https://orcid.org/0000-0001-5403-4730
-
PDF
- Split View
-
Views
-
Cite
Cite
Madeline A Lipp, Ryan L Crass, Linda J Fitzgerald, Twisha S Patel, Richard H Simon, Blair E Lenhan, MeiLan K Han, Shijing Jia, Acute kidney injury in cystic fibrosis patients treated with intravenous colistimethate sodium or tobramycin, Journal of Antimicrobial Chemotherapy, Volume 77, Issue 9, September 2022, Pages 2516–2521, https://doi.org/10.1093/jac/dkac187
Close - Share Icon Share
Abstract
Colistimethate sodium and tobramycin are important systemic antibiotics for treatment of cystic fibrosis (CF) pulmonary exacerbations but can induce acute kidney injury (AKI). We characterize the rate of AKI in CF patients treated with systemic colistimethate sodium compared with tobramycin.
This single-centre, retrospective cohort study included hospitalized CF patients treated with IV colistimethate sodium or tobramycin. The primary outcome was AKI defined using the RIFLE criteria. Multivariate logistic regression using a mixed model was performed to identify variables that were independently associated with AKI.
Overall, 156 patients representing 507 care encounters were included. The OR of AKI was not increased with IV colistimethate sodium relative to IV tobramycin after adjusting for other potential predictor variables (aOR 1.00; 95% CI 0.16–6.03). The frequency of AKI was 9.5% across all encounters, 6.9% with IV colistimethate sodium and 9.9% with IV tobramycin, with RIFLE category R (risk) being the most common stage, accounting for 4.2% of encounters with IV colistimethate sodium and 9.2% with IV tobramycin. The concomitant use of another nephrotoxin (aOR 2.51; 95% CI 1.27–4.95) or the combination of vancomycin and piperacillin/tazobactam (aOR 5.95; 95% CI 2.05–17.3) were both associated with increased odds of AKI.
Systemic treatment with colistimethate sodium or tobramycin in the CF patient population is associated with a similar rate of nephrotoxicity. However, clinicians should be mindful of the increased risk for AKI in patients treated with either IV colistimethate sodium or IV tobramycin when used concurrently with other nephrotoxic agents, particularly the combination of vancomycin and piperacillin/tazobactam.
Introduction
Cystic fibrosis (CF) is characterized by repeated pulmonary exacerbations, in which treatment with antibiotics leads to bacterial infection with MDR organisms, such as Pseudomonas aeruginosa.1 IV therapy for MDR P. aeruginosa commonly consists of an antipseudomonal β-lactam in combination with either colistimethate sodium or tobramycin, both of which have maintained excellent in vitro activity against MDR P. aeruginosa, but are associated with nephrotoxicity.2–4 With progressive lung disease and repeated courses of antibiotic treatments, the use of these nephrotoxic IV antibiotics increases in the ageing adult population.1
Colistimethate sodium accumulation in the proximal renal tubules leads to a dose- and time-dependent acute tubular necrosis (ATN) that has been reported in up to 30%–60% of patients.5,6 Predictors for increased nephrotoxicity risk are not well described.7,8 While the majority (up to 90%) of patients with colistimethate sodium nephrotoxicity are reported to recover renal function, the effects of repeated courses of colistimethate sodium therapy on subsequent risk for acute or chronic renal dysfunction are not known. Tobramycin nephrotoxicity is primarily due to direct proximal tubular injury and also leads to ATN.9,10 The reported incidence of acute kidney injury (AKI) is much lower, but not negligible, in the general patient population compared with colistimethate sodium and ranges from 10% to 25%. Again, nephrotoxicity is dependent on dosing and dose strategy, with a majority of patients experiencing resolution within weeks of cessation of therapy.
People with CF are a patient group distinct from the general patient population receiving IV colistimethate sodium or tobramycin, who tend to be the critically ill, and as such have unique drug usage indications and nephrotoxicity risk.2,3,8,9 First, CF patients prescribed IV colistimethate sodium or tobramycin tend to be not critically ill and also younger compared with other study populations. Due to the absence of multisystemic illness, CF patients tend to receive fewer concomitant nephrotoxins during a hospitalization for CF pulmonary exacerbations compared with those without CF who receive these drugs. However, while non-CF patients tend to require single courses of therapy with broad-spectrum IV antibiotics, the typical CF patient can receive many courses of IV colistimethate sodium or tobramycin over just a few years.1 Another consideration is the drastically different pharmacokinetic profiles of many drugs in those with and without CF, likely owing to both drug metabolism and clearance.11 For these reasons, data of AKI rates and risks in the general patient population are not readily extrapolatable to the CF patient.
