Colistin-Associated Acute Kidney Injury in Burn Patients: Incidence, Associated Factors and Prognosis

SUMMARY

Colistin is often essential in burn patients presenting infections caused by multidrug-resistant organisms. A retrospective study was conducted in the intensive burn care department in Tunisia over 14 months to assess prevalence and predictive factors of colistin-related acute kidney injury (AKI). Among 70 patients (mean age 34±13 years), 28.6% developed AKI, with onset at 5±2 days. Associated factors included total burned surface area >41% (OR=2.6), UBS score >58 (OR=6.8), metabolic acidosis at admission (OR=28) and contrast agent injection (OR=107). AKI was reversible in 40% of cases. Colistin-associated AKI remains a frequent complication in severely burned patients, and in cases of contrast agent use.

Introduction

Colistin, an antibiotic from the polymyxin family, is effective against multidrug-resistant (MDR) Gram-negative bacilli (GNB) by destabilizing their cell membrane. Discovered in the 1940s, its use declined due, essentially, to its nephrotoxicity.¹ Mechanisms of renal damage are multiple, involving oxidative stress, inflammation and apoptosis in proximal tubular cells.²

Because of the emergence of MDR GNB such as Acinetobacter baumannii, Pseudomonas aeruginosa and Klebsiella pneumoniae, colistin has been reintroduced despite reports of nephrotoxicity in 10% to 60% of cases, influenced by factors such as dosage, age and comorbidities.^3-6 Among burn patients, who are highly susceptible to severe MDR GNB infections, the use of colistin is often unavoidable, even though this toxicity has been associated with increased mortality.⁷

The literature on acute kidney injury (AKI) in burn patients treated with colistin remains limited, despite an incidence as high as 38% reported in some studies.⁸ This study aims to determine incidence, associated factors of this complication, and its prognostic impact on this vulnerable population.

Methods

A monocentric, observational and retrospective study was conducted over 14 months (October 1, 2022, to November 30, 2023) in the Intensive Burn Care department in Tunisia.

Burn patients aged over 18 years who received colistin treatment for a minimum of 48 hours were included. Patients aged under 18 years, those with preexisting renal failure at the initiation of colistin, pregnant women, non-burned patients, and those with incomplete medical records were not included. After inclusion, demographic characteristics (age, sex, pre-existing comorbidity), clinical parameters (total burned surface area, type and location of burns, severity scores), and treatment-related parameters (indication for colistin, dosage, duration, co-administrated nephrotoxic agents, and therapeutic interventions) were recorded.

The primary endpoint was the occurrence of acute kidney injury (AKI), defined according to the RIFLE criteria, based on serum creatinine changes (Fig. 1). Secondary endpoints were risk factors for colistin-associated AKI and its prognostic impact. Biological parameters were monitored daily, and patients were followed for the reversibility of AKI, need for renal replacement therapy, or worsening outcomes. Statistical analyses included univariate and multivariate logistic regression to identify associated factors.

Fig. 1.

RIFLE classification

To identify associated factors, a comparison was made between patients who developed AKI (AKI+) and those who didn’t (AKI-). All statistical analyses were conducted using SPSS 26.0. Ethical standards were strictly observed, ensuring data anonymity and the confidentiality of patient information.

Results

During the study period, 464 patients were admitted for burns; among them 107 received colistin (23%). Seventy patients were included (Fig. 2).

Fig. 2.

Study flowchart

The mean age of our patients was 34±13 years, with a sex-ratio of 3,4. Obesity (14%), depression (13%) and diabetes (6%) were the most frequent comorbidities. Total burned surface area (TBSA) was 39±16%. The mean Baux Index and Unit Burn Standard (UBS) scores were 73±21 and 75±64, respectively.

More than two-thirds of the patients (n=54, 77%) were transferred from another hospital within a median of 48 [24-48] hours of burn onset. None of these patients had been treated with colistin before admission. Twenty-three percent (n=16) were admitted through the emergency department.

