Prognostic value of preoperative nutritional status for postoperative moderate to severe acute kidney injury among older patients undergoing coronary artery bypass graft surgery: a retrospective study based on the MIMIC-IV database Authors: Peng Bao, Peng Qiu, Tao Li, Xue Lv, Junyu Wu, Shaojie Wu, Hao Li, Zhiping Guo Published: 2024 ABSTRACT AbstractObjectiveTo investigate the association between preoperative nutritional scores and moderate-to-severe acute kidney injury (AKI) after coronary artery bypass graft (CABG) surgery and the predictive significance of nutritional indices for moderate to severe AKI.MethodsThis study retrospectively included older patients underwent CABG surgery from the Medical Information Mart for Intensive Care (MIMIC) database. Nutritional scores were calculated by the Geriatric Nutritional Risk Index (GNRI) and the Prognostic Nutritional Index (PNI), respectively. Moderate-to-severe injury was determined by KDIGO criteria. Logistic regression, subgroup analysis, and restricted cubic splines were utilized to investigate the association. The predictive value was also assessed by the area under the curve (AUC), net reclassification index (NRI), and integrated discrimination improvement (IDI).ResultsA total of 1,007 patients were retrospectively included, of which 100 (9.9%) and 380 (37.7%) had malnutrition calculated by GNRI and PNI scores. The incidence of moderate-to-severe AKI was 524 (52.0%). After adjustment for selected risk factors, worse nutritional scores were associated with a higher incidence of moderate-to-severe AKI (PGNRI<0.001; PPNI=0.001). Integrating these indices into different base models improves their performance, as manifested by significant improvements in AUCs and NRIs (p < 0.05).ConclusionWorse preoperative nutritional status was associated with an elevated risk of postoperative moderate-to-severe AKI. Integrating these indices into base models improve their predictive performance. These results highlight the importance of assessing nutritional status among older patients had CABG surgery.Keywords: Acute kidney injury, CABG, malnutrition, prediction model Introduction Acute kidney injury (AKI) after coronary artery bypass graft (CABG) surgery is a widespread complication with an incidence between 19% and 28% [1–3], and it is particularly common in older patients [4]. In different types of cardiac surgery, the AKI incidence in patients with CABG surgery was lower compared to those who had aortic surgery or valve surgery [5]. Its severe form is independently associated with elevated morbidity and mortality in the cardiac surgery patients [6,7]. The mortality risk associated with AKI following cardiac surgery persists for 10 years and is independent of additional risk factors, even for patients who have recovered their kidneys completely [8]. Moderate-to-severe (stage 2 to 3) AKI has attached greater attention due to the complexity of its treatment. Timely recognition of AKI is important in order to initiate a comprehensive treatment strategy to minimize further damage and decreased associated mortality [9]. However, the pathogenesis of AKI following cardiac surgery is complicated and incompletely understood [10], early identifying predictive factors associated with AKI is urgent. Malnutrition involves multi-dimensional conditions characterized by a reduction of the body fat and/or muscle mass as a result of insufficient nutrient intake or utilization [11]. The incidence of malnutrition in the elderly is 25% [12]. Worse nutritional status is associated with many unfavorable outcomes in older patients, including delirium, in-hospital mortality, re-hospitalizations rate and long-term mortality [13–15]. The assessment methods based on the questionnaires to screen nutritional status have previously been utilized extensively. However, these approaches are inappropriate for geriatrics because of the recall bias and communication problems [13,14]. To overcome these limitations of questionnaires, some objective nutritional indicators based on routine laboratory tests have been validated and utilized, including the geriatric nutritional risk index (GNRI) and prognostic nutritional index (PNI). The sensitivity and specificity of the GNRI for assessing the malnutrition were 62.2% and 97.7% [16]. As for the PNI, the sensitivity and specificity for evaluating the malnutrition were 74.1%–80.8% and 66.4%–88.2% due to different subjects [17,18]. In addition, preoperative poor nutritional status calculated by GNRI or PNI was significantly associated with elevated AKI incidence [19–21], and higher mortality in cardiac surgery patients [22–24]. However, these studies failed to sufficiently take into account the complexity of nutritional scores being both a categorical and a continuous variable, nor did they consider the potential predictive value of GNRI and PNI. Small sample sizes are issues that cannot be disregarded. In addition, older patients rarely got much attention in patients only underwent CABG surgery. The purpose of this study is to completely investigate the association between the preoperative nutritional status, as evaluated by GNRI and PNI, and moderate-to-severe AKI following CABG surgery among older patients. Furthermore, we also determine whether GNRI and PNI can improve the model performance based on the traditional risk factors. Methods Study design The retrospective