Despite these differences, there is a paucity of data describing the frequency of colistimethate sodium- or tobramycin-associated AKI in patients with CF. The limited literature suggests a lower frequency of colistimethate sodium nephrotoxicity (0%–29%) than would be expected in non-CF patients.2,12–15 The incidence of nephrotoxicity related to tobramycin in the CF patient population is better described, ranging from 8% to 42%.3,16,17 However, these variable reports of AKI rates in CF and non-CF patients can generate uncertainty in the selection of colistimethate sodium or tobramycin as the second antipseudomonal IV antibiotic, particularly when the clinical decision is focused on minimizing nephrotoxicity. To our knowledge, no studies have directly compared the relative nephrotoxicity of IV colistimethate sodium with IV tobramycin in CF patients.
Methods
Data collection
This was a retrospective cohort study of adult (aged ≥18 years) inpatients with CF treated between 1 January 2008 and 30 June 2019 at a single CF centre. Patient encounters were eligible for inclusion if the patient received ≥48 h of inpatient treatment with either IV colistimethate sodium or IV tobramycin. Patient encounters were excluded if colistimethate sodium or tobramycin were administrated within 14 days prior to the index treatment event, if the patient required renal replacement therapy within 14 days prior to therapy, if there was a change in serum creatinine (SCr) of >0.5 mg/dL in the 48 h prior to therapy initiation, if there were no baseline or subsequent SCr results available to determine the onset of AKI, if both colistimethate sodium and tobramycin were administered during a single course of therapy, or if there was a history of solid organ transplantation. Concomitant exposure to other antimicrobials and/or nephrotoxins was classified as a dichotomous variable if any doses were administered during the index care encounter.
At our CF centre, IV tobramycin is initiated at 10 mg/kg every 24 h (extended interval dosing) with goals of 20–30 mg/mL peak concentration, <1 mg/mL concentration at 18 h and <0.25 mg/mL trough concentration by homogeneous enzyme immunoassay. If dosing met these goals, subsequent monitoring was transitioned to 18 h levels twice weekly. If dosing did not meet these goals, then repeat three-level monitoring was done after dose adjustment. IV colistimethate sodium was administered at 3–5 mg/kg/day of colistin base activity (CBA) in divided doses. No therapeutic drug monitoring was routinely completed at our centre during the period of this study. Daily SCr monitoring is standard at our centre for patients receiving either IV tobramycin or IV colistimethate sodium.
The primary outcome was AKI, defined using the risk, injury, failure, loss, end-stage (RIFLE) criteria.18 An AKI diagnosis was assigned if patients met RIFLE criteria as measured by SCr from the initiation of colistimethate sodium or tobramycin drug therapy through hospital discharge or 48 h after therapy completion, whichever occurred first. Baseline SCr was defined as the minimum value obtained within the 48 h prior to treatment initiation. Secondary outcomes of interest included stage of AKI, peak SCr and hospital length of stay.
Statistical analysis