Burns were secondary to suicide attempt by self-immolation in 50% of cases (n=35), followed by domestic accidents in 40% of cases (n=28). Burns were thermal in most cases (94%, n=66), all caused by direct flames. Burns involved multiple body areas in 76% of cases (n=53), and 46 patients (66%) had burns affecting the face and neck. Deep burns were observed in most patients (74%, n=52).

Clinical and biological features on admission are presented in Table I and Table II.

Table I.

Clinical features on admission

Clinical features	Value
Glasgow coma scale (med, IQR)	15 [15-15]
Respiratory rate (cpm) (Med, IQR)	22 [20-24]
SpO2(%)(Med, IQR)	100 [98-100]
invasive mechanical ventilation on admission	36; 51
Systolic blood pressure (mmHg) (Med, IQR)	120 [110-140]
Diastolic blood pressure (mmHg) (Med, IQR)	70 [60-80]
Shock on admission (n; %)	21; 30
Type of shock:
Septic (n; %)	4; 6
Hypovolemic (n; %)	17; 24

Cpm: cycles per minute; Med: median; IQR:IQR:interquartile

Table II.

Initial laboratory results

Biological Parameters	Med, IQR
Leukocytes (cells/mm3)	19000 [13000-26000]
Hemoglobin (g/dL)	17 [15-19]
Hematocrit (%)	50 [44-56]
Urea (mmol/L)	5 [4-6]
Creatinine (μmol/L)	70 [54-83]
Proteins (g/L)	50 [44-54]
CPK (U/L)	409 [244-1590]
pH	7,33 [7,27-4,4]
PC02 (mmHg)	37 [31-44]
HC03- (mmol/L)	21 [17-24]
Pa02/Fi02 ratio	370 [188-428]

CPK: creatine phosphokinase: Med: median, IQR: Interquartile

The main biological abnormalities were hypoproteinemia (83%), hemoconcentration (70%), metabolic acidosis (43%) and rhabdomyolysis (26%).

Sepsis occurred with a delay of 7±4.5 days after burns. The sources of infection were cutaneous (46%), pulmonary (31%) and vascular (19%). The most common pathogens were Acinetobacter baumannii (50%), Klebsiella pneumoniae (14%) and Pseudomonas aeruginosa (13%). Colistin was prescribed for sepsis or septic shock, mainly empirically (59%). It was administered as a 9 MU loading dose, followed by 4,5 MU twice a day for 8±3 days, always associated with another antibiotic which was in most cases carbapenem (54%), aminoglycoside (16%) and teicoplanin (12%).

Twenty patients (28.6%) developed acute kidney injury (AKI) based on RIFLE criteria (Fig. 2). The average onset of AKI was 5±2 days after the initiation of colistin. According to the RIFLE classification, 40% of cases were classified as Risk, 35% as Injury, and 25% as Failure.

In univariate analysis, associated factors for AKI were age, IB score, UBS score, TBSA, high-voltage electrical injury, the occurrence of metabolic acidosis, hemoconcentration, and concomitant use contrast agent injection (Table III).

Table III.

Univariate analysis: associated factors for colistin-associated acute kidney injury

