Genetic predisposition and high exposure to colistin in the early treatment period as independent risk factors for colistin‐induced nephrotoxicity

Abstract

Colistin is known to cause nephrotoxicity due to its extensive reabsorption and accumulation in renal tubules. In vitro studies have identified the functional role of colistin transporters such as OCTN2, PEPT2, megalin, and P‐glycoprotein. However, the role of these transporter gene variants in colistin‐induced nephrotoxicity has not been studied. Utilizing targeted next‐generation sequencing, we screened for genetic polymorphisms covering the colistin transporters (SLC15A1, SLC15A2, SLC22A5, LRP2, and ABCB1) in 42 critically ill patients who received colistimethate sodium. The genetic variants rs2257212 ((NM_021082.4):c.1048C>G) and rs13397109 ((NM_004525.3):C.7626C > T) were identified as being associated with an increased incidence of acute kidney injury (AKI) on Day 7. Colistin area under the curve (AUC) was predicted using a previously published pharmacokinetic model of colistin. Using logistic regression analysis, the predicted 24‐h AUC of colistin was identified as an important contributor for increased odds of AKI on Day 7. Among 42 patients, 4 (9.5%) were identified as having high predisposition to colistin‐induced AKI based on the presence of predisposing genetic variants. Determination of the presence of the abovementioned genetic variants and early therapeutic drug monitoring may reduce or prevent colistin‐induced nephrotoxicity and facilitate dose optimization of colistimethate sodium.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Colistin is known to cause acute kidney injury (AKI) due to its extensive reabsorption and accumulation in renal tubules. Studies have identified the functional role of colistin transporters such as OCTN2, PEPT2, megalin, and P‐glycoprotein.

• WHAT QUESTION DID THIS STUDY ADDRESS?

In this study we investigated the role of transporter gene variants and systemic exposure to colistin in causing AKI within 7 days of initiating treatment with colistimethate sodium.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

Presence of genetic variants rs2257212 and rs13397109 may increase the risk for colistin‐associated AKI. Systemic exposure to colistin (as 24‐h colistin AUC) in the early treatment period (within 6 days) is another factor that may be independently associated with colistin‐induced AKI.

• HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

The toxicodynamics of colistin in causing nephrotoxicity may be affected by the polymorphisms affecting renal transporters. Determination of polymorphisms in renal transporters along with estimating 24‐h colistin AUC in the early treatment phase may help to prevent colistin‐induced AKI.

INTRODUCTION

Colistin is an antibiotic of last resort for resistant Gram‐negative infections. It is increasingly used in critical care units due to the widespread antibiotic resistance towards carbapenems. ¹ Colistin is administered to patients as colistimethate sodium, a prodrug, and a significant portion of the administered colistimethate sodium is eliminated through the kidney. Colistimethate sodium undergoes spontaneous hydrolysis to form colistin in the body. ² Colistin is extensively reabsorbed in renal tubules. ³ , ⁴ Many patients develop acute kidney injury (AKI) after colistin administration, and this may be attributed to various factors such as age, sex, co‐administered nephrotoxic drugs, underlying chronic kidney disease, and cumulative exposure to colistin. ⁵ , ⁶ , ⁷ , ⁸

The extensive reabsorption of colistin in renal tubules and its active transport out of the tubular lumen is caused by tubular transporters such as OCTN2, PEPT2, megalin, and P‐glycoprotein. ⁹ , ¹⁰ , ¹¹ , ¹² The functional role of these transporters in colistin disposition is characterized by in vitro studies. However, to the best of our knowledge, the influence of genetic variants affecting these transporters on the incidence of AKI with colistimethate sodium treatment has not yet been studied. Identifying these genetic variants and determining their cumulative influence in critically ill patients will also help to delineate the role of other predisposing factors in the development of AKI while on treatment with colistimethate sodium.

