Risk Factors for Teicoplanin-Associated Acute Kidney Injury in Patients with Hematological Malignancies: Focusing on Concomitant Use of Tazobactam/Piperacillin

Abstract

Patients with hematological malignancies (HM) often receive tazobactam/piperacillin (TAZ/PIPC) and glycopeptide antibiotics for febrile neutropenia. The effect of concomitant use of TAZ/PIPC on risk of teicoplanin (TEIC)-associated acute kidney injury (AKI) remains unclear. We investigated the impact of concomitant TAZ/PIPC use on TEIC-associated AKI in HM patients and identified the risk factors. In this retrospective, single-center, observational cohort study, 203 patients received TEIC, 176 of whom satisfied the selection criteria and were divided into TEIC cohort (no TAZ/PIPC; n = 118) and TEIC + TAZ/PIPC cohort (n = 58). AKI was defined as serum creatinine increase ≥0.3 mg/dL within 48 h or ≥50% from baseline. Incidence of AKI in TEIC cohort before and after propensity score matching was 9.3 and 5.9%, respectively, and that in TEIC + TAZ/PIPC cohort was 10.3 and 11.8%. AKI incidence and risk were not significantly different between two cohorts before (p = 0.829; odds ratio (OR) 1.122, 95% confidence interval (CI) 0.393–3.202) and after matching (p = 0.244; OR 2.133, 95% CI 0.503–9.043). Logistic regression analysis with factors clinically or mechanistically potentially related to TEIC-associated AKI, including concomitant TAZ/PIPC use, as independent variables identified baseline hemoglobin level as the only significant risk factor for TEIC-associated AKI (p = 0.011; OR 0.484, 95% CI 0.276–0.848). In HM patients treated with TEIC, concomitant TAZ/PIPC use did not increase AKI risk whereas lower hemoglobin levels had higher risk for TEIC-associated AKI development, suggesting the necessity to monitor serum creatinine when using TEIC in patients with anemia.

INTRODUCTION

Tazobactam/piperacillin (TAZ/PIPC), an antibiotic containing a beta-lactamase inhibitor, has a broad antibacterial spectrum against many Gram-positive and Gram-negative aerobic and anaerobic bacteria, including many pathogens producing beta-lactamases.¹⁾ It is widely used for treating various infections and frequently used as a first-line drug for febrile neutropenia (FN) treatment. In addition, concomitant use of a glycopeptide antibiotic; vancomycin (VCM) or teicoplanin (TEIC), is recommended for FN treatment when methicillin-resistant Staphylococcus aureus (MRSA) infection is suspected, or response to broad-spectrum antibiotics such as TAZ/PIPC is inadequate. Therefore, patients with hematological malignancies have many opportunities to receive VCM or TEIC in addition to TAZ/PIPC.

VCM has broad-spectrum activities for Gram-positive bacteria such as MRSA, but accumulates in renal tubular epithelial cells and exerts direct toxicity to damage renal tubules, frequently causing acute kidney injury (AKI).²⁾ Histologically, AKI is characterized by acute interstitial nephritis and acute tubular necrosis, and the mechanism of cast nephropathy has also been demonstrated.³⁾ Risk factors for VCM-associated AKI include elevated trough levels, concomitant use of nephrotoxic agents such as aminoglycoside, amphotericin B, furosemide, non-steroidal anti-inflammatory drugs (NSAIDs), contrast media, and vasopressor; patients with impaired renal function, dehydration, and severe illness; use of high-dose VCM; admission to ICU; and treatment period of more than one week.⁴⁾

VCM alone has a risk of renal damage, but many reports in recent years have shown that concomitant use of TAZ/PIPC increases that risk. Burgess and Drew⁵⁾ reported that concomitant use of VCM and TAZ/PIPC increased the incidence of nephrotoxicity (odds ratio 2.48) compared to VCM alone. Oda et al.⁶⁾ revealed that AKI incidence in the VCM plus TAZ/PIPC group was 44.6%, compared to 17.6% in the group with either VCM or TAZ/PIPC replaced. Systematic reviews have also shown that concomitant use of VCM and TAZ/PIPC increases the frequency of renal damage.^7,8) Furthermore, multiple reports suggest that compared with other broad-spectrum antibiotics such as cefepime and meropenem, VCM increases the risk of AKI when concomitantly administered with TAZ/PIPC.^9–11)

TEIC has an antibacterial spectrum similar to VCM and is used to treat infections caused by resistant Gram-positive bacteria. Because of the lower incidence of AKI compared to VCM, TEIC can be used as an alternative when renal damage is a concern.¹²⁾ Therefore, TEIC is sometimes selected when adding a glycopeptide antibiotic to TAZ/PIPC in patients with FN having high risk of renal damage by using VCM concomitantly. A study comparing the efficacy and safety between VCM and TEIC in FN patients showed similar clinical efficacy for both drugs, whereas serum creatinine increase tended to be higher in VCM-treated than in TEIC-treated group.¹³⁾ Although a study by Shao et al.¹⁴⁾ showed no significant difference in AKI risk between TEIC plus TAZ/PIPC combination and VCM plus TAZ/PIPC combination, several recent reports have shown a significantly lower incidence of AKI in TEIC plus TAZ/PIPC than in VCM plus TAZ/PIPC.^15,16) Tai et al.¹⁷⁾ demonstrated no difference in AKI risk between concomitant use of TEIC with TAZ/PIPC and that of TEIC with other anti-pseudomonal β-lactams. In contrast, a single-center retrospective cohort study of 4202 patients suggested an association of TEIC plus TAZ/PIPC with a higher prevalence of AKI compared with monotherapy, although the overall decline in renal function was small.¹⁸⁾ However, there is no report evaluating whether TAZ/PIPC is a risk factor of TEIC-associated AKI.

Given the above background, this study compared AKI incidence between TEIC (no TAZ/PIPC) therapy and TEIC plus TAZ/PIPC combination therapy in patients with hematological malignancies and identified the risk factor of TEIC-associated AKI using patient background and clinical laboratory parameters including concomitant use of TAZ/PIPC.