Variable	AKI–(n=50)	AKI + (n=20)	P
Age (mean ± SD)	32 ± 11	40 ± 16	0,027
Sex-ratio	3, 16	4,20	NS
IB score (mean ± SD)	67 ± 18	87 ± 20	<10-3
UBS score (mean ± SD)	64 ± 53	102 ± 83	0,023
Diabetes n (%)	2(4)	2(10)	NS
Hypertension n (%)	1(2)	3(15)	0,06
TBSA (mean ± SD)	36 ± 15	48 ± 14	0,004
High-voltage electrical injury n (%)	0	4(20)	0,005
Shock on admission n (%)	16(32)	5(25)	NS
Biological abnormalities on admission n (%)
Metabolic acidosis	16(32)	14(70)	0,007
Rhabdomyolysis	10(20)	8(40)	0,08
Hemoconceritration	31(62)	18(90)	0,04
Hypoproteinemia	43(86)	15(75)	NS
Intubation on admission n (%)	30(60)	6(30)	NS
Concomitant use of a nephrotoxic agent n (%)
Aminoglycoside	7(14)	4(20)	NS
Carbapenem	28(56)	9(45)	NS
Teicoplanin	5(10)	3(15)	NS
Acyclovir	1(2)	2(10)	NS
Fluconzole	9(18)	2(10)	NS
Contrast agent injection	1(2)	9(45)	<10-3
Use of vasopressors n (%)	17(34)	10(50)	NS
Duration of treatment with colistin (mean ± SD)	7,8 ± 2,5	8,2 ± 2,4	NS

SD; standard derivation: NS: not significant; TBSA; total burned surface area: AKI: acute kidney injury

In multivariate analysis, associated factors were: TBSA >41%; UBS score >58; Metabolic acidosis on admission and concomitant use of contrast agent injection (Figs. 3 to 6) (Table IV).

Fig. 3.

ROC curve of age as a predictive factor for colistin-related acute kidney injury

Fig. 4.

ROC curve of Total Burned Surface Area as a predictive factor for colistin-related acute kidney injury

Fig. 5.

ROC curve of IB score as a predictive factor for colistin-related acute kidney injury

Fig. 6.

ROC curve of UBS score as a predictive factor for colistin-related acute kidney injury

Table IV.

Multivariate analysis: independent associated factors of acute kidney injury induced by colistin

Risk factors	OR	CI 95%	P
Inravenous contrast agent use	107	2,9-600	0,011
Metabolic acidosis on admission	28	2,7-286	0,007
UBS score > 58	6,8	1,5-47	0,049
TBSA > 41%	2,6	1,5-23	0,004

TBSA: total burned surface area; OR: odds ratio: CI: Confidence interval

Colistin was stopped in four patients, and dose adjustments were made for 16 others based on creatinine clearance. AKI was reversible in eight patients (40%) with normalization of renal function with a delay of 5±2.5 days.

Renal replacement therapy (RRT) was required in 12 patients, but three showed rapid deterioration. RRT was contraindicated in nine cases due to coagulation disorders. In our series, mortality was 60%. AKI was associated with a higher mortality (AKI+: 60% versus AKI-: 44%; p=0.04).

Discussion

Colistin causes cell death by binding the lipopolysaccharide of gram-negative bacteria cell walls, increasing their permeability. Similarly, renal toxicity occurs through increased permeability of renal tubular epithelium, leading to cellular lysis and acute tubular necrosis.⁹ In the literature, few studies dealing with this complication in burn patients have been published. An American study reported an incidence of 38% of AKI associated to colistin in burn patients, slightly higher than in our study (23%).⁸

Incidence of colistin nephrotoxicity ranges from 35% to 70%,^10-12 variability explained by differences in patient populations and definitions used for AKI (e.g., RIFLE vs. KDIGO). Both RIFLE and KDIGO classifications are commonly used to stage colistin-associated AKI, and we adopted RIFLE due to its widespread use in published studies.^3,13-15

In our study, colistin was prescribed mainly for treating sepsis from cutaneous or pulmonary sources, with the most common pathogens being Acinetobacter and Pseudomonas aeruginosa, which is comparable to the literature.^16,17 The mean onset of AKI in our study was 5±2 days, whereas the literature reported a delay of 3¹⁸ to 10 days.¹⁹ Pogue²⁰ found that AKI was developed within the first 7 days in 78% of patients, Gunay²¹ within 9 days in 77%, and Deryke²² within the first 5 days in all patients.

