering conversion factors, including respective molecular weights (43), 300 mg of CBA is equivalent to ∼740 mg of the chemical CMS, which in turn contains ∼495 mg of “colistin” in situ (i.e., the amount of colistin that would be generated if CMS was completely converted to colistin). If 70% of this is cleared renally (Fig. 1), the kidneys are exposed to a large amount of CMS (equivalent to ∼345 mg/day of “colistin”), with some of the CMS undergoing conversion to colistin within the kidneys, as discussed above. It is not known whether CMS or the numerous partially methanesulfonated derivatives formed during the conversion to colistin (1) are themselves nephrotoxic. The kidneys would also be exposed to additional colistin from the filtration and reabsorption (i.e., recycling) of colistin formed systemically (Fig. 1). If for example ∼15% of the administered CMS daily dose (495 mg “colistin” in situ) is converted systemically to colistin (1), then the patient is actually receiving systemically only ∼75 mg/day of colistin, and this will be recycled through the kidneys. This low extent of systemic conversion to colistin for circulation of the latter throughout the body illustrates why CMS is regarded as an extremely inefficient prodrug (1). In patients with creatinine clearance of >80 ml/min receiving the CMS regimen described above, only ∼30% of patients would achieve an average steady-state plasma concentration (C_ss,avg) of colistin ≥2 mg/liter, regarded as a reasonable target concentration (10). In contrast, for the same type of patient receiving a typical daily dose of 200 mg PMB sulfate (equivalent to ∼165 mg of PMB) where no conversion is required, the entire dose is available for circulation throughout the body, including the kidneys. It would be expected that ∼90% of such patients would achieve a plasma PMB C_ss,avg of ≥2 mg/liter (19). Rigatto et al. reported that the incidence of AKI (“F” of RIFLE) in patients with creatinine clearance of ≥90 ml/min was ∼5-fold higher (P < 0.001) with CMS (daily dose [mean ± standard deviation {SD}], 318.4 ± 67.3 mg CBA) than with PMB (162.5 ± 40.3 mg PMB) (39). Plasma polymyxin concentrations were not determined in the study of Rigatto et al., but it is very likely that with the aforementioned doses administered to patients with creatinine clearance of ≥90 ml/min, plasma C_ss,avg of PMB would have been substantially higher than the plasma C_ss,avg of formed colistin in the patients administered CMS (10, 19, 41). Thus, the finding of an ∼5-fold-higher incidence of AKI in the CMS-treated patients strongly implicates the delivery of CMS to the kidneys and its processing within that organ, probably involving intrarenal conversion to colistin, as having an important role in the renal injury that may occur with CMS. As noted above, the extent to which CMS or the numerous partially methanesulfonated derivatives formed during the conversion to colistin contribute to nephrotoxicity is not known. CMS may be like a Trojan horse that delivers large amounts of colistin-related material to the kidneys. Given that the ratio of systemically available active drug (and the respective C_ss,avg) to the amount of xenobiotic delivered to the kidneys is substantially lower for CMS, it is possible (but as yet unproven) that PMB may have a wider therapeutic window, especially for patients with good renal function.

CLINICAL IMPLICATIONS

AKI has long been associated with worse outcomes in hospitalized patients, including greater length of stay, increased hospital costs, higher risk for mortality (44, 45), and longer-term sequelae (46–48). It is a logical deduction that a drug with comparable therapeutic efficacy and lower nephrotoxic potential would be preferable over its comparator, because if it were the single difference between them, better short- and long-term outcomes would be expected with the less nephrotoxic agent. Regrettably, there is still no clinical study with adequate statistical power specifically designed to compare efficacy between the polymyxins and the impact of developing AKI during therapy in these patients, who are often severely ill and present several comorbidities. In one study with PMB, development of AKI during therapy was barely associated with increased risk for mortality in multivariate analysis (HR, 1.35; 95% CI, 0.99 to 1.85; P = 0.06) (49), but there was no comparison with CMS.