In this study we quantified and predicted the exposure to colistin over the initial 6 days of treatment with colistimethate sodium, screened for genetic variants of the transporters, OCTN2, PEPT1, PEPT2, megalin, and P‐glycoprotein which were identified to influence renal handling of colistin, determined the baseline characteristics of patients, and studied their collective influence on the incidence of AKI on Day 7 (AKI_{Day 7}) of colistimethate sodium treatment.

METHODS

Patient recruitment

Critically ill patients older than 18 years of age, admitted to the surgical intensive care unit from April 2016 to February 2022, and who were administered intravenous colistimethate sodium were recruited for this study. The Institutional Ethics Committee approved this study (IRB Min No. 13463) for patient recruitment after obtaining written informed consent from patients or from their legally acceptable representatives. Quantification of colistin was performed using a liquid chromatography with tandem mass spectrometry (LC–MS/MS) method described previously, ¹³ where the colistin subcomponents A and B were quantified separately and summed to determine the total colistin concentration in milligrams per liter (mg/L). A minimum of two blood samples (2 mL each) were collected at convenient timepoints for measuring colistin concentrations, and one sample (2 mL) to determine the genetic variants. AKI_{Day 7} was calculated using Kidney Disease|Improving Global Outcomes (KDIGO) AKI staging. A KDIGO stage of ≥1 on Day 7 was used to categorize patients into those who developed and those who had not developed AKI. Information collected from the patient's bedside included Acute Physiology and Chronic Health Evaluation II (APACHE II) score, age, sex, creatinine clearance calculated using the Jelliffe–Jelliffe equation, serum albumin, Sequential Organ Failure Assessment (SOFA) score on Days 3 and day 7, the dose of colistimethate sodium, time, and duration of infusion and presence of other nephrotoxic drugs. The list of co‐administered drugs is given in Data S1. The Jelliffe–Jelliffe equation was used to estimate creatinine clearance since this study was performed in critically ill patients with AKI. ¹⁴ Patients who developed AKI before colistimethate sodium was administered were excluded. The treating clinicians were informed about the measured and predicted colistin concentrations, and the decision on dose modification was based on the clinician's discretion.

Prediction of 24‐h colistin AUC _{Day 6}

Colistin exposure was predicted from the measured concentrations using our previously validated pharmacokinetic model ¹³ included in the BestDose™ software ¹⁵ (https://www.lapk.org/training.php, available online on request) on a research basis and was used for the Bayesian prediction of colistin concentrations. Information such as age, sex, serum albumin, and creatinine clearance calculated using the Jelliffe–Jelliffe equation, the dose of colistimethate sodium, time and duration of infusion, and measured colistin concentrations were used for the prediction of colistin area under the curve (AUC).

Genetic screening

DNA extraction

Whole blood (2 mL) EDTA samples were collected from the patients, and to extract DNA the Maxwell® 16 Blood DNA Purification Kit with the Maxwell® 16 Instrument was used. Following quantification the samples were stored at −20°C.

Target enrichment

A custom targeted gene panel was designed using Ampliseq technology covering the following gene panel SLC22A5 (OCTN2), SLC15A1 (PEPT1), SLC15A2 (PEPT2), ABCB1 (P‐glycoprotein), and LRP2 (megalin). Target amplification was carried out using Ion AmpliSeq Library kit2.0. with 10–15 ng/μL of DNA.

Next‐generation sequencing library preparation and sequencing

The amplified polymerase chain reaction (PCR) products as two pools were combined and carried forward for library preparation. These steps include partial digestion of the PCR primer and adapter‐ligation with Ion Xpress™ barcoded adapters. Following this the samples are purified with Agencourt™ AMPure beads. For template preparation, an equimolar concentration of libraries was used for emulsion PCR on Ion OneTouch™ OT2, followed by the enrichment of ion sphere particles on Ion OneTouch ES. The enriched template was further loaded on to an Ion Torrent Personal Genome Machine next‐generation sequencing system.

Data analysis

Preliminary and coverage analysis were performed with the Ion Torrent suite software. Further, variant calling and annotation were also carried out using DNASTAR software. Variants classification for novel or rare variants was based on the American College of Medical Genetics and Genomics (ACMG) 2015 guidelines.