MATERIALS AND METHODS

Subjects

The study design was a retrospective, single-center, observational cohort study. The electronic medical records at the Department of Hematology in Oita University Hospital from April 2014 to November 2020 were reviewed to identify patients who received intravenous TEIC. The exclusion criteria were as follows: age under 18 years; admission to intensive care unit or emergency center during TEIC treatment; no measurement of trough TEIC concentration; addition or discontinuation of TAZ/PIPC during TEIC treatment; switch from VCM; and death during TEIC treatment. The study was conducted in accordance with the ethical standards of our institute and the principles of Helsinki Declaration of 1975, as revised in 2013. The study was started following approval by the Ethics Committee of Oita University Faculty of Medicine (Review Reference Number: 2254).

Data Collection

The following demographic and laboratory parameters at the initiation of TEIC treatment were collected: sex, age, body weight, height, highest daily body temperature, albumin (ALB), white blood cell count (WBC), hemoglobin (HGB), platelet count (PLT), C-reactive protein (CRP), blood urea nitrogen (BUN), serum creatinine (Scr), and primary disease. Creatinine clearance (Ccr) was calculated with Cockcroft–Gault formula.¹⁹⁾ Primary disease was extracted from doctor’s medical records. Primary disease and infection type were extracted from doctor’s medical records. Infection type includes the suspected case. Concomitant drugs other than TAZ/PIPC included aminoglycosides, amphotericin B, loop diuretics, calcineurin inhibitors, and cisplatin, which are listed in the Japanese package insert of TEIC (Targocid^®) as having an increased risk for renal impairment; as well as sulfamethoxazole/trimethoprim, acyclovir, NSAIDs, angiotensin-converting enzyme inhibitors, and angiotensin receptor blockers, which possibly cause renal impairment. Concomitant use was defined as when one or more of these drugs were administered for at least three consecutive days during TEIC treatment.²⁰⁾ Topical formulations were not included in concomitant drugs. Serum TEIC concentrations were measured by a latex immunoturbidimetric assay using a TBA-25FR^® analyzer (Canon Medical Systems Co., Otawara, Japan). The highest trough concentration after more than three days of TEIC administration was used for data analysis.

Definition of AKI

AKI was defined as an increase in Scr level by ≥0.3 mg/dL within 48 h or by ≥50% from baseline, following the Kidney Disease Improving Global Outcomes (KDIGO) criteria.²¹⁾ Urine volume was not considered in the definition of AKI, because the data was not available. Scr levels were monitored from the beginning of TEIC treatment to the subsequent blood sampling after the end of treatment.

Statistical Analysis

Predictive Analytics Software (PASW) Statistics version 27 (SPSS Inc., Chicago, IL, U.S.A.) was used for statistical analysis. Data are expressed as number (%) for categorical variables and median [interquartile range] for continuous variables. Patients who met the selection criteria were classified into TEIC and TEIC + TAZ/PIPC cohorts as well as AKI and non-AKI cohorts. Data normality was assessed by Shapiro–Wilk test. Categorical variables were analyzed by chi-square test, or by Fisher’s exact test when more than 20% of the cells have expected frequencies <5. A p-value less than 0.05 was considered statistically significant.

One-to-one propensity score matching was implemented to adjust for differences in patient characteristics between TEIC (no TAZ/PIPC) and TEIC + TAZ/PIPC cohorts. A logistic regression model was adopted to estimate propensity scores, and c-index was calculated to assess the goodness of fit. Using the nearest neighbor matching method, one patient in one cohort was matched to one patient in the other cohort based on estimated propensity scores with the caliper width set at 0.2 standard deviations without replacement. The following demographic and laboratory variables potentially associated with AKI were incorporated into the one-to-one propensity score matching: sex, age, ALB, HGB, CRP, BUN, Ccr, infection type (febrile neutropenia and sepsis), and concomitant drugs. After propensity score matching, adjusted p-values and odds ratios were estimated using the same statistical methods as for the data before propensity score matching. The risk factors for the development of TEIC-associated AKI were analyzed by univariate and multivariate logistic regression model. For multivariate logistic regression analysis, independent variables comprising clinically or mechanistically potentially relevant factors with AKI development such as sex, age, highest trough TEIC concentration ALB, HGB, CRP, BUN, infection type (febrile neutropenia and sepsis), and concomitant drugs were inserted into the model by forced entry using Schwarz’s Bayesian information criterion, and variables remained in the model if p < 0.05. Aminoglycosides were excluded from the variable as they were used in only one patient.

RESULTS

Patient Characteristics

Of 203 patients who received TEIC, 176 satisfied the inclusion criteria and were divided into those who received TEIC but not TAZ/PIPC (TEIC cohort, n = 118) and those who received concomitant TEIC and TAZ/PIPC (TEIC + TAZ/PIPC cohort, n = 58). The background and clinical laboratory parameters of all patients analyzed are summarized in Table 1. The male:female ratio was 54 : 46; median age was 57.0 [45.0–66.0] years; and median body weight was 55.7 [48.5–63.3] kg. The median highest trough TEIC concentration was 20.7 [16.7–24.7] µg/mL, and median Ccr before TEIC initiation was 84.3 [59.7–120.2] mL/min. The most frequent primary disease was acute myeloid leukemia (34.1%), followed by lymphoma (20.5%). The most frequent infection type was febrile neutropenia (64.2%), followed by fever during neutropenia without fulfilling diagnostic for febrile neutropenia (17.0%). NSAIDs (53.4%) were the most frequently used concomitant drugs with AKI risk.

Table 1. Patient Characteristics and Clinical Laboratory Parameters of All Patients Analyzed

Parameter	Value
Patients; n	176
Sex (male/female); n (%)	95 (54.0)/81 (46.0)
Age (years)	57.0 [45.0–66.0]
Body weight (kg)	55.7 [48.5–63.3]
Height (cm)	161.2 [155.8–168.5]
Highest body temperature (°C)	38.2 [37.8–38.7]
Highest trough teicoplanine concentration (µg/mL)	20.7 [16.7–24.7]
Albumin	3.03 [2.65–3.32]
White blood cell count (×103/mm3)	0.64 [0.15–1.84]
Hemoglobin (g/dL)	8.1 [7.6–9.0]
Platelet count (×103/µL)	29.0 [15.8–49.0]
C-reactive protein (mg/dL)	4.61 [2.06–8.77]
Blood urea nitrogen (mg/dL)	14.8 [11.0–21.7]
Serum creatinine (mg/dL)	0.68 [0.53–0.91]
Creatinine clearance (mL/min)	84.3 [59.7–120.2]
Primary disease; n (%)
Acute myeloid leukemia	60 (34.1)
Acute lymphocytic leukemia	20 (11.4)
Adult T-cell leukemia	4 (2.3)
Multiple myeloma	13 (7.4)
Lymphoma	36 (20.5)
Myelodysplastic syndromes	13 (7.4)
Others	30 (17.0)
Infection type; n (%)
Febrile neutropenia	113 (64.2)
Pneumonia	8 (4.5)
Sepsis	18 (10.2)
Fever during neutropenia without fulfilling diagnostic for febrile neutropenia	30 (17.0)
Others	7 (4.0)
Concomitant drug; n (%)
Aminoglycoside	1 (0.6)
Sulfamethoxazole/trimethoprim	70 (39.8)
Amphotericin B	7 (4.0)
Acyclovir	87 (49.4)
Loop diuretic	58 (33.0)
Calcineurin inhibitor	56 (31.8)
Non-steroidal anti-inflammatory drugs	94 (53.4)
Angiotensin-converting enzyme inhibitors or angiotensin receptor blocker	11 (6.3)
Tazobactam/piperacillin	58 (33.0)

