Risk factors of colistin safety according to administration routes: Intravenous and aerosolized colistin

Background Nephrotoxicity of intravenous (IV) colistin has impeded its clinical use; aerosolized (AS) colistin may be an alternative, but safety data are lacking. Therefore, this study aimed to evaluate the incidence of acute kidney injury (AKI) and risk factors associated with IV and AS colistin administration. Methods A retrospective study was performed in a tertiary referral hospital. Data were collected before and after colistin administration between October 2012 and April 2016. Exclusion criteria were as follows: age less than 18 years, previous colistin administration, concurrent use of IV and AS colistin, dialysis before colistin use, and colistin use for less than 3 days. We compared AKI incidence following administration of IV versus AS colistin and analyzed risk factors for colistin-associated nephrotoxicity. Results A total of 464 patients were enrolled (n = 311, IV group; n = 153, AS group). Incidence of AKI was significantly higher in the IV group (IV vs AS, 20.26% vs 7.84%, p-value < 0.001). Duration of colistin use (OR 1.033, 95% CI 1.009–1.058, p-value 0.008) and presence of chronic kidney disease (OR 2.710, 95% CI 1.348–5.448, p-value 0.005) were associated with nephrotoxicity. There were no significant risk factors associated with AS colistin. Conclusions Although AS colistin was not associated with any significant risk factors for nephrotoxicity, duration of colistin use and baseline kidney function may affect AS colistin-associated nephrotoxicity.


Subjects
This single-center retrospective study was conducted in Severance Hospital, Seoul, South Korea. In total, 893 patients who were administered IV or AS colistin between October 2012 and April 2016 were evaluated. Subjects who were outpatients, younger than 18 years, had been previously administered colistin, used colistin for less than 3 days, received dialysis before colistin use, and who used IV and AS colistin concurrently were excluded from the study. According to the product label in Korea, 3 million units (MU) equals 100 mg. In Korea, the recommended dose is 2.5-5 mg/kg/d (75,000-150,000 units/kg/d) of IV colistin in 2-4 doses without a loading dose, and dose can be adjusted according to kidney function. Weight is based on actual body weight (ABW) except for obese patients, who use ideal body weight (IBW) for dosing [26]. Different from IV colistin, AS colistin is used as an off-label drug, thus, dosage range varies because clinicians usually rely on their experience to determine the dose. However, 75 mg (2.25 MU) of AS colistin was usually administered 3 to 4 times per day. The study protocol was approved by the institutional review board (IRB) of Yonsei University Health System (Seoul, South Korea, IRB No.4-2016-0490) and all data provided form hospital were anonymized and informed consent was waived by IRB.

Data collection
The following patient information was gathered from electronic medical records (EMRs): age, sex, actual body weight (ABW), height, colistin administration route, colistin dose, duration of colistin use, SCr before and after colistin administration, blood urea nitrogen (BUN) before and after colistin administration, dialysis record after colistin use, underlying diseases (hypertension, diabetes mellitus, and chronic kidney disease), infected pathogens, APACHE2 (Acute Physiology and Chronic Health Evaluation) score, and other concomitant nephrotoxins [7,27]. If the EMRs contained several data points on the first or last day of colistin administration, the mean value was used; if records were not available on the first or last day of colistin administration, the nearest record was used. APACHE2 score was calculated only in ICU patients.

Definitions
IBW was calculated by using the Devine formula for height of more than 5 feet (for men, 50 kg + 2.3 kg for each inch over 5 feet; for women, 45.5 kg + 2.3 kg for each inch over 5 feet). IBW was calculated by the equations of less than 5 feet (for men, 50 kg -2.3 kg for each inch less 5 feet; for women, 45.5 kg-2.3kg for each inch less 5 feet [26]. Accumulation dose, daily dose, daily dose per ABW, and daily dose per IBW were calculated to determine if these doses could affect nephrotoxicity. The estimated glomerular filtration rate (eGFR) was calculated to assess kidney function by using the abbreviated MDRD study equation [eGFR = 175 × SCr -1.154 × Age -0.203 × Sex (0.742 for female, 1 for male) × Race (1.21 for black, 1 for nonblack)] [28].
Nephrotoxicity by colistin was originally defined using the KDIGO 2012 guideline [29]. However, to reflect clinical situations that SCr is elevated, but kidney function is normal, nephrotoxicity was defined by SCr exceeding 2 mg/dL (176.8 μmol/L) after colistin use when patients' baseline eGFR was higher than 60 mL/min/1.73 m 2 [30]. Patients with baseline eGFR less than 60 mL/min/1.73 m 2 , and elevated SCr by more than 1.5 times control, or 0.3 mg/dL, were considered to have colistin-induced nephrotoxicity per KDIGO 2012 guideline [29]; All subjects that underwent dialysis after colistin use were also considered to have nephrotoxicity.