Summary statistics were calculated for baseline characteristics and crude outcomes between care encounters with colistimethate sodium or tobramycin. Multiple encounters could be included for each patient, and patients could cross over between treatment groups, provided the crossover occurred as a separate course of therapy. Logistic mixed-effects models were fitted with AKI as the outcome variable described as a binary occurrence. Models included a random intercept term for the patient to account for repeated measures within an individual with fixed effects for drug treatment and clinically relevant covariates. Covariates incorporated into the models were predetermined based on risk factors identified via literature review of colistimethate sodium-associated nephrotoxicity in CF patients. Univariable and multivariate modelling was performed using all covariates listed in Table 1, with significant predictors or those of special clinical interest reported in Table 2. Analyses were performed using R version 3.6.1 and STATA version 12.0.
| Characteristic . | All encounters N = 507 . | CMS N = 72 . | TOB N = 435 . |
|---|---|---|---|
| Demographics | |||
| Age (years), median (IQR) | 26 (21–30) | 28 (24–33) | 25 (21–30) |
| Male sex, n (%) | 237 (46.7) | 25 (34.7) | 212 (48.7) |
| Caucasian race, n (%) | 500 (98.6) | 71 (98.6) | 429 (98.6) |
| Height (cm), median (IQR) | 165 (158–172) | 163 (158–170) | 167 (158–173) |
| Weight (kg), median (IQR) | 54.2 (48.2–61.7) | 54.8 (48.5–61.1) | 53.8 (48.1–61.8) |
| BMI (kg/m2), median (IQR) | 19.9 (18.2–22.5) | 20.3 (18.9–21.6) | 19.8 (18.0–22.6) |
| Homozygous F508del, n (%) | 238 (46.9) | 23 (31.9) | 215 (49.4) |
| Heterozygous F508del, n (%) | 210 (41.4) | 39 (54.2) | 171 (39.3) |
| CFTR modulator use, n (%) | 11 (2.2) | 2 (2.8) | 9 (2.1) |
| Clinical characteristics | |||
| ICU admission, n (%) | 27 (5.3) | 3 (4.2) | 24 (5.5) |
| Mechanical ventilation, n (%) | 17 (3.4) | 2 (2.8) | 16 (3.7) |
| Charlson comorbidity index, median (IQR) | 1 (0–1) | 1 (1–2) | 1 (0–1) |
| Sepsis, n (%) | 27 (5.3) | 1 (1.4) | 16 (3.7) |
| Diabetes, n (%) | 66 (13.0) | 19 (26.4) | 47 (10.8) |
| Pancreatic insufficiency, n (%) | 470 (92.7) | 72 (100) | 398 (91.5) |
| Hypertension, n (%) | 21 (4.1) | 4 (5.6) | 17 (3.9) |
| Baseline SCr (mg/dL), median (IQR) | 0.69 (0.59–0.82) | 0.70 (0.59–0.83) | 0.69 (0.59–0.81) |
| SCr measurements standardized to 7 days, median (IQR) | 7 (5–9) | 7 (6–9) | 7 (5–9) |
| Baseline albumina (g/dL), median (IQR) | 3.8 (3.4–4.2) | 3.7 (3.4–3.8) | 3.8 (3.4–4.2) |
| Baseline ppFEV1, median (IQR) | 28 (16–45) | 28.5 (16–43) | 27 (15–46) |
| Concomitant antipseudomonal antibiotics, n (%) | |||
| Aztreonam | 36 (7.1) | 8 (11.1) | 28 (6.4) |
| Cefepime | 128 (25.2) | 14 (19.1) | 114 (26.2) |
| Ceftazidime | 129 (25.4) | 20 (27.8) | 109 (25.1) |
| Meropenem | 136 (26.8) | 22(30.6) | 114 (26.2) |
| Piperacillin/tazobactam | 112 (22.3) | 5 (6.9) | 108 (24.8) |
| Ciprofloxacin | 49 (9.7) | 9 (12.5) | 40 (9.2) |
| Levofloxacin | 30 (5.9) | 9 (12.5) | 21 (4.8) |
| Inhaled TOB/CMS | 110 (21.7) | 2 (2.8) | 108 (24.8) |
| Inhaled TOB | 31 (16.2) | 2 (2.8) | 29 (6.7) |
| Inhaled CMS | 82 (16.2) | 0 (0.0) | 82 (18.9) |
| High-dose study drugb | 111 (21.9) | 3 (4.2) | 108 (24.8) |
| Concomitant nephrotoxic medications, n (%) | |||
| ACE inhibitors | 3 (0.6) | 0 (0) | 3 (0.7) |
| Contrast dye | 3 (0.6) | 0 (0) | 3 (0.7) |
| Loop diuretics | 10 (2) | 1 (1.4) | 9 (2.1) |
| NSAIDs | 173 (34.1) | 28 (38.9) | 145 (33.3) |
| Sulpha antibiotics | 77 (15.2) | 7 (9.7) | 70 (16.1) |
| Vancomycin | 167 (32.9) | 27 (37.5) | 140 (32.2) |
| Vasopressors | 5 (1) | 1 (1.4) | 4 (0.9) |
| Characteristic . | All encounters N = 507 . | CMS N = 72 . | TOB N = 435 . |
|---|---|---|---|
| Demographics | |||
| Age (years), median (IQR) | 26 (21–30) | 28 (24–33) | 25 (21–30) |
| Male sex, n (%) | 237 (46.7) | 25 (34.7) | 212 (48.7) |