Data are expressed as number (%) for categorical variables, and median [interquartile range] for continuous variables.

Comparison between TEIC and TEIC + TAZ/PIPC Cohorts with Propensity Score Matching

Fifty-one patients in each cohort were matched by propensity scores. Tables 2, 3 present univariate analyses comparing patient characteristics and clinical laboratory parameters between TEIC and TEIC + TAZ/PIPC cohorts before and after propensity score matching, respectively. The highest trough TEIC concentration was significantly higher in TEIC cohort than in TEIC + TAZ/PIPC cohort before matching (p = 0.045), while the difference was no longer significant after matching (p = 0.216).

Table 2. Comparison of Patient Characteristics and Clinical Laboratory Parameters between Teicoplanin and Teicoplanin + Tazobactam/Piperacillin Cohorts before Propensity Score Matching

Parameter	TEIC	TEIC + TAZ/PIPC	p-Value
Patients; n	118	58
Sex (male/female); n (%)	65 (55.1)/53 (44.9)	30 (51.7)/28 (48.3)	0.674
Age (years)	57.0 [45.0–66.0]	56.0 [46.0–64.8]	0.760
Body weight (kg)	56.4 [48.5–63.7]	53.7 [48.4–62.4]	0.446
Height (cm)	161.9 [156.4–168.4]	160.9 [154.1–168.5]	0.582
Highest body temperature (°C)	38.3 [37.9–38.6]	38.2 [37.7–38.8]	0.627
TEIC administration period	9.0 [6.0–14.0]	8.0 [5.0–12.0]	0.069
Highest trough teicoplanine concentration (µg/mL)	20.7 [18.1–25.1]	19.9 [15.7–22.3]	0.045
Albumin	2.98 [2.64–3.27]	3.11 [2.73–3.47]	0.112
White blood cell count (×103/mm3)	0.67 [0.20–1.82]	0.61 [0.13–1.85]	0.363
Hemoglobin (g/dL)	8.2 [7.7–8.9]	8.0 [7.6–9.1]	0.935
Platelet count (×103/µL)	29.0 [15.0–45.8]	28.5 [16.0–51.3]	0.552
C-reactive protein (mg/dL)	5.18 [1.82–9.76]	4.45 [2.50–7.43]	0.761
Blood urea nitrogen (mg/dL)	15.3 [11.1–23.7]	13.6 [10.3–17.6]	0.052
Serum creatinine (mg/dL)	0.68 [0.55–0.91]	0.70 [0.51–0.87]	0.695
Creatinine clearance (mL/min)	86.9 [58.3–118.0]	79.1 [60.3–120.4]	0.885
Infection type; n (%)
Febrile neutropenia	76 (64.4)	37 (63.8)	0.936
Pneumonia	6 (5.1)	2 (3.4)	0.475
Sepsis	11 (9.3)	7 (12.1)	0.572
Fever during neutropenia without fulfilling diagnostic for febrile neutropenia	22 (18.6)	8 (13.8)	0.421
Others	3 (2.5)	4 (6.9)	0.163
Concomitant drug; n (%)
Aminoglycoside	1 (0.8)	0 (0.0)	0.670
Sulfamethoxazole/trimethoprim	44 (37.3)	26 (44.8)	0.337
Amphotericin B	6 (5.1)	1 (1.7)	0.265
Acyclovir	59 (50.0)	28 (48.3)	0.830
Loop diuretic	44 (37.3)	14 (24.1)	0.081
Calcineurin inhibitor	35 (29.7)	21 (36.2)	0.381
Non-steroidal anti-inflammatory drugs	62 (52.5)	32 (55.2)	0.742
Angiotensin-converting enzyme inhibitors or angiotensin receptor blocker	11 (9.3)	5 (8.6)	0.879

Data are expressed as number (%) for categorical variables, and median [interquartile range] for continuous variables. Categorical variables were analyzed by chi-square test or Fisher’s exact test. Continuous variables were analyzed by Mann–Whitney U test. TEIC, teicoplanin; TAZ/PIPC, tazobactam/piperacillin.

Table 3. Comparison of Patient Characteristics and Clinical Laboratory Parameters between Teicoplanin and Concomitant Teicoplanin + Tazobactam/Piperacillin Cohorts after Propensity Score Matching