Statistical analysis
Statistical analyses were performed using SPSS ver. 23 (IBM corporation, Armonk, New York, U.S.). All tests were two-sided and the significance level was set at 0.05 (α = 0.05). Continuous variables were presented as mean and standard deviation (SD), while categorical variables were presented as number (n) and proportion (%). Nephrotoxicity was evaluated based on AKI incidences resulting from IV or AS colistin administration, and the risk factors for IV or AS colistin-associated nephrotoxicity were analyzed. An independent t-test was used for continuous variables and Fisher's exact test was used for categorical variables. Binary multivariate logistic regression was used to evaluate risk factors for colistin-associated nephrotoxicity. Covariates that showed statistical significance in univariate logistic regression were selected for multivariate logistic regression.

Patient characteristics
A total of 893 subjects were screened. In total, 385 subjects were excluded for the following reasons (Fig 1): outpatient status (n = 3), age less than 18 years (n = 147), previous use of colistin or concurrent use of IV and AS colistin (n = 80), dialysis prior to colistin use (n = 44), and colistin use for less than 3 days (n = 155). Thus, 464 subjects were included in the study and were grouped as follows: 311 subjects were treated with IV colistin and 153 subjects were treated with AS colistin. Colistin was used when other antibiotics were ineffective and culture results were sensitive to colistin. Most patients were in sepsis. The baseline characteristics of the patients treated with IV and AS colistin are summarized in Table 1

Kidney function assessment
AKI developed in 63 (20.26%) of the 311 patients in the IV group and 12 (7.84%) of the 153 patients in the AS group (Fig 2). AKI was significantly more frequent in the IV group than in the AS group (p-value < 0.001).

Risk factors for IV and AS colistin-associated nephrotoxicity
Binary univariate and multivariate logistic regression analyses were performed to determine the risk factors for IV and AS colistin-associated nephrotoxicity (Tables 2 and 3). Age, sex, colistin administration route, height, ABW, IBW, duration of colistin use, baseline SCr, baseline eGFR, baseline BUN, accumulated dose, daily dose, daily dose per ABW, daily dose per IBW, hypertension, diabetes mellitus, chronic kidney disease (CKD), number of nephrotoxins, and each individual nephrotoxin were analyzed as the risk factors.
In the AS group, there were no significant nephrotoxicity-associated risk factors as determined by multivariate analysis. Angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), and cyclosporine were not analyzed due to the low number of cases of use concurrent with colistin. However, duration of colistin use, baseline SCr, baseline eGFR, and baseline BUN were determined to be significant risk factors using univariate analysis.

Comparison of AKI and non-AKI patients
Between AKI and non-AKI patients, duration of colistin use, kidney function, and accumulated dose were significantly different. In total subjects, the AKI group used colistin for one week longer than the non-AKI group (18.77 ± 18.92 vs 11.79 ± 9.75 days, p-value < 0.001). Factors associated with kidney function were worse in the AKI group (AKI vs non-AKI;  blood urea nitrogen; NSAIDs, non-steroidal anti-inflammatory drugs; ACE inhibitors, acetylcholinesterase inhibitors; ARBs, angiotensin receptor blockers a p-value was calculated by logistic regression using SPSS ver.23 (IBM corporation, Armonk, New York, U.S.) b ACE inhibitors, ARBs, and cyclosporine and were not analyzed due to very low number of cases.