Determination of factors associated with acute kidney injury (AKI _{Day 7})

A Fisher's exact p‐value threshold of ≤0.05 was used to identify single nucleotide polymorphisms (SNPs) which were homozygous that had an association with AKI_{Day 7}. A Pearson correlation coefficient cutoff of >0.8 was used to identify linkage disequilibrium among the identified SNPs. SNPs at coding regions which were non‐synonymous were given preference compared with SNPs from non‐coding regions or synonymous variants in their association with AKI, in case of linkage disequilibrium, and were chosen for further analysis. Multiple testing p‐value correction was not done while performing the abovementioned statistical tests.

To determine other factors associated with AKI_{Day 7}, simple logistic regression was performed and those which had a p < 0.10 were chosen for multiple logistic regression. Factors which were considered for logistic regression includes APACHE II score, age, sex, baseline creatinine clearance calculated using the Jelliffe–Jelliffe equation, baseline serum albumin, SOFA score on Day 3, dose of colistimethate sodium on Days 1 and day 6, genetic variants which were identified based on the abovementioned method, and presence of drugs which can increase serum creatinine. The co‐administered drugs which can increase serum creatinine include nephrotoxic drugs, drugs which can decrease creatinine excretion, and drugs which can enhance creatinine production. The model was assessed using the pseudo‐R ² statistic quantified by McFadden's and Tjur's method. The variance inflation factor was determined for each of these variables. The model which gave the least Akaike information Criterion (AIK) value, best pseudo‐R ², and had a variance inflation factor of one for all parameters, was selected as the final model. Genetic variants retained in the final model were considered as important predictors of AKI_{Day 7} and those patients who had those variants were considered as having a high genetic predisposition to AKI.

ROC threshold cutoff for predicted colistin AUC _{day 3} and predicted colistin AUC _{day 6}

The receiver operator characteristics (ROC) curve was determined to identify the cutoff of predicted 24‐h colistin AUC_{Day 3} and predicted 24‐h colistin AUC_{Day 6} for the incidence of AKI_{Day 7}.

RESULTS

From the critically ill patients who received colistimethate sodium, 68 patients were screened for the study. Among these individuals, 14 patients who had AKI even before administration of colistimethate sodium were excluded (Figure 1). The characteristics of the participants are given in Table 1. None of the participants recruited into the study underwent dialysis, at least within the first 7 days of treatment with colistimethate sodium. None of the participants recruited received aminoglycoside antibiotics or other highly nephrotoxic drugs prior to the administration of colistimethate sodium. Among the 54 participants finally recruited into the study, genetic variants were screened and analyzed for 42 participants and colistin exposure was determined in 48 participants. Only in 36 of 54 participants were both colistin exposure and genetic variants determined. Among the 42 participants in whom genetic screening was performed six developed AKI_{Day 7} (14.3%), and three other participants had AKI_{Day 7} among those whom genetic screening was not performed.

FIGURE 1.

Flowchart of patient recruitment. AKI, acute kidney injury.

TABLE 1.

Characteristics of the study population.

Clinical parameter	Median (IQR)
Male:Female	38:16
Age (years)	40.0 (28.3–54.8)
APACHE II score	17.0 (13.0–23.0)
Creatinine clearanceBaseline (mL/min/1.73 m2)	90.0 (49.3–129.3)
Serum albuminBaseline (g/dL)	2.3 (1.9–2.7)
CMS doseDay 1 (MU)	15.0 (12.7–15.0)
CMS doseDay 6 (MU)	9.0 (6.0–9.0)
SOFADay 3	4.0 (3.0–7.0)
SOFADay 7	3.0 (2.0–6.0)
Predicted 24‐h colistin AUCDay 6	43.6 (31.1–70.2)

Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; AUC, area under the curve; CMS, colistimethate sodium; IQR, interquartile range; MU, Million Units; SOFA, Sequential Organ Failure Assessment score.