Parameter	TEIC	TEIC + TAZ/PIPC	p-Value	Std diff
Patients; n	51	51
*Sex (male/female); n (%)	27 (52.9)/24 (47.1)	26 (51.0)/25 (49.0)	0.843	0.039
*Age (years)	53.0 [40.0–66.0]	57.0 [46.0–64.5]	0.715	0.003
Body weight (kg)	54.2 [47.5–62.5]	52.3 [47.9–62.8]	0.875	—
Height (cm)	162.5 [156.4–168.3]	161.0 [154.0–168.5]	0.585	—
Highest body temperature (°C)	38.3 [38.0–38.6]	38.1 [37.7–38.8]	0.620	—
TEIC administration period	9.0 [6.0–14.0]	7.0 [4.0–12.0]	0.063	—
Highest trough teicoplanine concentration (µg/mL)	21.2 [18.6–24.5]	20.1 [15.7–22.5]	0.216	—
*Albumin (g/dL)	2.98 [2.70–3.38]	3.11 [2.65–3.46]	0.763	0.010
White blood cell count (×103/mm3)	0.70 [0.20–1.81]	0.63 [0.13–1.73]	0.154	—
*Hemoglobin (g/dL)	8.0 [7.5–8.6]	7.9 [7.6–9.1]	0.896	0.020
Platelet count (×103/µL)	32.0 [17.0–45.5]	29.0 [16.5–49.0]	0.875	—
*C-reactive protein (mg/dL)	5.31 [2.26–9.98]	4.39 [2.54–7.38]	0.720	0.054
*Blood urea nitrogen (mg/dL)	16.5 [11.4–21.7]	13.6 [10.3–17.8]	0.098	0.287
Serum creatinine (mg/dL)	0.68 [0.55–1.02]	0.66 [0.49–0.85]	0.518	—
*Creatinine clearance (mL/min)	90.4 [58.8–111.0]	84.7 [60.4–123.4]	0.792	0.154
Infection type; n (%)
*Febrile neutropenia	33 (64.7)	35 (68.6)	0.674	0.083
Pneumonia	5 (9.8)	2 (3.9)	0.218	—
*Sepsis	3 (5.9)	3 (5.9)	0.661	0.000
Fever during neutropenia without fulfilling diagnostic for febrile neutropenia	8 (15.7)	10 (19.6)	0.603	—
Others	2 (3.9)	1 (2.0)	0.500	—
Concomitant drug; n (%)
*Aminoglycoside	0 (0.0)	0 (0.0)	—	—
*Sulfamethoxazole/trimethoprim	24 (47.1)	22 (43.1)	0.691	0.079
*Amphotericin B	1 (2.0)	1 (2.0)	0.752	0.000
*Acyclovir	23 (45.1)	25 (49.0)	0.692	0.079
*Loop diuretic	18 (35.3)	13 (25.5)	0.282	0.214
*Calcineurin inhibitor	24 (47.1)	16 (31.4)	0.105	0.326
*Non-steroidal anti-inflammatory drugs	29 (56.9)	28 (54.9)	0.842	0.039
*Angiotensin-converting enzyme inhibitors or angiotensin receptor blocker	4 (7.8)	4 (7.8)	0.642	0.000

Data are expressed as number (%) for categorical variables, and median [interquartile range] for continuous variables. Categorical variables were analyzed by chi-square test or Fisher’s exact test. Continuous variables were analyzed by Mann–Whitney U test. * indicates factors used for propensity score matching. TEIC, teicoplanin; TAZ/PIPC, tazobactam/piperacillin; Std diff, standardized difference.

Table 4 compares the incidence of AKI between TEIC and TEIC + TAZ/PIPC cohorts before and after propensity score matching. Eleven (9.3%) of 115 patients in TEIC cohort and 6 (10.3%) of 58 in TEIC + TAZ/PIPC cohort developed AKI, with no significant difference in incidence of AKI between the two cohorts (p = 0.829). Similarly, no significant difference between cohorts was found after matching (p = 0.244).

Table 4. Comparison of Incidence of Acute Kidney Injury between Teicoplanin and Concomitant Teicoplanin + Tazobactam/Piperacillin Cohorts before and after Propensity Score Analysis

	Before propensity score matching					After propensity score matching
	TEIC	TEIC +  TAZ/PIPC	p-Value	Odds ratio	95%CI	TEIC	TEIC +  TAZ/PIPC	p-Value	Odds ratio	95%CI
Acute kidney injury (+)	11 (9.3)	6 (10.3)	0.829	1.122	0.393–3.202	3 (5.9)	6 (11.8)	0.244	2.133	0.503–9.043
Acute kidney injury (−)	107 (90.7)	52 (89.7)				48 (94.1)	45 (88.2)

Data are expressed as number (%). p-values, odds ratio, and 95% CI were obtained from chi-square test of TEIC vs. TEIC + TAZ/PIPC cohorts before and after propensity score matching. TEIC, teicoplanin; TAZ/PIPC, tazobactam/piperacillin; CI, confidence interval.

Identification of Risk Factors for AKI by Univariate and Multivariate Logistic Regression

Of all patients, 17 (9.7%) satisfied the definition of AKI. In univariate analysis, the highest trough TEIC concentration was significantly higher and HGB level was significantly lower in AKI cohort than in non-AKI cohort (highest trough TEIC concentration: p = 0.016; HGB level: p = 0.004) (Table 5). In addition, AKI cohort used loop diuretics concomitantly with TEIC more frequently than non-AKI cohort (p = 0.017). A logistic regression analysis using 16 factors potentially related to TEIC-associated AKI as independent variables identified HGB level prior to TEIC initiation as the only significant risk factor for the development of AKI when using TEIC (p = 0.011; OR 0.484, 95% CI 0.276–0.848) (Table 6).

Table 5. Comparison of Patient Characteristics and Clinical Laboratory Parameters between Patients Who Developed Acute Kidney Injury (AKI Cohort) and Those Who Did Not (Non-AKI Cohort)

Parameter	Non-AKI	AKI	p-Value
Patients; n	159	17
Sex (male/female); n (%)	88 (55.3)/71 (44.7)	7 (41.2)/10 (58.8)	0.265
Age (years)	57.0 [45.0–66.0]	55.0 [45.0–58.0]	0.341
Body weight (kg)	56.0 [48.3–63.8]	55.2 [51.6–60.4]	0.822
Height (cm)	161.5 [156.2–168.5]	160.8 [152.5–165.5]	0.236
Highest body temperature (°C)	38.2 [37.8–38.7]	38.3 [37.5–38.7]	0.930
TEIC administration period	8.0 [5.0–13.0]	9.0 [5.0–20.0]	0.284
Highest trough teicoplanine concentration (µg/mL)	20.5 [16.4–23.5]	24.9 [20.7–27.1]	0.016
Albumin (g/dL)	3.04 [2.67–3.33]	2.89 [2.56–3.21]	0.410
White blood cell count (×103/mm3)	0.58 [0.15–1.80]	0.79 [0.28–2.37]	0.336
Hemoglobin (g/dL)	8.3 [7.7–9.1]	7.8 [7.2–8.0]	0.004
Platelet count (×103/µL)	30.0 [16.0–49.0]	16.0 [15.0–40.0]	0.104
C-reactive protein (mg/dL)	4.72 [2.16–8.87]	2.50 [1.30–7.66]	0.454
Blood urea nitrogen (mg/dL)	14.5 [11.0–21.7]	17.0 [11.1–20.9]	0.335
Serum creatinine (mg/dL)	0.69 [0.54–0.92]	0.64 [0.52–0.77]	0.531
Creatinine clearance (mL/min)	83.9 [59.3–118.6]	103.4 [75.4–122.9]	0.510
Infection type; n (%)
Febrile neutropenia	101 (63.5)	12 (70.6)	0.564
Pneumonia	8 (5.0)	0 (0.0)	0.436
Sepsis	17 (10.7)	1 (5.9)	0.459
Fever during neutropenia without fulfilling diagnostic for febrile neutropenia	27 (17.0)	3 (17.6)	0.583
Others	6 (3.8)	1 (5.9)	0.515
Concomitant drug; n (%)
Tazobactam/piperacillin	52 (32.7)	6 (35.3)	0.829
Aminoglycoside	0 (0.0)	1 (5.9)	0.097
Sulfamethoxazole/trimethoprim	63 (39.6)	7 (41.2)	0.901
Amphotericin B	5 (3.1)	2 (11.8)	0.138
Acyclovir	77 (48.4)	10 (58.8)	0.415
Loop diuretic	48 (30.2)	10 (58.8)	0.017
Calcineurin inhibitor	49 (30.8)	7 (41.2)	0.156
Non-steroidal anti-inflammatory agent	84 (52.8)	10 (58.8)	0.638
Angiotensin-converting enzyme inhibitors or angiotensin receptor blocker	16 (10.1)	0 (0.0)	0.182