study used data derived from the Medical Information Mart for Intensive Care IV (MIMIC-IV, version 2.2) database, encompassing detailed medical information from patients admitted to the Intensive Care Units (ICUs) of Beth Israel Deaconess Medical Center from 2008 to 2019. The author passed the Collaborative Institutional Training Initiative (CITI) examination, received a certification (ID number: 60692864), and then accessed this database, which obviated the need for individual patient consent. This study included 5,956 patients who were hospitalized to the Cardiac Surgery ICU following CABG in the MIMIC IV database. The following exclusion criteria were applied for patients: (1) Not first hospital admission for CABG; (2) ICU admission before CABG; (3) ICU admission beyond the first one after CABG; (4) Age < 65 years; (5) Failure to evaluate postoperative AKI stage; (6) Lossing key information of preoperative serum albumin and lymphocyte counts. After applying these criteria, the final sample comprised 1,007 patients (Figure 1). Flow diagram for patient selection. ICU, intensive care unit; MIMIC, medical information mart for intensive care; CABG, coronary artery bypass grafting; AKI, acute kidney injury. Data collection The study involved the extraction of comprehensive patient data, including demographics (gender, age, heigh, weight, and body mass index [BMI]) and comorbidities (congestive heart failure, peripheral vascular disease, diabetes, hypertension, chronic pulmonary disease, chronic liver disease, stroke, chronic kidney disease, cerebrovascular disease, and Charlson comorbidity index). The definition of chronic kidney disease is an abnormality in the structure or function of the kidney that persists for more than three months and has an impact on health [25]. Additionally, preoperative clinical status indicators such as the Sequential Organ Failure Assessment (SOFA) score, Glasgow Coma Scale (GCS), heart rate, oxyhemoglobin saturation (SPO2), mean blood pressure (MBP), use of intra-aortic balloon pump (IABP), ventilators, and vasopressors were carefully documented. Preoperative laboratory parameters, including albumin, serum creatinine (Scr), hematocrit, blood urea nitrogen (BUN), potassium, sodium, bicarbonate, and lymphocyte counts, were also recorded. Lastly, the first postoperative laboratory values for creatinine, BUN, sodium, potassium, and bicarbonate, along with the duration between CABG surgery completion and the initial renal function measurement, were systematically collected. Outcome assessments The primary outcome was the severity of AKI within seven days following CABG operation, classified into stages 0, 1, 2, or 3 according to the Kidney Disease Improving Global Outcomes (KDIGO) criteria [26]. The highest Scr level recorded within this period determined the AKI stage, with stage 2 to 3 considered moderate-to-severe AKI. The baseline Scr level was evaluated using the most recent measurements prior to surgery. Secondary outcomes encompassed the length of stay (LOS) in ICU and in hospital, as well as hospital mortality, 30-day and 90-day mortality. Nutritional evaluation and categorization The GNRI was computed utilizing the formula: 1.489 × serum albumin (g/L) + 41.7 × (actual body weight [kg]/ideal body weight [kg]). In addition, this formula for calculating the ideal weight varies according to gender: 0.75 × height (cm) − 62.5 for males and 0.60 × height (cm) − 40 for females. Based on GNRI scores, patients were categorized into two groups reflecting their risk of malnutrition: no nutritional risk group (GNRI > 98) and nutritional risk group (GNRI ≤ 98) [27]. The PNI was evaluated by the formula: serum albumin (g/L) + 5 × total lymphocyte count (×109/L), dividing patients into following two groups: no nutritional risk group (PNI ≥ 48) and nutritional risk group (PNI <48) [19,24,28]. Statistical analysis The R software (version 4.2.1) and SPSS (version 25.0) were utilized to carry out the significant difference analysis. The Shapiro–Wilk test was utilized across all continuous variables to determine whether it is normally or abnormally distributed. They were expressed as mean ± standard deviation or medians (interquartile ranges [IQR]) and compared by the Mann–Whitney U-test. Afterwards, categorical data were stated as frequencies (percentages) and assessed utilizing the chi-square test or Fisher exact test. The relationship between the GNRI and the PNI was depicted by a Venn diagram. The Spearman’s correlation coefficients (r) were utilized to investigate the correlation between the GNRI and PNI. The association between nutritional condition and the incidence of moderate to severe AKI after CABG surgery was explored through logistic regression, treating nutritional indices as both continuous and categorical data. Covariables that were associated with moderate-to-severe AKI following CABG surgery in the adjusted model were calculated by multivariate logistic regression. To eliminate multicollinearity, components used in calculating malnutrition scores, such as weight, height, BMI, lymphocyte counts and serum albumin were omitted from the multivariate regression analysis. An initial univariate analysis was followed by a multivariate approach to adjust for screened confounders. Three multivariate models were developed to validate the association: Model 1 adjusted by demographic factors including