Discussion
Use of AS colistin for improved efficacy and reduction of systemic toxicity in pulmonary infection has increased. However, studies of nephrotoxicity resulting from AS colistin treatment are sparse, and those that were reported contained only a small number of patients [15,16,18,19]. In one study, AS colistin monotherapy resulted in less nephrotoxicity than IV colistin monotherapy [20]. However, nephrotoxicity associated factors were not evaluated. In our study, the theoretical safety of AS colistin with regard to nephrotoxicity was evaluated and compared to nephrotoxicity resulting from IV colistin treatment. In addition, we attempted to elucidate risk factors that can affect AS or IV colistin-induced nephrotoxicity.
In this study, we defined nephrotoxicity as follows: SCr elevated more than 1.5 times or 0.3 mg/dL in patients having baseline eGFR less than 60 mL/min/1.73 m 2 (compared to the control), according to KDIGO 2012 guidelines [29]; SCr exceeding 2 mg/dL (176.8 μmol/L) after colistin use in patients having baseline eGFR higher than 60 mL/min/1.73 m 2 [30]. We defined patients with nephrotoxicity as having baseline eGFR higher than 60 mL/min/1.73 m 2 because factors that affect renal function could not be precisely controlled because this was a retrospective study. In addition, we felt that simply increasing the SCr threshold by 50% in patients with normal kidney function would be unlikely to be representative of true nephrotoxicity. Despite a 1.5-times increase in SCr in the normal kidney function group, many patients presented with low Scr. Individuals with low SCr were not considered to have impaired renal function and were not treated in clinical practice. According to these standards, the results of our study demonstrated that nephrotoxicity occurred twice as frequently with IV versus AS colistin treatment (20.26% vs 7.84%; p-value < 0.001, Fig 2). Few studies have compared nephrotoxicity between IV colistin monotherapy and AS colistin monotherapy. One retrospective observational study showed no nephrotoxic adverse events with AS colistin treatment. However, only six patients received AS colistin [18]. In another retrospective study, nephrotoxicity incidence was 4 times higher with IV colistin compared to AS colistin [20]. These studies agree with the results of our retrospective study. In the multivariate analyses performed to determine risk factors for AS colistin-associated nephrotoxicity, no significant risk factors were identified. However, duration of colistin use, baseline SCr, baseline eGFR, and baseline BUN were associated with nephrotoxicity resulting from AS colistin treatment as determined by univariate analyses (Table 3). Similarly, these factors were also significantly associated with nephrotoxicity in the univariate analysis of IV colistin. This indicates that duration of colistin use and baseline kidney function may be associated with colistin-induced nephrotoxicity.
IV colistin-associated nephrotoxicity risk may increase with duration of colistin use. A second risk factor for IV colistin-associated nephrotoxicity was the presence of CKD. The OR of CKD was high at 2.710 (range, 1.348-5.448), meaning that patients with CKD have a high risk of nephrotoxicity. As such, colistin should be used with caution in CKD patients. Contrary to our findings, previous studies have described old age, longer duration of colistin use, diabetes mellitus, high dose per IBW, and septic shock as risk factors [10,31,32].
Several studies have examined association of IV colistin-associated AKI with nephrotoxins. Use of vancomycin, glycopeptides, NSAIDs, loop diuretics, and rifampin were reported to induce IV colistin-associated AKI [7,10,[31][32][33]. Furthermore, co-administration of more than three nephrotoxins was reported to increase the risk of nephrotoxicity [24]. However, an increase in the number of nephrotoxins did not increase risk in our study. However, univariate analysis showed a significant increase in IV colistin-associated nephrotoxicity with co-administration of NSAIDS. Thus, additional studies on nephrotoxicity resulting from co-administration of NSAIDs and colistin are needed to clarify this relationship.
Several studies have examined dose-dependence of IV colistin-induced nephrotoxicity. Kwon et al. [34] reported that dose per IBW was related to nephrotoxicity risk in patients administered IV colistin. Lee et al. [7] reported that dose per IBW was a risk factor only in patients with impaired kidney function. Two other studies found that use of more than 5 mg/ kg/day (IBW) IV colistin increased nephrotoxicity risk [10,32]. However, dose was not related to colistin-associated nephrotoxicity in our study. Additionally, old age, prolonged colistin administration, hypoalbuminemia, low serum albumin level, high Charlson Comorbidity Index, and the presence of septic shock were reported to be related to nephrotoxicity [7,8,31,32,34].
Comparison of the AKI group and non-AKI group showed that the AKI group used colistin for one week longer. From these results, it could be inferred that duration of colistin use affects nephrotoxicity. Accumulated dose is also significantly higher in the AKI group, but this may be due to the longer use of colistin in AKI. These results showed the same trend with AS colistin administration (Table 4). Colistin is not commonly used in early phases of treatment. Colistin is used when other antibiotics have failed or colistin is suitable based on a pathogen susceptibility test. It is possible that colistin was used in many cases because it was the only alternative. As such, it is possible that colistin was used for long periods of time despite symptoms of nephrotoxicity because there are no other viable options or treatment.
Our study has some limitations that merit discussion. As this was a retrospective study, inaccurate patient information may have been provided. If data were missing, data collected most closely in time to the missing data were used. Furthermore, the mean value was used if there were multiple values obtained for the same factor on the same day. In addition, Patient clinical status was not evaluated using an APACHE II score, which would have been helpful in assessing kidney function and risk of AKI. Despite these limitations, this study, to our knowledge, is the first to attempt to determine risk factors associated AS colistin-induced nephrotoxicity. Furthermore, this is the first study to evaluate nephrotoxicity risks between IV and AS colistin treatment in a large number of patients.

Conclusion
From the analysis, duration of colistin use and baseline kidney function may be associated with AS colistin-associated nephrotoxicity. Therefore, to reduce nephrotoxicity, short term use of AS colistin and evaluation of baseline kidney function prior to treatment are recommended.
Supporting information S1 Dataset. Colistin_Data.sav. All data for this study were included and personal information were anonymous.