In fact, secondary analysis of comparative studies showing higher AKI incidence in patients treated with CMS could not show any significant difference in mortality rates, whenever described, between PMB- and CMS-treated patients (34, 38, 39). Counterintuitively, overall mortality rates tended to be higher in PMB-treated patients than in CMS-treated patients. Phe et al. (34) found that in-hospital mortality was significantly lower in CMS-treated patients compared with PMB-treated patients (12.0% versus 30.8%, respectively; P = 0.003) in the cohort without cystic fibrosis patients. Tuon et al. (38) observed that patients who developed AKI presented significantly higher in-hospital mortality compared with patients without AKI (67.6% and 39.8%, respectively; P = 0.009), but no mortality difference was demonstrated between PMB (45.8%) and CMS (50.0%) groups (P = 0.48) despite a trend to increased risk for AKI in the CMS group in multivariate analysis (HR, 1.74; 95% CI, 0.82 to 3.69; P = 0.15). Finally, in the study of Rigatto et al. (39), 30-day mortality rates were 30.9% and 43.4% in the CMS and PMB groups, respectively (P = 0.083), even with the observation of significantly higher mortality rates in patients who developed renal failure (50.6%) compared with those who did not (38.0%) (P = 0.044). The results of the recent meta-analysis (40), including the three studies discussed above (34, 38, 39), indicated no difference in the risk for overall mortality but a trend to lower risk with CMS therapy (RR, 0.71; 95% CI, 0.45 to 1.13), in agreement with the study of Oliveira et al. in which no difference was observed in 30-day mortality (36).

Despite these apparently contradictory findings, it is imperative to note that, with the exception of the study of Oliveira et al. (36), none of these studies were designed to compare mortality between polymyxins. Thus, critical covariates that may affect this outcome have not been considered in these studies. Factors influencing clinical outcomes, such as proportion of patients with presumed versus confirmed infection, primary infection site, time to start polymyxin regimen, polymyxin MIC, use of a second active antibiotic in combination, among others, were not addressed in these studies. Such factors may potentially have affected the overall mortality of patients, thereby overcoming the expected higher risk for mortality posed by the increased incidence of AKI in patients treated with CMS. In addition, it is possible that the same potential bias that might partially explain lower AKI rates with PMB in some studies, as discussed above (i.e., relatively lower PMB doses administered to patients with low baseline creatinine clearance and dose reduction with even minimal increases in serum creatinine concentration, both resulting in lower plasma PMB concentrations), may also explain this trend to higher mortality rates in these patients.

There is no doubt that AKI may impair the general condition especially in critically ill patients (44), leading to fluid overload, cardiac insufficiency, and electrolyte and acid-base disorders, among others. Thus, although it has not been demonstrated in clinical studies comparing CMS and PMB, the development of AKI during therapy is expected to impair short- and long-term outcomes (44, 45, 48). Clearly, both polymyxins are potentially nephrotoxic, and therefore, strict attention must be paid to monitoring of renal function and fluid and electrolyte balance, avoidance of concomitant nephrotoxic agents when possible, and provision of supportive measures as needed (6). Doses of both polymyxins should be optimized; those for CMS should be adjusted for baseline renal function (10, 41), but this is not appropriate for PMB (19). If AKI occurs, the dose of CMS can be adjusted to accommodate the reduced renal clearance while maintaining the preexisting plasma colistin concentration; if the dose of PMB is reduced in the face of AKI, the plasma PMB concentration will decline and potentially impact the antibacterial effect. Any consideration given to ceasing either polymyxin should recognize that polymyxin-associated AKI is usually mild to moderate and reversible (6) and balance risk to benefit in the light of the condition of the patient and the availability of other antimicrobial options. While concomitant administration of the antioxidant ascorbic acid has been shown to protect against colistin-induced nephrotoxicity in animals (50), clinical studies have provided contradictory findings (51, 52), necessitating further examination.

CONCLUSION

The renal handling mechanisms of the polymyxins favor their extensive accumulation in kidney proximal tubular cells. There is a growing body of data suggesting that CMS is more nephrotoxic than PMB in patients. Such an outcome is not unexpected given the differences in the overall pharmacokinetics and renal handling mechanisms involved, and the much greater load of colistin-related material delivered to the body, the kidneys in particular, in the form of the inefficient prodrug CMS. A lesser risk of AKI with PMB is one of several potential advantages over CMS in certain clinical circumstances (1). These aspects and the impact of any differential nephrotoxicity of CMS and PMB on prognosis and mortality require assessment in well-designed clinical studies that should include determination of the concentration of each polymyxin in plasma.