gender, congestive heart failure, and chronic pulmonary disease; Model 2 expanded on Model 1 by incorporating preoperative conditions such as the SOFA score, MBP, preoperative hematocrit, and BUN; Model 3 further included postoperative creatinine and BUN concentrations. These results were reported as odds ratios (ORs) and 95% confidence intervals (CIs). The multivariable association between preoperative nutritional assessments and the incidence of moderate to severe AKI following CABG surgery was also elucidated using restricted cubic splines (RCS), which necessitated the treatment of the GNRI and PNI exclusively as continuous variables for statistical analysis. The RCS could not be displayed when the nutritional indices were regarded as categorical variables. Similarly, this approach was utilized to determine the optimal cutoff points of GNRI and PNI [29]. In order to further explain the importance of every categorical variable to the moderate-to-severe AKI after CABG surgery, subgroup analyses were conducted. There was an interaction effect when the p for interaction across subgroups < 0.05 [30]. Furthermore, the incremental predictive utility of these nutritional indices for moderate-to-severe AKI after CABG surgery was assessed by integrating them into three base models (Model 1, 2, and 3), thereby generating refined predictive models. The discriminative capability of these nutritional metrics was determined through the area under the curve (AUC), the net reclassification index (NRI), and integrated discrimination improvement (IDI). In addition, calibration plots, utilizing bootstrapping methods with 1000 resamples, were utilized to compare the predicted risks with the observed frequency, and Brier scores < 0.25 demonstrated goodness of fits. Decision curve analyses were subsequently applied to ascertain the clinical utility. The minimum sample size was determined by the package ʻpmsamplesizeʼ in the R software [31]. The minimum sample size was 915 individuals, according to the AUC of 0.79, 28.7% incidence of postoperative moderate to severe AKI and 14 variables [21,32]. The p-value < 0.05 denotes statistical significance. Results Clinical characteristics The selection process for patients was depicted, with a total of 1007 patients retrospectively included (Figure 1). As the result, the median age was 73.46 years (IQR: 69.1–79.02), with a predominance of male patients (73.1%). The median BMI was recorded at 28.96 (IQR: 25.89–32.49). Hypertension, diabetes, and congestive heart failure emerged as the most prevalent comorbidities. The median lengths of stay (LOS) in ICU and hospital were 1.98 days (IQR: 1.26–3.22) and 8.05 days (IQR: 5.74–11.45), respectively. Observed mortality rates for the hospital, 30-day, and 90-day periods were 1.3%, 1.6%, and 1.8%, respectively. Among all eligible patients, 798 (79.2%) had AKI following CABG surgery within 7 days, with 27.2% at stage 1 (n = 274), 44.7% at stage 2 (n = 450), and 7.3% at stage 3 (n = 74). The incidence of moderate-to-severe AKI (stages 2–3) was 524 (52.0%). The characteristics of the study cohort were showed, ranging from global to stratified data (Table 1 and Table S1). Patients with moderate-to-severe AKI typically exhibited a higher BMI, a shorter LOS in ICU and in hospital, and a higher prevalence of comorbidities, including congestive heart failure, hypertension, chronic pulmonary disease, and diabetes. These patients also presented a lower MBP and hematocrit levels prior to CABG surgery, whereas the SOFA score, utilization rate of IABP, the creatinine, and BUN were elevated compared to those without moderate to severe AKI. Furthermore, creatinine and BUN levels were higher in the first renal function test after CABG surgery. Additionally, the score of GNRI was higher and the metric of PNI was significantly lower among patients with moderate-to-severe AKI. Table 1.Baseline characteristics of study participants. overallno moderate to severe AKImoderate to severe AKIOR (95% CI)p value1007483524Gender (Male)736 (73.1)371 (76.8)365 (69.7)0.69 (0.52-0.92)0.013Age (year)73.46 [69.1–79.02]72.80 [68.91, 78.62]74.01 [69.21, 79.51]1.02 (1-1.04)0.128BMI (kg/m2)28.96 [25.89–32.49]27.64 [24.84, 30.97]30.20 [27.20, 33.91]1.12 (1.09-1.15)<0.001Height (cm)170.00 [165.00, 178.00]170.00 [165.00, 178.00]172.41 [164.29, 178.00]1.01 (1-1.02)0.157Weight (kg)84.05 [74.10–96.45]80.00 [71.50, 90.97]88.10 [77.00, 101.30]1.04 (1.03-1.04)<0.001Comorbid disease Charlson comorbidity index5.00 [4.00–7.00]5.00 [4.00, 6.00]5.00 [4.00, 7.00]1.17 (1.1-1.25)<0.001Congestive heart failure287 (28.5)98 (20.3)189 (36.1)2.22 (1.67-2.95)<0.001Peripheral vascular disease153 (15.2)71 (14.7)82 (15.6)1.08 (0.76-1.52)0.74Hypertension588 (58.4)298 (61.7)290 (55.3)0.77 (0.6-0.99)0.048Chronic pulmonary disease212 (21.1)86 (17.8)126 (24.0)1.46 (1.07-1.99)0.019Diabetes427 (42.4)186 (38.5)241 (46.0)1.36 (1.06-1.75)0.019Chronic liver disease16 (1.6)4 (0.8)12 (2.3)2.81 (0.9-8.76)0.109Stroke75 (7.4)35 (7.2)40 (7.6)1.06 (0.66-1.7)0.909Chronic kidney disease227 (22.5)92 (19.0)135 (25.8)1.47 (1.09-1.99)0.013Cerebrovascular disease131 (13.0)63 (13.0)68 (13.0)0.99 (0.69-1.44)1GCS14.00 [13.00–15.00]14.00 [14.00, 15.00]14.00 [12.00, 15.00]0.97 (0.94-1)0.007SOFA6.00 [4.00–8.00]5.00 [4.00, 7.00]6.00 [4.00, 9.00]1.15 (1.1-1.2)<0.001Heart rate (beat)82.00 [77.00, 87.00]82.00 [77.00, 87.00]81.00 [77.00, 88.00]1 (0.99-1.02)0.968MBP (mmHg)73.00 [69.00, 76.00]74.00 [70.00, 77.00]72.00 [69.00, 75.00]0.94 (0.92-0.96)<0.001SPO2 (%)98.00 [97.00, 99.00]98.00 [97.00, 99.00]98.00 [97.00, 99.00]1.01 (0.92-1.11)0.872IABP63 (6.3)20 (4.1)43 (8.2)2.07 (1.2-3.57)0.011Ventilator872 (86.6)414 (85.7)458 (87.4)1.16 (0.8-1.66)0.488Vasopressor876 (87.0)411 (85.1)465 (88.7)1.38 (0.95-2)0.104Preoperative laboratories Albumin (mg/dL)4.00 [3.60–4.30]4.00 [3.70–4.40]3.90 [3.50–4.20]0.5 (0.39-0.65)<0.001Creatinine (mg/dL)1.00 [0.90–1.30]1.00 [0.90, 1.20]1.10 [0.90, 1.30]1.51 (1.2-1.9)0.004Hematocrit (%)37.60 [33.20–41.10]38.50 [34.50–41.50]36.35 [31.80–40.50]0.96 (0.94-0.98)<0.001BUN (mg/dL)20.00 [16.00–26.00]20.00 [16.00, 25.00]21.00 [16.00, 27.00]1.02 (1.01-1.03)0.063Sodium (mmol/L)139.00 [137.00–141.00]139.00 [137.00, 141.00]139.00 [137.00, 141.00]0.99 (0.95-1.02)0.437Potassium (mmol/L)4.20 [4.00–4.50]4.20 [4.00, 4.50]4.20 [4.00, 4.50]0.97 (0.73-1.29)0.36Bicarbonate (mmol/L)26.00 [24.00–28.00]26.00 [24.00, 28.00]26.00 [24.00, 28.00]0.97 (0.94-1.01)0.215Lymphocyte (103/μL)1.63 [1.24–2.12]1.68 [1.29, 2.18]1.57 [1.19, 2.00]0.83 (0.71-0.96)0.002GNRI114.27 (12.57)112.8 (11.46)115.62 (13.37)1.02 (1.01-1.03)0.001GNRI ≤ 98100 (9.9)49 (10.1)51 (9.7)1.05 (0.69-1.58)0.91PNI48.39 [44.15–52.64]49.45 [45.64, 53.91]47.02 [42.83, 51.65]0.95 (0.93-0.97)<0.001PNI < 48480 (47.7)194 (40.2)286 (54.6)0.56 (0.43-0.72)<0.001Postoperative laboratories Creatinine (mg/dL)0.90 [0.70–1.10]0.90 [0.70, 1.00]0.90 [0.70, 1.10]1.83 (1.37-2.45)<0.001BUN (mg /dL)17.00 [14.00–21.00]16.00 [13.00, 20.00]18.00 [14.00, 23.00]1.03 (1.02-1.05)<0.001Sodium (mmol/L)139.00 [137.00–141.00]139.00 [137.00, 141.00]139.00 [137.00, 141.00]0.97 (0.93-1.01)0.089Potassium (mmol/L)4.30 [3.90–4.60]4.20 [3.90, 4.60]4.30 [4.00, 4.70]1.23 (0.98-1.55)0.069Bicarbonate (mmol/L)23.00 [21.00–24.00]23.00 [21.00–24.0023.00 [21.00–24.000.98 (0.93-1.04)0.931Time interval (h)3.50 [2.48–4.86]3.39 [2.48, 4.76]3.57 [2.50, 4.99]1.02 (0.95-1.09)0.364outcome LOS in ICU (day)1.98 [1.26–3.22]1.35 [1.16, 2.25]2.37 [1.37, 4.17]1.34 (1.24-1.45)<0.001LOS in hospital (day)8.05 [5.74–11.45]7.05 [5.30, 9.91]9.13 [6.61, 12.70]1.11 (1.08-1.14)<0.001Hospital mortality (%)13 (1.3%)3 (0.62%)10 (1.91%)3.11 (0.85-11.38)0.126Mortality 30‐day (%)16 (1.6%)5 (1.04%)11 (2.10%)2.05 (0.71-5.94)0.273Mortality 90‐day (%)38 (3.8%)12 (2.48%)26 (4.96%)2.05 (1.02-4.11)0.058Values are expressed as median (interquartile range) or mean (standard deviation) and frequencies (percentages). AKI, acute kidney injury; BMI, body mass index; GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; MBP, mean blood pressure; SPO2, oxyhemoglobin saturation by pulse oximetry; IABP, intra-aortic balloon pump; GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index; BUN, blood urea nitrogen; LOS, length of stay. Baseline characteristics of study participants. Values are expressed as median (interquartile range) or mean (standard deviation) and frequencies (percentages). AKI, acute kidney injury; BMI, body mass index; GCS, Glasgow coma scale; SOFA, sequential organ failure assessment; MBP, mean blood pressure; SPO2, oxyhemoglobin saturation by pulse oximetry; IABP, intra-aortic balloon pump; GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index; BUN, blood urea nitrogen; LOS, length of stay. Prevalence and distribution of malnutrition The distributions of GNRI and PNI were indicated. The prevalence of malnutrition among the patients was found to be 9.9% when assessed by the GNRI and 47.9% when evaluated through the PNI (Figure 2). There was a significant correlation between these two indicators (p < 0.001, r = 0.541). Patients with malnutrition evaluated by GNRI and PNI were older, had a lower BMI, a worse SOFA score, and a higher utilization rate of IABP compared with patients in normal nutritional status. In addition, they tend to have a higher probability of having congestive heart failure and a lower percentage of combining hypertension (Tables S2–S3). Population distribution and incidence of malnutrition. Histograms show the population distribution of GNRI (a), and PNI (B). Venn diagram of malnutrition risk assessed by the GNRI and PNI(C). The frequency of malnutrition is shown by the frequencies and percentages outside of each circle. The area where two circles overlap indicates the frequency and percentage of a malnutrition diagnosis with GNRI and PNI coincide. GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index. The association between nutrition scores and moderate-to-severe AKI Covariables that were associated with moderate-to-severe AKI following CABG surgery in the adjusted model were screened (Table S4). GNRI and PNI were firstly incorporated as continuous variables for logistic regression, revealing that they were significantly associated with the odds ratio of moderate to severe AKI, ranging from the univariate model to different multivariate models (all p < 0.05; Table 2). After adjusting the screened variable by multivariable analysis, the RCS was utilized to illustrate the association between them. A linear association was indicated between nutrition scores and the OR of moderate to severe AKI (p for nonlinearity > 0.05). Specifically, the OR of moderate-to-severe AKI increased as GNRI score elevated. Regarding PNI, there was a significant decline in the OR of moderate to severe AKI with increasing PNI. The optimal cutoff points of GNRI and