Data are expressed as number (%) for categorical variables, and median [interquartile range] for continuous variables. Categorical variables were analyzed by chi-square test or Fisher’s exact test. Continuous variables were analyzed by Mann–Whitney U test.

Table 6. Multivariate Logistic Regression Analysis of Factors Associated with Incidence of Acute Kidney Injury

Covariate (multivariate analysis)	Coefficient	p-Value	Odds ratio
			Estimate	95% CI
Sex	−0.721	0.279	0.486	0.132 to 1.792
Age	−0.014	0.543	0.986	0.944 to 1.031
Highest trough teicoplanin concentration	0.07	0.139	1.073	0.978 to 1.177
Albumin	−0.455	0.536	0.635	0.15 to 2.682
Hemoglobin	−0.727	0.011	0.484	0.276 to 0.848
C-reactive protein	−0.039	0.466	0.961	0.865 to 1.069
Blood urea nitrogen	0.04	0.111	1.041	0.991 to 1.094
Febrile neutropenia	0.138	0.859	1.147	0.253 to 5.204
Sepsis	1.384	0.295	3.991	0.299 to 53.185
Tazobactam/piperacillin	−0.703	0.284	0.495	0.137 to 1.793
Sulfamethoxazole/trimethoprim	0.361	0.577	1.435	0.402 to 5.12
Amphotericin B	−1.774	0.117	0.170	0.019 to 1.555
Acyclovir	−0.003	0.996	0.997	0.270 to 3.681
Loop diuretic	−1.221	0.055	0.295	0.085 to 1.026
Calcineurin inhibitor	−0.276	0.652	0.759	0.229 to 2.517
Non-steroidal anti-inflammatory drugs	−0.592	0.354	0.553	0.158 to 1.936
Angiotensin-converting enzyme inhibitorsor angiotensin receptor blocker	19.987	0.997	<0.001	—

CI, confidence interval.

DISCUSSION

TAZ/PIPC combined with glycopeptide antibiotics are often used in patients with hematological malignancies who develop severe infections. Many previous reports have revealed significant increases in AKI incidence with concomitant use of VCM and TAZ/PIPC.^5–11) Hence, TEIC that has lower AKI risk than VCM is often selected when used with TAZ/PIPC in patients with hematological malignancies. However, to our knowledge, no reports have evaluated whether TAZ/PIPC is a risk factor for AKI associated with TEIC. This study retrospectively explored the impact of concomitant TAZ/PIPC use on TEIC-associated AKI in patients with hematological malignancies and identified potential risk factors. The following results were obtained: (1) there were no significant differences in AKI incidence and odds ratio between TEIC and TEIC + TAZ/PIPC cohorts, even after propensity score matching; (2) univariate and multivariate analyses identified baseline HGB level as a significant risk factor for TEIC-associated AKI, but not concomitant use of TAZ/PIPC.

No difference in AKI incidence between TEIC and TEIC + TAZ/PIPC cohorts was found even after propensity score matching. This result is consistent with a previous report.¹⁷⁾ Although the underlying pathogenesis of VCM-related AKI is not fully understood, experimental studies support the involvement of mitochondrial dysfunction, proinflammatory oxidative stress, and renal tubular cell apoptosis.²²⁾ On the contrary, there are no reports of in vitro and in vivo experimental studies on the mechanism of TEIC-associated AKI, although it cannot be ruled out that TEIC causes AKI by similar mechanisms as VCM. Beta-lactam antibiotics including TAZ/PIPC are well-known to cause drug-associated acute interstitial nephritis.²³⁾ No data from previous experimental models, human renal biopsies, and biomarker studies demonstrate a synergistic increase in acute tubular injury by combining TAZ/PIPC with VCM compared to VCM alone.²⁴⁾ The mechanisms of renal damage caused by VCM and TAZ/PIPC are different, and co-administration may increase the risk additively. A systematic review and meta-analysis demonstrated less frequent nephrotoxicity with TEIC than with VCM (relative risk 0.44; 95% CI 0.32 to 0.61).²⁵⁾ Hence, the risk of developing TEIC-associated AKI may not increase by combining TAZ/PIPC with TEIC, unlike with VCM. On the other hand, both TAZ and PIPC are known substrates of OAT1 and OAT3,²⁶⁾ and vancomycin partially downregulates OAT1 and OAT3 mRNA levels and protein expression.²⁷⁾ These drugs could diminish tubular transporters causing impaired creatinine secretion and elevated Scr, which resembles AKI, and is referred to as pseudo-toxic drug interaction. However, whether exposure of renal tubules to TEIC downregulates OAT1 and OAT3 expression remains unclear. While the possibility of pseudo-toxic drug interaction between TEIC and TAZ/PIPC is still controversial, our results suggest that the interaction would be unlikely because AKI risk associated with TEIC did not increase when TAZ/PIPC was used concomitantly.