A significant but poor negative correlation (r) of −0.34 (p = 0.02, Pearson correlation) was observed between the baseline creatinine clearance calculated using the Jelliffe–Jelliffe equation and the predicted 24‐h colistin AUC_{Day 6} (Figure 2).

FIGURE 2.

Correlation between Jelliffe–Jelliffe calculated baseline creatinine clearance and the predicted 24‐h colistin area under the curve (AUC) on Day 6.

Of the 182 genetic variants which were identified, 12 were found to be significantly associated with AKI (Table 2). Among these, we identified linkage disequilibrium among rs2257212, rs2257214, rs1143670, rs1143671, rs874741, rs7637569, rs2293616, rs1316397, rs3817601, and rs2257109 (Pearson r > 0.8). Since rs2257212 (a non‐synonymous SNP in the coding region) was known to be functionally important, ¹⁶ only this SNP was retained for further analysis. Two SNPs which were not in linkage disequilibrium were also considered for further analysis.

TABLE 2.

Single nucleotide polymorphisms associated with acute kidney injury on Day 7 (AKI_{Day 7}).

Variant	HGVS nomenclature	Gene	Region	P‐value a	Patients (of 55) (n)	Final selection
rs2257212	(NM_021082.4):c.1048C>G	SLC15A2 (PEPT2)	Coding (non‐synonymous)	0.002	3	Yes
rs13397109	(NM_004525.3):c.7626C>T	LRP2 (megalin)	Coding (synonymous)	0.02	2	Yes
rs1339067	(NM_005073.4):c.1347T>G	SLC15A1 (PEPT1)	Coding (synonymous)	0.02	16	No
rs1143671	(NM_021082.4):c.1225C>A	SLC15A2 (PEPT2)	Coding (non‐synonymous)	0.02	2	No
rs1316397	(NM_021082.4):c.1342‐47C>G	SLC15A2 (PEPT2)	Non‐coding	0.02	2	No
rs3817601	(NM_021082.4):c.958‐21A>C	SLC15A2 (PEPT2)	Non‐coding	0.02	2	No
rs2257109	(NM_021082.4):c.1003‐28A>C	SLC15A2 (PEPT2)	Non‐coding	0.002	3	No
rs2257214	(NM_021082.4):c.1124+13T>A	SLC15A2 (PEPT2)	Non‐coding	0.002	3	No
rs1143670	(NM_021082.4):c.1161A>G	SLC15A2 (PEPT2)	Coding (synonymous)	0.002	3	No
rs874741	(NM_021082.4):c.1762‐114A>C	SLC15A2 (PEPT2)	Coding (non‐synonymous)	0.02	2	No
rs7637569	(NM_021082.4):c.781‐94G>A	SLC15A2 (PEPT2)	Coding (non‐coding)	0.02	2	No
rs2293616	(NM_021082.4):c.852G>A	SLC15A2 (PEPT2)	Coding (synonymous)	0.02	2	No

Abbreviation: HGVS, Human Genome Variation Society.

^a Fisher's exact test p‐value.

Simple logistic regression was performed to determine other factors associated with AKI_{Day 7}. Only predicted 24‐h AUC_{Day 6} of colistin (p = 0.05) and baseline serum albumin (p = 0.06) were moderately associated (p < 0.10) with AKI_{Day 7}. Multiple logistic regression model details are described in Table 3. The final selected model had a variance inflation factor of 1.0 for all the predictors.

TABLE 3.

Multiple logistic regression: comparison among models with lowest Akaike information criterion values.