PNI were 113.73 and 48.28, respectively (Figure 3). Table 2.Univariable and multivariable analyses of GNRI and PNI to predict moderate to severe AKI after CABG. Univariable analysisMultivariable analysis model 1Multivariable analysis model 2Multivariable analysis model 3 OR (95%) P OR (95%) P OR (95%) P OR (95%) P Nutritional indices as continuous variables GNRI1.018 (1.008–1.029)<0.0011.022 (1.011–1.033)<0.0011.03 (1.018–1.041)<0.0011.027 (1.016-1.039)<0.001 PNI0.947 (0.928–0.966)<0.0010.955 (0.936–0.974)<0.0010.964 (0.944–0.984)0.0010.963 (0.943-0.984)0.001Nutritional indices as category variables GNRI > 98Ref. Ref. Ref. Ref. GNRI ≤ 980.955 (0.632–1.444)0.8270.803 (0.524–1.23)0.3130.69 (0.441–1.079)0.1040.72 (0.458-1.132)0.155 PNI ≥ 48Ref. Ref. Ref. Ref. PNI < 481.79 (1.394–2.299)<0.0011.577 (1.218–2.042)0.0011.332 (1.013–1.751)0.041.338 (1.016-1.763)0.038Multivariable analysis model 1: adjusting for gender, congestive heart failure and chronic pulmonary disease. Multivariable analysis model 2: adjusting for variables in model 1 as well as SOFA score, MBP, preoperative hematocrit, and BUN. Multivariable analysis model 3: adjusting for variables in model 2 as well as postoperative creatinine and BUN. GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index; AKI, acute kidney injury; CABG, coronary artery bypass grafting; OR, odds ratio, Ref, reference. Univariable and multivariable analyses of GNRI and PNI to predict moderate to severe AKI after CABG. Multivariable analysis model 1: adjusting for gender, congestive heart failure and chronic pulmonary disease. Multivariable analysis model 2: adjusting for variables in model 1 as well as SOFA score, MBP, preoperative hematocrit, and BUN. Multivariable analysis model 3: adjusting for variables in model 2 as well as postoperative creatinine and BUN. GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index; AKI, acute kidney injury; CABG, coronary artery bypass grafting; OR, odds ratio, Ref, reference. Dose–response relationships between GNRI (A) and PNI (B) and moderate to severe AKI in older patients undergoing CABG. The red bold line illustrates the or, while the shaded area indicates the 95% CI. They were adjusted for gender, congestive heart failure, chronic pulmonary disease, SOFA, MBP, preoperative hematocrit, preoperative BUN, postoperative creatinine and postoperative BUN. Vertical dotted lines reflect the minimal threshold for the beneficial association with estimated or = 1. Abbreviations: GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index; AKI, acute kidney injury; CABG, coronary-artery-bypass-grafting. OR, odds ratio; CI, confidence interval; SOFA, sequential organ failure assessment; MBP, mean blood pressure; BUN, blood urea nitrogen. GNRI and PNI scores were also distinguished into categorical variables to further study the association between nutrition risk and moderate-to-severe AKI after CABG surgery via logistic regression. As the results demonstrated, there was no significant association between the nutrition risk calculated by GNRI and the incidence of moderate to severe AKI (p > 0.05). In contrast, the PNI-defined nutrition risk was significantly associated with the elevated risk of moderate to severe AKI (p < 0.05; Table 2). To be specific, the odds ratio of moderate-to-severe AKI was more than 1.338(1.016–1.763) times greater than those without nutrition risk. Assessing the association between the malnutrition and the risk of moderate-to-severe AKI among geriatric patients underwent CABG surgery, while taking into categorical variances such as gender, congestive heart failure, and chronic pulmonary disease. The results demonstrated that there were no significant interaction effects across different subgroups in the subgroup analysis (Figure 4). Subgroup analysis for the association of malnutrition regrading as categorical variables and moderate to severe AKI in different subgroups. OR of GNRI (a), and PNI (B) were adjusted by the variables that are stratified by itself only. OR, odds ratio; GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index. Incremental value of nutrition scores and comparative analysis We firstly consider the GNRI and PNI scores as continuous data. When adding them separately to three base models, the results demonstrated that the addition of GNRI or PNI to the base model 1 significantly elevated AUC and IDI, but the NRI weren’t significantly improved (Table 3 and Figure S1A). The AUC and IDI of the base model 2 elevated significantly when the two indices were included, but only GNRI significantly improved NRI (Table 3 and Figure S1B). Similarly, adding the two indices to the base model 3 significantly improved AUC and IDI without significant NRI changes (Table 3 and Figure S1C). When adding two indicators simultaneously to three base models, the results indicated that it can significantly improve the AUCs, NRIs, and IDIs (Table 3 and Figure S1). When the two nutritional indicators were compared, the results revealed that their AUCs were similar and that their differences of NRI and IDI were not statistically significant in different models (Table S5 and Figure S1). Table 3.Performance of models with GNRI and PNI to predict the postoperative moderate to severe AKI in older patients undergoing CABG. AUCNRIIDI AUC (95%) P index (95%) P index (95%) P Nutritional indices as continuous variablesmodel10.607 (0.574 − 0.64) model1 + GNRI0.635 (0.601 − 0.67)0.0090.027 (-0.012 − 0.067)0.1760.017 (0.009 − 0.024)<0.001model1 + PNI0.64 (0.606 − 0.674)0.0040.024 (-0.019 − 0.067)0.2660.024 (0.015 − 0.034)<0.001model1 + GNRI+PNI0.708 (0.676 − 0.739)<0.0010.097 (0.039 − 0.155)0.0010.083 (0.067 − 0.100)<0.001model20.659 (0.626 − 0.693) model2 + GNRI0.685 (0.653 − 0.718)0.0070.059 (0.009 − 0.110)0.0210.026 (0.016 − 0.035)<0.001model2 + PNI0.674 (0.641 − 0.706)0.050−0.0134 (-0.056 − 0.029)0.5340.016 (0.008 − 0.024)<0.001model2 + GNRI+PNI0.731 (0.70 − 0.761)<0.0010.0953 (0.036 − 0.154)0.0020.075 (0.059 − 0.091)<0.001model30.674 (0.642 − 0.707) model3 + GNRI0.694 (0.661 − 0.726)0.0200.0082 (-0.034 − 0.050)0.7010.020 (0.012 − 0.029)<0.001model3 + PNI0.688 (0.656 − 0.721)0.0420.0227 (-0.019 − 0.064)0.2830.016 (0.008 − 0.024)<0.001model3 + GNRI+PNI0.737 (0.706 − 0.767)<0.0010.1042 (0.049 − 0.160)0.0000.067 (0.052 − 0.082)<0.001Nutritional indices as category variablesmodel10.607 (0.574 − 0.64) model1 + GNRI0.611 (0.577 − 0.644)0.143−0.007 (-0.015 − 0)0.0450.001 (-0.001 − 0.003)0.320model1 + PNI0.622 (0.588 − 0.656)0.141−0.006 (-0.048 − 0.035)0.7700.011 (0.005 − 0.018)0.001model1 + GNRI+PNI0.627 (0.593 − 0.661)0.0570 (-0.039 − 0.039)0.9840.015 (0.007 − 0.022)<0.001model20.659 (0.626 − 0.693) model2 + GNRI0.662 (0.628 − 0.695)0.512−0.006 (-0.027 − 0.015)0.5720.002 (-0.001–0.005)0.158model2 + PNI0.663 (0.629 − 0.696)0.486−0.036 (-0.075 − 0.003)0.0690.004 (0 − 0.008)0.049model2 + GNRI+PNI0.666 (0.633 − 0.699)0.270−0.018 (-0.061 − 0.024)0.3950.008 (0.002 − 0.013)0.006model30.674 (0.642 − 0.707) model3 + GNRI0.676 (0.643 − 0.708)0.655−0.005 (-0.028 − 0.018)0.6800.002 (-0.001 − 0.004)0.258model3 + PNI0.678 (0.645 − 0.71)0.444−0.023 (-0.059 − 0.014)0.2210.004 (0 − 0.008)0.043model3 + GNRI+PNI0.68 (0.647 − 0.712)0.332−0.005 (-0.045 − 0.034)0.7920.007 (0.002 − 0.012)0.009Multivariable analysis model 1: adjusting for gender, congestive heart failure and chronic pulmonary disease. Multivariable analysis model 2: adjusting for variables in model 1 as well as SOFA score, MBP, preoperative hematocrit and BUN. Multivariable analysis model 3: adjusting for variables in model 2 as well as postoperative creatinine and BUN. GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index; AKI, acute kidney injury; CABG, coronary artery bypass grafting; AUC, area under curve; NRI, Net reclassification improvement; IDI, Integrated discrimination improvement. Performance of models with GNRI and PNI to predict the postoperative moderate to severe AKI in older patients undergoing CABG. Multivariable analysis model 1: adjusting for gender, congestive heart failure and chronic pulmonary disease. Multivariable analysis model 2: adjusting for variables in model 1 as well as SOFA score, MBP, preoperative hematocrit and BUN. Multivariable analysis model 3: adjusting for variables in model 2 as well as postoperative creatinine and BUN. GNRI, geriatric nutritional risk index; PNI, prognostic nutritional index; AKI, acute kidney injury; CABG, coronary artery bypass grafting; AUC, area under curve; NRI, Net reclassification improvement; IDI, Integrated discrimination improvement. Next, we calculated the incremental predictive value of the GNRI and PNI regarding as categorical variables. When adding them respectively to three base models, the addition of everyone to the base model 1 did not significantly improve the AUC and NRI, but PNI merely leads to a significant increment in terms of IDI (Table 3 and Figure S2A). This outcome was consistent with model 2 and 3 (Table 3 and Figure S2B–C). When adding two indicators to three base models at the same time, the results showed that it can significantly improve the IDIs (Table 3 and Figure S2). When the two nutritional indicators were compared, the results indicated that they have no significant differences in terms of AUCs, NRIs, and IDIs (Table S5 and Figure S2). In addition, calibration curves also demonstrated good performance between model-predicted moderate to severe AKI risk and the actual moderate to severe AKI probability in three models (Figure S3–S4). Finally, multivariable models integrating different nutritional scores showed higher net benefits at decision thresholds between 30% and 80%, according to DCA (Figure S5–S6). Discussion Main findings In the retrospective study, the results revealed that preoperative nutritional status, a potential independent predictor, was significantly associated with moderate-to-severe AKI following CABG surgery in elderly patients. Additionally, we discovered that integrating nutritional indices into the basic predictive models enhanced their predictive performance. There was no difference in the incremental prediction significance of the two indicators. Possible explanations for findings The precise mechanisms by which malnutrition contributes to AKI remain to be fully determined. However, the pronounced catabolism, a characteristic of critical illness following CABG surgery, provides a partial explanation [33]. Excessive catabolism can damage renal and adrenal cells through increased inflammatory responses and oxidative stress. This not only directly precipitates AKI but also disrupts systemic glucose and lipid metabolism and induces insulin resistance by impacting adrenal function, thereby exacerbating malnutrition [34,35]. Furthermore, heightened catabolism leads to reduced serum albumin