| Caucasian race, n (%) | 500 (98.6) | 71 (98.6) | 429 (98.6) |
| Height (cm), median (IQR) | 165 (158–172) | 163 (158–170) | 167 (158–173) |
| Weight (kg), median (IQR) | 54.2 (48.2–61.7) | 54.8 (48.5–61.1) | 53.8 (48.1–61.8) |
| BMI (kg/m2), median (IQR) | 19.9 (18.2–22.5) | 20.3 (18.9–21.6) | 19.8 (18.0–22.6) |
| Homozygous F508del, n (%) | 238 (46.9) | 23 (31.9) | 215 (49.4) |
| Heterozygous F508del, n (%) | 210 (41.4) | 39 (54.2) | 171 (39.3) |
| CFTR modulator use, n (%) | 11 (2.2) | 2 (2.8) | 9 (2.1) |
| Clinical characteristics | |||
| ICU admission, n (%) | 27 (5.3) | 3 (4.2) | 24 (5.5) |
| Mechanical ventilation, n (%) | 17 (3.4) | 2 (2.8) | 16 (3.7) |
| Charlson comorbidity index, median (IQR) | 1 (0–1) | 1 (1–2) | 1 (0–1) |
| Sepsis, n (%) | 27 (5.3) | 1 (1.4) | 16 (3.7) |
| Diabetes, n (%) | 66 (13.0) | 19 (26.4) | 47 (10.8) |
| Pancreatic insufficiency, n (%) | 470 (92.7) | 72 (100) | 398 (91.5) |
| Hypertension, n (%) | 21 (4.1) | 4 (5.6) | 17 (3.9) |
| Baseline SCr (mg/dL), median (IQR) | 0.69 (0.59–0.82) | 0.70 (0.59–0.83) | 0.69 (0.59–0.81) |
| SCr measurements standardized to 7 days, median (IQR) | 7 (5–9) | 7 (6–9) | 7 (5–9) |
| Baseline albumina (g/dL), median (IQR) | 3.8 (3.4–4.2) | 3.7 (3.4–3.8) | 3.8 (3.4–4.2) |
| Baseline ppFEV1, median (IQR) | 28 (16–45) | 28.5 (16–43) | 27 (15–46) |
| Concomitant antipseudomonal antibiotics, n (%) | |||
| Aztreonam | 36 (7.1) | 8 (11.1) | 28 (6.4) |
| Cefepime | 128 (25.2) | 14 (19.1) | 114 (26.2) |
| Ceftazidime | 129 (25.4) | 20 (27.8) | 109 (25.1) |
| Meropenem | 136 (26.8) | 22(30.6) | 114 (26.2) |
| Piperacillin/tazobactam | 112 (22.3) | 5 (6.9) | 108 (24.8) |
| Ciprofloxacin | 49 (9.7) | 9 (12.5) | 40 (9.2) |
| Levofloxacin | 30 (5.9) | 9 (12.5) | 21 (4.8) |
| Inhaled TOB/CMS | 110 (21.7) | 2 (2.8) | 108 (24.8) |
| Inhaled TOB | 31 (16.2) | 2 (2.8) | 29 (6.7) |
| Inhaled CMS | 82 (16.2) | 0 (0.0) | 82 (18.9) |
| High-dose study drugb | 111 (21.9) | 3 (4.2) | 108 (24.8) |
| Concomitant nephrotoxic medications, n (%) | |||
| ACE inhibitors | 3 (0.6) | 0 (0) | 3 (0.7) |
| Contrast dye | 3 (0.6) | 0 (0) | 3 (0.7) |
| Loop diuretics | 10 (2) | 1 (1.4) | 9 (2.1) |
| NSAIDs | 173 (34.1) | 28 (38.9) | 145 (33.3) |
| Sulpha antibiotics | 77 (15.2) | 7 (9.7) | 70 (16.1) |
| Vancomycin | 167 (32.9) | 27 (37.5) | 140 (32.2) |
| Vasopressors | 5 (1) | 1 (1.4) | 4 (0.9) |
ACE, angiotensin-converting enzyme; NSAIDs, non-steroidal anti-inflammatory drugs; ppFEV1, percent predicted forced expiratory volume in 1 s; TOB, tobramycin; CMS, colistimethate sodium.
Encounters with missing values (n = 110) were categorized as normal (>3 g/dL) for inclusion in the model.
Dose >10 mg/kg tobramycin or >5 mg/kg CBA.
| Characteristic . | All encounters N = 507 . | CMS N = 72 . | TOB N = 435 . |
|---|---|---|---|
| Demographics | |||
| Age (years), median (IQR) | 26 (21–30) | 28 (24–33) | 25 (21–30) |
| Male sex, n (%) | 237 (46.7) | 25 (34.7) | 212 (48.7) |
| Caucasian race, n (%) | 500 (98.6) | 71 (98.6) | 429 (98.6) |
| Height (cm), median (IQR) | 165 (158–172) | 163 (158–170) | 167 (158–173) |
| Weight (kg), median (IQR) | 54.2 (48.2–61.7) | 54.8 (48.5–61.1) | 53.8 (48.1–61.8) |
| BMI (kg/m2), median (IQR) | 19.9 (18.2–22.5) | 20.3 (18.9–21.6) | 19.8 (18.0–22.6) |
| Homozygous F508del, n (%) | 238 (46.9) | 23 (31.9) | 215 (49.4) |
| Heterozygous F508del, n (%) | 210 (41.4) | 39 (54.2) | 171 (39.3) |
| CFTR modulator use, n (%) | 11 (2.2) | 2 (2.8) | 9 (2.1) |
| Clinical characteristics | |||
| ICU admission, n (%) | 27 (5.3) | 3 (4.2) | 24 (5.5) |