Parameter	OR	P‐value	VIF
Model 1 (final selected)
Intercept	0.01	0.003
rs2257212	4.1*109	0.996	1.00
rs13397109	3.2*109	0.996	1.00
Predicted 24‐h AUCDay 6	1.02	0.09	1.00
Pseudo‐R 2
McFadden	0.62
Tjur (coefficient of discrimination)	0.67
AIC	21.51
Model 2 (lowest AIC)
Intercept	0.00	0.996
rs2257212	8.4*1017	0.997	1.19
rs13397109	3.3*109	0.998	1.00
rs1339067	3.3*108	0.997	1.19
Pseudo‐R 2
McFadden	0.68
Tjur (coefficient of discrimination)	0.67
AIC	20.24
Model 3 (full model)
Intercept	0.00	0.997
rs2257212	7.2*1016	0.997	1.22
rs13397109	1.2*1011	0.998	1.00
rs1339067	2.2*108	0.997	1.22
Predicted 24‐h AUCDay 6	1.02	0.19	1.44
Serum albumin	0.16	0.25	1.44
Pseudo‐R 2
McFadden	0.75
Tjur (coefficient of discrimination)	0.77
AIC	22.87

Abbreviations: AIC, Akaike information criterion; AUC, area under the curve; OR, odds ratio; VIF, variance inflation factor.

Of 42 participants, 4 (9.5%) were identified to have a high predisposition to colistin‐induced AKI. All of them developed AKI, and among them those who had a predicted 24‐h colistin AUC of >50 mg*h/L developed AKI of KDIGO Stage 2 and 3. There was no significant difference in the predicted 24‐h colistin_{Day 6} between participants with a low predisposition and high predisposition (Wilcoxon rank sum test, p = 1.0). Categorical analysis with the cumulative presence of two final selected SNPs (based on the model) estimated significantly higher risk (p < 0.001) of developing AKI_{Day 7} for participants in the high predisposition category (n = 4/4) compared with those in the low predisposition category (n = 2/38; Table 4). The two participants in the low predisposition group who developed AKI_{Day 7} had high exposure to colistin (predicted 24‐h AUC_{Day 6} >64.1 mg∙h/L), presence of drugs which can increase serum creatinine, or very low albumin (≤2.5 g/dL) on Days 3 and 6 of colistin administration or a combination of the abovementioned factors.

TABLE 4.

Clinical parameters in high versus low genetic predisposition to colistin‐induced acute kidney injury.

Clinical parameter	High predisposition (n = 4)	Low predisposition (n = 38)	P‐value (Wilcoxon rank sum test)
Male:Female	3:1	27:11	1.0 a
Age (years) (median, IQR)	47 (38.8–55.5)	36 (26.3–52.5)	0.29
APACHE II score (median, IQR)	26.5 (25.8–27.3)	16 (13.0–23.0)	0.06
Creatinine clearanceBaseline (mL/min/1.73 m2) b (median, IQR)	93.5 (67.3–115.0)	90.0 (51.3–129.3)	0.76
Serum albuminBaseline (g/dL) (median, IQR)	2.1 (1.6–2.7)	2.5 (2.1–2.8)	0.44
CMS doseDay 1 (MU) (median, IQR)	15.0 (12.8–15.0)	15.0 (12.6–15.0)	0.64
CMS doseDay 6 (MU) (median, IQR)	9.0 (4.5–10.5)	9.0 (6.0–9.0)	0.61
Predicted 24‐h colistin AUCDay 6 (median, IQR)	40.2 (24.5–58.4)	38.1 (23.7–59.9)	1.00
SOFADay 3	7.5 (6.0–8.3)	3.0 (2.0–5.8)	0.10
SOFADay 7	5.5 (1.5–9.8)	2.0 (1.0–4.0)	0.44
KDIGO scoreDay 7	1.5 (1.0–2.3)	0.0 (0.0–0.0)	<0.001
Number of nephrotoxic drugs	1.5 (1.0–2.0)	1.0 (1.0–1.0)	0.16

Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; AUC, area under the curve; CMS, define; IQR, interquartile range; KDIGO, Kidney Disease|Improving Global Outcomes; MU, define; SOFA, Sequential Organ Failure Assessment score.

^a Fisher's exact test.

^b Creatinine clearance was calculated using the Jelliffe–Jelliffe equation.