levels and plasma colloid osmotic pressure, causing intravascular water to shift to peripheral tissues. This shift results in a decreased effective circulating blood volume, subsequently impairing renal perfusion and precipitating AKI. Additionally, the inflammatory response can induce vasoconstriction, further leading to renal hypoperfusion [33,36]. Albumin has been conventionally regarded as an indicator of patients’ nutritional status. However, this perspective is increasingly challenged by contemporary studies [37,38]. Recent studies suggest that albumin characterizes inflammatory processes rather than nutritional status or protein-energy malnutrition. Inflammatory mediators significantly suppress the synthesis of albumin and induce the redistribution of serum proteins by increasing capillary permeability as part of the inflammatory response [39,40]. Therefore, albumin more accurately predicts the risk of adverse outcomes in patients, diverging from its traditional associated with malnutrition. In addition, lymphocytes are the important laboratory results to assess nutritional status. Due to the characteristics of immunosuppression in the critical illness, lymphocytes played an integral part in the occurrence, development and recovery of AKI [34,41]. Many clinical studies indicated that decreased preoperative lymphocyte counts were associated with incremental risk of postoperative AKI and in-hospital mortality [42,43]. The widely used cutoff values of GNRI to assess malnutrition was established by Bouillanne et al. [27]. However, the cutoff values may vary across different patient populations and diseases. For instance, the optimal GNRI cutoff to predict overall survival among older patients with diffuse large B-cell lymphoma was 106.26 [29], whereas for elderly patients who received curative resection for colorectal cancer, the cutoff was 101.1 [44]. Furthermore, the ideal GNRI cutoffs for predicting mortality among adult hemodialysis patients varied according to age and gender [45]. In addition, the above studies primarily examined the relationship between GNRI and mortality, while this study concentrates on the association between GNRI and the incidence of moderate to severe AKI. In our findings, the optimal GNRI cutoff value for predicting moderate to severe AKI is determined to be 113.42 by RCS. This may be the reason why there was no significant association between malnutrition and moderate-to-severe AKI when 98 was regarded as the cutoff value. We also find that the OR for moderate-to-severe AKI increases as the GNRI value raises, which is consistent with previous studies [46,47]. This underscores the significance of considering the GNRI cutoff variability when assessing risk and prognosis across different medical conditions and patients. The GNRI is largely determined by actual body weight. Increased GNRI may have elevated the risk of moderate to severe AKI in patients underwent the CABG surgery. The pathophysiological mechanism behind this may be associated with obesity-related renal dysfunction. Several possible mechanisms have been suggested to explain. On the one hand, obesity directly contributes to kidney damage by altering hemodynamics. It leads to a state of ‘hyperperfusion, hyperpressure, and hyperfiltration’ in the glomeruli, which increases the tension in the glomerular capillary walls and compromises the glomerular filtration barrier. This alteration causes the leakage of proteins, which are toxic to mesangial cells and podocytes in the kidneys, thereby exacerbating tubulointerstitial damage and accelerating glomerulosclerosis, which ultimately leads to kidney injury [48,49]. On the other hand, obesity is commonly associated with metabolic dysregulation, characterized by insulin resistance and a low-grade inflammatory state. These conditions collectively exacerbate renal damage by promoting systemic and renal-specific inflammatory pathways [50,51]. Such mechanisms underscore the intricate link between obesity and kidney health, underscoring the significance of managing nutritional status by regulating obesity to mitigate its adverse effects on renal function. Clinical implications This study has underlying clinical impact, especially for elderly patients who will undergo the CABG surgery. First, it underscores the significant association between preoperative nutritional condition and postoperative moderate to severe AKI. Previous studies have indicated that most patients after cardiac surgery exist late nutritional support and inadequate feeding [52]. Screening and managing of malnutrition were essential parts of the surgical enhanced recovery program. Utilizing GNRI and PNI facilitates earlier and more precise assessment of nutritional status, providing new insights for perioperative nutritional intervention. Second, extensive studies have indicated that preoperative malnutrition is a major risk factor contributing to postoperative AKI [19,20,53]. This study further confirms that the predictive performance of models incorporating nutritional scores is superior. This aids in prediction of moderate-to-severe AKI in advance, providing novel theoretical support for comprehensive management of AKI. The guideline has also recommended to evaluate and manage preoperatively malnutrition before cardiac surgery to decrease the incidence of postoperative complications [54]. Third, the specialized