| Mechanical ventilation, n (%) | 17 (3.4) | 2 (2.8) | 16 (3.7) |
| Charlson comorbidity index, median (IQR) | 1 (0–1) | 1 (1–2) | 1 (0–1) |
| Sepsis, n (%) | 27 (5.3) | 1 (1.4) | 16 (3.7) |
| Diabetes, n (%) | 66 (13.0) | 19 (26.4) | 47 (10.8) |
| Pancreatic insufficiency, n (%) | 470 (92.7) | 72 (100) | 398 (91.5) |
| Hypertension, n (%) | 21 (4.1) | 4 (5.6) | 17 (3.9) |
| Baseline SCr (mg/dL), median (IQR) | 0.69 (0.59–0.82) | 0.70 (0.59–0.83) | 0.69 (0.59–0.81) |
| SCr measurements standardized to 7 days, median (IQR) | 7 (5–9) | 7 (6–9) | 7 (5–9) |
| Baseline albumina (g/dL), median (IQR) | 3.8 (3.4–4.2) | 3.7 (3.4–3.8) | 3.8 (3.4–4.2) |
| Baseline ppFEV1, median (IQR) | 28 (16–45) | 28.5 (16–43) | 27 (15–46) |
| Concomitant antipseudomonal antibiotics, n (%) | |||
| Aztreonam | 36 (7.1) | 8 (11.1) | 28 (6.4) |
| Cefepime | 128 (25.2) | 14 (19.1) | 114 (26.2) |
| Ceftazidime | 129 (25.4) | 20 (27.8) | 109 (25.1) |
| Meropenem | 136 (26.8) | 22(30.6) | 114 (26.2) |
| Piperacillin/tazobactam | 112 (22.3) | 5 (6.9) | 108 (24.8) |
| Ciprofloxacin | 49 (9.7) | 9 (12.5) | 40 (9.2) |
| Levofloxacin | 30 (5.9) | 9 (12.5) | 21 (4.8) |
| Inhaled TOB/CMS | 110 (21.7) | 2 (2.8) | 108 (24.8) |
| Inhaled TOB | 31 (16.2) | 2 (2.8) | 29 (6.7) |
| Inhaled CMS | 82 (16.2) | 0 (0.0) | 82 (18.9) |
| High-dose study drugb | 111 (21.9) | 3 (4.2) | 108 (24.8) |
| Concomitant nephrotoxic medications, n (%) | |||
| ACE inhibitors | 3 (0.6) | 0 (0) | 3 (0.7) |
| Contrast dye | 3 (0.6) | 0 (0) | 3 (0.7) |
| Loop diuretics | 10 (2) | 1 (1.4) | 9 (2.1) |
| NSAIDs | 173 (34.1) | 28 (38.9) | 145 (33.3) |
| Sulpha antibiotics | 77 (15.2) | 7 (9.7) | 70 (16.1) |
| Vancomycin | 167 (32.9) | 27 (37.5) | 140 (32.2) |
| Vasopressors | 5 (1) | 1 (1.4) | 4 (0.9) |
| Characteristic . | All encounters N = 507 . | CMS N = 72 . | TOB N = 435 . |
|---|---|---|---|
| Demographics | |||
| Age (years), median (IQR) | 26 (21–30) | 28 (24–33) | 25 (21–30) |
| Male sex, n (%) | 237 (46.7) | 25 (34.7) | 212 (48.7) |
| Caucasian race, n (%) | 500 (98.6) | 71 (98.6) | 429 (98.6) |
| Height (cm), median (IQR) | 165 (158–172) | 163 (158–170) | 167 (158–173) |
| Weight (kg), median (IQR) | 54.2 (48.2–61.7) | 54.8 (48.5–61.1) | 53.8 (48.1–61.8) |
| BMI (kg/m2), median (IQR) | 19.9 (18.2–22.5) | 20.3 (18.9–21.6) | 19.8 (18.0–22.6) |
| Homozygous F508del, n (%) | 238 (46.9) | 23 (31.9) | 215 (49.4) |
| Heterozygous F508del, n (%) | 210 (41.4) | 39 (54.2) | 171 (39.3) |
| CFTR modulator use, n (%) | 11 (2.2) | 2 (2.8) | 9 (2.1) |
| Clinical characteristics | |||
| ICU admission, n (%) | 27 (5.3) | 3 (4.2) | 24 (5.5) |
| Mechanical ventilation, n (%) | 17 (3.4) | 2 (2.8) | 16 (3.7) |
| Charlson comorbidity index, median (IQR) | 1 (0–1) | 1 (1–2) | 1 (0–1) |
| Sepsis, n (%) | 27 (5.3) | 1 (1.4) | 16 (3.7) |
| Diabetes, n (%) | 66 (13.0) | 19 (26.4) | 47 (10.8) |
| Pancreatic insufficiency, n (%) | 470 (92.7) | 72 (100) | 398 (91.5) |
| Hypertension, n (%) | 21 (4.1) | 4 (5.6) | 17 (3.9) |
| Baseline SCr (mg/dL), median (IQR) | 0.69 (0.59–0.82) | 0.70 (0.59–0.83) | 0.69 (0.59–0.81) |
| SCr measurements standardized to 7 days, median (IQR) | 7 (5–9) | 7 (6–9) | 7 (5–9) |
| Baseline albumina (g/dL), median (IQR) | 3.8 (3.4–4.2) | 3.7 (3.4–3.8) | 3.8 (3.4–4.2) |
| Baseline ppFEV1, median (IQR) | 28 (16–45) | 28.5 (16–43) | 27 (15–46) |
| Concomitant antipseudomonal antibiotics, n (%) | |||
| Aztreonam | 36 (7.1) | 8 (11.1) | 28 (6.4) |
| Cefepime | 128 (25.2) | 14 (19.1) | 114 (26.2) |
| Ceftazidime | 129 (25.4) | 20 (27.8) | 109 (25.1) |
| Meropenem | 136 (26.8) | 22(30.6) | 114 (26.2) |