This retrospective study recruited patients with hematological malignancies, whose renal function is likely to augment. This may lead to a pseudo-decrease in the incidence of AKI. However, the incidence of AKI in TEIC and TEIC + TAZ/PIPC cohorts was, respectively, 9.3 and 10.3% before propensity score matching, and 5.9 and 11.8% after matching, which is in line with previous literature.^{13,15–17,21)} In addition, 17 of a total of 176 patients receiving TEIC developed AKI, which is comparable with the incidence in a previous systematic review and meta-analysis by Svetitsky et al. (94 of 914 patients).²⁵⁾ Univariate analysis found that the highest trough TEIC concentration in AKI cohort was significantly higher than that in non-AKI cohort. Clinical practice guidelines for therapeutic drug monitoring of TEIC revised by the Japanese Society of Chemotherapy and the Japanese Society of Therapeutic Drug Monitoring recommend maintaining the trough concentration within 15–30 µg/mL for uncomplicated MRSA infections and within 20–40 µg/mL for complicated or severe MRSA infections.²⁸⁾ Maintaining these ranges is suggested to improve treatment success without increasing the risk of adverse effects including AKI. The median highest trough TEIC concentration in AKI cohort was 24.9 µg/mL, and no patient exceeded 40 µg/mL. In addition, logistic regression analysis did not identify the highest trough TEIC concentration as a risk factor for AKI. These findings suggest a low impact of TEIC trough concentration on AKI development.

Univariate analysis showed that patients receiving loop diuretics were more likely to develop TEIC-associated AKI than those without loop diuretics (p = 0.017). Loop diuretics suppress sodium reabsorption by inhibiting the Na-K-2Cl cotransporter expressed in the ascending limb of the loop of Henle. Loop diuretics are the most commonly prescribed fluid management agents with indication for water and sodium retention due to heart disease, hypertension, and kidney diseases including AKI.²⁹⁾ However, many studies suggested an increased risk for AKI development with loop diuretic use. Levi et al.³⁰⁾ reported that furosemide use was associated with increased risk of AKI development in critically ill patients with sepsis or septic shock. A retrospective case-control study of 487372 individuals by Lapi et al.³¹⁾ demonstrated an increased risk for AKI development by concomitant use of diuretics with NSAIDs and angiotensin receptor blockers/angiotensin-converting enzyme inhibitors. Another case-control study suggested synergistically amplified nephrotoxicity associated with combined use of NSAIDs with renin–angiotensin system inhibitors and/or diuretics.³²⁾ The underlying biological mechanism for loop diuretic-associated AKI remains unclear. However, loop diuretics can improve filling pressure and relieve water load in the thick ascending limbs of the loop of Henle by suppressing sodium reabsorption, which could cause volume depletion and possible renal hypoperfusion.³³⁾ In addition, loop diuretics could bind to antigens in the kidney or act as antigens, thereby depositing in the interstitium to induce immune reactions causing acute interstitial nephritis.³⁴⁾ A recent systematic review and meta-analysis identified loop diuretics as a risk factor for VCM-associated acute kidney injury.³⁵⁾ Loop diuretics reduce volume distribution of VCM by lowering the extracellular volume and deteriorate glomerular filtration rate by increasing urine volume.³⁶⁾ This could lead to increased VCM exposure and nephrotoxicity. Since TEIC is also primarily excreted by glomerular filtration, exposure to TEIC may increase when used with loop diuretics. However, this possibility would be unlikely because TEIC has lower urinary excretion rate and higher threshold for nephrotoxicity than VCM. In fact, loop diuretics were not identified as a risk factor for TEIC-associated AKI in multivariate logistic regression analysis.

Univariate and multivariate analysis identified only HGB level prior to TEIC initiation as a significant risk factor for AKI (p = 0.011). Many prospective and retrospective studies have demonstrated that low HGB is a significant risk factor for AKI associated with sepsis,³⁷⁾ cardiovascular surgery,³⁸⁾ and medications that possibly cause renal impairment.^39–41) Sreenivasan et al.³⁹⁾ reported that severity of anemia defined by hemoglobin level was a risk factor for contrast-induced AKI following coronary angiography. In addition, Chou et al.⁴¹⁾ identified low albumin or hemoglobin levels as relevant risk factors for parenteral aminoglycosides-associated AKI in a large study of 43259 individuals without a history of chronic kidney disease. Low HGB level causes oxygen depletion as a result of reduced oxygen-carrying capacity in the blood^42,43) and decreased hemoglobin level in overt bleeding may reduce renal perfusion pressure.⁴⁴⁾ As a result, insufficient hemoglobin cannot deliver sufficient oxygen to the kidney tissue, leading to subsequent kidney damage. As aforementioned, the molecular mechanism of TEIC-associated AKI has not been elucidated, but assuming that it is similar to VCM, renal tubular apoptosis and mitochondrial dysfunction are likely to be involved. Therefore, low baseline HGB level was identified as a risk factor presumably reflecting an additive effect of AKI pathogenesis derived from different mechanisms.

This retrospective, single-center, observational cohort study has some limitations. First, the sample sizes in both cohorts were small, with only 51 patients in each cohort after propensity score matching. Many factors were potentially associated with AKI relative to the sample size, resulting in a standardized difference of several covariates after matching >0.1, suggesting inadequate variable balancing. Second, subject selection bias may have occurred when physicians chose TEIC rather than VCM for the treatment regimen. Concrete evidence of increased VCM-associated AKI risk with concomitant use of TAZ/PIPC has been established. Hence, physicians may avoid using VCM in patients receiving TAZ/PIPC, who had a high risk of AKI such as those with reduced kidney function. In other words, the TEIC + TAZ/PIPC cohort possibly had more patients with AKI risk at baseline. This study included only patients with hematological malignancies, and confounding factors were adjusted for patient background and clinical laboratory parameters using propensity score. However, since this propensity score analysis cannot include all the relevant factors in the model, some confounding factors would remain. Third, urine volume was not included in the definition of AKI in this study, despite using AKI criteria according to KDIGO, because these data were not available in many patients. This study could have missed patients with AKI who would have been included if urine volume were considered. Fourth, the severity of the infection is associated with the risk of developing AKI, but this study excluded patients admitted to the emergency center or intensive care unit during TEIC administration. Therefore, the Glasgow Coma Scale, arterial blood gases, and the severity of organ dysfunction were not measured in some patients, and the severity of infection could not be evaluated by APACHE II and SOFA scores in all patients. Finally, serum cystatin C is a better indicator than serum creatinine for the early diagnosis of AKI and may, therefore, provide a more accurate assessment of AKI associated with TEIC. However, it was not possible to obtain sufficient serum cystatin C levels from electronic medical records to assess AKI in many patients.