nutritional interventions were required for patients with severe AKI to correct electrolyte disorders and negative nitrogen balance [55,56]. Although studies have suggested that the timing of parenteral nutrition intervention may affect the time required for renal function recovery [57], literature regarding nutritional intervention’s safety and efficacy among AKI patients is limited and a suitable approach for enhancing nutritional status remains for patients with severe AKI was unclear. There was no standard strategy for improving nutritional status because every patient’s nutritional condition and comorbidities vary. Further studies are needed to explore the effects of different nutritional managements on renal function recovery in patients with severe AKI, especially for older patients after CABG surgery. Strengths and limitations Our study possesses several notable strengths. Primarily, it concentrated on the high-risk individuals: older patients undergoing CABG surgery. Additionally, our research evaluated the association between preoperative nutritional status and postoperative moderate to severe AKI in the elderly underwent CABG surgery, as well as the predictive performance of various nutritional indicators for moderate to severe AKI through comprehensive approaches. Lastly, we made a comparative analysis to have a better understanding of the predictive efficacy of GNRI and PNI, which contributed valuable insights into their relative performance. The study is not without limitations. First, being a single-center, retrospective, and observational study, it didn’t establish a causal link between malnutrition and moderate to severe AKI, which multi-center studies are necessary to further validation. Second, the absence of CONUT, MUST, and NRS-2002 scores limited the generalizability of our findings, as these indices are more commonly employed in clinical settings. Third, the nutritional indices used may not capture the entire spectrum of malnutrition, omitting aspects like vitamin deficiencies [58,59], skeletal muscle index [60,61], and frailty [19], which are known to be associated with AKI. Fourth, cardiopulmonary bypass is an independent risk factor for postoperative AKI. However, the relevant information was lacked in the MIMIC database, and we cannot further explore the predictive performance of preoperative nutritional status in off-pump or on-pump patients. Last, the exclusion of many patients due to incomplete preoperative nutritional information or AKI data could introduce selection bias. Furthermore, potential influences on GNRI and PNI indices from internal factors such as hormonal fluctuations, inflammation, and changes in the circulating blood volume were not accounted for, possibly leading to another selection bias. Conclusion This study reported malnutrition status, evaluated by GNRI and PNI, was independently associated with the incidence of moderate to severe AKI in geriatrics following CABG surgery. Adding two indicators to base models of moderate to severe AKI made the predictive performance better. Collectively, these results underscored the potential value of GNRI and PNI monitoring in forecasting moderate to severe AKI following CABG, thereby paving the way for future research on their clinical utility and implementation. Supplementary Material supplementary material.docx Acknowledgement We would like to thank the Massachusetts Institute of Technology and the Beth Israel Deaconess Medical Center for the MIMIC Project. Funding Statement This research was supported by the National Key R&D Program of China [number 2022YFC2010001]; Henan Provincial Natural Science Foundation under Grant [number 242300421291]; Henan Province Young and Middle-aged Health Science and Technology Innovation Outstanding Talent Training Project Grant [number YXKC2022043]; and Henan Provincial Health Commission Young and Middle-aged Discipline Leader Project Under Grant [number HNSWJW-2022010]. Authors’ contributions P.B.: study design, data extraction, data analysis, and writing the manuscript. P. Q. and T.L.: data processing and data analysis. X.L., J.W., and S.W.: data processing and literature search. H.L. and Z.G.: study design, supervision, funds supporting, and manuscript revision. All authors read and approved the final manuscript. Ethics approval and consent to participate The establishment of this database was approved by the Massachusetts Institute of Technology (Cambridge, Massachusetts) and Beth Israel Deaconess Medical Center (Boston, Massachusetts), and consent was obtained for the original data collection. One author has received the certification (ID number: 60692864). Therefore, the ethical approval statement and the need for informed consent were waived for this manuscript. Consent for publication Not applicable. Disclosure statement No potential conflict of interest was reported by the author(s). Data availability statement Data are available in a public, open access repository. The datasets utilized during the current study are available in the MIMIC- IV repository (https://physionet.org/content/mimiciv/2.2/). References Associated Data Supplementary Materials supplementary material.docx Data Availability Statement Data are available in a public, open access repository. The datasets utilized during the current study are available in the MIMIC- IV repository (https://physionet.org/content/mimiciv/2.2/).