| Piperacillin/tazobactam | 112 (22.3) | 5 (6.9) | 108 (24.8) |
| Ciprofloxacin | 49 (9.7) | 9 (12.5) | 40 (9.2) |
| Levofloxacin | 30 (5.9) | 9 (12.5) | 21 (4.8) |
| Inhaled TOB/CMS | 110 (21.7) | 2 (2.8) | 108 (24.8) |
| Inhaled TOB | 31 (16.2) | 2 (2.8) | 29 (6.7) |
| Inhaled CMS | 82 (16.2) | 0 (0.0) | 82 (18.9) |
| High-dose study drugb | 111 (21.9) | 3 (4.2) | 108 (24.8) |
| Concomitant nephrotoxic medications, n (%) | |||
| ACE inhibitors | 3 (0.6) | 0 (0) | 3 (0.7) |
| Contrast dye | 3 (0.6) | 0 (0) | 3 (0.7) |
| Loop diuretics | 10 (2) | 1 (1.4) | 9 (2.1) |
| NSAIDs | 173 (34.1) | 28 (38.9) | 145 (33.3) |
| Sulpha antibiotics | 77 (15.2) | 7 (9.7) | 70 (16.1) |
| Vancomycin | 167 (32.9) | 27 (37.5) | 140 (32.2) |
| Vasopressors | 5 (1) | 1 (1.4) | 4 (0.9) |
ACE, angiotensin-converting enzyme; NSAIDs, non-steroidal anti-inflammatory drugs; ppFEV1, percent predicted forced expiratory volume in 1 s; TOB, tobramycin; CMS, colistimethate sodium.
Encounters with missing values (n = 110) were categorized as normal (>3 g/dL) for inclusion in the model.
Dose >10 mg/kg tobramycin or >5 mg/kg CBA.
Multivariate analysis (logistic mixed-effects model) for development of AKI
| Model covariates . | Univariate . | Multivariate . | ||
|---|---|---|---|---|
| OR . | 95% CI . | aOR . | 95% CI . | |
| Tobramycin (versus CMS) | 1.31 | 0.46–3.69 | 1.00 | 0.33–3.02 |
| Age (per year) | 0.97 | 0.93–1.02 | 0.98 | 0.93–1.02 |
| Diabetes | 1.13 | 0.41–3.06 | 1.14 | 0.40–3.22 |
| Any nephrotoxina | 2.33 | 1.22–4.43 | 2.51 | 1.27–4.95 |
| High-dose study drugb | 1.02 | 0.47–2.22 | 1.01 | 0.45–2.31 |
| Vancomycin + piperacillin/tazobactam | 5.45 | 1.92–15.4 | 5.95 | 2.05–17.3 |
| Model covariates . | Univariate . | Multivariate . | ||
|---|---|---|---|---|
| OR . | 95% CI . | aOR . | 95% CI . | |
| Tobramycin (versus CMS) | 1.31 | 0.46–3.69 | 1.00 | 0.33–3.02 |
| Age (per year) | 0.97 | 0.93–1.02 | 0.98 | 0.93–1.02 |
| Diabetes | 1.13 | 0.41–3.06 | 1.14 | 0.40–3.22 |
| Any nephrotoxina | 2.33 | 1.22–4.43 | 2.51 | 1.27–4.95 |
| High-dose study drugb | 1.02 | 0.47–2.22 | 1.01 | 0.45–2.31 |
| Vancomycin + piperacillin/tazobactam | 5.45 | 1.92–15.4 | 5.95 | 2.05–17.3 |
CMS, colistimethate sodium. Model contains a different random-effects intercept for each subject.
≥1 dose of nephrotoxin during index encounter, not including vancomycin.
Dose >10 mg/kg tobramycin or >5 mg/kg CBA.
Multivariate analysis (logistic mixed-effects model) for development of AKI
| Model covariates . | Univariate . | Multivariate . | ||
|---|---|---|---|---|
| OR . | 95% CI . | aOR . | 95% CI . | |
| Tobramycin (versus CMS) | 1.31 | 0.46–3.69 | 1.00 | 0.33–3.02 |
| Age (per year) | 0.97 | 0.93–1.02 | 0.98 | 0.93–1.02 |
| Diabetes | 1.13 | 0.41–3.06 | 1.14 | 0.40–3.22 |
| Any nephrotoxina | 2.33 | 1.22–4.43 | 2.51 | 1.27–4.95 |
| High-dose study drugb | 1.02 | 0.47–2.22 | 1.01 | 0.45–2.31 |
| Vancomycin + piperacillin/tazobactam | 5.45 | 1.92–15.4 | 5.95 | 2.05–17.3 |
| Model covariates . | Univariate . | Multivariate . | ||
|---|---|---|---|---|
| OR . | 95% CI . | aOR . | 95% CI . | |
| Tobramycin (versus CMS) | 1.31 | 0.46–3.69 | 1.00 | 0.33–3.02 |
| Age (per year) | 0.97 | 0.93–1.02 | 0.98 | 0.93–1.02 |
| Diabetes | 1.13 | 0.41–3.06 | 1.14 | 0.40–3.22 |
| Any nephrotoxina | 2.33 | 1.22–4.43 | 2.51 | 1.27–4.95 |
| High-dose study drugb | 1.02 | 0.47–2.22 | 1.01 | 0.45–2.31 |
| Vancomycin + piperacillin/tazobactam | 5.45 | 1.92–15.4 | 5.95 | 2.05–17.3 |
CMS, colistimethate sodium. Model contains a different random-effects intercept for each subject.