In participants with low genetic predisposition, the ROC cutoff was 64.1 mg∙h/L for predicted 24‐h colistin AUC_{Day 6}, with 83% specificity and 100% sensitivity to predict AKI_{day 7} (AUROC = 0.90). For earlier days of treatment with colistimethate sodium (Day 3) the threshold cutoffs for predicted 24‐h colistin AUC_{Day 3} was 74.93 mg∙h/L with 85% specificity and 100% sensitivity (AUROC = 0.89) for the low predisposition group. Since the number of participants who developed AKI was small, this cutoff value should be confirmed in larger clinical trials.

DISCUSSION

This study was performed on 54 critically ill patients who received intravenous colistimethate sodium for clinically suspected sepsis. Colistin exposure was reported as 24‐h predicted AUC, and any decision on dose modification was based on the clinical judgment of the physician. Among the participants, 9 (16.6%) developed AKI on Day 7. Among the factors studied, only genetic predisposition and colistin exposure measured as predicted 24‐h AUC_{Day 6} were identified to be associated with AKI.

Baseline renal function was identified as a significant predictor of colistin‐induced AKI in previous studies. ¹⁷ We observed a significant but poor negative correlation between baseline creatinine clearance and predicted 24‐h AUC_{Day 6} of colistin. However, in the regression analysis, baseline creatinine clearance calculated using the Jelliffe–Jelliffe equation was not identified as a significant predictor of AKI. This could be attributed to the toxicodynamics of colistin‐induced nephrotoxicity, which is more dependent on the tubular reabsorption process rather than the simple glomerular filtration of colistin.

Colistin is eliminated through glomerular filtration and the colistin formed from colistimethate sodium in the renal tubules is almost completely reabsorbed by the proximal convoluted tubules. ³ , ⁴ , ¹⁸ Various studies have implicated the role of transporters such as P glycoprotein, peptide transporter (PEPT2), OCTN2, and megalin in handling colistin in the renal tubules. ⁹ , ¹⁰ , ¹¹ , ¹² We identified two polymorphisms [rs13397109, affecting LRP2 and rs2257212, affecting PEPT2] which may increase the risk of AKI. The polymorphism rs2257212 has previously been shown to improve the response to sorafenib treatment in patients with hepatocellular carcinoma. ¹⁶ This polymorphism may increase the stability of SLC15A2 protein and thus may contribute to increased reabsorption of colistin in the renal tubule and thus nephrotoxicity. We could not identify from the literature any other polymorphisms (from among the ones which we have identified) which could have caused a functional influence on these transporters. The predicted Day 6 colistin AUC among participants with a high and low predisposition to colistin‐induced AKI was similar (Wilcoxon rank sum test, p = 1.00). Nevertheless, all participants in the high predisposition category had AKI on Day 7 (100.0%) compared with those in the low predisposition category (5.3%) (p < 0.001). Thus, the presence of harmful genetic variants seems to have imposed lower tolerance to colistin exposure. This needs to be confirmed in larger studies performed on critically ill patients who are administered colistin.

One major drawback of our study is the absence of a validation group to confirm the association between AKI_{Day 7} and the identified SNPs (rs13397109 and rs2257212). This was not possible as the prescribing pattern was modified in the hospital and colistin was being replaced by polymyxin. Since the study was exploratory in nature, p value correction for multiple testing was not performed for any of the statistical tests mentioned in this study. The association between the aforementioned SNPs and AKI_{Day 7} needs to be confirmed in future prospective clinical trials. We will confirm the presence of the previously mentioned three polymorphisms using Sanger sequencing. Polymyxin B also has a high nephrotoxic potential. ¹⁹ Since both colistin and polymyxin B have similar chemical structures, both may be handled by similar transporters ²⁰ , ²¹ and therefore the identified variants may also be associated with polymyxin B‐induced renal injury.