Regarding associated factors for nephrotoxicity associated to colistin, studies published in the literature reported many factors:

Age: identified as a risk factor for AKI in our study, and in most studies.^13,16,23,24 However, Deniz³ and Hartzell²⁵ found no such relationship. Some studies suggest that being over 60 years old predisposes individuals to renal damage due to glomerulosclerosis, interstitial fibrosis, tubular atrophy, and decreased glomerular filtration rate.²⁶ In our study, an age above 35 was associated with AKI, likely due to the younger average age of our population (34 years).

Sex: despite known pharmacokinetic differences, sex was not identified as a significant risk factor for AKI in our study or in the literature.^10,19,20,25

Medical History: pre-existing medical conditions were not associated with AKI risk in our study, similar to results reported by Deniz³ and Razan,¹ who reported no correlation between diabetes, hypertension and AKI.

Severity scores: high severity scores at admission were associated with AKI. Ghafur⁷ reported higher APACHE II and Charlson scores in patients with AKI. Similarly, in our study, burn severity scores (UBS and IB) were linked to AKI risk.

Total burn surface area (TBSA): In our study, TBSA >41% was predictive of renal injury with colistin. Literature associates extensive TBSA with acute renal injury (AKI), whether drug-related or not. Gharsallah and Clark linked AKI in burn patients to cellular lysis releasing substances like myoglobin that alter renal tubular function.^27,28 Wu²⁹ and Folkesdad³⁰ identified extensive TBSA as a predictor of AKI in this population, increasing the risk 15.6 and 4.7 times, respectively.

Initial hemoconcentration and metabolic acidosis were identified as associated factors with AKI. Metabolic acidosis was an independent risk factor, highlighting the importance of early and adequate fluid resuscitation.³¹ However, initial renal failure at admission was not associated with AKI, contrary to Deniz’s findings.

Concomitant nephrotoxic agents:

Antibiotics: we found no association between the concurrent use of other nephrotoxic antibiotics and nephrotoxicity with colistin. Most studies report no significant effect when colistin and aminoglycosides are used together.^20,25 Doshi³² noted a trend towards significance for an association with aminoglycosides (P=0.069).

Contrast agents: we identified intravenous contrast injection as an independent risk factor for colistin-associated AKI (OR=79, p=0.004), consistent with Doshi’s findings.³² Other studies, however, did not find an association between contrast agents and nephrotoxicity.^15,22

- Vasopressors: we did not identify vasopressors as a risk factor for AKI. Some studies found no significant association, with only Deryke et al.²² reporting a significant link.

- Colistin dose: in our study, the colistin dose was standardized for all patients. Literature suggests that higher doses of colistin are associated with increased AKI risk.³³ However, a Spanish study did not find a link between dose and nephrotoxicity risk.¹⁴

- Treatment duration: the duration of colistin treatment was not associated with AKI in our study or in several others.^1,11

In our study, 40% of patients recovered normal renal function with dose reduction. However, 60% required renal replacement therapy (RRT), with 11.5% receiving it successfully. Mortality was 60% in those with AKI, primarily due to the severity of burns. Mortality related to AKI in the literature ranges from 45% to 70%,^20,24,33 and AKI was found to be a significant risk factor for mortality.^3,6,7

Our study reports an entity rarely described in burn patients with minimal missing data despite its retrospective nature. However, its limitations include monocentric design, the absence of pharmacovigilance to confirm causality, and the lack of nephrotoxic drug level monitoring, such as gentamicin. Multicenter studies and further research into protective therapies are necessary to validate and expand these findings.

Conclusions

Acute kidney injury is a common complication in severely burned patients in intensive care treated with colistin. This complication significantly increases the morbidity and mortality of these patients. Therefore, periodic assessment of serum creatinine levels, modification of the colistin dose according to renal function, shortening the duration of antimicrobial treatment and avoidance of coadministration of other nephrotoxic agents (if possible) could minimize potential risk for nephrotoxic effects of this valuable old antibiotic.