≥1 dose of nephrotoxin during index encounter, not including vancomycin.
Dose >10 mg/kg tobramycin or >5 mg/kg CBA.
Results
A total of 156 patients representing 507 care encounters were included for analysis. Baseline characteristics and concomitant medication administration are presented in Table 1. Use of concomitant nephrotoxic medications was common. Concomitant antibiotics piperacillin/tazobactam (24.8% versus 6.9%) and inhaled colistimethate sodium/tobramycin (24.8% versus 2.8%) were used more frequently among patients treated with IV tobramycin.
Median (IQR) dose of IV tobramycin was 8.9 (7.7–10.0) mg/kg/day. The median (IQR) dose of IV colistimethate sodium was 3.6 (2.9–4.3) mg/kg/day over three divided doses of CBA. The crude frequency of AKI was 9.5% across all encounters, occurring in 5 encounters (6.9%) with colistimethate sodium and 43 encounters (9.9%) with tobramycin. RIFLE category R (risk) was the most common stage of AKI, occurring in 3 (4.2%) encounters with colistimethate sodium and 40 (9.2%) with tobramycin. Median (IQR) peak SCr and time to peak SCr were not different between the two groups [peak: 0.70 (0.60–0.90) mg/dL for colistimethate sodium versus 0.76 (0.60–0.90) mg/dL for tobramycin; time to peak: 4.6 (4.3–5.9) versus 4.2 (1.6–7.8) days]. Similarly, median (IQR) hospital length of stay was 8 (5–14) days for both colistimethate sodium and tobramycin encounters.
Univariate analysis of predictors of clinical interest or of significance are included in Table 2. AKI development was not significantly different between colistimethate sodium and tobramycin [adjusted OR (aOR) 1.00; 95% CI 0.16–6.03] (Table 2). Variables independently associated with increased odds of AKI in either IV colistimethate sodium or tobramycin groups included concomitant nephrotoxin use (aOR 2.51; 95% CI 1.27–4.95) and concomitant use of the combination of vancomycin and piperacillin/tazobactam (aOR 5.95; 95% CI 2.05–17.3).
Discussion
In this study, AKI occurred in only 6.9% of IV colistimethate sodium encounters and 9.9% of IV tobramycin encounters in our CF study population, which is similar to previously reported rates.12–17 Multivariate analysis showed that AKI did not occur more frequently in patients treated with IV colistimethate sodium compared with those treated with IV tobramycin.
As expected, the rate of IV colistimethate sodium nephrotoxicity in our study was significantly lower than previously reported in the non-CF population, likely due to the differences in study population characteristics and the pharmacokinetics differences observed in CF patients compared with the non-CF population.11 Colistimethate sodium is administered as a pro-drug that is converted to the active drug colistin, the nephrotoxic agent. However, increased renal elimination of colistimethate sodium in the CF population may lead to decreased exposure and nephrotoxicity due to colistin.19–21 Data from Susanto and colleagues22 suggest that enhanced tubular secretion in CF is likely a factor in enhancing renal elimination of drugs; furthermore, a compensatory mechanism for the deficiency in the cystic fibrosis transmembrane conductance regulator (CFTR) protein is an up-regulation of the gut and renal P-glycoprotein efflux pump, of which colistimethate sodium is a substrate. This increases efflux of colistimethate sodium and depletes the amount of active drug available to injure the kidneys.23
Secondly, the median (IQR) dose of IV colistimethate sodium used at our institution in this patient population of 3.6 (2.9–4.3) mg/kg/day CBA is lower than historical publications using doses up to 5–8 mg/kg/day CBA in CF patients.14,15 However, it is in keeping with more recent evaluations of broad dosing practices of ∼3.2 mg/kg/day CBA in the CF population,24 despite the dose being lower than the standard dose of 5 mg/kg/day CBA used for a Gram-negative infection in the non-CF patient population. Thus, the lower rate of AKI in CF patients compared with the very high rates in non-CF patients may be a reflection of the routine lower dosing.