Hypoalbuminemia is another factor proposed to be associated with AKI during colistin treatment. ²² However, in our population, simple logistic regression analysis did not identify serum albumin as a significant clinical parameter affecting AKI_{Day 7} (p = 0.06). Including the baseline serum albumin in logistic regression worsened the Akaike Information criteria (with vs without albumin: 22.9 vs. 21.7), suggesting an independent role of the SNPs and colistin exposure in causing the AKI. However, the role of serum albumin in colistin‐induced AKI needs to be confirmed in large clinical trials. It would be prudent to be cautious when administering colistin to patients with very low serum albumin (≤2.5 g/dL). Albumin in the peritubular capillaries may help with redistribution of accumulated colistin from renal tissue and thus normal serum albumin levels may prevent nephrotoxicity. Hypoalbuminemia may also cause accumulation of reactive oxygen species in renal tubules, which may lead to apoptosis of renal tubular cells predisposing the patient to AKI. ²³

Multiple studies performed in critically ill patients ¹⁷ , ²⁴ , ²⁵ have identified the role of colistin exposure as a risk factor for AKI. Systemic colistin exposure may not be affected by the presence of SNPs predisposing to nephrotoxicity due to the fast distribution of reabsorbed colistin from the renal tubules. A threshold cutoff of 64.1 mg∙h/L for the predicted 24‐h AUC_{Day 6} was identified to predict AKI_{Day 7} with reasonable sensitivity and specificity in patients with a low genetic predisposition to colistin‐induced AKI. The inter‐patient variability in predicted 24‐h AUC_{Day 6} was 78.6% among the patients recruited for the study. These findings call for therapeutic drug monitoring and early dose optimization of colistimethate sodium, as high exposure to colistin can predispose patients to AKI and low exposures may lead to therapeutic failure. We have observed a fall in threshold cutoffs for AKI, in the low predisposition group, on Day 6 of treatment compared with Day 3. For longer duration of treatment with colistimethate sodium, the colistin concentration threshold for colistin‐associated AKI may be lower.

Routine estimation of the genetic predisposition prior to initiating colistimethate sodium may not be feasible in all healthcare settings. But its role in treating those severely critically ill patients who have a potential for an AKI should not be undermined. This study supports the view that pharmacogenomics plays an independent role in pharmacodynamics and therefore the target concentrations of a drug need to be individualized based on genetic constitution. However, the genetic variants identified as high‐risk variants for AKI were not validated in a separate group of patients, and therefore we do not recommend the use of the proposed genetic variants for routine patient care, unless validated in the intended patient population. In conclusion, genetic variants were screened and assessed to determine the combined effect of genetic predisposition and colistin exposure on the incidence of AKI. To our knowledge, this is the first study to determine the genetic predisposition of critically ill patients towards colistin‐induced AKI. The combined effect of genetic variants and the threshold cutoff for colistin AUC_{Day 6} to predict colistin‐induced AKI needs to be confirmed in a future prospective study with a larger cohort of patients.

AUTHOR CONTRIBUTIONS

S.K.M., A.C., and M.N.N. wrote the manuscript. S.K.M., A.C., B.W.A., and B.S.M. designed the research. S.K.M., A.C., P.V., V.E., R.P., S.V.R., S.K., and B.S.M. performed the research. S.K.M., A.C., P.V., V.E., and B.W.A. analyzed the data. S.K.M., A.C. and M.N. contributed analytical tools.

FUNDING INFORMATION

This work was supported by the Indian Council of Medical Research (File No. AMR/Adhoc/239/2020‐ECD‐II). The funding source had no role in the study design; collection, analysis, and interpretation of the data; in the writing of the report; and in the decision to submit the article for publication. S.K.M. received the grant as principal investigator. The funders had no role in the study design, data collection, analysis, decision to publish, or preparation of the manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declared no competing interests for this work.

Supporting information

Data S1.

Mathew SK, Chapla A, Venkatesan P, et al. Genetic predisposition and high exposure to colistin in the early treatment period as independent risk factors for colistin‐induced nephrotoxicity. Clin Transl Sci. 2024;17:e13764. doi: 10.1111/cts.13764

DATA AVAILABILITY STATEMENT

Data available in the supplementary file.