In conclusion, a propensity score matching analysis found no significant increase in incidence of TEIC-associated AKI by concomitant use of TAZ/PIPC. Univariate and multivariate regression analyses identified baseline HGB level as a risk factor for developing TEIC-associated AKI, but not concomitant TAZ/PIPC use. Given these results, use of TAZ/PIPC in combination did not increase the risk of developing TEIC-associated AKI in patients with hematologic malignancies. On the other hand, lower HGB level had a higher risk of developing TEIC-associated AKI, suggesting the necessity to carefully monitor Scr levels when using TEIC in patients with anemia.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflict of Interest

The authors declare no conflict of interest.

REFERENCES

• 1) Perry CM, Markham A. Piperacillin/tazobactam: an updated review of its use in the treatment of bacterial infections. Drugs, 57, 805–843 (1999).
• 2) Sawada A, Kawanishi K, Morikawa S, Nakano T, Kodama M, Mitobe M, Taneda S, Koike J, Ohara M, Nagashima Y, Nitta K, Mochizuki T. Biopsy-proven vancomycin-induced acute kidney injury: a case report and literature review. BMC Nephrol., 19, 72 (2018).
• 3) Luque Y, Louis K, Jouanneau C, Placier S, Esteve E, Bazin D, Rondeau E, Letavernier E, Wolfromm A, Gosset C, Boueilh A, Burbach M, Frère P, Verpont MC, Vandermeersch S, Langui D, Daudon M, Frochot V, Mesnard L. Vancomycin-associated cast nephropathy. J. Am. Soc. Nephrol., 28, 1723–1728 (2017).
• 4) Elyasi S, Khalili H, Dashti-Khavidaki S, Mohammadpour A. Vancomycin-induced nephrotoxicity: mechanism, incidence, risk factors and special populations. A literature review. Eur. J. Clin. Pharmacol., 68, 1243–1255 (2012).
• 5) Burgess LD, Drew RH. Comparison of the incidence of vancomycin-induced nephrotoxicity in hospitalized patients with and without concomitant piperacillin-tazobactam. Pharmacotherapy, 34, 670–676 (2014).
• 6) Oda K, Hashiguchi Y, Katanoda T, Nakata H, Jono H, Saito H. Lowered risk of nephrotoxicity through intervention against the combined use of vancomycin and tazobactam/piperacillin: a retrospective cohort study. Microbiol. Spectr., 9, e0035521 (2021).
• 7) Hammond DA, Smith MN, Li C, Hayes SM, Lusardi K, Bookstaver PB. Systematic review and meta-analysis of acute kidney injury associated with concomitant vancomycin and piperacillin/tazobactam. Clin. Infect. Dis., 64, 666–674 (2017).
• 8) Luther MK, Timbrook TT, Caffrey AR, Dosa D, Lodise TP, LaPlante KL. Vancomycin plus piperacillin-tazobactam and acute kidney injury in adults: a systematic review and meta-analysis. Crit. Care Med., 46, 12–20 (2018).
• 9) Gomes DM, Smotherman C, Birch A, Dupree L, Della Vecchia BJ, Kraemer DF, Jankowski CA. Comparison of acute kidney injury during treatment with vancomycin in combination with piperacillin-tazobactam or cefepime. Pharmacotherapy, 34, 662–669 (2014).
• 10) Rutter WC, Cox JN, Martin CA, Burgess DR, Burgess DS. Nephrotoxicity during vancomycin therapy in combination with piperacillin-tazobactam or cefepime. Antimicrob. Agents Chemother., 61, e02089-16 (2017).
• 11) Mullins BP, Kramer CJ, Bartel BJ, Catlin JS, Gilder RE. Comparison of the nephrotoxicity of vancomycin in combination with cefepime, meropenem, or piperacillin/tazobactam: a prospective, multicenter study. Ann. Pharmacother., 52, 639–644 (2018).
• 12) Wood MJ. The comparative efficacy and safety of teicoplanin and vancomycin. J. Antimicrob. Chemother., 37, 209–222 (1996).
• 13) Hahn-Ast C, Glasmacher A, Arns A, Mühling A, Orlopp K, Marklein G, Von Lilienfeld-Toal M. An audit of efficacy and toxicity of teicoplanin versus vancomycin in febrile neutropenia: is the different toxicity profile clinically relevant? Infection, 36, 54–58 (2008).
• 14) Shao CH, Tai CH, Lin FJ, Wu CC, Wang JT, Wang CC. Comparison of risk of acute kidney injury between patients receiving the combination of teicoplanin and piperacillin/tazobactam versus vancomycin and piperacillin/tazobactam. J. Formos. Med. Assoc., 121, 117–125 (2022).
• 15) Sazanami K, Inose R, Yagi T, Dote S, Horiuchi N, Kobayashi Y, Muraki Y. Incidence of acute kidney injury after teicoplanin- or vancomycin- and piperacillin/tazobactam combination therapy: a comparative study using propensity score matching analysis. J. Infect. Chemother., 27, 1723–1728 (2021).
• 16) Aslan AT, Pashayev T, Dağ O, Akova M. Comparison of teicoplanin versus vancomycin in combination with piperacillin-tazobactam or meropenem for the risk of acute kidney injury. Eur. J. Clin. Microbiol. Infect. Dis., 40, 1953–1961 (2021).
• 17) Tai CH, Shao CH, Wang CC, Lin FJ, Wang JT, Wu CC. Nephrotoxicity of teicoplanin-based combination therapy: focus on piperacillin/tazobactam and other anti-pseudomonal β-lactams. J. Antimicrob. Chemother., 76, 499–506 (2021).
• 18) Workum JD, Kramers C, Kolwijck E, Schouten JA, de Wildt SN, Brüggemann RJ. Nephrotoxicity of concomitant piperacillin/tazobactam and teicoplanin compared with monotherapy. J. Antimicrob. Chemother., 76, 212–219 (2021).
• 19) Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron, 16, 31–41 (1976).