Compared with the few studies assessing IV colistimethate sodium nephrotoxicity in the CF patient, our study population is quite similar. In an early study by Bosso and colleagues,14 only 5.3% of subjects experienced nephrotoxicity, despite doses ranging from 3 to 8 mg/kg/day CBA. Similarly, nephrotoxicity was reported in 3.7% of subjects in a study with lower doses of CBA.15 The largest report of IV colistimethate sodium in CF patients described AKI occurring in 29.8% of subjects.13 However, there were significant higher rates of concomitant nephrotoxin usage and a slightly higher dose CBA (mean ± SD: 4.1 ± 1.1) compared with our study population.
When compared with reports of IV tobramycin in the CF population, our study also describes rates of nephrotoxicity similar to prior reports. AKI occurred in 8.7% of a cohort of CF patients with similar baseline characteristics.17 Other reports of higher rates of AKI with the use of IV tobramycin in CF patients selected for more advanced disease with repeated pulmonary exacerbations and are difficult to compare directly.3,16 Not surprisingly, our rate of tobramycin AKI is lower than in non-CF populations, likely as a result of decreased renal tubular accumulation due to enhanced clearance in the CF kidney.25
In this study, utilization of univariate and multivariate mixed-effects modelling allowed for the inclusion of recurrent events, thereby incorporating maximal data inclusion into the study while maintaining the ability to analyse relevant covariates independently.26 By including repeated encounters for IV colistimethate sodium or IV tobramycin usage, we provide a real-life estimate for the rate of AKI in CF patients, which includes first-time usage and longer-term repeated use.
In our multivariate analysis, concomitant use of any additional nephrotoxic medication was associated with an increased risk of AKI, as was concomitant use of the combination of vancomycin and piperacillin/tazobactam. While vancomycin is well established as a nephrotoxin, the data associating piperacillin/tazobactam alone with AKI are mixed, and some animal studies suggest a protective effect on the kidneys.27,28 However, a growing body of clinical evidence suggests an increased risk of AKI with the combination of vancomycin and piperacillin/tazobactam,29 as was the case in the CF patient population in the present study.
This study has limitations, including a small sample size typical for CF literature, no availability of detailed outpatient follow-up data, and missing data, as is the nature of retrospective studies. A pre-defined covariate for multivariate analysis was low albumin, yet a significant proportion of missing values precluded incorporation into the final model. Another variable not available, given the retrospective nature of this study, was urine output, thus AKI in our study was defined by SCr trends alone. Another consideration is the definition of concomitant nephrotoxins as at least one dose during the index encounter, which was necessary to facilitate consistent extraction of data from different electronic health record systems, but may not represent clinically significant exposure to the concomitant nephrotoxin in all cases.
Another limitation is the lack of multiple groups using other antibiotic agents for comparison. Our approach was to prioritize data modelling to help provide evidence for the selection of a second antipseudomonal agent, between IV colistimethate sodium versus IV tobramycin, in CF patients with pulmonary exacerbations when nephrotoxicity risk is part of the consideration. As such, we do not compare the presented groups with other treatment approaches such as inhaled antibiotics or other aminoglycosides and polymyxin B. Crass and colleagues13 described similar rates of nephrotoxicity with the use of polymyxin B versus IV colistimethate sodium in CF patients, despite more commonly reported higher rates of nephrotoxicity with IV colistimethate sodium in the non-CF population.30 While some reports suggest at least some risk for inhaled colistin or an aminoglycoside, a head-to-head comparison is beyond the scope of this study.31
Despite these limitations, this is among the largest studies assessing AKI in CF patients treated with IV colistimethate sodium. It also offers a unique design by including multiple encounters per patient, maximizing the amount of data available for interpretation. Our data suggest that treatment with either systemic colistimethate sodium or tobramycin in the CF patient population is associated with a similar rate of nephrotoxicity, and can provide reassurance that selection of a second antipseudomonal antibiotic between these two agents should not depend on this specific safety concern. However, the concomitant use of other nephrotoxic agents, particularly the combination of vancomycin and piperacillin/tazobactam, significantly increases the risk for AKI. Thus, mindful consideration of the entire medication profiles during treatment for pulmonary exacerbations warrants attention.
Funding
This study was carried out as part of our routine work.
Transparency declarations
S. Jia reports grants and clinical trials support from the Cystic Fibrosis Foundation, and clinical trials support from Insmed Inc, outside the submitted work. M. K. Han reports grants or contracts from NHLBI, Sanofi, Novartis, Nuvaira, Sunovion, Gala Therapeutics, COPD Foundation and AstraZeneca; royalties or licenses from UpToDate, Norton Publishing and Penguin Random House; consulting fees from AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, Novartis, Pulmonx, Teva, Verona, Merck, Sanofi, DevPro, Aerogen, Polarian, United Therapeutics and Regeneron; payment or honoraria from Cipla, Chiesi, AstraZeneca, Boehringer Ingelheim and GlaxoSmithKline; participation on data safety monitoring or advisory boards with Novartis and Medtronic; stock or stock options with Meissa Vaccines; and receipt of materials or services from GSK, Boehringer Ingelheim, AstraZeneca and Novartis. Other authors have nothing to disclose.