• 20) Tanaka R, Morinaga Y, Iwao M, Tatsuta R, Hashimoto T, Hiramatsu K, Itoh H. Comparison of incidence of hyponatremia between linezolid and vancomycin by propensity score matching analysis. Biol. Pharm. Bull., 46, 1365–1370 (2023).
• 21) Kellum JA, Lameire N. Diagnosis, evaluation, and management of acute kidney injury: a KDIGO summary (Part 1). Crit. Care, 17, 204 (2013).
• 22) Bamgbola O. Review of vancomycin-induced renal toxicity: an update. Ther. Adv. Endocrinol. Metab., 7, 136–147 (2016).
• 23) Moledina DG, Perazella MA. Drug-induced acute interstitial nephritis. Clin. J. Am. Soc. Nephrol., 12, 2046–2049 (2017).
• 24) Blair M, Côté JM, Cotter A, Lynch B, Redahan L, Murray PT. Nephrotoxicity from vancomycin combined with piperacillin-tazobactam: a comprehensive review. Am. J. Nephrol., 52, 85–97 (2021).
• 25) Svetitsky S, Leibovici L, Paul M. Comparative efficacy and safety of vancomycin versus teicoplanin: systematic review and meta-analysis. Antimicrob. Agents Chemother., 53, 4069–4079 (2009).
• 26) Wen S, Wang C, Duan Y, Huo X, Meng Q, Liu Z, Yang S, Zhu Y, Sun H, Ma X, Yang S, Liu K. OAT1 and OAT3 also mediate the drug-drug interaction between piperacillin and tazobactam. Int. J. Pharm., 537, 172–182 (2018).
• 27) Wen S, Wang C, Huo X, Meng Q, Liu Z, Yang S, Zhu Y, Sun H, Ma X, Liu K. JBP485 attenuates vancomycin-induced nephrotoxicity by regulating the expressions of organic anion transporter (Oat) 1, Oat3, organic cation transporter 2 (Oct2), multidrug resistance-associated protein 2 (Mrp2) and P-glycoprotein (P-gp) in rats. Toxicol. Lett., 295, 195–204 (2018).
• 28) Hanai Y, Takahashi Y, Niwa T, Mayumi T, Hamada Y, Kimura T, Matsumoto K, Fujii S, Takesue Y. Clinical practice guidelines for therapeutic drug monitoring of teicoplanin: a consensus review by the Japanese Society of Chemotherapy and the Japanese Society of Therapeutic Drug Monitoring. J. Antimicrob. Chemother., 77, 869–879 (2022).
• 29) Grams ME, Estrella MM, Coresh J, Brower RG, Liu KD. Fluid balance, diuretic use, and mortality in acute kidney injury. Clin. J. Am. Soc. Nephrol., 6, 966–973 (2011).
• 30) Levi TM, Rocha MS, Almeida DN, Martins RT, Silva MG, Santana NC, Sanjuan IT, Cruz CM. Furosemide is associated with acute kidney injury in critically ill patients. Braz. J. Med. Biol. Res., 45, 827–833 (2012).
• 31) Lapi F, Azoulay L, Yin H, Nessim SJ, Suissa S. Concurrent use of diuretics, angiotensin converting enzyme inhibitors, and angiotensin receptor blockers with non-steroidal anti-inflammatory drugs and risk of acute kidney injury: nested case-control study. BMJ, 346 (jan08 12), e8525 (2013).
• 32) Dreischulte T, Morales DR, Bell S, Guthrie B. Combined use of nonsteroidal anti-inflammatory drugs with diuretics and/or renin-angiotensin system inhibitors in the community increases the risk of acute kidney injury. Kidney Int., 88, 396–403 (2015).
• 33) Dikshit K, Vyden JK, Forrester JS, Chatterjee K, Prakash R, Swan HJ. Renal and extrarenal hemodynamic effects of furosemide in congestive heart failure after acute myocardial infarction. N. Engl. J. Med., 288, 1087–1090 (1973).
• 34) Naughton CA. Drug-induced nephrotoxicity. Am. Fam. Physician, 78, 743–750 (2008) ‹https://pubmed.ncbi.nlm.nih.gov/18819242/›.
• 35) Kim JY, Yee J, Yoon HY, Han JM, Gwak HS. Risk factors for vancomycin-associated acute kidney injury: a systematic review and meta-analysis. Br. J. Clin. Pharmacol., 88, 3977–3989 (2022).
• 36) Hirai T, Hanada K, Kanno A, Akashi M, Itoh T. Risk factors for vancomycin nephrotoxicity and time course of renal function during vancomycin treatment. Eur. J. Clin. Pharmacol., 75, 859–866 (2019).
• 37) Ruan X, Wang B, Gao Y, Wu J, Yu X, Liang C, Pan J. Assessing the impact of transfusion thresholds in patients with septic acute kidney injury: a retrospective study. Front. Med. (Lausanne), 10, 1308275 (2023).
• 38) Tseng PY, Chen YT, Wang CH, Chiu KM, Peng YS, Hsu SP, Chen KL, Yang CY, Lee OK. Prediction of the development of acute kidney injury following cardiac surgery by machine learning. Crit. Care, 24, 478 (2020).
• 39) Sreenivasan J, Zhuo M, Khan MS, Li H, Fugar S, Desai P, Yadav N. Anemia (Hemoglobin ≤13 g/dL) as a risk factor for contrast-induced acute kidney injury following coronary angiography. Am. J. Cardiol., 122, 961–965 (2018).
• 40) Mu F, Cui C, Tang M, Guo G, Zhang H, Ge J, Bai Y, Zhao J, Cao S, Wang J, Guan Y. Analysis of a machine learning-based risk stratification scheme for acute kidney injury in vancomycin. Front. Pharmacol., 13, 1027230 (2022).
• 41) Chou CL, Chuang NC, Chiu HW, Liao CT, Hsu YH, Chang TH. Aminoglycosides use has a risk of acute kidney injury in patients without prior chronic kidney disease. Sci. Rep., 12, 17212 (2022).
• 42) Abuelo JG. Normotensive ischemic acute renal failure. N. Engl. J. Med., 357, 797–805 (2007).
• 43) Rosenberger C, Rosen S, Heyman SN. Renal parenchymal oxygenation and hypoxia adaptation in acute kidney injury. Clin. Exp. Pharmacol. Physiol., 33, 980–988 (2006).
• 44) Walsh M, Garg AX, Devereaux PJ, Argalious M, Honar H, Sessler DI. The association between perioperative hemoglobin and acute kidney injury in patients having noncardiac surgery. Anesth. Analg., 117, 924–931 (2013).